p
UNIVERSITY OF
ILLINOIS LIBRARY
AT USBANA-CHAMPAIGN
UNIVERSITY LIBRARY
UNIVERSITY OF ILLINOIS AT URBANA-CHAMPAIGN
The person charging this material is responsible for its
renewal or return to the library on or before the due
date. The minimum fee for a lost item is $125.00,
$300.00 for bound journals.
Theft mutilation, and underlining of books are reasons
for disciplinary action and may result in dismissal froni
the University. Please note: self-stick notes may result
in torn pages and lift some inks.
Renew via the Telephone Center at 217-333-8400,
846-262-1510 (toll-free) orcirclib@uiuc.edu.
Renew online by choosing the My Account option at:
http://www.library.uiuc.edu/catalog/
410 ^
St92
20:1 SPR 1990 COPY 2
1)20, No. 1
Spring 1990
THE CONTRIBUTION OF AFRICAN LINGUISTICS
TO LINGUISTIC tH£ORY, VOL 2
Preface
Introduction
Part I: Plenary Address^
EYAMBAG. BOKAMBA: AfricarOangua^s and sc^Iinguistic theories 3
JOAN BRESNAN: African languagestend syi^tic theories 35
JOHN GOLDSMITH: Phonological theo^ and African language
phonology v>> 49
SalikOKO MUFWENE: African languag^, African linguistics, and
linguistic theory: A commentary on the plenary session papers 63
Part II: Phonology
Ellen BROSELOW & Alice NIYONDAGARA: Feature geometry of
Kirundi palatalization 71
G. N. CLEMENTS & REMl SONAIYA: Underiying feature
representation in Yoruba 89
Omar K A: Reduplication and prosodic constituents in Wolof 105
MANUELA NOSKE: Vowel Harmony in Turkana 1 23
METERWA A. OURSO & CHARLES H. ULRICH: Sonorant-
strengthening in Lama 135
Part III: Tonology
CHARLES KISSEBERTH & SHEILA MMUSI: The tonology of the
object prefix in Setswana 151
NGESSIMO MUTAKA: The tone bearing unit in Kinande 163
Part IV: Sociolinguistics & History of Linguistics
ANDRE MWAMBA KapanGA: Language variation and language
attitudes: A case study from Shaba Swahili 175
Margaret Wade-LEWIS: The contribution of Lorenzo Dow
Turner to African linguistics 1 89
/
Department of Linguistics
University of Illinois
STUDIES IN THE LINGUISTIC SCIENCES
PUBLICATION OF THE DEPARTMENT OF LINGUISTICS
COLLEGE OF LIBERAL ARTS AND SCIENCES
UNIVERSITY OF ILLINOIS AT URBANA-CHAMPAIGN
EDITOR: Hans Henrich Hock; REVIEW EDITOR: James H. Yoon
EDITORIAL BOARD: Eyamba G. Bokamba, Chin-chuan Cheng, Jennifer S.
Cole, Georgia M. Green, Hans Henrich Hock, Braj B. Kachru, Yamuna Kachru,
Chin-W. Kim, Charles W. Kisseberth, Howard Maclay, Jerry L. Morgan,
Rajeshwari Pandharipande, James H. Yoon, and Ladislav Zgusta.
AIM: SLS is intended as a forum for the presentation of the latest original
research by the faculty and especially students of the Department of Linguistics,
University of Illinois, Urbana-Champaign. Invited papers by scholars not as-
sociated with the University of Illinois will also be included.
SPECIAL ISSUES: Since its inception SLS has devoted one issue each year to
restricted, specialized topics. A complete list of such special issues is given on the
back cover.
BOOKS FOR REVIEW: Review copies of books may be sent to the Editor,
Studies in the Linguistic Sciences, Department of Linguistics, University of
Illinois, 4088 Foreign Languages Building, 707 S. Mathews, Urbana, Illinois
61801.
SUBSCRIPTION: Normally, there are two issues per year. Requests for sub-
scriptions should be addressed to SLS Subscriptions, Department of Linguistics,
University of Illinois, 4088 Foreign Languages Building, 707 South Mathews,
Urbana, Illinois 61801.
UPCOMING ISSUES: Vol. 22:2: Twenty-five years of Linguistic Research at
the University of Illinois at Urbana-Champaign (Post-graduate research by
Doctoral and Master's degree students in the Department of Linguistics). Editors:
Braj B. Kachru and Frances Vavrus; Vol. 23:1: Papers in General Linguistics.
Price: $7.50 (per issue)
STUDIES IN THE LINGUISTIC SCIENCES
The Contribution of African Linguistics
TO Linguistic Theory
Proceedings of the 20th Annual Conference on African Linguistics
VOL. H
EDITOR
Eyamba G. Bokamba
ASSOCIATE EDITORS
Amy C. Cheatham
Dorothy E. Evans
Rick Treece
VOLUME 20, NUMBER 1
SPRING 1990
department of linguistics, university of ILLINOIS
URBANA, ILLINOIS 61801
CONTENTS
Preface v
Introduction vii
Part I: Plenary Addresses
EyambaG. Bokamba: African languages and sociolinguistic theories 3
Joan Bresnan: African languages and syntactic theories 35
John Goldsmith: Phonological theory and African language phonology 49
Salikoko Mufwene: African languages, African linguistics, and linguistic
theory: A commentary on the plenary session papers 63
Part II: Phonology
Ellen Broselow & Alice Niyondagara: Feature geometry of Kirundi
palatalization 71
G. N. Clements & Remi Sonaiya: Underlying feature representation
in Yoruba 89
Omar Ka: Reduplication and prosodic constituents in Wolof 105
Manuela Noske: Vowel Harmony in Turkana 1 23
fvlETERWA A. OURSO & CHARLES H. Ulrich: Sonorant-strengthening
in Lama 135
Part ill: Tonology
Charles Kisseberth & Sheila Mmusi: The tonology of the object prefix
in Setswana 151
Ngessimo Mutaka: The tone bearing unit in Kinande 163
Part IV: Sociolinguistlcs & History of Linguistics
Andre Mwamba Kapanga: Language variation and language attitudes:
A case study from Shaba Swahili 1 75
Margaret Wade-Lewis: The contribution of Lorenzo Dow Turner to
African linguistics 189
PREFACE
This issue of SLS contains the second set of the papers selected from the
20th Annual Conference on African Linguistics held at the University of Illinois,
Urbana-Champaign, April 19-22, 1989. A number of factors, including delays in
the vetting of the papers, editorial work, and overwhelming new administrative
responsibilities for me account for this belated publication. I deeply regret this
development and extend my sincere apologies to the authors and the SLS
subscribers who have had to wait for so many years to see this issue in print.
Your patience is very much appreciated, and I hope the appearence of this fine
set of papers will merit it.
As in the previous issue (SLS 19:2), the papers included in this issue were
vetted by at least two faculty members selected from a number of universities in
the U.S. Their comments were initially reviewed by the editors and
subsequently forwarded anonymously to the authors of the papers concerned
for action. We relied heavily on the opinions of these reviewers in accepting or
rejecting any of the papers submitted for consideration. The revised
contributions were subsequently reviewed by the editor for consistency and
final editorial changes wherever this was necessary. Overall, our aim has been
to produce a volume that is representative of current research in African
languages that is informed by various contemporary theories.
We are most grateful to the following scholars for reviewing the papers in
this issue: A. Adun, E. G. Bokamba, J. Bresnan, C. C. Cheng, G. N. Clements, J.
Cole, C. E. DeBose, D. Evans, M. Goodman, H. H. Hock, J. I. Hualde, B. B.
Kachru, J. Karneva, M. Kenstowicz, C. W. Kim, W. Leben, S. S. Mufwene, D.
Odden, and C. H. Ulrich.
Several academic units at this University and elsewhere co-sponsored the
20th ACAL from which the papers included here were selected. We take this
opportunity to express our gratitude to these units for their financial and material
support: African Studies Center (Michigan State University, East Lansing), the
Center for African Studies (University of Illinois at Urbana-Champaign), the
Department of Linguistics (UIUC), the Miller Endowment Fund (Center for
Advanced Study, UIUC), the College of Liberal Arts and Sciences (UIUC), the
Office of the Vice-Chancellor for Academic Affairs (UIUC), the African Student
Organization (UIUC), and the Africa-Related Women Association (UIUC).
Work on this issue has been done in stages by different individuals since
the conclusion of the conference. As the proceedings' editor, I am indebted to
Dr. Dorothy E. Evans and Dr. Rick E. Treece for their various contributions in the
preparation of the manuscripts, and to Ms. Cathy Huffman and Eileen Sutton
(secretaries. Department of Linguistics) for reword processing three of the
papers included here. I owe a greater debt of gratitude to Ms. Amy C.
Cheatham, an MA candidate in linguistics and graduate assistant for SLS
(UIUC) for the final reformating and proof-reading of the entire manuscript;
without her meticulous rereading of the papers and hard work beyond the call of
vi Studies in the Linguistic Sciences 20:1 (Spring 1990)
duty, many of the contributions in this issue would have appeared with
unexplained gaps and errors of various sorts.
Once again, I take full responsibility for the long delay in the publication of
this issue. The associate editors and I hope that you will find these studies as
informative and stimulating as those in the first issue.
Urbana, Illinois Eyamba G. Bokamba
June 1993 Editor
INTRODUCTION
Eyamba G. Bokamba
University of Illinois at Urbana-Champaign
This issue of SLS contains the second set of papers which comprise the
second volume of what we have titled "The Contribution of African Linguistics to
Linguistic Theory." Salikoko S. Mufwene (The University of Chicago), in his
commentary on the plenary session papers (pp. 63-70), raises the question of
what actually constitutes this contribution. Is it, he asked, the contribution of
African language data to the analysis of standard problems in linguistics, or is it
the presentation of abnormal problems and the postulation of new theories for
their analysis?
The answer that emerges from the thirteen papers included in this volume
is that the contribution is both data-based and theoretically oriented. This point
is made first and forcefully in the plenary session papers by Bokamba (pp. 3-
34), Bresnan (pp. 35-48), and Goldsmith (pp. 49-62). Bokamba's paper offers
three types of discussions: First, it summarizes several areas of sociolinguistics
and indicates the contribution of African linguistic data in extending the
application of "standard" analyses to accommodate such facts. Second, it points
out, wherever relevant, the contributions that Africanist linguists have made
theoretically in forcing a redirection of certain analytical views/approaches as a
result of African data — the unusual problem. And third, the paper points out the
deficiencies in African sociolinguistic research and suggests a research agenda
in sociolinguistics, arguing that Africanist linguists are best placed to pursue
various issues related to language in its social context than many other area-
scholars.
Bresnan's study also takes a topical approach, focusing on four syntactic
phenomena that represent both the common and uncommon type of problems:
(1) logophoricity (uncommon), (2) topic, pronoun, and agreement (common), (3)
hierarchies and argument asymmetries (less common), and (4) syntax of verbs
(common). In discussing each of these Bresnan provides a critique of the
Chomskyan syntactic theory in handling or not dealing at all with these
phenomena, and points out the extent to which African data have forced
theoretical syntacticians to incorporate them in their analyses. The work of Alec
Marantz (1984), Hilda Koopman (1984), Mark C. Baker (1988), Bresnan's own
research and that of her colleagues (e.g., Sam Mchombo, Lioba Moshi, Alex
Alsina, Jonni Kanerva) on Bantu languages in the development of LFG (Lexical
Functional Grammar) in the past several years (cf. 1986, 1987, 1988, 1989,
1990) constitute eloquent examples of both data-driven changes and theory
formulation based on African languages. She observes that while the impact of
these changes have not been as profound as those in phonology, they have
nonetheless impacted significantly syntactic theory and will continue to do so in
the forseeable future.
viii Studies in the Linguistic Sciences 20:1 (Spring 1990)
John Goldsmith's paper approaches the question of African linguistics
from a "continuity" perspective. He states that "no historian of modern linguistics
can understand the continuities In our field without tracing them through fields
such as African linguistics, for that is where the important ideas of our times live,
prosper, and remain fertile, often despite the Balkanizing effects of linguistic
theory" (p. 50). Goldsmith focusses on prosodic phonology and traces
developments In this area from Firthian phonology to contemporary
autosegmental and metrical theory to show the direct and indirect contribution
of African linguistics to this evolution. He suggests that if one views African
language data and Africanist-originating theories as contributing ultimately to a
better understanding of language structure in general, then the dichotomy
between "descriptive" and "analytical" research becomes meaningless in the
assessment of area-specific contribution to linguistic theory.
The remaining nine papers by E. Broselow and A. Niyondagara (pp. 71-
88), G. N. Clements and R. Sonaiya (pp. 89-104), O. Ka (pp. 105-22), M. Noske
(pp. 123-34), M. Ourso and C. Ulrich (pp. 135-50), C. Kisseberth and S. Mmusi
(pp. 151-62), N. Mutaka (pp. 163-73), A. Kapanga (pp. 175-88), and M. Wade-
Lewis (189-204) offer specific substantiations of the analysis of African
language data which accommodate current theories, supplement or extend
them, and in some cases critique and offer alternative analyses (to) them. Taken
together, these studies present a microcosmic view of the continuing evolution
of African linguistics and its impact on general linguistic theory.
Mufwene (pp. 63-70) observes that the best measure of the impact of a
particular field is the extent to which theories/approaches that it has originated
or influenced are published as textbooks — the standard bearers and agents of
received wisdom in academia. It is gratifying to point out in this regard that
African linguistics has indeed reached this stage in its development: Since the
writing and submission of the papers assembled in this volume, the publication
of African languages-based and/or strongly influenced textbooks has increased
tremendously. In addition to Marantz (1984) and Baker (1988) mentioned
earlier, and J. Bendor-Samuel and R. Hartell, eds. (1989), The Niger-Congo
Languages: A Classification and Description of Africa's Largest Language
Fam/Vy (Lanham: University Press of America), recent publications include: John
Goldsmith (1990), Autosegmental and Metrical Ptionology (Oxford: Basil
Blackwell); George Poulos (1990), A Linguistic Analysis of Venda (Pretoria: Via
Afrika); Ayo Bamgbose (1991), Language and ttie Nation [in Africa] (Edinburgh:
Edinburgh University Press); Jenny Cheshire, ed. (1991), English Around the
World: Sociolinguistic Perspectives (Cambridge: Cambridge University Press);
Carol Myers-Scotton (1993), Social Motivations for Codeswitching: Evidence
from Africa (Oxford: Oxford University Press); Carol Myers-Scotton (1993),
Duelling Languages: Grammatical Structure in Codeswitching (Oxford: Oxford
University Press; D. Nurse and T. Hinnebusch (1993), Swahili and Sabaki: A
Linguistic History (Berkeley: University of California Press); Louis-Jean Calvet
(1992), Les Langues des Marches en Afrique (Paris: Didier Erudition); Sam
Mchombo, ed. (In Press), Studies in Bantu Syntax and Linguistic Theory
(Stanford University: CSLI); and Salikoko S. Mufwene and Lioba Moshi, eds.(ln
Press), Topics in African Linguistics (Philadelphia: The John Benjamins
Introduction i x
Publishers). As in the past, a number of anthologies containing substantial
African linguistic contributions have been published since 1989, and both of
these trends will likely continue. It is hoped that the studies in this volume will be
a small contribution to this burgeoning literature in African linguistics.
I
Plenary Addresses
Eyamba G. Bokamba
African languages and scx;iolinguistic theory
JOAN BRESNAN
African languages and syntactic theories
JOHN Goldsmith
Phonological theory and African language phonology
Salikoko Mufwene
African languages, African linguistics, and linguistic theory (commentary)
studies in the Linguistic Sciences
Volume 20, Number 1, Spring 1990
AFRICAN LANGUAGES AND SOCIOLINGUISTIC THEORIES
Eyamba G. Bokamba
University of Illinois at Urbana-Champaign
1. Introduction
Until the 1970s sociolinguistic research on African languages dealt mainly
with language planning and a restricted range of language contact phenomena,
with particular emphasis on pidgins and Creoles. The few studies undertaken on
language variation during the two preceding decades, for instance, focused
largely on English in West Africa (cf. Spencer 1971, Sebeok 1971). Since the
mid-1970s, however, the scope of sociolinguistic research has expanded to
include language variation in African languages, language spread/diffusion,
and code-switching. In this paper I will attempt to provide an overview of this
research, with particular attention to language spread, language policy and
planning, language variation, and code-switching, ^ as a way to ascertain the
nature of and extent to which African languages have influenced developments
in sociolinguistics. The examination of the contribution of African linguistics to
the sociolinguistic theories associated with the above-mentioned areas will
involve an assessment of the adequacy of such approaches in the study of
African languages and by implication those of other multilingual societies.
The task of providing an overview of the contribution of African linguistics
to linguistic theory is a difficult and challenging one for any specialist; and this is
particularly true in regard to the assessment of the contribution of African
languages to sociolinguistic theories. There are three principal difficulties in this
type of study. First, unlike phonology, semantics, and syntax, the field of
sociolinguistics is very broad with ill-defined boundaries between subfields.
While it is generally agreed that sociolinguistics is the scientific study of
language in relation to society (Fishman 1969, Hymes 1974, Hudson 1980,
Wardhaugh 1986), what this relation encompasses remains the subject of
considerable debate. The problem is amply illustrated in the views expressed
by five distinguished sociolinguistic scholars, viz., Hymes, Trudgill, Hudson,
Wardhaugh, and Fasold.
Hymes, in a paper delivered at the 1972 Georgetown University Round
Table conference on languages and linguistics, acknowledges that:
The term 'sociolinguistics' means many things to many people, and
of course no one has a patent on its definition. Indeed not everyone
whose work is called sociolinguistic is ready to accept the label, and
4 Studies in the Linguistic Sciences 20:1 (Spring 1990)
those who do not use the term include and emphasize different
things (Hymes 1974:195).
Hymes (1974) views the scope of sociolinguistics as consisting of three
major concerns: (1) "the social as well as the linguistics;" (2) the "socially
realistic linguistics;" and (3) "socially constituted linguistics." The social and
linguistic perspective deals with social problems involving language and its
use; this is what Fishman (1969) refers to as the "sociology of language" which
he characterizes as an area of inquiry that "focuses upon the entire gamut of
topics related to the social organization of language behavior, including not
only language usage per se but also language attitudes, overt behavior toward
language and toward language users" (Fishman 1972:45).
The second concern, viz. socially realistic linguistics, focuses on the
interplay of linguistic structure and social factors or variables; it addresses itself
to questions of the extent to which linguistic structure reflects and can be
explained by societal factors. This concern is termed "descriptive
sociolinguistics" by Fishman (1969) and is acknowledged by him and others (cf.
Labov 1972, Trudgill 1983) to be the main focus of sociolinguistics. The third
and final topic, "socially (re-)constituted linguistics," according to Hymes
(1974:196-97), "is concerned with social as well as referential meaning, and
with language as part of communicative conduct and social action." This area
has evolved in recent times into an independent subfield of linguistics known as
ethnography of communication as exemplified in recent publications by
Gumperz and Hymes (1972), Gumperz (1982), and Saville-Troike (1982).
Hymes' (1974) view of sociolinguistics is echoed by Trudgill (1983:1) who
states that "It has become apparent that [sociolinguistics] is a term which means
many different things to many people." In particular, the term "appears to have
different implications in Britain and North America than those it has in Europe."
Trudgill attributes this multiplicity of interpretations and the confusion they
engender to "the fact that different scholars draw the line" between language
and society, on the one hand, and sociolinguistics, on the other, "in different
places." He observes that:
while everybody would agree that sociolinguistics has something \o
do with language in society, it is equally clearly not concerned with
everything that could be considered under the heading of 'language
and society'. (Trudgill 1983:1)
According to Trudgill (1983:2-4) this delimitation of the field is motivated by
the various objectives envisaged by scholars. These objectives fall into three
groups: (1) "purely linguistic " objectives; (2) partly linguistic and partly social;
and (3) sociological. The first set of objectives is exemplified in the work of
scholars such as Labov who seeks to secularize linguistics; the second by the
bulk of scholars who engage in the descriptive study of language in its social
context with particular reference to phenomena such as language variation,
code-switching, pidgins and Creoles, language planning, and other aspects of
bi- and multi-lingualism (cf. e.g., B. Kachru 1982, 1983, 1986; Ferguson 1968,
Bokamba: African languages and sociolinguistic theories 5
Trudgill 1983a&b. Mufwene 1979, 1988, 1989; Romaine 1982, 1986, 1988).
The third category of objectives is addressed by individuals concerned with the
sociology and/or functions of language in society. The v^^ork of Fishman and
Hymes, among others, appear to be characteristic of this group.
In contrast to these two views (viz. Hymes and Trudgill), Hudson (1980)
considers sociolinguistics as that field of linguistics which deals with all aspects
of language excluded by autonomous syntax, phonology, and historical
linguistics. This perspective is shared by Wardhaugh (1988) and Fasold (1984,
1990), among others, who include in their respective studies of sociolinguistics
topics such as dialectology, language variation and change, multilingualism,
pidgin and Creole languages, pragmatics, ethnography of communication,
language and sex, code-switching, and language planning.
In spite of the recognition of this multiplicity of subfields which are
generally assumed to constitute the domain of sociolinguistic research, many
sociolinguists continue to view the field as one that is amenable to a simple
definition and therefore capable of offering a unified theoretical model. Fasold's
statement in this regard is illustrative of both the dilemma faced and frustration
experienced by many theoretically-inclined sociolinguists. After arguing for the
need to treat different topics of sociolinguistics separately, Fasold (1990:viii)
points out and laments that,
Although it might make some sense to equate 'inguistic proper' with
'sociolinguistics, '2 no unified theory of sociolinguistics will be found
here. Instead, I present sociolinguistics as a series of topics with
some connections between them, as was done in the companion
book [The Sociolinguistics of Society]. The reason for this is that I am
not able to detect an overall theory, even of the portion of
sociolinguistics that is addressed here.
That there is no unified theory of sociolinguistics is not at all surprising in
view of the facts pointed out above: the field encompasses many topics and
subfields which often entail different research methods and theories, as Trudgill
(1983) correctly observes. It is therefore unreasonable to expect such a macro-
field to evolve a unified theory. Further, the conclusion that there is no unified
theory in sociolinguistics cannot be construed to imply that there are no theories
for the constituent subfields; on the contrary, theories abound in most of these
subfields, as will be seen later for the ones selected for this study. Thus, instead
of a single theory, I will assume in this paper that there are several theories of
sociolinguistics.
The second major difficulty in providing an overview of African
sociolinguistics is a derivative of the first: the vastness of the field makes it
impossible to adequately cover all the subfields. The best that one can do is to
highlight a few salient research developments that illustrate the contribution
made by African linguistics. The third and final difficulty is that mainstream or
descriptive sociolinguistics (viz., language variation, including code-
switching/code-mixing and creolistics) has generally assumed knowledge of
6 Studies in the Linguistic Sciences 20:1 (Spring 1990)
syntax and phonology, and thereby the various theories that have developed in
these fields. In other words, descriptive sociolinguistics is to syntax and
phonology as physics is to applied physics. To adequately review
developments in the applied aspect of the field one must be not only conversant
with the basic theories, but also the ways in which they are applied in the
derived or associated fields and the methodologies utilized in conducting the
research in question. Thus in making the assessments of the research in the
subfields under consideration here I have had to remain attentive to the
theoretical frameworks under which such research was written.
In summary, I view sociolinguistics as a rigorous and macro-field of inquiry
within linguistics. The kinds of phenomena that it addresses and attempts to
explicate can be seen as representing a continuum that ranges from purely
linguistics at the top end to sociological concerns at the bottom end. As one
moves from one end of the continuum to the other, the extent to which linguistic
theories apply decline accordingly. Hence research resulting from the bottom
end of this continuum cannot be assessed in purely linguistic evaluation
measures but in terms of a combination of measurements. For instance, the
study of language policy and planning cannot, in my view, be characterized as
Involving theoretical constructs as for example in phonology, semantics, syntax,
and historical linguistics, but instead in terms of "approaches"; whereas the
study of different aspects of language variation can be measured in terms of
theoretical frameworks.
2. Sociolinguistic theories and African languages
Judging from Clement's (1989:23-28) survey of doctoral theses submitted
in American universities from 1933 to 1987, the sociolinguistic study of African
languages represents the third most important area of research in African
linguistics after syntax (which has 80 theses, including grammar), 3 phonology
(44), and sociolinguistics (41, including dialectology, discourse analysis, and
pragmatics). If one were to conduct a cursory survey of the non-doctoral
dissertation literature (viz., M.A. theses and article-length studies) throughout
African, European, and North American institutions and journals, however, it is
very likely that sociolinguistics would emerge as the most studied aspect of
African linguistics in terms of volume of studies.
African sociolinguistics has focused mainly on five major areas: (1)
language spread, (2) language policy and language planning, (3) language
variation, (4) code-switching and code-mixing, and (5) pidgin and Creole
languages. Very little has been done on bilingualism, discourse analysis,
ethnography of communication, and language shift and maintenance for which
African languages offer excellent and extremely rich data. As indicated in the
first footnote, this study excludes a review of the research on topic five (5);
individuals interested in this area are referred to the excellent work of Hancock
(1971, 1979, 1986); Mufwene (1988, 1989); and Samarin (1960, 1962, 1984,
1985, 1986, 1989). The discussion in this section addresses itself to the first
four topics by summarizing the results of these studies and offering a
Bokamba: African languages and sociolinguistic theories 7
preliminary assessment of their contribution to the general sociolinguistic
research. Consider, first, the case of language spread.
2.1 Language spread
Sociolinguistics research on language spread in general addresses itself
to the question of the expansion of certain languages over geographical and
population areas by examining the psychological, social, and linguistic
phenomena that account for such an expansion. In an interesting article that
introduces the first collection of in-depth study of language diffusion from a
cross-linguistic perspective. Cooper (1982:6) defines language spread as
an increase, over time, in the proportion of a communicative network
that adopts a given language or language variety for a given
communicative function.
Cooper (1982:5-36) suggests, and the other contributors in the volume
concur, that the study of language spread can best be approached by
examining three broad questions: "(1) what is spreading, (2) the notion of
spread as a time-dependent phenomenon, and (3) the medium through which
language spread occurs" (p. 6). More specifically, research on language spread
seeks to provide answers to a set of fundamental questions that include the
following:
[4] What are the psychological, social, and linguistic phenomena
which, in interaction, account for language spread? [5] Do languages
spread in the same way as a single item of vocabulary,
pronunciation, or grammar? [6] Do languages spread according to
the same laws as innovations more generally? [7] What variables
entering into what equations can predict the rate and extent of
spread of a given language among a given group of speakers? (cf.
Cooper 1982:5)
In answer to the first four questions (i.e., (1) - (4)), it has been determined
in the general literature that language spread involves the diffusion of two or
more varieties of a given language over specific communities of individuals who
are willing and capable of learning that language as an additional medium of
communication for perceived socioeconomic benefits (cf. Brosnahan 1963,
Tabouret-Keller 1968, Fishman, Cooper & Conrad 1977, Kachru 1978, 1983,
Kahane & Kahane 1979, Scotton 1982, and Cooper 1982). The acquisition of
the language concerned as a second, third or nth language occurs gradually
over a period of time, and is said to be facilitated by five major agents: (1)
military conquest, (2) colonization, (3) religion, (4) trade, and (5) education.
Writing systems, music, and intermarriage have also been mentioned
sporadically as two of the minor agents of language spread (cf. Cooper, 1982b,
Bokamba 1984c).
It has been further determined in regard to questions (5) and (6) that
language spread, as defined above, differs significantly from the diffusion of
8 Studies in the Linguistic Sciences 20: 1 (Spring 1 990)
innovations in pronunciation, vocabulary, and grammar; the two processes are
therefore not analogous although some of the social variables are arguably
similar (Cooper 1982b). Language spread typically occurs across speech
communities via specifically targetted population groups through the five agents
of spread mentioned above; whereas the spread of innovations takes place
through individual speakers under naturalistic, rather than forced or planned,
language contact situations. Question (7), however, has not been directly
addressed in the general literature, but Section's (1971, 1972) studies, as will
be seen shortly, attempt to deal with it on the basis of African languages.
Sociolinguistic research on language spread In Africa has focused mainly
on the expansion of lingua francas either in certain regions (e.g., northern,
western, central, and eastern Africa) or countries (Central African Republic,
Ethiopia, Kenya, Sudan, Tanzania, Uganda, Zaire, Zambia). Except for a few
studies (e.g., Hulstaert 1946; Samarin 1962, 1984/1985, 1986, 1989;
Greenberg 1965; Tabouret-Keller 1968; Whiteley 1969; Heine 1970; Scotton
1971, 1972; Cooper 1978; Calvet 1982; Fabian 1986), the bulk of this research
has commonly been undertaken in the context of the general question of
language planning vis-a-vis education. This orientation, as will be seen later,
accounts in part for the lack of discussion of the structural properties of the
spreading language.
As in the general sociolinguistic literature, research on language spread in
the African context has sought to address the same questions identified in
paragraph one above. Over-all, the findings of these studies confirm those
encountered in non-African languages. For instance, Hulstaert (1946) and
Samarin (1962, 1984/85, 1986, 1989) in their studies of the spread of lingua
francas such as Lingala, Sango, Kikongo/Kituba, Swahili, etc., in Central Africa
(viz. Zaire, Congo, and Central African Republic) found the same agents of
spread to be operative. The same conclusion is reached by several other
scholars, including Greenberg (1965) in his study of the effect of urbanism and
migration on language diffusion in West Africa, Tabouret-Keller (1968) in his
comparative analysis of the sociological factors that contribute to language
maintenance and shift in Africa and Europe, Whiteley (1969) and Mazrui &
Zirimi (1978) in their study of the spread of Swahili in Eastern Africa. While most
of the other studies seem to suggest that military conquest, colonization, and
commerce played an equal role in the spread of African lingua francas, Cooper
(1978) shows that the primary impetus for the spread of Amharic in Ethiopia was
due initially to military and political domination of the non-Amharas by Emperor
Haile Selasie and his predecessor. Commerce and education subsequently
strengthened further the process and consolidated the gains made during the
initial phases. In the case of the spread of Mandingo in West Africa, Calvet
(1982) demonstrates that military conquest and commerce played a primary
role in the diffusion process.
In the most comprehensive study on this topic in the African context, Heine
(1970) documents forty different African lingua francas throughout the continent,
groups them by region, and discusses their spread in the respective regions
(pp. 46-125). Heine addresses here essentially the same major questions that
I
Bokamba: Al'rican languages and sociolinguistic theories 9
Cooper (1982) and his colleagues were still concerned with about fourteen
years later. In regard to the sociology of the spread of a lingua franca, he
confirms Brosnahan's (1963a) and Greenberg's (1965) findings regarding the
prototypical agents of spread of languages of wider communication and lingua
francas and the socioeconomic conditions under which the phenomenon
arises. Heine (1970:36) points out that the extent to which each of these factors
were influential in each of the forty lingua francas varied according to historical
and geographical factors. For instance, citing Westermann (1940), he states that
religion (i.e., Islam in this case) was very influential in the expansion of Arabic,
Hausa, Mandingo, and Swahili. This conclusion has been corroborated by
other Africanist scholars with respect to Arabic (cf. Greenberg 1965), Mandingo
(Calvet 1982), Swahili (Whiteley 1969, Mazrui & Zirimu 1978). A further
conclusion that Heine draws regarding the medium through which a lingua
franca , and by extension any language of wider communication, spreads in
Africa is that men are more involved than women in serving as agents of the
diffusion. He attributes this difference to the greater mobility that men have in
accessing education, employment as well as in undertaking long distance
trading (Heine 1970:29-30).
In addressing the structural and socio-psychological characteristics of
language spread, Heine (1970) suggests with respect to the former that most
lingua francas, like other languages of wider communication, develop at least
two forms: a "basis form" and a "lingua franca form" (p. 25ff). The basis form is
that variety spoken by the native speakers, whereas the lingua franca form is
that which is used by those who speak the lingua franca in question as a
second or additional language. The lingua franca form, according to Heine
(1970), characteristically undergoes and exhibits "simplification" via
"pidginization" of the grammar due to the influence of the speakers of other
languages. Throughout the study Heine (1970) shows that the socio-
psychological determinants of lingua franca spread are similar: Diffusion and
infusion occur much more rapidly and spontaneously in the direction of related
languages and among speakers of related languages who perceive benefits
(social, economic, educational) for learning an additional language than in
cases where these conditions do not obtain (pp. 34-35 ff). The spread of a
lingua franca is often impeded or inhibited altogether if the languages spoken in
the region or community concerned are unrelated to it and/or there is a
competing lingua franca (see also Greenberg 1965).
If, as it seems to be generally agreed, infusion precedes diffusion, then the
choice of which lingua franca to learn among competing ones must be guided
at least by some discernable sociolinguistic criteria or variables. An
identification of these criterial variables would permit us to answer question (7)
in the statement cited above from Cooper (1982): "What variables entering into
what equations predict the rate and extent of spread of a given language
among a given group of speakers?"
Scotton (1971, 1972) addresses this question from the perspective of code
selection in a multilingual setting where three lingua francas, viz., English,
Luganda, and Swahili, compete as languages of wider communication in an
1 Studies in the Linguistic Sciences 20: 1 (Spring 1990)
African capital city: Kampala (Uganda, East Africa). In carrying out this study
among a stratified and random sample of two hundred and twenty-three (223)
heads of households, Scotton had established that 97% of the respondents
claimed to speak Swahili to a certain extent versus 76% for English and 75% for
Luganda (pp. 113-16). The remainder of the interviewees spoke no Swahili
(viz., 3%), English (24%), and Luganda (25%). What this meant, Scotton
observed, is that 24-25% of the time that a conversation was initiated in English
and Luganda it met with "incomprehension", whereas this was not the case with
Swahili which appeared to be the ideal lingua franca that "presume[d] nothing"
about the interlocutor's socioeconomic background (p. 116). Initiating a
conversation in English or Luganda had socioeconomic connotations.
To investigate code-selection in this sociollnguistic context, Scotton
(1971:122) made two major conceptual assumptions:
(1) that supra-individual norms arise in the group situation and
influence the individual to act in ways he might not choose for
himself, and (2) that social interaction is a balancing of rewards and
costs by each individual so that he achieves the highest rewards
possible.
She found that speakers' choice of a particular code/language in a face-
to-face interaction was dictated by a strong sense of accommodation and
perceived rewards and costs in terms of group membership, irrespective of the
recognized "prestige" of the other available code(s) to the speaker-hearer. She
showed that speakers were rather consistent in their choice of language in
speaking with lower status interlocutors, peers, strangers, and people of higher
statuses (pp. 117-21).
Scotton's (1971, 1972) findings provide empirical support from an African
multilingual setting for the long-held hypothesis initially advanced by Firth
(1935) that code-selection in social interactions is determined by the context of
situation. A similar finding based on an American datum is presented in
Fishman (1969). Scotton's contribution lies in the fact that the respondents were
observed making code-selections in real-life situations, rather than simply
responding to a questionnaire about the potential, and therefore, hypothetical,
choice they might make. In this regard, Scotton appears to have been the first
scholar to have introduced the theory of accommodation that was subsequently
popularized in the work of scholars such as Giles (1973, 1979).
A second major contribution that Scotton made in this study concerns the
determinants and direction of language spread. She found, for example, that
75% of her respondents chose Swahili as a language that their children should
study; 98% of them considered it very useful over-all; 84% of them considered it
to be the language that the government should require of all its employees;
77% viewed it as the language that is capable of serving as a medium of
administration in Uganda; and 43% chose it as a potential official/national
language, comparing favorably in this regard with the 50% preference
expressed for English (p. 117). According to the study, these choices were
Bokamba: African languages and sociolinguistic theories 1 1
motivated by perceived socioeconomic benefits. If this is correct, then such
considerations would influence and predict the direction of language infusion
and spread as the author subsequently argued in part in a recent study (Scotton
1982). '^
2.2 Language planning
Language spread as a sociolinguistic area of inquiry represents one of the
phenomena that bears crucially on the issue of language planning (LP) to
which we now turn. As currently understood, language planning is
characterized as
an explicit and systematic effort to resolve [perceived] language
problems and achieve related goals through institutionally organized
intervention in the use and usage of languages [or language
varieties] (Christian 1988:197).
As this definition suggests, it has been assumed since the early study of LP that
the formulation of language policy (LPo), which is the outcome of planning,
proceeds in a systematic and step-wise manner from the perception of a
language structure and/or language function problem. Once such a problem is
perceived at the local, regional or national level, an organized effort is made to
address it through a constituted body at the societal level concerned. Now,
while in general LP is often initiated and carried out by and through
authoritative institutions, not all cases LP or aspects thereof, as Cooper (1989)
convincingly argues, are undertaken by institutions: individuals can initiate LP
or aspects of it. Noted historical cases where this has occurred include Ben
Yehuda's efforts in the revival of Hebrew in Palestine (now modern Israel) as
amply documented in Fellman (1974) and Cooper (1989), Samuel Johnson's
(of England) dictionary work (Bates 1975), Aasen's work on Norwegian
(Haugen 1966), and Webster's work on the dictionary of the American English.
These facts, as Cooper (1989:29-45) correctly maintains, argue for a less
restrictive and yet expanded definition of LP such as the following:
Language planning refers to deliberate efforts to influence the
behavior of others with respect to the acquisition, structure, or
functional allocation of their language codes (Cooper 1989:45).
Both Christian's (1988) and Cooper's (1989) definitions have evolved from
a body of key studies some of which require summation here in order to better
contextualize the developments in this area of research.
The classical conception proposed in Haugen (1966a-b, 1969) on the
basis of Norwegian identified four major tasks in the process of LP: (1)
code/language selection, (2) codification, (3) implementation, and (4) code
elaboration. This model concerned itself primarily with language form, rather
than function. Code or language selection involved the choice of a particular
language or variety for the purpose of communication at the national, regional
or local level; and codification dealt with the standardization procedures (e.g.,
12
Studies in the Linguistic Sciences 20:1 (Spring 1990)
graphicization, spelling, and policy regarding the selection of vocabulary,
including loanwords). Implementation referred to the act of executing the
adopted LPo, whereas code elaboration involved the modernization of the
language in terms of the expansion of the registers and styles to accommodate
the communicative needs to which the language is put. Tasks (1) and (3) came
to be known subsequently as status planning (SP), and tasks (2) and (4) as
corpus planning (CP) (cf. Kloss 1969) where the former set involves largely
decision-making by organized authoritative bodies and the latter set addresses
linguistic issues.
A number of studies undertaken in the 1970s (e.g., Rubin 1971; Das Gupta
1973; Fishman 1974, 1978; Karam 1974; Neustupny 1974; etc.) led to important
modifications in the classical conceptualization of LP by emphasizing equally
the significance of SP and CP. The currently accepted view of LP that
incorporates Haugen (1983) and subsequent research (e.g., Kennedy 1982;
1984; Neustupny 1983; Rubin 1984) is essentially as follows:
Form
(policy planning)
Status planning: (1)Code selection
(decision-making
procedures):
i
a. Identification of
problems
b. Allocation of norms
i
Corpus planning: (2) Codification
(standardization
procedures):
i
a. Graphicization
b. Grammatication
0. Lexication
â– * FEEDBACK/EVALUATION
Function
(language cultivation)
4'
(3) Implementation
(educational spread):
a. Correction procedures
b. Feedback/evaluation
(4) Code elaboration
(functional development):
a. Terminological
modernization
b. Stylistic development
J'
Figure 1: Schematized view of Language Planning
A few comments are necessary here to flesh out what this elaborated view
of LP assumes and represents. Specifically, if we regard Figure 1 as a four-cell
matrix, four conceptual facts are observable here. First, language policy is
understood as the outcome of language planning that affects language form or
structure through the decisions of an individual, group of individuals, or some
authoritative body at the micro and/or macro level. Second, status planning,
which is seen in Figure 1 as encompassing code selection and implementation,
is regarded as an aspect of decision-making that is driven by underlying socio-
cultural and economic goals and is acted upon largely by the society concerned
through individuals or groups thereof at the micro level and through organized
institutions at the macro level. The decisions made at this phase are
Bokamba: African languages and sociolinguistic theories 1 3
characteristically societal or political, and therefore language external. As
indicated earlier, these decisions are assumed to be initiated by the perception
of language-related problems and are informed by fact-finding and planning
(Rubin 1971). The fact-finding subphase involves the gathering of demographic
and attitudinal data about the language situation; whereas the planning
subphase concerns the formulation of goals, means to achieve them, and the
articulation of expected outcomes. Cost-benefit analysis is seen as a crucial
aspect of the planning sub-phase. Third, corpus planning, which comprises
codification and code elaboration, is viewed as the phase that addresses
language-internal issues and thus falls within the purview of linguists. Fourth,
language cultivation is motivated by functional objectives and demands on the
language(s) under consideration. Finally, each phase of LP requires both
decision-making and evaluation of some sort, with an ultimate assessment of
the success of the LPo being undertaken after it has been implemented for a
determined period of time (e.g., ten, fifteen years). Thus feedback/evaluation
loops around the four cells in a U-shape fashion. It is this kind of step-wise
progression and interrelationship of the phases that characterizes language
planning in theory as a "systematic effort" undertaken "to resolve [perceived]
language problems and achieve related goals" (Christian, op.cit.)
The study of LP in the African context constitutes one of the most important
aspects of sociolinguistic research on African languages. There is a significant
body of descriptive literature which offers partial support for as well as
contradicts the LP model presented in Figure 1. Several conclusions emerge
from this research which began to appear in the early 1960s following the
accession to independence of many of the countries in the continent.
The first conclusion is that LP in Africa has been less systematic than
Christian's (1988) definition and the model presented in Figure 1 above
suggest. This conclusion is supported by the fact that most of the language
policies in the continent have been imposed, rather than developed
systematically as discussed above, by the various governments that have come
to power and the private institutions (e.g., churches and commercial
enterprises) that have supported them since the European occupation of the
continent in the 17th and 18th centuries (cf. Brosnahan 1963; Spencer 1963,
1971; Alexandre 1963, 1968; Ansre 1968; Foster 1971; Whiteley 1971; Gorman
1974; Bokamba & Tlou 1977; Scotton 1978; Cooper 1976, 1989; Turcotte
1981a-b; Marshad 1984). For example, the LPo of Ethiopia which involves the
use of Amharic as the official/national language, and which is often cited as an
example of success in the adoption of an indigenous language, was imposed
by the ruling Amhara Emperors Minilik II and Haile Silasie (Cooper 1978, 1989;
Bender 1986; Seyoum 1989). Similarly, the French and Portuguese colonial
administrations imposed French and Portuguese, respectively, to their
respective colonies as the exclusive languages of administration and
education. In the case of the French a 16th century decree, known as the
"ordinance" of Villers-Cotteret issued in 1539 by King Francois I, forbidding the
use of languages other than French in all official functions of the state within
French territories was extended to the African colonies without debate in 1826
(Spencer 1971; Turcotte 1981b; Bokamba 1984). The decree was further
1 4 Studies in the Linguistic Sciences 20: 1 (Spring 1990)
strengthened by subsequent colonial administration executives orders and
decrees throughout the colonial era (Turcotte 1981b; Bokamba 1984).
The LP literature in the African context reveals further that while the
Belgian, British, and German colonial administrations had adopted a laissez-
faire policy vis-a-vis the education of the colonized by relegating this
responsibility to religious organizations, they did nonetheless impose from the
outset their own languages as the media of administration while allowing the
missionaries to develop and implement language policies which advocated the
use of selected African linguae francae in primary education (Ansre 1968;
Spencer 1971; Whiteley 1971; Apronti 1974; Gorman 1974; Awoniyi 1976;
Bokamba & Tlou 1977; Ndoma 1977; Yates 1981). When these colonial
administrations eventually became involved in education, initially through the
provision of subsidies towards the end of the 19th century and then more
significant involvement in the first quarter of the 20th century (1920-24), they
over-rode the religious establishments-based language policies by extending
the imposition of the official languages into this sector (Awoniyi 1976; Ndoma
1977; Yates 1981).
The second major conclusion that emerges from the African LP literature is
that religious organizations developed and implemented language policies vis-
a-vis education and religious work, often with the acquiescence of the colonial
administrators and without significant input from the Africans (Ndoma 1977;
Yates 1981; Awoniyi 1976; Scotton 1978; Stumpf 1979; Marshad 1984). LP and
LPo formulation by missionaries were mainly driven by one major objective and
related subsidiaries considerations: evangelization and provision of pre-
university education to maintain and attract converts. Fact-finding in regard to
demographic and attitudinal considerations was rarely undertaken to inform
LPo; missionaries did, however, carry out investigations on CP and had
considerable success in all aspects of this dimension of LP. The general African
linguistic and the sociolinguistic literature consistently show that religious
organizations were the first ones to undertake the study of African languages,
produce monographs, dictionaries, grammars, and other reference tools (cf.
Birnie & Ansre 1969; Cole 1971; Bendor-Samuel & Hartell 1989).
The third major conclusion which arises from the LP literature in Africa
since the 1970s is that post-colonial African states, with a few exceptions, have
continued the inherited colonial era language policies in spite of perceived and
demonstrated problems with such policies (Champion 1974; Apronti 1974;
Ansre 1976; Bokamba 1976, 1984a-b; Bokamba & Tlou 1977; Mateene 1980).
The exceptions to this fact are Rwanda, Tanzania, Somalia, Malagasy,
Botswana, and Lesotho which have each adopted Kinyarwanda, KiSwahili,
Somali, Malagasy, SeTswana, and SeSotho, respectively, as national/official
languages. Ethiopia, which was never colonized, but freed itself from the Italian
occupation in W.W.I, has adopted Amharic as the national/official language. A
few other countries, including the Arabic-speaking North African states, Nigeria,
Togo, Kenya, and South Africa have adopted multilingual policies that
recognize European and African languages as co-official languages in specific
domains. A number of studies have argued persuasively that the retention of the
Bokamba: African languages and sociolinguistic theories 1 5
Status quo on the inherited colonial languages policies demonstrates both the
lack of political will on the part of African political elites (Ansre 1976; Kashoki
1978; Bokamba 1976, 1981, 1984a) and the existence of an "elite closure"
mentality (Scotton 1978).
Fourth and finally, unlike in most other regions, LP in Africa has been
hampered further by pervasive multilingualism, combined with unstable political
structures. Bendor-Samuel (1989: vi) states that there are 1900 languages
spoken in Africa, and several other previous studies have documented the
extent of multilingualism of most of the continent's nations (cf. e.g., Ladefoged &
Cripper 1972; Whiteley 1974; Bender, et al., 1976; Hansford, et al., 1976;
Ohannessian & Kashoki 1978; Polome & Hill 1980; Confermen 1986).
Pervasive multilingualism within each nation has been used by most African
political leaders as an excuse and a rationale to avoid the initiation of
indigenously-based language policies, even in cases where certain languages
have served as dominant linguae francae for decades (cf. Spencer 1971;
Apronti 1974; Ansre 1976; Bokamba & Tlou 1977; Scotton 1978; Dumont 1983).
As a consequence, most African states have retained the status quo whereby
English, French, and Portuguese continue to serve as the exclusive official
languages and media of instruction in education.
Four main arguments, which Ansre (1976) characterizes as
"rationalizations", have been advanced by various African political leaders to
support their retention of the pre-independence language policies: (1) national
unity, (2) national development/progress, (3) efficiency of European languages
of wider communication (ELsWC), and (4) cost-effectiveness of ELsWC
(Spencer 1971; Foster 1971; Whiteley 1971; Ansre 1976; Bokamba & Tlou
1977; etc.). The national unity argument holds that the adoption of a language
policy which involves the use of an African language as a national/official
language in a multilingual nation will lead to ethnic conflicts which could
destroy the delicate national unity created since the liberation struggles. To
prevent such social unrests, it is argued, African states must choose ELsWC,
because they are perceived as neutral or "atribal". The national development
argument maintains that progress, understood as "westernization," for the
African people will be possible and unimpeded only if ELsWC are used as
media of instruction (cf. Ansre & Birnie 1968; Spencer 1971; Gorman 1974);
African languages are perceived by such political leaders as obstacles to
"progress". Similarly, African languages are viewed as less-developed (in the
sense of Ferguson 1968) than ELsWC and are, therefore, unsuitable as media
of instruction. In contrast, it is argued that ELsWC have had a long tradition of
writing and research that make them efficient for teaching of all sorts that is
difficult to carry out in African languages. A closely related argument maintains
that the adoption of indigenously-based language policies would be extremely
costly as they will necessitate the translation and writing of textbooks, reference
tools, and the training of teaching personnel. The adoption of ELsWC-based
policies, in contrast, would be cost-effective as the requisite pedagogical
materials and personnel are available and can be imported from elsewhere,
especially Western Europe.
1 6 Studies in the Linguistic Sciences 20: 1 (Spring 1990)
I have shown elsewhere (Bokamba 1981, 1984a,b) that while these
arguments have some validity, the conclusions that African languages are
unqualified at this stage to serve as national/ official languages are
unwarranted on empirical grounds. At least two sets of facts can be cited in
support of this claim. The first set of data include the successful utilization of
selected national or state languages in India and the USSR; the revival of
Hebrew by Yehuda in 1898 as a medium of instruction; the use of Japanese
and Korean by Japan and Korea, respectively, as languages of education, and
the recent adoption and utilization of languages such as Amharic (Ethiopia) and
Swahili (Tanzania). These facts demonstrate that any language can be
cultivated to serve as a vehicle of education without impeding developments of
all sorts (e.g., educational, economic, and political, etc.). The second set of data
consist of the well-documented high failure, class repetition, and drop-out rates
in the educational systems (from elementary to university) in African states, and
the extremely high illiteracy rates resulting in part from such systems (cf. Foster
1971; Bhola 1990; Bokamba 1984a; UNESCO 1981, 1988).
It has been argued rather convincingly in several studies that the
inefficiency of African states educational systems that is characterized by high
failure and attrition rates is largely ascribable to poor mastery of the official
language by both teachers and students (Champion 1974; Awoniyi 1976;
Barnes 1981; Bokamba 1981, 1984a). Educational inefficiencies, combined
with structural disarticulations, have in turn made African education extremely
less cost-effective: Governments expand disproportionately high percentages of
their national budgets on education only to have incommensurably small
numbers of students complete their education. The arguments about the level of
development and efficiency of ELsWC become, therefore, vacuous in the face of
such facts, and the envisaged benefits to be accrued to the nations concerned
are mere mirages. A careful reading of the research on LP/LPo in the African
context shows that the vast majority of language policies in the continent are
more driven by historically-based external political and economic
considerations, including direct pressure from former colonial powers, than by
internal factors as Fishman's (1968) typology of LPo and Christian's (1989)
definition would lead us to believe. African states characteristic political
instability due to lack of general mass support further makes it difficult for
political leaders to develop the political will to initiate and implement the kinds
of language policies that would be consonant with the aspirations of the African
people. Thus in Africa the perception of the language-related problems to which
LP is assumed to offer solutions is colored, if not highly constrained, by external
politico-economic considerations.
2.3 Language variation
Consider now the phenomenon of language variation. The research
paradigm that prevailed in the 1960s and 1970s and Is still influential today is
predicated on the major assumption that each language consists of two phases:
an invariant or homogeneous phase and a variant or heterogeneous phase.
Variation under this perspective, therefore , is explained as a deviation from the
perceived or hypothesized norm arising from a combination of language
Bokamba: Aincan languages and sociolinguistic theories 1 7
internal mechanisms and the context of situation (interlocutors, topics,
context/settings) (Labov 1966, 1972; King 1969; Giles 1979; etc.)- This
deviationist view is based on the historical linguistic approach as modified by
dialectology studies.
Under this deviationist framework, the language internal mechanisms that
contribute to and/or cause variation are structural rules from all levels of the
grammar: phonology, morphology, syntax and semantics. For example, the non-
occurrence of certain types of consonant clusters in Black English (in America)
and West African English, as in the illustrations below, is viewed as a case of
consonant cluster simplification:
(1) a. next -> nest
b. needs -> nees
Similarly, the occurrence of the so-called equational sentences in Black
American English and other inner city Englishes, as in examples (2)-(5) below,
is analyzed as an instance of copula deletion in various structural contexts (cf.
Labov 1972):
(2) NP:
a. She the first one started us off.
b. Means he a faggot or sump'm like that.
(3) Predicate Adjective:
a. He fast in everything he do.
b. I know, but he wild, though.
(4) Locative:
a. You out the game.
b. We on tape.
(5) Negation;
a. But everybody not black.
b. They not caught .
At the level of the lexicon, however, various semantic processes such as
semantic extension, semantic shift, and semantic change are regarded as
language internally and externally motivated variation. It is argued in this
connection that semantic variation in any language cannot be adequately
accounted for simply through an examination of language-internal mechanisms:
one must take into consideration the sociolinguistic context of the language
acquisition and use (Kachru 1982, 1983, 1986; Bokamba 1982, 1991; and
Kapanga 1991). This argument is amply supported in the literature, including
the works of scholars such as Labov (1972), Kachru (1982, 1983, 1986), and
Trudgill (1983), where it is shown that acquisitional, socio-cultural, motivational
and functional factors combine with other parameters such as interlocutors,
topics and context/setting to structure the form that a given language takes.
1 8 Studies in the Linguistic Sciences 20: 1 (Spiing 1990)
Linguistic rules, while important and crucial, cannot alone explicate the
phenomenon of language variation and the eventual change it entails.
The framework summarized here has been utilized successfully in
accounting for language variation in a variety of languages, including notably
English and to a certain extent French, in their global contexts. Much of the
research on variation in the African context, however, has focused mainly on
English as exemplified by anthologies edited by scholars such as Spencer
(1963), B. Kachru (1982), Lanham (1985), Cheshire (1991), McArthur (1992),
and doctoral dissertations by Sey (1973), Chishimba (1984), Magura (1985),
among others.
The main change in orientation that has occurred to this deviationist
paradigm is the recognition of the fact that language variation constitutes an
inherent property of all languages: Homogeneity or invariation is in fact a
linguistic construct that has little basis in reality (cf. Guy 1986, 1989). Under this
view, language variation is increased or enhanced, but not initiated, by
sociolinguistic factors and the grammatical rules involved in such change
represent essentially an encoding of these otherwise non-linguistic factors
(Labov 1972, Kachru 1982, 1986; Bokamba 1977, 1982; Guy 1989; Kapanga
1991). Substratal influence, or the influence of mother tongues on a
second/foreign language, which lead to processes such as nativization of
lexical elements at the phonetic, semantic and morphological levels, serves as
the best illustration of this view.
Except for English and English-based pidgin languages (e.g. Krio,
Cameroonian Pidgin English and Nigerian Pidgin English), there have been
very few studies of language variation involving African languages. In depth
studies of variation in African languages include Derek Nursse and Thomas
Hinnebusch (forthcoming) on Swahili in Kenya; Jan Fabian (1986) on Zairean
Swahili; Timothy Wilt (1989) on Kenya Swahili; Mwamba Kapanga (1991) on
Shaba Swahili; and Janice Bernstein (1991) on Shona. Several article-length
studies have appeared on several African languages (cf. e.g., Gilman 1979;
Bokamba 1977, 1982; Stucky 1978; Mufwene 1979, 1988). This research is
presented largely within the deviationist paradigm summarized earlier while
taking cognizance of the inherent variability hypothesis. The authors provide
sociolinguistic support for the hypothesis that language variation is both
inherent and externally induced under conditions of language contact
(Bloomfield 1933; Weinreich 1953; Weinreich, Labov & Herzog 1968.)
Over-all, research on language variation in the African context has made two
major contributions to sociolinguistics theories with particular reference to the
phenomena of language variation and change: (1) the provision of cross-
linguistic data and descriptions thereof, and (2) the reaffirmation of the
mechanisms of language variation, especially in pervasively multilingual
societies. Perhaps the most important aspect of these contributions is the
demonstration of the role played by non-linguistic factors in the structuring of
language form.
Bokamba: African languages and sociolinguistic theories 1 9
2.4 Code-switching
Research on code-switching and code-mixing in the past has focused on
the characterization of these phenomena from the socio-psychological and
syntactic perspectives. Initially perceived as non-rule governed and
pathological language behaviors by bilinguals (cf. Weinreich 1953; Lambert
1967; Gumperz 1967), code-switching and code-mixing, as defined below,
have come to be regarded as natural and rule-governed behaviors that reflect
the multilingual speaker's creativity (cf. Kachru 1978; Lipski 1978; Pfaff 1979;
etc.).
Since the late 1970s research on bilingualism and multilingualism has
provided a number of insightful findings about the nature and type of speech
produced by speakers in bi- and multi-lingual communities. One of these
findings is the determination that code-switching and code-mixing are two
distinct aspects of the same general phenomenon of bi- and multi-lingual
creativity triggered by specific conversational factors. Thus in view of the
structural and psychological properties manifested and/or inferred from such
linguistic behavior, the following operational definitions of these phenomena
have been formulated and widely accepted (Sridhar & Sridhar 1980; Bokamba
1988, etc.):
{6)Code-switching is the mixing of words, phrases and sentences
from two distinct grammatical (sub-) systems across sentence
boundaries within the same speech event. In other words, CS is
intersentential switching.
{7)Code-mixing is the embedding of various linguistic units such as
affixes (bound morphemes), words (unbound morphemes),
phrases and clauses from two distinct grammatical (sub-)systems
within the same sentence and speech event. That is, CM is
intrasentential switching.
For the purpose of this study, the focus will be on code-mixing (viz. definition 7).
Of the two research approaches that have been pursued with regard to
code-mixing since the late 1970s, the most interesting and relevant from the
perspective of this study has been the so-called syntactic constraints paradigm.
The ultimate objective of this research paradigm which began in the mid-1970s
is to provide a characterization of what is 'a possible sentence' in code-mixed
speech (Gingras 1974; Pfaff 1979; Wentz 1977; Lipski 1978; Sridhar & Sridhar
1980; Sankoff & Poplack 1981; Poplack 1980, 1982; Woolford 1983; Ewing
1984; and Di Sciullo et al. 1986). Most of these studies, especially beginning
with Lipski (1978), have attempted to address themselves to two fundamental
questions: (1) how to describe and explain the morpho-syntactic characteristics
exhibited in CM data; and (2) what kinds of psycholinguistic inferences can be
drawn from the grammatical properties exemplified in code-mixed speech.
2 Studies in the Linguistic Sciences 20: 1 (Spring 1990)
Several hypotheses have been advanced regarding the first question.
First, it has been suggested that the production of well-formed code-mixed
sentences necessitates the integration of the grammatical rules of the language
pair involved in the speech event (Lipski 1978; Pfaff 1979; and Sridhar &
Sridhar 1980). This integration is generally assumed to occur at the syntactic
level, but may also extend to the morphological level (Gumperz 1982; Bokamba
1985, 1988, etc.), as illustrated in the Kin Lingala (KL) sentence (8), and the
Puerto Rican English (PRE) sentence (9):
(8) KL: Mo-bali na yo a-telephon-ak^ y6 deux fois par jour.
'Your husband calls you twice per day." (Luambo, 1985)
(9) PRE: But I wanted to fight her con los punos you know.
'I wanted to fight her with my fists, you know.' (Poplack, 1980)
It is claimed in regard to facts such as these that, in the production of both
sentences, the placement of the constituents from the guest language in each
case (viz. French and Spanish, respectively) obeys the constituent structure of
the host language (Lingala and English). In particular, the French verb root in
(8) occurs precisely where a typical Lingala verb would appear, and it shows
the characteristic subject-verb agreement and tense/aspect morphosyntactic
markings. Further, the phrase deux fois par jour occurs in the same position
where the equivalent Lingala phrase would have appeared. There is, however,
a minor syntactic violation here in the placement of the numeral adjective before
the modified noun fois , to which I shall return later. In sentence (9), the
placement of the PP (prepositional phrase) con los punos is similarly consistent
with the rules of English syntax.
The analysis of sentences such as (8) and (9), however, has not posed
any problems for researchers, especially in view of the fact that little attention
has been paid to derivational questions. What has concerned specialists in CM
the most are putatively ill-formed utterances such as the following:
(10) a. *l want a motorcycle verde. (Pfaff, 1979)
'I want a green motorcycle.'
b. *EI hombres car. 'The man's car' (Sankoff & Poplack, 1981).
c. She sees lo. 'She sees it.' (Timm, 1975).
d. *Yo went. 'I went' (Timm, 1975).
e. *EAT-iendo. 'Eating' (Poplack, 1980).
f. I told him that ram bahut bimar hai. (Di Sciullo et al., 1986
Ram very sick Aux
{*ki
that}
'I told him that Ram was very sick.'
g. mujhe lagta hai Ram will come tomorrow
{ki
*that }
me seem Aux
'I feel that Ram will come tomorrow.' (Di Sciullo et al., 1986)
Bokamba: African languages and sociolinguistic theories 2 1
Sentence (10a) is said to be unacceptable because of the occurrence of
the Spanish adjective verde in an adjectival NP controlled by English; (10b)
violates a similar restriction in that the English noun car occurs in a Spanish
genitival NP. The next two sentences (10c, d) are ill-formed because the
occurrence of an object and subject pronouns, respectively, in an otherwise
English sentence. (lOe) violates the restriction against the mixing of bound and
free morphemes from two different languages, whereas (lOf) and (lOg) violate
the complementizer constraint.
To account for the ungrammaticality and/or non-attestation of sentences
such as these, several language-specific and general surface morpho-syntactic
constraints have been proposed. They include the following constraints which
are commonly viewed as general or universal restrictions on CM: (1) the-size-
of-constituent constraints (Gumperz & Hernavez 1975; Poplack 1989), (2) the
conjunction and/or complementizer constraints (Gumperz 1977; Kachru 1978;
Singh 1981; Di Sciullo et al. 1986), (3) the adjectival phrase constraint (Pfaff,
1979), (4) the clitic pronoun constraint (Pfaff 1979), (EC) (Lipski 1978; Poplack
1980, 1982); and (7) the dual structure principle (DSP) (Sridhar & Sridhar
1980). 5 To my knowledge, all these constraints have been invalidated on the
basis of either language-specific and/or cross-linguistic data (Bokamba 1987;
Kamwangamalu 1989).
Research on CM in African languages by scholars such as Nartey (1982),
Scotton (1983b, 1987), Bokamba (1987, 1988), and Kamwangamalu (1989),
among others, has made significant contribution to the analysis and
understanding of the phenomenon of code-mixing by demonstrating not only its
morphological and syntactic complexity when the language-pairs comprise
African languages, but also the invalidity of the putative universal syntactic
constraints proposed on the basis of English-Spanish data (cf. Nartey 1982,
Scotton 1983b, Bokamba 1987, 1988; Kamwangamalu 1989). These studies,
especially Bokamba & Kamwangamalu (1988), and Kamwangamalu (1989),
have further raised very interesting questions concerning the processing of
code-mixed speech and its implications for syntactic theory. Bokamba (1988),
for example, proposes an innovative model in which the production/processing
of code-mixed speech across languages is analogized to the processing of
cross-dialectal speech within an individual's code repertoire. He argues in
essence that the main differences between the two cases lie in the manner in
which the brain treats the lexicon and morphosyntactic rules, but not in the
actual encoding and decoding of the messages. He suggests further that the
multilingual lexica internalized by the code-mixer is treated as a thesaurus that
can be accessed simultaneously, whereas the syntax of each language is
treated separately and is accessed either sequentially or simultaneously
depending on the type of speech (i.e., monolingual or mixed) is being
processed (Bokamba 1988).
2.5 Summary
To summarize this part of the study, I began by pointing out the difficulties
involved in the delimitation of the field of sociolinguistics. During the discussion
22 Studies in the Linguistic Sciences 20:1 (Spring 1990)
I have attempted to demonstrate that sociolinguistics is a macro-field consisting
of several major subfields which have different research objectives and utilize
different analytical frameworks. As a macro-field, it is impossible to make it
amenable to a single theory or approach. The development of these different
approaches to the study of language in its social context has been considerably
aided by research on African and non-African languages such as English and
French. Perhaps the most important contributions made by Africanist scholars
has been in language planning and code-mixing where established paradigms
have been seriously questioned and shown to be inadequate.
3. Research agenda for the 1990s
Having now reviewed critically the research on four major areas of
sociolinguistics, we are now led to raise the question of where to go from here.
What, in other words, should be the direction for future research and what
should the contribution of Africanist scholars be in this regard? These questions
entail first and foremost an examination of the significance of sociolinguistic
research to the entire enterprise of the linguistic sciences to which I now turn.
3.1 Significance of sociolinguistic research
It is commonly argued that the ultimate objective of linguistic research is
the construction of a general theory of language knowledge: What is it that a
person knows when (s)he knows a language. The attempt to discover and
characterize this internal knowledge takes different directions and dimensions
which have commonly been recognized as fields and subfields of linguistics:
phonetics, phonology, morphology, syntax, semantics, historical linguistics,
psycholinguistics, sociolinguistics, etc. In order to provide a descriptively
adequate theory of language knowledge, linguists must examine both linguistic
and related non-linguistic factors exemplified by the classical dichotomy of
competence/performance, because the interplay of these factors encompasses
what we know intuitively as language knowledge. Further, such a study from
this kind of perspective is bound to be more insightful than one that restricts
itself to one of these factors or dimensions, contrary to Chomsky (1965)
pronouncements.
In my considered opinion, there is no better subfield of linguistics to
undertake this kind of research than sociolinguistics. The reasons for this are
not difficult to find: sociolinguistics deals with the study of language structure,
function, and usage in social context. It, therefore, subsumes the study of
aspects of phonology, phonetics, morphology, syntax, and semantics. In this
regard it reflects better than any other subfield of linguistics, the totality of
language knowledge.
3.2 Role of African(ist) linguists
If the view offered here is correct, then African(ist) linguists have a major and
essential role to play in the advancement of linguistics. With its estimated 1900
languages (Bendor-Samuel & Hartell 1989), Africa constitutes both a gold mine
Bokamba: African languages and sociolinguistic theories 23
and an unparalleled laboratory for linguistic discoveries and research. India,
with its estimated 1600 plus languages is the closest competitor. What the facts
of multilingualism in Africa indicate in practical terms is that African(ist) linguists
are in a much better position to undertake sociolinguistic research of all sorts,
because they understand better than anyone else the issues of multilingualism
and should, therefore, be able to provide needed insight to this phenomenon
and others that it entails. Thus far African(ist) linguists have done relatively little
in the study of individual and societal multilingualism.
Research in sociolinguistic is a necessity for Africanists not simply on the
account of theoretical considerations, but also and perhaps more importantly,
because of practical considerations. African linguists must understand
sociolinguistics in order to deal adequately with the various issues raised by
multilingualism: language variation, pidginization/creolization, language policy
and planning vis-a-vis education, language and literacy, second/foreign
language teaching and acquisition, language and communication, language
and developments of all sorts, including cultural, political and economic. The
study of various aspects of multilingualism constitutes the agenda for the 1990s.
If we do not undertake it, who else will and with what understanding of the
issues involved? If we engage in such research, the practical benefits to be
accrued will be considerable, and the role that African linguistics will play in the
evolution of a sound and cross-linguistically informed theory of sociolinguistics
will be pivotal as has been the case in contemporary phonological theory (see
Goldsmith, this issue) and is increasingly becoming the case in syntactic theory
(see Bresnan, this issue).
NOTES
I am indebted to Braj B. Kachru and Salikoko S. Mufwene for comments
on an earlier version of this paper. I alone, however, am responsible for the
analysis presented here.
1 Unless otherwise specified, I will use the term code-switching in this
study in its generic sense to cover both intersentential and intrasentential
language switching which are defined in (6) and (7) below.
2 Fasold is alluding here to the British school of linguistics which views
sociolinguistics and general linguistics as two aspects of the same, rather than
separate, fields. This perspective is exemplified in the works of Firth (1957),
Halliday (1973, 1974), and Hudson (1980).
3 Clements' grouping of sociolinguistics theses include research on Pidgin
and Creole languages which this study has excluded because of space
limitations. Some scholars consider this area of study as a separate/distinct
(sub)field from sociolinguistics (e.g., Mufwene, in personal communication),
while others view it as an integral aspect of sociolinguistics (Hudson 1980;
2 4 Studies in the Linguistic Sciences 20: 1 (Spring 1990)
Fasold 1984, 1990; Wardhaugh 1986; B. Kachru (in personal communication),
and this writer).
^ An interesting finding/datum is that those who claimed knowledge of
English learned it at school, whereas "almost all [those] claiming ability in
Swahili reported learning it on their own" (p. 119). Hence the latter learning
mode is likely to contribute more to the spread of Swahili than the restrictive
classroom learning mode.
5 Of these syntactic constraints and principles on code-mixing, the last five
have been treated in the literature as if they were universal. They are stated as
follows:
{a)The Adjectival [Noun] Phrase Constraint Adjective/noun mixes
must match the surface word order of both the language of the
adjective and the language of the head [noun]. (Pfaff 1979:306).
(b)TheClitic Pronoun Constraint: Clitic pronoun objects are realized
in the same language as the verb to which they are cliticized, and
in the same position required by the syntactic rules of that
language (Pfaff 1979:303).
(c) Ttie Free Morplieme Constraint Codes may be switched after any
constituent in discourse provided that [that] constituent is not a
bound morpheme (Poplack 1980:585-86).
{d)The Equivalence Constraint Code-switches tend to occur at
points in discourse where the juxtaposition of LI and L2 elements
does not violate a syntactic rule of either language, i.e., at points
around which the surface structures of the two languages map on
to each other (Poplack 1980:586).
{e)The Dual Structure Principle: The internal structure of the guest
constituent need not conform to the constituent structure rules of
the host language, so long as its placement in the host sentence
obeys the rules of the host language (Sridhar & Sridhar
1980:412).
REFERENCES
Afolayan, a. 1978. Towards an adequate theory of bilingual education for
Africa. International Dimensions of Bilingual Education. Georgetown
University Round Table Conference on Languages and Linguistics (GURT
1978), ed. by J.E. Alatis, 330-390. Washington, D.C.: Georgetown
University Press.
Bokamba: African languages and sociolinguistic theories 2 5
ALEXANDRE, P. 1963. Les problemes linguistiques africaines vus de Paris.
Language in Africa, ed. J. Spencer, 53-59. London: Cambridge University
Press.
. 1967. Langues et langages en Afrique noire. Paris: Payot.
Ansre, G. 1969. The need for a specific and comprehensive policy on the
teaching of Ghanaian languages, ed. by J.R. Birnie & G. Ansre, 5-1 1 .
. 1976. Four rationalizations for maintaining the European languages in
education in Africa. Paper presented at the International Seminar on
African languages in education; Kinshasa, Zaire. Mimeographed.
Apronti, E. 1974. Sociolinguistics and the question of a national language; The
case of Ghana. Studies in African Linguistics, Supplement 5, 1-20.
AWONIYI, T. A. 1976. Mother tongue education in West Africa: A historical
background, ed. by A. Bamgbose, 27-42.
Bamgbose, a. (ed.) 1976. Mother Tongue education: The West African
experience. London: Hodder and Stoughton.
BARNES, V. A. 1982. A review of literature on efficiencies in primary education
for Sub-Saharan Africa. Paper presented at the A.I.D. seminar on
education training in Africa; Baltimore, Maryland, March 1982.
Bates M., & Dudley-Evans A. 1975. Notes on the introduction of English
courses for students of science and technology of the University of Tabriz,
Iran. ELT Documents (74/4), London.
BENDER, M. L. 1986. Ethiopian language policy 1974-81. Anthropological
Linguistics 27.273-79.
, J. D. Bowen, R. L. Cooper, & C. A. Ferguson (eds.) 1976. Language in
Ethiopia. London: Oxford University Press.
Bendor-Samuel, J., & R. L Hartell (eds.) 1989. The Niger-Congo languages: A
classification and description of Africa's largest language family. Lanham,
MD: University Press of America.
Bernstein, J. G. 1990. The integration of loans in Shona: Social correlates and
linguistic consequences. Michigan State University at East Lansing Ph.D.
dissertation in Linguistics.
Bhola, H. S. 1990. An overview of literacy in Sub-Sahara Africa — Images in
the making. African Studies Review 33:3. Special Issue for the
International Year of Literacy, ed. by H. S. Bhola.
Birnie, j. R., & G. Ansre (eds.) 1969. Proceedings of the Conference on the
Study of Ghanaian Languages. Legon: Ghana Publishing Corp.
Bloomfield, L 1933. Language. New York: Holt, Rinehart and Winston.
Bokamba, E. G. 1976. Authenticity and the choice of a national language: The
case of Zaire. Presence Africaine 99/100:104-43.
. 1977. The impact of multilingualism on language structures: The case of
Central Africa. Anthropological Linguisitcs 19:5.181-202.
. 1981. Language and development in Sub-Saharan Africa: A progress
report. Studies in the Linguistic Sciences 11:1.1-26.
. 1982. The Africanization of English. The Other Tongue: English Across
Cultures, ed. by Braj B. Kachru, 77-98. Champaign, IL; University of Illinois
Press.
. 1984a. Language and literacy in West Africa. Annual Review of Applied
Linguistics 1983, ed. by Robert B. Kaplan, 40-75. Rowley, MA: Newbury
House Publishers.
2 6 Studies in the Linguistic Sciences 20: 1 (Spring 1 990)
. 1984b. French colonial language policy in Africa and its legacies.
Studies in the Linguistic Sciences 14:2.1-36.
_. 1986. Education and development in Zaire. The Crisis in Zaire: Myths
and Realities, ed. by Nzongola-Ntalaja, 191-218. Trenton, NJ: Africa World
Press.
_. 1987. Are there syntactic constraints on code-mixing? Variation in
Language: NWAV-15 at Stanford, ed. by Keith M. Denning, et al., 34-51.
Stanford, CA: Department of Linguistics, Stanford University. Reprinted in
Tej K. Bhatia & William Ritchie (eds.) Code-Mixing: English Across
Languages. Special issue of World Englishes 8:3.277-92.
_. 1988. Code-mixing, language variation, and linguistic theory: Evidence
from Bantu languages. Lingua: International Journal of General Linguistics
71:1.21-62.
_. 1991. [English in] West Africa. English Around the World: Sociolinguistic
Perspectives, ed. by Jenny Cheshire, 493-508. London Cambridge
University Press.
_. In press. African English. The Oxford Companion to the English
Language, ed. by Tom McArthur. Oxford: Oxford University Press.
_, & J. S. Tlou. 1977. The consequences of the language policies of
African states vis-a-vis education. Language and Linguistic Problems in
Africa, ed. by Haig der Houssikian & Paul A. Kotey, 35-53. Columbia, S.C:
Hornbeam Press. Reprinted in Kahombo Mateene & John Kalema (eds.)
1980. Reconsideration of African Linguistic Policies. Kampala, Uganda:
The OAU Inter-African Bureau of Languages, 43-66.
Brosnahan, L. F. 1963. Some historical cases of language imposition.
Language in Africa, ed. by John Spencer, 7-24. Cambridge: Cambridge
University Press.
Calvet, L-J. 1982. The spread of Mandingo: Military, commercial, and colonial
influence on a linguistic datum. Language Spread: Studies in Diffusion
and Social Change, ed. by Robert L. Cooper, 184-97. Bloomington, IN:
Indiana University Press.
Champion, J. 1974. Les langues africaines et la francophonie. Paris: Mouton.
Cheshire, J. (ed.) 1991. English around the world: Sociolinguistic perspectives.
London: Cambridge University Press.
Chishimba, M. M. 1984. African varieties of English: Text in context. University of
Illinois at Urbana-Champaign unpublished Ph.D. dissertation in
Linguistics.
. 1991. [English in] Southern Africa, In Jenny Cheshire, ed. op. cit., 435-
45.
Chomsky, N. 1965. Aspects of the theory of syntax. Cambridge, MA: M.I.T.
Press.
Christian, D. 1989. Language planning: The view from linguistics. Linguistics:
The Cambridge Survey IV: Language, The Socio-Cultural Context, ed. by
Frederick J. Newmeyer, 193-209. Cambridge: Cambridge University
Press.
Clements, G. N. 1989. African linguistics and its contribution to linguistic theory.
In Eyamba G. Bokamba, Rick E. Treece, and Dorothy E. Evans (eds.) The
Contribution of African Linguistics to Linguistic Theory: Proceedings of the
Bokamba: African languages and sociolinguistic theories 2 7
20th ACAL, Vol 1. Special Issue of Studies in the Linguistic Sciences
19:2.3-42.
COLE, D. T. 1971. The history of African linguistics to 1945. In Thomas A.
Sebeok, ed. op. cit., 1-29.
CONFERMEN. 1986. Promotion et integration deslangues nationales dans les
systemes educatifs. Paris: UNESCO.
Cooper, R. L. 1978. The spread of Amharic in Ethiopia. Advances in the Study
of Societal Multilingualism, ed. by Joshua A. Fishman, 459-76. The
Hague: Mouton.
, & S. Carpenter. 1976. Language in the market. Language in Ethiopia,
ed. by M. L. Bender, et al., 244-55. London: Oxford University Press.
. 1982a. A framework for the study of language spread. In Robert L.
Cooper (ed.), 5-36.
. (ed.) 1982b. Language spread: Studies in diffusion and social change.
Bloomington, IN: Indiana University Press.
. 1989. Language planning and social change. Cambridge: Cambridge
University Press.
Das Gupta. 1973. Language planning and public policy: Analytical outline of
the policy process related to language planning in India. Georgetown
University Roundtable Conference on Languages and Linguistics, ed. by
Roger Shuy, 157-65.
Dl SciULLO, A. M., P. Muysken, & R. Singh. 1986. Government and code-mixing.
Journal of Linguistics 22:1-24.
DUMONT, P. 1983. Le francais et les langues africaines au Senegal. Paris:
Editions Karthala.
Eastman, C. M. 1983. Language planning: An introduction. San Francisco:
Chandler & Sharp Publishers.
EwiNG, A. 1984. Polish-English code-switching: A clue to constituent structure
and processing mechanism. Papers from the Regional Meeting of the
Chicago Linguistic Society, ed by J. Drogo, V. Mishra, & D. Testen, 52-64.
Chicago: Department of Linguistics, University of Chicago.
Fabian, J. (1986). Language and colonial power: The appropriation of Swahili
in the Belgian Congo 1880-1938. Cambridge: Cambridge University
Press.
Fasold. 1984. The sociolinguistics of society. Oxford: Basil Blackwell.
. 1990. The sociolinguistics of language. Oxford: Basil Blackwell.
Fellman, J. 1974. The role of Eliezer Ben Yehuda in the revival of the Hebrew
language: An assessment, ed. by J. A. Fishman, 427-55.
Ferguson, C. A. 1968. Language development. In J. A. Fishman, C. A.
Ferguson, & J. Das Gupta (eds.), 27-36.
Firth, J. R. 1957. Papers in Linguistics 1934-51. London: Oxford University
Press.
Fishman, J. a. 1968. Some contrasts between linguistically homogeneous and
linguistically heterogeneous polities. In J. A. Fishman, C. A. Ferguson, and
J. Das Gupta (eds.), 53-68.
, C. A. Ferguson, & J. Das Gupta (eds.) 1968. Language problems of
developing nations. New York: John Wiley & Sons.
28 Studies in the Linguistic Sciences 20: 1 (Spring 1990)
. 1969a. National languages and languages of wider communication In
the developing nations. Anthropological Linguistics 11:111-135
(Bloomington, Indiana)
_. 1972a. The relationship between micro- and macro-sociolinguistics in
the study of who speaks what language to whom and when.
Sociolinguistics, ed. by J. B. Pride & J. Holmes, 15-32. Harmondsworth,
U.K.: Penguin Books.
_. 1972b. Language and nationalism: Two integrative essays. Rowley, MA:
Newbury House.
_. 1974. A sociology of bilingual education. Final Report to Div. of Foreign
Studies, HEW-OE, OECO-73-0588.
_, R. L. Cooper, & A. W. Conrad, (eds.) 1977. The spread of English: The
sociology of English as an additional language. Rowley, MA: Newbury
House.
Foster, P. J. 1971. Problems of literacy in Sub-Saharan Africa. In T.A. Sebeok,
ed. op. cit., 587-617.
Giles, H., D. Taylor, & R. Bourhis. 1973. Towards a theory of interpersonal
accommodation through speech: Some Canadian data. Language in
Society 2.177-92.
, & P. M. Smith. 1979. Accommodation theory: Optimal levels of
convergence. Language and social psychology, ed. by H. Giles & R. St.
Clair, 45-65. Oxford: Basil Blackwell.
Oilman, C. 1979. Convergence in Lingala and Zairean Swahili. Anthropological
Linguistics 21.99-109.
GiNGRAS, R. C. 1974. Problems in the description of Spanish-English
intrasentential code-switching. Southwest Areal Linguistics, ed. by G. Bills,
167-74. San Diego, CA: Institute for Cultural Pluralism.
Gorman, T. P. 1974. The development of language policy in Kenya with
particular reference to the educational system. In W. H. Whiteley (ed.),
397-453.
Greenberg, J. H. 1965. Urbanism, migration, and language. Urbanization and
Migration in West Africa, ed. by H. Kuper, 50-59.
GUMPERZ, J. J. 1967. On the linguistic markers of bilingual communication.
Journal of Social Issues 23.48-57.
. 1982. The sociolinguistic significance of conversational code-swithcing.
Discourse Strategies, 55-99. Cambridge: Cambridge University Press.
, & E. Hernandez-Chavez. 1975. Cognitive aspects of bilingual
communication. El Lenguaje de los Chicanos, ed. by E. Hernandez-
Chavez and A. Beltramo, 154-63. Arlington, VA: Center for Applied
Linguistics.
Guy, G. R. 1980. Variation in the group and the individual: The case of final stop
deletion. Locating Language in Time and Space, ed. by W. Labov, 1-36.
New York: Academic Press.
. 1989. Language and social class. Linguistics: The Cambridge Survey,
Vol. IV. Language: The Socio-Cultural Context, ed. by F. J. Newmeyer, 37-
63. Cambridge: Cambridge University Press.
, B. Horvath, J. Vonwiller, E. Daisley, & I. Rogers. 1986. An intonational
change in progress in Australian English. Language in Society 15:1.23-
51.
Bokamba: African languages and sociolinguistic theories 29
HANCOCK, I. F. 1971. A Study of the sources and development of the lexicon on
Sierra Leone Krio. SOAS, University of London unpublished Ph.D.
dissertation in Linguistics.
. 1979. Readings in Creole Studies. Ghent: Story Scientia.
. 1986. The domestic hypothesis, diffusion, and componentiality: An
account of Atlantic Anglophone Creole origins. Substrata Versus
Universals in Creole Genesis, ed. by P. Muysken and N. Smith, 71-102.
Amsterdam: John Benjamins Publishing Co.
, & R. Angogo. 1982. English in East Africa. English as a World
Language, ed. by R. Bailey and f^. Gorlarch, 306-23. Ann Arbor: University
of (Vlichigan Press.
, & R. Angogo. 1979. African English: Converging or diverging varieties?
HANSFORD, K., J. Bendor-Samuel, & R. Stanford. 1976. Index of Nigerian
Languages. Ghana: Summer Institute of Language.
Haugen, Einar. 1966. Dialect, language, nation. American Anthropologist
68:922-935 (Menasha, Wisconsin).
. 1983. The implementation of Corpus Planning: Theory and practice.
Progress in language planning: International perspectives, ed. by Juan
Cobarrubias and Joshua Fishman, 269-89. Berlin: Mouton.
HEINE, B. 1970. Status and use of African languages. Munchen: Weltforum
Verlag.
HUDSON, R. A. 1980. Sociolinguistics. Cambridge: Cambridge University Press.
HULSTAERT, G. 1946. Les langues indigenes et les Europeens au Congo.
African Studies 5:126-35.
HYMES, D. 1973. The scope of sociolinguistics. Georgetown University 23rd
Annual Round Table. Sociolinguistics: Current Trends and Prospects, ed.
by R. W. Shuy, 313-334. Washington, D.C.: Georgetown University Press.
. 1974. Foundations in Sociolinguistics. Philadelphia: University of
Pennsylvania Press.
Kachru, B. B. 1978. Code-mixing as a communicative strategy in India.
International Dimensions of Bilingual Education. GURT, ed. by J. E. Alatis,
107-124. Washington, D.C.: Georgetown University Press.
. 1982a. The bilingual's linguistic repertoire. International Bilingual
Education: The role of the vernacular, ed. by B. Hartford & A. Valdman, 25-
52. New York: Plenum.
. 1983. The Indianization of English: The English language in India.
Delhi: Oxford University Press.
. 1986. The alchemy of English: The spread, functions, and models of
non-native Englishes. Oxford: Pergamon Press.
. (ed.) 1982b. The other tongue: English across cultures. Champaign, IL:
University of Illinois Press.
Kahane, H., & R. Kahane. 1979. Decline and survival of Western prestige
languages. Language 55:1.183-98.
Kamwangamalu, N. M. 1987. French-vernacular code-mixing in Zaire:
Implications for syntactic constraints. Papers from the 23rd Regional
Meeting of the Chicago Linguistic Society, 166-80. Chicago: Department
of Linguistics, University of Chicago.
30 Studies in the Linguistic Sciences 20:1 (Spring 1990)
. 1989. Code-mixing across languages: Structure, functions, and
constraints. University of Illinois at Urbana-Champaign unpublished Ph.D.
dissertation In Linguistics.
Kapanga, M. T. 1991. Language variation and change: A case study of Shaba
Swahlll. University of Illinois at Urbana-ChampaIgn unpublished Ph.D.
dissertation In Linguistics.
Karam, F. X. 1974. Towards a definition of language planning. Advances In
language planning, ed. by J. A. FIshman. The Hague.
Kashoki, M. E. 1978. The language situation In Zambia. In S. Ohannesslan and
M. E. Kashoki (eds), 9-46.
Kazadi, N. (ed.) 1987. Utilisation des langues natlonales: Actes du collogue sur
les langues natlonales; Kinshasa, [Zaire], 11-16 mars 1985. LIngulstlque
et Sciences Humalnes 27, no special. Kinshasa: CELTA.
Kennedy, C. 1982. Language planning. Language Teaching 15.266-73.
. 1984. Language planning and language education. London: George
Allen and Unwin.
King, R. D. 1969. Historical Linguistics and Generative Grammar. New Jersey:
Prentice Hall.
LABOV, W. 1966. The social stratification of English In New York City.
Washington, D.C.: Center for Applied Linguistics.
. 1972a. Language In the Inner city. Philadelphia: University of
Pennsylvania Press.
. 1972b. Soclollngulstic patterns. Philadelphia: University of
Pennsylvania Press.
. (ed.) 1980. Locating language In time and space. New York: Academic
Press.
Ladefoged, p. R. Gllck, & C. Criper. 1971. Language In Uganda. Nairobi:
Oxford University.
LAMBERT, W. E. 1967. A social psychology of blllnguallsm. Journal of Social
Issues 23.91-109.
LANHAM, L W., & C. A. Mcdonald. 1985. The standard In South African English.
Heidelberg: Groos.
LiPSKi, J. 1978. Code-switching and the problem of bilingual competence.
Aspects of Blllnguallsm, ed. by M. Paradls, 250-264.
LUAMBO, M. F. 1985. Tres Impoll [a 33 RPM album]. Bruxelles: African Sun
Music.
Magura, B. J. 1984. Style and meaning In Southern African English. University
of Illinois, Urbana-ChampaIgn unpublished Ph.D. dissertation In
Linguistics.
Marshad, H. 1984. An approach to code elaboration and Its application to
Swahlll. University of Illinois, Urbana-ChampaIgn unpublished Ph.D.
dissertation in Linguistics.
Mateene, K. 1980. Failure In the obligatory use European languages in Africa
and the advantages of a policy of linguistic independance. In K. Mateene
and J. Kalema (eds.), 9-44.
, & J. Kalema. (eds.) 1980. Reconsideration of African linguistic policies.
Kampala: GAU-Bureau of Languages.
Mazrui, a. a., & P. ZIrlml. 1978. Church, state and marketplace In the spread of
Kiswahlll: Comparative educational Implications. Case Studies In
Bokamba: African languages and sociolinguistic theories 3 1
Bilingual Education, ed. by B. Spolsky and R. L. Cooper, 427-53. Rowley,
MA: Newbury House.
McARTHUR, T. In press. The Oxford companion to the English language. Oxford:
Oxford University Press.
MUFWENE, S. S. 1979. Some grammatical changes and variations in Kikongo-
Kituba and their implications for language planning. Paper read at the
10th ACAL, University of Illinois, Urbana-Champaign (April).
. 1988. Formal evidence of pidginization/creolization in Kituba. Journal of
African Languages and Linguistics 10.33-51.
. 1989. La creolisation en bantou: Les cas di kituba, du lingala urbain, et
du swahili du Shaba. Etudes Creoles 12.74-106.
MUMBANZA, B. 1973. Y a-t-il des Bangala? Zaire-Afrique 78.471-83.
. 1974. Les Bangala du fleuve sont-ils apparentes au Mongo? Zaire-
Afrique 90.625-32.
. 1980. Histoire des peuples riverains de I'entre Zaire-Ubangi: Evolution
sociale et economique (ca. 1700-1930). These doctorale; Universite
Nationale du Zaire, Lubumbashi.
NARTEY, J. A. 1982. Code-switching, interference or fadism: Language use
among educated Ghanaians. Anthropological Linguistics 24.183-92.
Ndoma, U. 1977. Some aspects of planning language policy in education in the
Belgian Congo: 1906-1960. Northwestern University at Evanston, Illinois
unpublished Ph.D. dissertation in Linguistics.
NEUSTUPNY, J. V. 1974. Basic types of language treatment of language
problems. Advances in language planning, ed. by J. A. Fishman, 37-48.
. 1983. Towards a paradigm for language planning. Language Planning
Newsletter 9:4.1-4.
NURSE, D., & T. Hinnebusch. In press. Swahili and Sabaki: A linguistic history.
Berkeley: University of California Press.
Ohannessian, S., & M. E. Kashoki (eds.) 1978. Language in Zambia. London:
International African Institute.
PFAFF, C. W. 1979. Constraints on language mixing. Language 55.291-318.
POLOM^, E. C, & C. P. Hill. 1980. Language in Tanzania. London: Oxford
University Press.
POPLACK, S. 1980. Sometimes I'll start a sentence in Spanish y termino en
Espanol: Toward a typology of code-switching. Linguistics 18.581-618.
. 1982. Bilingualism and the vernacular. In B. Hartford and A. Valdman,
eds., op. cit., 1-23.
. 1989. Language status and language accommodation along a linguistic
border. GURT '87. Language Spread and Language Policy: Issues,
implications, and case studies, ed. by P.H. Lowenberg, 90-118.
Washington, D.C.: Georgetown University Press.
ROMAINE, S. 1982. Sociolinguistic variation in speech communities. London:
Edward Arnold.
. 1988. Pidgin and Creole languages. London: Longman.
. 1982. Colonization and pidginization on the Ubangi River. Journal of
African Languages and Lingusitics 4.1-42.
. 1984/1985. Communication by Ubangian water and word. Sprache und
Geschichte in Afrika 6.309-73.
32 Studies in the Linguistic Sciences 20:1 (Spring 1990)
. 1986. Protestant missions and the history of Lingala. Journal of Religion
in Africa 16.138-63.
1989. The Black man's burden: African colonial labor on the Congo and
Ubangi rivers, 1880-1900. Boulder, CO.: Westview Press.
Rubin, Joan, & B. H. Jernudd. 1971. Can language be planned? Honolulu:
University of Hawaii Press.
Samarin W. J. 1966. A grammar of Sango. The Hague: Mouton.
. 1982a. Colonization and pidginization on the Ubangi River. Journal
African Languages and Linguistics 4.1-42.
. 1982b. Goals, roles, and skills in colonizing Central Equatorial Africa.
Anthropological Linguistics 24.410-22.
. 1984/1985. Communication by Ubangi water and word. Sprache und
Geschiste in Afrika 6.309-373.
. 1986. Prostestant missions and the history of Lingala. Journal of religion
in Africa 16.138-63.
. 1989a. 'Official language': The case of Lingala. Status and Function of
Languages and Language Varieties, ed. by Ulrich Ammon, 386-98. Berlin:
Walter de Gruyter.
. 1989b. The Black man's burden: African colonial labor on the Congo
and Ubangi rivers, 1880-1900. Boulder, Co: Westview Press.
Sankoff, D., & S. Poplack. 1981. A formal grammar of code-switching. Papers
in Linguistics 14.3-46.
Saville-TROIKE, M. 1982. The ethnography of communication: An introduction.
Oxford: Basil Blackwell.
SCOTTON, C. M. 1971. Towards a sociolinguistic theory of choosing a lingua
franca (a system of costs and rewards in Kampala where Swahili is often a
"best buy"). Studies in African Linguistics. Supplement 2, 109-30.
. 1972. Choosing a lingua franca in an African capital. Edmonton,
Canada: Linguistic Research, Inc.
. 1978. Language in East Africa: Linguistic patterns and political
ideologies. Advances in the Study of Societal Multilingualism, ed. by J. A.
Fishman, 719-60. The Hague: Mouton.
. 1979. Elite closures as boundary maintenance: The case of Africa.
Manuscript.
. 1982. Learning lingua francas and socioeconomic integration: Evidence
from Africa. In R. L. Cooper, ed., op.cit., 63-94.
. 1983. Roots in Swahili? A locative copula becomes a stative marker.
Paper read at the LSA Annual Meeting; Minneapolis, Minnesota.
. 1988. Code-switching and types of multilingual communities. In P. H.
Lowenberg, ed., op. cit., 61-82.
Sebeok, T. a. (ed.) 1971. Current Trends in Linguistics, vol 7: Linguistics in
Sub-Saharan Africa. The Hague: Mouton.
Sey, K. a. 1973. Ghanaian English: An exploratory survey. London: Macmillan.
Seyoum, Mulugeta. 1989. The development of the national language policy in
Ethiopia. Georgetown University unpublished Ph.D. dissertation in
Linguistics.
Singh, R. 1981. Grammatical constraints on code-switching. Recherches
Linguistiques a Montreal 17:155-63.
Bokamba: African languages and sociolinguistic theories 3 3
SPENCER, J. (ed.) 1963. Language in Africa. London: Cambridge University
Press.
. 1971a. Colonial languages and their legacies. In T. A. Sebeok, ed.,
op.cit., 537-47.
. 1971b. West Africa and the English language. In J. Spencer, ed., op. cit.,
xzw.
_. (ed.) 1971c. The English language in West Africa. London: Longnnan.
SRIDHAR, S. N., & K. K. Shdhar. 1980. The syntax and psycholinguistics of
bilingual code-nnixing. Canadian Journal of Psychology 34.407-16.
Stucky, S. U. 1978. How a noun class system may be lost: Evidence from
Kituba (Lingua franca Kikongo). Studies in the Lingustics Sciences
8:1.216-33.
Stumpf, R. 1979. La politique linguistique au Cameroun de 1884-1960:
Comparaison entre les administrations coloniales allemande, francaise, et
britannique et du role joue par les societes missionnaires. Berne, W. G.:
Peter Lang.
Tabouret-Keller, A. 1968. Sociological factors of language maintenance and
language shift: A methodological approach based on European and
African examples. In Fishman, Ferguson, and Das Gupta, op. cit., 107-18.
TiMM, L. A. 1975. Spanish-English code-switching : El proque y how not-to.
Romance Philology 28.473-82.
Trudgill, p. 1983a. On dialect. Oxford: Basil Blackwell.
. 1983b. Sociolinguistics: An introduction to language and society.
Revised edition. Harmondsworth: Pelican.
TURCOTTE, D. 1981a. La politique linguistique en Afrique francophone: Une
etude comparative de la Cote d'lvoire et de Madagascar. Quebec: Presses
de rUniversite Laval.
. 1981b. Repertoire chronologique de la politique linguistique en Afrique
francophone. Manuscript.
UNESCO. 1981. Statistical Yearbook. Paris: UNESCO
. 1988. Compendium of statistics on illiteracy. Paris: UNESCO
WARDHAUGH, R. 1986. An introduction to sociolinguistics. Oxford: Basil
Blackwell.
Weinreich, U. 1953. Languages in contact: Findings and problems. The Hague:
Mouton Publishers.
, W. Labov, & M. Herzog. 1968. Empirical foundations for a theory of
language change. Directions for Historical Linguistics, ed. by W. P.
Lehmann & Y. Malkiel, 95-188. Austin, TX.: University of Texas Press.
Wentz, J. 1977. Some considerations in the development of a syntactic
description of code-xwitching. University of Illinois at Urbana-Champaign
unpublished Ph.D. dissertation in Linguistics.
Whiteley, W. H. 1969. Swahili: The rise of a national language. Studies in
African History 3. London: Methuen.
. 1971a. Some factors influencing language policies in East Africa. Can
language be planned? ed. by J. Rubin & Bjorn Jernudd, 141-58. Honolulu:
University of Hawaii Press.
. 1971b. Language policies of independent African states. In T. A.
Sebeok, ed., op. cit., 548-58.
34 Studies in the Linguistic Sciences 20:1 (Spring 1990)
. (ed.) 1971c. Language use and social change: Problems of
multilingualism with special reference to Eastern Africa. Studies presented
and discussed at the ninth International African Seminar at the University
College, Dar es Salaam [Tanzania], December 1968. London: Oxford
University Press.
. 1974. Language in Kenya. Nairobi: Oxford University Press.
Wilt, T. 1988. Bukavu Swahili: A sociolinguistic study of language change.
Michigan State University at East Lansing unpublished Ph.D. dissertation
in Linguistics.
WOOLFORD, E. 1983. Bilingual code-switching and syntactic theory. Linguistic
Inquiry 14.520-35.
WORLD BANK. 1981. Accelerated development in Sub-Saharan Africa.
Washington, D.C.: The World Bank.
YATES, B. 1980. The origins of language policy in Zaire. Journal of Modern
African Studies 18.257-79.
. 1981. Educating Congolese abroad: An historical note on an African
elite. The International Journal of African Historical Studies 14:1.34-64.
Studies in the Linguistic Sciences
Volume 20, Number 1, Spring 1990
AFRICAN LANGUAGES AND SYNTACTIC THEORIES
Joan Bresnan
Stanford University
Linguistic research on African languages has led to a number of
discoveries that have important implications for syntactic theories.
Nevertheless, this research has not yet had the same revolutionary
impact upon syntax that Africanist research has had on phonology,
where a fundamental restructuring of phonological theories was
brought about. After a review of some of the syntactic research, I will
discuss possible directions for the future. The four cases I will review
are (1) logophoricity; (2) topic, pronoun, and agreement; (3)
hierarchies and argument asymmetries; and (4) the syntax of verbs.
Case 1: Logophoricity
The phenomenon of logophoricity first came to the attention of theoretical
syntacticians in the 70's with the report in work by Hagege (1974) and Clements
(1975) that some West African languages have a morphologically distinct series
of pronouns whose use differs from both personal pronouns and reflexive
pronouns. These 'logophoric' pronouns are used in particular indirect discourse
contexts to represent the speech, consciousness, or point of view of a person
other than the speaker. Clements' work, "The Logophoric Pronoun in Ewe: Its
Role in Discourse," published in the Journal of West African Languages in
1975, showed quite clearly that the phenomenon cannot be explained in terms
of standard syntactic analysis, either by deriving the logophoric pronouns from
underlying first person pronouns, or by analyzing their antecedents as the
'deep' or 'logical' subject at an underlying level of syntactic structure. Clements
also suggested that certain puzzling properties of the reflexive pronoun in Latin
and Icelandic could be explained if they were seen as having a logophoric use.
A variety of descriptive studies subsequently amassed evidence that
certain non-clause-bounded uses of reflexive pronouns in European, East
Asian, and South Asian languages are actually logophoric.'' While
morphologically distinctive logophoric marking has been found in American
languages (e.g. O'Connor (1987)), the greatest number of cases has been
discovered in African languages (Hyman (1979), Hyman & Comrie (1981),
Frajzyngier (1985), von Roncador (1988)). However, until recently, the
phenomenon of logophoricity had never been integrated into formal linguistic
36 Studies in the Linguistic Sciences 20: 1 (Spring 1990)
theories of pronominal reference. Syntactic theories (such as Chomsky's (1981;
1986) binding theory) have defined the properties of pronominal reference
completely in terms of syntactic structures, in which notions such as
"represented consciousness" and "discourse context" have no place. Hence,
syntactic analyses of logophoric pronouns within Chomsky's framework have
generally attempted (without great success) to reduce their distribution to factors
(such as grammatical mood and tense) that can be represented in syntactic
structures (Kayne (1981;1983), Everaert (1984), Anderson (1986), Koopman &
Sportiche (1989)). (For criticism of some of these works see Bresnan,
Halvorsen, & Maling (1983), Maling (1984), Rognvaldsson (1986), and
SigurSsson (1986).)
On the other hand, semantic theories of pronominal reference (most
notably Kamp's (1981) discourse representation theory) have successfully
characterized the truth conditions for quantification and pronominal reference
relative to discourse structures. But formal semantic theories have nevertheless
lacked a rich enough linguistic structure to characterize nonquantificational
concepts like logophoricity. Thus, until recently, logophoricity had the status of a
widespread linguistic phenomenon which had not been explicitly
characterizable in any theoretical framework. An important recent development
is Peter Sells' "Aspects of Logophoricity," published in Linguistic Inquiry in
1987. Sells proposes an explicit formal framework for representing and
interpreting logophoricity, based on an extension of Kamp's theory of discourse
representation structures. Sells also shows that logophoricity can be factored
into three more primitive concepts (the source of communication, the
represented consciousness, and the point of view or deictic center) that explain
its common features as well as its variations across languages as diverse as
Ewe, Japanese, and Icelandic. (See also Abaitua (1988) for work in this
framework on Basque.) Another fascinating development is Culy, Kodio, &
Togo's (1989) use of formal analysis of anaphoric binding systems to
reconstruct the historical relations among three closely related dialects of
Dogon, spoken in Mali, and to trace the evolution of a logophoric system to a
nonlogophoric system.
What is notable about the African cases of logophoricity is that they exhibit
distinctive grammatical marking of an important discourse-dependent
dimension of pronominal reference that is indistinctly expressed almost
everywhere else. I think it is quite likely that without the Africanist research cited,
logophoric reference in all languages would remain obscured and confused
with other phenomena, and our conception of binding theory would continue to
be skewed to the purely syntactic factors that have been proposed on the basis
of more familiar, especially European, languages.
Case 2: Topic, pronoun, and agreement
A second case where Africanist research has made a significant contribu-
tion to syntactic theory is in our understanding of agreement and its relation to
pronominal anaphora. The idea that verbal agreement affixes are incorporated
pronouns appeared early in descriptions of American Indian languages
Bresnan: African languages and syntactic theories 37
(DuPonceau (1819:xxxi) cited by Mithun (1987), Boas (1911:646), Bloomfield
(1927, 1933, 1962)) as well as In the comparative grammar of Indo-European,
and typologists have often noted the relation of agreement systems to
pronominal anaphora across languages (Givon (1976), Lehmann (1982, 1984),
Greenberg (1977, 1978)). But the evolution of anaphoric pronouns into
agreement markers is nowhere more clearly exemplified than in the case of
Bantu subject and object agreement markers, which often morphologically
resemble reduced forms of independent pronouns or demonstratives. In his
1976 paper "Topic, Pronoun, and Agreement," Talmy Givon hypothesized that
verb agreement arises when a subject or object pronoun used for reference to
topics has become cliticlzed and then morphologically bound to its verb. The
resulting grammatical agreement between a verb and its subject or object, he
claimed, is indistinguishable from the anaphoric relation between a
morphologically bound pronoun and a discourse topic. But research by
Byarushengo, Hyman, & Tenenbaum (1976), Byarushengo & Tenenbaum
(1976), Wald (1979), Allan (1983), Bresnan & Mchombo (1986,1987), and
Demuth & Johnson (forthcoming) has shown that the two types of relations can
indeed be distinguished within the grammars of individual languages. There
are a variety of grammatical tests that clearly distinguish a topic from a subject
or object: these involve word order, phrasal phonology, the discourse function
of independent pronouns, interactions with relativization and question
formation, and contrastive focus constructions. By these tests the Bantu subject
and object markers are in some cases clearly incorporated pronouns in
anaphoric agreement with topics, while in other cases they have evolved into
markers of grammatical agreement with a subject or object.
In sum, there are clear formal differences between grammatical and
anaphoric agreement, yet the one merges into the other historically. This state
of affairs poses a problem for syntactic theories, formulated by Bresnan &
Mchombo (1986) as follows: " ... exactly how does this change from a pragmatic
to a syntactic relation take place? By what formal steps does pronominal
anaphora to a topic become subject-verb agreement? Current linguistic
theories have provided too little insight into the nature of agreement to answer
this question, for the simple reason that these theories have ignored the
relations between the morphological and syntactic subsystems of formal
grammar on the one hand, and, on the other, the discourse functions such as
topic and focus that these subsystems express."
Bresnan & Mchombo (1986, 1987) proposed a solution to this problem
within a formal theory of syntax (LFG). In their framework, functional ambiguity
(as between subject and topic, or agreement and pronoun) does not imply
structural ambiguity — structure and function are independent planes of
grammatical organization. In such a framework, one and the same
morphosyntactic structure can be functionally ambiguous. In this situation the
minimal change required for a pronominal affix to become an agreement
marker is a change in one single functional property of the affix: when it loses
the referential attribute that gives it pronominal meaning. The emergence of
grammatical agreement then follows from general principles relating structure
and function: since the affix lacks pronominal meaning of its own, but preserves
38 Studies in the Linguistic Sciences 20: 1 (Spring 1990)
the pronominal classificatory attributes (person, number, and gender) of a
subject or object, principles of completeness and uniqueness^ require that it be
identified ('unified') with a meaningful grammatical argument, necessarily giving
rise to grammatical agreement.
This account provides a simple formal relation between grammatical and
anaphoric agreement, but it does not explain why the pronominal affix loses its
referential attribute In the first place. As Givon maintained, such an explanation
must be functional. The agreement affix originates as a topicanaphoric pronoun,
and the gradual process of erosion into an agreement marker begins with
reference to the most highly topical types of arguments: subjects before objects,
and among objects, those that are superior on a hierarchy of person or animacy
(Givon (1976), Wald (1979), Allan (1983)). Although this progression is quite
visible cross-linguistically, it has been largely ignored in formal syntactic
theories, which have instead constructed mechanical accounts of special cases.
(Indeed, phenomena that might be analyzed as cases of morphological
agreement of a verb with the topic have been reported in several Bantu
languages (Givon (1973), Bokamba (1980, 1985), Zaenen (1981)).)
Case 3: Hierarchies and argument asymmetries
This brings us to the third case I wish to consider: hierarchies and
argument asymmetries. A major problem of contemporary syntactic theory has
been to explain the syntactic asymmetries that occur among different arguments
of a verb, such as subject-object asymmetries. Idioms are generally formed, for
example, from verbs together with their objects, but not from verbs together with
their subjects (Marantz (1984)). Grammatical agreement arises with the subject
before the object, as pointed out by Givon (1976) and others. Extractions often
treat subjects and nonsubjects differentially. Two entirely different traditions of
syntactic theory have drawn heavily on the evidence in African languages in
accounting for such argument asymmetries.
One tradition, functionalist in orientation, has maintained that semantic
and pragmatic hierarchies determine which arguments of verbs can be subjects
or objects (Hawkinson & Hyman (1974), Givon (1976), Morolong & Hyman
(1977), Duranti (1979), Hyman & Duranti (1982), Kidima (1987)). The semantic
hierarchy, for example, ranks agents above beneficiaries and recipients, and all
of these above instruments and patients. Such a hierarchy has been shown by
functional theorists to have quite general crosslinguistic validity in determining
subject and object assignment. ^
Within the formalist tradition of syntactic theory, in contrast, argument
asymmetries have been attributed to hierarchical arrangements of syntactic
functions. In Relational Grammar there is a hierarchy of syntactic functions (the
relational hierarchy) which determines 'advancements' and 'demotions' among
arguments (Perlmutter & Postal (1983)). Asymmetries among arguments arise
from this hierarchical structure, together with other axioms of the theory, such as
the uniqueness of relations at each stratum. The exact nature of this hierarchy
was thrown into question in an important study by Gary & Keenan (1977) of
Bresnan: African languages and syntactic theories 39
Kinyarwanda applicatives. Their study brought out the fact that Bantu languages
vary in the extent of asymmetry among objects: in some, like Kinyarwanda,
different arguments display nearly identical syntactic behavior. As the properties
of objects in Bantu have been more widely studied, they have deepened the
formalist inquiry into fundamental nature of the object relation (Kimenyi (1976,
1980), Kisseberth & Abasheikh (1977), Bokamba (1981), Hyman & Duranti
(1982), Dryer (1983), Perlmutter & Postal (1983), Marantz (1984), Baker
(1988a,c), and Bresnan & Moshi (1990)).
Within the formalist tradition, GB also accounts for argument asymmetries
by means of a hierarchy of syntactic functions: syntactic functions are
represented in different hierarchical positions in syntactic structure (Marantz
(1981; 1984), Baker (1988a,b)). Thus the subject argument is represented as
the NP external to the VP, and the agent role is projected into this structural
position in D-structure. Among nonsubject arguments, asymmetries are derived
from asymmetrical syntactic representations as well: for example, asymmetries
in the behavior of applied objects in Fula (Sylla (1979)) and Chichewa (Baker
1988a,b) are traced by Marantz and Baker (in somewhat different ways) to the
presence of an underlying preposition which assigns a theta role in one case
but not the other before undergoing structural incorporation into the verb as a
suffix. (A rather different account employing the thematic hierarchy is given by
Machobane (1989).)
Quite recently work has appeared which attempts to combine aspects of
both the functionalist and the formalist traditions in explaining argument
asymmetries in Bantu and other languages (see Bresnan & Kanerva (1989),
Alsina & Mchombo (1989), Bresnan & Moshi (1990), Harford (1988), and the
references cited in these). This work appeals to semantic (and in some cases
pragmatic) hierarchies which are independent of syntactic structure, but it aims
to map these hierarchies explicitly onto formal morphosyntactic structures.
Case 4: The syntax of verbs
Most of the research I have discussed so far deals with nominal and
pronominal arguments. But Africanist studies have also led to important
advances in our understanding of the syntax of verbs. An impressively well-
argued work in this area is Hilda Koopman's The Syntax of Verbs: From Verb
Movement Rules in the Kru Languages to Universal Grammar, published in
1984 by Foris. Although the verb second rules of Germanic have been familiar
topics in syntactic theory, Koopman showed that the Germanic phenomena can
be viewed as a special case of a more general theory of verb movement more
richly manifested in the previously undescribed Kru languages Vata and Gbadi.
Partly through the influence of her work, the rule of verb movement has recently
begun to play a central role in theoretical work in GB. (See Pollock (1988),
Chomsky (1988).) However, not all instances of putative verb movement can be
naturally analyzed in this way: unlike predicate clefting in Kru, for example,
predicate clefting in some other African languages involves a change from
verbal to nominal morphology (Mufwene (1987)).
40 Studies in the Linguistic Sciences 20:1 (Spring 1990)
Verb movement may be regarded as a special case of head-to-head
movement in GB, and one of the most fascinating theoretical works in this area
is Mark Baker's Incorporation: A Theory of Grammatical Function Changing,
published in 1988 by Chicago University Press. Baker aims to show that all
relation changing rules (e.g. passive, dative, causative) arise from general
principles regulating the movement of lexical categories (verbs, prepositions,
and nouns) into other lexical categories to form complex predicates. A
centerpiece of this work is his analysis of causatives, applicatives, and passives
In Bantu, which Is partly revised and elaborated In Baker (igSSb)."*
Under the heading of the syntax of verbs I also include studies of serial
verb constructions in the Kwa languages of West Africa and Caribbean Creole.
These could have a major impact on syntactic theory In the future because they
raise the issues of where the boundary is between syntax and the lexicon, and
how general the principles of the X theory of phrase structure are across
languages. 5
Conclusion: A look to the future
There are many other specific contributions of Africanist studies to
syntactic theories. The cases I have reviewed here are chosen to illustrate the
four themes I have selected, and indicate, as I have suggested, that the
linguistic study of African languages has the potential to take syntax to its
theoretical boundaries and to transform it radically, just as it has transformed
generative phonology. However, this potential has not yet been fully realized,
and in closing I would briefly like to consider why not.
It is clear that practical considerations are an important part of the answer,
as suggested to me by Eyamba Bokamba:^
There are also two practical considerations, however, that have
slowed down progress in this area: (1) the paucity of Africanist
syntacticians in the U.S. and Europe (I am discounting colleagues in
Africa here because their research is highly restricted due to working
conditions and lack of publication opportunities), and (2) the
demanding requirements for syntactic investigation.
Until the beginning of the last decade, the number of established
Africanist linguists conducting research on the syntax of African
languages was negligible. As a result, there were very few
publications in this area in the 1970s, and the teaching of Bantu
syntax or the syntax of selected African languages became an
extremely frustrating exercise. This situation, however, has changed
and is changing dramatically in the last decade as the recent
literature shows. The completion of advanced graduate studies by a
significant number of younger scholars in the the syntax of African
languages in the 1980s, combined with the expansion of
specialization by established scholars .... has rejuvenated and
provided a new impetus to the field. The impact of this research is
i
Bresnan: African languages and syntactic theories 4 1
beginning to be felt and is expected to grow this decade as more and
more scholars engage in such work.
The paucity of Africanist syntacticians is exacerbated by the
demanding requirements for syntactic investigation. In particular, the
syntactic study of African languages, unlike its phonological
counterpart, requires a much more extensive understanding of the
language(s) concerned both in terms of structural characteristics and
the development of a certain level of intuition as to "what is a
grammatical sentence" in such (a) language(s). What this means is
that the researcher has to collect a considerable amount of data and
possibly gain some speaking knowledge of the language(s) under
investigation. These requirements do not obtain in phonological
research, especially in studies involving segmental and
suprasemental phenomena based on the lexicon: All that one needs
is to collect an adequate sample of lexical elements in isolation and
limited context to permit a far-ranging analysis of the phenomenon
under consideration. Now, tonal spread phenomena have recently
pushed tonologists to go beyond the isolated lexical element to the
phrasal/contextualized lexicon with some attention being given to
syntax, but this work has not (to my knowledge) made any new
discoveries that have significantly modified previous work based on
the lexicon.
In fact, the study of phonology in syntactic context has recently been
extremely productive and exciting. Significantly, much of this work draws
heavily on research in African languages. For example, of the nineteen papers
published in Sharon Inkelas and Draga Zee's volume on The Phonology-
Syntax Connection (Inkelas & Zee, eds. (1990)), seven are devoted to African
languages (all Bantu), and four more have substantial discussions of issues
posed by African languages. But Bokamba's point still holds, in that the authors
of these works in general have much more in-depth knowledge of the
languages studied than is required elsewhere in phonology. The discovery that
discourse factors as well as syntax are directly implicated in phonological
phrasing in Bantu (Byarushengo, Hyman, & Tenenbaum (1976), Kanerva
(1989)) will undoubtedly reinforce this development in phonology.
But apart from these practical problems, there is the question of whether
current syntactic theory provides the appropriate intellectual tools for research
on African syntax. It can be argued that the architecture of Universal Grammar
as it is conventionally conceived is biassed against integrating what is most
informative in African linguistic structure (cf. Bresnan (1988, 1990)). If we look at
phonology, we see that a major development has been the factorization of
different aspects of linguistic "substance" into partial structures which are
co-present — that is, they are related nontransformationally by principles of
structural correspondence (called 'association'). Each partial structure — such
as the skeletal tier that represents "timing slots," or the feature geometry
structures that represent "phonetic substance" — has its own distinct geometry.
42 Studies in the Linguistic Sciences 20: 1 (Spring 1990)
Thus the modularity of phonological theory is embodied in the distinct geometry
of the structures and their principles of correspondence.
Syntax also aims to represent different aspects of linguistic substance, but
it has done so in a very different way. In conventional syntactic theory, all kinds
of linguistic substance are represented in structures of the same formal
geometry. For example, in GB we find semantic roles (9-roles) represented in
D-structures, surface arrangements of syntactic functions in S-structures, and
semantic scope relations in LF structures; all three structures share essentially
the same formal geometry. Moreover, this uniformity of representation is
necessary given the stipulated nature of the computations that relate syntactic
structures — iterated movements and adjunctions. Hence the modularity of
syntactic theory is not embodied in the hypothesized immanent structures of
language, as in phonology, but in the groups of principles that regulate the
stipulated computations and representations.
We see, then, that although a great deal is written about modularity in
syntax, current syntactic theory lacks true structural modularity. It is a closed
system — a very coherent, richly deductive system, but a closed system that
cannot naturally encompass other dimensions of linguistic substance. It is this
adherence to the basic computational architecture of transformations that I
would argue has prevented syntax itself from undergoing a true theoretical
transformation. If this is so, then African linguistics could have its largest impact
on syntactic theory in the development of new architectures for Universal Gram-
mar. This is the direction that I have taken in collaboration with my colleagues
and students at the Center for the Study of Language and Information at
Stanford. Our work on Chichewa, Kichaga, and other Bantu languages has led
to a deepening and generalization of what I once believed to be the universal
design of grammar. The organization of linguistic structure that is being
explored in our research project departs from the conventional Chomskyan
view. Semantic roles, syntactic constituents, and grammatical functions belong
to parallel information structures of very different formal character. They are
related not by proof-theoretic derivation but by structural correspondences, as a
melody is related to the words of a song. The song is decomposable into par-
allel melodic and linguistic structures, which jointly constrain the nature of the
whole. In the same way, the sentences of human language are themselves
decomposable into parallel systems of constraints-structural, functional, se-
mantic, and prosodic — which the whole must jointly satisfy. In this framework,
linguistic dimensions such as logophoricity, topicality and focus, and semantic
role hierarchies, which are highly important in African languages, can be
studied in relation to syntactic phrase structures, morphology, and phonology.
Such work could bridge between the rich and insightful descriptive traditions
and the powerful formal and computational approaches. How can we judge
which directions future research in syntactic theory should take in the domain of
African linguistics? Here we can rely on a simple test: how well does current
theoretical research elucidate African languages? how deep an understanding
of African language structures is gained from theoretical study? Here the best
judges here, I think, are those linguists who are speakers of African languages.
They alone possess the deep, contextual knowledge required to evaluate the
Bresnan: African languages and syntactic theories 43
alternative syntactic hypotheses. If they find intellectual satisfaction in the
discoveries and results, and if they are drawn to make use of the conceptual
tools in their own researches, then the approach is a good one.
NOTES
*This paper is a revised version of an invited address to the Plenary
Session of the 20th Annual Conference of African Linguistics at the University of
Illinois, Urbana-Champaign, April 19-22, 1989. I am grateful to Salikoko
Mufwene and Eyamba Bokamba for comments on the address, to Mark Baker,
Katherine Demuth, and Sam Mchombo for suggestions in preparing it, and to
Bill Poser and Eyamba Bokamba for suggestions for revisions. This paper is
based in part on work supported by the National Science Foundation under
Grant No. BNS-8609642.
1 See Maling (1984), Barnes (1984), SigurSsson (forthcoming),
Kameyama (1984; 1985), Kuno (1986), von Bremen (1984)).
2 The functional uniqueness condition of LFG requires that, regardless of
where it may be expressed in the word and phrase structure, information about
the same function must be consistent — and, in the case of meaning, unique.
The completeness condition requires that every argument which is lexically
required must be present (Bresnan & Mchombo (1987:745). See Kaplan &
Bresnan (1982) for one formalization.
3 See especially Dik (1978), as well as Givon (1984) and Foley & van
Valin (1984).
4 For criticism of this theory, see Alsina & Mchombo (1989), Alsina (1989),
Bresnan & Moshi (1990), Machobane (1989), and Perlmutter (1988).
5 See Sebba (1987), Baker (1989) and the references cited therein.
6 In comments in a letter to the author dated January 27, 1990.
REFERENCES
ABAITUA, Joseba. 1988. Complex predicates in Basque: From lexical forms to
functional structures. University of Manchester Ph.D. dissertation in
Linguistics.
ALLAN, Keith. 1983. Anaphora, cataphora, and topic focusing: Functions of the
object prefix in Swahili. Current Approaches to African Linguistics vol. I,
ed. by Ivan R. Dihoff, xxx-xxx. Dordrecht: Foris Publications.
44 Studies in the Linguistic Sciences 20:1 (Spring 1990)
Alsina, Alex. 1990. Where's the mirror principle? Evidence from Chichewa.
Paper presented at the 13th GLOW Colloquium at St. John's Codllege,
Cambridge University, April 6-8, 1990.
, & Sam A. Mchombo. 1988. Lexical mapping in the Chichewa applicative
construction. Paper presented at the 19th Annual African Linguistics
Conference, Boston University, April 14-17, 1988.
, & . 1989. Object asymmetries and the Chichewa applicative
construction. To appear in Theoretical Aspects of Bantu Grammar, CSLI,
Stanford University, ed. by S. Mchombo.
& . 1990. The syntax of applicatives in Chichewa: Problems for a
Theta theoretic asymmetry. To appear in Natural Language and Linguistic
Theory.
ANDERSON, Stephen. 1986. The typology of anaphoric dependencies: Icelandic
(and other) reflexives. Journal of Linguistic Research 2.1-22. Also in
Topics in Scandanavian syntax, ed. by Hellan & Koch Christensen, 65-88.
Baker, Mark. 1988a. Incorporation: A theory of grammatical function changing.
Chicago: The University of Chicago Press.
. 1988b. Theta theory and the syntax of applicatives in Chichewa. Natural
Language and Linguistic Theory 6.373-89.
. 1988c. Elements of a typology of applicatives in Bantu. Paper presented
at the 19th Annual African Linguistics Conference, Boston University, April
14-17, 1988, Special Syntax Panel.
. 1989. Object sharing and projection in serial verb constructions.
Linguistic Inquiry 20:4.513-53.
Barnes, Michael. 1984. Reflexives in the Scandinavian languages: A review of
some new ideas. Duplicated MS, University College London.
BLOOMFIELD, Leonard. 1933. Language. New York: Holt, Rinehart and Winston.
. 1962. The Menomini language. New Haven: Yale University Press.
BOAS, Franz. 1911. Handbook of American Indian languages. Bureau of
American Ethology Bulletin 40, Smithsonian Institution.
BOKAMBA, Eyamba G. 1980. Verbal agreement as a non-cyclic rule in Bantu.
Papers from the Sixteenth Regional Meeting, Chicago Linguistic Society,
April 17-18 1980, ed. by Jody Krieman & Almerindo E. Ojeda, 10-29.
Chicago: Chicago Linguistic Society.
. 1981. Aspects of Bantu Syntax. Preliminary edition for comments,
Department of Linguistics, University of Illinois, Urbana.
. 1985. Verbal agreement as a non-cyclic rule in Bantu. African
Linguistics: Essays in Memory of M. W. K. Semikenke. Studies in the
Sciences of Language Series, ed. by Didier L. Goyvaerts, 11-54.
Amsterdam: John Benjamins Publication Co. A revised and expanded
version of Bokamba (1980).
BREMEN, K. von. 1984. Anaphors: Reference, binding and domains. Linguistic
Analysis 14.191-229.
Bresnan, Joan. 1988. The design of grammar. Paper presented at the
Interdisciplinary Conference, Women on the Frontiers of Research, in
celebration of the Centennial of Clark University, March 31 and April 1,
1988, Worcester, Massachusetts.
. 1990. Levels of representation in locative inversion: A comparison of
English and Chichewa. Revised and expanded version of an invited
Bresnan: African languages and syntactic theories 4 5
address to the 13th GLOW Colloquium at St. John's College, Cambridge
University, on April 6,1990; duplicated, CSLI, Stanford University.
_, Per-Kristian Halvorsen, & Joan Maling. 1983. Invariants of anaphoric
binding systems, Unpublished MS, Stanford University.
_, & Jonni M. Kanerva. 1989. Locative inversion in Chichewa: A case study
of factorization in grammar. Linguistic Inquiry 20:1.1-50. Also forthcoming
Syntax and Semantics: Syntax and the lexicon, ed. by E. Wehrii & T.
Stowell. New York: Academic Press.
_, & . Forthcoming. The thematic hierarchy and locative inversion in
UG. A reply to Paul Schachter's comments. Syntax and Semantics: Syntax
and the lexicon, ed. by E. Wehrii & T. Stowell. New York: Academic Press.
_, & S. A. Mchombo. 1986. Grammatical and anaphoric agreement. CLS
22 Part 2: Papers from the parasession on pragmatics and grammatical
theory at the Twenty-Second Regional Meeting, ed. by A. M. Farley, P. T.
Farley, & K.-E. McCullough, 278-297. Chicago: Chicago Linguistic
Society.
_, & . 1987. Topic, pronoun, and agreement in Chichewa, Language
63, 741-782.
_, & Lioba Moshi. 1990. Object asymmetries in comparative Bantu syntax,
Linguistic Inquiry 20:2.147-185.
BYARUSHENGO, Ernest A., Alessandro Duranti, & Larry M. Hyman (eds.) 1977.
Haya grammatical structure. Southern California Occasional Papers in
Linguistics 6, June 1977.
, Larry M. Hyman, & Sarah Tenenbaum. 1976. Tone, accent, and
assertion in Haya. Studies in Bantu Tonology, Southern California
Occasional Papers in Linguistics 3, ed. by Hyman, 183-205.
, & Sarah Tenenbaum. 1976. Agreement and word order: A case for
pragmatics in Haya. Proceedings of the Second Annual Meeting of the
Berkeley Linguistics Society, 89-99.
Chomsky, Noam. 1981. Lectures on Government and Binding. Dordrecht: Foris.
. 1986. Knowledge of language: Its nature, origin, and use. New York:
Praeger.
. 1988. Some notes on economy of derivation and representation. MIT
Working Papers in Linguistics 10.43-74.
Clements, George N. 1975. The logophoric pronoun in Ewe: Its role in
discourse. Journal of West African Languages 10:2.141-77.
CULY, Christopher, Koungarma Kodio, & Patrice Togo. 1989. Reflexive and
logophoric pronouns in Dogon. Unpublished MS, Department of
Linguistics, Stanford University.
Demuth, Katherine, & Mark Johnson. Forthcoming. Interaction between
discourse functions and agreement in Setawana. Journal of African
Languages and Linguistics.
DiK, Simon C. 1978. Functional Grammar. Amsterdam: North-Holland.
DrYER, M. S. 1983. Indirect objects in Kinyarwanda revisited. Studies in
Relational Grammar, ed. by D. Perlmutter, 129-40.
DuPONCEAU, Peter S. 1819. Report of the historical and literary committee to the
American Philosophical Society. Transactions of the Historical and
Literary Committee of the American Philosophy Society vol. 1.
46 Studies in the Linguistic Sciences 20:1 (Spring 1990)
DURANTl. Alessandro. 1979. Object clitic pronouns in Bantu and the topicality
hierarchy. Studies in African Linguistics 10:1.31-45.
EVERAERT, Martin. 1984. Icelandic long reflexivization and tense-
connectedness. Working Papers in Scandinavian Syntax 12, Department
of Lingusitics, University of Trondheim, Norway.
Foley, William, & Robert Van Valin. 1984. Functional syntax and Universal
Grammar. Cambridge: Cambridge University Press.
Frajzyngier, Z. 1985. Logophoric systems in Chadic. Journal of African
Languages and Linguistics 7.23-37.
Gary, Judith, & Edward Keenan. 1977. On collapsing grammatical relations in
Universal Grammar. Syntax and Semantics: Grammatical Relations, ed. by
Peter Cole & Jerry Sadock, 83-120. New York: Academic Press.
GIVON, Talmy. 1976. Topic, pronoun, and grammatical agreement. Subject and
topic, ed. by Charles N. Li, 149-188. New York: Academic Press.
(ed.) 1979. Syntax and Semantics 12: Discourse and syntax. New York:
Academic Press.
1984. Syntax: A functional-typological introduction. Amsterdam:
Benjamins.
Greenberg, Joseph. 1977. Niger-Congo noun class markers: Prefixes, suffixes,
both or neither? Studies in African Linguistics, Supplement 7, 97-104.
. 1978. How does a language acquire gender markers? Universals of
human language, ed. by Joseph H. Greenberg et al., 47-82.
, et al. (eds.) 1978. Universals of human language, vol. 3. Stanford:
Stanford University Press.
Hagege, C. 1974. Les pronoms logophoriques. Bulletin de la Societe de
Linguistique de Paris 69.287-310.
Harford, Carolyn. 1988. Locative inversion in Chishona. Current Approaches
to African linguistics 7, ed. by J. Hutchison & V. Manfredi, 137-144.
Dordrecht: Foris Publications, 1989.
HAWKINSON, A., & L. Hyman. 1974. Hierarchies of natural topic in Shona.
Studies in African Linguistics 5.147-170.
Hellan, Lars, & Kirsti Koch Christensen (eds.) 1986. Topics in Scandinavian
syntax. Boston: D. Reidel, 1986.
Hopper, P. J., & S. A. Thompson (eds.) 1982. Studies in transitivity: Syntax and
Semantics 15. New York: Academic Press.
Hyman, Larry (ed.) 1979. Aghem grammatical structure, with special reference
to noun classes, tense-aspect and focus marking. Southern California
Occasional Papers in Lingustics No. 7, The Department of Linguistics,
University of Southern California, Los Angeles.
, & Bernard Comrie. 1981. Logophoric reference in Gokana. Journal of
African Languages and Linguistics 3.19-37.
, & Alessandro Duranti. 1982. On the object relation in Bantu. Studies in
transitivity: Syntax and Semantics 15, ed. by Paul Hopper & Sandra
Thompson, 217-239.
Inkelas, Sharon, & Draga Zee (eds.) 1990. The phonology-syntax connection.
Chicago: The University of Chicago Press.
Kameyama, Megumi. 1984. Subjective/logophoric bound anaphor zibun. Papers
from the 20th Regional Meeting, Chicago Linguistic Society, ed. by J.
Bresnan: African languages and syntactic theories 4 7
Drogo, V. Mishra, & D. Testen, 228-238. Chicago: Chicago Linguistic
Society.
_. 1985. Zero anaphora: The case of Japanese. Stanford University Ph.D.
dissertation in Linguistics.
Kamp, J. A. W. 1981. A theory of truth and semantic representation. Formal
methods in the study of language (Part 1), ed. by J. Groenendijk, T.
Janssen, & M. Stokhof, 277-321. Amsterdam: Mathematical Centre Tracts.
Also in Truth, interpretation and information, ed. by J. Groenendijk, T.
Janssen, and M. Stokhof, 1-31. Dordrecht: Foris Publications, 1984.
Kanerva, Jonni M. 1989. Focus and phrasing in Chichewa phonology. Stanford
University Ph.D. dissertation in Linguistics.
Kaplan, R. M., & J. Bresnan. 1982. Lexical-functional grammar: A formal system
for grammatical representation. The mental representation of grammatical
relations, ed. by J. Bresnan, 173-281. Cambridge: MIT Press.
KAYNE, R. 1981. ECP extensions. Linguistic Inquiry 12.93-133.
. 1983. Connectedness. Linguistic Inquiry 14.223-249.
KIDIMA, L. 1987. Object agreement and topicality hierarchies in Kiyaka. Studies
in African Linguistics 18.175-209.
KlMENYi, A. 1976. Subjectivization rules in Kinyarwanda. BLS 2.258-68.
Berkeley: Berkeley Linguistics Society.
. 1980. A relational grammar of Kinyaruwanda. University of California
Publications in Linguistics 91. Berkeley: University of California Press.
KISSEBERTH, C. W., & M. I. Abasheikh. 1977. The object relationship in
chi-Mwi:ni, A Bantu Language. Syntax and Semantics 8: Grammatical
relations, ed. by P. Cole and J. M. Sadock, 179-218. New York: Academic
Press.
KOOPMAN, Hilda. 1984. The syntax of verbs: From verb movement rules in the
Kru languages to Universal Grammar. Dordrecht: Foris Publications.
, & Dominique Sportiche. 1989. Pronouns, logical variables, and
logophoricity in Abe. Linguistic Inquiry 20:4.555-588.
KUNO, Susumu. 1986. Anaphora in Japanese. Papers from the First SDF
Workshop in Japanese Syntax, ed. by S.-Y. Kuroda, 11-70. Department of
Linguistics, University of California at San Diego, La Jolla.
Lehmann, Christian. 1982. Universal and typological aspects of agreement.
Apprehension, vol. II, ed. by H. Seller and F. J. Stachowiak, 201-267. Narr,
Tubingen.
Machobane, 'Malillo. 1989. Some restrictions on the Sesotho transitivizing
morphemes. McGill University Ph.D. dissertation in Linguistics.
Maling, Joan. 1984. Non-clause-bounded reflexives in Modern Icelandic.
Linguistics and Philosophy 7.211-241.
, & Annie Zaenen (eds.) Forthcoming. Modern Icelandic syntax. Syntax
and Semantics 24. Orlando, FL: Academic Press.
Marantz, a. p. 1984. On the nature of grammatical relations. Cambridge: MIT
Press.
MiTHUN, Marianne. 1986. When zero isn't there. Proceedings of the Twelfth
Annual Meeting of the Berkeley Linguistic Society, University of California,
Berkeley.
MOROLONG, 'Malillo, & Larry M. Hyman. 1977. Animacy, objects, and clitics in
Sesotho. Studies in African linguistics 8.199-218.
48 Studies in the Linguistic Sciences 20:1 (Spring 1990)
MUFWENE, Salikoko. 1987. An issue on predlcate-clefting: Evidence from
Atlantic Creoles and African languages. Varia Creolica, ed. by Philippe
Maurer & Thomas Stolz, 71-89. Bochum: Brockmeyer.
O'Connor, Catherine. 1987. Topics in Northern Pomo grammar. University of
California at Berkeley Ph.D. dissertation in Linguistics.
Perlmutter, David M. (ed.) 1983. Studies in Relational Grammar 1. Chicago:
The University of Chicago Press.
, & Paul Postal. 1983. Some proposed laws of basic clause structure.
Studies in Relational Grammar 1, ed. by D. Perlmutter, 81-128.
. 1988. Inactive 'objects' in Kinyarwanda: Chomage or incorporation?
Paper presented at the Nineteenth Annual Meeting of the North Eastern
Linguistics Society, Cornell University. Revised and expanded manuscript
under revision, Department of Linguistics, University of California, San
Diego.
Pollock, Jean-Yves. 1988. Verb movement. Universal Grammar, and the
structure of IP. Linguistic Inquiry 20:3.365-424.
ROGNVALDSSON, Eri'kur. 1986. Some comments on reflexivization in Icelandic.
Topics in Scandinavian syntax, ed. by Hellan & Koch Christensen, 89-102.
RONCANDOR, Manfred von. 1988. Zwischen direkter und indirekter Rede:
Nichtwortliche direkte Rede, eriebte Rede, logophorische Konstruktionen
and Verwandtes. TiJbingen: Max Niemeyer Verlag.
Sells, Peter. 1987. Aspects of logophoricity. Linguistic Inquiry 18:3.445479.
SigurSsson. 1986. Moods and (long distance) reflexives in Icelandic. Working
Papers in Scandinavian Syntax, Department of Linguistics, The University
of Trondheim. Also in Maling & Zaenen, eds. (in press).
SEBBA, Mark. 1987. The syntax of serial verbs: An investigation into se-
rialisation in Sranan and other languages. Amsterdam: John Benjamins
Publishing Company.
Wald, Benji. 1979. The development of the Swahili object marker: A study of
the interaction of syntax and discourse. Syntax and Semantics 12, ed. by
Talmy Givon, 505-524.
Zaenen, Annie. 1981. Characterizing syntactic binding domains. Occasional
Paper No. 17. Cambridge: The MIT Center for Cognitive Science.
Studies in the Linguistic Sciences
Volume 20, Number 1, Spring 1990
PHONOLOGICAL THEORY AND AFRICAN LANGUAGE
PHONOLOGY*
John Goldsmith
The University of Chicago
1. introduction
What is the relationship between phonological theory and African
language phonology? And what does it matter? These are the two questions I
would like to consider in this note. To some degree, the difference between
phonological theory and African language phonology or more generally,
phonological work in an areal subdiscipline — lies in the eye of the beholder.
What may strike one reader today as a highly theoretical work may seem in forty
years' time to be hardly theoretical at all, perhaps no more than a passing
description of some facts, while another analysis — ostensibly a simple account
of some observations — may be seen decades after the fact to be heavy-laden
with new and original perspectives going well beyond the immediate subject
matter of the paper.
A part — a large part — of the reason we may have so much difficulty in
determining whether a particular work is a contribution to theory or to African
linguistics derives from our unanalyzed assumptions regarding what the
difference between linguistic theory and a descriptive/historical field such
African linguistics is. It is difficult for most of us, I daresay, to remove ourselves
from what we may call the "data versus analysis" myth: the myth that holds that
there is in principle, or in practice, a line that can be drawn between linguistic
description, which focuses on work with informants, and linguistic analysis,
which consists of two parts: first, producing analyses of the data that have come
from the informants, and second, producing and testing theoretical models
which bear on the analyses of the data that the field-workers have so graciously
provided us with; meanwhile, the theoretical models may bear on analyses by
encouraging, discouraging, or even eliminating various such analyses.
The data vs. analysis myth encourages a particular view of what the
relation must be between linguistic theory and African linguistics: African
linguistics must be primarily data-collection, and linguistic theory must be
primarily analysis-production. If we start with assumptions such as these, then
we may end up with surprising conclusions, such as "how theoretical African
linguistics has become in the last ten years," or "African linguistics is certainly
making a major contribution to linguistic theory these days". Now, we do hear
such things, and not infrequently (underscoring the sway of this myth); and
while there is a good deal of truth to such statements, and while the self-
5 Studies in the Linguistic Sciences 20: 1 (Spring 1 990)
congratulatory back-patting that such statements lead to may well be in some
measure justified, I would like to offer a different perspective on the relation of
linguistic theory to African linguistics, which has as its central theme the
following idea: that one of the functions^ of linguistic theories is to establish
professional affiliations and distances. Thus, while linguists working within a
single theoretical framework may make serious efforts to remain knowledgeable
about the work of their colleagues within the same framework, this effort is often
counterbalanced by an unspoken sense that work which is not within one's own
framework falls beyond one's immediate responsibility.
Theory, in such a way, can have the definite effect of fragmenting the field.
A professional group such as constitutes the field of African linguists serves the
opposite function: it serves to unite, over space and over time, the work of
linguists in highly divergent theoretical frameworks. I will summarize my point
briefly as follows: no historian of modern linguistics can understand the
continuities in our field without tracing them through fields such as African
linguistics, for that is where the important ideas of our times live, prosper, and
remain fertile, often despite the Balkanizing effects of linguistic theory.
My goal, then, is to illustrate this perspective with a limited case study, the
relationship between Firthian prosodic phonology and current theories of
autosegmental and metrical phonology. I will suggest that the only way to make
sense of the historical facts of the matter is to understand the competing and
conflicting business of linguistic theory, on the one hand, and African
descriptive linguistics, on the other.
2. Firthian phonology
The British linguist J. R. Firth established a way of thinking about
phonological problems which is today generally referred to as "Firthian
phonology" or "prosodic phonology", or as "the London School". As Hill (1966)
wrote, "Prosodic Analysis made its effective debut with J. R. Firth's 'Sounds and
Prosodies' in 1948 — effective, in the sense that from this point on there has
been a continuous flow of published work from linguists practicing it." (223) As
this dating suggests, Firthian developments were contemporaneous with similar
developments in the United States of the sort discussed in Zellig Harris" (1944)
work on long and simultaneous components, Charles Hockett's (1947)
developments of this, and Bernard Bloch's (1948) work as well. I will not discuss
this American development here, in part because I have discussed it elsewhere
(Goldsmith 1976). Firth's work was also roughly contemporaneous with much of
the work in this area by Kenneth Pike, though Pike's work continued after Firth's
own ended; for a practical summary of Pike's work in the area of African
linguistics (covered virtually not at all in his well-known Tone Languages 1948,
see Pike 1966.)
The concerns that are central to papers written within the Firthian tradition
in the 1940s, 1950s, and 1960s are much more in tune with the current spirit of
theoretical concerns than are those of the bulk of papers written in other
theoretical frameworks of the time.^
Goldsmith: Phonological theory and African language phonology 5 1
Firth's approach to phonological analysis began with a division of the
sound features of a given language into phonematic units and prosodies. The
phonematic units we might think of as corresponding to the elements of a
skeletal tier in autosegmental terms, though typically they would have some
phonological substance. Another useful analogy would be to equate Firth's
phonematic units to a melody tier in an autosegmental model, a tier which was
distinguished for only consonant and vowel features, as in, for example,
McCarthy's work on Arabic (McCarthy 1979). Firth did see these phonematic
units as being the core, irreducible point-like units of phonological analysis;
prosodies used them to spread over. Firthian analysis also includes a kind of
prosody that consists of C and V patterns, as when a particular grammatical
pattern is always expressed with a CVCCVC pattern, which would be a Firthian
prosody. Prosodies more generally correspond to autosegmentalized features,
as well as to metrical structure, such as syllable, foot, and grid structure.
The notion of phonematic unit was not an easy one, it would seem, for
American linguists to grasp, and indeed, little or no use was made of such
notions in the North American context; Gleason has recently written about this,
noting that from the point of view of American structuralism, it was only natural to
interpret the word "phonematic" as an idiosyncratic variant of "phonemic", which
was not at all what Firth meant; but American linguists were accustomed to
Joshua Whatmough's inveighing against the word "phonemic", which he
thought should be "phonematic" on purely etymological grounds. In short,
Americans were equipped to misunderstand some of Firth's terminological
decisions.
Now, to understand any linguistic movement, we must understand what it
is a reaction against, of course; to understand phonemics, we must understand
that it was in part a reaction to the deluge of irrelevant phonetic information that
phoneticians were immersed in (see R. H. Robins' remarks, in Robins 1970, pp.
170-71, 210-13); to understand prosodic phonology, we must recognize that it
was in part a reaction to phonemics, whose concern for determining what was
phonologically contrastive within the segmentable speech signal was so
thorough-going that it left little or no room for considerations of higher-level
phonological structure. A phonemic analysis requires that phonetic information
be sorted into the constrastive and the non-contrastive, and requires that the
phonologically non-contrastive not be represented on the phonemic level. Firth
saw that the move of eliminating all of the phonologically predictable material
frequently made it impossible to draw the generalizations that involved higher
level structure, such as that brought in with considerations of syllable structure
or vowel harmony. Thus Firth was more interested in determining the broader
sound patterns of a language than he was in developing a model or a notation
in which all and only constrastive information would be represented. To put it
another way, the phonemicists' move to eliminate redundant phonetic
information was eventually viewed by the phonemicists themselves as a goal in
itself, while for a Firthian such a step was a reasonable one, but only as a
means to a higher end, the determination of the larger phonological pattern of
the language.3
52 Studies in the Linguistic Sciences 20:1 (Spring 1990)
As Robins has observed, the phonemicists' emphasis on matters of
contrast placed the focus on paradigmatic questions, and Firth, like the
phonologists of the 1980s, was equally concerned with (if not more concerned
with) questions of syntagmatic relations in phonology. Paradigmatic questions
would naturally focus on issues of inventory, and in particular on inventory of
sounds (rather than, say, inventories of syllable types or word-level tone
melodies). Phonologists today have by no means lost this concern of the
structuralists, of course; the renewed interest in the underspecification of
features that plays a major role in discussions of both lexical anu
autosegmental phonology is the direct descendant of this issue. Any version of
lexical phonology includes the premise that a rule of 'allophony' — as a
structuralist would have put it — introducing non-contrastive phonetic
differences — may not precede a rule sensitive to word-level morphological
considerations. Within the African context, we can find cases that illustrate
difficulties for that position, appealing though it is in general; for example, in
KiRundi, the rule weakening voiceless stops to breathy aspiration (i.e., an /h/)
after a nasal is an 'allophonic' post-lexical rule; however, it bleeds (i.e., blocks
from applying) a well-known rule of Bantu lexical phonology, Dahl's Rule, which
in KiRundi voices an obstruent in a morpheme that immediately precedes a
noun or verb stem. Thus Dahl's Rule applies in the negative subordinate
(present tense) in changing u ta tern a\o u da tern a ('that you not cut') but it fails
to apply in n-ta-tem-a) where instead of voicing, the form ta undergoes the post-
nasal softening, becoming [n hi tem a].
Returning to the Firthian school, central to their concerns were
phonological 'features' that spread over such units as the syllable and the word.
Of these, clear cases that could be handled directed were vowel harmony,
nasal harmony, and certain other harmonies of this sort. (The interested reader
may consult various references in Bazell 1966, Langendoen 1968, and Palmer
1970).
2.1 Tone
But perhaps surprisingly, the Firthian treatment of tone (at least in the
African context) was a good deal less insightful than its treatment of other
prosodic effects, and I think that one of the reasons for this was that tone is not
just like vowel harmony; it is not just something that spreads over a large
domain in a homogeneous fashion. The Firthian approach encouraged noticing
respects in which a tone pattern was a property of an entire word, and in the
case of African languages, this was an important step toward the correct
analysis, a step which permitted a correlation to be established between
grammatical and lexical dimensions and the tone melody of the word abstracted
away from the syllable template. But tone is not like nasalization, even when
nasalization is as grammaticized as Bendor-Samuel showed that it is in Terena,
where the first-person singular is marked by a prosody of nasalization (see
Bendor-Samuel 1960). For in tone systems, it is necessary to come to grips with
a kind of internal segmentation within the tonal melody or envelope. As we
have come to see in the last ten or fifteen years, this autonomous segmentation
of tone, and other prosodic levels, is an important characteristic of African tone
I
Goldsmith: Phonological theory and African language phonology 5 3
systems, and autosegmental analyses specifically differ from their Firthian
counterparts in insisting on segmentation of a uniform sort on each tier. Indeed,
it is fair to say that this is the central idea of autosegmental phonology: that the
effects impressionistically called "suprasegmental" are still just as "segmental"
as anything else, in the sense that they consist of linear sequences of more
basic units which can be treated analytically.
But that kind of segmentation of prosodies has been quite foreign in spirit
to prosodic analysis, I think it is fair to say; indeed, it was a Firth's antipathy,
hostility, and mistrust of the segmentation that had led to traditional phonemic
segmentation in the first place that brought him to the postulation of prosodies.
This difference between the conception of autosegments and that of prosodies
Is one of the most important and distinctive. The end result was that Firthian
tonal analysis was practical and insightful when applied to the treatment of tone
languages with short words (such as many Asian languages) (Sprigg 1955,
Scott 1956, e.g.), but of more limited practical and theoretical success when
applied to the analysis of African tone languages, where the domain across
which tones may be mapped, and may interact, is frequently much larger — as
is certainly the case in the Bantu languages, as well as in Igbo and a number of
other West African languages.
2.2 Degree of specification
Firthian analysis addressed a question that is very much with us today,
that of the number of "values" that are specified for a given feature. In a general
essay on prosodic analysis by Hill (1966) published in the collection dedicated
to Firth after his death. Hill writes:
...there is nothing about the incidence of frontness and backness in
the native Turkish word that would lead us to treat either one as the
marked member of an opposition. The case of roundedness,
however, is different: we can state a rule for its occurrence in the
word, but there is no complementary rule, of the same order of
simplicity, for the occurrence of spreadness....To illustrate the point
further, we may take verbal tone in Nyanja [here the writer bases
himself on his own work]. In Nyanja words. ..each syllable has a high
or low tone: there are virtually no restrictions on sequences, except
that final low-high does not occur.. .our natural inclination [would] be
to treat high/low as a pair of equipollent alternant features. However,
each Nyanja verb tense has a characteristic tone pattern. If we
examine its operation with verb stems containing varying numbers of
syllables, we shall see that the tense tone pattern is a set of high
tones: so many syllables must have them, the rest are unmarked,
therefore low.
This discussion is by no means isolated in the Firthian literature, and we
see that the nature of specification — whether something akin to features
should be monovalent (or privative), as Hill suggests, or bivalent (or
equipollent) is an important question, one which still remains unresolved in its
54 Studies in the Linguistic Sciences 20:1 (Spring 1990)
entirety. Without reading too much into this passage, I think that one gets a
sense that the issue is even more alive as a matter of the active architecture of
the grammar, writing within a Firthian context, than the question would be for
Trubetzkoy, for Hill does not simply want to conclude whether the feature is
monovalent or bivalent — that is, privative or equipollent — he wants us to
understand that this decision has further consequences with regard to other
principles down the line that appear in our grammar.
2.3 Quantity and syllable structure
An insightful and influential paper on syllable structure in Luganda, and in
fact more generally in Bantu, was published by A. N. Tucker in 1962. In this
extended discussion, Tucker develops an account which brings out an
"aesthetically satisfying" (122) picture, as he puts it, of the syllable in Luganda,
treating a number of problems that have traditionally been recognized to be
especially problematic areas for segmentally-oriented theories of phonology:
the problem of geminate consonants; the problem of the long/short vowel
contrast; the nature of syllable weight or quantity, and its relation to tone. There
are a certain number of "dynamic" aspects to his analysis, that is, places where
his analysis speaks of one thing "becoming" another under various conditions
or subject to various constraints, and in this respect his analysis is amenable to
a generative reinterpretation.
Tucker's conclusions focus on aspects such as the following: he says that
"one of the outstanding characteristics of Luganda is that, although
compensation for elision or contraction is made, this compensation must never
allow a long syllable to contain more than two morae. Consequently if two
lengthening features come together, their effect is not cumulative." While we
might expect this to be stated in turn as a condition on what a possible syllable
then is in Luganda, Tucker does not ever in fact do that; he does not take the
step of equating limitations on dynamic processes to constraints on possible
structures, though put in this way we may have little doubt that he would agree
on the natural connection between the two.
Such notions of derivation subject to cumulative restrictions were quite
uncharacteristic of most phonological theories of the time, including generative
theories; Kisseberth's discussion of "conspiracies" was perhaps the first clear
discussion in generative terms, and it was not published until close to a decade
later.4
Tucker is at pains In this paper to motivate the notion of mora as the
appropriate analytical tool for understanding vowel length, consonant length,
and tonal association, and makes arguments that sound quite contemporary in
this respect. For example (145), he argues that the first half of a geminate
voiceless stop (as in ku-coppa 'to become a pauper') is to be associated with a
mora, and is thus tone-bearing; analytically we would associate this mora with a
Low tone, and the surprising consequence of this. Tucker notes, is that this Low
tone does Indeed trigger downstep on following overt High tones. He proceeds
to argue (155), as well, on tonal grounds that the syllable must be maintained
Goldsmith: Phonological theory and African language phonology
55
as a distinct unit, as well as on the grounds that the syllable is the unit which
cannot contain more than two moras.
2.4 Vowel Harmony
The treatment of vowel harmony in Igbo was an important example in the
armamentarium of the Firthian linguist. Treatments by Ward 1936 and
Carnochan 1960 were significant steps, and deserve our attention. Carnochan
analyzed the Igbo vowel system into three distinct equipollent (i.e., bivalent)
features which he called: L7R (today we would call +/- ATR), I/A (high/low), and
Y/W (front/back, or unround/round). Two of these are prosodies — L/R, and Y/W;
but I/A (high/low) is not prosodic, presumably because there is relatively little
evidence that it spreads from one phonematic unit to another. Had there been
more evidence of the feature low/high spreading, he would have extracted, or
factored out, three prosodies, leaving him with abstract V elements. This would
have been a good thing, I think it is fair to say, because there is a suffix which
Carnochan suggests cannot be defined as anything but an empty V-slot (as we
might put it today); Carnochan indicated this with a schwa (s). Thus we end up
with the following representation, in (1), which is Carnochan's, which we may
compare with an autosegmental rewriting of this, as in (2), or a more thorough-
going autosegmental reinterpretation, as in (3). Perhaps Carnochan felt some
theoretical discomfiture with the idea of extracting out all phonological material
into prosodies in the general case; in any event, he did not do so, even when it
seems attractive to us today. Perhaps his notation encouraged the choice he
made, because he expresses prosodies in the established Firthian way,
resembling a kind of logical notation, with phonematic units being written as if
they were arguments, and the prosodies were the functors, as in (1).
(1) osiri R[(A)w(Clr8)y]
(2) R
(3)
56 Studies in the Linguistic Sciences 20:1 (Spring 1990)
3. Effects of Firthian work on autosegmental and metrical studies
What, now, have been the modes and manners of the influence of Firthian
thought on current autosegmental and metrical theory? One thing is certain:
working linguists in our tradition feel little obligation to offer any citation of Firth's
theoretical or descriptive work, or to acknowledge a debt to London modes of
thought. I have not found any references to Firth in any papers ever published in
Linguistic Inquiry. But of the papers on African linguistics directly inspired by
Firth's proposals, many are cited and developed at length in the theoretical
literature. This ironic situation deserves our attention.
3.1 Tone
I shall begin with one personal example. In my own first work on
autosegmental phonology, I was influenced by work by Will Leben, whose work
in turn was a development within a generative framework of the work of such
linguists as R. C. Abraham and J. T. Bendor-Samuel, to mention just two. When
I looked for additional resources to develop the theory further, I went to find
good grammars, and good grammars are necessarily based, to be sure, on the
good linguistic insights of their authors. In the event, I found the grammar of Igbo
published by M. M. Green and G. E. Igwe (1963), which in turn was heavily
influenced by the earlier work of another linguist from the School of Oriental and
African Studies, Ida Ward (1936), a linguist influenced, in turn, by J. R. Firth at
SOAS.
Green and Igwe made little or no effort to develop a set of general rules for
the material that they gathered, but the care and attention they gave to the tonal
material, and the weight that they assigned to tone in the organization and
presentation of their material, showed clearly their sense of the importance of
these tonal factors for understanding the underlying phonological structure of
the Igbo language. They also had a clear sense that apparent allomorphy in the
language could reflect at times the syntactic structure of the Igbo sentence, and
a sense that the apparent variety of surface tonal patterns on the verb in the
various tenses must actually be the reflection of some deeper set of regularities
in the language. It was this sense, I am convinced, that made subsequent
autosegmental analysis using an autonomous tonal tier possible within an
autosegmental framework.
Early generative thinking about African tone was not very successful, and
much of it had little effect even on generative thought. ^ Carroll's (1966)
generative account of Igbo syntax and phonology, for example, did not apply
early generative techniques to the point of developing new insights into the
language. Work such as that of Edmondson and Bendor-Samuel (1966) on
Etung, and Arnott (1964) on Tiv, work that was prosodically based, was more
influential, even among generativists. Arnott's work on Tiv led to a reanalysis by
McCawley (published, 1978), which in turn drew the attention of Leben in his
influential dissertation (1973), and of Goldsmith (1976), and most recently
Pulleyblank (1983). In the treatment of Tiv, for example, if we look for it, we can
be struck — and I believe we should be — by the continuity in the description
Goldsmith: Phonological theory and African language phonology 5 7
and the analyses of these authors. In the case of all the authors but the last,
Pulleyblank, the focus was on the "tonal melody" as a unit, and how to treat this
object that is distinct from the string of segments or phonematic units. There is a
constant core of a body of data to be attended to, and to be reworked with tools
that varied from case to case; but concern for the same core phenomenon
lurking behind the data links all these analyses, despite changes in theoretical
stance.
This is a prime example of the coherence that African linguistics lends to
linguistic research and scholarship.
3.2 Quantity and syllable structure
The work on Luganda syllable structure by Tucker that I mentioned above
has been fruitful in its effects on recent work in theoretical phonology. In a paper
that circulated in a number of drafts before being published in 1985, Clements
developed an account within an autosegmental model utilizing a skeletal tier
that incorporated Tucker's insights and developed more deeply our
understanding of syllable structure, and a number of researchers have
subsequently pursued these notions additionally, in the Bantu context and
elsewhere. 6 More generally, of course, the syllable as a unit in phonological
theory has become indispensable, in one form or another.
3.3 Vowel Harmony
A striking example of acknowledged influence of prosodic thinking on
generative theory can be found in Fromkin's 1965 article, in which she studies
the segmental inventory of Twi, the morphological and syllable structure of the
language, and its system of vowel harmony, drawing on her own work on Twi as
well as work by such Firthians as Berry (1957) and Carnochan (1960), and work
as well by Boadi (1963), which is more Harrisian in its phonological tone.
4. Prosodies, autosegments, and rules
It might become easy — too easy — to draw the conclusion that Firthian
phonology already contained, in its essence, the key ideas in autosegmental
theory. I have already suggested one reason why I do not believe that this is
correct, and in general it is important, when looking at the history of linguistic
theories, not to jump from the first step, in which we find scholarly continuity, to
the second, which holds that the two are just one. Perhaps it is the fear of this
admittedly illogical jump that drives some linguists to exaggerate the lack of
scholarly continuity with the past in their own work. Be that as it may, we would
be wise to recognize some major differences between Firthian and current
autosegmental and metrical theory.^ The Firthian approach to word-level
regularities of any kind was to posit a prosody — even a regularity of the
mundane sort in which a syllable-final consonant was devoiced. In our current
conception of phonological theory, the part of the grammar responsible for such
generalizations is quite separate from the strictly autosegmental part, i.e., the
part which up till now we have seen as most directly tied to firthian prosodies. In
58 Studies in the Linguistic Sciences 20: 1 (Spring 1990)
our current view, there are phonological rules, segregated into various
components, which interact with well-formedness conditions on phonological
structures such as the syllable and the foot; there are, in addition, several levels
of phonological representation, though the details remain here a matter of
considerable disagreement and research. In short, while the continuities
between Firth and current work is real enough, no one should allow themselves
to overlook the even greater disparities that separate prosodic analysis from the
more articulated theories of our present decade.^
5. Conclusions
In this brief note, I have discussed the continuity that African linguistics
offers to linguistics, and focused on the relationship between Firthian linguistics
and current autosegmental and metrical phonology. I could have chosen other
examples, to be sure; African linguistics has equally served as a link between
the work of French and Belgian Bantuists and that of current theoreticians, as is,
I believe, well-known; that would be a story for another day, with a similar moral,
and similar stories could be told regarding Pike's work on syllable structure, and
so forth and so on.
My review has attempted to be descriptive rather than normative — to
provide a perspective from which the continuities that we perceive in our
professional lives make sense, and from which there are, correspondingly,
fewer ironies — ironies like the "rebirth" of the study of the syllable, or of tone, or
of prosodies more generally. In acknowledging that I am being descriptive
rather than normative, I trust it is nonetheless clear that I personally believe that
the continuity that African linguistics provides is a good thing; what remains an
open question, in all seriousness, is whether the divisive effects of linguistic
theory are avoidable. I certainly do not wish to be taken to be saying working on
linguistic theory makes a person narrow-minded, and unaware of what happens
outside of their own framework, nor do I wish to be understood as saying that
theorists are that way. After all, many linguists feel comfortable wearing both the
hat of the African linguist and the hat of the linguistic theorist. What I do believe
(though I have not substantiated this in these pages) is that as a professional
and social matrix, linguistic theory can all too easily be taken, and has often
been taken, to provide a rationalization and a justification for what I referred to
before as the Balkanization of linguistics — the unfortunate lack of
communication across frameworks or paradigms. It is not the theory perse that
causes the fragmentation; it is rather that theory provides a convenient way for
justifying an otherwise unfortunate, and ultimately unhealthy, narrowness.
But it can only be healthy to be aware of the nature of discontinuity in
linguistic theory, so that we may not share the misplaced outrage and apparent
frustration of a writer such as Geoffrey Sampson, who, writing in 1980, spoke
despairingly of autosegmental phonology as a set of "half-baked ideas" that
were "anticipated by far more solid work done in the "wrong" places" [meaning
outside of MIT, of course], work that is "not rejected, just ignored" (235), and
which is a reinvention of Firthian phonology "without acknowledgement to Firth"
(258). 9 Autosegmental phonology is not a reinvention of prosodic phonology; it
Goldsmith: Phonological theory and African language phonology 59
is a different model which has intentionally maintained the insights of the
prosodic school, while providing additional analytic possibilities for the
treatment of tone, vowel harmony, syllable structure, and so on, in a number of
areas where Firthian phonology had not succeeding in shedding light.
NOTES
I am grateful to Stephen Anderson and Eric Hamp for comments on an
earlier version of this paper.
"• Several of my colleagues have expressed puzzlement or dismay at my
use of the term function here, rather than some milder term, such as effect. They
have raised the question as to whether I am endorsing, in what follows, a
thorough-goingly sociological — and perhaps to that extent, non-rational —
view of linguistic science. I can only remind such readers that science, like
language itself or any other human field of endeavor, is a gridwork of motives
subject to many simultaneous levels of analysis, none of which replace the
other. A phonological account of a language does not, generally speaking,
replace a syntax or a morphology; the one supplements the other. In certain
notorious areas, these familiar components of the grammar can begin to
impinge on each other, and affect their individual autonomies; so too for the
levels of analysis of our field, as I indicate in the text. In linguistics, though, we
may go so far as to draw normative conclusions as to how why prefer our field to
operate, and I will do just that below, and suggest how some quite human and
natural functions might just as well be less noticeably represented in our
professional matrix.
2 Which is not to say that American theorists did not have a good deal to
say of relevance to current autosegmental and metrical models. I have
attempted to illustrate that point — implicitly, but in some detail — in Goldsmith
1989, referring to the work of Hockett, Bloch, and others. Nonetheless, the
heavy effect of Bloomfieldian assumptions about phonological representation in
the United States made American work largely less relevant to our current
interests when compared to Firthian work.
3 A particular case of this kind of problem is well-illustrated and discussed
by Fudge (1976), in a discussion of a thorough-going phonemicization of Bella
Coola by S. S. Newman. (See also Hill 1961.)
4 The importance of tactics as guiding rule application was first
emphasized in Lamb 1966; I discuss this, and some other points related to the
matter in the text, in Goldsmith (Forthcoming).
5 One of the few extended discussions of African tone in the heyday of
SPE phonology is in the West African context, Tone in generative phonology
(1970), edited by Ian Maddieson, Research Notes vol. 3, parts 2 and 3, from the
60 Studies in the Linguistic Sciences 20:1 (Spring 1990)
Department of Linguistics and Nigerian Languages, University of Ibadan.
Olasope Oyelaran reminds us in his paper of an interesting passage from
Gleason 1961, one of the American linguists more aware of prosodic, Firthian
trends, and also working on African languages:
It is obvious that linguists in general have been less successful in
coping with tone systems than with consonants or vowels. ...The
...need is for better theory. We should expect that general phonologic
theory should be as adequate for tone as for consonants and vowels,
but it has not been. This can be only for one of two reasons: either
the two are quite different and will require totally different theory (and
hence techniques) or our existing theories are insufficiently general.
If, as I suspect, the problem is largely of the second sort, then
development of a theory better able to handle tone will result
automatically in better theory for all phonologic subsystems."
6 See Katamba 1985, Borowsky 1983, for example.
7 I was tempted to write, "Was SOAS the Port Royal of nonlinear
phonology?" But in the case of the Port Royal grammar, too, all the questions
about measuring continuity over disparate traditions remain thorny and
unsettled.
8 This Is hardly the place for developing what I take to be the current view
of phonology, but I have done this elsewhere; cf. Goldsmith 1989.
9 This is not true, I might add; on p. 15 (Goldsmith 1976), I observe that a
prime motivation for the study of suprasegmentals within the framework of
generative phonology is that generative phonology is not as equipped as
Firthian analysis to treat problems of suprasegmentals.
REFERENCES
Arnott, D. W. 1964. Downstep in the Tiv verbal system. African Language
Studies 5.34-51.
Bazell, C. E. et al. (eds.) 1966. In memory of J. R. Firth. London: Longmans.
Bendor-Samuel, J. T. 1960. Some problems of segmentation in the
phonological analysis of Terena. Word 16:3.348-55.
Berry, J. 1957. Vowel Harmony in Twi. Bulletin of the School of Oriental and
African Studies 19.124-30.
Bloch, Bernard. 1948. A set of postulates for phonemic analysis. Language
24.3-46.
BOADI, L. 1963. Palatality as a factor in Twi Vowel Harmony. Journal of African
Languages 2.133-39.
Goldsmith: Phonological theory and African language phonology 6 1
BOROWSKY, Toni. 1983. Geminate consonants in Luganda. Current Approaches
to African Linguistics, ed. by Jonathan Kaye, Hilda Koopman, Dominique
Sportiche, & Andre Dugas. Dordrecht: Foris Publications.
Carnochan, J. 1960. Vowel Harmony in Igbo. African Language Studies 1:155-
63.
Carroll, Patricia L. 1966. A Transformational Grammar of Igbo. London:
Cambridge University Press.
Clements, George N. 1985. Compensatory Lengthening and Consonant
Gemination in Luganda. In Wetzels & Sezer 1985.
EDMUNDSON, T., & J. T. Bendor-Samuel. 1966. Tone pattern of Etung. Journal of
African Languages 5.1-6.
Firth, J. R. 1948. Sounds and Prosodies. In Palmer 1970.
Fromkin, Victoria. 1965. On System-Structure Phonology. Language 41:601-9.
. ed. 1978. Tone: A linguistic survey. New York: Academic Press.
Fudge, Erik. 1976. Phonotactics and the syllable. Linguistic Studies offered to
Joseph Greenberg vol. 2, ed. by Alphonse Juilland. Saratoga: Anma Libri.
Gleason, H. a. 1961. An introduction to Descriptive Linguistics. New York: Holt,
Rinhart and Winston.
Goldsmith, John. 1976. Autosegmental Phonology. MIT dissertation. Published
by Garland Press, New York, 1979.
. 1990. Autosegmental and Metrical Phonology. Oxford: Basil Blackwell.
. Forthcoming. Phonology as an intelligent system. Bridges between
Psychology and Linguistics: A Swarthmore Festschrift for Lila Gleitman,
ed. by D. J. Napoli & J. Kegl. Lawrence Eribaum.
Green, M. M. & G. E. Igwe. 1963. A Descriptive Grammar of Igbo. London:
Oxford University Press.
Harris, Zellig. 1944. Simultaneous components in phonology. Language.
20.181-205.
HILL, A. A. 1961. Suprasegmentals, Prosodies, Prosodemes: Comparison and
discussion. Language 37:4.457-68.
HILL, T. 1966. The technique of prosodic analysis. In Bazell 1966.
HOCKETT, Charles. 1947. Componential analysis of Sierra Popoluca. UAL
13.259-67.
Katamba, Francis. 1985. A non-linear account of the syllable in Luganda.
African Linguistics, ed. by Didier L. Goyvaerts. Amsterdam: John
Benjamins.
Lamb, Sydney. 1966. Prolegomena to a theory of phonology. Language 42:536-
73.
Langendoen, D. Terence. 1968. The London School of Linguistics. Cambridge:
MIT Press.
Leben, William. 1973. Suprasegmental Phonology. MIT Ph.D. dissertation in
Linguistics.
McCarthy, John. 1979. Formal problems in Semitic phonology and
morphology. MIT Ph.D. dissertation in Linguistics. Distributed by Indiana
University Linguistics Club.
McCawley, James. 1978. What is a Tone Language? In Fromkin 1978.
Palmer, F. R. (ed.) 1970. Prosodic analysis. London: Oxford University Press.
Pike, Kenneth. 1948. Tone languages. Ann Arbor: University of Michigan Press.
6 2 Studies in the Linguistic Sciences 20: 1 (Spring 1 990)
. 1966. Tagmemic and Matrix Linguistics applied to selected African
languages. November 1966. U.S. Department of Health, Education, and
Welfare. Ann Arbor: Center for Research on Language and Language
Behavior.
PULLEYBLANK, Douglas. 1983. Tone in Lexical Phonology. MIT Ph.D.
dissertation In Linguistics. Published by D. Reldel, Dordrecht, 1986.
ROBINS, R. H. 1970. Diversions of Bloomsbury. North-Holland Publishing
Company. Amsterdam.
Sampson, Geoffrey. 1980. Schools of Linguistics. London: Hutchinson and
Company.
Tucker, A. N. 1962. The syllable in Luganda: A Prosodic approach. Journal of
African Languages 1:2.122-66.
WARD, Ida C. 1936. An Introduction to the Ibo language. London: Heffer.
i
studies in the Linguistic Sciences
Volume 20, Number 1, Spring 1990
AFRICAN LANGUAGES, AFRICAN LINGUISTICS, AND LINGUISTIC
THEORY*
Salikoko S. Mufwene
University of Chicago
The title of this discussion reflects variation in the ways the target papers
by Joan Bresnan, John Goldsmith, and Eyamba Bokamba have been titled.
They respond to an invitation by the organizers of the 20th Annual Conference
on African Linguistics (ACAL) that we focus on contributions which the study of
African languages have made to developments in syntactic, phonological, and
sociolinguistic theories. Such variation is natural. Assuming linguistics to be an
empirical discipline, any sound theory of language must be based on actual
languages, especially if its hypotheses are claimed to have universal or
typological significance. On the other hand, languages alone may not bear on
theory unless they have been investigated. Thus, the notion 'African linguistics'
as that area! subfield of linguistics dealing with African languages may, for the
purposes of the theme of the 20th ACAL, be blurred with that of 'African
languages'. The latter notion makes possible the identification of the subfield as
a unified body of scholarship (one of the observations by Goldsmith), but it is
through the subfield that African languages can bear on linguistic research.
'Linguistic theory' will be interpreted here in more or less the same
intuitive way as in the target papers, viz., a body of hypotheses on the subject
matters of a field of inquiry. The contribution that a particular area! subfield has
made to the field at large may be interpreted as insights that the subfield has
contributed toward our understanding of the field's subject. 1 see two
perspectives from which the contribution of African languages and linguistic
theory may be assessed. Though they provide pictures which are sometimes
not convergent, together they reflect more accurately how much attention
research on African languages has been given in syntax, phonology, and
sociolinguistics, among other topical subfields.^ The first perspective should
assess the contribution that African linguistics has made to determining
standard analyses of African languages, hypotheses based only on, e.g..
Western European languages. 2 The second perspective should focus on the
role of African linguistics in the identification of what philosophers of science
call "anomalous problems," i.e., data which any analytical framework that claims
to be more adequate than its competitors must account for, in addition to
everything that they all can explain.
One way or another, the target papers focus mostly on research on
aspects of African languages that has, or should have, had a significant bearing
64 Studies in the Linguistic Sciences 20:1 (Spring 1990)
on linguistic theory. As informative as they are, the papers do not provide a
consistent measure^ of what African linguistics has contributed to either the
development of standard analyses or the identification of anomalous problems.
Perhaps there is really no across-the-board measure, as different topical
subfields of linguistics do not approach their subject matters in the same ways
and as syntacticians, phonologists, and sociolinguists do not form the same
kinds of collaborative affiliations (in the sense suggested by Goldsmith). ^
Nonetheless, I think that textbooks on syntax, phonology, and
sociolinguistics, regardless of their analytical frameworks, may provide just the
kind of obvious evaluation needed. The reason is that normally they summarize
the knowledge that may be assumed as established in an analytical framework;
they may be expected to include the standard problems and analyses, as well
as to identify the anomalous problems. They determine the picture that the
novice develops of the subfield and, in a way, his/her future contribution to the
field. Which languages are particularly well cited on particular topics in the
relevant textbooks determines what kinds of literature the motivated novice will
read, even without advice from the instructor, to further his/her knowledge. ^
Using textbooks, one develops a differential assessment of the
contribution of African linguistics to linguistic theory, consistently with Bresnan's
initial observation (p. 35, this volume) that:
Linguistic research on African languages has led to a number of
discoveries that have important implications for syntactic theories.
Nevertheless, this research has not yet had the same revolutionary
impact upon syntax that Africanist research has had on phonology,
where a fundamental restructuring of phonological theories was
brought about.
Some of the authoritative works cited by Bokamba suggest that Bresnan
could likewise have contrasted the general failure of syntactic theory to
incorporate insights from African linguistics with the relative success of
sociolinguistics. Just as it is difficult to read a good phonology textbook that
does not cite examples from African languages regarding, e.g., prosodic
features, vowel harmony, and rule ordering, it is difficult to think of a decent
sociolinguistics textbook that does not cite African languages and countries
regarding societal and individual multilingualism, lingua francas, language
planning, code-switching and mixing, diglossia and related phenomena, and
the role of colonization, of religion, and of trade in language spread. On the
other hand, it is virtually impossible to find a syntax textbook that discusses
serial verb constructions, even though the literature on this topic in the Kwa
languages is older than today's syntactic theories; in fact studies of it have
increased since the 1970s. Ironically, current research on the subject matter is
even likely to discover the relevant West African literature backward through
Creole studies! I cannot think of any textbook that mentions Bantu languages
with regard to agreement or cites them regarding the ability to drop the subject.
The same applies to the other topics discussed by Bresnan, though no one will
dispute her observation that they have been well studied in African linguistics.
Mufwene: African languages, African linguistics, and linguistic theory 65
While what I say above is true of syntactic theory in terms of textbooks
representing standard analyses of standard problems, there is the other sense
in which Bresnan is absolutely right in noting that research in African linguistics
has started making the impact expected from the long tradition of research on
African languages. The theoretically-oriented studies cited in her paper (many
of them recent) converge to build the body of anomalous problems that may no
longer be ignored, especially after Bresnan herself has done a great job of
stating the problems explicitly and of summarizing competing analyses thereof,
and as more and more eminent theoreticians like herself continue research
both on African languages and in linguistic theory. The impact will be easier to
measure when the insights she identifies in these studies find their way into
textbooks; new scholars will learn these phenomena as standard problems with
hopefully standard analyses.
It now seems relevant to ask why so much work on the syntax of African
languages has failed "to take syntax to its theoretical boundaries and to
transform it radically" (Bresnan). It seems to me that an important factor in this
state of affairs is how the problems that have preoccupied syntacticians,
phonologists, and sociolinguists have been determined from the beginning of
each subfield.
It is not surprising that the contribution of African linguistics to
sociolinguistics is not in variation analysis or the covariance of language with
sociological factors such as gender and status, but rather in areas such as
multilingualism, language planning/development, colonialism and language
spread, diglossia, and code-switching. These areas developed not only at
about the same time current trends in linguistic theory began but also when
several African and other Third World countries becoming independent needed
quick solutions for their developments. Part of the problems for most countries
were located in societal multilingualism. African new nations and other Third
World countries then determined the standard problems and African linguistics
contributed to developing the standard explanations and the relevant theory.
Many new expansions of sociolinguistics, some of them interfacing with other
topical subfields of linguistics, have developed since then. Among these are
concerns with the nativization of European languages in Third World countries
and code-switching and mixing. They have given a new dimension to
sociolinguistics, to which many of the studies cited by Bokamba have
contributed with leadership.
The story is similar for phonology, a subfield whose subject matter was
already well-defined before the Chomskyan revolution, as evidenced by
Goldsmith's paper. While the subfield has definitely been affected by this
revolution, much of what has happened in it probably was bound to happen
even without the changes that took place in syntactic analysis. In terms of
feature analysis, for instance, the Prague School had already laid the
groundwork for the development. This set aside, both the standard and
anomalous problems of phonology have, since the American descriptivist and
the European structuralist schools, been defined significantly by non-European
languages. While the Chomskyan revolution may be credited with novel
6 6 Studies in the Linguistic Sciences 20: 1 (Spring 1990)
approaches and solutions to them, the problems themselves have continued to
be defined by the (kinds of) languages in which phonologists have traditionally
been interested. Thus being a phonologist has in most cases meant being an
Africanist, or a Sinologist, or an expert in some other non-Indo-European
language(s). Having played a central role in the development of phonology, it is
normal that African languages and linguistics have contributed significantly to
shaping phonological theory.
Things are different for syntax, where since the Chomskyan revolution, the
standard and anomalous problems have been defined primarily by English and
a handful of Western European languages. ^ This is the picture one gets from
surveying both the technical literature and textbooks. It is only recently that
some non-European languages such as Chinese, Japanese, and Korean have
figured in textbooks and one can only hope that African languages will join their
company. If they do, I hope that it will be primarily for the reasons advocated by
Bresnan, viz., to adjust linguistic theory, rather than simply to increase variety in
the sample of languages cited.
In all the above situations, syntax shares something with phonology and
sociolinguistics: The standard problems of each subfield date largely from its
beginnings and it takes a strong push to add new data and research questions
to the initial research agenda. While it is normal to expect those working in the
hard core of syntactic theory to show more interest in African languages and
linguistics, it is also the responsibility of Africanists to try to break the barrier that
has prevented their works from having the impact they should have had on
syntactic theory.
NOTES
* I am grateful to the organizers of the 20th Annual Conference on African
Linguistics for inviting me to participate in the meeting and discuss the target
papers, in writing this final version of the discussion, I have benefited from
comments by Eyamba Bokamba, Lioba Moshi, and Jessica Cooper. All the
shortcomings are my sole responsibility.
1 The term "sociolinguistics" is used loosely here, as in Bokamba's target
paper, for convenience, to subsume also the subfield that is more properly .
identified as "sociology of language", which deals, for instance, with language (
development and planning in multilingual countries.
2 Clements (1989:14) recognizes the need "to single out those respects in â–
which Africanists have been producers rather than just consumers of linguistics M
theories."
3 I use "consistent measure" with emphasis on "consistent" rather than on
"measure". I do not dispute the fact that by reading these papers one may
Mufwene: African languages, African linguistics, and linguistic theory 6 7
determine, in different ways, the extent to which syntactic, phonological, and
sociolinguistic theories have been influenced by research on African
languages.
4 By the end of this essay, I speculate that the different ways in which the
subfields have developed in the generative era account in part for the
differential contribution that African linguistics has made to linguistic theory.
5 One major shortcoming of this evaluation metric is that it says nothing
about frameworks for which there are no textbooks yet. Thus, although much of
what is being proposed in Lexical-Functional Grammar to remedy analytical
shortcomings of competing frameworks is based on African languages, the
contribution of African linguistics to the framework cannot be assessed without
reviewing the original, more technical literature itself. Likewise, this evaluation
metric does not do justice to reference works such as Givon (1979, 1984, 1990)
and Baker (1988), which advocate theoretical adjustments dictated in part by
African languages. Undoubtedly, such literature has contributed to linguistic
theory at least by identifying anomalous problems to which it proposes
solutions. However, it does not seem to have exerted the same kind of impact
that good textbooks institutionalize by compelling competing frameworks to
address the same problems.
6 The considerations given here are intended to complement those of
Clements (1989:19-20), especially regarding the limited number of extensive
descriptions of the grammars of African languages.
REFERENCES
Baker, Mark. 1988. Incorporation: A theory of grammatical function changing.
Chicago: University of Chicago Press.
BOKAMBA, Eyamba G. 1990. African languages and sociolinguistic theories.
This volume.
Bresnan, Joan. 1990. The contribution of African linguistics to linguistic theory:
African languages and syntactic theories. This volume.
Clements, G. N. 1989. African linguists and its contribution to linguistic theory.
Studies in the Linguistic Sciences 19:2.3-39.
GiVON, Talmy. 1979. On understanding grammar. New York: Academic Press.
. 1984. Syntax: A functional-typological introduction. Vol. 1. Amsterdam:
John Benjamins.
. 1990. Syntax: A functional-typological introduction. Vol. 2. Amsterdam:
John Benjamins.
Goldsmith, John. 1990. Phonological theory and African language phonology.
This volume.
II
Phonology
Ellen Broselow & Alice Niyondagara:
Feature geometry of Kirundi palatalization
G. N. Clements & Remi Sonaiya
Underlying feature representation in Yoruba
Omar Ka
Reduplication and prosodic constituents in Wolof
Manuela NOSKE
Vowel Harmony in Turkana
Meterwa a. Ourso & Charles H. Ulrich
Sonorant-strengthening in Lama
studies in the Linguistic Sciences
Volume 20, Number 1, Spring 1990
FEATURE GEOMETRY OF KIRUNDI PALATALIZATION
Ellen Broselow and Alice Niyondagara
State University of New York at Stony Brook
This paper provides an analysis of palatalization in Kirundi as
the coalescence of a front vowel and a preceding consonant. Since
consonants undergoing this coalescence with a front vowel exhibit a
coronal articulation, the paper argues for a feature geometry in which
both front vowels and coronal consonants are associated with a
coronal articulator node.
1. Introduction
One aspect of the distinctive feature theory of Chomksy and Halle 1968 is
an asymmetry in the sets of features used to characterize consonants and
vowels, an asymmetry which has been maintained in most current extensions of
this theory. One consequence of this asymmetry is a difficulty in describing
certain consonant-vowel interactions. In this paper we examine the interactions
between consonants and vowels in the Bantu language Kirundi, and argue for a
revision of the theory which provides for a closer match between the feature
characterizations of these two classes of sounds.
In SPE, vowel quality is described in terms of the tongue body features
[back], [high] and [low] and the labial feature [round]. Consonantal place of
articulation, on the other hand, is described in terms of two sets of features.
Consonants are specified, first, for primary stricture features ([coronal] and
[anterior]), features which are mainly irrelevant for vowels, which are assumed
to be redundantly [-anterior] and, except for retroflex vowels, [-coronal]. Second,
consonants are specified for tongue body features which serve to characterize
secondary articulations (described as superimposition of a vocalic articulation
on a consonant) or to make finer distinctions in place of articulation than are
permitted by the stricture features.
Recent revisions of feature theory potentially obviate the need for vocalic
features to characterize consonantal place of articulation, though this argument
has not, to our knowledge, been made. The revision we will be concerned with
here is that of Sagey (1986), which specifies place of articulation primarily in
terms of (privative) articulator nodes [labial], [coronal], and [dorsal]. ^ Since
Sagey allows a segment to be characterized by more than one articulator node.
72 Studies in the Linguistic Sciences 20:1 (Spring 1990)
many of the secondary articulations that in SPE are described in terms of
vocalic features can in Sagey's approach be described directly in terms of
multiple articulator nodes (for example, a palatalized velar may be described as
simultaneously [dorsal] and [coronal]) — thereby eliminating much of the
contrastive function of the vocalic features on consonants. This development
raises the question of the relationship between vocalic features and
consonantal place features in an articulator-based theory.
One area in which assumptions concerning consonantal and vocalic
features has clear consequences is in the description of the relationships
between particular classes of consonants and vowels: for example, the common
cross-linguistic association of front vowels and palatal consonants. As Clements
(1976) argued persuasively, the SPE system fails to reflect this association in
any satisfying manner. In this paper we examine the effect of vowels of different
classes on adjacent consonants and argue that Sagey's system fares no better
in reflecting the relationships between consonantal place of articulation and
vowel quality. Kirundi exhibits a pattern of consonant mutation before front
vowel suffixes that has traditionally been palatalizational. We argue below that
these consonant changes, which range from addition of a secondary
articulation in a consonant to a complete change in place of articulation, pose
problems for Sagey's model of feature geometry. We argue further that
palatalization in general is problematic for this model, and we propose a
revision of the feature geometry that reflects the connection between front
vowels and coronal consonants while also avoiding the redundancy inherent in
a theory that characterizes consonants in terms of both multiple articulator
nodes and vocalic tongue body features.
We begin with a discussion of the changes associated with each place of
articulation in the formation of the perfective verb. We show that Sagey's model
does not allow for a unified account of these changes, and we propose a
revision of this model and an analysis of the Kirundi facts within this framework.
Throughout we follow Kirundi orthography in the presentation of data, except
where the orthography provides insufficient information concerning phonetic
value; deviations from the standard orthography are identified as such in the
text.
2. Consonant changes in the perfective
Palatalization occurs in a number of morphological environments, each
associated with a suffixal front vowel. We will begin our examination with the
perfective forms of basic (nonderived) verbs, since the infinitive/perfective
alternation is a productive one, and the consonant changes of the perfective are
generally representative of other palatalizations, with one exception noted
below. The infinitive/perfective paradigm is illustrated in (1), where the root kam
ends in a labial before the final vowel -a of the infinitive, but is transformed into
my before the final vowel -e of the perfective suffix. (A detailed account of the
perfective morphology is provided in section 3; at present our focus is on the
output of palatalization.)
Broselow & Niyondagara: Feature geometry of Kirundi palatalization 7 3
(1) a. bukama 'to dry' (intransitive)
b. yakamye 'he dried' (intransitive)
Here we follow Sagey in assuming that my, like other multiply articulated units,
represents a single segment in the phonological system of Kirundi. Sagey
provides convincing evidence from closely related languages that such multiply
articulated elements pattern like single segments with respect to processes like
syllabification and reduplication, and that the various articulations of these
segments are phonologically unordered. Similar arguments for Kirundi are
provided by Bothner-By (1988), who argues that Kirundi syllable structure is
uniformly CV, and that orthographic sequences of consonants actually
represent single, multiply articulated segments. We therefore assume a
structure for my roughly like that in (2):
(2) root
I
. . . place . . .
/ \
labial coronal
I
-ant
(2) illustrates a single segment with two simultaneous places of
articulation, palatal and labial. Because my is a single segment, it has a single
root node, but because it has two places of articulation, the place node
branches into two articulator nodes, one labial and one coronal.
As (3b) illustrates, the labial obstruent also acquires a palatal articulation
before the final vowel of the perfective:
(3) Labials
a. m -^ my
gukama/yakamye 'to dry/he dried'
b. b (p) ^ vy
kuraaba/yaraavdye 'to look at/he looked at'
Orthographic b in (3b) represents a labial fricative, which is realized
allophonically as a labiodental fricative when combined with a palatal stop (dy);
this combination is represented in the orthography as vy. The alternation ku/gu
in the infinitive prefix is due to the well-known phenomenon of Dahl's Law: gu
appears before roots beginning in voiceless consonants, while ku appears
before voiced consonants.
As we would expect, palatalization does not affect consonants that are
already palatal, as shown in (4). {sh represents a voiceless palatal fricative, c
and y palatal affricates):
7 4 Studies in the Linguistic Sciences 20: 1 (Spring 1 990)
(4) Palatals
a. sh -^ sh
kuglsha/yagishe "drive cattle'
b. j^j
kumiija/yamilje 'sprinkle'
And glottal h becomes the palatal fricative sh:
(5) Glottals
gutaaha/yataashe 'go home'
However, not all consonants exhibit such straightforward behavior in the
perfective. Alveolars vary according to manner of articulation, as illustrated in
(6):
(6) Alveolars
a. n ^ ny
gukina/yakinye 'play'
b. t-^s/sh
kuroota/yaroose 'dream'
kumata/yamashe 'stick'
c. s ^ sh
kumesa/yameshe 'do laundry'
d. r -^ dz/y
kurira/yaridze 'cry'
kubarira/yabariye 'tell'
While alveolar nasals become palatal (as shown in (6a)), obstruents may either
remain alveolar, or become entirely palatal, as in (6b-d). The choice of svs. sh
as the palatalized varient of t is determined by the number of moras in the verb
root, as is the choice of palatalized variant of r (6d), which may surface either as
y or as an alveolar affricate (which is represented in the orthography as z,
reflecting the fact that z and dz are in free variation in this language).
Velars also exhibit a somewhat surprising pattern:
(7) Velars
a. k^ts
guteeka/yateetse 'cook'
b. g -^ dz
kwooga/yoodze 'swim'
c. k-^c
iceera 'white ones'
iki +eera
class marker +white
d. b -> vdy
ivdyeera 'white ones'
ibi +eera
class marker, pi. +white
Broselow & Niyondagara: Feature geometry of Kirundi palatalization 7 5
Velars are generally replaced by alveolar affricates ts and dz (= orthohgraphic
z), as In (7a) and (b). However, km a noun class prefix is replaced by a palatal
affricate, as shown in (7c). We assume that the palatalization rules are sensitive
to the difference in morphological structure: in the noun class marker, the front
vowel and the palatalizing consonant are contained within a single morpheme,
while in the perfective they are in separate morphemes. A labial consonant
contained in a class marker, however, undergoes the usual palatalization,
illustrated in (7d). (See Niyondagara 1989 for a fuller account of these facts.)
Thus while the facts of palatalization are complex, exhibiting sensitivity to
morphological and prosodic factors, one clear generalization emerges: the
output of palatalization is always a coronal — a palatal segment, a doubly-
articulated segment of which one articulation is palatal, or an alveolar segment.
Oral segments also acquire a continuant articulation under palatalization,
becoming either affricates, fricates, or glides. We will now consider what sort of
theory can provide a unified account of the coronalization involved in the
palatalization process.
As Clements (1976) points out, palatalization, when viewed as an
assimilation of consonants to [-back] vowels, is problematic in an SPE model of
phonology. Many languages have rules which front velars (for example k) to
palatals (for example c), but the assignment of the feature [-back] to a velar the
form of linking rules changing the value for [coronal] (at least), is required to
convert a fronted velar into a palatal, as shown in (8):
(8) k k< c
Cor - - +
Ant - - - [-ant,-bk] -^ [+cor]
Back +
The palatalization (or more accurately, coronalization) of Kirundi velars exhibits
a similar and even more severe instance of this problem, since the output of
palatalization may be a [+coronal, +anterior] segment.
The effect of palatalization is also a problem for Sagey's model of segment
structure. Sagey, following Clements (1985), assumes that the features
associated with a segment are organized in a hierarchical structure. Sagey's
revision of Clements' model represents multiply articulated segments as those
in which the place node dominates more than one articulator node. The
maximal expansion of the place node is as shown in (9):
76 Studies in the Linguistic Sciences 20:1 (Spring 1990)
(9) Sagey's Model of Feature Geometry
root
/ I \
laryngel i continuant, consonantal
supralaryngeal
I \
I soft palate
place I
/ I \ nasal
labial i dorsal
/ I / I \
round i high low back
coronal
/ \
anterior distributed
Palatalization of labials and of n can be described in this framework as
spreading of the feature [-back] from the following vowel onto the preceding
consonant. Sagey argues that spread of a dependent feature triggers
interpolation of the articulator node needed to support that feature; therefore,
spread of [back] will be accompanied by addition of an interpolated dorsal node
(represented in parentheses) on an original n, as in (10a), or on m or b, as in
(10b).
(10) a. n ^ ny: place place
I \ I
coronal (dorsal) dorsal
^-bk
b. m, b -^ my, by: place place
I \ I
labial (dorsal) dorsal
^--^^ I
^-bk
The structures shown in (10), however, require some additional interpretation.
First, the output of palatalizing an alveolar nasal is a palatal nasal (that is, a
nonanterior coronal sound), rather than the doubly-articulated segment shown
in (10a). Discussing similar palatalization of a coronal in Zoque, Sagey argues
that "the coronal and dorsal articulations, because they are so close to each
other, are not pronounced as two independent constrictions, but rather fuse to a
single, [-anterior] coronal articulation" (p. 110): in other words, the dorsal
articulation associated with the palatalizing vowel must be deleted. Thus
Sagey's system requires the analogue of SPE linking rules to transform the
output of palatalization to a coronal, since neither the SPE system nor Sagey's
system makes any necessary connection between [-back] and coronality.
Broselow & Niyondagara: Feature geometry of Kirundi palatalization 7 7
The doubly-articulated segments in (10b), on the other hand, are
apparently not reanalyzed; rather, the added palatal articulation remains dorsal.
Thus while palatals are characterized as coronal, palatalization is represented
as addition of a dorsal articulation. This vitiates one of the most attractive
features of Sagey's system: the parallelism between primary and secondary
places of articulation in consonants. Furthermore, the assignment of vowel
features to the dorsal node means that if the glide y is considered coronal, it
cannot be seen as the nonsyllabic counterpart of /, which must of necessity be
dorsal. Yet it is well known that /and y alternate in a number of languages (for
example, in conditioning the alterntion of /and rin Ewe).
Furthermore, this approach is even less successful with other places of
articulation. Once again, palatalization — spread of [-back] from a front vowel —
is analyzed as addition of a dorsal node. But while consonants which are
originally coronal remain coronal under palatalization, velars — segments
with an original dorsal articulation — become coronals when palatalized. It is
a mystery in Sagey's system why addition of a second dorsal articulation to an
original dorsal should change the dorsal to a coronal. Similarly, there is no
apparent reason why the palatalized counterpart of /7 — a segment with no
inherent place feature — should surface as a coronal segment, sh.
The creation of coronal segments under the influence of a neighboring
front vowel is of course by no means restricted to Kirundi; in many languages
velars are fronted to palato-alveolars by front vowels. And, as Mester and Ito
(1989) note, coronals are more likely than other segments to serve as landing
sites for palatalizing autosegments. We conclude, therefore, that the stipulation
that the feature [back] must be dominated by the dorsal node prevents us from
giving a natural account of the prevailing association between front vowels and
coronals in a wide array of languages.
As a possible but ultimately unsatisfactory means of overcoming this
inadequacy in the formalism, we might try moving the feature [back] under the
coronal node. In this case, spreading of [-back] would cause interpolation of a
coronal node, producing a segment with at least one coronal articulation.
However, this would simply shift the problem to another area of the grammar:
while providing an adequante account of the effect of front vowels on preceding
consonants, it would then create a similarly knotty problem in describing the
effect of back vowels. Kirundi has, along with palatalization, a parallel
consonant mutation caused by a following nonlow back vowel. This can be
illustrated by the effect of the passive suffix -u, which is added to a verb before
the final vowel. This suffix is realized in its effect on the preceding consonant.
When added to a velar, it produces an additional labial articulation, an effect of
the roundness of this vowel, as shown in (11):
(11) Active/Passive
a. Velars
guteeka/guteekwa 'to cook/to be cooked'
kuraga/kuragwa 'to bequeath/to be bequeathed'
78 Studies in the Linguistic Sciences 20:1 (Spring 1990)
This is analyzed in Sagey's systenn as spread of the feature [+round],
dominated by [labial]. The labial node is interpolated on the consonant to carry
the [round] feature, and the resulting segment thus displays two articulator
nodes, the original dorsal node and the added labial node, as shown in (12):
(12) kw: place
/ \
(labial) dorsal
I I
+rd +bk
All consonants other than velars show up with not only a labial articulation but
also an added velar articulation. This is the process known as velar fortition,
illustrated in (13). Labials receive only the additional velar articulation, since
they are already labial. Coronals take on both a labial and a velar articulation,
as shown in (13b) and (c). (Here we depart from the orthography, which does
not indicate the presence of the velar nasal in (1 3a) or the velar stop in (1 3b,c).
(13) a. Labials
gukama/gukamna 'to dry/to be dried'
kuraaba/kuraabga 'to look at/to be looked at'
b. Alveolars
kubona/kubonrjwa 'to see/to be seen'
gukubita/gukubitkwa 'to beat/to be beaten'
kugura/kugurgwa 'to buy/to be bought'
kuramutsa/kuramutskwa 'to visit/to be visited'
gutereredza/gutereredzgwa 'to ask/to be asked (for help)'
c. Palatals
gukoresha/gukoreshkwa 'to hire/to be hired'
kwiica/kwiickwa 'to kill/to be killed'
It seems clear that velar fortition is an effect of the following back vowel;
coalescence of the back vowel with the preceding consonant adds a velar
articulation to that consonant. In Sagey's analysis this is handled by spread of
[+back], with interpolation of a dorsal node to carry the back feature. But given
her assumpions that [back] is dominated by [dorsal], front vowels and back
vowels should be equally likely to induce velar fortition.
The generalization is clear, however: a front vowel induces coronality,
while a back vowel induces velarity.^ This is not easily accounted for in a
system in which [back] is lodged under the dorsal node. Nor is this effect
expected in the revision of Sagey's system proposed by Steriade (1987), which
goes even further in segregating vocalic and consonantal features by locating
all vowel features, including [back], under a dorsal node, while consonants are
characterized by labial, coronal, and velar nodes.
Broselow & Niyondagara: Feature geometry of Kirundi palatalization 7 9
It might seem at this point that the way to reflect the widespread
association of front vowels with coronality and back vowels with velarity is to
assign [-back] to the coronal node and [+back] to the dorsal node. To do this,
however, would obviously be to give up the binary nature of [back] as a feature.
Alternatively, we might consider front vowels to be doubly-articulated, both
dorsal and coronal. Under this assumption, front vowels would have the
representation in (14a), while back vowels would be represented as in (14b):3
(14) a. place b. place
/ \ I
coronal dorsal dorsal
I I
-bk +bk
However, this account maintains the association of front vowels and [dorsal],
with the result that front vowels should be as likely to induce velar fortition as
back vowels, since the spread of [-back] should still trigger interpolation of a
dorsal (velar) node. The representations in (14) also introduce a problem of
redundance: the feature specification [-back], while dominated by the dorsal
node, is always associated with the presence of an additional coronal
articulation.
Our proposal, then, is to simply eliminate the feature [back] from the
inventory and to assign front vowels and back vowels to different articulator
nodes: back vowels to the dorsal node and front vowels to the coronal node, as
in (15):
(15) a. front vowels b. back vowels
place place
coronal dorsal
I
-ant
Clements' (1976) arguments for the assignment of the feature [+coronal] to both
front vowels and coronal consonants are arguments for this proposal as well.
Furthermore, Clements' proposed characterization of retroflex vowels as
[+disteributed] — motivated both on articulatory grounds and on the basis of the
distribution of retroflexed vowels and consonants — would require that
coronality be associated with at least some vowels, since in Sagey's system
[distributed] is a dependent of the coronal node, and therefore cannot be
specified for solely dorsal segments. In fact, the correct characterization of
retroflex vowels (often described as similar to schwa in tongue body position but
with an additional retroflex articulation) may in fact be as doubly-articulated
segments, both coronal and dorsal. Thus the theory allows both doubly
articulated consonants and doubly articulated vowels.
The proposal outlined above is consistent not only with the assimilatory
nature of palatalization but also with the common tendency of palatalization
processes to single out coronal segments as their target. Mester and Ito (1989)
in their discussion of Japanese palatalization analyze the process as docking of
a [-anterior] autosegment on an appropriate consonant; since [anterior] is
80 Studies in the Linguistic Sciences 20:1 (Spring 1990)
dominated by the coronal node, the palatalizing autosegment seeks out
coronals."*
Furthermore, in giving up [back] as a separate feature we are eliminating a
potential redundance in Sagey's system. One function of the features [back],
[high], and [low] on consonants was to further subdivide the articulations
described by the stricture features [coronal] and [anterior]. Thus velars, uvulars,
and pharyngeals, which shared the same specifications for stricture features,
could be distinguished by their specifications for tongue body features.
However, this system of classification has been argued to be inadequate, at
least in the description of pharyngeals. ^ Recently, McCarthy (1989) has
suggested the addition of a fourth major articulator node, [pharyngeal] to the
inventory, and he provides compelling arguments that the distinction between
velars and uvulars in Arabic is best described not as a difference in tongue
body features but rather in terms of an added pharyngeal articulation in uvulars
which is not present in velars.^
One major argument for retaining the feature [back], on the other hand,
harks back to one of the original motivations for employing vowel features as
dependents of articulator nodes: the postulation of separate vowel features
makes possible a description of vowel harmony in which vowel features spread
across intervening consonants, regardless of the place of these consonants. If
backness harmony is analyzed as spread of the articulator node of a vowel,
then we would expect harmony to be blocked by intervening consonantal
articulations. However, this is a problem only under the assumption that
consonants and vowels always occupy a single plane, except when they
constitute separate morphemes. If, however, as McCarthy (1989) argues, this
assumption is too strong, then it is possible that any language that has
backness harmony independent of the articulation of intervening consonants
arranges its vowels and consonants on separate planes at some point in the
derivation. This is clearly too broad a topic for us to explore here, so we will
simply present a feature geometry for Kirundi and offer an analysis of
palatalization using this geometry. ^
The feature geometry we are proposing, then, is as in (16):
(16) root (cont,cons)
/ \
laryng. supralaryngeal
I \
place soft palate (nasal)
/ \
labial lingual
I / \
round backness height
/ \ / \
cor dors high (low)
/ \
ant dist
In this model, the place node branches into the labial and the lingual nodes.
The lingual node in turn branches into the backness node, which dominates
Broselow & Niyondagara: Feature geometry of Kirundi palatalization 8 1
[coronal] and [dorsal], and the height node, which dominates the features [high]
and [low]. (We assume, following McCarthy (1989), that a pharyngeal node is
necessary at least for languages with pharyngeal consonants, and leave open
the possibility that certain vowel contrasts might be a function of pharyngeal
articulation.) Association of the features [high] and [low] with a height node
rather than with the coronal or dorsal nodes ensures that backness harmony
and height harmony can take place independently: backness harmony is
spread of the lingual node. (If, as seems likely, low vowels can be characterized
as vowels with an added pharyngeal articulation, the feature [low] can be
dispensed with; under this account, the pharyngealized vowels of Arabic will be
doubly articulated (corono-pharyngeal or dorso-pharyngeal)). Front vowels will
be charactehzed by the presence of a coronal node dominating [-anterior],
while back vowels will be characterized by the presence of a dorsal node. We
turn now to an analysis of the Kirundi palatalization facts discussed above.
3. Palatalization in perfective and transitive verbs
Following Sagey's analysis of Kinyarwanda (Sagey 1986, Walli-Sagey
1985; see also a similar analysis of Kirundi by Kenstowicz & Kisseberth 1979 in
a linear framework), we assume that palatalization involves the coalescence of
a vowel with a preceding consonant. Sagey argues convincingly that
Kinyarwanda syllable structure is strictly CV, with no underlying distinction
between vowels and glides. Sequences of unlike vowels are not permitted;
CVV sequences may be syllabified by coalescence of the consonant and the
first vowel, as most clearly illustrated in forms (6c) and (6d), repeated below as
(17a,b). (17c,d) illustrate roots begining with back vowels, while (17e,f) illustrate
consonant-initial roots:
(17)
a.
iceera
iki+eera
'white person'
b.
ivdyeera
ibi+eera
'white person'
c.
icoobo
iki+oobo
'pit'
d.
ivdyoobo
ibi+oobo
'pits'
e.
ikiraato
'shoe'
f.
ibiraato
'shoes'
As the forms in (17) illustrate, a single following front vowel is not a sufficient
condition for palatalization, since a single vowel is syllabifiable. Similar facts
obtain in the verbal system. (18) shows the derivation of perfective and
transitive verb forms. The final consonant of the verb root is palatalized in (18b),
the basic perfective, and in (18c), the transitivized form of the verb, but in the
perfective transitive (18d) the underlying m re-emerges.
(18) a. gukama 'to dry (intransitive)'
b. yakamye 'he dried (intransitive)'
c. gukamya 'to dry (transitive)'
d. yakamije 'he dried (transitive)'
82 Studies in the Linguistic Sciences 20:1 (Spring 1990)
In light of the forms in (18), we assume that the perfective takes an /suffix before
the final vowel -e which is realized as palatalization on the root-final
consonant. 8 Similarly, the transitives are formed by a suffix / placed before the
final vowel (-a for infinitives, -e for perfectives). Since Kirundi insists on CV
syllables, the first of the two suffixes coalesces with the preceding consonant, as
represented in (19):
(19) syll
/ I
cvv
l\ I
m i e/a
The coalescence of the /with the m in (19) leaves a vacant vowel slot which
would normally be filled by lengthening the e or a; however, Kirundi does not
normally permit long vowels in final position. The underlying representations of
the forms in (17), then, are as in (20):
(20) a. basic infinitive: INF ku- + ROOT + FINAL VOWEL -a
ku+kam+a -> gukama
b. basic perfective: 3MSG ya- +ROOT + PERF -i +FV -e
ya+kam+i+e -^ yakamye
c. trans, infinitive: INF + ROOT + TRANS -i +FV -a
ku+kam+i+a -^ gukamya
d. trans, perfective: 3MSG + ROOT + TRANS -i + PERF -i +FV e
ya+kam+i+i+e -^ yakamije
For (20d), we assume a rule which takes intervocalic i\o j:^
(21) i^j/V_V
After the application of this rule to (20d), all vowels and consonants in the word
may be accommodated in a CV syllable structure. Rule (21) does not apply in
the (b) and (c) forms, however, since these have only two vocalic suffixes, which
can be incorporated into a well-formed prosodic structure by coalescence of the
first vowel with the preceding consonant and retention of the second vowel as a
syllable nucleus.
The output of the consonant-vowel coalescence illustrated in (19) is a
consonant which always has a coronal component (it is either palatal or
alveolar) and, except for nasals, a [+continuant] component (it is either a
fricative, affricate, or glide). We interpret this coalescence formally as spreading
of the place node from the vowel to the preceding consonant, with delinking of
the rest of the vowel features from their original timing slot. For the labials m and
b, this gives the structure in (22):
Broselow & Niyondagara: Feature geometry of Kirundi palatalization 8 3
(22) m ^ my, b ^ by
m, b i
ROOT ROOT
place place
labial lingual
I
coronal
I
-ant
For alveolar n, since both the n and the following vowel are coronal, the
pruning of shared features will combine the two place nodes into a single
coronal articulation. While n is [+anterior] and / is [-anterior], the result of
palatalization is nv, a [-anterior] segment. We can account for this by assuming
that [-i-anterior] is the default setting for coronals, and that therefore n is
unspecified for [anterior], in which case the [-anterior] specification of the vowel
is the one assigned to the resulting consonant. Or alternatively, we might
assume that docking features win out over those already associated to a node.
We will not attempt to choose between these alternatives here, but simply note
that either one will also account for the change of s to sh under palatalization.
Sounds which are already palatal will not be affected, since the place features
spread to them will be identical to those already specified.
(23) n ^ ny, s ^ sh
place ^ place
lingual lingual
I I
coronal coronal
I
-ant
Palatalization of non-nasal stops requires an additional wrinkle, since the
output of the palatalization of obstruents or h is always either a fricative or an
affricate, as illustrated in (24):
(24) a. gutaaha (basic inf: ku+taah+a) 'to go back home'
b. yataashe (basic pert: ya+taah+i+e) 'he went back home'
c. gutaashe (trans inf: ku+taah+i+e) 'to greet'
d. yataahije (trans pert: ya+taah+-i-e) 'he greeted'
84 Studies in the Linguistic Sciences 20:1 (Spring 1990)
We assume, therefore, that palatalization assigns the feature [+continuant] to
oral consonants, perhaps as spread of continuancy from the palatalizing vowel.
Once [+continuant] is assigned, the spread of the place features of /to h will
result straightforwardly in sh; since h has no supralaryngeal features of its own,
the spread of a minus anterior coronal articulation to a voiceless continuant
produces sh with no additional machinery:
h ^ sh
h
i
root
root
1 ^
+cont *" ""
place
coronal
1
1
-ant
Next to be accounted for are the alveolar stops r (which becomes y or dz in
the perfective), and t (which becomes sh or s in the perfective). It is apparently
unpredictable for any given lexical item which alternant it will take; examples of
each are illustrated in (5), and (26) provides a full paradigm of perfective and
transitivized forms of an r-final and a ^final verb:
(26) r ^ y/dz
a. gukira/yakidze "recover (inf./perf.)*
b. gukidza/yakirije 'heal (inf./perf.)' (-yakijije)
t -4 sh/s
c. kuruta/yaruse 'be bigger than (Inf./perf.)'
d. kurusha/yarushije 'do better than (inf./perf.)'
The change of the flap (that is, noncontinuant) rto y and /to sh represents the
triumph of [-anterior] over [+anterior] and the triumph of [+continuiant] over
[-continuant]. For r, the change to y maintains the original sonority of r, while
simple addition of [+continuant] gives dz. We assume that certain verb roots are
lexically marked to maintain [+anterior]. An additional rule of consonant
harmony takes rto /before /, transforming yakihje to yakijije in (26b). This rule is
blocked, however, in certain syntactic contexts, in which the form with r surfaces
(see Niyondagara 1989 for discussion).
The final question to account for is why the velar stops surface as alveolar
affricates, as illustrated in (6) and in (27):
(27) k,g ^ ts/c, dz
a. kwaaka/yaatse 'light (intrans.)'
b. kwaatsa/yakije 'light (trans.)'
c. iki+era -^ iceera 'white one'
Broselow & Niyondagara: Feature geometry of Kirundi palatalization 8 5
The change from stop to affricate is consistent with the behavior of the other
sounds: all non-nasals acquire a [+continuant] articulation under palatalization.
What is unusual about the velars is that the original dorsal articulation is lost
entirely, leaving only the added coronal articulation. We assume that velars
undergo the spread of the place node from the following vowel that we have
seen for other sounds, illustrated in (28):
(28) k i
root root
[-cont, -i-cont]
place ""^^ place
I I
dorsal cor
The structure shown in (28) requires an additional rule delinking [dorsal]:
(29) Dorsal Delinking
root
place
/ \
dorsal coronal
Rule (29) transforms segments which are simultaneously dorsal and coronal
into solely coronal segments. Of course, dorsal and coronal articulations do
cooccur in certain segments — specifically, in the output of velar fortition,
discussed above, where the coalescence of the passive suffix u with an
alveolar, for example, creates an alveolar-velar {gukubita/gukubitkwa 'to beat/to
be beaten). However, in these cases the segment always has a third
articulation, labial. If we assume that the association lines in rule (29) must be
interpreted as exhaustive (Hayes 1986), rule (29) will not apply to these triply-
articulated segments. Thus (29) will convert segments that are exhaustively
dorso-coronal to pure coronals, but will not apply to segments that are
simultaneously labial, dorsal, and coronal.
The output of coalescence and dorsal delinking is either an alveolar or a
palatal affricate, depending on word structure. The maintenance of the expected
[-anterior] articulation seems to be specific to pre-root morphemes, and possibly
to the morpheme ki (which may be used either as a class marker on nouns or as
an object pronoun on verbs). Root-final velars and velars in morphemes
following the root shift to a [-i-anterior] articulation. What is needed is a rule of
anteriorization, sensitive to word structure; exact statement of this rule will
require a more complete account of Kirundi word structure than we can provide
here.'io
8 6 Studies in the Linguistic Sciences 20: 1 (Spring 1 990)
4. Conclusion
To conclude, then, we have argued that palatalization is best described as
addition of a coronal articulation to a consonant. The added coronal articulation
either coexists with the original consonantal articulation or replacaes it (by
means of a rule delinking the original articulator node). Formally, palatalization
is accomplished by spreading the place node of a following vowel onto a
preceding consonant, an approach which could be extended to the description
of velar fortition as well-''^ The feature geometry we have proposed provides a
unified description of vowels and consonants which reflects the widespread
relationship between front vowels and palatalization.
NOTES
* This is a slightly revised version of a paper delivered at the 20th Annual
Conference on African Linguistics, April 1989. We would like to thank Mark
Aronoff, Christina Bethin, Su-I Chen, Elan Dresher, Dan Finer, John Golsdmith,
Marcia Haag, Larry Hyman, Alexandre Kimenyi, and Moira Yip for comments
and discussion, with the usual disclaimers. Kirundi is the native language of the
second author, and the major language of Burundi.
1 Though additional nodes have been motivated by McCarthy (1989); see
discussion below.
2 In a small number of cases — specifically CV morphemes consisting of a
coronal consonant followed by a front vowel — coalescence of the consonant
and vowel is associated with velar articulation of the resulting complex segment
in addition to the expected palatal articulation (for example, ku+se+a -^ guskya
'to grind'; of. kumesa/yameshe 'to do laundry/he did laundry'). However,
noncoronals in CV morphemes do not undergo velarization in the same context:
ku+ke+a -^ guca 'to set (of sun'); ku+vi+aar+a -^ kuvdyaara 'to give birth".
Therefore, this velahzation is not an example of the general case of velar
fortition, but instead requires rules specific to morpheme-internal coronal-V
coalescence (Niyondagara 1989).
3 A proposal somewhat similar to this is made in Pulleyblank 1989.
^ However, since the palatalizing morpheme is not an autonomous
segment, Mester and Ito are not committed to any particular representation of
vowels.
5 See, for example, Broselow 1976, 1979.
6 McCarthy's account differs in other respects from the one adopted here,
however.
(
Broselow & Niyondagara: Feature geometry of Kirundi palatalization 8 7
7 See Broselow (in preparation) for a discussion of this question in terms
of vowel and consonant harmony and interactions in Turkish and Kirundi.
8 Meeussen 1959 assumes a y suffix; however, since the syllabicity of high
vowels and glides is entirely predictable in this language, there is in fact no
evidence for positing underlying glides.
9 This applies only when the first vowel is not part of the verb root; in the
latter case, / is realized as v, as in kumnwa/yamoye 'to shave/he shaved', or
gusha/yahiye 'to burn/he burned'. As suggested to us by one reviewer, the rule
may actually be more general: v becomes; after /.
â– 10 See Myers 1987 for a suggestive account of Shona word structure.
^^ Sagey's arguments against velar fortition as spread of the place node
rely on the use of the feature [high] to distinguish velars and uvulars;
specifically, spread on the place features of a mid vowel should induce a uvular
rather than a velar articulation. We find this argument unconvincing for two
reasons: first, these languages do not employ a velar/uvular contrast, and
second, as discussed above, this contrast is probably best characterized not by
a difference in the specification of value of [high].
REFERENCES
Bothner-By, a. 1988. Phonological representations of Kirundi. Unpublished
M.A. thesis, SUNY Stony Brook.
Broselow, E. 1976. The phonology of Egyptian Arabic. University of
Massachusetts at Amherst unpublished Ph.D. dissertation in Linguistics.
. 1979. Cairene Arabic syllable structure. Linguistic Analysis 5.345-382.
. In preparation. Consonant and vowel harmony in Turkish and Kirundi.
Chomsky, N., & M. Halle. 1968. The sound pattern of English. Harper and Row.
Clements, G. N. 1976. Palatalization: Linking or assimilation? CLS 12.90-109.
. 1985. The geometry of phonological features. Phonology Yearbook
2.225-252.
Hayes, B. 1986. inalterability in CV phonology. Language 62.321-51.
Kenstowicz, M., & C. Kisseberth. 1979. Generative phonology. Orlando,
Florida: Academic Press.
KiMENYi, A. 1979. Studies in Kinyarwanda and Bantu phonology. Carbondale:
Linguistic Research Inc.
McCarthy, J. 1989. Linear order in phonological representation. Linguistic
Inquiry 20.71-100.
. 1989. Guttural phonology. Unpublished ms.
Meeussen, A. 1959. Essai de grammaire Rundi. Tervuren: Annales du Musee
Royal du Congo Beige.
Mester, a., & J. Ito. 1989. Feature predictability and underspecification palatal
prosody in Japanese mimetics. Language 65.258-293.
88 Studies in the Linguistic Sciences 20: 1 (Spring 1990)
Myers, S. 1987. Tone and word structure in Shona. University of
Massachusetts at Amherst unpublished Ph.D. dissertation in Linguistics.
NiYONDAGARA, A. 1989. Phonological mechanisms in Kirundi perfective and
transitive forms. Ms., SUNY at Stony Brook.
PULLEYBLANK, E. 1989. The role of coronal in articulator based features.
Delivered at the Chicago Linguistics Society Meeting.
Sagey, E. 1986. The representation of features and relations in non-liner
phonology. MIT unpublished Ph.D. dissertation in Linguistics .
Steriade, D. 1987. Locality conditions and feature geometry. NELS 17.595-
618.
Walli-Sagey, E. 1985. On the representation of complex segments and their
formation in Kinyarwanda. Studies in compensatory lengthening, ed. by L.
Wetzels & E. Sezer, 251-278. Dordrecht: Foris Publications.
i
Studies in the Linguistic Sciences
Volume 20, Number 1, Spring 1990
UNDERLYING FEATURE SPECIFICATION IN YORUBA*
G. N. Clements, Cornell University
Remi Sonaiya, Obafemi Awolowo University
This paper discusses the nature of feature specification in
Standard Yoruba in the context of the theory of lexical phonology.
We postulate that Standard Yoruba has no underlying nasal
consonants in major lexical classes (nouns, verbs, and adjectives). In
these classes, nasals arise from two lexical rules, the first nasalizing
/b/ and /I/ before nasal vowels, and the second nasalizing /I/ before
l\l. Exceptions to these rules are restricted to loanwords. We further
examine a lexical constraint prohibiting /a, o/ as the first vowel in
VCV stems in which the second vowel is a high oral vowel /i, u/. On
the basis of this analysis, we show that Yoruba gives evidence for a
theory of feature specification in which distinctive feature values are
specified and redundant values are omitted (cf. Steriade 1987,
Clements 1987).
1. Introduction
This paper concerns the question of phonological feature specification in
Standard Yoruba (SY) within the context of lexical phonology. Recent work has
shown that phonological segments must be underspecified for certain features
at the level of underlying representation and for portions of the lexical and
postlexical phonology. However, theories differ as to the extent to which
representations can be underspecified. Some theories have proposed that only
the marked or non-default value of each feature appears in underlying
representations, the other value being introduced by redundancy rules (for
variants of this position see Kiparsky 1982, 1985, Archangeli 1984, Archangeli
& Pulleyblank 1986); let us call this theory Radical Underspecification. Other
theories have proposed that underlying representations contain more feature
information, and in particular that all distinctive feature values are underlyingly
present (Steriade 1987), at least in the representation of vowels at the
segmental level (Clements 1987); let us call these theories Distinctive Feature
Specification. This paper presents evidence that the vowel [i] in Yoruba must be
distinctively specified in the lexical phonology, that is, sufficiently specified for it
to be distinguished from all other vowels. While this result is strictly speaking
9 Studies in the Linguistic Sciences 20: 1 (Spring 1 990)
compatible with both theories of feature specification, it is not consistent with the
claim that [i] is totally unspecified for place features in the lexical phonology of
Yoruba, a claim which has been recently offered in support of the theory of
radical underspecification (Pulleyblank 1988).
For the purposes of this discussion, we adopt a "generic" theory of lexical
phonology in which only the following principles are crucial:
(1) a. All phonological rules are assigned to one or more (lexical or
postlexical) strata.
b. Application of lexical rules takes place before application of
postlexical rules.
c. Only lexical rules have lexically- or morphologically-defined
exceptions.
For further discussion of lexical phonology, see e.g. Mohanan (1982, 1986) and
Kiparsky (1982, 1985). A further assumption, discussed and motivated in
Clements (1987), is that phonological rules do not make crucial reference to
zero, i.e. to the absence of a feature or class of features. We state this constraint
as follows:
(2) Invisibility of Zero:
Phonological rules cannot refer explicitly to the absence of a feature
(or class node, in the sense of Clements 1985).
Within the context of a theory of distinctive feature specification such as that
assumed here, this constraint expresses the generalization that natural classes
are not defined in terms of redundant feature values. Thus, for example, in a
language in which /p/ is redundantly voiceless (due to the absence of /b/) and
sonorants are redundantly voiced, the set of all sounds in which the feature
[voiced] is redundant (namely, the set consisting of all sonorants and /p/) is not a
natural class. Constraint (2) makes it impossible for a phonological rule to
designate such a class by referring to [0 voiced]. ^
The consonant system of Yoruba is as follows (all symbols have IPA
values):
(3) a.
t
k
kp
b
f
d
s
d3
1
g
gb
m
n
I, r j w h
We follow the standard orthography in representing [kp] by the letter <p>, [j] by
<s>, [dj] by <j>, and [j] by <y>.
The vowel system is given below, with distinctive feature values for each
vowel. It will be seen that just the minimum number of feature values is given to
permit the unique identification of each vowel. For example, /O/ is the only
[-high, -low, +back] vowel in Yoruba; [-high] is required to distinguish it from /u/,
[-low] to distinguish it from /a/, and [+back] to distinguish it from /E/. All other
features are redundant. Note that /e/ occurs marginally in a few lexical items
i
Clements & Sonaiya: Underlying feature specification in Yoruba 9 1
only, /a/ is frequently realized as [o], especially after labial consonants. The
segments /E, 0/ are realized as either the [+ATR] vowels [e, o] or the [-ATR]
vowels [£, o] according to the system of morpheme-level vowel harmony
(Archangeli & Pulleyblank 1989).
(3) b. i u E O a T u a (e)
high + + - - + +
low - + +
back - + - + - +
nasal - - - + + + +
We follow the standard orthography in writing [e o] as <e, o> and [rij a] as
<in un an>, respectively, except for forms cited in slants or brackets, which are
given in phonetic characters. Note that vowels are redundantly nasalized after
nasal consonants (except for a few exceptions to be noted below) and are not
specially indicated as nasal in the standard orthography; thus <din> stands for
[di] and <mu> for [mu].
Our discussion proceeds in three parts. First, we discuss the status of the
nasal consonants [m, n] and argue that they are not underlying segments, but
derived by lexical rules from the archisegments /B, U, underspecified for
nasality (§2). We then examine alternations involving [n] and [I], and show that
the alternations are created by a lexical rule that applies in the context of the
vowel [i] (§3). Third, we discuss the consequences of this analysis for theories of
feature specification (§4). We show that the vowel [i] must be specified for
distinctive features if the full set of generalizations concerning [n] and [I] are to
be captured; further support for this conclusion is adduced from a constraint on
nominal stems.
2. Status of the nasal consonants [m, n]
For the most part, [m] occurs only before nasal vowels and [b] only before
oral vowels: mu [mti] 'to soak', bu 'to add'. This complementary distribution is
broken in the native lexicon by a small number of exceptions:
(4) a. [b] before nasal vowels:
bun [bu] 'to give' (and its derivatives)
ibon [ibo] 'gun'
ibon [ibo] 'type of disease'
obinrin [obfri] 'woman'
abon [abo] 'unripe'
b. [m] before oral vowels:^
i. verbal prefixes:
the negative imperative prefix [ma - maa ~ moo]
the habitual prefix [maa ~ moo]
ii. lexical items:
amodi 'sickness' (dialectally aboi, abori)
maalu 'cow, beef
92 Studies in the Linguistic Sciences 20:1 (Spring 1990)
These forms constitute a very small number of exceptions to what is otherwise a
strongly supported generalization in the Yoruba lexicon. To account for this
generalization, we postulate the following lexical rule (which we state
informally); "B" represents a labial stop unspecified for nasality:
(5) Nasal Spread 1 :
B V
\l
[+nasal]
Elsewhere, /B/ is realized as oral. The examples in (4a) are marked as
exceptions to the rule, and the forms in (4b) are regarded as containing
marginal occurrences of underlying /m/.
The relation between [n] and [I] is similar to that between [m] and [b],
except that in this case the complementary distribution is almost totally
complete, [n] occurs only before nasal vowels and [I] only before oral vowels: na
[na] 'to flog', Ip [b] 'to go'. The only exceptions known to us (outside obvious
loanwords) are nagudu 'Jodhpur-trousers', a possible loanword (although we
do not know the source) and naira 'Naira' (unit of national currency), a recent
coinage, not specific to Yoruba, possibly created from (English) Ni(ge)r(i)a.
Notice that Nigeria itself is often regularized to [nodgfria], which conforms to the
rule. We postulate the following rule of Nasal Spread, in which "L" represents a
lateral sonorant unspecified for nasality. This rule carries out both nasal spread
and delateralization (the latter change is required to create [n] rather than [I]).
(6) Nasal Spread 2:
L ^ [-lateral] / _ V
[+nasal]
Elsewhere, L is normally realized as oral, although postlexically, [I] may occur
as the result of a rule of sonorant nasalization (see below). (Nasal Spread 1
and 2 cannot be collapsed as the segments they apply to, /B/ and /!_/, are not a
natural class in Yoruba.)
There is a further argument for deriving [m, n] from underlying consonants
unmarked for nasality: such an analysis explains the fact that [n, m] do not occur
before mid vowels; thus, we do not find the syllables [ne, no, me, mo] in the
native lexicon. The source for these syllables would have to be /le, lo, be, bo/ in
our analysis, but as noted in (3) there are no nasalized mid vowels in Yoruba,
with the marginal exceptions of /£/ and the phonetic vowel [o] which is the
realization of /a/ after labials. 3 This gap follows from an analysis which derives
[m, n] from /B, U before nasal vowels.
Clements & Sonaiya: Underlying feature specification in Yoruba 93
There are reasons to believe that the rules of Nasal Spread described
above are lexical rules. The first evidence is that Nasal Spread 1 has a small
number of lexical exceptions, as noted in (4a). By (1c), this requires us to
consider it as a lexical rule. (Nasal Spread 2 is exceptionless, however.) The
lexical status of both rules is confirmed by the rule of gerundive reduplication.
Gerundives are formed in Yoruba by prefixing a syllable consisting of a copy of
the first consonant followed by the oral vowel [f]. Ordinarily, this vowel is oral
even if the vowel of the stem is nasal, as the examples in the right-hand column
of (7a,b) show:
(7)
dun [du]
'to be sweet'
di'-dun [didu]
'being sweet'
sin [si]
'to sneeze'
sf-sin [sfsff
'sneezing'
win [wi]
'to borrow'
wi-win [wfwff
'borrowing'
yon [p]
'to yawn'
yi'-ypn [JTjo]
'yawning'
ranti [rati]
'to remember'
ri-ranti [riratf]
'remembering'
na [na]
'to flog'
ni'-na [nfnl]
'flogging'
mo [mo]
'to know, understand'
mi-mo [mimo]
'knowledge'
We therefore consider the prefix vowel as inherently specified for [-nasal]. We
see further in (7b) that the sonorants /w j r/ are nasalized before a nasal vowel,
but remain oral before the oral vowel of the reduplicative prefix. (7c) shows that
the stem-initial consonants /m, n/ are copied as such; thus Nasal Spread 1, 2
must have already applied at the time reduplication applies. But as
reduplication is itself a lexical process, this means that Nasal Spread is also a
lexical rule. We will assume that the nasalization of the sonorants in (7b) and of
the prefix vowel in (7c) results from the operation of postlexical nasalization
rules, which spreads nasality onto sonorants before nasal vowels and onto high
vowels after nasal consonants.'*
This analysis gives us the representative derivations in (S):^
(8) du
di du
3. Alternations between [I] and [n]
We now consider a further type of relation between [I] and [n]. We find that
the syllable [li] is highly restricted in native Yoruba words, occurring only in the
reduplication of roots beginning with [I]: li-lq 'the going', li-la 'the saving', etc.,
and in a small number of roots, mostly in noninitial syllables: bggli 'long and
big', bogl'i 'roasted plantain', paal'i-paali 'hardened' (as in 'hardened
unbeliever'), paali 'shriveled', lik'i {mq) 'stay for a long time in an enclosed
place'. It might seem at first that we require a morpheme structure constraint to
express this restriction. However, an alternative explanation is suggested by the
existence of alternations between [nff and [I] involving the formatives n/''have',
n/' 'locative particle', n/'to say', n/' 'predicative particle', n/'to own,' and the prefix
m
La
underlying representation
na
lexical: Nasal Spread
wi wi
ni na
Reduplication
wi w"
nfna
postlexical nasalization rules
94 Studies in the Linguistic Sciences 20:1 (Spring 1990)
onf 'owner of. The following representative examples are taken from Folarin
(1987) and Ward (1952):
(9) [nl] ~ [I] alternations:
a. nominals formed with the prefix /oLi-/:
bata 'shoes' onfbata 'owner of shoes'
ija 'fight' onfja 'someone who loves to fight'
eso 'fruit' eleso 'owner of fruit'
Oku 'dead person' oloku 'owner of dead person'
omo 'child' olomo 'owner of a child, parent'
asp 'clothes' alaso 'owner of clothes'
itoju 'care' olutoju ~ onitoju 'caretaker'
b. sentences with n/'to have':
mo nf aso ~ mo I'asp 'I have clothes'
ni ese ~ o I'ese 'you have legs'
6 ni ewe ~ 6 I'ewe 'he has leaves'
c. sentences with /c/'-n/ 'what':
ki-ni n ta ~ ki-l'o n ta 'what is it you are selling?'
ki-ni omode na fe ~ kf-l'omode na fe 'what is it the child wants?'
ki-ni eyii ~ ki-l'eyif 'what is this?'
As these examples show, [n] appears if the following vowel (after the application
of a rule of vowel elision which deletes the first of two vowels in a sequence) is
[i], and otherwise [I] appears.
We have already established that [n] is not an underlying phoneme of
Yoruba. Since [n] alternates with [I] in (9), it would at first seem reasonable to
derive it from underlying /Li/ as we have done earlier. However, evidence that
the vowel must be underlyingly oral comes from the rule of Vowel Deletion,
which deletes the first of two vowels in sequence (Pulleyblank, Forthcoming;
Sonaiya, Forthcoming). When this rule applies to sequences of nasal vowels
followed by oral vowels, nasality is transferred to the second vowel, as in
examples like din [di] + ejo [edgo] realized in combination as denjo [dSdjo] 'fry a
snake' (cf. Pulleyblank 1988:251). When it applies to /oLf + aso/ 'owner of
clothes', on the other hand, it leaves an oral vowel: [aiaso], not *[alaso]. If we
assumed a base form of /\V, we would have to postulate a special rule of
denasalization to account for the oral vowel (note that denasalization cannot be
considered an effect of structure-preservation since nasal vowels occur
underlyingly). But what about the surface realization of this vowel? There is
some unclarity regarding its phonetic status in the literature. Abraham
(1958:xxii) reports that the vowel in n/' 'possess' is normally oral; however. Ward
finds that the vowel of n/'be' has "considerable nasality" (13). Our observation,
based on a survey of several speakers Obafemi Awolowo University, is that it is
very difficult, if not impossible to get an oral vowel in these forms: the vowel is
nasalized in both n/' 'possess' and ni 'be', as far as we can tell. We explain this
as an effect of the rule of postlexical nasalization, independently required as
shown in (8) above to account for the nasal vowels in reduplication.
Clements & Sonaiya: Underlying feature specification in Yoruba 9 5
We propose, then, that the underlying representation of surface [nf] is /Li/,
and that [L] is assigned nasality by the following rule:^
(10) L Nasalization (lexical, level 1):
L -> [-lateral, +nasal] / [+high,-back]
Instances of lU not undergoing this rule or that of Nasal Spread 2 are realized
as the oral lateral [I]. L Nasalization must be a lexical rule, since as noted, it has
lexical exceptions. We further assume that it is a level 1 rule, since it does not
apply to the output of the productive rule of gerundive reduplication in forms like
li-lo, which we have assigned to level 2 (see note 5).
We now come to the crux of any analysis of Yoruba nasality. As a lexical
rule, L Nasalization should change /Li/ to [ni] before the operation of phrase-
level phonological rules. But then we appear to have no way of accounting for
the reduced form [I] appearing in examples such as mo I'asg 'I have clothes' in
(9b,c) in which [I] replaces the expected [n]. In such examples, the alternation
between [I] and [n] is determined by the identity of the immediately following
vowel only after the application of Vowel Deletion at the phrase level. But if L
Nasalization is a lexical rule, it must change underlying /Li/ to [ni] before the
phrase level rules apply, and we have no way of changing [n] back to [I].
In order to account for this apparent paradox, we propose that the [I] in
forms life mo I'asq represents a suppletive morphological alternant /!_/ which
replaces [ni] in the postlexical phonology just in case it is followed by a nonhigh
vowel in the next word. If this condition is not satisfied (that is, if the next word
begins with [i]; note that [u] never occurs in word-initial position), [n] remains in
place. Thus the lexical entry of a form like ni'Xo have' contains the information
that beside its regular phonological shape /Li/, the suppletive alternant l\J is
available for substitution in the syntax if the appropriate phono-syntactic frame
is satisfied; this alternant is a lexical property of ni'Xo have' and of those of its
homophones that show the same pattern of alternation. Interestingly, there is a
further set of homophones that do not show this pattern of alternation, including
ni'Xo make uncomfortable', n/''to come to the aid of, en/' 'mat', and en/ 'person';
these always retain [n] whatever the following vowel (Oyelaran 1986). Thus, for
example, depending on whether the vowel elision rule applies or not, we find
olu ni qga re lara or olu n'qga re lara 'Olu bugs his boss', but in neither case is
I'oga possible. Whether a given morpheme of underlying form /Li/ has the
suppletive alternant [L] or not is an unpredictable, idiosyncratic fact about the
morpheme.''
Many languages give evidence of syntactically-conditioned lexical
alternations of a very similar type. In French, for example, we find two lexical
variants of certain high-frequency adjectives and determiners, one selected just
in case a vowel follows in the next word. Thus the masculine forms beau 'pretty',
vieux 'old' are replaced by bel, vieil before vowel-initial (masculine) nouns, as
in bel enfant, vieil homme. These morphological alternants cannot be derived
by a postlexical phonological rule, given the assumptions in (1), since the rule
would have to be sensitive to the lexical identity of the items in question.
96 Studies in the Linguistic Sciences 20:1 (Spring 1990)
Instead, both alternants are listed in the lexicon, and the appropriate surface
allomorph is selected depending on the phonosyntactic context. ^ Whether a
given determiner has a suppletive alternant or not is an idiosyncratic property of
that morpheme; thus joli 'pretty' has the same shape whether a consonant or
vowel follows. The alternation pattern of /Li/ in Yoruba seems quite parallel.
Hayes (1988), in a review of similar patterns of syntactically-determined
alternations in other languages, concludes that syntactically-conditioned
allomorphs may be generated in the lexicon by phonological rules sensitive to
certain lexically-specified syntactic frames. The process of lexical insertion
checks to see whether any such syntactic frame is satisfied in the particular
structure into which substitution takes place: If it is, the lexically-generated
allomorph is substituted, and otherwise, the unconditioned allomorph is
selected. Applying this model to Yoruba, we would say that the allomorph [I] is
lexically precompiled by the rule [Li] -^ [L], the latter allomorph being marked for
substitution in the frame [ [-high]]. Both forms undergo parallel lexical
derivations, in particular receiving the feature [-nasal] by Nasal Spread 2. The
syntactically-conditioned allomorph is then available for postlexical substitution
if its context satisfies the syntactic frame.
A lexical analysis of this sort seems clearly superior to a purely
phonological analysis in which a phrase-level rule of N Denasalization is
postulated to change derived [n] to [I] before nonhigh vowels following vowel
deletion. Although this would be descriptively adequate, N Denasalization
would be the synchronic inversion of the rule of L Nasalization that created the
[n] in the first place. Furthermore, though this rule would have to apply
postlexically to account for the examples in (9b, c), the forms ni 'to make
uncomfortable', n/''to come to the aid of, en/ 'mat', and en/ 'person' would now
constitute lexical exceptions to it, in violation of principle (1c); in the lexical
analysis, in contrast, these forms are simply marked as exceptions to the lexical
allomorphy rule. Since a model of lexical phonology must allow for phrasally-
determined suppletion in any case, the lexical analysis seems to allow a more
principled analysis in this case.
To summarize the discussion of this section, the proposed analysis
explains both the near-absence of the syllable [li] in the native lexicon, and the
[n] ~ [I] alternation, which is created by the rule of L Nasalization (10) and (in the
phrase-level phonology) by the rule of suppletion. Thus the analysis of [ni] ~ [I]
alternations does not provide an argument for underlying /n/, but on the contrary
shows that in these cases, too, surface [n] derives from underlying /L/,
consistently with our earlier analysis.
The evidence from loanwords is consistent with our analysis. We assume,
following Mohanan 1982, that the level of lexical representation (defined as the
output of the lexical rules) has a psychological status similar to that of the
phoneme level in classical phoneme theory. Thus we expect postlexical rules
(like allophonic rules) to be highly productive, applying freely to novel
combinations created in the syntax as well as to loanwords, new coinages,
speech errors, and the like. On the other hand, lexical rules (like
Clements & Sonaiya: Underlying feature specification in Yoruba 9 7
morphophonemic rules) are generally less productive, and their productivity
depends on factors like rule opacity, number of exceptions, and the like. We
thus expect loanwords to undergo postlexical rules quite regularly, but to be
more resistant to lexical rules.
This is just what we find in Yoruba. Thus the lexical rules of Nasal Spread
1, 2 and L Nasalization have lexical exceptions in the loanword vocabulary, as
the following examples show (Amadou 1989):
(11) a. [b] before nasal vowel:
Bi'ntu [bitu] (female name) < Hausa < Arabic 'daughter'
bandi'ri 'boundary'
bongalo 'bungalow'
alubansi 'advance'
[1] before nasal vowel:
balonon [balool
'ball'
< French ballon
[l£5]
'bedlinen'
< French linge
ilenti'ri'ki
'electric'
< English
sflenda
'slender
< English
[1] before [i]:
III]
'lily'
All
'AM'
olide
'holiday'
wuli'ni
'woolen'
telifoonu
'telephone'
Further instances of syllable [li] are created by a rule of epenthesis applying
after syllable-final consonants:
(12)
kalikiili
'calculation'
< Fr. calcul
baali
'ball'
< Fr. bal
alifabeeti
'alphabet'
beliiti
'belt'
peeli
'pail'
mfliki
'milk'
(In contrast, as far as we have been able to determine, the postlexical rules of
nasalization mentioned in connection with (8) have no lexical exceptions). In
other languages, too, loanwords frequently constitute exceptions to lexical
rules. Thus in Mixtec, the underspecified lateral /L/ is generally realized as [n]
before nasal vowels due to the operation of a lexical rule of Nasal Spread
similar to the one of Yoruba, and is othenwise realized as [I] (Marlett 1989).
Loanwords present exceptions, however: thus the [I] in [malfj (< Sp. comadre )
fails to undergo the rule.
In our analysis, [m] and [n] must have phonemic status in loanwords
wherever they precede the oral mid vowels [e,o], since our rules cannot derive
98 Studies in the Linguistic Sciences 20:1 (Spring 1990)
them from underlying /B, U in this context. Thus we find a small number of
examples like the following (Amadou 1989):
(13)
penee
•tire*
< Fr. pneu
mesie
'sir'
< Fr. monsieur
emee
'love'
< Fr. aimer
serimonl
'ceremony'
<Eng.
sfnetp
'senator'
< Eng.
Moosiisi
'Moses'
<Eng.
This behavior is also not unexpected, as loanwords commonly introduce new
phonemes, especially when they are identical to the surface realizations of
native phonemes or phoneme sequences. In English, for example, [t)] can
normally be derived from underlying /ng/ in the native lexicon, but has
phonemic status in loanwords such as dinghy (from Hindi), hangar (from
French), gingham (Dutch, from Malay), orangutan (Malay), and otherwise
occurs phonemically only in onomatopoeic words {ting-a-ling) or foreign proper
names (Singapore). Such examples suggest that /t)/ is slowly entering English
as a phoneme, but still remains outside the native "core." The nasals /m,n/ in
Yoruba appear to have the same status.
4. Theoretical considerations
We now consider how the analysis presented above bears on the theory
of feature specification. In Pulleyblank's analysis of Yoruba (1988), [i] is the
"default" vowel in the sense that it is uncharacterized for any place features in
underlying representation. The default values of [i] are filled in late in the
postlexical derivation. Therefore [i] remains featureless (as far as place features
are concerned) throughout the lexical derivation, and much of the postlexical
derivation.
As we have seen, however, the rule L Nasalization requires crucial
reference to [i] in the lexical phonology. Under Pulleyblank's analysis, this rule
must be able to identify the vowel [i] at a stage where it has no place features,
and is therefore nondistinct from all other vowels. The problem for radical
underspecification theory is obvious: the only way to do this is to make explicit
reference to the fact that [i] has no place features, but this is prohibited, as we
have seen, by the Invisibility of Zero Principle (2). This principle is required in a
theory of radical underspecification to express the generalization that natural
classes of sounds cannot be designated by the set of redundant and default
values of a given feature. Without it, rules could designate unnatural classes of
sounds by referring to all missing values of [voiced], both redundant and default,
and could thus refer, for example, to the set of (distinctively) voiceless
obstruents and (redundantly) voiced sonorants in a language, while excluding
voiced obstruents, through the mention of [Ovoiced]. If we were to allow classes
of sounds to be defined by missing values in this way, we would make false
predictions about the natural classes that occur in the rules of the world's
languages.
Clements & Sonaiya: Underlying feature specification in Yoruba 99
Further evidence that lexical rules must be able to refer to [i] comes from
the study of a constraint on VCV nominal stems. In such stems, if the first vowel
is [a] or [o], the second cannot be one of the oral vowels [I] or [u] (Fresco 1970,
Oyelaran 1973). We state is informally as follows:
(14) *[{a,o}C{i,u}]N (level 1)
This constraint is exceptionless in roots beginning with [a], and has only one
exception in roots and stems beginning with [o]. (14) states it as a constraint on
the output of the level 1 phonology and morphology, where it has only the
following exceptions, as far as we have been able to determine:
(15) a. roots:
pti 'spirits'
b. derived stems:
a-du 'one or something that is black'
a-tu 'that which easily falls apart; soft type of yam'
a-df ~ a-din 'oil from palm kernel'
a-bi 'placenta'
c. loanword:
Alf (male name)
This constraint cannot be explained in terms of phoneme frequency, since /a, o,
i, u/ are all high-frequency vowels in Yoruba, and their random combination
could be expected to yield many examples of the prohibited sequence. As an
indication of this, we give in (16) a few of the many examples of well-formed
VCV nouns in which the second vowel is a nasal vowel (recall that in the ortho-
graphy, <mV> and <nV> represent nasals followed by nasal vowels). ^
(16) ami 'sign'
ami 'scout, spy'
amu 'large waterpot'
arun 'five'
ahun ~ awun 'miser; tortoise'
akin 'manly fellow'
a-win 'buying on credit'
Similarly, violations are common in forms other than nominals. All verbs are
consonant-initial, and therefore cannot present violations of (14) in principle, but
(17) gives representative examples from other parts of speech:
(17) abi 'or'
abi 'that which possesses'
afl 'except, unless, only'
atl 'and'
anf 'even, in like manner'
This constraint must make reference to the features [+high] and [-nasal] in
order to designate the natural class of high oral vowels {i,u}. Thus, we must
100 Studies in the Linguistic Sciences 20:1 (Spring 1990)
assume that [i] bears the feature [+high] at the point where the constraint
applies, that is, at level 1 .
5. Conclusion
We conclude that lexical rules in Yoruba must be able to refer to the
following classes of vowels, among others:
(18) {i}: [+high, -back]
{ i, u } : [+high, -nasal]
This shows that [i] cannot be the default vowel in the lexical phonology. But if [i]
is not the default vowel, what is? It cannot be [u], since [u] is subject to constraint
(14), and to the further constraint that words cannot begin with [u] (Fresco 1970).
It cannot be [a] or [o], due again to constraint (14). It cannot be one of the mid
vowels or [a], since these vowels are crucially characterized as [-high] for the
purposes of vowel harmony (Archangeli & Pulleyblank 1989). We are led to the
conclusion that all vowels bear at least those features that distinguish them from
other vowels in the lexical phonology — which is just the claim made by the
theory of Distinctive Feature Specification. ^o
NOTES
* We thank Yetunde Laniran for commenting on an earlier draft of this
paper, and Wahab Amadou, Yetunde Folarin, and Olasope Oyelaran for making
their unpublished work available to us. We are also grateful to two anonymous
SLS reviewers for their comments on this version of the paper. The first author's
research was supported in part by grant no. INT-8807437 from the National
Science Foundation.
1 In the discussion following the oral presentation of this paper, Will Leben
raised the question whether (2) might be too strong, in view of the fact that
certain types of phonological rules appear to require a distinction between
segments bearing a certain feature and segments not bearing that feature. The
most familiar rules of this type are those which spread a given feature onto
neighboring segments which do not bear a specification for the feature in
question. For example, tone spreading rules are often restricted to apply only to
toneless vowels, and rules of vowel harmony are typically restricted to apply
only to underspecified vowels. We would argue, however, that these examples
do not require that we abandon principle (2), given the independent proposal of
Archangeli & Pulleyblank 1986 that spreading of a feature F onto segments not
bearing F represents the default, or unmarked case of spreading rules, and may
be considered the unmarked value of a spreading rule parameter. If we accept
this position, then it is not necessary to stipulate in the formal statement of a
spreading rule that its target does not bear the spreading feature, and such
rules will not violate principle (2). Similarly, default rules, which introduce
default values of features onto segments not yet specified for those features, are
Clements & Sonaiya: Underlying feature specification in Yoruba 101
generally thought to be constrained by the Strict Cycle Condition, which
prevents the feature-changing application of (cyclic) lexical rules (Kiparsky
1982), so that again it need not be stipulated that the target vowel is
underspecified for the feature in question. We are aware of only a few proposed
rules referring crucially to unspecified feature values that cannot be accounted
for in one of these two ways, and alternative analyses seem to be available in
most or all of these cases. It seems, then, that (2) represents a strong and well-
supported constraint on rules that we would not want to abandon unless forced
to by strong and uncontrovertible evidence.
2 Other apparent counterexamples of this type can be analyzed as
complex forms in which [m] precedes a nasal vowel. Numerals such as meji
'two', meta 'three' can be analyzed into a prefix and a root, cf. eyV 'two', eta
'three' (Fresco 1970). The first person singular prefixes mo, ma in forms like mo
fe Ip 'I want to go', ma a Ig 'I will go' can similarly be derived from the syllable
[mi] followed by the subject prefix [6] and the future tense prefix [a], respectively.
Proper names such as Monilpla, Malqmq, Oladimeji are sentential in structure,
and contain mo-, ma-, meji as components.
3 As our analysis also predicts, the labial nasal [m] occurs freely before [o]
representing /a/: mo 'to know,' mo 'to build', omo 'child', somq 'sky'.
'^ These two rules cannot be collapsed into a single, bidirectional rule of
Sonorant Nasalization due to the rare forms in which nasals precede nonhigh
oral vowels in surface representation (see the examples in (4b) and (13)). We
thank Moira Yip for calling this point to our attention.
5 We assume that the productive rule of gerundive reduplication is a level
2 rule, along with certain compounding rules (for the latter, see Folarin 1987).
Among other things, this explains why gerundive reduplications, unlike other
reduplications, cannot be prefixed, with the isolated exception of dtito'truW,
from to 'be correct', which we treat as a lexicalized form.
6 There is a further reason for stating the rule as one nasalizing [L] before
[i] instead of one oralizing [n] before [u e o e o a]: the latter rule is not formulated
on a natural class of vowels. For discussion of variants like [olutoju - onitoju],
showing that [I] appears before the stem vowel [u], see Folarin (1987). Of
course, if (contrary to our hypothesis) [i] were underlyingly featureless, and thus
lacked a place node, the rule could oralize [n] before vowels with place nodes;
however, we know of no precedent for rules which crucially mention class
nodes as a contextual element (as opposed to an affected segment), and
believe that phonological theory may be able to exclude such rules in principle.
'' We further assume that the alternant /L/ goes through a parallel
morphological derivation of its own, acquiring a reduplicated form which may be
substituted in the same way if it comes to precede a nonlow vowel in the
phrase-level part of the derivation; this will account for the reduced reduplicated
form liTowd 'having money' beside the unreduced ni'ni 'having' (Pulleyblank
1988:266).
102 Studies in the Linguistic Sciences 20:1 (Spring 1990)
8 The substitution takes place not only postlexically, but also lexically, as is
shown by derivatives like bellot 'bonny', embellir 'to make pretty', vieillot
'antiquated', and vieillir 'to age'. The substitution also takes place in the
feminine forms belle, vieille, which (historically, at least) involved a final suffix
vowel [-a].
9 These forms provide further support for an analysis in which surface
nasals are derived from oral consonants followed by nasal vowels. If instead,
nasal consonants were followed by oral vowels in underlying representation,
we could not explain why forms like ami etc. are exceptions to constraint (14).
1° We emphasize that our conclusion is not inconsistent with the theory of
Radical Underspecification, which does not predict that every language has a
default vowel or consonant throughout the lexical phonology but only requires
that rules assigning default values of a feature must be assigned to the first
stratum in which reference is made to that value, in the limit case stratum 1
(Archangeli & Pulleyblank 1989). The facts discussed here can be
accommodated on the assumption that /i/ is partly specified as early as level 1
in accordance with this principle, and do not require us to abandon or modify
Radical Underspecification theory. However, they do show that Pulleyblank's
proposal that [i] is unspecified for place features in all lexical strata in Yoruba is
inconsistent with a fuller range of observations in this language. Thus Yoruba
does not provide crucial evidence for Radical Underspecification theory, and
the evidence for the special status of /i/ in Yoruba must receive another
explanation.
REFERENCES
Amadou, W. 1989. "L'integration linguistique des emprunts," chapter of
University of Paris Ph.D. dissertation in Linguistics, in preparation.
Abraham, R. C. 1958. Dictionary of Modern Yoruba. London: University of
London Press.
Archangeli, D. 1984. Underspecification in Yawelmani phonology and
morphology. MIT Ph.D. dissertation in Linguistics.
, & D. Pulleyblank. 1986. The content and structure of phonological
representations. University of Arizona and University of Ottawa, MS.
, & D. Pulleyblank. 1989. Yoruba Vowel Harmony. Linguistic Inquiry
20:173-217.
Clements, G. N. 1985. The geometry of phonological features. Phonology
Yearbook 2:225-52.
. 1987. Towards a substantive theory of feature specification. NELS
18:79-93.
FOLARIN, A. Y. 1987. Lexical phonology of Yoruba nouns and verbs. University
of Kansas Ph.D. dissertation in Linguistics.
Clements & Sonaiya: Underlying feature specification in Yoruba 1 03
Fresco, E. 1970. Topics in Yoruba Dialect Phonology. Studies in African
Linguistics, supplement 1. Department of Linguistics, UCLA, Los Angeles.
Hayes, B. 1988. Precompiled Phrasal Phonology. UCLA, MS. To appear in
CSLI workshop on the syntax/phonology connection, ed. by S. Inkelas &
D. Zee. CSLI, Stanford, CA.
KiPARSKY, P. 1982. From Cyclic Phonology to Lexical Phonology. The Structure
of Phonological Representations, part 1, ed. by H. van der Hulst & N.
Smith, 131-75. Foris Publishing, Dordrecht.
. 1985. Some consequences of Lexical Phonology. Phonology Yearbook
2:85-138.
Marlett, S. a. 1989. Nasalization in Mixtec. S.I.L. and University of North
Dakota, MS.
MOHANAN, K. P. 1982. Lexical Phonology. MIT Ph.D. dissertation in Linguistics.
[Revised version published 1986 by Reidel Publishing, Dordrecht.]
Oyelaran, O. O. 1973. Yoruba vowel co-occurrence restrictions. Studies in
African Linguistics 4:155-182.
. 1986. The alveolar /n/ in Yoruba: A bad dream which refuses to
dissipate. Paper presented to the 17th West African Languages Congress,
University of Ibadan, Ibadan, Nigeria.
PULLEYBLANK, D. 1988. Vocalic Underspecification in Yoruba. Linguistic Inquiry
19:2.233-70.
SONAIYA, R. Forthcoming. Vowel contraction in Yoruba. Dept. of Foreign
Languages, Obafemi Awolowo University, lle-lfe, Nigeria, MS. To appear
in Linguistique africaine.
Steriade, D. 1987. Redundant values. Parasession on Autosegmental and
Metrical Phonology (CLS 23.2), Chicago Linguistic Society, Chicago, ed.
by A. Bosch.
Ward, I. C. 1952. An introduction to the Yoruba language. Cambridge
University Press.
i
i
Studies in the Linguistic Sciences
Volume 20, Number 1, Spring 1990
REDUPLICATION AND PROSODIC CONSTITUENTS IN WOLOF
Omar Ka
University of Maryland
A straightforward account of reduplication processes in Wolof —
a Northern West Atlantic language spoken primarily in Senegambia
— has been to consider those processes as always involving the
copying of an entire morpheme (cf. Sauvageot 1965, Diagne 1971,
Ka 1981), or the addition of a morpheme skeleton to a stem (cf. Ka
1988). This account however, did not examine language-external
data such as those provided by linguistic games. In this paper, I
attempt to show that, in fact, Wolof reduplication also involves the
copying of prosodic constituents such as the syllable, the foot, and
the prosodic word. I stress the central role played by those
constituents in the understanding of the prosody of the language.
Part of the data are drawn from different varieties of a Wolof secret
code called Kail.
1. Background
In this section, I provide some relevant information concerning the syllable
structure of the stems, the representation of complex segments and the types of
reduplication found in the language.
1.1 The syllable structure of the stems
Stems susceptible of reduplication show the following surface syllable
types:
— monosyllabic stems:
C V V
M/
a
dee 'to die'
C V
gis 'to see*
men 'can'
Kow 'hinterland'
C V V
a
gaafi 'to hurt'
tooy 'to be wet"
Siin (Senegalese province)
V c
begg 'to want'
lakk 'to burn'
wicc 'to shake head'
106 Studies in the Linguistic Sciences 20:1 (Spring 1990)
— disyllabic stems:^
e. C V C V C Jolof (Senegalese province)
\/ \|/^ Bawol (Senegalese province)
o c patam (patam) 'hasty work'
lambar (lambar) 'useless agitation'
f. C V V C V C Saalum (Senegalese province)
\|/^ XL/' raakam (raakam) 'back and forth walk'
o o xaatar (xaatar) 'angry walk'
g. C V C C V C takkam (takkam) 'formula to propose
\|/ \J/^ ^ guessing game'
a a nokkos (nokkos) 'slow and hesitating
walk'
h. C V C V V C
\y ^^"""^/^ Kajoor (Senegalese province)
a o
1. C V C V
Sy Sy poto (poto) 'muddy location'
a a
j. C V V C V Waalo (Senegalese province)
\|/^ \/ baana (baana) 'street peddler'
o a taaba (taaba) 'to live on an
occasional income'
As the data show, reduplicated stems belong to a variety of syllable types;
only C V and C V C V V stems cannot be reduplicated. The data also reveal the
existence of complex segments, such as long vowels, geminated consonants
and prenasal consonants. I deal below with the representation of those
segments.
1.2 Complex segments
The Wolof phonological system includes simple segments such as short
vowels and simple consonants, and complex segments such as long vowels,
geminated and prenasal consonants (cf. Ka 1988).
The phonological system of consonants is the following:2
(1) Simples p t c k
b d j g
m n n h
f s X
V w
prenasals mp nt nc nk nq
mb nd nj ng
geminates pp tt cc kk qq
bb dd jj gg
mm nn fin hh
II yy WW
I
i
Ka: Reduplication and prosodic constituents in Wolof 107
The system of short vowels is depicted below:
(2) front central back
high i u
mid e e 6
e
low a
Only e (/9/) has no corresponding long counterpart, except in borrowings.
In accordance with the Obligatory Contour Principle, which prohibits
sequences of adjacent identical elements, geminate consonants and long
vowels will be treated as simple elements of the segmental tier that are
associated to two consecutive slots on the CV-tier:
(3) a. X b. Y
A A
C C V V
Stems such as gaan, begg, nokkos, and /<ayoor will be represented as in (4):
(4) a. beg n o k o s
II N I I A II
CVGC CVCCVC
b. g a n k a j o r
I A I I I I A I
CVVC GVCVVC
Prenasal consonants will be analyzed as sequences of a nasal and a stop
segment mapped to a single C slot: /mb/ for instance, will have the following
representation:
(5) m b
V
C
A stem such as lambarvj\\\ be represented as in (6):
(6) I a m b a r
II V I I
C V C V c
A number of compelling arguments can be given to support the above
representations of complex segments; for a detailed discussion, I refer the
reader to Ka (1988).
108 Studies in the Linguistic Sciences 20:1 (Spring 1990)
1.3 Types of reduplication
Two types of reduplication exist in Wolof: one that is dubbed "ordinary"
reduplication in this study, and another one that occurs within the secret code
Kall.3
Ordinary reduplication always involves the copying of an entire morpheme
of a word. There is no copying of parts of a morpheme, such as a phoneme, a
syllable, or a metrical foot. It is thus possible to characterize ordinary
reduplication as a case of total reduplication. "* In contrast, the secret code Kail
involves the copying of prosodic constituents such as the syllable, the foot, or
the prosodic word.
In the next sections, I provide an analysis of each type of reduplication.
Section 2 describes ordinary reduplication in terms of the affixation of a
morphemic skeleton to a stem. Section 3 analyzes the different varieties of Kail
as cases of transposition and reduplication of specific prosodic constituents.
The definition of those constituents entails the examination of the syllabification
principles and the stress algorithm of the language. I make several references
to McCarthy and Prince (1986, 1987) and Ka (1988, Forthcoming).
2. Reduplication of a morpheme
In Wolof, "ordinary" reduplication is a derivational process, in that it is used
to derive nouns from verb, noun or ideophonic stems; it can also be combined
with suffixation to form derived verbs. Let us examine the data.
2.1 The data
2.1.1 Derived nouns
To form a derived noun, the entire stem is always copied; it is a verb, a
noun, or an ideophone.
(7)
gis-gis
'vision'
(gis 'to see')
begg-begg
'desire'
(begg 'to want')
lakk-lakk
'burn'
(lakk 'to burn')
gaah-gaah
'wound'
(gaan 'to hurt')
tooy-tooy
'wet spot, wetness'
(tooy 'to be wet')
tang-tang
'heat'
(tang 'to be hot')
gakk-gakk
'stain'
(gakk 'to be stained")
ben-ben
'hole'
(benn 'to pierce')
deg-deg
'understanding'
(degg 'to hear')
(In the last two examples only, degemination will take place before
reduplication in stem-final position. This phenomenon is no longer productive,
and appears to have been morphologized in the language.)
Ka: Reduplication and prosodic constituents in Wolof
109
The noun stem refers to a region, a city or an ethnic group. The
reduplicated from has the meaning 'inhabitant of, originating from' that region,
city or ethnic group.
(8) Pel-Pel
Kow-Kow
Ndar-Ndar
Siin-Siin
Bawol-Bawol
Jolof-Jolof
Saalum-Saalum
Waalo-Waalo
'race of sheep'
'inhabitant of the hinterland'
'inhabitant of Ndar'
'inhabitant of Siin'
'inhabitant of Bawol'
'inhabitant of Jolof
'inhabitant of Saalum'
'inhabitant of Waalo'
(Pel 'ethnic group
of herders')
(Kow 'hinterland')
(Ndar 'Saint-Louis'
(Siin)
(Bawol)
(Jolof)
(Saalum)
(Waalo)
In contrast with the verb and noun stems, ideophonic stems never appear
in isolation, they are always copied.
(9)
rah-rah
'loud metallic noise'
(*rah)
nes-nes
'brightness'
(*nes)
yopp-yopp
'cross-country run'
(*yopp)
nukk-nukk
'short steps run'
(*nukk)
waah-waah
'back and forth walking
a woman in labor'
of
(*waah)
yoor-yoor
'morning'
(*yoor)
takkam-takkam
'formula to propose a guessing game'
(*takkam)
lambar-lambar
'useless agitation'
(*lambar)
The data in (7) through (9) show a single process of stem copying.
However, that copying may be combined with suffixation to form derived verbs,
as is shown in the next section.
2.1.2 Derived verbs
To form derived verbs, the entire stem is copied before the addition of the
following suffixes:
â– al
'intensive'
â– aat
'iterative'
â– le
'participant'
â– lu
'benefactive-reflexive'
â– i
'verbalizing'
The reduplicated stem is either a verb (A) or an ideophone (B). Here too, the
ideophonic stem is never found in isolation, it is always reduplicated.
wis-wisal
toq-toqal
wax-waxaat
bey-beyaat
xar-xarle
far-farle
baax-baaxlu
dee-deelu
Studies in the Linguistic Sciences 20:1 (Spring 1990)
(10) wis-wisal 'to rain lightly and continuously' (wis 'to rain lightly')
'to drop slowly and continuously" (toq 'to drop one
by one')
'to repeat again and again' (wax 'to say')
'to cultivate repeatedly' (bey 'to cultivate')
'to agonize' (xar 'to break')
'to take side with passion' (far 'to be allied to'
'to pretend to be good' (baax "to be good')
'to simulate being dead' (dee 'to die')
(11) tes-tesi 'to be very active' (*tes)
kuus-kuusi 'to try with great effort' (*kuus)
hakk-hakki 'to laugh uncontrollably' (*hakk)
yeger-yegeri 'to walk heavily' (yeger)
nokkos-nokkosi 'to walk slowly and with hesitation' (*nokkos)
neb-nebal 'to fix grossly' (*neb)
tucc-tuccal 'to smash' (*tucc)
2.2 The analysis
From the data in (7) through (11), it is clear that the stems are always
copied, regardless of their phonemic makeup or syllable structure. Inf fact, the
copied constituent borrows its phonemic melody, its CV structure and its syllabic
structure from the stem to which it is attached. Therefore, we will conceive of
ordinary reduplication in Wolof as involving the prefixation of a morphemic
skeleton. That skeleton comprises a phonemic melody, a CV skeleton and a
syllabic skeleton borrowed from the stem, as is shown in (12).
(12) [ ] [ ] [ ] [ ] phonemic melody
I I I I
C V V C CV skeleton
o syllabic skeleton
[i morphemic skeleton
This structure will yield the following representations:
(13) sin s i n s i n
I A I I A I
cvvc cvvc
a
1 I
^ + [i
(14) I a k I a k I a k
â– â– â– I I A I I A
C V CC C V CC
1 I
^ + ^
i
Ka: Reduplication and prosodic constituents in Wolof 111
(15) lambar lambar lambar
IIVII IIVII IIVII
CVCVC -> CVCVC CVCVC
a (5 o
I I I
^ + ^ ^l + ^i
So far, the analysis of reduplication in Wolof seems to be straightforward,
since it always involves a full copying of a stem. However, the purpose of this
paper Is to show that in fact Wolof reduplication also involves the copying of
constituents of a prosodic nature: the syllable, the foot, or the prosodic word. Let
us consider the language-external evidence coming from the secret code Kail.
3. Reduplication of a prosodic constituent
Specific prosodic constituents are reduplicated in Kail. This name refers
to a secret code used for various purposes, all related to a need for
unintelligibility (cf. Ka to appear). Kail is still routinely used in the areas of
Silmaxa and Ceneba, in the Kajoor-Bawol region of Senegal. Illustrative data^
are provided below.
3.1 The data
Let us first examine the data in their "original" (I.e., non secret) form.
(16) a. jaay ma yapp^
sell 3 ps meat
obj. pro.
'sell me (some) meat'
b. ma yobbu ko sama ker
1 ps bring 3 ps 1 ps home
subj. pro. obj. pro. poss.
'I bring it (to) my home'
c. jox ko sama jabar
give 3 ps 1 ps wife
obj. pro. poss.
'give it (to) my wife'
d. mu toggu ko
1 ps cook 3 ps
subj. pro. obj. pro.
'she cooks it'
e. samay doom lekk
1 ps-plural child eat
poss.
'my children eat'
112 Studies in the Linguistic Sciences 20: 1 (Spring 1990)
Before comparing these data with the corresponding "secret" forms, I shall
review the Wolof syllabification principles and the prosodic constituents that will
be relevant to the analysis.
3.2 Syllabification principles
In CV terms (cf. Clements & Keyser 1983, Ka 1988), the syllable structure
of Wolof lexical items obeys the following syllabification principles:
(17) a. The syllable peak may consist of a short vowel V or a long vowel
VV;
b. Each syllable begins with a consonant, hence the syllable left
margin is an obligatory constituent: it consists of either a simple
consonant or a prenasal;
c. The syllable right margin is an optional constituent (if it is
present, it may consist of a simple consonant, a geminate or a
prenasal), hence the "minimal" syllable is either CV or CVV;
d. Neither the LM nor the RM may consist of a sequence of
consonants that is not a prenasal or a geminate structure.
3.3 Relevant prosodic constituents
In Kail, the following prosodic constituents are reduplicated:
(18) Wd "prosodic word"
F "foot"
a "syllable"
Oc "core syllable"
A prosodic word consists of a lexical item and an optional clitic^ (cf. Ka
1988). If the clitic precedes, it constitutes a separate prosodic word. The
bracketing of the data in (16a) - (16e) will be as in (19a) - (19e).
(19) a. [jaay ma] [yapp]
b. [ma] [yobbu ko] [sama] [ker]
c. Oox ko] [sama] Oabar]
d. [mu] [toggu ko]
e. [samay] [doom] [lekk]
Ka: Reduplication and prosodic constituents in Wolof 113
In (19a) - (19e), the lexical items jaay, yobbu, jox, toggu form prosodic words
with the clitics that follow them, i.e., the object pronouns ma and ko. However, in
(19a) - (19e), the subject pronoms ma, mu, and the possessives sama, samay
precede lexical items, and thus will form separate prosodic words.
The metrical foot is another prosodic constituent that will be relevant to the
discussion. Following the analysis of stress given in Ka (1988), I will
characterize the foot in Wolof as binary, left-dominant:
(20) F
The core syllable Oc is equated in McCarthy & Prince (1986) to the
"minimal" syllable in a language. Recall the Wolof syllabification principle in
(17c) which stated the optionality of the syllable right margin: the "minimal"
syllable is then CV or CVV, and will correspond to Oc .
Let us now turn to the analysis of the different varieties of Kail.
3.4 Types of Kail
Two major processes are involved in the different varieties of Kail: one in
which a syllable is transposed within a particular domain, and one requiring the
copying of a designated prosodic constituent. I will call the first process True
Transposition, and the second one Reduplication (cf. Ka Forthcoming).
3.4.1 True transposition
Consider the data in (21a) - (21 e) which correspond to the "original" forms
in (16a)- (16e):
) a.
ye
ma
ja
pe
yaa
b.
ma
buko
yoo''
masa
reke
c.
ko
jo
masa
barja
d.
mu
guko
too7
e. masa medoo kele
A comparison of the two sets of data clearly indicates that syllables (or parts of
syllables) are transposed; in fact, it is not always the original syllable that is
transposed, but specific parts of it, i.e., the left margin and the peak.
Interestingly, these correspond to the core syllable: CV or CVV (cf. 17a) - (17c)).
Some examples are given below:
114 Studies in the Linguistic Sciences 20:1 (Spring 1990)
(22) jaay -^ jaa
yapp -^ yaa
ker -^ ke
jox -^ jo
jabar -^ ja
lekk -^ le
(The reason for the deletion of /x/ in jox (21c), and of /y/ in samay (21 e) remains
unclear at this point).
After transposition has taken place, the syllabification principles of the
language will fail to syllabify exhaustively the new segmental string. Syllable-
sensitive rules will then apply, in accordance with Wolof syllable structure; those
rules are: schwa insertion and degemination:
— schwa insertion:
After transposition, the original right margin becomes a left margin; in order to
be syllabified, that LM needs a peak. A rule of schwa insertion will apply; it is
independently motivated (cf. Ka 1985, 1988, Forthcoming), and has the form:
(23) ^ V y —
The following examples illustrate the rule:
(24) jaay (ma) ye (ma) jaa
ker reke
doom medoo
— degemination:
If the original hght margin was a geminate, it will degeminate in its new left
margin position, since only a single C is allowed in syllable-initial position:
recall the syllabification principle in (17b). The rule can be formulated as
follows:
X X • r
/\ ^ \ / [
(25) X X
/\ - I
c c c
The rule is illustrated in (26):
(26) yobbu (ko) bu (ko) yoo
toggu (ko) gu (ko) too
yapp pe yaa
lekk ke le
Ka: Reduplication and proscxlic constituents in Wolof 115
(Notice that, in the last two examples, both degemination and schwa insertion
will apply).
The next question is to identify the donnain within which True Transposition
takes place. Recall the bracketing of prosodic words in the "original" data in
(19a) - (19e). The same bracketing is preserved after transposition of the core
syllable, as shown in (27a) - (27e):
(27) a. [ye ma jaa] [peyaa]
b. [ma] [bu ko yoo] [masa] [reke]
c. [ko jo] [masa] [barja]
d. [mu] [gu ko too]
e. [masa] [medoo] [kele]
Thus, it appears that the core syllable (CV or CVV) is transposed from the left
end to the right end of the prosodic word. The transposition rule will be
formulated as follows:
(28) Move Oc from the left edge to the right edge of Wd.
Note that an alternative analysis — such as the one proposed in McCarthy
& Prince (1986) — is also possible. It would consider this variety of Kail as
simply involving a copying of the prosodic word, since the same bracketing is
preserved. Recall the bracketing of the "original" data in to prosodic words in
(19a) - (19e); in (27a) - (27e), that bracketing is simply reduplicated, leaving
intact the number of prosodic words. Compare for example (19b) and (27b):
(19) b. [ma] [yobbu ko] [sama] [ker]
(27) b. [ma] [bu ko yoo] [masa] [reke]
McCarthy and Prince (1986) would therefore consider this is an instance of
reduplication of the prosodic word.
3.4.2 Reduplication
Two varieties of Kail involving the reduplication of a prosodic category will
be examined here: One requires the copying of only a single mora of Oc, the
other one entails infixation and copying of the metrical foot.
3.4.2.1 Monomoraic copy of the core syllable
Consider the data in (29), which correspond to the "original" form in (19a):
(29) [7ay ma jaa] [7app yaa]
(19) a. Oaay ma] [yapp]
116 Studies in the Linguistic Sciences 20:1 (Spring 1990)
Here again, the core syllable CVV is transposed from the left edge to the right
edge of the prosodic word, in a manner similar to the transposition rule in (28).
However, in (29), the transposed CVV leaves behind a copy of only one mora
(assuming that in Wolof a heavy syllable contains two moras and a light syllable
one mora: cf. Ka 1988, Forthcoming). The copied mora is followed of course by
the original right margin. The process will look like this:
C. V V Cr ^ V Cr C. V V
xv^ V N^
a CO
(where C| = left margin, and Cr = right margin).
Taking into account the fact that in other cases of transposition the core
syllable (i.e., the left margin and the peak) is moved, one would have expected
the copied V in (29) to be heavy, but this is not the case. To account for this,
McCarthy and Prince (1986) posit a Copy-Base Complementary Principle,
paraphrased as in (30):
(30) If the base is heavy, consider the copy as light.
This principle would apply to the reduplication data found in several languages:
cf. for instance Sanskrit verb reduplication (Steriade 1982, McCarthy & Prince
1986), Ponapean durative verb reduplication (McCarthy & Prince 1986),
Southern Paiute reduplication (McCarthy 1983). In this paper, I will propose a
somewhat more prosodic explanation of the mora problem. Note that, although
vowel length is distinctive in Wolof, segmental quantity is not preserved in the
copy. The transposition process is clearly not sensitive to the segmental make-
up of the word. Within a prosodic framework, this is accounted for: Syllable size
takes precedence over segmental size; the loss of vowel quantity still preserves
the minimum size of the syllable, i.e., CV. The following principle of Prosodic
Precedence will be posited:
(31) V < Oc
(where < reads: "is secondary to")
A last problem to account for is that of the left margin: The transposition of the
core syllable leaves the copied mora and the right margin without a left margin.
To obey the syllabification principle in (17b) which requires syllables in Wolof to
have a left margin, the empty position in syllable-initial position will be filled In
by a prothetic glottal stop:
(32) -^ ? / [ _
The rule in (32) is independently motivated: It applies as a default rule on
borrowed words that are vowel-initial in the source language (cf. Ka 1988,
Forthcoming).
Ka: Reduplication and prosodic constituents in Wolof 117
3.4.2.2 Infixation and reduplication of the foot
Consider the data in (33a) - (33d), which correspond to the "original" form
in (16a):
(16) a.
jaay
ma
yapp
(33) a.
jaray
mara
yara
pere
b.
jalfay
malfa
yalfa
pelfe
c.
jancay
manca
yanca
penc(
d. jariatlahay mahatlaha yahatlaha pehetiehe
On the surface, the data seem to involve the insertion of specific Cs after the
peak of every core syllable; those Cs are followed by a copy of the V of the core
syllable. These configurations are attested in various languages: Cuna,
Javanese, Saramaccan Creole, Tagalog (cf. McCarthy 1982), English, German
(McCarthy 1984, McCarthy & Prince 1986).
The analysis will be as in (34):
(34) a. determine a core syllable CV;
b. prespecify on a different tier a melody consisting of fixed
consonants: rifnc htlh, and affix that melody to the CV tier;
c. spread the peak of the core syllable to all available V slots, from
the left to right;
d. "linearize" the original and the infixed melodies on a single tier.
(34a) will trigger both degemination of a right margin when it is transfered to a
left margin position, and schwa insertion;
(35) a. yapp -^ yara pere (33a)
-^ yalfa peife (33b)
-> yanca pence (33c)
-^ yahatlaha pehetiehe (33d)
In contrast, if the nght margin consists of a single C, no transfer will occur:
(35) b. jaay -^ jaray (33a)
^ jalfay (33b)
-^ jancay (33c)
-> jahatlahay (33d)
This again illustrates the difference in behavior between single and geminate
consonants on the CV tier, and further confirms the conspicuous absence of the
following disyllabic stem structures in Wolof: A form such as *yarapp, therefore,
is not licensed.
118
Studies in the Linguistic Sciences 20:1 (Spring 1990)
(36) C V C V C C
In (34b), prespecification takes precedence over any melodic element that is
outside the core syllable (cf. Marantz 1982, McCarthy & Prince 1986). In (34c),
spreading provides a peak to the prespecified consonants that have become
left margins.
The representations in (37) and (38) will correspond to (33a) and (33d):
(37) CVVC
ja y
(38)
CVVC
\v\
j a y
cvvcc
\vv
ya p
cvcvc
ll/l
j a y
CV->CVCV
1! II''
ma ma
ri t I ri
I I I I
CVCVCCVCVC
\V-''"\
j a y
ri t I ri
I II I
CVCVCCVCV
V^'"-
cvvcc
ya p
r r
I I
CVCV CVCV
\v w
ya pe
CV
ri t I fi
till
o\ic:,\iQ.c:,MOM
lU
ri t I ri
I II I
CVCVCCVCV
IUi='"
pe
The important question to be answered then, is that of the nature of the prosodic
category that is reduplicated in this type of KaW. Notice first that each core
syllable of the "original" form triggers a separate prosodic word in the "secret"
form. Consider for instance (39), in which the core syllable CV is underlined:
(39)
jaay
[jaray] (33a)
[jalfay] (33b)
[jancay] (33c)
[jahatlahay] (33d)
Notice that the core syllable may be one created by degemination (cf. (25)) and
schwa insertion (cf. (23)):
(40)
a. yapp
[pere]
(33a)
[peife]
(33b)
[pence]
(33c)
[pehetlehe]
(33d)
In each of the newly created prosodic words, the structure of the metrical foot is
uniform: it is always binary (and left-dominant), in conformity with the Wolof
Ka: Reduplication and prosodic constituents in Wolof 119
Stress algorithm (cf. (20) and Ka 1988). Consider (41), in which phonological
words comprise either a single binary foot or two binary feet:
(41) a. Oa ray] [ma ra] [ya ra] [pe re]
F F F F
b. [jariat laiiay] [mariat lafia] [yariat lafia] [periet lerie]
vv vv vv vv
FF FF FF FF
Since all feet have the same labelling in each sentence, ^ and since that
labelling conforms to the canonical form of the foot in Wolof, uniformly
replicating it, it is possible to characterize this Ka// variety as an instance of foot
reduplication.
4. Conclusion
In this paper, I have shown that reduplication in Wolof involves more than
a mere copying of a morphemic skeleton. The examination of language-
external evidence — in particular the secret code Kail — reveals that other
constituents, of a prosodic nature, are also replicated. They are the core syllable
and the metrical foot. This constitues another confirmation of the importance of
the role of prosodic units in the phonology and morphology of the language.
NOTES
I would like to thank two anonymous reviews for their comments and
suggestions on an earlier version of this paper.
â– â– Disyllabic stems susceptible of reduplication are underlyingly
monsoyllabic; they become disyllabic on the surface to obey the syllabification
principles of the language. The only exceptions are borrowed words and names
of locations that can historically be analyzed as involving more than one
morpheme (cf. discussion in Ka 1988).
2 I use here the orthography adopted in Senegal. The pharyngeal
geminate /qq/ is represented in that orthography as q. Notice that geminate
consonants do not exist in the Gambian variety.
3 a is an orthographic device representing the low long vowel /aa/ before
geminate and prenasal consonants (cf. discussion in Ka 1988).
'♦McCarthy and Prince (1987) propose to consider total reduplication as
involving no copying at all, but rather compounding of a word with itself.
120 Studies in the Linguistic Sciences 20: 1 (Spring 1 990)
5 The data are extracted from an interview of a Kail speaker conducted by
Mr. Abdoul Aziz Diaw, researcher at the Centre de Linguistique Apliquee de
Dakar {C.L.A.D.).
6 Clitic is being defined here from a syntactic point of view, as any element
that does not constitute the head of a phrase (in Wolof, only lexical items may i
have a head status). "
7 The first vowel in yobbu and toggu is underlyingly long and shortens
before a geminate consonant. Thus, the underlying representations of the
above items are respectively /yoobbu/ and /toogguA
8 McCarthy and Prince (1986) posit a Uniformity Parameter to explain this
regularity.
REFERENCES
Clements, G., & J. Keyser. 1983. CV Phonology. Linguistic Inquiry Monograph.
Cambridge: MIT Press.
DiAGNE, P. 1971. Grammaire du Wolof moderne. Paris: Presence Africaine.
Hayes, B. 1986. inalterability in CV-Phonology. Language 62:2.321-51.
Hyman, L. 1985. A theory of phonological weight. Dordrecht: Foris.
Ka, O. 1981. La derivation et la composition en Wolof. Dakar: C.L.A.D. 77.
. 1985. Syllable structure and suffixation in Wolof. Studies in the
Linguistic Sciences 15:1.61-90.
. 1988. Wolof phonology and morphology: A nonlinear approach.
University of Illinois at Urbana-Champaign Ph.D. dissertation in
Linguistics.
_. Forthcoming. Wolof syllable structure: Evidence from a secret code.
Proceedings of the Fifth Eastern States Conference on Linguistics.
University of Pennsylvania.
_, & C. Kisseberth. Forthcoming. Prosodic domains in Wolof vowel
harmony.
Marantz, a. 1982. Re-reduplication. Linguistic Inquiry 13.483-545.
McCarthy, J. 1982. Prosodic templates, morphemic templates, and morphemic
tiers. The structure of phonological representations 1.191-223. Dordrecht:
Foris.
. 1983. Morpheme form and phonological representation. University of
Texas at Austin, MS.
. 1984. Prosodic organization in phonology. Language sound structure,
ed. by M. Aronoff & R. Oehrle. Cambridge: MIT Press.
, & A. Prince. 1986. Prosodic morphology. University of Massachusetts &
Brandeis University, MS.
, & A. Prince. 1987. Quantitative transfer in reduplicative and templatic
morphology. University of Massachusetts & Brandeis University, MS.
Ka: Reduplication and prosodic constituents in Wolof 121
Sauvageot, S. 1965. Description synchronique d'un dialecte wolof: le parler du
Dyolof. Dakar: Memoires de l"l. F.A.N. 73.
Steriade, D. 1982. Greek prosodies and the nature of syllabification. MIT Ph.D.
dissertation in Linguistics.
Steriade, D. 1986. Yokuts and the vowel plane. Linguistic Inquiry 17.129-46.
Yip, M. 1982. Reduplication and CV skeleta in Chinese secret languages.
Linguistic Inquiry 13.637-62.
studies in the Linguistic Sciences
Volume 20, Number 1, Spring 1990.
VOWEL HARMONY IN TURKANA
Manuela Noske
University of Chicago
In this paper I describe two processes of vowel harmony in
Turkana. Turkana has a nine vowel system which can be
distinguished by the features [ATR] (Advanced Tongue Root),
[round] and [low]. While the features [ATR] and [round] are
equipollent features in Turkana, [low] is employed as a privative
feature. I argue that [ATR] harmony operates to spread the feature
[+ATR] to a vowel which is specified for either value of the
equipollent feature [round] and to spread the feature [-ATR] to a
vowel specified for the privative feature [low]. The features [round]
and [low] are therefore conditioning factors in [ATR] harmony. While
there is a two-way distinction underlyingly in roots ([+ATR] or
unspecified), there is a three-way underlying distinction in suffixes
([+ATR], [-ATR], or unspecified). I claim that vowel harmony in
Turkana is feature-changing.
Turkana is an Eastern Nilotic language which is spoken by approximately
250,000 people in Turkana District, Kenya. The aim of this paper is to analyze
the most salient properties of the vowel harmony system of this language. The
vowel harmony system of Turkana is unique in that both the plus and the minus
value of the harmonizing feature [Advanced Tongue Root] spread. The data
used in this description are from fieldnotes which I collected in Kenya in 1986
and from Dimmendaal's 1983 grammar of the Turkana language.
1. Vowel system
Turkana has a nine vowel system, which can be divided into two sets by
the feature [Advanced Tongue Root] ([ATR]).
(1) [+ATR] [-ATR]
i u ! M
e e 9
a
124 Studies in the Linguistic Sciences 20: 1 (Spring 1 990)
The vowel system is asymmetrical since it lacks a [+ATR] low vowel [a].
The low vowel [a] is phonetically always [-ATR] and patterns phonologically with
the set of [-ATR] vowels.
Instead of using a set of binary features, I analyze the Turkana vowel
system using a privative feature [low] and an equipollent feature [round]. The
equipollent feature [round] gives rise to a ternary distinction. A vowel can be
specified as [+round], [-round], or it can bear no specification for this feature at
all. Rather than specifying the absence of rounding, a [-round] vowel is opposed
not only to vowels which are specified as [+round] but yet to a third category:
those vowels for which the absence of any such feature specification is
distinctive.
(2) a. [+F] b. [-F] c.
I I
V V V
All vowels which are specified as [-round] are assigned the phonetic
feature [front] by a late redundancy rule:
(3) Default [front] Assignment: [-round] -^ [front]
In contrast, the feature [low] gives rise to a binary distinction between
those vowels which are underlyingly specified as [+low] and those vowels
which do not bear a specification for this feature.
(4) a. [F] b.
I
V V
The combination of an equipollent feature [round] with a privative feature
[low] gives rise to six possible vowels of which only five occur in Turkana. It has
to be assumed that Turkana does not allow a completely unspecified segment
In its inventory. Hence the five-vowel system in (5) emerges:
(5) [+rd] [-rd] [+rd] [-rd]
I I I I
V V V V V V
I I I
[low] [low] [low]
/U/ /I/ /O/ /E/ /A/
The advantage of using a privative and an equipollent feature to analyze
a five vowel system over a traditional approach which uses the three binary
features [high], [low] and [round] is that the first approach gives rise to less
possible combinations. While the choice of features opted for here limits the
number of possible vowels to six, the three binary features give rise to eight
Noske: Vowel harmony in Turkana 125
possible combinations, of which two ([+high, +low, +round] and [+high, +low,
-round]) have to be ruled out as impossible.
On top of a five vowel system Turkana has a distinctive autosegmental
feature [ATR]. I will assume that [ATR] just like [round] is an equipollent feature,
and that both the [+ATR] and the [-ATR] values are specified underlyingly. In
addition, vowels can be unspecified for the feature [ATR] and receive a
specification for this feature either by spreading or by complement assignment.
In this treatment of the Turkana vowel system I follow a suggestion made
in Goldsmith (1985) and (1987). The choice of the features [ATR] and [round] as
equipollent features is language specific. As will be seen later, this feature
specification allows for a natural description of the rules of vowel harmony.
2. Vowel harmony
In general all vowels in a word must be chosen from the first or the second
set of vowels in (1). Which kind of vowels occur in the word can either be
determined by the root or by the suffix. I will first consider cases of root
controlled harmony. If the word contains a [+ATR] root, then the prefixes and
suffixes are [+ATR], and if the word contains a [-ATR] root then both prefixes and
suffixes will be realized as [-ATR].
(6) a. [agolun] 'to close in' INF: close: MT^
b. [adokun] 'to climb down' INF: climb: MT
Roots can therefore be classified into two groups depending on their
specification for tongue root advancing. They are either lexically specified for
the feature [+ATR] or they are unspecified for this feature. The reason for
treating roots as lexically unspecified rather than as being [-ATR] is to avoid
changing their feature specification when they are adjacent to a dominant
[+ATR] suffix. If [+ATR] Spreading were a feature-changing rule, then it would
not only delete the [-ATR] specification of a [-ATR] root, but it would also delink
the [-ATR] specification of a dominant [-ATR] suffix. I will return to this point
below.
(7) a. [+ATR] b.
I
CVroot CVroot
The [+ATR] feature of the root spreads to all prefix and suffix vowels which
have no specification for this feature. In this respect vowel harmony in Turkana
is root controlled.
(8) [-hA] [+A]
I , (^) r\\
A-gol-Unr -^ A-gol-uni [agolun] 'to close in'
INF-close-MT
126 Studies in the Linguistic Sciences 20: 1 (Spring 1 990)
The rule of [+ATR] Spreading is formalized as in (9) below.
(9) [+ATR] Spreading: [+ATR]
W
V V (bidirectional)
If a root is unspecified for the feature [ATR] it receives the feature value
[-ATR] by a complement rule:
(10) Complement [-ATR] Assignment: [ ] -^ [-ATR]
An example of Complement [-ATR] Assignment is given in (11). The root
as well as the prefix and suffix vowels in this example surface as [-ATR].
(11)
[-A]
(10)
/ \ \
A-dOk-UnI
->
A-dok-uni
[adokun] 'to climb down'
INF-climb-MT
In addition Turkana has dominant suffixes which cause the root and prefix
vowels to assimilate. If a dominant [+ATR] suffix is added to the root, it will
surface as [+ATR] and if a dominant [-ATR] suffix is added, the root will surface
as [-ATR].
(12) a. [egole] 'way of closing' M: close: GER
b. [edoke] 'way of climbing' M: climb: GER
(13) a. [agolere] 'to close' INF: close: SBJV
b. [adokere] 'to climb' INF: climb: SBJV
Suffixes, unlike roots, have a three-way distinction for the feature [ATR].
Suffixes can be lexically specified as either [+ATR], [-ATR], or are unspecified for
this feature.
(14) a. [+ATR] b. [-ATR] c.
I I
CVsuffix CVsuffix CVsuffix
If a suffix which is unspecified for the feature [ATR] is added to a root, it
can receive a specification for this feature in one of two ways. The suffix either
associates with the [+ATR] feature of the root as in (8) or it is assigned a [-ATR]
feature by the complement rule, if the root itself is unspecified for [ATR] as in
(11) above.
Suffixes which are specified for the feature [+ATR] spread this feature onto
the preceding root and prefix vowels.
(15) [+A1[+A]
[*A1=
, II
(9)
/7\
E-gol-e
—>
e-gol-e
Noske: Vowel harmony in Turkana 127
[egole] 'way of closing'
M-close-GER
(16) [+A] [+A]
I (^) .vl
E-dOk-e -^ e-dok-e [edoke] 'way of climbing'
M-climb-GER
No effect of the dominant [-hATR] suffix can be observed in example (15)
because both the suffix and the root are underlyingly specified as [+ATR]. In
example (16), however, it can be seen that the [+ATR] feature of the suffix
spreads to all preceding vowels, thereby causing the unspecified root to
become [+ATR]. In these cases Turkana vowel harmony is of the dominant type.
Suffixes can be dominant and determine the ATR-category of the word.
Finally there are those suffixes which are underlyingly specified as [-ATR].
Not only do these suffixes always surface with [-ATR] vowels, but they even
spread this feature onto the preceding root and prefix vowels.
(17) [+A][-A] [+A][-A]
I /\ <-. t//\
A-gol-ere -^ A-gol-ere [agolere] 'to close'
INF-close-SBJV^
(18) [-A] [-A]
/\ (-) /7\
A-dOk-ere -^ A-dok-ere [adokere] 'to climb'
INF-climb-SBJV
In example (17) the [-ATR] feature of the subjunctive suffix associates with
the root vowel in a process of delinking and spreading. Since the underlying
[+ATR] vowel surfaces as [-ATR], [-ATR] Spreading must be a feature-changing
process. In example (18) a [-ATR] suffix is added to an unspecified root. All
vowels in the word associate with the [-ATR] feature of the suffix. Apart from the
rule of [-t-ATR] Spreading and Complement [-ATR] Assignment, a rule of [-ATR]
Spreading is therefore needed. This process is formalized in the following rule:
(19) [-ATR] Spreading: [-ATR] (to be revised in (22))
V V (bidirectional)
[-ATR] Spreading is, however, subject to the following restriction: the
feature [-ATR] spreads only to vowels which have a specification for the feature
128
Studies in the Linguistic Sciences 20:1 (Spring 1990)
[low]. In Turkana not only IN but also the mid vowels /E/ and 101 are specified as
[low] and are therefore a target for [-ATR] Spreading. Since high vowels have
no specification for this feature, [-ATR] can spread neither to /I/ nor /U/.
(20)
[+A] [-A]
I /\
A-buk-ere
'\ /
[low]
[abukere]
'to pour'
INF-build-SBJV
The [-ATR] feature of the subjunctive marker /-ere/ in (20) does not spread
to the root, because the high root vowel has no specification for the feature
[low]. If the same suffix, however, is added to a [+ATR] mid vowel root, the
feature [-ATR] can spread and the underlying [+ATR] root becomes [-ATR].
(21 :
l+AH-A]
[tA][-Al
1 /\
(22)
t/7\
A-gol-ere
— >
A-gol-ere
[agplere]
'to close'
w//
\\//
INF-close-SBJV
[low]
[low]
The rule of [-ATR] Spreading in (19) therefore has to be revised in order to
apply only to vowels which are specified for the feature [low].
(22) [-ATR] Spreading (revised):
â– a'
V V
\/
[low]
(bidirectional)
The different underlying specifications of roots and suffixes, as well as
their surface forms are summarized in (23):
(23)
suffix:
root:
[+ATR] /-gol/
unspecified /-dOk/
[+ATR] /-buky
unspecified /-dUk/
motion towards gerund subjunctive
unspecified
/-Unl/
[agoluni]
[adokuni ]
[abukuni]
[adukuni ]
[+ATR] [-ATR]
l-el l-^'^^l
[egole] [agolere]
[edoke] [adokere]
[ebuke] [abukere]
[eduke] [adukere]
3. The analysis of low vowels
As was mentioned above Turkana does not have a [+ATR] low vowel [s ]
in its segment inventory. The low vowel always surfaces as [-ATR] [a]
phonetically. This means that the low vowel cannot associate with the feature
[+ATR] and therefore does not undergo harmony. IN also blocks the spreading
Noske: Vowel harmony in Turkana 129
of the feature [+ATR]. Words consisting of a [-ATR] prefix and a [+ATR] root are
common. They occur if the first vowel in the root is the low vowel IN.
(24) [emakuk] 'chair' M: chair: SG
[ngimakukyo] 'chair' M: chair: PL
(25) [+A] [-A] [+A]
I (10) I I
E-mAkuk -> e-mAkuk [emakuk] 'chair'
M-chair: SG
In (25) the [+ATR] specification of the root does not spread to the prefix
across the low root vowel. The prefix vowel surfaces with the complement value
[-ATR].
Further evidence to show that IN is opaque in the vowel harmony system
of Turkana comes from words which consist of a low vowel root to which a
dominant [+ATR] suffix is added, in (26) the [+ATR] feature of the genjnd suffix
does not spread past the low vowel to the prefix vowel. The prefix vowel
therefore surfaces with the complement value [-ATR].
(26) [M] [-A] [+A]
I (10) I I
E-rAm-e -^ e-rAm-e [erame] 'way of killing'
M-kill-GER
The question is how to analyze the behavior of the low vowel in a
framework which uses both equipollent and privative features. One possibility is
to represent the low vowel as underlyingly linked to a feature [-ATR]. If [+ATR]
Spreading is a feature-filling and not a feature-changing rule, then spreading of
the [+ATR] feature cannot delink the [-ATR] specification of a low vowel and the
low vowel will surface as [-ATR].
(27) [-A] [+A] [-A] [+A]
\ I /\ I
E-makuk -^ e-makuk [emakuk] 'chair'
M-chair: SG
(28) [-A] [-.A] [-A] [+A]
\ I /I I
E-ram-e -^ e-ram-e [erame] 'way of killing'
M-kill-GER
This solution is, however, untenable for Turkana since both the [+ATR]
and the [-ATR] values spread. If IN is underlyingly specified as [-ATR] and if
[-ATR] spreads in a feature-changing fashion, then we would expect the [-ATR]
feature of the low vowel to spread to the surrounding vowels. While the [-ATR]
specification of a low vowel can be prevented from associating to the following
130 Studies in the Linguistic Sciences 20:1 (Spring 1990)
root vowel in (27) by invoking the Strict Cycle Condition, it cannot be prevented
from spreading to the suffix vowel in example (28). In (28) the [-ATR] feature
should spread to the suffix vowel and thereby delink its underlying [+ATR]
specification.
Another possibility to account for the opacity of the low vowel is to m
represent /A/ as underlyingly unspecified for [ATR] and to make it subject to a ^
negative co-occurrence constraint which prohibits the occurrence of the feature
[ATR] with the low vowel as in (29).
(29) Negative Co-occurrence Constraint: *[ATR]
I
V
I
[low]
Since both the low and the mid vowels are specified for [low], this
constraint would also apply to the latter and prohibit the association of the
feature [ATR] with /E/ and /O/. This effect is clearly unwanted. A co-occurrence
constraint of the form stated in (29) does not therefore express the correct
generalization.
Instead of a negative constraint a positive co-occurrence constraint can be
formed which makes the association of the feature [ATR] dependent on the
presence of the feature [R round]. If [ATR] can only associate with a vowel which
is specified for [B round], then all vowels except the low vowel itself represent
suitable targets for the association with this feature.
(30) Positive Co-occurrence Constraint: [ATR]
I
V
I
[P round]
The filter in (30) states that only vowels which bear a specification for
either value of the equipollent feature [13 round] can associate with the feature
[ATR]. Since vowel harmony is structure preserving, (30) expresses both an
existing constraint on the underlying representation of vowels as well as a
constraint which holds throughout the lexical phonology. The low vowel /A/
therefore does not constitute a suitable target for either [+ATR] Spreading or
[-ATR] Spreading. In addition, I assume that Turkana has a constraint which d
prohibits vowels from being skipped in the association process. If a vowel "
cannot be associated with the [ATR] feature, then spreading to the following j
vowel is blocked. 1
I will tentatively assume that the low vowel is underlyingly unspecified for
the feature [ATR], although the behavior of low vowels in suffixes indicates that
this restriction might be too strong and that Turkana does indeed have two low
vowels in its inventory: a low vowel which is underlyingly unspecified for [ATR]
Noske: Vowel harmony in Turkana 131
and a low vowel which Is underlyingly linked to a feature [-ATR]. I will return to
this point below. If the low vowel IN is represented as bearing no specification
for the feature [ATR] underlyingly, then a default rule is needed which assigns
the low vowel the feature value [-ATR] after round vowels have received their
[-ATR] specification by the complement rule in (10).
(31) Default [-ATR] Assignment: [low]
4. Low vowel rounding
[-ATR]
Suffixes which contain a low vowel underlyingly undergo a
morphophonemic rule which changes IN to a low round vowel 101 if the suffix
follows a [+ATR] root.
(32) Low Vowel Rounding:
In (33) the low suffix vowel is rounded after a [+ATR] root. The rounded
vowel then meets the structural description for [+ATR] Spreading and
associates with the [+ATR] feature of the root, so that it will surface as [o]. No
rounding takes place in (34) in which the root is unspecified for [ATR] and in
which the structural description for rounding is therefore not met.
[ajulot] 'hair'
F-hair-SG
(9)
[low]
(34)
(10)
A-mUk-At
[-A]
I
A-muk-At
[amukat] 'shoe'
F-shoe-SG
132
Studies in the Linguistic Sciences 20:1 (Spring 1990)
However, a slightly different process takes place when the suffix {-ari}5 is
added to a [+ATR] root. In this case the low vowel turns into a low [-ATR] round
vowel [p]. I assume that the suffix {-ari} is underlyingly linked to a [-ATR] feature
since it spreads this feature backwards and thereby delinks the underlying
[+ATR] specification of the root.
[+A][-A]
I A
A-gol-ori
[+rd][+rd]
[low]
(22)
[+A][-A]
+-''A
â– A-gol-ori-
[+rd] h-
[low]
[agplor] 'to close out'
INF-close-MA
in example (35) [+ATR] Spreading fails to apply since the suffix {-ah} is
already linked to a [-ATR] feature and [+ATR] spreading is not a feature-
changing rule. However, in the same example the structural description for
[-ATR] Spreading is met. The [-ATR] feature of the suffix spreads backwards to
the root vowel, thereby delinking its [+ATR] feature. It is crucial to note that in this
example Low Vowel Rounding applies to the suffix {-ari} although it is
Invariantly [-ATR]. This shows that it is not the spreading of a [-hATR] feature to a
low vowel which causes it to be rounded, but that we are dealing instead with
an insertion rule whose only condition is that the preceding morpheme be
linked to the feature [+ATR]. Example (35) therefore provides another instance
of feature-changing harmony.
(36)
â– A-dUk-ari-
y \
[+rd]-
[low]
[-A][-A]
1 A
A-duk-ari-
[+rd] —
[low]
(10)
[adukar] 'to build over there'
INF-build-rvlA
Noske: Vowel harmony in Turkana 133
In contrast, in example (36) the root is unspecified for [ATR] and Low
Vowel Rounding fails to apply. The suffix in this example surfaces with a [-ATR]
low vowel [a].
[-ATR] Spreading (22) applies after the rule of Low Vowel Rounding (32),
while [+ATR] Spreading (9) in its turn applies after [-ATR] Spreading.
(37) Low Vowel Rounding (32)
[-ATR] Spreading (22)
[+ATR] Spreading (9)
Complement [-ATR] Assignment (10)
Default [-ATR] Assignment (31)
Default [front] Assignment (3)
In conjunction with the spreading processes described above it is of
interest to note that all dominant suffixes in Turkana are mid vowel suffixes. ^ By
first assuming that all vowels which are underlyingly linked to the [round] tier
and the [low] tier also have to be associated to the [ATR] tier, the dominance of
the mid vowel suffix is explained. It is then the multiple association of a segment
to the different vowel tiers that causes it to be dominant and to spread the
feature [ATR] onto preceding vowels.
NOTES
I am thankful to two SLS reviewers for their useful comments and
suggestions on an earlier version of this paper. Any remaining weaknesses or
error of facts remain solely my responsibility.
■• MT stands for the suffix /-Unl/ which denotes a motion towards the
speaker. The underlying final high vowel of this suffix is devoiced or deleted
before a pause.
2 Capital letters indicate vowels which are lexically unspecified for the
feature [ATR].
3 Two adjacent identical autosegments are prohibited according to the
OCR.
^ The subjunctive is formed with the help of the voice marker /-A/ plus a
word-final instrumental marker {-re}. The voice marker /-A/ is contracted with a
preceding high front vowel to form the subjunctive marker {-ere}. I represent the
[-ATR] feature as being linked to both mid vowels, though strictly it should only
be associated with the latter.
5 The suffix {-ari} indicates a motion away from the speaker. The
underlying final high vowel is devoiced or deleted before a pause.
134 Studies in the Linguistic Sciences 20:1 (Spring 1990)
6 Exceptions to this claim are the nominal plural marker {-i} and the
abstract noun marker {-u}. Dimmendaal and Breedveld (1986) claim that the two
suffixes contain voiced vowels underlyingly, which are devoiced or deleted in
pre-pausal position. They are, however, fully voiced if followed by a vowel-initial
word. My data do not include any examples formed with the abstract noun
marker {-u} which would allow me to judge its behavior. I do, however, have
doubts about the claim that the nominal plural marker {-i} is dominant in
Turkana vowel harmony. Suffixation of the the plural marker {-1} is not
productive in Turkana.
REFERENCES
Dimmendaal, Gerrit Jan. 1983. The Turkana language. Dordrecht: Foris.
, & Anneke Breedveld. 1986. Tonal influence on vocalic quality. The
phonological representation of suprasegmentals, ed. by Koen Sogers,
Harry van der Hulst & Maarten Mous, 1-33. Dordrecht: Foris.
Goldsmith, John. 1985. Vowel harmony in Khalkha Mongolian, Yaka, Finnish
and Hungarian. Phonology Yearbook 2.259-275.
. 1987. Toward a theory of vowel systems. Papers from the 23rd annual
regional meeting of the Chicago Linguistic Society. Part two: Parasession
on Autosegmental and Metrical Phonology, ed. by Anna Bosch, Barbara
Need & Eric Schiller, 116-133. Chicago: Chicago Linguistic Society.
Studies in the Linguistic Sciences
Volume 20, Number 1, Spring 1990
SONORANT-STRENGTHENING IN LAMA
Meterwa A. Ourso and Charles H. Ulrich
In Lama, a Gur language, two distinct rules change sonorants to
stops after another sonorant. W-Strengthening changes /w/ to /p/
after /m/, and R-Strengthening changes /r/ to a retroflex stop after any
sonorant consonant. These two rules, along with a rule deleting
glides in post-consonantal position, are motivated by the Syllable
Contact Law (Murray & Vennemann 1983), which states that
heterosyllabic consonant clusters are preferred in which the second
consonant is less sonorous than the first. R-Strengthening interacts
with a rule of Schwa Deletion that exhibits a relatively uncommon
type of compensatory lengthening, where a vowel is lengthened to
compensate for the loss of a vowel in the following syllable.
1. Introduction
Lama (also known as Lamba) belongs to the Grusi (or Gurunsi) branch of
the Gur (or Voltaic) language family. It is spoken in northern Togo. Previous
work on Lama includes a grammar by Prost (1963).
Lama has the following underlying consonants:
(1) p t c k kp
p
t
c
f
s
m
n
n
w
1
r
y
Orthographic d represents a retroflex stop (derived from underlying /r/), not a
voiced alveolar stop. All consonants can occur in syllable onsets. All non-
palatal sonorants can occur in codas.
Lama has eleven underlying vowels. The [-ATR] non-low vowels /i. e a p u/
have [+ATR] counterparts /i e s o u/. The [-ATR] low vowel a has no contrastive
[+ATR] counterpart, although a occurs through vowel harmony (Ourso 1988).
Tones, which are irrelevant to the phenomena under discussion, are not
marked in this paper (but see Kenstowicz et al. 1988, Ourso 1989).
136 Studies in the Linguistic Sciences 20:1 (Spring 1990)
Lama has ten noun classes: four singular (classes 1, 3, 5, 7), four plural
(classes 2, 4, 6, 8), and two mass (classes 9, 10). Noun class is indicated by a
suffix on the noun root. Most singular nouns in class 1 have plurals in class 2,
and so on for classes 3/4, 5/6, and 7/8. Third-person personal pronouns are in
most cases homophonous with the class suffixes. A number of phonological
rules apply at the boundary between noun root and class suffix (Ourso 1989,
forthcoming).
2. Labial Sonorant/Obstruent Alternations
The labial stop p and the labial-velar glide w contrast in word Initial
position (2) and in intervocalic position (3):
(2) wetu 'to sell'
pelu 'to cut'
(3) awpr 'place, position'
apeer (proper name)
Elsewhere, the opposition between /p/ and /w/ is neutralized.
Only sonorant consonants can occur in syllable codas in Lama, /p/
becomes /w/ before a consonant, as in the causative form in (4), and word-
finally, as in the imperative form in (5):
(4) kpap-3 'to be similar'
kpaw-s-u 'to reconcile'
(5) yap-8 'to buy'
yaw 'buy!'
This alternation can be expressed with a rule weakening an unsyllabified /p/
(indicated in the rule by circling the unsyllabified segment):
(6) P-Weakening (Stratum 1 only)^
p -> [+sonorant]
As will be shown below, P-Weakening must be restricted to the first stratum of
the lexical phonology.
The demonstrative pronouns illustrate a converse alternation. After the
homorganic nasal of the demonstrative, /w/ becomes /p/ and Ixl becomes the
retroflex stop 161:
(7) wa 'they (class 2)' mpa 'those (class 2)'
rs 'it (class 7)' nda 'that (class 7)'
Note, however, that /y/ does not become Id in the same environment:
Ourso & Ulrich: Sonorant-Strengthening in Lama 137
(8) ya 'they (class 8)' nya 'those (class 8)'
While it is difficult to identify a natural class including /w/ and /r/ but excluding /y/,
there is an obvious historical explanation for the failure of /y/ to strengthen.
Lama /w/ and /r/ correspond to /b/ and 161 (i.e. a voiced alveolar stop) in related
languages, and are derived from Proto-Gurunsi */b/ and */d/ (Manessy 1969).
Lama /y/, on the other hand, corresponds to /y/ in related languages. Thus, it is
only those sonorants derived diachronically from stops that alternate with stops
synch ronically.2
Two homophonous {-w^a} suffixes illustrate the behavior of morpheme-
initial /w/ in different environments. These are the past tense suffix and one
allomorph of the class 2 noun suffix.^ Both are inflectional suffixes of the second
lexical stratum. The /w/ surfaces unchanged after vowel-final roots:
(9) na- 'to see' nawa 'saw'
CO- 'to listen' cowa 'listened'
(10) apu- apuwa 'headlice'
alu- aluwa 'devils'
After a verb root ending in /m/, the /w/ of either suffix is strengthened to /p/:
(11) hom- 'to pull' hpmpa 'pulled'
ram- 'to bite' rampa 'bit'
(12) rantam- rsntampa 'deaf people'
yadsm- yadampa 'co-wives'
This strengthening could be analyzed in either of two ways: as assimilation of
the feature [-continuant], or as dissimilation of the feature [+sonorant]. It will be
shown below that a similar process of R-Strengthening is unambiguously
dissimilatory. Therefore, we analyze W-Strengthening also as a dissimilatory
process, changing a sonorant consonant to an obstruent after another sonorant:
(13) W-Strengthening
w -> [-sonorant] / m
When the past tense suffix or the class 2 suffix follows any consonant other
than /m/, the /w/ is deleted:
4) tatsr-
'to crush'
tatara
'crushed'
wil-
'to hunt'
wila
'hunted'
kpew-
'to fasten a
belt'
kpewa
'fastened a belt'
san-
'to help'
Sana
'helped'
sap-
'to die'
sapa
•died'
wat-
'to sell'
wata
•sold'
wos-
'to wake'
wosa
'woke'
138 Studies in the Linguistic Sciences 20:1 (Spring 1990)
(15) yir- yira 'people'
waal- waala 'husbands'
Because /w, y/, and /r/ are never found after consonants in Lama, we can
formulate the following rule:
(16) Glide Deletion^
+sonorant
+consonantal
-nasal
-lateral
/ [+consonantal]
We follow Hayes (1989) in analyzing glides as [+consonantal], in order to
distinguish them from vowels without positing a skeleton.
Glide Deletion also deletes /y/ from the noun class 8 suffix {-ya}
(17) se- seya 'fieldmice'
waas- waasa 'vipers'
It never deletes r, as it is always bled by R-Strengthening or Schwa Epenthesis
(to be discussed below). It is bled by W-Strengthening after m.
Derivations of words exhibiting /p/ ~ /w/ alternations are given in (18):
(18) hpm-i-wa(ll) sap+wa(14) kpap+s+u (4)
— — kpawsLJ P-Weakening (str. 1)
hompa — — W-Strengthening
— sapa — Glide Deletion
[hompa] [sapa] [kpawsu]
P-Weakening does not apply to [ssp+wa] because it is a first stratum rule, and
[-wa ] is suffixed on the second stratum.
3. Retroflex sonorant/obstruent alternations
Two homophonous suffixes exhibit alternations between /r/ and the
retroflex stop 161. These are the agentive suffix and the noun class 7 suffix. The
agentive suffix derives nouns from verbs on the first lexical stratum. The class 7
suffix is an inflectional suffix of the second lexical stratum. After a vowel-final
root, either of these suffixes is realized as lengthening of the vowel plus Ixl:
(19)
ha-
'to give'
haar
'benefactor'
se-
'to run'
seer
'runner'
(20)
mi-
miir
'nose'
se-
seer
'field mouse'
Ourso & Ulrich: Sonorant- Strengthening in Lama 139
After a root ending in an obstruent other than /p/, these suffixes are
realized as [sr].
(21) nos-
rik-
wet-
•to fix'
'to herd'
•to sell'
npsar
rikar
wetar
â– fixer^
•herdsman^
•seller'
(22) akpet-
kank-
waas-
akpetar
kankar
waassr
'baboon'
'snake species'
'viper'
After a root ending in /I/ or a nasal, these suffixes are re
(23) kal-
S9n-
yim-
'to count'
'to help'
'to bury'
kalda
sanda
yimda
'accountant
'helper'
'burier'
(24) msl-
sen-
yem-
msda
sends
yenda
'millet'
'bean'
•hippo^
No further rules (other than Vowel Harmony) affect the agentive forms in (23). In
the class 7 forms in (24), a root-final nasal assimilates in place to the following
stop, becoming retroflex, and a root-final /I/ is deleted. These processes are
limited to the second stratum.
Roots ending in /r, w/, and /p/ present further complications. The agentive
forms of roots ending in Irl contain a geminate /rr/ that resists strengthening:
(25) mar- 'to tell a story^ marra 'storyteller'
On the other hand, the class 7 suffix does undergo strengthening after a root
ending in Irl.
(26) sar- sada 'frog^
After triggering R-Strengthening, the root-final 1x1 is deleted by the same rule that
deletes /I/ in (24):
(27) Oral Sonorant Deletion (Stratum 2 only)
-t-consonantal
-i-sonorant
-nasal
+voiced
-lateral
/
140 Studies in the Linguistic Sciences 20:1 (Spring 1990)
Since heteromorphemic geminate Irxl is inalterable in one case, but not
the other, we can posit a rule turning fake geminates into true geminates. That
is, a sequence of two identical melodic units, the first in the rime of one syllable
(linked to a mora, m), the second the onset of the following syllable (linked to a
syllable, s), is converted into a single melodic unit occupying both prosodic
positions:
(28) Geminate Consolidation (Stratum 1 only)
a a
\^
'â– /
Geminate Consolidation applies on the first stratum, where the agentive suffix is
added, but not on the second stratum, where the class 7 suffix is added. After
Geminate Consolidation, a form will no longer meet the structural description of
Glide Deletion (16) or R-Strengthening (33, below), since the /r/ will follow a
vowel.
Root-final /w/ is deleted after triggering strengthening of the /r/ in the
agentive suffix:
(29) kpew- 'to fasten a belt' kpedu 'belt'^
(30) W-Deletion
w -» 0/ d
Notice that W-Deletion (30) deletes /w/ before 161 in the agentive suffix on the
first stratum. Oral Sonorant Deletion (27) deletes any oral sonorant in the same
environment, but only on the second stratum.
Before the agentive suffix, root-final /p/ undergoes P-Weakening (6), and
then behaves like underlying /w/ in triggering strengthening of the /r/ and then
being deleted:
(31) lap- 'to do' lada 'worker'
sap- 'to die' sada 'corpse'
On the other hand, noun roots ending in /p/ behave like roots ending in other
obstruents before the class 7 suffix:
(32) asap- asapar 'red ant'
Thus, P-Weakening must be restricted to the first lexical stratum.
Ourso & Ulrich: Sonorant-Strengthening in Lama 141
Unlike W-Strengthening, there is no uncertainty about the nature of R-
Strengthening. /r/ is strengthened after nasals, liquids, and the labial-velar
glide. Because R-Strengthening is triggered by /w/, it cannot be interpreted as
spreading the feature [-continuant]: w has the opposite value. Moreover, notice
that R-Strengthening, when not bled by Geminate Consolidation, is triggered by
a root-final /r/. Clearly, a feature-changing operation triggered by an identical
segment must be dissimilatory, not assimilatory. Thus, we can formulate R-
Strengthening as follows:
(33) R-Strengthening
r -^ [-sonorant] / Uconsonantal
Usonorant
Although the two strengthening rules cannot be collapsed into a single schema,
the dissimilatory nature of R-Strengthening suggests that W-Strengthening too
is a dissimilation of the feature [sonorant]. Both rules change a sonorant
consonant into a stop after a sonorant consonant, reversing a diachronic
change in the development of Lama. They differ in the class of sonorants
triggering each rule: any sonorant triggers R-Strengthening, while only /m/
triggers W-Strengthening.
4. Schwa deletion and epenthesis
Having accounted for the strengthening of /r/, we must account for the
placement of schwa in the agentive suffix and the class 7 suffix. Given that
schwa sometimes appears before and sometimes after the retroflex consonant,
four possible analyses suggest themselves: the underlying representation might
be[-ar], [-sr?], [-r], or[-r9].
If the suffixes were underlyingly [-sr], they would surface unchanged after
an obstruent-final root (21, 22). Vowel-final roots (19, 20) could be handled
simply enough by having the root-final vowel spread into the mora of the suffix's
schwa. But a metathesis rule would be necessary after a root ending in a
sonorant:
(34) yim+9r(23)
yimgr
Vowel Harmony
yimra
Metathesis
yimdg
R-Strengthening
[yimd9]
Given that the input to such a metathesis rule is syllabifiable, and that
metathesis would apply after sonorants but not obstruents, the rule seems
rather ad hoc.
142
Studies in the Linguistic Sciences 20:1 (Spring 1990)
No metathesis rule would be needed if the underlying form [-sra] were
posited. Instead, various rules would be required to delete one or the other of
the two schwas (without compensatory lengthening). However, most noun class
suffixes in Lama are homophonous with the corresponding independent
pronouns, and the class 7 pronoun is [ra], not '[srs]. Moreover, most class
suffixes in Lama and in other Gur languages have the shape CV (Bendor-
Samuel 1971). Manessy (1979) reconstructs this particular noun class suffix as
*[-de] in Proto-Gurunsi. Thus, there is neither synchronic nor diachronic support
for positing [-ars].
If the underlying representation of the suffixes was [-r],epenthesis rules
would be necessary to insert schwa before or after the Irl whenever the root
ended with a consonant. Such rules would be motivated by considerations of
syllable structure: word-final Irl cannot be syllabified after a consonant. But the
lengthening of a root-final vowel would be totally unmotivated under this
analysis.
In fact, the one constant across the allomorphs of these suffixes, aside
from the presence of some retroflex consonant, is that the suffixed form is
always one mora longer than the root. (Coda consonants do not constitute
moras in Lama.) Thus, the underlying representation of these suffixes should be
one mora long. Since [-sr] has already been rejected, the most likely choice is
[-ra]. This is the form proposed by Kenstowicz (1989). And it is identical to the
surface form of the independent pronoun.
Given the underlying form [-ra], a rule of Schwa Deletion must apply after a
vowel-final root, with compensatory lengthening. Such a rule will derive haar,
for example, from underlying ha+rd. This type of compensatory lengthening,
where the trigger and target are vowels separated by a consonant, is not
predicted by Steriade (1982), but will follow from a morale account of
compensatory lengthening (Hock 1986, Hayes 1989, Kenstowicz 1989).
The rule of Schwa Deletion can be stated as follows:
(35) Schwa Deletion
9-^0/ +consonantal #
+sonorant
The derivation of haar (19) is as follows:
(36) ^ Q (5
h a + r 9 h a r
h a r
Ourso & Ulrich: Sonorant-Strengthening in Lama 143
Schwa Deletion also deletes the schwa of the {-rs} suffixes after an
obstruent-final root. In this case, there is no compensatory lengthening. Instead,
a schwa is inserted to break up the final consonant cluster and bear the mora
left by Schwa Deletion:
(37) Schwa Epenthesis
-> 9 / [+consonantal] [+consonantal] #
The derivation of nosar (21) is as follows:
(38) CJ G o
I / I I \
nos+rs fipsr
The inserted schwa also allows syllabification of the root-final obstruent.
After a root ending in a sonorant consonant, however. Schwa Deletion will
be bled by R-Strengthening.^ The derivation of kaldd (23) is as follows:
(39) kai+rg
kaldg R-Strengthening
— Schwa Deletion
— Schwa Epenthesis
[kalda]
After R-Strengthening, the schwa is not preceded by a sonorant, so it is not
deleted.
Schwa Deletion applies not only to the two {-ra} suffixes, but also to the
noun class 4 suffix {-na}:
(40) na
'they (class 4)'
(41) ti-
tiin
'elephants'
(42) kaas-
kaassn
'crocodiles'
(43) fiar-
fiarsn
•buffalo'
Because the palatal nasal never becomes an obstruent, the class 4 suffix
undergoes Schwa Deletion not only after vowel-final roots (41) and obstruent-
final roots (42), but also after sonorant-final roots (43):
(44) ti+ns kaas+na nar+na
tiiri kaasn fiarn Schwa Deletion
— kaasafi narafi Schwa Epenthesis
tiin kaasan narsn Depalatalization
[tiin] [kaassn] [naran]
In each case, the palatal nasal becomes alveolar in coda position.
144 Studies in the Linguistic Sciences 20:1 (Spring 1990)
5. Conclusion
We have shown that Lama has two rules that strengthen sonorants to
stops after another sonorant. W-Strengthening changes /w/ to /p/ after /m/, and
R-Strengthening changes /r/ to the retroflex stop 161 after any sonorant
consonant. These two rules can be seen as part of a more general
phenomenon: the disfavoring of weak consonants in post-consonantal position.
Glide Deletion, which deletes a glide after any consonant, has a similar
function.
Murray and Vennemann (1983:520) propose the following principle:
(45) The Syllable Contact Law
The preference for a syllabic structure A$B, where A and B are
marginal segments and a and b are the Consonantal Strength of A
and B respectively, increases with the value of b minus a.
That is, in a heterosyllabic consonant cluster, the first consonant should ideally
be of high sonority, and the second should be of low sonority.
The following scale of consonantal strength — consistent with the one
proposed by Murray and Vennemann for Icelandic and Faroese, but making
fewer distinctions — will work for Lama:
(46) glides, r < I, nasals < obstruents
Glides, the weakest consonants, are not permitted after any consonant (except
for the homorganic sequence hy). This is not a prohibition on glides in onset
position; they are allowed after a vowel or in word-initial position, /mw/
sequences, with the second consonant weaker than the first, undergo W-
Strengthening, which results in the second consonant being stronger than the
first. All other consonant + glide sequences are resolved by Glide Deletion. It
should be noted that while W-Strengthening and R-Strengthening are
examples of what Vennemann (1988:50) terms "calibration". Glide Deletion
illustrates a rule type missing from his catalog of syllable contact changes,
namely onset deletion: A.B > .A0.7
Similarly, /r/, also a weak segment, is not permitted after any consonant in
Lama. R-Strengthening makes an Ixl stronger than a preceding sonorant.
R-Strengthening would not make an /r/ stronger than a preceding obstruent; in
these cases. Schwa Epenthesis applies, breaking up the cluster.
Nasals and /I/, of medium sonority, are allowed as either the first or the
second consonant in a cluster. The only deletion rule affecting /I/, Oral Sonorant
Deletion, is unrelated to the Syllable Contact Law, and applies only on the
second stratum, comparable clusters derived on the first stratum being
tolerated. No other rules affect the sonority of nasals or /I/.
Ourso & Ulrich: Sonorant-Strengthening in Lama 145
Obstruents, the strongest consonants, can follow sonorants. but cannot
precede any consonant. The prohibition of obstruents in codas holds for word-
final as well as preconsonantal position, and so cannot be attributed to the
Syllable Contact Law. Still, this prohibition, the related rule of P-Weakening,
and Schwa Epenthesis help to remove disfavored clusters.
The type of compensatory lengthening that accompanies the Lama
process of Schwa Deletion, where a vowel is lengthened to compensate for the
deletion of a vowel in the next syllable, is relatively uncommon, at least as a
synchronic rule. (f\/lost of the examples cited by Hayes (1989) are diachronic
changes.) Hayes has argued that compensatory lengthening can best be
handled within a theory with moras but no segmental skeleton. The
abandonment of the skeleton requires that glides be analyzed as
[+consonantal]; otherwise glides not in onsets could not be distinguished from
vowels. In Lama, glides pattern with other consonants in triggering Glide
Deletion and R-Hardening. These rules have been written to refer to the feature
[+consonantal].
Unfortunately, treating glides as [+consonantal] makes it difficult to identify
them as a natural class to the exclusion of liquids. The feature [+high] is surely a
prime candidate for underspecification in a language where /w/ alternates with
/p/, but not with /u/. Fortunately, there is no evidence that Glide Deletion does
not apply to /r/ as well as /w/ and /y/, so we need only exclude /I/. But it is not
difficult to imagine a language where such a solution is not possible, where /w/
and /y/ behave as a natural class to the exclusion of Irl. Such cases will have to
be solved as they arise.
NOTES
In addition to the ACAL, versions of this paper have been presented at
the University of Connecticut; Yale University; California State University,
Fresno; and the University of British Columbia. We are grateful to Eulalia Bonet,
Jennifer Cole, Hans Hock, Larry Hyman, Stanley Insler, Mike Kenstowicz,
Chuck Kisseberth, Brian McHugh, and others in those audiences for helpful
discussion, and to Diane Lillo-Martin and an anonymous reviewer for comments
on earlier versions of this paper.
^ We will write rules in linear form wherever doing so will not result in
unclarity. All deletion mles in this paper apply to segments (i.e., root nodes), not,
to individual features or to moras, and adjacency is determined on the root tier
for all rules.
We assume that the [continuant], [dorsal], [high], [back], and [round]
features of /w/ are filled in by redundancy rules. Within the theory of radical
underspecification, the underlying features of Lama labials would be as follows:
146 Studies in the Linguistic Sciences 20: 1 (Spring 1990)
/p/, [labial]; /f/, [labial, +continuant]; /m/, [labial, +nasal]; /w/, [labial, +sonorant];
IV.pl, [labial, dorsal].
2 The third voiced stop of Proto-Gurunsi, *g, has merged with /k/ in Lama.
3 The other lexically-conditioned allomorph of the noun class 2 suffix is
{-na}, which retains its consonant in all environments. The pronoun wa is used
to refer back to any class 2 noun.
4 The only apparent exception to Glide Deletion is /hy/, a homorganic
sequence resulting from place assimilation, as in (8). The only homorganic
sequence known to undergo the rule is the fake geminate /ww/, in kpewa (14). If
the failure of homorganic sequences to undergo Glide Deletion cannot be
predicted by universal principles, it must be stipulated as part of the rule of Glide
Deletion.
The given name of the first author of this paper might also appear to be an
exception to this rule. However, it is a three-word sentence, with a word
boundary before the /w/. Glide Deletion does not apply post-lexically.
5 The word for 'belt' contains the noun class 3 suffix {-u} after the agentive
suffix. The loss of schwa before a vowel-initial inflectional suffix is completely
regular.
6 Where Geminate Consolidation (28) has bled R-Strengthening, Schwa
Deletion will be blocked by the Linking Constraint (Hayes 1986).
7 Alternatively, this could be seen as a sequence of two processes:
tautosyllabification and glide deletion. Tautosyllabification, deriving a complex
onset by moving the syllable boundary, is motivated by the Syllable Contact
Law. Subsequent glide deletion would be motivated by the Head Law
(Vennemann 1988:13), which disfavors complex onsets.
REFERENCES
Bendor-Samuel, John T. 1971. Niger-Congo, Gur. Linguistics in Sub-Saharan
Africa, ed. by Thomas A. Sebeok, Current Trends in Linguistics 7.141-
178. The Hague: Mouton.
Hayes, Bruce. 1986. Inalterability in CV phonology. Language 62:2.321-352.
. 1989. Compensatory lengthening in morale phonology. Linguistic
Inquiry 20:2.253-306.
HOCK, Hans H. 1986. Compensatory lengthening: In defense of the concept
'mora'. Folia Linguistica 20.431-460.
Ourso & Ulrich: Sonorant-Strengthening in Lama 147
Kenstowicz, Michael. 1989. Compensatory lengthening: From West Africa to
East Slavic. Paper presented at the University of Connecticut.
Kenstowicz, Michael, Emmanuel Nikiema, & Meterwa Ourso. 1988. Tonal
polarity in two Gur languages. SLS 18:1.77-103.
Manessy, Gabriel. 1969. Les langues gurunsi. Paris: SELAF.
. 1979. Contribution a la classification genealogique des langues
voltaiques. Paris: CNRS.
Murray, Robert W., & Theo Vennemann. 1983. Sound change and syllable
structure in Germanic phonology. Language 59:3.514-528.
Ourso, Meterwa A. 1988. Root control, underspecification, and ATR harmony.
SLS 18:2.111-127.
. 1989. Lama phonology and morphology. University of Illinois Ph.D.
dissertation in Linguistics.
. Forthcoming. Phonological processes in the noun class system of Lama.
Studies in African Linguistics 20:2.
Prost, Andre. 1963. Lamba. Documents Linguistiques. Universite de Dakar.
Steriade, Donca. 1982. Greek prosodies and the nature of syllabification. MIT
Ph.D. dissertation in Linguistics.
Vennemann, Theo. 1988. Preference laws for syllable structure and the
explanation of sound change. Berlin: Mouton de Gruyter.
<
Ill
Tonology
Charles Kisseberth & Sheila Mmusi
The tonology of the object prefix in Setswana
NGESSIMO MUTAKA
The tone bearing unit in Kinande
4
Studies in the Linguistic Sciences
Volume 20, Number 1, Spring 1990
THE TONOLOGY OF THE OBJECT PREFIX IN SETSWANA
Charles W. Kisseberth and Sheila O. Mmusi
University of Illinois at Urbana-Champaign
In this paper we will explore the tonal behavior of object prefixes
in Setswana. Object prefixes are underlyingly associated with a H
tone. When the object prefix appears in conjunction with a toneless
verbstem, this H will spread onto a following vowel (or vowels) by
virtue of the independently motivated rules of High Tone Spread and
Phrasefinai Spread. The resulting multiply-linked H structures are
then subject to independently-motivated rules of delinking.
After analyzing the interaction between object prefixes and
toneless verbstems, we examine the case where a H object prefix
precedes a H verbstem. We argue that these two H tones are subject
to the OCP and are reduced to a single, mutiply-linked H tone. We
base this argument on the fact that these structures are subject to
rules that affect multiply-linked H tones (and not singly-linked H
tones). Finally, we establish that when Setswana has two object
prefixes in front of the verbstem, the OCP does not affect the first
object prefix H. Rather, the first object prefix maintains its own H tone,
while the second object prefix and the verbstem have their H's
reduced to a single H by the OCP.
This paper explores the tonal behavior of object prefixes in Setswana. In
order to properly understand and analyse the behavior of the object prefixes,
we must venture into several of the most essential tonal phenomena in
Setswana. Considerations of space force us to provide merely a sketch of the
motivation for some of the tonal principles that we identify. It is our hope,
however, that we have provided a basic introduction not only to the object prefix
tonology of Setswana, but also to some more general features of the tonal
structure of the language as a whole.
We will adopt an analysis of Setswana tonology wherein the underlying
representations of morphemes are only High tones and no Low tones. We will
assume that Low tones are inserted by default on syllables that are not
associated with High tones. Given this basic assumption, it is fairly easy to
motivate the following claims about the nature of Setswana tonology:
152 Studies in the Linguistic Sciences 20:1 (Spring 1990)
Hypothesis A
Verb stems fall into two lexical tonal types: High verb stems and Toneless
verb stems. The high tone of the High verb stems can be presumed to be
unassociated underlyingly.
The first person, present indicative affirmative verbal forms in (1) illustrate
this basic division of the verb stems.
(1) a. ke-a-wa 'I am falling'
ke-a-lwa 'I am fighting'
ke-a-lema 'I am cultivating
ke-a-baka 'I am praising'
ke-a-lebala 'I am forgetting'
ke-a-hun£la 'I am tying'
b. ke-a-ja
'1 am eating'
ke-ja
ke-a-rska/rskai
'1 am buying'
ke-r£ka
ke-a-r£kfsa/r£kfsa
'1 am selling'
ke-r£kisa
ke-a-b£r£k£la
'1 am working for'
ke-b£r£k£la
In (1a) we have examples of Toneless verb stems, whereas (lb) illustrates
High verb stems. In (lb) we have shown H verb stems both phrase-finally and
phrase-medially for reasons that will be developed below.
Hypothesis B
Subject prefixes in certain verbal forms (e.g. the present indicative
affirmative) fall into two tonal types: High subject prefixes and Toneless
subject prefixes. There is no particularly compelling basis for deciding
whether the H-toned subject prefixes have an unassociated H or an
associated H. In either case, we must guarantee that this H ultimately
associates to the subject prefix.
The Toneless subject prefix /ke/ was illustrated in (1). A High toned subject
prefix /o/ is illustrated in (2).
(2) a. 6-a-wa/wa '(s)he is falling'
6-a-lema '(s)he is cultivating'
6-a-lebala '(s)he forgets'
b. 6-a-ja '(s)he is eating'
6-a-r£ka/r£ka '(s)he is buying'
6-a-r£kisa/r£kisa '(s)he is selling'
6-a-b£r£k£la '(s)he is working for'
Hypothesis C
The lexical H of a verb stem associates to the first stem Tone-Bearing-
Unit (TBU) of the verb stem — presumably the result of a universal
principle of tone association that anchors a free tone to the leftmost free
TBU in the relevant domain ( in this case, the verbstem).
The validity of this claim is demonstrated by the fact that all H verbs have a
H tone associated with their first stem TBU in the data in (1 ) and (2).
Kisseberth & Mmusi: The tonology of the object prefix in Setswana 153
Hypothesis D
There is a rule of High Tone Spread, which is formulated as in (3)
(3) High Tone Spread (HTS)
H
N
X X
The motivation for (3) can be seen in the fact that a H verb stem will
regularly not only have a H tone on the first stem TBU, but on the second as
well. This fact is illustrated by the data in (1) and (2) above. A second piece of
evidence in favor of (3) is provided by the data in (2a) — e.g. 6-a-lebala —
where we see that the H tone contributed by the subject prefix has been
extended over onto the tense/aspect marker /a/. That /a/ is essentially toneless
as is demonstrated by the data in (1).
Hypothesis E
There is a rule of Optional Phrase-Final Spread, which is formulated in
(4), which has the effect of spreading a H associated to the penult TBU of
the intonational phrase onto the final TBU of that phrase. Optional
Phrase-Final Spread must be applied to the output of High Tone Spread,
as shown in (5).
(4) Optional Phrase-Final Spread (PFS)
H
K
X X % (where %= intonational phrase boundary)
(5) H
ke-a-rskisa (underlying)
H
I
ke-a-r&kisa (initial tone association)
H
f\
ke-a-rdkisa (High Tone Spread)
rv.
ke-a-r£kisa (Optional Phrase Final Spread)
The evidence that rule (4) operates only phrase-finally is provided by the
phrase-medial data cited in (lb) — e.g. ke-r&kisa. — where we see that the only
154 Studies in the Linguistic Sciences 20:1 (Spring 1990)
High-toned syllables are the stem-initial syllable and the immediately following
syllable (which gets its H through the rule of High Tone Spread).
Hypothesis F
There is a rule of Left-Branch Delinking, which is formulated in (6)
(6) Left Branch Delinking
H H
Left-Branch Delinking (LBD) says that if there is a multiply-linked H tone,
and if that H tone is preceded by another H tone, the leftmost branch of the
multiply-linked H is deleted.
The evidence for (6) is quite strong. Consider, for example, the potential
construction. This construction is characterized by a prefix /ka/ that is High-
toned. In (7) we illustrate what happens when that prefix is adjoined to a High
verb stem.
(7) n-ka-ja 'I can eat'
n-ka-r&ka 'I can buy'
n-ka-r£kisa/r£kisa 'I can sell'
n-ka-b&r£k£la 'I can work for'
We will not discuss the case of a monosyllabic verb such as /ja/, but rather
concentrate on the remaining examples. Notice that in every case the first
syllable of the High verbstem is realized with a low tone. Rule (6) explains this
as follows: These verbstems, of course, have a H tone associated with their first
stem vowel (after initial tone association takes place ). Then High Tone Spread
extends this H tone onto the second stem vowel. But this application of High
Tone Spread creates the environment for Left-Branch Delinking to apply. The
result is that the initial vowel of the verbstem is delinked from the H and
subsequently receives a low tone by default.
Hypothesis G
Either, High Tone Spread is constrained so as not to produce an output
that violates the Obligatory Contour Principle (OCP) (i.e. High Tone
Spread is constrained so that it will not spread a H onto a TBU that is
followed by a H-toned TBU), or alternatively, High Tone Spread applies
wherever possible and there is a rule of Right-Branch Delinking (RBD)
formulated in (8).
Kisseberth & Mmusi: The tonology of the object prefix in Setswana 155
(8) Right-Branch Delinking (RBD)
H H
N
XXX
For the sake of ease of discussion, we will assume the Right-Branch
Delinking analysis. (8) of course is simply the mirror image of (6). Evidence for
Right-Branch Delinking comes from the data in (2b) — e.g. o-a-reka, where we
see that the H tone of the subject prefix is not also associated with the following
tense/aspect prefix /a/ even though the rule of High Tone Spread predicts that it
should be. We suggest that it is the rule of Right-Branch Delinking in (8),
operating on the output of High Tone Spread, that predicts that /a/ will be
delinked from the H of the subject prefix.
Given the preceding discussion, we are now in a position to examine the
tonology of the object prefixes in Setswana. Because of space limitations we
will not illustrate all of the object prefixes in Setswana. The few object prefixes
that we use in the examples are intended to be taken as generally
representative of all of the object prefixes.
First of all, it is easy to determine immediately that the object prefixes in
Setswana are fundamentally High-toned and that they behave much like other
High syllables. The data in (10), showing object prefixes in front of Toneless
verb stems in the present indicative affirmative, are designed to establish these
points.
(10) ke-a-mo-tswa 'I am staying away from him'
ke-a-e-lema/lema 'I am cultivating it'
ke-a-ba-lem£la 'I am cultivating for them'
The data in (10) demonstrates that the object prefix is associated to a H
tone and that this H tone spreads onto the immediately following TBU by virtue
of HTS (see the example ke-a-ba-lemela). The example ke-a-e-lema shows that
when the H of the object prefix has spread onto the penult TBU of the phrase,
the final TBU of the phrase can receive the H tone by virtue of Optional Phrase-
Final Spread.
At this point we should note two important points. First, it is possible for the
object prefix to optionally delink from its H tone after that H has spread onto the
first TBU of the verbstem. The only case where this is not possible in our data is
when the first stem TBU is in fact a monosyllabic stem. We have not studied in
any detail the question of whether there is any functional difference between
forms where the object prefixes delinked from its H and forms where the object
prefix remains linked. The second point is that in some varieties of Setswana
object prefixes may be treated as toneless underlyingly. If they are so treated
then of course there is no H to spread onto a following TBU. Throughout this
paper we ignore both the cases where the object prefix optionally delinks and
cases where the object prefix is treated as toneless.
156 Studies in the Linguistic Sciences 20:1 (Spring 1990)
Given that the object prefixes, when they appear in conjunction with a
Toneless verb stem, are associated with a H tone (which spreads according to
the same pattern as the lexical H of a verb stem), it is imperative that we
immediately consider the case where an object prefix is used in conjunction
with a High verbstem. Example (1 1) illustrates this situation:
(1 1 ) ke-a-di- ja/ja 'I am eating them'
ke-a-df-r£ka/r£ka 'I am buying them'
ke-a-di'-r£k[sa/r£kisa 'I am selling them'
Notice that the stem tone pattern here is exactly the same as when no
object prefix is present. Furthermore, the object prefix is realized on a high pitch.
Assuming that there is a H tone contributed by the object prefix and also a H
tone contributed by the verb stem proper, and assuming that we could predict
that the first of these H tones would anchor to the object prefix and the second to
the verb stem proper, then the surface forms shown in (11) do not require that
we postulate any rules in addition to HTS and Optional PFS.
However, it might be suggested that a H object prefix followed by a H verb
stem represents a violation of the Obligatory Contour Principle (which disallows
successive identical autosegments on the same tier). Let us now raise the issue
of whether there is any way to invoke the OCR and yet derive the forms in (1 1 ).
Suppose that the OCR would collapse these two H tones to a single H
tone before the tonal association principle works. That is, suppose we were to
first apply the OCR and then associate the resulting single H.
We will consider two possibilities. The first possibility is that the OCR fuses
the H of the object prefix and the H of the verb stem before either of these
Highs has undergone the initial association process. The result would be that in
the verbal complex consisting of the object prefix and the verb stem, there
would just be a single H tone at the point where initial tone association takes
place. Presumably the result of tone association would be to link that H tone to
the first element in the morphological domain in question — namely, the object
prefix. While this will correctly yield a H tone on the object prefix, it will fail to
derive the correct pronunciation in a case like ke-a-e-reki'sa. The partial
derivation in (12) illustrates this point:
(12) H
ke-a-e-r£kisa OCR result
H
I
ke-a-e-r£kisa Initial Tone Association
H
ke-a-e-r£kisa HTS
inapplicable Optional PFS
Kisseberth & Mmusi: The tonology of the object prefix in Setswana 157
The form * ke-a-e-rikisa is incorrect in that the H tone does not extend to
the second syllable as well as all the way to the final syllable through Optional
Phrase-final Spread to derive ke-a-e-rekisa/ke-a-e-rskisa which is the correct
pronunciation.
It cannot be the case then, that the OCP operates on two unassociated H's
and reduces them to one, with subsequent association. What sort of alternative
story can we tell? Suppose that we adopt a cyclic application of the universal
tone association principle so that on the first cycle, the H contributed by the verb
stem proper associates to the first TBU of that stem, while on the second cycle
the H contributed by the object prefix anchors to the object prefix. If we then
allow the application of the OCP to reduce these two H tones to one, while
retaining all of the linkages that these H tones have, we will end up with the
appropriate input to HTS and Optional PFS. This is demonstrated in (13).
(13) H H
[ ke a [ e [ rskisa] ] ]
H H
I
[ ke a [ e [ rskisa ] ] ] first cycle tone association
H H
I I
[ ke a [ e r&kisa ] ] ] second cycle tone association
H
/\
[ ke a [ e [ r£ kisa ] ] ] OCP
n
[kea[e[r£kisa]]] HTS
H
/Tnn
[kea[e[r£kisa]]] Optional PFS
We shall next argue that (a) we cannot simply do nothing and allow the H
of the object prefix and the H of the verb stem to stand as two successive
separate H tones, and (b) the correct solution is to reduce the two H tones to a
single, multiply-linked H tone as in the derivation above. In order to motivate the
claim that the H of the object prefix and the H of the verbstem must be fused into
a single, multiply-linked H tone, we need to consider certain cases where the
158 Studies in the Linguistic Sciences 20:1 (Spring 1990)
object prefix obligatorily is realized on a low tone rather than on the expected
High tone. Consider the data in (14).
(14) n-ka-e-ja "I can eat It'
n-ka-di-r£ka 'I can buy them'
n-ka-e-r&kisa/rskfsa "I can sell if
n-ka-mo-tswa 'I can stay away from him'
n-ka-e-lema/lema 'I can cultivate it'
n-ka-mo-lemila "I can cultivate for him'
Notice that when the H-toned potential marker precedes the object prefix,
the object prefix is in every case pronounced on a low tone. The data from the
Toneless verb stems establishes clearly that the object prefix in this construction
is fundamentally High-toned, since the H of the object prefix has spread onto
the Toneless verbstem producing forms such as n-ka-mo-tswa and
n-ka-mo-lemela. But even though the object prefixes are basically High, they
are pronounced here on a low tone.
There is of course a very easy solution to this problem. Recall the rule of
Left-Branch Delinking discussed earlier. This rule delinks the element on the
leftmost branch of a multiply-linked H tone just in the event a H tone is anchored
to the preceding TBU. In the case of the toneless verb stems in (14) above, it is
obviously Left-Branch Delinking that is at work. Example (15) illustrates.
(15) H H
n-ka-mo-lem£la (underlyingly after tone association)
H H
h
n-ka-mo-lem£la High Tone Spread
H H
f\
n-ka-mo-lem£la Left Branch Delinking
We see that since the H of the object prefix spreads onto the Toneless
verbstem, there is a branching H tone preceded by a H and therefore Left-
Branch Delinking will come into play to delink the object prefix.
But now we must consider the case of the H verb stems in (14). In order for
Left-Branch Delinking to apply and account for the low-tone realization of the
object prefix, the object prefix must be on the leftmost branch of a multiply linked
H structure. If we assume that the object prefix H and the verbstem H are not
affected by the OCP — i.e. that they remain two separate High tones on the
tonal tier, then the object prefix will not be on the left-branch of a multiply-linked
â– /r^!
Kisseberth & Mmusi: The tonology of the object prefix in Setswana 159
H stmcture. As a result, Left-Branch Delinking should not apply to it. Thus such
an approach would fail to account for the data in (14). On the other hand, if the
OCP fuses the object prefix H and the verb stem H into a single H tone, we
correctly predict that the object prefix will be on the left-branch of a multiply-
linked H tone and thus will be subject to the rule of Left-Branch Delinking. And
thus we will make the correct predictions about the data in (14) like
n-ka-e-rekisa/rekisa.
The above argument for the fusion of the object prefix H and the verb stem
H into a single multiply-linked H tone depends on the claim that Left-Branch
Delinking affects only branching H tones. There is a fair amount of evidence to
support this claim since there are several situations where a sequence of
singly-linked H tones do not undergo any sort of delinking. We will confine
ourselves to one example involving object prefixes.
(16) ga-ke-e-leme 'I am not cultivating it'
ga-ke-ba-lem£le/lem£le 'I am not cultivating for them'
The data in (16) involve a negative tense where (a) the subject prefix /ke/
is associated with a H tone and (b) where there is a morphological H tone that is
assigned to the second syllable of the Toneless verbstem. Because the lexically
Toneless verbstem has a morphologically-induced H tone on its second TBU,
the H of the object prefix is unable to be linked to the following TBU (due to the
effect of Right-Branch Delinking). The result is that the object prefix remains
(singly-) linked to a High tone. Notice that this singly-linked object prefix is able
to remain High-toned after the H subject prefix. We conclude that it is only
multiply-linked H tones that are subject to Left-Branch Delinking and therefore
that the object prefix must undergo the OCP-induced fusion.
There is one very interesting case that we will take up at this point.
Consider the present indicative affirmative third person subject form:
(17) 6-a-di-ja (preferred, but also: 6-a-di-ja possible)
'(s)he is eating them (food)'
6-a-e-r£ka, or 6-a-e-rdka '(s)he is buying it'
6-a-mo-r£k£la or 6-a-m6-r£k£la '(s)he is buying for...'
6-a-mo-tswa (preferred, but also: 6-a-m6-tswa)
'(s)he is getting rid of him/her'
6-a-e-lema or 6-a-e-lema '(s)he is cultivating it'
6-a-mo-lem£la or 6-a-m6-lem£la '(s)he is cultivating for ..."
These data involve an issue that we do not have space to explore here —
namely, the issue of the interaction of Left-Branch Delinking and Right-Branch
Delinking. But what seems apparent from (17) is that in the preferred
pronunciation, application of Left-Branch Delinking to the object prefix allows
the /a/ prefix to avoid being delinked by Right-Branch Delinking. In the
alternative pronunciation, it seems that the delinking of /a/ by Right-Branch
160 Studies in the Linguistic Sciences 20:1 (Spring 1990)
Delinking removes the object prefix from the environment for Left-Branch
Delinking.
We now turn to our concluding observation about object prefixes in
Setswana. Setswana differs from the other Sotho languages in allowing two
object prefixes in the same verbal word. For example,
(18) a. ke-a-mo-e-fa 'I am giving him/her it' (preferred)
ke-a-e-mo-fa (ibid)
ke-a-mo-e-bontsha 'I am showing him/her it'
ke-a-e-mo-bontsha (ibid)
ke-a-m6-e-r£k£la 'I am buying it for him/her*
ke-a-e-mo-r&k&la (ibid)
ke-a-mo-e-rdkis&tsa 'I am selling it for/to him/her*
ke-a-e-mo-r£kis£tsa (ibid)
b. ke-a-m6-e-tl£la 'I am bringing it for/to him/her'
ke-a-e-mo-tl£la (ibid)
ke-a-m6-e-lem£la 'I am cultivating it for him/her*
ke-a-e-mo-lem£la (ibid)
ke-a-m6-e-kg6rom£l£tsa 'I am pushing it to/for him/her'
ke-a-e-mo-kg6rom£l£tsa (ibid)
In (18a) we illustrate High verb stems with two object prefixes and in (18b)
we illustrate Toneless verbstems with two object prefixes. Notice that the H tone
associated to the first OP triggers LBD of the second OP. This requires that (a)
the H of the first OP does not collapse via the OCP with the H of the second OP,
and (b) the H of the second OP must be involved in a multiply-linked structure
(originating either from HTS or from the OCP-induced fusion of an object prefix
H and the verb stem H). Notice that the retention of the High tone on the first
object prefix provides further evidence that delinking rules do not affect singly-
linked High tones.
What can we conclude from the data in (18)? It appears quite clear: the
OCP-induced fusion affects the object prefix immediately adjacent to the verb
stem. It does not fuse the two Highs of the two object prefixes. And we think that
further study will demonstrate that it does operate anywhere outside the
complex consisting of the object immediately in front of the verb stem and the
verb.
In conclusion, we have demonstrated that object prefixes in Setswana are
High-toned and that their H tone spreads and delinks according to the same
pattern as other High tones in the language. Furthermore, we have argued that
the H of the object prefix and the H of an immediately following verb stem must
be fused by the OCP into a single, multiply-linked H. Finally we have argued
that in a sequence of object prefixes, only the last of the sequence is subject to
Kisseberth & Mmusi: The tonology of the object prefix in Setswana 161
this OCP-induced fusion with a H verb stem. Furthermore, the two object
prefixes H's are not themselves fused together into a single H.
NOTES
1 The dialect of Setswana under discussion has the option of surfacing
with a high tone or low tone in phrasefinal position. This matter is not addressed
in this paper. For further information concerning this phenomenon the reader is
referred to the unpublished University of Illinois Ph.D dissertation by Sheila
Mmusi (1992).
REFERENCES
COLE, D. T. 1955. An introduction to Setswana grammar. Cape Town, South
Africa: Longmans, Green and Co.
Mmusi, Sheila O. 1992. An autosegmental analysis of some aspects of
Setswana tone. University of Illinois at Urbana-Champaign Ph.D.
dissertation in Linguistics.
Studies in the Linguistic Sciences
Volume 20, Number 1, Spring 1990
THE TONE BEARING UNIT IN KINANDE
Ngessimo Mutaka
University of Southern California, Los Angeles
The Tone Bearing Unit in Kinande must be one which is comprised
of vowel features only in the melodic structure. Evidence for this is
presented from the consideration of the gliding of the causitive and
passive morphemes, providing that this process is postlexical.
This paper addresses the status of the Tone Bearing Unit (TBU) in
Kinande, a Bantu language spoken in Eastern Zaire, with respect to the vowels
of the causative morpheme {i} and the passive morpheme {u} that devocalise
lexically. The example in (1) illustrates better the problem.
(1) a. tu-a-gul-irir-a-y-a-a [twaguliriraya] 'we sold anyway'
b. tu-a-gul-irir-a-y-a Valinande 'we sold Valinande anyway'
The glide [y] is the causative morpheme {i} as in eh-gul-i-a 'to sell'. In the
recent and remote past tense, it behaves like a consonant and triggers the
insertion of the vowel [a] at its left. The question is: can this causative {i} be a
TBU and behave like a consonant at the same time? In order to answer this
question, I will first show that a TBU in Kinande is invariably a vowel, then I will
discuss some rules specific to the causative morpheme, and finally I will show
that a vowel that devocalises lexically maintains its status of TBU as long as it
does not contain any consonantal feature in its melodic structure.
Let us first explore what a TBU in Kinande is by considering the phrasal H
tone (discussed in Hyman in press; Hyman & Valinande 1985) that docks on the
penultimate vowel at the end of a phrase. This H will be underlined throughout
the examples.
(2) a. eri-hum-i'r-a 'to hit for'
R EXT FV
tu-kandi-hiim-a 'we'll hit'
eri-tum-a 'to send'
eri'-to-er-a [eritwera] 'to dig for'
eri-tu-ir-a [eri'twira] 'to cut for'
tu-kandf-tum-a 'we'll send'
b. eri-hum-ir-a Valinande 'to hit for Valinande'
tu-kandi-hum-a Valinande 'we will hit Valinande'
eri-tum-a Valinande 'to send Valinande'
164 Studies in the Linguistic Sciences 20:1 (Spring 1990)
In these forms, the root -hum- is underlyingly toneless and the roots -to-,
-tu-, -turn- are underlyingly H. The morphemes {ir} and {er} are extensions. The
phrasal H docks on the penultimate vowel in (2a) but this H is not assigned on
the verb in (2b), because the verb is not at the end of a phrase. Consider, for
instance, the utterances in (3).
(3) a. eri-s6-a [eriswa] 'to grind"
eri-tu-a [eritwa] 'to cut'
eri-to-a [eritwa] 'to dig'
b. tu-ka-hum-a-a [tukahuma] 'we are hitting'
tu-ka-tum-a-a [tukatuma] 'we are sending'
In these forms, the phrasal H appears on the penultimate vowel as the first
member of the falling tone on the FV. At this point, I would like to point out that
gliding in Kinande is postlexicaP. The phrasal H is assigned to the strings in
(3a) before gliding takes place. Consider, in contrast, the data in (4).
(4) eri-way-a 'to wander'
/way-a/ */uai-a/
eri-hay-a 'to storm at someone'
/hay-a/ */hai-a/
eri-goy-a [erigoya] 'to woo a girl'
eri-tey-a [eriteya] 'to take'
I assume that the glide in these forms is underlying. An underlying glide
does not act as a TBU. It behaves like a consonant with respect to the phrasal
H. Support for the fact that it behaves like a consonant also comes from the
processes of height harmony in Kinande whereby /!/ becomes [e] in the
extension if the root vowel is [-hi, -lo]. This is illustrated in (5).
(5) eri-goy-er-a /e-ri-goy-ir-a/ 'to woo a girl for someone'
eri-tey-er-a /e-ri-tey-ir-a/ 'to get tired for'
The forms below show that a glide acts as a TBU if it is a vowel underlyingly.
(6) a. eri-wS
'to fair
/eri-u-a/
~ V -cv -cv
III 1
e r i u
*/eri-wa/
eri-owS 'to hear'
/ou-a/ ~ V -CV -VCV -V
I II I I I
e r i o u a */ow-a/
eri-mu-ou-a [erimwowa] 'to hear him'
b. eri-otV-a [eryota] 'to light'
eri-mu-6t-a [erimwota] 'to light him'
eri-ot-er-a [eryotera] 'to light for'
Mutaka: The Tone Bearing Unit in Kinande 165
In the word eryota, I have introduced an extra V that is inserted before the
FV to meet a constraint in Kinande that a stem must have at least two vowels.
The analysis I assume is that Kinande has two lexical strataJ At the first stratum,
the stem formatives and the object marker are affixed. At stratum 2, the other
formatives are affixed. The form erimwota meets that constraint because, at
stratum 2, the object marker will have been affixed to the stem and its consonant
is therefore able to serve as the onset to the initial vowel of the stem. Since the
phrasal H surfaces on the FV in en'-mu-ow-a, we must conclude that its
penultimate segment is a vowel. If it were a consonant, the phrasal H would
have appeared on the penultimate vowel as in eri-mu-6t-a. It should also be
pointed out that no data were found where the glide is underlyingly a vowel in
disyllabic words or in words of more than two syllables. It may be possible that
the strings where I have posited a vowel underlyingly may also have a
consonant as indicated in (6a) on the right of the tilda symbol. The vowel would
be branching on both the vowel and the consonant underlyingly. In sum, we
have established that the TBU in Kinande is a vowel and that it cannot be a
glide.
Before showing whether a vowel that devocalises remains a TBU, we also
need to establish some rules that are morphologically conditioned by the recent
or remote past tense in a form with the causative vowel. Consider the following
examples:
(7) Recent past tense with mo-
a. mo-tu-a-mu-hum-ir-a 'we hit for him'
mo-tu-a-hiim-ir-a 'we hit for'
mo-tu-a-gul-ir-a 'we bought for'
mo-tu-a-mu-gul-ir-a 'we bought for him'
mo-tu-a-giil-ir-a Valinande 'we bought for Valinande'
b. mo-tu-a-mu-gul-irir-i'-a [motwamugulirirya]
'we sold him anyway'
mo-tu-a-gul-irir-i-a [motwagulirirya]
'we sold anyway'
mo-tu-agul-irir-i-a [motwagulirirya] Valinande
'we sold Valinande anyway'
Assuming Mutaka (1991), at stratum 2, the recent past tense assigns a
suffixal HL. The H docks onto the first root vowel and the L is inserted onto the
final vowel by rule in a toneless verb as illustrated in (7a). The H in the falling
tone on the final vowel must be the result of a phrasal H as it disappears if the
word is not at the end of a phrase. The H on the right of the root vowel must be
caused by the causative morpheme in the recent past tense as it does not
surface in a form without the causative morpheme. Since the phrasal H appears
in this form with the causative morpheme, two conclusions appear to be
unavoidable here: a morphologically conditioned rule deletes the L assigned by
the recent past tense and a H is also assigned to the causative vowel [i]. Notice
that this H of the causative morpheme cannot be underlying as illustrated in (8).
166 Studies in the Linguistic Sciences 20:1 (Spring 1990)
(8) tu-kandi-gul-a 'we will buy'
tu-kandl-gul-i-a [tukandigulya] 'we will sell'
tu-kandi-gul-irir-i-a 'we will sell anyway'
The vowel preceding the causative {i} does not surface with a H. A first
approximation of the formulation of these rules is given in (9).
(9) a. Causative H: V Condition: V is the causative {i}
I Tense: Recent Past
I
H
b. Causative L deletion:
V V] Condition: the preceding V is the causative {i}
I
L ^ Tense: Recent past
These rules are illustrated in (10)
(10) mo-tu-a-gul-ir-a mo-tu-a-gijl-irfr-i'-a
Stratum 1 :
gul-ir-a gul-irir-i-a
Stratum 2:
m6-tu-a-[mu-gul-ir-a m6-tu-a-[mu[gul-irir-i-a
II II
H L H L
(Assignment of suffixal HL)
m6-tu-a-[mu-[gul-ir-a m6-tu-a-[mu-[gul-irir-i- a
II III
H L H H L
(Causative L deletion + causative H) i
Stratum 3:
mo-tu-a-mu-gul-ir-a mo-tu-a-mu-gul-irir-i-a
II II
H L H H
(H spreading)
mo-tu-a-mu-gul-ir-a mo-tu-a-mu-gul-irir-i-a
II II
H L H H
(PHA)
(Other rules: e.g. gliding, default L)
Output: motwamugulira motwamiJguliriryS
In these derivations, it is not clear that the H on the final vowel is the result
of the PHA rule. The following forms in another recent past tense shows it more
clearly.
Mutaka: The Tone Bearing Unit in Kinande 167
(11) Recent past tense.
a. tu-a-mu-hum-ir-a-a [twamuhumira]
'we hit for hinn* from eri-hum-a 'to hit'
tu-a-hum-ir-a-a [twahumira]
'we hit for'
tu-a-gul-ir-a-a [twagulira]
'we bought for' from eri-gul-a 'to buy'
tu-a-mu-gul-ir-a-a 'we bought for him'
b. tu-a-gul-ir-a-i-a-a [twaguliraya]
'we sold for' from eri-gul-f-a 'to sell'
tu-a-mu-giil-ir-a-i-a-a [twamuguliraya] 'we sold for him'
c. tu-a-tum-irir-a-a 'we sent anyway'
tu-a-mu-tum-irir-a-a 'we sent him anyway'
The tones in the stem of the facts in (11) resembles the ones seen in the
recent past tense with mo-. The only difference here is that there is an extra FV
as best illustrated with a H tone verb in (1 1 c). The glide [y] in the stem in (1 1 b) is
the causative morpheme. It is preceded by the vowel [a] which is epenthetic.
The H on this vowel must also be the result of the H Spreading at Stratum 3 and
it originates from the causative vowel. The H on the penultimate vowel in (1 lb)
is the phrasal H which disappears when the word is not at the end of a phrase.
Two questions necessitate answers with respect to these data in (lib). First,
why is there an epenthetic vowel? Secondly, where does the H on the
epenthetic vowel [a] originate?
The answer to the first question is simple. In Kinande, there is a prohibition
of a sequence of consonants (unless it is a homorganic nasal with a consonant
as in e-ndi 'belly', e-mbene 'goat' or a consonant with a semivowel as in
eri-to-a [erftwa] 'to dig' or a combination of Nasal + homorganic consonant +
semivowel as in embw^ 'dog'. The semivowel in a cluster consonant +
semivowel will always result from an underlying vowel. In order to explain the
presence of the epenthetic vowel [a], we must assume that the causative {i}
becomes the onset of a syllable and thus behaves like a consonant by the
causative syllabification rule formulated in (12).
(12) Causative syllabification
C V
\
\
i Condition: certain tenses e.g. Recent past.
That this causative syllabification rule is restricted to certain tenses is
illustrated in (13).
168 Studies in the Linguistic Sciences 20:1 (Spring 1990)
(13) a. tu-ka-gul-a-y-a-a /tu-ka-gul-i-a/
[tukagulaya] 'we are selling' (Present tense)
tu-a-gul-a-y-a /tu-a-gul-i-a/
[twagulaya] 'we bought' (Remote past)
tu-ana-gul-a-y-a /tu-ana-gul-i-a/
[twanagulaya] 'we did sell'
b. tu-ka-gul-i'-a /tu-ka-gul-i-a/
[tukagulya] 'while we were selling'
tu-a-gul-i'-a /tu-a-gul-i-a/
[twagulya] 'in that case, we will sell'
tu-ana-gul-i'-a /tu-ana-gul-i-a/
[twanagulya] 'as we were selling'
As can be seen in these forms, the cCausative syllabification cannot be
phonological because it does not occur in other tenses (as in 13b) that have
similar segments underlyingly (shown here in the forms between back slashes)
as the tense in which it applies. Notice also that the causative H posited above
does not necessarily appear in the forms where the causative syllabification
rule applies as the vowel that precedes this causative does not always surface
with a H as illustrated in tu-ka-gul-a-y-a 'we are selling'. It should also be
pointed out that the vowel [a] is the default vowel in Kinande as argued in
Mutaka (1987; Schlindwein 1988; and Archangel! 1988). If any vowel would be
inserted to break up a cluster, we would rightly expect it to be the default vowel
[a].
To answer the second question, we can construe that the H on the
epenthetic vowel [a] originates from the causative vowel [i] by means of the rule
of H spreading that was discussed earlier. This means that the causative H also
applied in this form. The question now is: since H spreading applies at stratum
3, and the causative syllabification rule applies at stratum 2, could this H spread
from the causative {i} that is already the onset of a syllable? In other words,
could this H originate from the causative {i} that has presumably the feature
[-syll] which is a consonantal feature?
Assuming that the causative syllabification rule applies at stratum 2, notice
that we cannot construe that, after the causative {i} devocalized at stratum 2 to
become the onset of the syllable, the H that was assigned to it by the causative
H, linked to this default vowel. If it were so, we would have expected this H to
spread one more time at stratum 3 and surface on the vowel that precedes the
epenthetic vowel as in *twamuguliraya (cf. twamuguliraya). A second possibility
would be that, after the causative syllabification rule, the H assigned to the
causative vowel by the causative H would dock on the head of the syllable, that
is, on the vowel [a] at the right of {y}. This would mean that the H in the falling
tone in a form like twamuguliriraya is the result of this causative H. That this is
not so is illustrated by the forms in (14).
Mutaka: The Tone Bearing Unit in Kinande
169
(14) a. tw-a-mu-gul-irir-a-y-a-a Valinande
[twamuguliriraya] 'we sold him Valinande anyway'
tw-a-mu-gul-irir-a-y-a-a
[twamuguliriraya] 'we sold him anyway"
b. tw-a-mu-tum-iri'r-a-a
[twamutmirfra] 'we sent him any way'
tw-a-mu-tum-irir-a-a Valinande
[twamutumirira] 'we sent him Valinande anyway'
In the H tone form in (14b), the H in the falling tone on the FV is the result
of the suffixal H at stratum 2. When the word is not at the end of a phrase, this H
does not disappear. However, the one in (14a) disappears. This contrast shows
that this H must be the result of the phrasal H unlike the one in (14b), indicating
that the H of the causative {i} must have originated from this causative {i}
although it underwent causative syllabification to become the onset of a
syllable. 2 This observation, however, appears to be a counterintuitive
conclusion in that it amounts to saying that a TBU can bear a consonantal
feature in Kinande. In what follows, I would like to argue that this can happen
only if the consonantal feature is not a melodic feature but a prosodic feature (a
la Archangeli & Pulleyblank 1988). By melodic features, I mean the features that
are below the skeletal tier as opposed to prosodic features like [-syll] or syllable,
foot and minimal word.
Consider the following hierarchical representation that is adapted^ from
Clements (1985) and Sagey (1986) in (15).
(15)
ant dist / / \ \
(Foot)
(Syllable)
skeletal tier or CV tier
Root node
Laryngeal node
Tonal node
Supralaryngeal node
Labial node
Coronal node
Dorsal node
hi bk lo ATR
Notice that the array of features used in this hierarchical representation
are about the melodic structure of a segment. A segment will be identified as a
170
Studies in the Linguistic Sciences 20:1 (Spring 1990)
consonant or a vowel as long as it has consonantal or vowel features in its
structure. A feature like [-syll] cannot be viewed as a melodic feature in that it
encodes a position on the skeleton about what can constitute the onset of a
syllable but it tells us nothing about the melodic content of the segment. If the
condition to be a TBU is for a segment to have only vowel features and if a
vowel that has the feature [-syll] only gets a consonantal feature when it
undergoes the rule of gliding, then we can construe that, as long as gliding has
not applied, the causative morpheme in (14a) will safely behave as a TBU
although it is assigned the feature [-syll] lexically. Presumably, the consonantal
feature to appear in the melodic content of a [+hi] vowel as caused by gliding
will be [cons]. A sample derivation of this form can now be presented in (16).
(16)
tu-a-mu-gul-ir-a-f-
tu-a-mu-gul-irir-a-a
Stratum 1 :
Stem: gul-ir-i-a
OM: mu-gul-ir-i-a
gul-irir-a
mu-gul-irir-a
Stratum 2: affixation of the TM + SM
tu-a-[mu-[gul-ir-i-a-a
I I
H L
(Suffixal H + Suffixal L)
tu-a-[mu-[gul-irir-a-a
I I
H L
@ (Syllable)
-syll I
XXX (skeleton)
I
tu-a-[mu-[gul-ir-a-i-a-a
I I
H H
(Causative Syllabif. + Epenthetic -a- + Causative H + L Del.)
Stratum 3 (or postlexically)
tu-a-mu-gul-ir-a-i-a
^Nl M
H H
(H Spreading)
tu-a-mu-eul-ir-a-i-a-a
-^l \l /
H HH
tu-a-mu-gul-irir-a-a
H L
tu-a-mu-gul-irir-a-a
^1 I
H L
(PHA + Gliding + Default L + Vowel shortening)
Output: twamuguliraya twamugulirira
Mutaka: The Tone Bearing Unit in Kinande 171
I would like to point out that it is not only the causative morpheme that
triggers these rules of causative syllabification, causative H, seietion of L and
the insertion of the epenthetic {a}: The passive morpheme triggers all of these
rules also under the same conditions as illustrated in (17) where the forms are
in the recent past tense as well.
(17) tu-a-hum-a-u-^-a [twahumawa] 'we were hit'
from eri-hum-ii-a 'to be hit'
mo-tu-a-hum-u-a [motwahumwa] 'we were hit'
mo-tu-a-hum-irir-u-a [motwahumirirwa] 'we were hit anyway'
The question now is whether there is any relation between the causative
and the passive morpheme to trigger such similar rules. One relationship that
can be pointed out is that they are the only extensions to consist of a single
vowel [V] unlike the other extensions which have the structure {VC}. Another
relationship is that they both do not undergo height harmony as in eri-log-es-f-a
'to make someone be bewitched' or eri-log-ol-u-a 'to be unbewitched'.
Extensions like {is} or {ir} in these forms undergo height harmony, but the
causative and the passive do not. What this suggests is that our earlier rules of
causative syllabification, causative H discussed in (9), should be rather called
vowel extension syllabification and vowel extension H.
Finally, I would like to point out that these rules also apply in the remote
past tense as illustrated in (18).
(18) Remote past tense
a. tu-a-gul-a-i'-a [twagulaya ~ twagulaya] "we sold'
tu-a-mu-gul-a-i'-a [twamugulaya -twamugulaya] 'we sold him'
tu-a-gul-ir-a-u-a [twagulirawa -twagulirawa]
'we were bought for'
b. tu-a-gul-irir-a 'we bought anyway'
tu-a-mu-gul-irir-a 'we bought him anyway'
As shown in (18b), no H surfaces in the remote past tense of a toneless
verb. Since the H does surface in the forms in (18a) at the left of the causative or
the passive morpheme, it must be concluded that this H is the result of the vowel
extension H posited above.
To sum up, it has been proposed in this paper that what qualifies a
Kinande segment to be a Tone Bearing Unit is to contain only vowel features in
its melodic structure. The feature [-syll] which is not a melodic feature, although
it may be assigned to a [+hi] vowel, will not prevent it from acting as a TBU as
long as gliding, which is postlexical, has not applied by assigning a
consonantal feature like [+cons] into the melodic content of the vowel.
172 Studies in the Linguistic Sciences 20:1 (Spring 1990)
NOTES
Special thanks are due to Diana Archangeli and also to Doug
Pulleyblank and Larry Hyman with whom I discussed the material of this paper.
My thanks are also due to the audience of the 20th African Linguistics
Conference at the University of Illinois at Urbana-Champaign where this paper
was read. I am also indebted to the editors of this issue for their editorial
comments and corrections. Any errors of facts and analysis are of course mine
alone.
1 See Mutaka(1991)
2 Another alternative that was briefly considered but rejected was to posit a
phantom consonant -C- on the left of the causative {i} in forms with an
epenthetic vowel [a]. The reason this solution was rejected was due to the
existence of other forms like eri-gul-i-bu-a "to be sold' where it could be argued
for that the passive morpheme in this form is {u} and not {bu} and that {b} is an
epenthetic consonant that gets part of its features from ju}. Given this fact, it
would then be difficult to explain why, in a form like tu-a-gul-a-wa which would
have tu-a-gul-a-C-u-a as an intermediate stage in the derivation, the consonant
-C- does not surface as {b} as in eh-gul-i-bwa.
3 Notice that the tonal node in this tree hangs directly onto the skeletal tier.
This is not assumed in Clements (1985) and Sagey (1986).
REFERENCES
ARCHANGELI, D., & D. Pulleyblank. 1988. Maximal and minimal rules: Effects of
tier scansion. Proceedings of NELS 17. Cambridge, MA: MIT
Archangeli, D. 1988. Maximal and minimal target. Paper presented at a
Phonology workshop at UCLA, Los Angeles. MS.
Clements, N. 1985. The geometry of distinctive features. Phonology Yearbook
2:225-252.
HYMAN, L. M. Forthcoming. Boundary tonology and the prosodic hierarchy.
Proceedings of Stanford Conference on the syntax-Phonology
Connection. CSLI.
Hyman, L., & N. Valinande. 1985. Globality in the Kinande tone system. African
Linguistics SSLS 6.239-260, ed. by D. Goyvaerts.
MUTAKA, M. N. 1987. The status of liquids in Kinande. Current Approaches to
African Linguistics, 6. Foris: Dordrecht.
. 1991. The lexical tonology of Kinande. University of Southern
California-Los Angeles Ph.D. dissertation in Linguistics.
Sagey, E. 1986. The representation of feature and relations in non-linear
phonology. MIT Ph.D. dissertation in Linguistics.
SCHLINDWEIN, D. 1987. P-bearing units: A study in Kinande Vowel Harmony.
Proceedings of NELS 17.
IV
Sociolinguistics & History of Linguistics
Andre Mwamba Kapanga
Language variation and language attitudes: A case study from Shaba Swahili
Margaret Wade-Lewis
The contribution of Lorenzo Dow Turner to African Linguistics
Studies in the Linguistic Sciences
Volume 20, Number 1, Spring 1990
LANGUAGE VARIATION AND LANGUAGE ATTITUDES:
A CASE STUDY OF SHABA SWAHILI*
Andre Mwamba Kapanga
University of Nebraska-Lincoln
This paper examines the differences between the native variety
of Swahili spoken in East Africa and its historical non-native variety
spoken in the southeastern region of Zaire known as, Shaba, from
the point of view of language variation. It shows that there are indeed
variations between ShS and its native variety counterpart, EAS. It will
be demonstrated that these variations are, by and large, due the
enculturation of ShS in the non-native context of use and usage.
These differences between these varieties will be shown to be the
result of linguistic and cultural contacts between ShS and the many
local languages and cultures of Shaba. Thus, the linguistic variations
exhibited by ShS are due not only to the acculturation of Swahili to fit
both the linguistic and communicative strategies proper to the
multilingual but also to sociocultural context of use in Shaba. This
claim is supported by these important findings. (1) the impact of the
substratum Bantu languages and the superstratum language which
have closely been in contact with ShS since its introduction in the
Shabian context; (2) the contextual rendering of texts in daily
interactions; and (3) an attitudinal survey of ShS speakers' attitudes
towards EAS and their own variety.
1. Introduction
The concept of variability in languages has been recognized by linguists
for many years. However, for a very long time, most studies have limited
themselves to the description of languages with the assumption that they were
homogeneous, i.e., speakers of a given language would all share internally
consistent structures in the use of that language. At the beginning of this
century, however, a new trend started to develop in the study of language
variation. Variation came to be considered as a phenomenon that occurs over
time and space. This view resulted in the study of variation in relation to
extralinguistic factors at the time of any speech event (Smith 1974). In his
pioneering work on the study of language variation in New York City, Labov
(1972) showed that linguistic output is generally correlated with extralinguistic
factors. Thus, in the process of collecting linguistic data, researchers need to
collect information not only about the characteristics of the speaker, but also
about the situation as well as the speakers perception of the situation in which
speech takes place. Therefore, such aspects as age, sex, social background,
education, ethnic and geographical background together with the context of the
speech act were to become very relevant not only in the collection of data, but
176 Studies in the Linguistic Sciences 20:1 (Spring 1990)
also in their analysis. This new perspective for analyzing variation has not been
limited to the study of monolingual communities; it has been extended to
multilingual communities as well, and to second language studies in both native
and non-native environments (Kachru 1982 and later; Lowenberg 1985, 1986;
Magura 1984).
2. Processes of linguistic variation
The study of language varieties has generally involved discussions of the
concept 'model'. This concept has proved to be important in that in any study of
language variation, one variety which is considered to be the model, is usually
taken as the reference point for the study of differences displayed by other
varieties. The concept model itself implies the existence of a certain linguistic
ideal which language learners have to attain in the learning process (Kachru
1986). However, it needs to be pointed out that the selection of a model is not
dependent on linguistic factors; rather extralinguistic factors such as language
attitude, language identity, prestige factors, to name a few, are generally what
determine the selection of one variety over another.
Swahili, as a language, displays a whole range of varieties among which
are found Kiungudja, Kimatumbi, Kimtangata, Kihamu, Kisangani Ki Swahili,
Kivu KiSwahili, Shaba KiSwahili, etc. As is well known, one variety has been
selected for extralinguistic reasons to be the model for all other varieties. This
variety is known as Kisanifu KiSwahili, also known as EAS. This model is
regulated by a set of norms to which the language speakers are expected to
conform. Norms are defined by Bedard and Maurais (1983:7) as "I'usage
valorise dans un groupe donne; le groupe socialement dominant produisant
alors le bon usage, qui eclipse les normes des autres groupes et reussit alors a
faire croire a leur non-existence" (For futher discussions of the models and
norms see Bedard, et al. 1983; Kachru 1986). In comparing the EAS model and
the Zairian variety spoken in Shaba, one notices that the latter displays many
instances that violates the norms that regulate the language at all linguistic
levels. The assumptions by many scholars who have dealt with ShS has been
that such violations are mainly due to linguistic corruption, substandardization
or acquisitional deficiencies by ShS speakers when they are learning EAS.
These views are found in works by Lecoste (1954), Harris (1956), Polome
(1968. 1969). Heine (1970), Fabian (1982, 1986). among others.
This section focuses on some of the linguistic patterns of ShS that deviate
from the norms of the model variety, EAS. These deviations are analyzed to see
whether they are the result of corruption, substandardness or acquisitional
deficiencies.
When introduced in Shaba, Swahili had a very limited functional range. It
was mainly used in the trading context (Polome 1967, 1968; Heine 1970;
Fabian 1982, 1986) and could in this capacity be referred to as a performance
variety. However, its later adoption as a lingua franca allowed it to increase its
functional uses. In Shaba today, Swahili is used for many functions that include,
in B. Kachru's (1982) terminology: instrumental, regulatory, interactional.
(
Kapanga: Language variation and language attitudes — Shaba Swahili 177
personal, heuristic, imaginative, and informative. That is, today ShS is a
"deregionalized variety which has become a vehicle for supra-regional
communication" (Hock 1986:485). This regionalization is mostly due to the
convergence of Swahili with Shabian local languages and its mechanism
seems to have been the result of Interlanguage early in the introduction of
Swahili in Shaba. French, the official language of the country, is generally used
to convey information about new technological development and to learn about
new concepts in the world. Thus, though Swahili is used in the heuristic and
informative functions, its function in this capacity is very limited and closely
linked to the non-western aspects of life in the community. The expansion of
functions as well as the new linguistic and sociocultural context of development
have resulted in nativization of the language. Such nativization is essentially
what distinguishes ShS from the native variety in East Africa. It has mostly
occurred as a set of processes that involve not only transfer of linguistic
patterns, but also transfer of cultural patterns of the new sociolinguistic context.
This has been the case at all linguistic levels, namely, phonological,
morphological, lexical, syntactic, semantic and pragmatic.
Of all these processes, the most interesting and noticeable are lexical
transfers. They involve many of the innovative processes discussed in Sey
(1973), Bamgbose (1982), Bokamba (1982), and Zuengler (1982) with regard to
African varieties of English. The first process refers to direct transfers. Here,
lexical items from language A are transferred to language B when B has no pre-
existing words. This has led to the introduction of words such as twika 'help
carry on head', papa 'carry the baby on the back with a piece of cloth', tula 'put
down what is carried on the head', shesheti {French.chaussettes) 'socks',
pantalo (FrencU.pantalon), jipe (French :ytype) 'skirt', shemize {French.chemise)
'shirt' and words that refer to animals, vegetables, and insects. Examples
include: nsombe (EAS:kisanvu) 'cassava leaves', lengalenga {EAS.mchicha)
'spinach', kabunji {EAS.mbweha) 'fox', mbandakwe {EAS.kobe) 'tortoise'
mpanjo {EAS:nzige) 'crickets'. Moreover, there are cases of direct transfer that
take place despite the fact that the borrowing language has a direct translation
for the borrowed item. In cases like these, ethnolinguistic vitality of an ethnic
group, as defined by Giles (1979, 1982) can be the determining factor in
influencing tranfer (Giles 1979).
The second type of transfer involves semantic deviations which can be
regarded as semantic shifts, semantic extensions, and semantic transfers.
Semantic shifts "involve a redefinition of the characteristic patterns of a word
within the semantic field to such an extent that its central context becomes
marginal" (Bokamba 1982:87). Examples of this kind of deviation can be seen
by comparing the meaning of words such as jamaa, bibl, bwana, masomo and
kichele in both EAS and ShS. In EAS these words mean, 'relatives', 'lady',
'mister', 'studies', and 'petty cash', respectively. In ShS, however, the EAS
meanings have become marginal, instead other meanings have become
prominent, jamaa is almost always associated with a religious group of the
Catholic church whose members consider themselves as belonging to one big
family , the jamaa. Bibi and bwana are, in contrast, generally used to mean
178 Studies in the Linguistic Sciences 20:1 (Spring 1990)
'wife' and 'husband', respectively. Masomo is always used for 'school' while
kichele is used for 'coins'.
The third type of transfer involves semantic extensions. These deviations
are generally talked about when loan words or L2 words acquire additional
meanings in the borrowing variety. The early non-native speakers of Swahili in
Shaba used the language as a second language. In the process, they extended
the meanings of many EAS words by adding new meanings to them, while
conserving the EAS meanings. Examples of these are words such as kuweza
'to be able', kupumzika 'to rest', mchele 'uncooked rice', and mpepo 'wind'. In
ShS these words have acquired additional meanings, namely, 'to succeed', 'to
breath', 'cooked rice', and 'air', respectively. These latter meanings are
expressed in EAS by such words as kufaulu, kupumua, wall and hewa,
respectively. The fourth type of transfer involve semantic transfers. This process
consists of a complete reassignment of the meanings of L2 words. Words that
underwent changes like these are numerous in ShS. Among them are: kuuza
'to sell', kushinda 'to defeat', magharibi 'west', kuvuma 'to blow', and kudhuru
'to harm'. In ShS these words mean 'to buy', 'to fail' 'afternoon', 'to be reputed'
and 'to insult'.
The fifth and last kind of transfer is known as coinage. This process refers
to outright invention of words, and extension of roots by affixation; it also
involves the formation of new words by the use of morpheme combinations
such as reduplication, compounding, and the use of onomatopoeic sounds
(Bokamba 1982). The reduplication process involves the formation of new
words by combining two identical morphemes, which need not necessarily bear
any meaning when they occur by themselves, into one word. For examples,
kutangatanga 'to wander', kuwayawaya 'hang around', and kusemasema 'to
complain' The morphemes tanga and waya do not bear any meanings by
themselves; whereas sema means 'say'.
Compounding, in contrast, is the process whereby existing lexical items
are combined or used in genitive construction type of structures to form new
words in the language. This process is the most productive in the formation of
new linguistic terms in ShS. Examples of these include: formation of numbers
beyond nineteen {makumi mbili: 20, makumi tatu: 30, makumi tisa: 90 etc.),
kifungula mimba 'first born child' from kufungula 'to open' and mimba
'pregnancy', vunjambanga 'very hard biscuits' from vunja 'to break' and
mbanga 'jaws'; this word literaly means 'jawbreaker'. Examples that involve the
use of the genitive construction includes: muntu wa kazi 'worker', mutoto wa
mayimayi 'baby', mutshi ya ndizi 'banana tree'; in EAS these words are
expressed with mfanyakazi, mtotomchanga, and mgomba. Onomatopoeics, on
their part, are also used to create new lexical items in ShS. In this process, the
sound produced by a given person or thing is used to refer to that person or
thing. For example, kokoriko 'rooster', tukutuku 'motorcycle', and matamata 'fat
person'.
Kapanga: Language variation and language attitudes — Shaba Swahili 179
3. Functional meaning of linguistic variation
Many of the transfers that have been observed in ShS have generally
been brandished as Instances of Inadequacies of this variety vis-a-vis EAS.
This section examines these transfers to see whether they carry any social or
cultural significance for their users in the Shablan context. The emphasis is
placed on the appropriateness of transferred Items in the new contextual
situation where Swahlll is used.
Both ShS and EAS develop In multilingual contexts. The languages that
comprise what can be termed "the language repertoire"^ of each of the areas
where these two varieties are spoken differ considerably. The Implication for the
linguistic co-existence in each of these two geographical areas is the mutual
influence at all linguistic levels of the languages in contact. However, this
coloring Is not limited to linguistic patterns: it extends to cultural patterns, for the
differences between these languages are not only linguistic, but also cultural.
That Is, language being part of the culture of the people who speak It, It Is
obvious that linguistic transfers that occur will Involve transfer of the cultural
aspects that are associated with those transfers. Since EAS and ShS co-exist,
each with languages spoken by people from linguistically and culturally
different backgrounds, some transfers of these languages' linguistic and cultural
patterns can be expected. In addition, given the limited contact between EAS
and ShS, each variety would take Its own path of development and Is likely to
converge with the languages of its own linguistic area of use. The danger for
language prescriptivists is to think of transfers from other languages in the non-
native variety as examples of substandardness, corruption or acquisitional
deficiency whereas in the native variety, such transfers are viewed simply as
instances of language enrichment.
A great deal of the differences between the Shablan variety of Swahlll and
its native counterpart can better be understood If one views the contextual
significance of in the Introduction of most of the deviations exhibited by ShS.
The countries of East Africa and Zaire are two areas of Africa that took two
separate paths at the advent of colonization. East Africa was colonized by Great
Britain, while Zaire was colonized by Belgium. While these two countries
basically brought the same new western concepts to their colonies, the former
did so by using English whereas the latter relied on French. This explains the
high occurrence of English borrowings in the variety of Swahlll spoken In East
Africa while In Zaire, most borrowings from the colonizers are In French. Thus,
EAS exhibits words such as refa, shati, sketi, dereva and soksi from English
referee, shirt, skirt, driver, and socks, respectively; their ShS counterparts are
arbitre, shemize, jipe, shofer, and shesheti which are derived from French
words arbitre, chemise, jupe, chauffeur, and chaussettes, respectively. For most
ShS speakers, words of French origin are not used to Impede communication
with native variety speakers: They are used as models of acculturation as well
as markers of membership by speakers who share the same superstratum
language. In their community, French Is the prestige language as well as the
language of socioeconomic advancement. Thus, the contextual significance
rather than the lack of EAS knowledge Is what determines the use of words
180 Studies in the Linguistic Sciences 20:1 (Spring 1990)
which are viewed by prescriptivists as instances of acquisitional deficiencies on
the part of ShS speakers. ShS has also been labelled a corrupt variety
because its extensive use of words of Bantu origin in lieu of those derived from
Arabic. As is well known, Swahili is a language that was originally spoken only
on the coast of East Africa where, because of comnnerce, it was in contact with
the countries of the Arabic peninsula (Haddad 1983). These contacts led to the
settlement of Arab merchants and the introduction of Islam in the area. In
addition, the Arab settlers were among the first to have provided the first
descriptive books of Swahili; these books are characterized by a heavy use of
words of Arabic origin in the language. The result has been a large body of
words of Arabic origin in Swahili.
In Shaba, Arabic and Islam have had very little impact on Swahili; their
influence had been curtailed by the Belgian colonizers and the missionaries
who have played a big role in shaping ShS (see Polome 1967, 1968; Haddad
1983). The occurrence of a very limited number of words of Arabic origin and
the low following of Islam attest to this fact. When these words are used in the
discourse, they are generally not associated with Arabic, for they are
considered Bantu in nature. The integration in ShS of words that are clearly
associated with Arabic is resisted by ShS speakers. The abandonment of
Arabic loans can be found in (a) the counting system, (b) the days of the week.
The counting system of EAS is based on that of Arabic starting from 20 on.
For example, 20=ishirini, 30=thelathini, AO=arobaini, etc. In ShS, Bantu based
numbers are used; they are realized as makumi mbili, makumi tatu, makumi ine,
respectively. The days of the week in EAS are also based on the Muslim
calendar; thus, the first day of the week is Saturday and the last day of the week
is Friday; in addition, words of Arabic origin are used for the different days.
These words are: Jumamosi 'Saturday', Jumapili 'Sunday', Jumatatu 'Monday',
Jumanne 'Tuesday', Jumatano 'Wednesday', Alhamisi 'Thursday' and Ijumaa
'Friday'. In ShS, however as in many other Zaierean Bantu languages, the
naming of the different days of the week is based on the working calendar.
Monday is considered the first day of the week because it is the first day when
workers start working. Thus, words used to express the days of the week are:
Kazi moya 'Monday', Kazi mbili 'Tuesday', Kazi tatu 'Wednesday', Kazi ine
'Thursday', Kazi tano 'Friday', Siku ya mpostio 'Saturday' and
Juma/Yenga/Siku ya Mungu 'Sunday'. The literary translations of these are: first
day of work, second day of work, third day of work etc.; the expression siku ya
posho refers to the day when the colonizers used to give the weekly food supply
to their workers. Siku ya Mungu refers to the day of God.
In essence, there is, in ShS, a strong emphasis to use words of Bantu
origin instead of Arabic ones. The use of words of Bantu origin is a way of
conveying a certain allegiance to the Bantu background to which Swahili
belongs. Bantu vocabulary is used to make a statement regarding the
language group membership. It is not uncommon to hear people talk of EAS as
Arabic Swahili; members of the Shabian community who excessively use words
of Arabic origin are ridiculed and thought of as snobbish and pretentious. This
perception has led to a slow but steady disappearance of words of Arabic origin
Kapanga: Language variation and language attitudes — Shaba Swahili
18]
which have corresponding Bantu synonyms in the speech of many ShS
speakers. This 'deliberate fore-grounding' of words of Bantu origin is one way of
separating Swahili from Arabic influence in the Shabian community, while
allowing it then to converge more with local cultures in the new context of use.
Thus variation in this case can be attributed to linguistic regionalism/
nationalism in the Shabian context.
4. Language attitudes and the study of linguistic variation
The impact of language attitudes on linguistic behavior is a phenomenon
that has been recognized by sociolinguists who have focused on the study of
language attitude as a research topic for the last decades (Shuy & Fasold 1973;
Cooper 1974, 1975; Cooper & Fishman 1974). These studies and others have
acknowledged that language attitudes can contribute to linguistic change and
speech community identification, as well as reflect intergroup communication
patterns within the same community (Tucker & Lambert 1969; Bradac 1982;
Caranza 1982). During the summer of 1987 I conducted an attitudinal survey
among three hundred randomly selected subjects representing most of the
different ethnic groups and social strata found in Shaba to ascertain the role of
attitudes on the linguistic behavior in the Shabian environment. The social
factors in this selection process included sex, age, education, profession, social
class, religion, and geographical background. First, the subjects were asked to
identify all varieties of Swahili they were aware of and classify them according
to their prestige in the Swahili community at large. Four varieties were identified
by most subjects: EAS, Kivu Swahili (KS), Kisangani Swahili (KiS), and Shaba
Swahili. Table 1 shows the percentage of recognition among the subjects for
each variety identified.
Table 1 ; Varieties recognized by ttie subjects
ShS
100%
EAS
68%
KS
29%
KiS
7%
That is, in addition to their own variety, all the subjects were aware of the
existence of several other varieties; 68% knew of the existence of EAS, while
only 29% and 7% were aware of KS and KiS, respectively. Second, the
subjects were asked to rank those varieties according to the prestige associated
with each of them in the Swahili community in general. The results were as
given in table 2.
Table 2: Ranking of the recognized Varieties
Rank
Variety
%
1
EAS
54
2
KS
23
3
ShS
17
4
KiS
6
182
Studies in the Linguistic Sciences 20:1 (Spring 1990)
Third, the subjects were asked to rank the four varieties according to their own
preference and to give reasons for their first choice. The results are as given in
table 3.
Table 3: Ranking by subject's preference
Rank
Variety
%
1
ShS
73
2
EAS
12
3
KS
9
4
KiS
6
The reasons for preferring one variety over the others led to the identification of
four clusters of socio-psychological motivations. First, the speakers who
preferred ShS over other varieties did so for 4 main reasons:
1. It is the language they grew up speaking.
2. It is the language of their peers.
3. For integrative purposes.
4. It does not have too many Arabic words.
Second, EAS was preferred over the other varieties because:
1. It has more prestige than our Swahili.
2. It is the Standard variety.
3. It allows me to understand the news in Swahili.
4. It is the language of the church.
Third, KS was preferred over the other varieties because:
1 . It is closer to EAS in form.
2. They spent some time in Kivu
3. Of the influence of friends/relatives.
Finally, they preferred KiS over the other varieties because:
1. Of the influence of friends/relatives.
2. They spent some time in Kisangani.
These attitudes are of course natural and should be expected in any language
environment where instrumental and sentimental motivations are in constant
contact. They suggest that most of the speakers in Shaba want to feel they are a
part of the Shabian community. They are aware of the fact that social group
membership is closely associated with particular styles of speech; they are also
aware of the fact that listeners always make stereotypical inferences about their
interlocutors based on the speech markers that they perceive in the
interlocutors' actualization of speech (Bradac 1982). The disregard for the
native and more prestigious variety is not due to its linguistic complexity, but
Kapanga: Language variation and language attitudes — Shaba Swahili 183
results rather from its inappropriateness in the Shabian context. This
inappropriateness is not only linguistic, but also political, social, and cultural.
This attitude has ted to both a deliberate and non-deliberate effort by Shabians
to promote the use of their own model which is for them a sociolinguistic
prerequisite for complete integration in the community. The features of ShS that
deviate from EAS are what constitute its "Shabaness". Therefore, transfers from
local languages and French as well as the "foregrounding" of words of Bantu
origin at the expense of words of Arabic origin, whether intentional or not, are
the strategies used to represent identity maintenance among ShS speakers.
These features are not used to compensate for the speakers' putative
deficiencies in the acquisition of the native variety; rather, they are fundamental
to the authentication of the Shabian variety. This view is supported by Fishman
(1971:1) when he says: "language is itself a content, a referent for loyalties, an
indicator of social statuses and personal relationships, a marker of situations
and topics as well as of societal goals and large scale value-laden arenas of
interaction that typify every speech community".
5. Socio-psychological aspects of linguistic variation
Another aspect of language that has often been neglected by variation
studies of African languages is the socio-psychological aspects of speech
markers in speech communities. A speech community here is perceived as a
group of people who consider themselves as speaking the same language or
speech variety for which they share not only the norms for its appropriate use,
but also a set of social attitudes (Gumperz 1971; Corder 1973; Labov 1972;
Romaine 1982). Speech markers, in contrast, are the linguistic and
extralinguistic cues that differentiate between the different biological, social, and
psychological categories or characteristics of the speakers which in one way or
the other are actually or potentially of importance for social interaction or social
organization (Giles, et al. 1979). Generally, each speech community displays
certain markers that identify it; they include markers of speech identity which
have been classified into three categories (Abercrombie 1967; Laver & Trudgill
1979): (a) social markers; (b) physical markers; and (c) psychological markers.
Belonging to a speech community implies adhering to the norms expected
of the members in the community. Such norms involve the use of certain speech
markers: linguistic, social, physical and/or psychological, by the members of that
community. Thus, speech markers, whether used for communicative or
informative functions are generally used to convey the community's beliefs,
attitudes, and values (cf. Lyons 1972; Giles, et al. 1979). In communicating with
others, receivers generally infer from the speakers' language their attitudes,
mood, and affiliations (Bradac, et al. 1983). These inferences are generally
made from either one's style or the use of the aforementioned markers.
Speakers are generally aware of this fact and they use their speech to convey
their attitudes, mood, and affiliations. This task is mostly performed through the
use of two strategies of speech communication: convergence and divergence
(Giles, et al. 1977, 1979). The former involves the use of certain speech markers
by the speaker to gain social approval and/or to be socially integrated. The
latter involves the use of speech markers which allow speakers to dissociate
184
Studies in the Linguistic Sciences 20:1 (Spring 1990)
themselves from a given group or outgroup and associate themselves with a
group they positively identify with.
In the Shabian speech community, the existing norm for all speakers of
Swahili is the use of speech markers that characterize the community. Such
markers involve the use of linguistic items and expressions that typify the
community. For the sake of achieving total integration or being socially
accepted in the community, members find themselves using attenuation and
accentuation markers to reach a full social integration in the community. The
community does not encourage the acquisition of any variety other than ShS; it
completely discourages the use of features not found in the Shabian variety of
Swahili by stimulating the use of local features. A clear example of this can be
seen in the conversation of two ShS speakers who were talking about visiting a
common friend who makes excessive use of EAS.
A: unipeleke kwa X
'please accompany me to X's house'
B: sipende kwenda kwake.
'I don't want to go his house'
A: juu ya nini? 'why?'
B: kama unaenda kwake ataanza kukusemea Kiswaili kya mu
Tanzanie; ye anawaza asema tuko mu Tazanie?
'if you go to his house he'll start talking to you inTazanian Swahili;
does he think that we are in Tazania?'
The attitude of speaker B is very common among ShS speakers; it is found
among both intellectuals and non-intellectuals alike. As a matter of fact, speaker
B happens to be a university graduate who has studied Standard Swahili in
School and has a very good job in the highest paying company in Shaba:
Gecamine. For many people who speak varieties other than ShS there is a
tremendous pressure to speak ShS as the data in table 4 show. The subjects in
this case includes only those whose preferred dialects are different from ShS.
Table 4: Which dialect do you usually use in your daily life?
ShS
73%
EAS
17%
KS
7%
KiS
3%
Table 4 shows that there is a strong loyalty to ShS even among people who
prefer other varieties of Swahili. The subjects acknowledged that their
preference for ShS in their daily life is mostly the result of social pressures to
use the local variety. ShS is, by and large, a socially imposed dialect rather
than a personal preference for most of these speakers. The impact of social
pressures on the maintenance of low varieties has been recognized by
previous research (cf. Ryan, et al. 1977:9):
Kapanga: Language variarion and language attitudes — Shaba Swahili 185
"Ingroup solidarity or language loyalty reflect the social pressures
that operate to maintain language varieties even in the absence of
social prestige. The language or dialect of one's family life, intimate
friendships and informal interactions acquire vital social meanings
and comes to represent the social groups which one identifies with."
Given this situation, it would be inappropriate to claim that the speakers of the
Shabian variety of Swahili speak a corrupt, substandard and approximative
system; for societal pressure is what determines the dialect to be used in the
community. From the point of view of theories of linguistic variation (and second
language acquisition), preference of one dialect over another is determined by
sociolinguistic pressures. In Shaba, society is what imposes the ShS variety on
people despite the fact that many people are aware of and taught Standard
Swahili or EAS. Society denies them the use of the variety that is more
acceptable to the prescriptivists. Therefore, evaluation of the acquisition of
Swahili in the Shabian non-native context should be one that takes the local
variety as the sole goal to be reached by the learners. The reason for this is
basically because it is the only input available to the learners.
6. Conclusion
This paper has shown that ShS has significantly deviated from EAS as
reflected in part by the lexicon and lexical semantics. It has shown that there are
many processes that account for these deviations. These processes include
transfers from local languages and French, semantic deviations, and
foregrounding of words of Bantu origin at the expense of words of Arabic origin.
The paper also showed that these processes have actually become part of the
strategies used to represent identity maintenance among ShS speakers. These
features are not used to compensate for the speakers' acquisitional deficiencies
in the native variety, but rather, they are integral or fundamental to the
authentication of the Shabian variety. It has also been argued that the linguistic
characteristics of EAS and ShS cannot be delineated without discussing the
speakers' linguistic attitudes. Therefore, analysis of language variation should
not center on the study of form to the exclusion of its sociolinguistic context of
use. Language variation analysis should also include a discussion of the
attitudes of the speakers, as they are the reflections of the realities of the socio-
cultural context of language use. Such a step makes it feasible to not only focus
on the differences in the social powers likely to influence speech within a
community, but to also have a perspective on the dynamics of speech and
social structures as well as [the directionality of] linguistic change.
In principle, a good analysis of linguistic variation, especially in
multilingual communities, requires a good understanding of the social groups
that exhibit those variations. The researcher needs to understand the
community's attitudes vis-a-vis the linguistic, social, and cultural variations
peculiar to each constituent of that community. To attain this objective, the
researcher has to undertake two necessary steps. The first is to define the
community whose speech is to be analyzed; a speech community, in this
particular case, must be perceived as a group of people who consider
186 Studies in the Linguistic Sciences 20:1 (Spring 1990)
themselves as speaking the same language or speech variety for which they
share not only the norms for its appropriate use, but also a set of social and
cultural attitudes (Gumperz 1971; Labov 1972; Corder 1973; and Romaine
1982). In other words, any group which shares both linguistic resources and
rules for interaction in interpretation is what ought to be defined as a speech
community. The second step is to define the linguistic, social and cultural
parameters of the community whose language is being observed. These
parameters can be established by using attitudinal surveys of the type utilized in
this study as well as those used by social psychologists and neo-labovians.
Taking these two steps into consideration will ensure that two of the most crucial
aspects of linguistic variation will not be overlooked. In overlooking them, there
is a strong possibility of making generalizations which are not supported by the
facts, has happened in most studies of ShS.
NOTES
I would like to thank Professors Eyamba Bokamba, Hans H. Hock,
Yamuna Kachru and Braj Kachru for their helpful and valuable comments.
However, I alone, assume responsibility for any remaining weaknesses and
mistakes.
1 Language repertoire in this instance can be defined as a set of
languages found in a given speech community.
REFERENCES
Abercrombie, D. 1967. Elements of general phonetics. Edimburg.
Bamgbose, Ayo. 1982. Standard Nigerian English: Issues of identification. The
other tongue: English across cultures, ed. by Kachru, 99-111.
Bedard, Edith, & Maurais, Jacques. 1983. La Norme Linguistique.
Gouvernement du Quebec. Canada.
Bokamba, Eyamba G. 1982. The Africanization of English. The other tongue:
English across cultures, ed. by Kachru, 77-98.
Bradac, James J. 1982. A rose by another name: Attitudinal consequences of
lexical variation. Attitudes towards language variation: Social and applied
contexts, ed. by Ryan & Giles, 99-115.
, Bowers, J. W., & Courtright J. A. 1980. Lexical variations in intensity,
immediacy, and diversity: An axiomatic theory and causal Model. The
social and psychological contexts of language, ed. by St. Clair & Giles,
193-224.
Carranza, Miguel A. 1982. Attitudinal research on Hispanic language varieties.
Attitudes towards language variation: Social and applied contexts, ed. by
Ryan & Giles. 63-83.
Kapanga: Language variation and language attitudes — Shaba Swahili 187
Cooper, R. L. 1974. Language Attitudes I. International Journal of the
Sociology of Language 6:5-9.
. 1975. Language Attitudes II. International Journal of the Sociology of
Language 6.
, & Fishman, J. A. 1974. The study of language variation. International
Journal of the Sociology of Language 3.5-19.
CORDER, S. P. 1974. Idiosyncratic approach and Error Analysis. Error analysis:
Perspectives on second language acquisition, ed. by Richards, 19-27.
Fabian. 1982. Scratching the surface: Observations in the oetics of lexical
borrowing in Shaba Swahili. Anthropological Linguistics 24.14-50.
. 1986. Language and colonial power: The appropriation of Swahili in the
Belgian Congo 1880-1938. Cambridge University Press. Cambridge,
England.
Fishman, J. 1971. The Sociology of Language. Advances in the Sociology of
Language, vol.1, ed. by J. Fishman, 217-404. The Hague: Mouton.
Giles, Howard. 1977. Social psychology and Applied Linguistics: Towards an
integrative approach. ITL: Review of Applied Linguistics 33.27-42.
. Bourhis, R. Y., & Taylor D. M. 1977. Towards a theory of language in
ethnic group relations. Language, Ethnicity and Intergroup Relations, ed.
by Giles, 307-348. London: Academic Press.
. 1979. Ethnicity markers in speech. Social markers in speech, ed. by K.
R. Scherer & Giles, H, 251-290.
GUMPERZ, J. 1972. The speech community. Language and social context, ed. by
Giglioli, 219-231.
Haddad, Adnan. 1983. L'Arabe et Le Swahili dans La Republique du Zaire:
Etudes Islamiques (Histoire et Linguistiques). C.D.U et SEDES. Paris,
France.
Harris, Lyndon. 1956. Congo Swahili. Tanganyika Notes and Records 38-
45:44. Published for the Editorial Board, Dar es Salaam.
Hock, H. 1986. Principles of Historical Linguistics. Mouton De Gruyter. Berlin.
Kachru, B. B. 1982. The other tongue: English across cultures. Press. Oxford,
London.
. 1982. Models for non-native Englishes. The other tongue: English
across cultures, ed. by Kachru, 31-57.
. 1982. Meaning in deviations: Towards understanding non-native
English texts. The other tongue: English across cultures, ed. by Kachru,
325-350.
. 1983. The Indianization of English. Oxford University Press. Oxford,
England.
. 1986. The alchemy of English: The spread, functions, and models on
non-native Englishes. Oxford, England: Pergamon Institute of English.
Kachru, Yamuna. 1986. Applied Linguistics and foreign language teaching: A
non-Western perspective. Studies in Linguistic Sciences 16:1.35-51.
Labov, W. 1972. Sociolinguistic patterns. Philadelphia: University of
Philadelphia Press.
Laver, John, & Trudgill Peter. 1979. Phonetic and linguistic markers in speech.
Social markers in speech, ed. by K. R. Scherer & Giles, H, 1-32.
Lecoste, B. 1954. Variete Congolaise du Swahili. Congo Overzee 20.4-5. De
Sikkel Antwerp.
188 Studies in the Linguistic Sciences 20: 1 (Spring 1990)
LOWENBERG, Peter H. 1986. Sociolinguistic context and Second Language
Acquisition: Acculturation and creativity in Malaysian English. World
Englishes 5.71-83.
. 1985. Nativization and function in ESL: A case study. TESL Studies
6.88-107.
Magura, Benjamin. 1984. Style and meaning in African English. University of
Illinois at Urbana-Champaign Ph.D. dissertation in Linguistics.
POLOM^, Edgard. 1967. Swahili language handbook. Washington D.C.: Center
for Applied Linguistics.
. 1968. Lubumbashi Swahili. Journal of African Languages 7:1.15-25.
Hretford, England: Nimram Books.
. 1971. Multilingualism in an African urban center: The Lubumbashi case.
Language use and social change, ed. by Whiteley, 364-375. London:
Oxford University Press.
Richards, Jack. 1978. Models of language use and language learning.
Understanding second and foreign language learning: Issues and
approaches, ed. by Richards, 94-116. Rowley, MA: Newbury House
Publishers, Inc.
ROMAINE, Suzanne (ed.) 1982. Sociolinguistic variation in speech communities.
London: Edward Arnold.
RYAN, Ellen B., & Giles, Howard, 1982. Attitudes towards language variation:
Social and applied contexts. London: Edward Arnold.
Sankoff, Gillian. The social life of language. 1980. Pennsylvania: University of
Pennsylvania Press.
SCHERER, R. K., & Giles, H. (eds.) 1979. Social markers in speech. Cambridge:
Cambridge University Press.
Selinker, L. 1974. Interlanguage. Error analysis: Perspectives on second
language acquisition, ed. by Richards, 31-54.
Sey, K. a. 1973. Ghanaen English: An explanatory survey. London: Macmillan.
Shuy, R.W., & Fasold, R. W. 1973. Language attitudes, current trends and
prospects. Washington D.C.: Georgetown University Press.
Smith, P. M., Giles H., & Hewstone M. 1980. Sociolinguistics: A social
psychological perspective. The social and psychological contexts of
language, ed. by St Clair & Giles, 283-298.
Smith, Richards Rhodes. 1974. Awadi/Kannauji transition phenomena and their
correlates: A study in dialect geography. Cornell University Ph.D.
dissertation in Linguistics.
Sridhar, Kamal K. 1979. English in sociocultural context of India. Studies in
Language Learning 2:1.63-79. Unit for Foreign Language Study and
Research, University of Illinois at Urbana-Champaign.
, & Sridhar, S. N. 1986. Bridging the gap: Second language acquisition
theory and indigenized varieties of English. World Englishes 5:1.3-14.
St Clair, Robert N. 1980. The context of language. The social and
psychological contexts of language, ed. by St. Clair & Giles, 1 1-38.
Tucker, J. C, & Lambert, W. E. 1969. White and Negro listeners' reactions to
various American-English dialects. Social Forces 47.463-468.
ZUENGLER, Jane E. 1982. Kenyan English. The other tongue: English across
cultures, ed. by Kachru, 112-124.
Studies in the Linguistic Sciences
Volume 20, Number 1, Spring 1990
THE CONTRIBUTION OF LORENZO DOW TURNER TO AFRICAN
LINGUISTICS*
Margaret Wade-Lewis
SUNY, The College at New Paltz
Until fairly recently, the linguistic study of African languages has
focused on the African languages still spoken on the African
continent; very little serious attention has been given to the study of
remnants of African languages in the United States. This African
continent-based focus has underplayed the contribution of Africanist
linguists who uncovered interesting African linguistic research in the
U.S.A. Dr. Lorenzo Dow Turner who, in 1932, began the formal study
of Gullah, the South Carolina and Georgia coastal Creole which soon
led him to the study of West African languages and to Sierra Leone
Krio, is one such researcher. In this study the importance of his
contribution is assessed in light of contemporary scholarship.
1. Introduction
The first African American linguist to gain national and international
recognition was Dr. Lorenzo Dow Turner. Turner's reputation is largely the
result of his having been the first linguist to clearly and unequivocally establish
that African languages were not totally eradicated from the United States as a
result of African captivity. His book. Africanisms in the Gullah Dialect (1949),
documents the presence of African semantic retentions in the United States and
marks the beginning of Gullah Studies as a legitimate dimension of linguistics.
Africanisms outlines the phonology and some aspects of the morphology of
major West African languages from which Gullah developed. Furthermore, it
glosses at least 4,000 lexical items in Gullah from thirty African languages,
many of them personal names.''
Although Africanisms is well-known in linguistic circles. Turner's other
work end his life-long interest in African languages and culture are less familiar.
The purpose of this paper is to document the role he played in illuminating the
impact of African languages and culture on the United States, and his role in
collecting materials on African languages and culture in Africa and Brazil. An
examination of Turner's research is timely because of the renewed emphasis
on research on African languages in the United States, and the continuing
interest in Gullah Studies. Furthermore, focus on his work demonstrates the
190 Studies in the Linguistic Sciences 20:1 (Spring 1990)
value of cultural diversity in the professional ranks of American higher
education, and the enriching influence of multicultural data in the curricula.
Finally, Turner's role as a pioneer conscientiously dedicated to the pursuit of
data on New World African linguistic retentions, makes his life a model for
present day scholars to emulate.
2. Lorenzo Turner's education and early scholarship
Lorenzo Dow Turner was born on August 21, 1895, the last of four sons to
Rooks and Elizabeth (Freeman) Turner. He spent his first ten years in Elizabeth
City, North Carolina, after which his family moved to Rockville, Maryland.
Turner's family is an old and distinguished North Carolina clan, tracing its
American beginnings to the birth of Polly Rooks in Gates County in 1799
(Jones, 1952:12). Because the family members were born free, they were able
to experience greater opportunities than many of their peers, often becoming
professional persons such as seamstresses and cabinetmakers, doctors and
professors. Turner's father. Rooks, for example, a member of the third
generation, was one of the first men to gain a Master's degree from Howard
University. He later founded the Elizabeth City Normal School (Turner Williams,
1986; Bell, 1989). Both parents stressed educational attainment. Therefore, in
1910, Turner completed high school at the Howard University Academy, and
four years later he completed his A.B. degree from Howard University (1914) at
age nineteen. As a result of working on steamboats during the summer. Turner
was able to finance his study for the Master's degree at Harvard University,
completing the Master's of Arts in English in 1917.
During several summers and one leave year. Turner pursued the Ph.D. in
English from the University of Chicago, completing it in 1926 with the
dissertation Anti-Slavery Sentiment in American Literature Prior to 1865. His
early research concentrated on literature — first, the publication of Anti-Slavery
Sentiment (1929), followed by the publication of Readings from Negro Authors
for Schools and Colleges: With a Bibliography of Negro Literature (1931), an
anthology edited with Otelia Cromwell and Eva B. Dykes.
2.1 Turner's development as a linguist
According to his widow, Mrs. Lois Turner Williams, Turner's interest in
linguistics developed during the latter 1920's while he served on the faculty of
Howard University (1919-1928), and later at Fisk (1928-1946). Often Turner
taught summer school at other African American universities in the South.
Among them were Alcorn A. and M. College in Mississippi (1935), Atlanta
University in Georgia (1945) and Tuskegee Institute in Tennessee. During at
least one summer (1929), his assignment took him to South Carolina State
College at Orangeburg, which is sixty miles upcountry from Gullah territory
(Turner Williams, 1986). Gullah had often been the subject of folktales, novels
and comments by folklorists and linguists, among them Harris (1887), Mencken
(1929) and Johnson (1930), but there had been no scientific studies of it. The
usual explanation of its origins was that Gullah was the result of 'archaic
Wade-Lewis: The contribution of Lorenzo Dow Turner to African Linguistics 191
English dialects' being maintained among a group of Africans isolated from the
mainstream American culture (Gonzales, 1922; Krapp, 1924; Johnson, 1930).
When Turner was in Orangeburg, he was able to observe Guliah first-
hand. He noticed that his students and local residents spoke a dialect
phonologically, syntactically, and semantically different from any other he had
known. Over time he also learned some of their unusual personal names called
"basket names." As he listened and contemplated, the 'archaic English dialect'
hypothesis of the origins of Guliah seemed more and more implausible to him.
At that point, Turner began to suspect that the speech of his students had been
influenced by the languages of West Africa (Turner Williams, 1986a; See also
Wade-Lewis, 1988:8). Stoney and Shelby's Black Genesis, published a year
after Turner's obsen/ations, suggested a much more limited African language
influence than Turner was to find during his research. They concluded that
perhaps twenty Umbundu terms had been preserved in Guliah, six of seven of
which were in common usage (Stoney & Shelby, 1930:xv).
Turner took advantage of the first opportunity available to him to study
linguistics by attending the Linguistic Institute in New York during the Summer
of 1930. While history is now silent on this detail, it is possible that Turner was
encouraged to persue his interest by anthropologist Mark Hanna Watkins, one
of his colleagues at Fisk.2
At the Institute, Turner met Hans Kurath and other luminaries in the world
of American linguistics. In December of that same year, Kurath, Director of the
Linguistic Atlas of the United States and Canada Project, wrote Turner and a
number of other persons, inviting them to participate in the data collection for
the Atlas (Kurath, 1930). Turner replied immediately, indicating his interest in
attending the 1931 Institute in New York. Expressing a desire to gain further
skills in collecting and analyzing dialect data, he stated to Kurath, 'I like it better
than any work I have ever done' (Turner, 1930). The linguist in Turner
apparently took shape during the summers of 1930 and 1931.
Armed with the Atlas approach to interviewing informants, and an 800-item
questionnaire prepared for the Atlas work, spanning a range of lexical,
phonological and grammatical items. Turner collected data in the field in the
South for the Atlas project during several summers. Furthermore, he attended
the Linguistic Institute again in 1934 at Brown University and gained additional
background by taking other linguistics courses at Brown University (Turner,
1940:1).
2.2 Turner's Guliah research
By the Fall of 1932, Turner had applied for and received a grant from The
American Council of Learned Societies, the chief funding source for the Atlas
project to conduct the Guliah research which led to his famous study. He was
the first American linguist to carry out systematic interviews of speakers of
Guliah. Turner remained on the Sea Islands until latter December, 1932,
returned again in Summer, 1933 and during several subsequent Summers.
192 Studies in the Linguistic Sciences 20:1 (Spring 1990)
Based on one of his letters to Melville Herskovits, Turner collected data again in
1940, at which time he concluded his research and finished his manuscript for
publication (Turner, 1940b). In addition to conducting interviews, he also
collected over 600 wire and rubber tape recordings, some of which he played
when he made presentations on Gullah for the Linguistic Society of America
and the American Dialect Society and other organizations, beginning in
December, 1932 (Wade-Lewis, 1988:15), seventeen years before his book was
published.
3. Turner's study of African languages
Turner's second opportunity to continue his formal study of linguistics
came when he concluded that in order to properly analyze his Gullah data, he
needed a background in African languages. Therefore, in 1935, after his trip to
collect data on Louisiana Creole, 3 Turner applied for a fellowship from the
American Council of Learned Societies and a grant-in-aid from the Humanities
Institute of Fisk University. These enabled him to spend 1936-37 at the School
of Oriental and African Studies of the University of London.
On April 6, 1936, when Turner wrote Daniel Jones, Secretary of the
International Phonetic Association, to pay his membership dues, he informed
Jones that:.
Next year at the University of London, in addition to pursuing certain
courses in phonetics, I should like to study the phonetic structure of
certain West African languages with a view to determine, if possible,
the nature and extent of African survivals in Gullah (Turner, 1936).
Subsequently, Turner travelled to London where he studied under the
direction of Ida C. Ward, Head of the Department of African Languages. His
concentration was on Kimbundu, KiKongo, Yoruba, Efik, Ewe, Twi, Fante,
Hausa, Mende, Ga and Wolof. He also worked each day with informants from
West Africa and took advanced courses in phonetics with Daniel Jones and L.
E. Armstrong (Turner, 1940:1). During the summer of 1937, while Turner was
still in London, he travelled to Paris to interview more than twenty Africans from
the area once called French West Africa. These interviews increased his
appreciation and understanding of Francophone African culture (Turner, 1940).
According to another letter from Turner to Herskovits, Turner had hoped to
continue his study and research on African languages by spending the months
following his London trip in West Africa (Turner, 1936b). However, finances
apparently did not permit him to do so. His opportunity to conduct research in
Africa did not come until fifteen years later in 1951-52.
In the meantime. Turner returned to the United States, spending 1938-39
as a Research Fellow in Linguistics at Yale University. During that period, he
studied with Edward Sapir, took a course in Arabic, and learned Umbundu from
materials made available to him by Rev. Henry C. McDowell, a former
missionary to Angola (Turner, 1940:2; and Negro History Bulletin, 1957:26).
Wade-Lewis: The contribution of Lorenzo Dow Turner to African Linguistics 193
As was the case with many of the African Americans educated in the latter
19th and early 20th centuries, among them W. E. B. DuBois, Carter G. Woodson,
and Zora Neale Hurston, Turner felt a sense of social responsibility resulting in
a mission to produce as much scholarly data about African people as possible.
Turner's particular contribution was to be grammars, dictionaries, and books on
retained semantic items from African languages, as well as collections of
folklore and music. John Work, the arranger of spirituals, a member of the
Department of Music at Fisk, was to transcribe the musical scores. Turner
looked forward to conducting field work in Africa, Brazil, Jamaica, Haiti, British
and Dutch Guiana and Louisiana (Turner, 1946). Hence, on a regular basis he
applied for grants to make his research possible.
3.1 Turner's analysis of Africanisms in Brazil
In 1940 Turner applied for and received a grant from the Rosenwald Fund
to study African language and culture in Brazil. In preparation he studied
Portuguese and read the available books on the African influence in Brazil.
According to his 'Proposals':
From my study of the importation of Negroes from Africa to Brazil, and
from the knowledge I have at present of Negro speech in Brazil, I find
that with few exceptions the West African languages which have
influenced the sea-island speech of South Carolina and Georgia
appear likewise to have influenced the speech of Negroes in Bahia
and Pernambuco (Turner, 1940-1).
Mentioning that he had already located 4,000 Africanisms in Gullah, Turner
noted that his proposed research on African linguistic retentions in Brazil would
be "a valuable extension in another part of the world of the work in linguistic
geography which has been in progress in New England for the past few years
(Turner, 1940:2).
Turner was more than gratified to find that in Bahia many African people
had not only retained a form of Yoruba as their native language, but various
cultural practices as well. In the first article on his research in Brazil, Turner
noted that many of his Brazilian informants:
...were born in Africa, still speak fluently their native language,
frequently have in their possession valuable papers and pictures
relating to West Africa, and correspond regularly with their relatives
living there (Turner, 1942a:58).
Turner's data focused on Yoruba-Brazilians, since they comprised the
largest and strongest African influence in Northeastern Brazil. He noted that
their influence was followed by that of the Dahomeans and Angolans
(1942a:55-67). He also observed that at that time some Brazilian Africans
continued their contact with the African mainland. Some of them (1) made trips
to Africa for visits and to stay for periods of years; (2) corresponded by letter with
194 Studies in the Linguistic Sciences 20:1 (Spring 1990)
relatives and friends in Nigeria; and (3) conducted trade in tobacco, sugar, and
dried beef between Brazil and Nigeria. Within the text of the article. Turner
reproduced copies of letters, passports, birth, baptism, marriage and burial
certificates of Brazilian Africans who had maintained ties with relatives in
Nigeria. The focus of Turner's published articles on Brazil is linguistic —
anthropological. Unfortunately, his linguistic data on African retentions from the
Brazilian trip are still unpublished. Among them are a Yoruba dictionary,
folklore, particularly folk tales and stories, some history and music. The music
and some folklore are recorded on many 12-inch discs. These discs contain
material in five West African languages, but particularly Yoruba, Fon and
Kimbundu (Turner, 1958:107).
3.2 Turner's fieldwork In Africa
During the second stage of Turner's career, he was able to travel to
Nigeria in 1951-52, with the assistance of a Fulbright Fellowship. His purpose
was to study African languages first hand and to collect language and folklore
data. His major assignment was as Visiting Lecturer in the Department of
English at University College, Ibadan, Nigeria (Negro History Bulletin, 1957:
26). In addition, during the year Turner was able to purchase an automobile and
travel 20,000 miles to various other areas of West Africa, visiting universities
and mission schools, lecturing, attending cultural events and collecting data. He
spent two months in Freetown, Sierra Leone (Kansas City Times, June 22,
1951 :7; and Say: Alumni Magazine of Roosevelt University, Spring, 1951 :5.13).
One of the misfortunes of Turner's career was that while he published
regularly and a good deal, he was not able to publish the bulk of his linguistic
research data. Wherever he traveled, he utilized the Linguistic Atlas approach,
talking with informants and recording their speech, songs and folklore. His plan
to write studies of New World African-language-influenced Creoles In Jamaica,
Haiti, British and Dutch Guiana, Brazil, and Louisiana, as well as studies of
African languages in Africa, was only partially fulfilled by the time of his death in
1972. Turner did not have the opportunity to travel and conduct research in the
Caribbean. Among the major constraints Turner faced were the lack of funds for
assistance in typing and manuscript preparation, limited leave time to make
possible his meticulous phonetic transcriptions of data, and limited publication
opportunities. These, coupled with the demands on his schedule as English
Department Head and coordinator of African Studies for the most of his career
(Negro History Bulletin, 1957:26), and the academic, social and political
engagements resulting from his status as the best known African American
linguist, served to limit the time available to him to prepare his findings for
publication.
During 1951-52, for example, he collected a prodigious amount of
linguistic and cultural data in Africa. Among his unpublished manuscripts are:
(1) grammatical notes on the Temne and Mende languages of Sierra Leone; (2)
notes on Freetown Creole of Sierra Leone; (3) a Yoruba language course; (4)
Yoruba tales in translation; (5) Yoruba songs and stories; (6) assorted African
folktales; (7) Cameroon Creole proverbs, riddles, and stories; (8) a manuscript
Wade-Lewis: The contribution of Lorenzo Dow Turner to African Linguistics 195
entitled Folktales from Africa; (9) and a manuscript entitled Chronicles of Africa:
Ancient, Medieval and Modern. Many of the unpublished African folk tales have
now been collected and copyrighted by other researchers.
4. Turner's research on Krio for the Peace Corps
During the third stage of Turner's career, he left Fisk University in 1946 to
become one of the first faculty members of Roosevelt College in Chicago.
Except for some teaching at the University of Illinois, Chicago Circle, in the mid-
1960's, Turner spent the rest of his academic career at Roosevelt until his
retirement in 1970 and subsequent death on February 10, 1972.
During the Roosevelt College years. Turner continued to make
contributions to African linguistics. In the late 1940's he became a founding
member of the African Studies Program, one of the first in a non-African
American university in the United States. He had previously been coordinator of
African Studies at Fisk University from 1944-1946. Beginning in 1960, Turner
served as the Faculty Coordinator of the Peace Corps Project at Roosevelt,
having been awarded $54,579 from the Department of Health, Education and
Welfare. He was responsible for preparing volunteers for service in Sierra
Leone and for preparing language texts on Sierra Leone Krio. At that time
Turner noted that Krio was spoken by 16 ethnic groups in Sierra Leone {Hyde;
Park Herald, March 29, 1961 : 16). As a result of the grant. Turner compiled two
Krio texts. The first is An Anthology of Krio Folklore and Literature: With Notes
and Inter-linear Translations in English (1963), designed as a text for Peace
Corps volunteers to Sierra Leone.
Two years later. Turner wrote Krio Texts: With Grammatical Notes and
Translation in English (1965), also to be utilized to teach Krio to Peace Corps
volunteers. Krio Texts is divided into four chapters entitled: (1) A History of the
Krio People of Sierra Leone; (2) The Sounds of Krio; (3) Grammatical Notes
(including use of the major parts of speech); and (4) Krio Texts (including
greeting, numerals, familiar conversations and proverbs). The final portion of
the text contains the English translations. Turner also prepared tape recordings
for Chapters II, III and IV, with additional classroom drills. The tapes feature
native speakers (See Preface of Krio Texts, 1965). Both of Turner's Krio
publications are important documents for non-Krio speakers, but since they are
currently out of print, their use is limited to persons who are able to locate library
copies.
5. The relationship between Turner and Herskovits
Turner's growth as an Africanist linguist was complemented by the
mutually beneficial relationship he shared with anthropologist Melville
Herskovits. Between 1940 and 1960, they were the two most prominent
advocates of the African retentions hypothesis in the United States.
Turner and Herskovits were counterparts in many ways: (1) both were
born in 1895 of education conscious parents; (2) both developed interest in
196 Studies in the Linguistic Sciences 20:1 (Spring 1990)
Africa during the 1920's; (3) both served on the faculty of Howard University
(Herskovits as a Assistant Professor in Anthropology 1925-26; Turner as Chair
of English from 1917-1928); (4) both gained higher degrees from The University
of Chicago end later relocated to the Chicago area, Turner to Roosevelt (1946)
and Herskovits to Northwestern (1927); (5) both were instrumental in the
founding of African Studies in the United States (Turner chaired African Studies
at Fisk from 1944-1946 and become a founder of African Studies at Roosevelt;
Herskovits founded the first program at a non-African American University in
1948); (6) both travelled and studied in the South (Herskovits studying the
physical anthropology of Africans; Turner studying Louisiana Creole (1935) and
Gullah Creole (1932-1940)); (7) both travelled and collected data in
Pernambuco, Brazil (Turner in 1940-41; Herskovits in 1942-43); (8) both
travelled and studied in Africa (Turner spending 1951-52 in Nigeria and Sierra
Leone; Herskovits making his first trip to Dahomey in 1931 and several
subsequent trips to other areas); (9) they were two of the three founders of the
Negro Studies Committee of the American Council of Learned Societies, on
which both served from 1940-1950, and they appeared on numerous panels
together; (10) both conducted research on African retentions in the Americas
and were the leading scholars in their specialties during their lifetimes.
The interaction between the two was enriching and enlightening. They
met, if not before 1925, during academic year 1925-26 while Turner was
English Department Head at Howard University and Herskovits was a visiting
Assistant Professor in the Anthropology Department. Between 1925 and 1956,
their encounters were numerous. Because they were the strongest and best
known proponents of the African retentions hypothesis in linguistics and
anthropology, they were drawn together by common interests, as well as the
need to share data and ideas. They interacted with many of the same scholars,
participated in lectures at each other's universities, and defended their positions
against opposition from those who rejected the African retentions hypothesis.
One set of materials which documents their relationship is the letters they
wrote each other. The first correspondence of record is a letter dated
September 9, 1936 from Turner to Herskovits, informing Herskovits of his
approaching research trip to London. Herskovits responded on September 23,
1936 in a cordial and supportive letter. Thereafter, the two corresponded until
May 15, 1956. The final letter of record is written by Herskovits. It offers to
recommend Turner for grants other than the Ford Foundation grant for which
Turner had been turned down because of age.'* Herskovits died in 1963.
Analysis of their written correspondences and other unpublished
documents reveals that the influence Turner and Herskovits exerted on each
other, and the support which they gave each other has been underestimated.
For example, it is often noted that Herskovits utilized pages from Turner's
unpublished version of Africanisms (1949) to strengthen his arguments for
African cultural retentions in Tlie Myth of ttie Negro Past (1941). Herskovits
acknowledges this material which he quotes on pp. 276-279. He also cites
Turner on pp. 37, 191, and 316.
Wade-Lewis: The contribution of Lx)renzo Dow Turner to African Linguistics 197
Herskovits requested a copy of Turner's manuscript in a letter to Turner, as
a result of having heard Turner's presentation on Gullah at the Conference of
the Modern Languages Association in New York in December, 1938
(Herskovits, 1939). In a return letter, Turner agreed to have his quotes from his
manuscript printed, noting that Africanisms was almost ready for publication
(Turner, 1939).
Herskovits had mentioned African linguistic retentions in Caribbean
Creoles his book Suriname Folklore (1936) with Frances Herskovits, but he
recognized that in Myth (1941) a strong case for linguistic retentions in the
United States needed to be made. In his letter to Turner, Herskovits stated:
Your papers will give me some telling ammunition in establishing the
invalidity of the position of those who insist that everything in
southern speech is derived from European sources, and since the
results of this study will, undoubtedly, be important in giving direction
to support for future projects, I think you will agree with me that it is of
the utmost importance that the most forceful presentation possible be
made (Herskovits, 1939).
Often the two read each other's manuscripts and commented on them,
sometimes later writing reviews for journals. For example. Turner reviewed
Herskovits' Myth (1941) in The Journal of Negro H/sfo/y (April, 1942b: 185-187).
Later, when Africanisms (1549) appeared, Turner recommended to the
University of Chicago Press that Herskovits be selected as the reviewer for
either The Journal of American Folklore or American Anthropologist (Turner,
1949a). Together Turner and Herskovits insured the continued advancement of
the African retentions hypothetis in American scholarship. One of its strongest
contemporary proponents is Alleyne (1980), whose solid scholarship reflects
the influence of both.
6. Conclusion
Lorenzo Dow Turner was a well-known personage in the world of
linguistics during his life time. He took quite seriously the quest to expose
Americans to African languages and culture in both Africa and the New World.
Toward this end, he collected research data in the United States in the Sea
Islands and Louisiana, and in Brazil and Africa, to demonstrate the cultural
continuities between Africa and the New World. As a linguist he viewed folklore
and music as important cultural manifestations from which linguistic
grammatical and semantic content could be drawn. Therefore, wherever he
went, he participated in African religious and cultural ceremonies, devising
surveys for informants, tape recording data and selecting speech samples for
phonetic transcription. In publishing Africanisms he was able to influence
American linguistics and creole studies considerably, in the process becoming
the pioneer in the Gullah Studies specialty. Simultaneously, he dispelled the
notion that although African linguistic retentions were evident in Haiti
(Comhaire-Sylvain, 1936), Brazil (Mendonga, 1935), Suriname (Herskovits and
Herskovits, 1937), and other areas of the New World, they had been lost in the
198 Studies in the Linguistic Sciences 20:1 (Spring 1990)
United States. To date no scholar has written more extensively than Turner on
retained linguistic Africanisms.
Though there is continuing debate about the significance of some of the
data in Africanisms (Hair, 1965; Mufwene, 1985), it is clear that Turner opened
the way for the scientific study of Gutlah. This thrust is continued in the work of a
cadre of dedicated scholars, including Cunningham (1970), who wrote the first
transformational analysis of Gullah; Jones-Jackson (1978), who analyzed
convergent processes; Nichols (1976), who analyzed male/female speech
differences; Cassidy (1980, 1983), who placed Gullah in context with Caribbean
Creoles; and Holm (1983b), who compared Gullah and Barbadian. The work of
Mufwene (University of Georgia - Athens) under a National Science Foundation
and National Endowment for Humanities grants to conduct the first longitudinal
study of Gullah represents one of the most recent effor in the study of this
language.
Work on Sierra Leone Krio has also continued. Among the researchers
are E. D. Jones (1962, 1971), who studied 19th century Krio; Fyle and Jones
(1980), who compiled a Krio-English dictionary; F. C. Jones (1983), who studied
English semantics in Krio; and Harris (1984), who has recently analyzed the
spread of Krio.
In 1956 Turner applied for a Ford Foundation Grant in order to take a
leave to complete the Yoruba dictionary and other projects resulting from his
data collection in Africa (Turner 1955:1-3). The application was not considered
because Turner had passed his sixtieth birthday (Turner, 1956). No doubt one
day some of the emerging linguists will edit his unpublished manuscripts,
creating the opportunity for all to gain a more complete sense of Turner as one
of two early African American contributors to scholarship in African linguistics.
NOTES
The research in the Turner and Herskovits Collections of the Melville
Herskovits Library at Northwestern University was made possible by a 1986
Grant for Research and Creative Projects, State University of New York, College
at New Paltz. I am grateful to two SLS anonymous reviewers and the editors for
their comments and suggestions on an earlier version of this paper. Any errors
of facts and/or interpretation are mine alone.
1 Gullah is an African-English Creole spoken in the coastal South Carolina
and Georgia Sea Islands. Turner traced the name Gullah to two sources: 'a
Liberian tribe and its language' (/go la/, /go ra/, /gu la/, /gu ra/; and /r] go la/, oa
tribe in the Hamba basin cf Angola" (1949:194).
Some researchers, several of them contemporary, maintain that Turner's
claims for an African substrate in the United States are not as strong as it would
appear on the surface since the majority of the terms collected are personal
Wade-Lewis: The contribution of Lxjrenzo Dow Turner to African Linguistics 199
names. Often Gullah speakers no longer know the particular African language
semantics. See Hair (1965), Cassidy (1980), and Mufwene (1985). On the other
hand, Rickford (1979) maintains that naming practices are one valid means of
assessing African retentions.
2 Turner's work was preceded by that of Dr. Mark Hanna Watkins,
Professor in the Anthropology Department at Fisk University and then Howard
University. Watkins was a student of Edward Sapir. He received the Ph. D. from
the University of Chicago in 1933, with the dissertation, A Descriptive Grammar
of Chichewa (1933). This research made him both the first American and the
first African American to publish a grammar of an African language. His
dissertation was published by the Linguistic Society of America in 1937.
Watkins and Turner made a strong impression on Dr. Raleigh Morgan, a
former student of both end a graduate of Fisk University, who is now a recently
retired faculty member from the University of Michigan. According to Morgan, his
love for languages began with his first Latin course in high school. This love,
nurtured at Fisk by both Watkins and Turner, resulted in his decision to gain a
Ph. D. in linguistics. In 1946-47 Morgan conducted fieldwork in St. Martin's
Parish of Southwest Louisiana on Louisiana Creole in an attempt to write a
study locating Africanisms as Turner had in Gullah (Morgan, May 29, 1986; from
Morgan, June 28, 1985). Morgan completed his Ph.D. at the University of
Michigan and has had a long and distinguished career as a Romance linguist
and a specialist in African-French Creoles. He is best known for Tiie Regional
French of County Beauce Quebec (1975). The Hague: Mouton.
3 Turner studied Louisiana Creole as part of his long-term plan to analyze
New World Creoles for African linguistic content.
4 The Turner/Herskovits letters are the subject of another Wade-Lewis
article entitled: 'The African Substrate Hypothesis and the Turner/Herskovits
Connection' (Forthcoming).
REFERENCES
Alleyne, Mervyn. 1980. Comparative Afro-American: An historical comparative
study of English-based Afro-American dialects in the new world. Ann
Arbor: Karoma Publishers, Inc.
Bell, Charlotte. 1989. Interview with M. Wade-Lewis on une 2. Philadelphia.
Blok, H.P. 1959. Annotations to Mr. Turner's africanisms in the gullah dialect.
Lingua 8.306-321.
BOKAMBA, Eyamba G. 1982. The Africanization of English. The other tongue:
English across cultures, ed by Braj B. Kachru, 77-98. Champaign:
University of Illinois Press.
Cassidy, Frederic G. 1980. The place of Gullah. American Speech 55.3-16.
200 Studies in the Linguistic Sciences 20:1 (Spring 1990)
. 1983. Sources of the African element in Guilaln. Studies in Caribbean
language, ed. by Lawrence Carrington, D. R. Craig, & R. Todd Dandare.
75-81. St. Augustine: University of the West Indies. 75-81.
Comhaire-Sylvain, Suzanne. 1936. Le Creole haitien. Port-au-Prince:
L'Auteur.
Cunningham, Irma E. 1970. A syntactic analysis of Sea Island Creole (Gullah).
University of Pennsylvania Ph.D. dissertation in Linguistics.
Decamp, David. 1974. Foreword to Lorenzo Turner's Africanisms in the Gullah
dialect. Ann Arbor: University of Michigan Press, v-xi.
Frazier, E. Franklin. 1942. The Negro Family and Bahia, Brazil. American
Sociological Review 7.445-480.
Fyle, C. N., & E. D. Jones. 1980. A Krio-English Dictionary. New York: Oxford
University Press.
Gilbert, Glenn, ed. 1987. Pidgin and Creole languages: Essays in memory of
John E. Reinecke. Honolulu: University Press of Hawaii.
Gonzales, Ambrose E. 1922. The Black border: Gullah stories of the Carolina
coast. Columbia, South Carolina: The State Company.
Greenberg, Joseph. 1966. The Languages of Africa. Bloomington: University of
Indiana Press.
Hancock, Ian. 1970. Dictionary of Sierra Leone Krio. Unpublished manuscript.
. 1971. A study of the sources and development of the lexicon of Sierra
Leone Creole. Unpublished University of London School of Oriental and
African Studies Ph.D. dissertation in Linguistics.
. 1983. Aspects of standardization in West African Creole English.
Proceedings of the Mid-American Linguistics Conference, ed. by Frances
Ingemann, 295-307. Lawrence: University of Kansas Press.
Hair, P. E. 1965. Sierra Leone items in the Gullah dialect of American English.
Sierra Leone Review 4.79-84.
Hall, Robert A., Jr. 1950. The African substratum of Negro English (Review of
Turner, 1949). American Speech 25 (April):51-54.
Harris, Joel Chandler. 1887. Free Joe and other Georgia sketches. New York:
Holt.
Harris, J. W. 1984. Language contact, pidgins and the emergence of Kriol in
the Northern Territory: Theoretical and historical perspectives. University
of Oueensland Ph.D. dissertation in Linguistics.
Herskovits, Melville. 1939. Letter to Lorenzo Turner, October 17. Evanston:
Northwestern University Library, Herskovits Collection.
. 1941. The myth of the Negro past. New York: Harper and Brothers.
, & Frances S. Herskovits. 1937. Suriname folklore. New York: Harper
and Brothers.
HOLM, John. 1983b. On the relationship of Gullah and Bahamian. American
Speech. 58:4.303-318.
Hurston, Zora Neale. 1931. Hoodoo in America. Journal of America Folklore,
317-417.
Ivy, James. 1950. Africanisms in the Gullah Dialect by Lorenzo Turner. Crisis
57. (April):263-264.
Johnson, Guy. 1930. Folk culture on St. Helena Island, South Carolina. Chapel
Hill: The University of North Carolina.
Wade-Lewis: The contribution of Lorenzo Dow Turner to African Linguistics 201
Jones, Connie. 1953. The Rooks: Famous North Carolina family observes
153rd anniversary. Color (March):12-17.
JONES, E. D. 1962. Mid-nineteenth-century evidences of a Sierra Leone patois.
Sierra Leone Language Review 1.19-26.
. 1971. Krio: An English-based language of Sierra Leone. The English
language in West Africa, ed. by J. Spencer. London: Longman.
Jones, F. C. 1983. English-derived words in Sierra Leone Krio. University of
Leeds Ph.D. dissertation in Linguistics.
Jones-Jackson, Patricia. 1978. The status of Gullah: An investigation of
convergent processes. University of Michigan Ph.D. dissertation in
Linguistics.
Krapp, George P. 1924. The English of the Negro. American Mercury. 2:5.190-
195.
KURATH, Hans. 1930. Letter to Lorenzo Turner and Others. December 11.
Evanston: Northwestern University Library, Turner Collection.
McDavid, Raven I., Jr. 1950. Review of Africanisms in the Gullah dialect by
Turner. Language 26.323-333.
Mencken, H. L. 1929. The American language. New York: Alfred A. Knopf.
MENDONpA, Renato. 1935. A influencia africana no Portugues do Brasil. Rio de
Janeiro: Editora Civilizagao Brasileira.
MORGAN, Raleigh. 1985. Letter to M. Wade-Lewis. June 28.
. 1986. Interview with M. Wade-Lewis, May 29. Ann Arbor.
MuFWENE, Salikoko. 1985. The linguistic significance of African proper names
in Gullah. New West Indian Guide 146-166.
. 1986. The universalist and substrate hypotheses complement one
another. Substrata versus universals in Creole genesis, ed. by Pieter
Muysken & Nerval Smith, 129-162. Amsterdam: John Benjamin Publishing
Company.
NEGRO History Bulletin. 1957. Profile of a scholar: Lorenzo Turner 21:2
(November) 26.42.
New York Times. 1933. Records in Gullah heard by linguists. Sunday, January
1:16 Column 3.
Nichols, Patricia. 1976. Linguistic change in Gullah: Sex, age and mobility.
Stanford University Ph.D. dissertation in Linguistics.
Obituary notice for Lorenzo Turner. 1972. Jet 41 (March 2):14.
O'Reilly, Merry. 1952. Dr. Turner returns — 20,000 miles through West Africa.
Say: Alumni Magazine of Roosevelt University, 5:13.
Ortiz, Fernando. 1938. La mijsica sagrada de los Negros yoruba en cuba.
Estudios afro-cubanos. 2.89-104.
Raimundo, Jacque. 1933. O Elemento afro-negro no lingua portuguesa. Rio de
Janeiro.
Rickford, John. 1988. Dimensions of a Creole continiuum. Stanford: Stanford
University Press.
Roosevelt gets Peace Corps grant. 1961. Hyde Park Herald. March 29:16.
Stoney, Samuel G., & Gertrude M. Shelby. 1930. Black genesis. New York:
McMillan.
Turner, Lorenzo. 1930. Letter to Hans Kurath, December 24. Evanston:
Northwestern University Library, Turner Collection.
202 Studies in the Linguistic Sciences 20:1 (Spring 1990)
. 1936. Letter to Daniel Jones, April 6. Evanston: Northwestern University
Library, Turner Collection.
_. 1936b. Letter to Melville Herskovits, September 30. Evanston:
Northwestern University Library, Turner Collection.
_. 1939. Letter to Melville Herskovits, October 24. Evanston: Northwestern
University Library, Turner Collection.
_. 1940. Proposals by Lorenzo Turner for a study of Negro speech in
Brazil. January. Evanston: Northwestern University Library, Turner
Collection.
_. 1940b. Letter to Melville Herskovits. March 24. Evanston: Northwestern
University Library, Turner Collection.
_. 1941a. Letter to Melville Herskovits. February 21.
_. 1946. Application to the American Philosophical Society for a grant from
the research funds, January 8. Evanston: Northwestern University Library,
Turner Collection.
_. 1949a. Letter to Paul Corbett of the University of Chicago Press. June
15. Evanston: Northwestern University Library, Turner Collection.
Turner travels to Sierra Leone. 1951. Say: Alumni magazine of Roosevelt
University (Spring), 5:13.
Turner visits Sierra Leone. 1951. Kansas City Times, June 22, 7.
Turner, Lorenzo. 1955. Ford Foundation proposal for a plan of work on West
African folklore and languages, November. Evanston: Northwestern
University Library, Turner Collection.
Turner Williams, Lois. 1986. Telephone Interview with M. Wade-Lewis on
March 26.
. 1986. Interview with M. Wade-Lewis on May 26. Chicago.
Wade-Lewis, Margaret. 1988. Lorenzo Dow Turner: First African American
linguist. Occasional Paper #2. Philadelphia: Temple University Institute of
African and African American Affairs.
. Forthcoming. The substrate hypothesis and the Turner/Herskovits
connection.
APPENDIX
Publications by Lorenzo Turner:
Books:
Cromwell, Otelia, Lorenzo Turner, & Eva Dykes (eds.) 1931. Readings from
Negro authors for schools and colleges: With a bibliography of Negro
literature. New York: Harcourt, Brace.
Turner, Lorenzo. 1929. Anti-Slavery sentiment in American literature prior to
1865. Washington, D. C: Association for the Study of Negro Life and
History.
. 1949. Africanisms in the Gullah dialect. Chicago: University of Chicago
Press. (Reprinted 1973. Ann Arbor: University of Michigan Press).
Wade-Lewis: The contribution of Lorenzo Dow Turner to African Linguistics 203
_. 1963. An anthology of Krio folklore and literature: With notes and inter-
linear translations in English. Chicago: Roosevelt University.
_. 1965. Krio texts: With grammatical notes and translations in English.
Chicago: Roosevelt University.
Articles:
Turner, Lorenzo. 1941. Linguistic research and African survivals.American
Council of Learned Societies Bulletin. Washington, D. C. 32:68-89.
. 1942a. Some contacts of Brazilian ex-slaves with Nigeria, West Africa.
Journal of Negro History 27.55-67.
. 1945a. Comments on word lists from the South. Publications the
American Dialect Society 3.
. 1945b. Notes on the sounds and vocabulary of Gullah. Publications of
the American Dialect Society 3.13-28.
. 1947a. Problems confronting the investigator of Gullah. Gullah.
Publications of the American Dialect Society 9.74-84.
. 1947b. Roosevelt College: Democratic haven. Opportunity 25.223-225.
. 1953. New directions in jazz research — symposium. The Record
Changer, 14-17:48-49.
. 1955a. The impact of western education on the African's way of life.
Africa today, ed. by Charles Grove, 147-171. Baltimore: Johns Hopkins
Press.
. 1955b. The odyssey of a Zulu warrior. Journal of Negro History 40.305-
317.
. 1956. Welt Whitman and the Negro. Chicago Jewish Forum 15:1.5-11.
. 1957a. The Negro in Brazil. Chicago Jewish Forum 15:232-236.
. 1957b. Our African heritage. Say: Alumni Magazine of Roosevelt
University, 15-19.
Book Reviews:
Turner, Lorenzo. 1942b. The myth of the negro past, by Melville Herskovits.
Journal of Negro Education 11.185-187.
. 1942c. The Negro caravan: Writings by American Negroes, ed. by
Sterling A. Brown et al. Journal of Negro History 27.219-222.
. 1950a. The Negro in northern Brazil, by O. da Costa Eduardo. Journal of
American Folklore 63.490-492.
. 1950b. Some sources of southernisms, by Mitford M. Mathews.
Language 26.167-170.
. 1952. The sin of the prophet, by Truman Nelson. Journal of Negro
History 37.466-468.
. 1954. The outsider, by Richard Wright. Journal of Negro History 39.77-
80.
. 1955a. Dr. Dan: Poneer in American surgery, by Helen Buckler. Journal
of Negro History 40.191-193.
. 1955b. Libretto for the republic of Liberia, by Melvin B. Tolson. Poetry
86.174-176.
204 Studies in the Linguistic Sciences 20:1 (Spring 1990)
. 1955c. The Palm-Wine drinkard, by Amos Tutuola. Midwest Folklore
5.189-191.
_. 1957b. The Negro in American culture, by Margaret Just Butcher.
Chicago Jewish Forum 16.56-57.
All checks 'and money orders must be In U.S. Dollars, drawn on a U.S. Bank,
and made payable to University of Illinois . Sales Tax must be included as follows (ex-
cept tax exempt organizations: IL 7 1/4%, IN 5%, MI 4%, MN 6%, OH 5%, WI 5%.
While quantities last. Volumes 7:2 through 16:2 are being offered at half-price.
2
Fall 1977
1
Spring 1978
2
Fall 1978
1
Spring 1979
2
Fall 1979
0:1
Spring 1980
3:2
Fall 1980
1:1
Spring 1981
1:2
Fall 1981
2:1
Spring 1982
2:2
Fall 1982
^:1
Spring 1983
3:2
Fall 1983
4:1
Spring 1984
4:2
Fall 1984
5:1
Spring 1985
5:2
Fall 1985
5:1
Spring 1986
3:2
Fall 1986
Papers in General Linguistics $2.50 $5.00
Papers in General Linguistics $2.50 $5.00
Papers in General Linguistics $2.50 $ 5.00
Papers in General Linguistics $2.50 $5.00
Relational Grammar and Semantics $2.50 $5.00
(Editor: Jerry L. Morgan)
Papers in General Linguistics $2.50 $5.00
Studies in Arabic Linguistics $2.50 $5.00
(Editor: Michael J. Kenstowicz)
Papers in General Linguistics $2.50 $5t©G
Dimensions of South Asian Linguistics
(Editor: Yamuna Kachru) $2.50 $^t09
Papers in General Linguistics $2.50 $5.00
Papers on Diachronic Syntax: $2.50 $5.00
Six Case Studies
(Editor: Hans Henrich Hock)
Papers in General Linguistics $2.50 $3^QQ
Studies in Language Variation: $2.50 $5.00
Nonwestern Case Studies
(Editor: Braj B. Kachru)
Papers in General Linguistics $2.50 $#.00
Language in African Culture and Society $2.50 $5.00
(Editor: Eyamba G. Bokamba)
Papers in General Linguistics $2.50 $5.00
Linguistic Studies in Memory of $2.50 $5.00
Theodore M. Lightner
(Editor: Michael J. Kenstowicz)
Papers in General Linguistics $3.00 $6.00
Illinois Studies in Korean Linguistics $3.00 $6.00
(Editor: Chin-W. Kim)
STUDIES IN THE LINGUISTIC SCIENCES
The following issues are available
Vol. 17:1
Spring 1987
Vol. 17:2 Fall 1987
Vol. 18:1 Spring 1988
Vol. 18:2 Fall 1988
Vol. 19:1 Spring 1989
Vol. 19:2 Fall 1989
Vol. 20:2
Fall 1990
Vol. 20:3 Spring 1991
Vol. 21:1 Spring 1991
Vol. 21:2 Fall 1991
Vol. 22:1 Spring 1992
IN PREPARATION:
Vol. 22:2 Fall 1992
Papers from the 1986 South Asian
Languages Analysis Roundtable
(Editor: Hans Henrich Hock)
Papers in GeneraJ Linguistics
Papers in General Linguistics
Papers in General Linguistics
Papers in General Linguistics >
The Contribution of African Linguistics
to Linguistic Theory, Vol. 1
(Editor: Eyamba G. Bokamba)
Linguistics for the Nineties:
Papers from a Lecture Series in Celebration
of the Department's Twenty-Fifth Anniversary
(Editor: Hans Henrich Hock)
Meeting Handbook: Thirteenth South Asian
Languages Analysis Roundtable, 25 - 27 May
1991, University of Illinois at Urbana-Champaign
Papers in General Linguistics
Illinois Studies in Korean Linguistics, 2
(Editor: Chin-W. Kim)
Papers in General Linguistics
Twenty-Five Years of Linguistic Research
at the University of Illinois at Urbana-Champaign:
Post-graduate research by Doctoral and Master's
degree students in the Department of Linguistics
(Editors: Braj B. Kachru and Frances Vavrus)
Vol. 23:1
Spring 1993 Papers in General Linguistics
$6.00
$6.00
$6.00
$6.00
$7.50
$7.50
$7.50
$5.00
$7.50
$7.50
$7.50
$7.50
$7.50
FOR EARLIER ISSUES AND ORDER INFORMATION SEE INSIDE COVER
Orders should be sent to:
SLS Subscriptions, Department of Linguistics
University of Illinois
4088 Foreign Languages Building
707 S. Mathews
Urbana, Illinois 61801
tudies in
he Lineuistic Sciences
410 STX:
St92
20:2 FALL 1990 COPY 2
OF
THE DEPARTMENT'S TWENTY-FIFTH ANNIVERSARY
PLUS A CUMULATIVE INDEX OF
STUDIES IN THE LINGUISTIC SCIENCES
VOLUMES 1 TO 19
Preface v
Introduction by Elmer H. Antonsen (Acting Head) ix
HENRY KaHANE (University of Illinois):
The establishment of Linguistics at Illinois 1
RONALD W. LANGACKJZR (University of California, San Diego):
Cognitive Grammar: The symbolic alternative 3
JERROLD M. SadoCK (University of Chicago):
A trimodular account of Yiddish syntax 31
FREDERICK J. NEWMEYER (University of Washington):
Some issues in language origins and evolution 51
DAVID Odden (Ohio State University):
Phonology and its interaction with syntax and morphology 69
LlSE MENN (University of Colorado, Boulder):
Aphasic language under discourse pressure: Functional syntax
vs. psycholinguistic function 109
PETER H. LOWENBERG (Georgetown University):
Standards and norms for World Englishes: Issues and
attitudes 123
CiABRIELLA HermON (University of Delaware):
Syntactic theory and language acquisition: A case against
parameters 139
S. N. Sridhar (State University of New York, Stony Brook):
What are applied linguistics? 165
Index to Studies in the Linguistic Sciences, Volumes 1-19
A. Author index 177
B. Title index 197
Department of Linguistics
University of Illinois
STUDIES IN THE LINGUISTIC SCIENCES
PUBLICATION OF THE DEPARTMENT OF LINGUISTICS
COLLEGE OF LIBERAL ARTS AND SCIENCES
UNIVERSITY OF ILLINOIS AT URBANA-CHAMPAIGN
EDITOR: Hans Henrich Hock
EDITORIAL BOARD: Eyamba G. Bokamba, Chin-chuan Cheng, Georgia M.
Green, Erhard W. Hinrichs, Hans Henrich Hock, Braj B. Kachru, Yamuna
Kachru, Chin-W. Kim, Charles W. Kissebcrlh, Howard Maclay, Jerry L.
Morgan, Rajeshwari Pandharipandc, and Ladislav Zgusta.
AIM: SLS is intended as a forum for the presentation of the latest original
research by the faculty and especially students of the Department of Linguistics,
University of Illinois, Urbana-Champaign. Invited papers by scholars not as-
sociated with the University of Illinois will also be included.
SPECIAL ISSUES: Since its inception SLS has devoted one issue each year to
restricted, specialized topics. A complete list of such special issues is given on the
back cover.
BOOKS FOR REVIEW: Review copies of books may be sent to the Editor,
Studies in the Linguistic Sciences, Department of Linguistics, University of
Illinois, 4088 Foreign Languages Building, 707 S. Mathews, Urbana, Illinois
61801.
SUBSCRIPTION: Normally, there are two issues per year. Requests for sub-
scriptions should be addressed to SLS Subscriptions, Department of Linguistics,
University of Illinois, 4088 Foreign Languages Building, 707 South Mathews,
Urbana, Illinois 61801.
UPCOMING ISSUES: Vol. 21:1: Papers in General Linguistics: Vol. 21:2:
Illinois Studies in Korean Linguistics, 2 (edited by Chin-W. Kim).
Price: $7.50 (per issue)
STUDIES IN THE LINGUISTIC SCIENCES
LINGUISTICS FOR THE NINETIES:
PAPERS FROM A LECTURE SERIES IN CELEBRATION OF
THE DEPARTMENT'S TWENTY-FIFTH ANNIVERSARY
PLUS A CUMULATIVE INDEX OF
STUDIES IN THE LINGUISTIC SCIENCES
VOLUMES 1 TO 19
EDITOR
Hans Henrich Hock
EDITORIAL ASSISTANT
Lynne Murphj
VOLUME 20, NUMBER 2
FALL 1990
DEPARTMENT OF LINGUISTICS, UNIVERSITY OF ILLINOIS
URBANA, ILLINOIS 61801
i
i
CONTENTS
Preface v
Introduction by Elmer H. Antonsen, Acting Head ix
Henry Kahane (University of Illinois):
The establisfiment of Linguistics at Illinois 1
Ronald W. Langacker (University of California, San Diego)
Cognitive Grammar: Thie symbolic alternative 3
Jerrold M. Sadock (University of Chicago)
A trimodular account of Yiddish syntax 31
Frederick J. Newmeyer (University of Washington)
Some issues in language origins and evolution 51
David Odden (Ohio State University)
Phonology and its interaction with syntax and morphology 69
Use Menn (University of Colorado, Boulder)
Aphasic language under discourse pressure: Functional syntax vs.
psycholinguistic function 109
Peter H. Lowenberg (Georgetown University)
Standards and norms for World Englishes: Issues and attitudes 1 23
Gabriella Hermon (University of Delaware)
Syntactic theory and language acquisition: A case against parameters 139
S. N. Sridhar (State University of New York, Stony Brook)
What are applied linguistics? 165
Index to Studies in the Linguistic Sciences, Volumes 1-19
A. Author index 177
B. Title index 197
Preface
This volume marks — and celebrates — the completion of twenty years of
Studies in the Linguistic Sciences. The journal was started in 1971 at the
instigation of Braj B. Kachru, head of Linguistics. The idea was to publish two
issues a year, one on general themes in linguistics, the other a special issue
dedicated to a specific topic.
During its early years, the journal was edited by Charles W. Kisseberth,
assisted by an Editorial Board consisting of Chin-chuang Cheng, Braj. B.
Kachru, Michael Kenstowicz, Chin-W. Kim, Jerry Morgan, and Ladislav Zgusta.
Over the years, the editorial organization has undergone a number of changes.
In 1975, Braj B. Kachru, Jerry Morgan, and Ladislav Zgusta joined Charles W.
Kisseberth as editors, and all other faculty members with linguistics appoint-
ments became members of the Editorial Board, reflecting the fact that as the
occasion required, they all were called upon to referee submitted contributions.
1978 marked the appointment of Robert N. Kantor and Ladislav Zgusta as
Review Editors. In 1980, Chin-W. Kim replaced Robert N. Kantor as one of the
Review Editors. A major change took place in 1986, when Michael J. Ken-
stowicz took over as sole Editor. Since 1988, the editorship has been the
responsibility of Hans Henrich Hock. With Volume 21, Number 1, he will be
joined by Charles W. Kisseberth as Review Editor.
While the major responsibility of the Editor(s) is concerned with general
issues, special issues are the responsility of guest editors. Since the inception
of the journal, the following special issues have been produced:
Papers on Hindi syntax, edited by Yamuna Kachru. (1 :2, Fall 1971)
Papers on syntax and semantics, edited by Georgia M. Green. (2:1,
Spring 1972)
Studies in Baltic linguistics, edited by Hans Henrich Hock and Mi-
chael J. Kenstowicz. (2:2, Fall 1972)
Papers on South Asian linguistics, edited by Braj B. Kachru. (3:2, Fall
1973)
Papers on phonetics and phonology, edited by Charles W. Kisse-
berth and Chin-W. Kim. (4:2, Fall 1974)
Papers on historical linguistics: Theory and method, edited by Ladis-
lav Zgusta and Hans Henrich Hock. (5:2, Fall 1975)
Topics in Relational Grammar, edited by Jerry Morgan, Georgia
Green, and Peter Cole. (Special section of 6:1, Spring 1976)
Papers on African linguistics, edited by Eyamba G. Bokamba and
Charles W. Kisseberth. (6:2, Fall 1976)
Studies in East Asian linguistics, edited by Chin-chuan Cheng and
Chin-W.Kim. (7:2, Fall 1977)
vi Studies in the Linguistic Sciences 20:2 (Fall 1990)
Linguistics in the seventies: Directions and prospects (Forum lec-
tures presented at the 1978 Linguistic Institute of the Linguistic
Society of America), edited by Braj B. Kachru. (8:2. Fall 1978)
Relational grammar and semantics, edited by Jerry L. Morgan. (9:2,
Fall 1979)
Studies in Arabic linguistics, edited by Michael J. Kenstowicz. (10:2,
Fall 1980)
Dimensions of South Asian linguistics, edited by Yamuna Kachru.
(11:2, Fall 1981)
Papers on diachronic syntax: Six case studies, edited by Hans Hen-
rich Hock. (12:2, Fall 1982)
Studies in language variation: Non-western case studies, edited by
Braj B. Kachru. (13:2, Fall 1983)
Language in African culture and society, edited by Eyamba G. Bo-
komba. (14:2, Fall 1984)
Linguistic studies in memory of Theodore M. Lightner, edited by
Michael J. Kenstowicz. (15:2, Fall 1985)
Illinois studies in Korean linguistics, edited by Chin-W. Kim. (16:2,
Spring 1986)
Papers from the 1986 South Asian Languages Analysis Roundtable,
edited by Hans Henrich Hock. (17:1, Spring 1987)
The Contributions of African linguistics to linguistic theory: Proceed-
ings of the 20th Annual Conference on African Linguistics, Vol. I,
edited by Eyamba G. Bokamba with Rick Treece and Dorothy E.
Evans (Associate Editors). (19:2, Fall 1989)
The companion volume of the last-mentioned item (to appear as SLS
20:1) is now being completed. Together with volume 19, number 2, it cele-
brates twenty years of the tradition of Annual Conferences on African Linguistics
that was established by the Department of Linguistics.
In addition, all of Volume 20 celebrates the twenty-fifth anniversary of the
Department of Linguistics. (More on this can be found in Elmer H. Antonsen's
Introduction and in Henry Kahane's contribution to this volume.) And, in order
to more fully celebrate that anniversary. Volume 20 consists of three issues:
The just-mentioned second part of the proceedings of the 20th Annual
Conference on African Linguistics (20:1), the present issue (20:2), and the
Meeting Handbook of the Thirteenth South Asian Languages Analysis Round-
table (25-27 May), edited by Hans Henrich Hock with editorial assistance by
Lynne Murphy, and published as a special issue (20:3).
The present issue contains papers presented in a special lecture series
celebrating the silver jubilee of the Department of Linguistics. The series was
planned in 1989/90, when Charles W. Kisseberth was Executive Officer of the
Department, and took place in the fall of 1990, when Elmer H. Antonsen was
Acting Head. Its success owes much to the support of these two colleagues.
The Department also owes gratitude to Lynne Murphy who took care of much of
Preface vii
the correspondence regarding the lecture series and the acquisition, con-
version, and preliminary editing of the manuscripts.
Some of the papers were presented under a title different from the one
under which they appear in this volume. Moreover, at his request, Henry
Kahane's contribution appears here in much reduced form. The original
program is given below.
Henry Kahane (University of Illinois)
Linguistics as personal experience: The formation of a depart-
ment. 12 September 1990.
Ronald Langacker (University of California, San Diego)
Cognitive Grammar: The symbolic alternative. 24 September
1990.
Jerrold M. Sadock (University of Chicago)
Yiddish solved I 1 October 1990.
Frederick Newmeyer (University of Washington)
Some problems of language origins and evolution. 8 October
1990.
David Odden (Ohio State University)
The relation between phonology and other parts of a grammar.
15 October 1990.
Use Menn (University of Colorade, Boulder)
Aphasic language under discourse pressure: Functional syntax
vs. psycholinguistic function. 29 October 1990.
Peter Lowenberg (Georgetown University)
Standards and norms for World Englishes: Issues and attitudes.
5 November 1990.
Gabriella Hermon (University of Delaware)
Syntactic theory and language acquisition: Current issues and
prospects. 1 2 November 1 990.
S. N. Sridhar (State University of New York, Stony Brook)
What are Applied Linguistics? 26 November 1990.
Except for the opening lecture, presented by Henry Kahane, who joined
the University of Illinois faculty in 1941 and was the person most responsible for
the successful establishment of the Department in 1965, the papers were by
former students of the Department who have established a name for themselves
in the field. Brief biographical notes are given below:
Gabriella Hermon received her Ph.D. in Linguistics in 1981. Her dis-
sertation topic was 'Non-nominative subject constructions in the Government &
Binding framework" (advisor: Georgia Green). She currently is Assistant Pro-
fessor in the Department of Education Studies, University of Delaware, with joint
appointment in Linguistics.
Ronald Langacker joined the University of Illinois graduate program in
Linguistics in the fall of 1963, prior to the establishment of the Department, and
viii Studies in the Linguistic Sciences 20:2 (Fall 1990)
received his Ph.D. in 1966, just after the Department's establishment. His dis-
sertation, 'A transformational sketch of French syntax', was directed by Robert B.
Lees. He now is Professor and Chair of Linguistics, University of California at
San Diego.
Peter Lowenberg completed his Ph.D. dissertation, 'English in the Malay
archipelago: Nativization and its functions in a sociolinguistic area', in 1984 un-
der the supervision of Braj B. Kachru. He is Associate Professor of Linguistics
at Georgetown University.
Use Menn's dissertation, 'Pattern, control, and contrast in beginning
speech: A case study in the development of word form and word function", was
completed in 1976 under the direction of Howard Maclay. She is now Asso-
ciate Professor of Linguistics at the University of Colorado at Boulder.
Frederick Newmeyer joined the Department in 1967 and received his
Ph.D. in 1969, two years later! His dissertation, 'English aspectual verbs', was
written under the direction of Robert B. Lees. He is Professor of Linguistics at
the University of Washington.
David Odden's 1981 dissertation, 'Problems in tone assignment in Sho-
na', was directed by Charles W. Kisseberth. Odden now is Associate Professor
of Linguistics at Ohio State University.
Jerrold Sadock entered the graduate program just prior to the establish-
ment of the Department and received his Ph.D. in 1968. His dissertation bears
the title 'On the notion "Sentence type" ' (director: Henry Kahane). He is Profes-
sor and Chair of Linguistics at the University of Chicago.
S. N. Sridhar's dissertation, 'Cognitive determinants of linguistic struc-
tures: A cross-linguistic experimental study of sentence production', was com-
pleted in 1980 under the direction of Charles Osgood. He is Associate Profes-
sor of Linguistics at the State University of New York, Stony Brook.
As a special bonus, this issue of Studies in the Linguistic Sciences addi-
tionally contains an index for volumes 1 - 19 of the journal. The index has been
keyboarded by Eileen Sutton, alphabetized by Lynne Murphy, and proofed by
Amy Repp, all of whom deserve special thanks.
As always, the Department is grateful to the College of Liberal Arts and
Sciences for publication subvention, and to the Language Learning Laboratory,
especially James E. Gothard, for technical support. Last but not least I have the
pleasant task of thanking Beth Creek, Cathy Huffman, and Eileen Sutton of the
Department Office for their help and especially Lynne Murphy, my editorial
assistant.
May 1991 Hans Henrich Hock (Editor)
Introduction
When Hans Henrich Hock, Editor of Studies in the Linguistic Sciences,
invited me to contribute an Introduction to the present volume devoted to cele-
brations of the twenty-fifth anniversary of the Department of Linguistics, I was at
first uncertain whether I should accept the honor (and thus the task) or leave it to
someone else, since I am only temporarily more closely associated with the
Department (as Acting Head from May 1990 to January 1992). Nevertheless, I
have been intensely interested in the activities of the Department over a long
period of time, having held the title of Associate Professor and then of Professor
of Linguistics 'without budgetary implications* ever since I came to the Universi-
ty of Illinois in 1967 to accept a position in Germanic linguistics in the De-
partment of Germanic Languages and Literatures, just two years after the formal
founding of the Department of Linguistics.
In actual fact, however, my association with the Department is somewhat
older than the Department itself, strange as that may seem. During my graduate
student days at Illinois in the Department of German (as it was then called)
between 1956 and 1959, I was introduced by Frank Banta, then Associate
Professor and Chairman of German (and a linguist), to a small group of linguists
from various departments who met on a regular basis to present and hear talks
on linguistic topics. Duhng the academic year 1958-1959, I even had the honor
to serve as a research assistant in linguistics to Professor Henry Kahane. From
him and Renee Kahane I learned a great deal about scholarly research through
osmosis, simply by observing how they collaborated on major projects; and my
wife and I were the recipients of many personal kindnesses. After I had left
Champaign-Urbana to accept a full-time position at Northwestern University
and had defended my dissertation, I was invited by Henry Kahane to return to
the Campus in 1961 to present a talk before the Linguistics Club, then under the
presidency of Mary Temperley, who is now a member of the faculty of the
Division of English as an International Language. I thus had the distinction of
becoming the first former student-member of the Linguistics Club to be invited
back as an outside speaker. These then are my meager qualifications for writ-
ing the Introduction to this volume of Studies in the Linguistic Sciences devoted
to the celebration of the twenty-fifth anniversary of the Department of Linguistics
and to the twentieth anniversary of the journal itself (for which see the Preface to
this issue).
As a student, of course, I was not fully aware of all the behind-the-scenes
efforts to establish the Department of Linguistics, although there was no missing
the fact that Henry Kahane was diligently at work preparing the ground. A
search through the archives reveals that the actual proposal was worked out in
1963 by a committee chaired by Robert B. Lees (English, Communications
Research), then Director of the Program in Linguistics, and consisting further of
X Studies in the Linguistic Sciences 20:2 (Fall 1990)
Joseph H. D. Allen (Spanish, Italian, and Portuguese), Katherine Aston (Eng-
lish), Frank G. Banta (German), Joseph B. Casagrande (Anthropology), Kenneth
L. Hale (Anthropology), Lee S. Hultzen (Speech), Henry R. Kahane (Spanish,
Italian, and Portuguese), Frederic K. Lehman (Anthropology), Rado Lencek
(Russian), Howard S. Maclay (Communications Research), Charles E. Osgood
(Communications Research), Angelina R. Pietrangeli (Spanish, Italian, and
Portuguese), Victor Terras (Russian), and Willard R. Zemlin (Speech).
When I returned to Illinois as a faculty member after an absence of eight
years, the Department was already firmly established and flourishing under its
first Head, Robert B. Lees. (See Henry Kahane's contribution to the present vol-
ume and more fully elsewhere; cf. below).
The Department immediately assumed a leading role in the development
of linguistics in the United States, as can be seen from the COPE (Committee
on Program Evaluation) Report of August 1976, which had high praise for both
the administrative and the academic performance of the Department and placed
it clearly in the top 10 linguistics departments in the country, with a number of its
programs ranking considerably higher. Only three years after its founding, the
Department hosted two summer Linguistics Institutes of the Linguistic Society of
America, with a repeat performance in 1978. After a quarter-century, the De-
partment continues to play a leading role in the discipline, as witnessed by the
commemorative events of this past year. In the fall of 1 990, for the Twenty-Fifth
Anniversary Lecture Series, eight linguists of national and international
standing who had received their training in the Department were invited back to
the campus, and their papers are presented in this volume. In addition, two
state-of-the-art conferences and one national meeting were organized as part of
the celebration:
The Organization of Phonology: Features and Domains, 2 - 4 May 1 991 ,
sponsored by the Department of Linguistics, the College of Liberal Arts and
Sciences, the Cognitive Science/Artificial Intelligence Steering Committee,
and the Beckman Institute, with the co-operation of nine other campus units
(Organizing Committee: Charles W. Kisseberth (Chair), Jennifer Cole, and
Chin-W. Kim; INVITED SPEAKERS: Diana Archangel!, G. N. Clements, John
Goldsmith, Morris Halle, Bruce Hayes, Larry Hyman, Junko ltd, Patricia
Keating, Michael Kenstowicz, John McCarthy, Armin Mester, Alan Phnce,
Doug Pulleyblank, Donca Steriade, and Moira Yip).
The Thirteenth South Asian Languages Analysis (SALA) Roundtable, 25
- 27 May 1991, sponsored by the Department of Linguistics with the co-
operation of International Programs and Studies, the Division of English as
an International Language, and the Programs in Comparative Literature, in
Religious Studies, and in South and West Asian Studies (Local Organizing
Committee: Jennifer Cole, Hans Henrich Hock (Chair), Braj B. Kachru,
Yamuna Kachru, Rajeshwari Pandharipande, and Girdhari Tikku). At this
year's meeting, two internationally renowned scholars, Professors Bh.
Elmer H. Antonsen: Introduction xi
Krishnamurti and Ladislav Zgusta, were honored, as was (as an unan-
nounced surprise) the 'father' of all SALAs, Professor Braj B. Kachru.
Linguistics and Computation: Computational Linguistics and the Foun-
dations OF Linguistic Theory, 1 3 - 1 5 June 1 991 , sponsored by the Depart-
ment of Linguistics, the College of Liberal Arts and Sciences, the Cognitive
Science/Artificial Intelligence Steering Committee, and the Beckman Insti-
tute, with the co-operation of the Department of Computer Science, the De-
partment of Germanic Languages and Literatures, and the Language Learn-
ing Laboratory (Organizing Committee: Jerry Morgan (Chair), Jennifer Cole,
and Georgia Green; Invited Participants: Steven Abney, Robert Berwick,
Jennifer Cole, John Coleman, Sandiway Fong, Alan Frisch, Erhard Hinrichs,
David Johnson, Ronald Kaplan, Mark Lieberman, Ivan Sag, and Richard
Sproat.
A volume of approximately 300 pages, entitled Twenty-Five Years of
Linguistic Research: Post-Graduate Research by Students at the University of
Illinois, has been prepared as one of the major activities to celebrate the 25th
anniversary of the Department. Compiled and edited by Braj Kachru and
Frances Vavrus, a special feature of the volume is the inclusion of essays on the
history of the Department and its founding by Henry Kahane, Charles Osgood,
Robert B. Lees, Braj B. Kachnj, and Charles W. Kisseberth. These essays were
originally written for a volume planned by Thomas Sebeok in 1974, but for
certain reasons, the volume never materialized. The papers, therefore, have a
'flavor* of the early 1970's. In the introductory note to these papers, Henry
Kahane writes: "In certain ways, the birth and growth of a new academic
department at a large modern university reflects changes in the intellectual and
social climate of the Nation; it thus merits a modicum of general attention. A
new discipline usually exists long before its official recognition; it is hidden
under labels until the time when its inherent dynamics pushes it to the fore,
when, as the saying goes, its time is ripe." Thus the perspectives of five
individual personalities, and their impact on the Department, essentially up to
the 1970's, is presented in these papers. The second part of the volume con-
sists of (1) abstracts (ca. 300-500 words each) of approximately 170 doctoral
dissertations; (2) abstracts of approximately 50 masters theses; and (3) a list of
over 20 Ph.D. dissertations in progress. The third part contains the following
indices: (1) author index, (2) language index, (3) geographical region, (4) area
of concentration, and (5) advisors. This volume will be available from the
Department of Linguistics in the fall of 1991.
A particular strength of the Department has always been the diversity and
high quality of its programs, not only in theoretical, socio-, and historical-com-
parative linguistics, but also in non-Western languages and in applied linguis-
tics. The research productivity of the Department's faculty was displayed in
striking fashion through an exhibition of faculty publications in the University
Library during the month of March 1991, as part of the celebration of the Depart-
ment's twenty-fifth anniversary. The exhibit was expertly arranged and anno-
tated by Assistant Librarians Tom D. Kilton and Gail P. Hueting of the Modern
xii Studies in the Linguistic Sciences 20:2 (Fall 1990)
Languages and Linguistics Library. Even though ail six of the large display
cases in the main corridor of the University Library were utilized, only books
could be included, and even then each faculty member had to be restricted to
no more than four books apiece, forcing difficult choices upon many.
Under the imaginative leadership of Braj Kachru, who was Acting
Department Head from 1968 to 1969 and Head from 1969 to 1979, great strides
were taken to enhance the position of the Department not only on the national
and international scenes, but also within the campus community, even though
the Department had seemed to be in a critical state after the departure of Robert
Lees, Arnold Zwicky, and Theodore Lightner, and from the collapse of the
shortlived post-sputnik era of national educational enlightenment. He initiated
not only the journal Studies in the Linguistic Sciences, which has since attained
such stature as to be included among the journals surveyed by the Biblio-
graphie linguistique/Linguistic Bibliography, but also the annual departmental
Newsletter, now in its twenty-second year. He encouraged members of the
faculty to launch a conference on African linguistics and himself co-organized a
conference on South Asian linguistics. These conferences were the stimuli for
the development of permanent national and international traditions of regular
conferences attended by the leading authorities in these fields. We have just
witnessed the Twentieth Annual Conference on African Linguistics, the
proceedings of which are presented in SLS 19:2 and 20:1 (Fall 1989 and
Spring 1990, edited by Eyamba Bokamba), and the Thirteenth South Asian
Languages Analysis Roundtable, for which the Meeting Handbook is presented
in SLS 20:3 (Spring 1991, edited by Hans Henrich Hock).
The Division of Applied Linguistics, established in 1976, is a research unit
within the Department of Linguistics with a cross-cultural and cross-linguistic
focus. It coordinates and initiates research activities in the areas of bilingual-
ism/multilingualism, language and development, literacy, and English in a glo-
bal context. The Division has organized and partially supported numerous in-
ternational conferences, colloquia, and seminars, and collaborates with activi-
ties related to English in the international context initiated by Larry E. Smith of
the East-West Center in Honolulu. The Division has international academic
contacts with scholars and institutions in such countries as India, Nigeria, Paki-
stan, and Singapore, which facilitates the exchange of research in various
areas of applied linguistics, particularly in the study and analysis of English in
non-native contexts. Among other projects, the Division of Applied Linguistics is
at present taking a leading role in the establishment of a databank for various
types of non-native Englishes. Research projects initiated by the Division have
been supported in the past by the Ford Foundation, the American Institute of
Indian Studies, the Research Board of the University of Illinois at Urbana-Cham-
paign, among other agencies. Coordinator of the Division is Professor Braj B.
Kachru.
Under the leadership of Chin-W. Kim (Chair, 1979-1986) and Charles W.
Kisseberth (Chair, 1986-1989; Acting Head, 1989-1990), the Department main-
tained its position of eminence in the fields of phonology, syntax, and historical
Elmer H. Antonsen: Intrcxluction xiii
linguistics, and in the areas of applied linguistics and non-Western language
teaching (particularly African and South Asian). It sought and eventually re-
ceived a faculty position in semantics with formal ties to the Program in Cog-
nitive Science/Artificial Intelligence and the Beckman Institute for Advanced Sci-
ence and Technology.
The Department is intimately bound together with numerous other campus
units through formally organized joint programs, e.g. in Romance linguistics with
the Department of Spanish, Italian, and Portuguese, and the Department of
French; in teacher education with the College of Education; in psycholinguistics
with the Department of Psychology; and in applied linguistics with the Division
of English as an International Language. It has both formal and less formal
arrangements with units such as the Beckman Institute, the Center for African
Studies, the Center for East Asian and Pacific Studies, the Center for Latin
American and Carribean Studies, the Center for Russian and East European
Studies, and the programs in Cognitive Science/Artificial Intelligence, in
Religious Studies, and in South and West Asian Studies. In addition, the
Department shares faculty members through joint and/or adjunct appointments
with the following departments or programs: African Studies; Anthropology; the
Center for Advanced Study; Classics; Communications Research; Comparative
Literature; East Asian Languages; English; English as an International Lan-
guage; French; Germanic Languages and Literatures; Language Learning Lab-
oratory; Psychology; Slavic Languages and Literatures; Spanish, Italian, and
Portuguese; and Speech and Hearing Sciences.
The centrality of the Department of Linguistics within the University of Illi-
nois at Urbana-Champaign is clearly evident and accounts for the fact that an
extraordinarily large number of its faculty members currently hold administrative
appointments in campus units outside the Department: C. C. Cheng, Director of
the Language Learning Laboratory; Braj B. Kachru, Director of the Division of
English as an International Language; Chin-W. Kim, Director of the Program in
East Asian Studies; and Ladislav Zgusta, Director of the Center for Advanced
Study.
The student-body of the Department has a distinctly international make-up
and orientation, which is at least partly the result of close cooperation with the
Center for African Studies, the Program in East Asian Languages, and the Pro-
gram for South and West Asian Studies, as well as with the Division of English
as an International Language. Those graduate students who have not yet
chosen an advisor are counseled by a newly established Graduate Program
Coordinator, Professor Eyamba Bokamba, who also chairs the Examination and
Student Evaluation Committee, while the gradually increasing enrollment at the
undergraduate level has led to the formation this year of an Undergraduate
Program Committee and the appointment of an Undergraduate Program
Coordinator, Professor Braj Kachru, who acts simultaneously as the coordinator
for teaching assistants in the undergraduate linguistics courses. Coordinator for
Non-Western Languages is Professor Yamuna Kachru, assisted by Associate
Professor Rajeshwari Pandharipande. In addition to the resources of the Lan-
xiv Studies in the Linguistic Sciences 20:2 (Fall 1990)
guage Learning Laboratory, the Phonetics Laboratory, and the collections in the
Modern Languages and Linguistics Library, the Africana Library, the Asian
Library, and the University Library as a whole, linguistics students can avail
themselves of research materials in the Henry and Renee Kahane Linguistics
Research Room, which has a particularly outstanding collection of linguistics
offprints.
Over the first quarter-century, the Department has trained more than 170
Ph.D.s and 50 M.A.s. The quality of the graduate program in linguistics has
never been in doubt. While the number of undergraduates majoring in linguis-
tics has traditionally been quite small, i.e. a total of 21 for the academic year
1990/91 (partly a result of the first head's belief that students should have a
major in a particular foreign language), it is worth pointing out that this small
body of students has shown an unusually high level of academic achievement:
out of a total of eight graduating linguistics majors in 1991, two are Bronze
Tablet Scholars (i.e. they rank among the top 3% OF the entire University
GRADUATING CLASS), and FOUR were elected to Phi Beta Kappa (i.e., in addition
to other requirements, they ranked among the top 10% of graduates in the
College of Liberal Arts and Sciences, by far the largest college in the
University). This year the Department established an award for the Outstanding
Undergraduate Student in Lingustics. The names of recipients will be inscribed
on a plaque in the Henry and Renee Kahane Linguistics Research Room and
they will receive a certificate and a check for $100.
During my short tenure of one year as Acting Head of the Department of
Linguistics, I have come to know its faculty members much more intimately than
before and to recognize in them an unusual aggregate of highly competent and
dedicated scholar/teachers who will continue the proud tradition of excellence
that has been the hallmark of the Department over the past twenty-five years in
all its areas of responsibility. The highly successful, intellectually stimulating,
and socially uplifting year-long celebrations of the twenty-fifth anniversary of the
founding of the Department, so ably coordinated by Hans Henrich Hock, Chair
of the Anniversary Committee, and aided by Braj Kachru, Chin-W. Kim, Charles
Kisseberth, Jerry Morgan, Elmer H. Antonsen (ex officio), Rakesh Bhatt (Student
Advisory Panel), and Henry Kahane (honorary member), amply attest to the
energy of the individual faculty members and to the vitality of the Department as
a whole.
No one need fear for the ability of the Department of Linguistics at the
University of Illinois at Urbana-Champaign to face the new challenges that will
inevitably arise in the next quarter-century and the new millenium.
Urbana Elmer H. Antonsen
May 1991 Acting Head
Studies in the Linguistic Sciences
Volume 20, Number 2, Fail 1990
THE ESTABLISHMENT OF LINGUISTICS AT ILLINOIS*
Henry Kahane
(University of llinois at Urbana-Champaign)
At Illinois, as elsewhere. Linguistics grew not so much from a single de-
partment as from a constellation of them. By the mid- and late forties, we had a
group of faculty members, most of them associated with the Linguistic Society of
America, who, in their respective departments and through common gatherings,
tried to promote the cause of Linguistics. The most active were: in Speech, Lee
Hultzen (1896-1968), oscillating between phonetics and phonemics, and Grant
Fairbanks (1910-1964), an experimental phonetician and specialist in the
acoustics of speech; in Psychology and Communications, Charles Osgood, the
psycholinguist, and his faithful collaborator, Howard Maclay, who contributed to
the concept of hesitation phenomena; in Philosophy, Leonard Linsky (at the
University of Illinois from 1948 to 1967), the semanticist; in Anthropology,
Joseph Casagrande (with us since 1960), an ethnolinguist, with special interest
in the Amerindians, a scholar who activated the anthropologist's concern with
linguistics; and in Romance (Spanish and Italian), the present chronicler, Henry
Kahane, philologist. In terms of the general background, the terrible event of
World War II proved to be a boon for linguistics: The Linguistic Society de-
veloped the so-called Army Method for teaching foreign languages to enlisted
men, and through applied linguistics made university communities (among
them, Illinois) aware of the existence of linguistics itself.
We decided to launch a Department of Linguistics. The academic steps,
one after another, were the usual ones: (a) A small curriculum with a director
but minus a budget, using the available faculty members on released time; (b)
a modest budget for the curriculum; (c) an officially established Department un-
der a head and with members still largely from other departments; and finally
(d) a regularly constituted Department. The Department's foundation was a
long affair which took about eighteen years. We succeeded when, after many
hopeful and hopeless memoranda and frustrating nos, two Deans of our Liberal
Arts College sensed the potentialities of the newcomer: the late Lyie Lanier
(later Provost of the University) and Jack Peltason (later Chancellor of our Cam-
pus and now at the University of California at Irvine)). The program of the early
2 Studies in the Linguistic Sciences 20:2 (Fall 1990)
Stage was determined, first, by the constraint, in view of our budgetary con-
ditions, to use just the men and the courses available on campus, and second,
by our consensus to balance, within what was available, the various directions
of linguistics: These were, by then, psycholinguistics, phonology and experi-
mental phonetics, semantics, and historical linguistics. The addition of a theo-
retical linguist was the most urgent desideratum: We had no doubt that lin-
guistic theory would become the core of the curriculum. By 1961, the Program
in Linguistics was in existence as a graduate program; in 1965 departmental
status had been reached.
Five men have played a preponderant role in the early history of our
department: Henry Kahane, a historian and comparatist linking linguistics to the
humanities, the founder and first director, who put the curriculum on its feet and
established the basic design for his successors to build on; Charles Osgood,
who cooperated from the very beginning in the founding of the Department, the
widely known creator of psycholinguistics, whose influence and prestige greatly
helped to convince the skeptics, and whose field became one of the hallmarks
of the linguistic offerings at the University of Illinois; Robert Lees, the repre-
sentative of standard transformational theory, a brilliant intellectual, the first
head, who gave to the department its decisive direction and put it on the map;
Braj Kachru, our sociolinguist, who with extraordinary energy and never-failing
gentlemanliness steered the department from its modest beginnings to a com-
plex and flourishing University unit; and Charles Kisseberth. a leading neo-
transformationalist, who gathered about him an enthusiastic group of adepts
trying to push back the known frontiers. Nationwide, the curricula look very
much alike. The real image of the Department was to be found in the per-
sonalities that made up the team.
NOTE
* This contribution is a brief excerpt from the talk 'Linguistics as personal
experience: The formation of a department' with which Henry Kahane inaugur-
ated the fall 1990 lecture series in honor of the twenty-fifth anniversary of the
Department of Linguistics. It presents his personal reminiscences on the estab-
lishment of Linguistics at Illinois. The bulk of his presentation is to appear else-
where.
studies in th e Linguistic Sciences
Volume 20, Number 2, Fall 1990
COGNITIVE GRAMMAR: THE SYMBOLIC ALTERNATIVE
Ronald W. Langacker
(University of California, San Diego)
Whatever its reputation might be among those who have not been
privileged to study it, in actuality cognitive grammar is the simplest, most
natural, down-to-earth, and non-speculative account of linguistic structure that I
am aware of — the one most closely tied to obsen/able phenomena. Yet be-
cause it represents a distinct, non-standard vision of language and linguistic
investigation, a minimally adequate introduction is hardly possible in a short
paper, or even in a full course. The following should therefore be thought of as
a whirlwind tour that may afford some appreciation of its guiding spirit and
descriptive strategy. (A reasonable basis for assessing its insights and explan-
atory potential may be obtained through assiduous study of the selected refer-
ences in the Appendix.)
Language permits the symbolization of conceptual structures by means of
phonological sequences. Granted this characterization, cognitive grammar
takes the simplest, most straightforward approach possible to linguistic struc-
ture. Its central hypothesis is that language comprises semantic structures,
phonological structures, and symbolic links between them — nothing more. A
symbolic structure is said to be bipolar: A semantic structure functions as its
SEMANTIC POLE, and a phonological structure as its phonological pole, as
shown in Fig. 1(a). Semantic, phonological, and symbolic structures of any
degree of complexity are capable of being formed and coalescing as estab-
lished units, as sketched in Fig. 1(b). This much clearly has to be imputed to
language. The central thesis of cognitive grammar is that only this much need
be imputed to it. In particular, lexicon, morphology, and syntax are seen as
forming a gradation and as being fully describable by means of symbolic units
alone.
Very stringent limits are imposed on what kinds of units one can postulate.
The CONTENT requirement Specifies that the only units ascribable to a linguistic
system are semantic, phonological, and symbolic structures that are part of
overtly occurring expressions, SCHEMATIZATIONS of permitted structures, and
CATEGORIZING RELATIONSHIPS between permitted structures. To see what this
4 Studies in the Linguistic Sciences 20:2 (Fall 1990)
means, consider a phonological example. The syllables [tap], [bed], and [raen]
are parts of overtly occurring expressions. The syllable canon [CVC] represents
a schematization over such structures. And the following formula, with a solid
arrow, indicates the categorization of [tap] as an instance of the [CVC] category:
[[CVC]-^[tap]]. The content requirement rules out all descriptive constructs that
are arbitrary in the sense of not being immanent or directly discernible in the
primary data of actual expressions. For example, it precludes the use of empty
diachtics, or of any other construct attributed neither phonological nor semantic
content (e.g. phonologically null syntactic 'dummies'). It also prevents one from
generating every possible string of elements and then imposing the needed
restrictions by means of a set of 'filters' that specify what can not occur. No oth-
er framework imposes such a powerful constraint.
(a)
H
(b)
semantic pole
symbolic link
B phonological
pole
s s s s s
' 2s, 4 5 6 S, ^
— t — — t — 1 — » ■t * ^ S3
r
— i — * — i — i * i i
P P P P P
Figure 1
If it is workable, linguistic theorists ought to embrace this conception of
language structure and abandon it only with the greatest reluctance in the face
of overwhelming empihcal evidence. Why? Because it is intrinsically desirable
on grounds of conceptual unification, theoretical austerity, and naturalness.
The conceptual unification it achieves — positing only symbolic links between
semantic and phonological structures — is analogous to the 'Holy Grail' of
theoretical physics: a unified theory of the strong, weak, electromagnetic, and
gravitational forces. The content requirement imposes a severe austerity on the
types of constructs theorists are allowed to invoke. Cognitive grammar is natur-
al by virtue of being solely and squarely based on the function of language as a
way of expressing meaning, and also because it invokes only obvious or easily
demonstrable cognitive abilities, such as the following:
(1) a. the ability to form structured conceptualizations
b. the ability to perceive and articulate phonological
sequences
c. the ability to establish symbolic associations between
conceptual and phonological structures
d. the ability to use one structure as a basis for categorizing
another
e. the ability to conceive a situation at varying levels of
abstraction (schematization)
Langacker: Cognitive Grammar 5
f. the ability to detect similarities between two structures
g. the ability to establish correspondences between facets
of different stmctures
h. the ability to combine simpler structures into more
complex ones
1. the ability to impose figure/ground organization on a
scene
j. the ability to construe a conceived situation in alternate
ways
But is it in fact workable? I believe so, and along with others I have been
striving for over a decade to show just how — step by step and phenomenon by
phenomenon. But does it not conflict with a fundamental tenet of generative
theory, namely the thesis that grammar is autonomous? And has that thesis not
been established beyond all reasonable doubt? It does contradict the autono-
my thesis. However, I do not accept that thesis as having been established; I
would argue instead that consideration of the issue has been clouded by
erroneous assumptions and lack of imagination in regard to possible alterna-
tives.
For one thing, standard arguments for grammatical autonomy presuppose
an inappropriate view of linguistic semantics, namely an objectivist view based
on truth conditions. For instance, the fact that either a verb or a noun — such as
explode and explosion — can refer to the same event is taken as indicating that
they have the same meaning and consequently that the noun and verb classes
cannot be semantically definable. Suppose, however, that one adopts a
subjectivist or conceptualist view of meaning. One can then argue (and
intuitively I find it quite obvious) that explode and explosion have different
meanings — more specifically, that the nominalization of explode to form ex-
plosion involves a kind of conceptual reification. If so, semantic charac-
terizations of the noun and verb classes remain possible, at least in principle.
Also erroneous is the assumption that a grammatical morpheme must be mean-
ingless unless one can formulate a single meaning that accounts for all its uses.
We know, however, that lexical items are almost invariably polysemous, having
not just one meaning but a family of related senses. Why should the same not
be true of grammatical elements? In its different uses, for example, dative case
in German has such meanings as 'experiencer', 'recipient', and 'source of a
path'. There are plausible connections among these senses, and failure to
reduce the German dative to a single Gesamtbedeutung would not entail that it
is meaningless.
Consideration of the autonomy thesis has not been a model of conceptual
lucidity. I will understand that thesis as claiming that grammar constitutes a sep-
arate level or domain of linguistic structure — with its own primitives, repre-
sentations, and so on — that is properly described without essential reference
to meaning. Now it is commonly assumed (explicitly in Newmeyer 1983) that
6 Studies in the Linguistic Sciences 20:2 (Fall 1990)
such autonomy is established if any aspect of grammatical structure is less than
fully predictable on the basis of meaning or other independent factors — i.e. if
any facet of grammar has to be learned or stated explicitly instead of simply
'falling out' as an automatic consequence of other phenomena. And of course,
any clear-headed person must recognize that absolute predictability of this sort
cannot be achieved. Grammatical patterns and restrictions do have to be
specifically learned and explicitly described. In that sense, grammar is auto-
nomous. Crucially, however, this does not entail the autonomy thesis as
defined. To proceed from non-predictability to the further conclusion that gram-
mar represents a separate, asemantic domain of linguistic structure is to
embrace the type/predictability fallacy — it confuses two quite distinct issues,
namely what kinds of structures there are, and the predictability of their be-
havior.
Unconfusing these issues allows one to formulate a position that I will call
the SYMBOLIC ALTERNATIVE: that grammatical patterns and restrictions are in-
deed less than fully predictable, but that their description requires nothing more
than symbolic units, each with both conceptual and phonological import. This
represents a fundamental claim of cognitive grammar. In what follows, I will try
to show that a grammatical description employing only symbolic units is work-
able, at least in principle. To do this, I must start by sketching an appropriate
view of linguistic semantics.
I take a subjectivist approach to semantics, in which meaning is equated
with conceptualization in the broadest sense of that term (any kind of mental
experience). Moreover, a particular symbolic unit — such as a lexical item or a
grammatical morpheme — typically has more than one meaning. That is, its
meaning represents a complex category. Most linguistic categories are
complex in the sense that they do not reduce to any single structure — such a
category must instead by described by a network whose nodes are structural
VARIANTS and whose links are categorizing relations. Two basic types of
categorizing relationships can be distinguished. A solid arrow is used for
elaboration (or instantiation), where the categorizing structure is schematic
and its instantiation is characterized with greater precision and detail. A dashed
arrow stands for extension from a prototype, which — unlike instantiation —
implies some conflict in specifications between the two structures. A linguistic
expression having multiple, related senses is said to be polysemous: Se-
mantically it comprises a complex category representable as a network, as
illustrated in Fig. 2, where heavy lines indicate the prototypicality of certain
senses.
Cognitive semantics is encyclopedic, in that it denies the existence of any
sharp, motivated boundary between semantics and pragmatics, or 'linguistic'
and 'extra-linguistic' knowledge. Instead, an expression is thought of as flexibly
invoking a large array of potentially open-ended knowledge systems, which
provide the basis for its semantic characterization. I refer to these as cognitive
Langacker Cognitive Grammar 7
DOMAINS. For example, the conception of the overall configuration of an arm is
one cognitive domain invoked for the characterization of elbow. Similarly, the
meaning of onside kick presupposes substantial knowledge of the rules, strate-
gies, and objectives of football. Given the appropriate knowledge base,
describing the meaning of such expressions is fairly straightforward; without it,
the task is hopeless. Observe that a cognitive domain represents an /ntegrated
conception or conceptual complex — it is not equivalent to a bundle of semantic
features or criterial attributes. In this view, linguistic semantics cannot be
divorced from the study of conceptual structure and cognitive development.
GROUP OF PEOPLE
OPERATING TOGETHER
(CLANDESTINELY)
ring (as noun)
Figure 2
It is essential to realize, however, that an expression's meaning is more
than just an array of conceptual content. Linguistic meaning depends not only
on the content evoked, but also on how that content is construed. Commonly,
in fact, expressions that invoke roughly the same body of conceptual content
are nevertheless semantically distinct because they construe it in different
manners. There are many aspects or dimensions of construal, only a few of
which are singled out here for brief illustration.
First, a conceived entity or situation can be characterized at different levels
of specificity and detail. Listed in (2) are three sets of expressions related in this
fashion; within a given set, each expression is schematic with respect to the one
that follows (as indicated by the solid arrows).
(2) a. thing -^ animal -^ mammal -> dog -^ beagle
b. do ^ act -» propel -^ throw -> hurl
c. Something happened. -4 Someone did something. -^
An adult propelled a physical object. -> A women threw
a rock at a mammal. -^ A muscular woman hurled a
large, jagged rock at a vicious beagle that had been
growling at her.
Observe that such relationships hold not only between lexical items, but also
between novel expressions of any size, as in (c). Indeed, there is no fun-
damental distinction in this framework between 'lexical' and 'sentential' seman-
8 Studies in the Linguistic Sciences 20:2 (FaU 1990)
tics — the same constructs are used for the semantic description of semantic
structures at any level of organization.
A second aspect of construal is the assessment of one structure against
the BACKGROUND provided by another. Under this rubric fall such notions as
presupposition, metaphor, and construal relative to different assumptions and
expectations. Previous discourse constitutes another kind of background; it is in
this respect that sentences (3a-c) contrast semantically although they describe
the same event in the same words.
(3) a. Jack insulted Jill.
b. Jack INSULTED Jill.
c. Jack insulted Jill.
d. They {even/only} have three cars.
Note that certain expressions, such as even and only in (3d), have no other
function than to indicate where something falls in regard to expectations.
A third aspect of construal is what I refer to as scope. An expression's
scope is the extent of its coverage in relevant cognitive domains, i.e., how much
of those domains it specifically evokes and relies on for its characterization. For
example, the conception of an arm provides the immediate scope for the
characterization of hand, while the conception of a hand is the immediate scope
for finger, and that of a finger for knuckle. Though usually implicit and only
vaguely delimited, scope has important structural consequences — note, for
instance, that we say fingernail rather than *handnail or 'armnail. The same
expression can often be constmed with different scopes. Thus (4a) invokes the
minimal scope ior jump (it need only include the conception of someone leaving
the ground), whereas the scope of jump in (4b) subsumes an entire scenario of
preparation, running, leaving the ground, sailing through the air, landing, and
measurement.
(4) a. She jumped to a height of seventeen inches,
b. Carl Lewis is jumping now.
The fourth dimension of construal, perspective, includes such factors as
VANTAGE POINT, ORIENTATION, and DIRECTIONALITY. The first two terms are self-
explanatory and can be illustrated by the expression in back of. In some uses,
this expression invokes an implicit vantage point; thus, in Fig. 3, The tree is in
back of the rock is appropriate with respect to vantage point 1, but not with
respect to vantage point 2. In other uses, in back of relies on the orientation of
its object; it is Jill's orientation in Fig. 3(b) — the fact that she is facing away from
Jack — that makes the sentence Jack is in back of Jill felicitous.
The term 'directionality' is also self-explanatory in examples like (5), which
describe the physical motion of an explicitly-mentioned participant.
Langacker: Cognitive Grammar
(5)
The balloon rose swiftly from the valley floor.
The rocket fell to the ground.
(a) The tree is In back of the rock. (b) Jack is in back of Jill.
Figure 3
(6)
The hill gently rises from the bank of the river.
The hill gently falls to the bank of the river.
(7)
This nerve branches just below the elbow.
These nerves merge just below the elbow.
However, consider examples (6) and (7). In each case we find a pair of sen-
tences that describe the same situation yet differ in meaning. Intuitively, more-
over, the semantic contrast is in each case ascribable to a difference in direc-
tionality. But nothing moves, at least objectively — all four sentences describe
single, static configurations. The directionality responsible for the meaning con-
trasts must therefore be subjective, i.e. a matter of construal. What we want to
say (based on intuition) is that the speaker or conceptualizer (as opposed to the
subject) scans mentally through the scene in one direction or the other. In (7),
for example, (a) is appropriate when one is mentally tracing a nerve's outward
path from the central nervous system, whereas (b) would be used when tracing
its inward path from the periphery. I take this subjective directionality, residing
in the direction of mental scanning by the conceptualizer, to be an inherent
aspect of the linguistic semantic value of such expressions.
The last dimension of constnjal is the relative prominence accorded to the
various facets of a conceptualization. By itself, of course, the term 'prominence'
is vague and uninformative. There are numerous ways in which a conceived
entity can be considered prominent, so a substantive analysis has to sort these
out and properly distinguish them. I will concentrate here on just two kinds of
prominence, both essential to grammatical structure. These are designation
and FIGURE/GROUND ORGANIZATION.
10
Studies in the Linguistic Sciences 20:2 (Fall 1990)
As one aspect of its meaning, every linguistic expression is construed as
designating some entity within its scope. I will say that it imposes a particular
PROFILE on the BASE its scope provides. Intuitively, the entity accorded this spe-
cial kind of prominence is something like a focus of attention. An expression's
profile can also be thought of as its referent— not its referent in the 'world' (if
indeed it has one), but rather its referent within the conceptualization that func-
tions as its base. For example, consider the nouns hub, spoke, and rim. In the
pertinent sense, each invokes as its base the conception of a wheel; its role
within the overall configuration of a wheel is crucial to its semantic charac-
terization. The nouns differ semantically because they profile different substruc-
tures within this common base, as sketched in Fig. 4. (Observe that profiling is
indicated by heavy lines.) We see from this simple example that two or more
expressions may invoke essentially the same conceptual content, yet have
distinct meanings by virtue of their contrasting profiles.
Figure 4
I use the term predication for the meaning of any expression, irrespective
of its size or type. There are two basic kinds of predications: those which profile
things, and those which profile relations. The terms 'thing' and 'relation' are
used in a technical sense and defined quite abstractly. By 'thing' 1 do not mean
just a physical object, but rather anything that can be characterized as a region
in some domain. When used as a noun, for instance, yellow profiles (i.e., desig-
nates), a region in color space; a notation for this is given in Fig. 5(a). Similarly,
January profiles a region within the conception of the calendrical cycle; para-
graph designates a region within a written work; and intermission profiles a
region within some kind of performance — a region characterized by the ab-
sence of the specified activity.
YELIXDW (=ADJ)
o
SIMPIi: AIEMPORAL REUnCN
Figure 5
Langacker Cognitive Grammar
The term 'relation' is also used in a very general sense. We can think of a
conceived relationship as residing in cognitive operations assessing the loca-
tion, relative position, or interaction of entities within a domain. Like things, rela-
tions can stand in profile, i.e., they can be designate by linguistic expressions.
When used as an adjective, for example, yellow profiles the relationship
sketched in Fig. 5(b). Participating in this relationship are two things: One is the
same region in color space profiled by the noun yellow; and the other is an
object that is the locus of a light sensation (usually on its outer surface). The
dashed arrow stands for the profiled relation, namely the specification that the
light sensation in question falls within the yellow region of color space.
Crucially, the entities participating in a relationship need not be distinct, salient,
or mentioned individually. The adjective yellow can therefore be relational
even though it takes only one overt argument, corresponding to the locus of the
sensation (as in yellow shirt). Since the other relational participant (a region in
color space) is uniquely identifiable from the adjective itself, there is no need to
spell it out with a separate nominal argument. Likewise, the adjective square is
considered relational even though it too takes just one overt argument (e.g.
square table). The profiled relationship (equality of the sides, etc.) holds among
subparts of the single participant, not between distinct participants.
Like nominal predications (which profile things), relational predications
sometimes invoke the same conceptual content yet differ in meaning by virtue of
their profiles. In their prototypical senses, for example, both give and receive
evoke as their base the conception of a canonical act of transfer. They contrast
semantically because they profile different facets of this complex interaction, as
shown in Fig. 6: give focuses on the agent's interaction with the mover, and
receive on the recipient's.
A • ajeat H » mover H » recipieat i ■esersy transfer >. > aotloo
^> • pereejtloo/jossessloo » eorrespoodeaee/ldeatlV t ^ • sphere of eoatrol
Figure 6
Yet profiling is insufficient by itself to distinguish many sets of relational
predications that evoke the same conceptual content. Cmcial in this regard is a
final aspect of construal, namely the relative prominence accorded the various
12 Studies in the Linguistic Sciences 20:2 (Fall 1990)
relational participants. I interpret this as being a matter of figure/ground
organization. The term trajector (tr) is used for the participant serving as the
figure in a profiled relationship; a salient entity other than the trajector is referred
to as a LANDMARK (Im). Consider the expressions in front of vs. in back of They
are clearly not synonymous, but precisely how to characterize their semantic
difference is less than obvious. (The traditional practice of referring to them as
converses or relational opposites merely labels the difference without providing
a characterization.) As sketched in Fig. 7, the two expressions pertain to the
same configuration, each profiling the relationship wherein one participant is
roughly in the line of sight leading from a vantage point to the other participant.
The difference, I suggest, is that in front oHakes the far participant as a land-
mark for locating the near one, whereas in back oftakes the near participant as
the landmark. The other participant — the one being located — is the trajector,
which I characterize as the figure within the scene.
nr FRoirr of (ij) m back of
(5>-0-^0
Figure 7
A comparable analysis is offered for the examples in (8), which I regard as
non-synonymous despite their truth-conditional equivalence.
(8) a. Line A is parallel to line B.
b. Line B is parallel to line A.
c. Lines A and B are parallel.
When I say that A is parallel to B, I am concerned with locating A and use B as a
landmark for this purpose. Conversely, 6 is parallel to A makes B the figure
within the scene and locates it with reference to A. What about the third
example? I see no reason not to take the surface evidence at face value: The
figure within the profiled relationship is not either line individually, but rather the
higher-order entity comprising both lines. There is nothing mysterious or un-
natural about this — note that comparable higher-order entities are profiled by
nouns like pair, set, row, and colonnade. When the ensemble comprising A and
B is accorded the status of trajector (i.e. relational figure), the profiled rela-
tionship no longer holds between distinct participants, but rather between what
are construed as facets of a single higher-order participant (just as in the case
of square).
Given a conceptualist semantics of this sort, based on construal, it be-
comes feasible at least in principle to claim that all valid grammatical constructs
have some kind of meaning or conceptual import. In the symbolic alternative.
Langacker: Cognitive Grammar 13
grammatical structure itself is inherently meaningful, consisting solely in pat-
terns for the structuring and symbolization of conceptual content. By choosing
one grammatical constnjction or grammatical marker rather than another, one is
inherently choosing to constme and portray a situation in a particular way — the
difference in form symbolizes a meaning difference. Construal is especially im-
portant for understanding grammatical structure: Though lexicon and grammar
form a gradation, it is not a gross distortion to say that the primary function of the
lexicon is to provide conceptual content, and that grammar imposes a particular
construal on such content.
Importantly, it is not claimed that grammar is predictable from meaning —
and certainly not from meaning of the sort contemplated in objectivist or truth-
conditional semantics. The claim is rather that a grammatical element is
inherently symbolic, or bipolar: Its semantic pole embodies a particular way of
construing conceptual content, while its phonological pole provides a way of
symbolizing that construal. Moreover, we cannot determine construal simply by
consulting intuitions — indeed, we tend to be oblivious to construal (certainly
most traditional semantic theory has been), perhaps because we are more
concerned with the content conveyed. What construals expressions impose,
and the optimal way to describe them, are matters that have to be determined
by careful investigation and ultimately require some kind of explicit justification.
What this means in practice is that an account of meaning and an account of
grammar have to be developed simultaneously, each supported and informed
by how it articulates with the other. It is the insight and coherence of the overall
account that demonstrates the viability of the general approach.
What kinds of justification can in principle be offered for semantic des-
criptions of the sort proposed? One kind is intuitive naturalness — for whatever
that may be worth. A more substantive point is that the analyses rely only on
well-established cognitive phenomena (such as figure/ground organization, the
ability to focus attention on some limited aspect of a scene, our capacity to
conceive of a situation at different levels of specificity, and so on). Furthermore,
a particular, restricted set of descriptive constructs are employed that prove
systematically applicable to an extremely broad array of diverse data. For in-
stance, the notion of profiling is applicable to all expressions at every level of
organization (not just lexical items), and trajector/landmark organization holds
for all relational predications.
Another potential source of justification are predictions about distribution
and well-formedness that follow from the different construals imputed to other-
wise similar expressions. Consider the contrast in (9a) between few and a few.
(9) a. He has {few/a few} close friends.
b. {Few/*A few} linguists have any common sense.
14
Studies in the Linguistic Sciences 20:2 (Fall 1990)
In terms of absolute quantity, they may be the same — with either one, there
might be just three close friends, for instance. I would argue, however, that few
is negative in the sense that it construes the quantity as a downward departure
from some norm or expectation, whereas a few is positive because it views the
quantity relative to a baseline of zero. These characterizations afford the pre-
diction that few, but not a few, should be able to sanction a negative polarity
item, such as any. We see from (9b) that this is in fact the case.
I have in general concentrated more on two other sources of justification:
Proposed semantic descriptions must be able to support a revealing charac-
terization of grammatical structure, and they must allow one to represent, in a
non-ad-hoc way, both the similarities and the subtle differences among sets of
expressions that are comparable in the conceptual content they invoke. Illus-
trating both points are the examples in (10), involving different uses and senses
of open or the participle opened.
(10) a. A butler OPENed the door.
b. The door OPENed easily.
c. Just then the door OPENed.
d. The door was opened by a butler.
e. the opened door
f. the open door
The respective senses of open and opened are diagrammed in Fig. 8, where
heavy lines indicate profiling, ^/"identifies the trajector (relational figure), and a
circle or ellipse represents the scope of predication.
w^ 1 — ^-\
(^^ W>-,
Figure 8
Fig. 8(a) depicts open in its use as a transitive verb. It profiles both the
transmission of energy (indicated by the double arrow) and the motion that re-
sults (represented by the single arrow). Observe that the agent is chosen as tra-
Langacker: Cognitive Grammar 15
jector, and the mover is singled out as a participant with substantial prominence
(hence a landmark). I have argued elsewhere (1982) that the corresponding
passive — be opened — also profiles this full course of action, as shown in (d).
The active/passive contrast does not reside in content or profiling, but only in
the choice of trajector. Consequently, the distinctive property of (d) is simply
that the mover (rather than the agent) stands out as the figure within the scene.
Next consider (b), The door opened easily. This expression does invoke
the efforts of an agent (otherwise the adverb easily makes no sense), yet some-
how we also want to say that it only describes what the door does, not the
agent. In the present framework, this amounts to saying that the agent and the
force it exerts are included within the scope of predication but remain un-
profiled; what (b) profiles — designates — is merely the door's resultant motion.
This construction is like a passive in that the mover (or undergoer) is selected
as relational figure, but it differs from both a passive and an active transitive by
virtue of its limited profile.
Let us now examine the contrast between (b). The door opened easily, and
(c), Just then the door opened. Both involve an intransitive sense of open that
profiles only the motion of the door (the trajector). The difference is that (c) does
not necessarily invoke the conception of an agent or the transmission of energy
— the door's motion is portrayed more as a spontaneous occurrence. To be
sure, this is a matter of degree, strongly influenced by the accompanying ad-
verbs, but nothing hinges on there being a sharp or absolute distinction. To the
extent that we do observe the contrast, it is describable with reference to scope
of predication: whether (or to what degree) the scope extends beyond the
profiled movement to encompass the force that induces it.
Finally, we must consider the distinction between the stative participle and
the simple adjective, i.e. between the opened door and the open door Each
modifier profiles a particular spatial relationship involving its trajector — pre-
cisely the same relationship in both instances. The difference is that an opened
door has to have undergone the process of opening, whereas an open door
need not have (e.g., it may have been placed on its hinges in the open position
and never have been closed). In other words, opened evokes as part of its
base the conception of the transitive event of opening, and within that base it
profiles only the final, resultant spatial configuration of the door. By contrast, the
adjective open has the same profile but does not necessarily include within its
scope any conception of the process of opening.
What have we done here? We have taken some of the constructs required
for lexical semantics and used them to describe in conceptual terms the simil-
arities and differences among expressions representing distinct grammatical
constructions (active, passive, patient-subject construction, etc.). Perhaps this
affords an initial glimpse of how a particular type of semantic description can be
1 6 Studies in the Linguistic Sciences 20:2 (Fall 1 990)
said to articulate with a certain conception of grammatical structure in a mutually
supportive fashion.
Let us now direct our attention to grammar per se. The issue is whether —
as one would hope — a workable account of grammatical structure can in fact
be devised that posits only symbolic units. Such an account will have to handle
all of the phenomena listed in (1 1), which are generally taken as supporting the
autonomy thesis:
(11) Need to account for:
a. grammatical categories;
b. grammatical rules and constructions;
c. supposed representations and primitives specific to grammar;
d. "semantically empty" grammatical markers;
e. the semantically arbitrary fact that expressions often have to
take a certain form, even though another form could perfectly
well express the same meaning;
f. non-predictabilty of the class of elements that participate in a
particular morphological or syntactic construction;
g. the apparent ability to judge grammaticality independently of
meaning;
h. restrictions that apparently have to be stated in purely formal
terms.
I will now consider each matter in turn and indicate, at least in very broad terms,
how the symbolic alternative can in principle accommodate it.
The first phenomenon is the existence of basic grammatical categories,
such as noun, verb, adjective, etc. These are often considered grammatical
"primitives", on the grounds that they are not susceptible to semantic charac-
terization — and certainly they are not if one adheres to an objectivist view of
meaning. If, however, one adopts a subjectivist view of meaning that properly
recognizes the pivotal role of construal, semantic characterizations can be en-
visaged that are at least coherent (even if not demonstrably valid). In a recent
article (1987), I have made reasonably explicit proposals about what it is that all
nouns have in common semantically, and all verbs, as well as their major
subclasses (count vs. mass nouns, perfective vs. imperfective verbs). By way of
partial justification, I showed that the analysis makes it possible — in the man-
ner of (10) and Fig. 8 above — to give precise characterizations of the semantic
similarities and differences among various types of relational predications (such
as verbs, adjectives, prepositions, infinitives, present participles, and the sever-
al kinds of past participles), and that their meanings allow us to explain much of
their grammatical behavior as well as central features of the English auxiliary. I
cannot go through the analysis here, but I recommend it as an example worked
out in considerable detail of how a symbolic account of grammar that properly
Langacker: Cognitive Grammar 17
recognizes the role of construal is able to make sense of what are usually re-
garded as purely formal classes, patterns, and restrictions.
Adopting the perspective of cognitive grammar, we can make the gener-
alization that an expression's grammatical category is determined by the nature
of its profile — it is thus a matter of construal rather than of content per se. For
this reason a transitive verb like open, its intransitive counterpart, and the stative
participle opened formed on it can all represent distinct grammatical classes
despite invoking exactly the same conceptual content (as sketched in diagrams
(a), (b), and (e) of Fig. 8). Now, I have already made a broad distinction be-
tween expressions that profile things and those that profile relations, and em-
phasized that these are technical notions defined quite abstractly (e.g., a thing
is a region in some domain, not just a physical object). We can now charac-
terize a NOUN as an expression that profiles a thing, whereas other basic clas-
ses — such as adjectives, prepositions, participles, and verbs — designate
different sorts of relations. A verb profiles a complex relation that saliently
involves time in particular ways. I call this a PROCESS. Other relational predica-
tions profile ATEMPORAL RELATIONS.
Some notational abbreviations are given in Fig. 9. A circle abbreviates a
thing. A simple relationship is represented by a line connecting the relational
participants. Some relations are complex, in the sense that they do not reduce
to a single, consistent configuration but rather comprise a series of configura-
tions, or states. A process is a complex relation that further invokes the notion of
time, in two ways. First, the component states of the process are conceived as
being distributed through a continuous span of time, represented by the arrow
(how many states are depicted diagrammatically is arbitrary — three are shown
in this diagram, just one in others; the important thing is that they form a
continuous series). Second, a process is temporal in the sense that the con-
ceptualizer scans through the component states sequentially rather than con-
struing it in a purely holistic fashion.
THIHG H2IATI0N EEIATIOH 5R0CESS
o
i: m in
Figure 9
18
Studies in the Linguistic Sciences 20:2 (Fall 1990)
Some illustration is provided in Fig. 10. The preposition /n designates a
simple atemporai relation involving two things, prototypically a relationship of
spatial inclusion. Into, in contrast, profiles a complex relation, which does not
reduce to a single spatial configuration, but resides instead in a series of such
relations. Observe that the final component state of info's profile matches the
single component state profiled by in. The dotted lines represent correspond-
ences. Here they show that into has the same trajector in all its component
states, as well as the same landmark. At least in terms of the spatial path it
descnbes, the verb enter is the same as into. The major difference is that into is
merely a complex locative predication, while enter — as a verb — highlights the
temporal evolution of the spatial relationship, in the ways just described.
(a) IN f^?) [•b] grro <•=;
(e) SNTER Uv)
Figure 10
After this all-too-brief discussion of grammatical classes, let us now con-
sider how RULES and constructions can be handled in the symbolic ap-
proach. Rules and constructions are actually not distinguished in this
framework: In accordance with the content requirement, grammatical rules take
the form of constructions characterized schematically. That is, rules are simply
schematizations over sets of overtly-occurring expressions parallel in formation,
representing whatever commonality is observable in these expressions. I thus
refer to such a rule as a constructional schema. Internally, a constructional
schema is a complex symbolic structure directly analogous to the expressions it
schematizes — it is merely more abstract. The function of a constructional
schema is threefold: (i) it captures whatever generalizations are inherent in the
primary data; (ii) it is available as a template for constnjcting or evaluating other
expressions on the same pattern; and (iii) its categorization of such an expres-
sion constitutes the latter's structural description. (Let me note in passing that
cognitive grammar basically subsumes the theory of construction grammar be-
ing developed by Fillmore (1988) and others. The major difference is that
proponents of construction grammar would not necessarily accept my con-
ceptual characterization of basic grammatical categories, hence their approach
does not achieve the full reduction of grammar to configurations of symbolic
structures.)
Langacker: Cognitive Grammar 19
What do I mean by 'construction'? A construction is a specific, symbolically
complex expression, or else a schematization over such expressions at some
level of abstraction. In the simplest case, a constmction involves the combina-
tion, or integration, of two symbolic structures to form a symbolic structure of
greater complexity. I will say that two component structures are integrated to
form a composite stmcture. Their integration is bipolar, i.e. it takes place at
both the semantic pole and the phonological pole. Integration is effected by
correspondences established at each pole between substructures of the two
components. The composite structure results from merging the two component
structures through the superimposition of corresponding entitles.
An example should make this clear. Represented in Fig. 11(a) (next page)
Is a simple symbolic structure, namely the noun balloon. The picture at the
semantic pole Is purely mnemonic — it abbreviates the full, multlfaceted
conceptual complex that constitutes our understanding of this notion. (As an
aside, I should note that cognitive grammar makes no claim whatever that
meaning reduces to visual Images, or that drawings done for expository pur-
poses are the formal objects of semantic description. These common miscon-
ceptions have no basis in anything I have ever said or written.) The notation
given at the phonological pole similarly abbreviates a complex phonological
structure. Note In particular that the ellipse labeled W represents a 'word', and
that the arrow labeled T stands for 'speech time'.
In a simple construction, two symbolic structures of this sort function as
component structures, and are Integrated to form a composite structure, as
shown in Fig. 1 1 (b). The dotted lines Indicate the correspondences that effect
this integration at each pole. That is, some facet of si is put in correspondence
with some facet of S2, where si and S2 are the semantic poles of the two com-
ponent structures. Likewise, some facet of pi Is put in correspondence with a
facet of p2, where pi and p2 are the component structures' phonological poles.
By the superimposition of corresponding entitles, si and S2 merge to form S3,
while pi and p2 merge to form P3. This Is composition — It yields a composite
structure in which S3 Is symbolized by P3.
Consider the integration of the adjective yellow and the noun balloon to
form the phrase yellow balloon Yellow and balloon are the two component
structures. Their Integration at the semantic pole Is diagrammed In Fig. 12(a),
and their phonological integration in 12(b). At the semantic pole, yellow pro-
files a simple atemporal relation, as previously described (Fig. 5(b) above),
while balloon designates a thing. Recall that the landmark for yellow is a
region In color space, and Its trajector a physical object that Is the locus of a
light sensation. Semantic integration is effected by a correspondence that iden-
tifies this trajector with the thing profiled by balloon. Superimposing these en-
titles yields the composite structure shown at the top, in which the locus of the
color sensation Is specified as being a balloon in particular. Observe that the
20
Studies in the Linguistic Sciences 20:2 (Fall 1990)
(a)
a
semantic
pole
symbolic
1 •inl'
(^lo^
phonological
pole
T
symbolic
structure
(b)
^
Composite Structure
\
1
-3
Compositio
i^*\.
s ...
1
p l...
s
Integration
... 5^
p
lA
2
Component Structures
Figure 11
composite structure designates the balloon — its relationship to color space is
included within the scope of predication but is unprofiled at the composite-
structure level. Hence yellow balloon, taken as a whole, is categorized as a
(complex) noun.
This semantic integration is symbolized by the phonological integration of
ye//ow and balloon, sketched in Fig. 12(b). Specifically, balloon is identified as
the word that directly follows yellow along the temporal axis. That is, the tem-
poral contiguity and ordering of yellow and balloon symbolizes their semantic
relationship, wherein the property of being the locus for a yellow sensation is
attributed to the balloon rather than to some other object.
Langacker: Cognitive Grammar
21
Figure 12
Yellow balloon, of course, instantiates a general pattern for the integration
of adjectives with nouns in English. In cognitive grammar, that pattern — or rule
— takes the form of a constructional schema, which is nothing more than a
schematization of such expressions. This particular constructional schema is
diagrammed in Fig. 13. It is a complex symtjolic structure whose internal organ-
ization is directly analogous to yellow balloon and other instantiating expres-
sions, the only difference being that specific characterizations of the adjective
and noun are replaced by schematic charactehzations: Semantically, they res-
pectively profile a simple atemporal relation and a thing, while phonologically
each is described as a word. However, their integration and profiling at the
composite-structure level is just the same as in the specific expression. Yellow
balloon thus participates in a categorizing relationship with the constructional
schema, which thereby provides its structural description. Moreover, the
schema is available for use as a template in assembling other expressions on
the same pattern.
(b)
QO
T ^
^^\
G5^^^
â– â– â– 'o.
T
T
-
Figure 13
Besides rules and grammatical classes, the symbolic alternative will have
to account for other supposed primitives and representations specific to gram-
22 Studies in the Linguistic Sciences 20:2 (Fall 1990)
mar: notions like "head", 'modifier', 'subject', and 'object', as well as syntactic
phrase trees. With respect to 'head' and 'modifier', let me call attention to a
couple of additional features of diagrams 12 and 13. It was observed that in this
construction the composite semantic structure profiles a thing rather than a stat-
ive relation — that is, the composite structure inherits its profile from the noun
rather than from the adjective (yellow balloon designates the balloon, not its
coloring). It is typical in a construction for the composite-structure profile to be
inherited from one of the components. And in fact, it is this component that is
traditionally regarded as the head. The notion head is so defined in cognitive
grammar; diagrammatically, it is indicated by the box drawn with heavy lines.
Also observe the cross-hatching and solid arrows in Figs. 12-13. As be-
fore, the solid arrows indicate an elaborative relationship. In a construction, it is
typical for one component structure to elaborate a subpart of the other (this
subpart is indicated diagrammatically by the cross-hatching). For instance, ye/-
/ow characterizes its trajector only schematically, and in the construction, bal-
loon characterizes the corresponding entity with considerably greater speci-
ficity. We can now define a modifier as a component structure one of whose
substructures is elaborated by the head. Yellow {hus modifies the head balloon
in yellow balloon.
The notion 'complement' (or 'argument') can also be defined in these
terms. A complement is component structure that elaborates one of the sub-
structures of the head. Examples of complements include subjects and direct
objects. Consider the verb enter, diagrammed in Fig. 10(c), and the sentence
Sally entered the room. The clausal head is enter, since the process it desig-
nates is profiled by the clause as a whole. Sally elaborates the schematic tra-
jector of this process, and the room elaborates its schematic landmark. I would
argue that subjects and direct objects are properly characterized as clause-
level complements, specifically as nominal expressions that respectively ela-
borate the trajector and primary landmark of the clausal head. Observe that this
characterization is based on semantic notions — profiling, correspondence, lev-
el of specificity — not on any particular constituency or syntactic tree structure.
This has important consequences for its general applicability (e.g. in VSO lan-
guages).
What about syntactic tree structures? The information they represent
seems crucial to linguistic structure, and as conceived in transformational gram-
mar, trees are purely grammatical objects, neither semantic nor phonological
(although they are used in semantic and phonological interpretation). The
kinds of information represented in phrase trees are indeed important. I main-
tain, however, that such trees — conceived as separate, purely syntactic objects
— are superfluous and artifactual.
Phrase trees incorporate three kinds of information: constituency, category
membership, and linear order. All of these are accommodated in the present
Langacker: Cognitive Grammar 23
approach by positing only symbolic units. Constituency is simply a matter of
smaller symbolic units being successively integrated to form progressively larg-
er symbolic structures. This happens when the composite structure at one level
of organization functions as a component structure in a higher-order construc-
tion. Moreover, a component or composite structure inherently represents a
particular grammatical category by virtue of instantiating the schema defining
that category. In this approach, category membership is not represented by
contentless node labels, but instead resides in categorizing relationships be-
tween schematic and specific symbolic structures. Lastly, linear order is in
reality temporal order, one dimension of phonological space. Temporal order-
ing is specified as part of the internal structure of every expression's phono-
logical pole, it is the arrow labeled T in Figs. 11-13. Observe that temporal
ordering is distinguished from constituency. The symbolic structures functioning
as nodes in a constituency hierarchy are not temporally ordered with respect to
one another — rather, temporal ordering is specified internally to each node as
part of its phonological characterization.
What about so-called grammatical morphemes, often regarded as seman-
tically empty markings used exclusively for syntactic purposes? I believe that ail
such markers can in fact be attributed conceptual import and revealingly ana-
lyzed as symbolic units. I have tried to show this by taking many of the toughest
examples and describing in fairly explicit detail just what I think they mean and
how that meaning accounts for their grammatical behavior. Among the "gram-
matical" elements that I have described in this way (in one publication or
another) are be, the auxiliary do, the perfect have ... -ing, the past participial
morpheme, the nominalizer -er, gender markers, the passive by, of, the pos-
sessive morpheme, case markers, etc. There are various reasons why their se-
mantic import has not been generally recognized: Because they are highly
schematic; because their value is primarily a matter of construal; because they
are polysemous; and because they are fully overlapped by the meanings of
other elements. From the standpoint of cognitive semantics these reasons are
all invalid.
Consider the morpheme -er, as in killer, swimmer, complainer, driver, etc.
As shown in Fig. 14(a), it invokes as its base a highly schematic process, hence
it has nothing in the way of specific conceptual content. Its import resides in con-
strual: the fact that it profiles the trajector of the schematic process serving as its
base. That schematic process is elaborated by a verb stem, such as kill, and
since -eris the head in this construction, it imposes its own profile on the spe-
cific process supplied by the stem; a killer \s thus characterized as the trajector
with respect to the process kill.
Similarly, the auxiliary do is analyzed as profiling a fully schematic pro-
cess. When it combines with another verb, as in They do like her, this
schematic process is put in correspondence with, and elaborated by, the
24
Studies in the Linguistic Sciences 20:2 (Fall 1990)
Figure 14
specific process profiled by the other verb, as shown in Fig. 14(b). Do adds
neither content nor profiling — semantically it is fully subsumed by the main
verb. But that does not entail that it is meaningless: Meaningfulness is not the
same as non-overlapping meaning. There is semantic overlap of some sort in
every construction. The overlap between ye//ow and balloon in yellow balloon
was indicated by the correspondence line in Fig. 12(a); although each compon-
ent contributes conceptual content not evoked by the other, the former's
schematic trajector is equated the latter's profile. In Fig. 14(a), the conceptual
content of -er is completely subsumed by that of the verb stem, yet -er has a
discernible semantic effect owing to the distinct profile it imposes. The overlap
is even more extensive in 14(b) because the two profiles correspond. The
differences among such examples reside only in the extent (not the existence)
of their semantic overlap, and consequently in how "visible" the meaning of
yellow, -er, or do is to the analyst. Complete overlap, as with do, is merely an
expected limiting case.
Examples like pants, binoculars, tongs, pliers, scissors, glasses, shorts,
trousers, tweezers, etc. are often cited to show that the semantic and gram-
matical notions of plurality have to be distinguished: Such forms are gram-
matically plural but supposedly semantically singular. The argument is fall-
acious, for it ignores the possibility that the plural morpheme might be
polysemous. In its prototypical sense, the plural morpheme designates a set of
distinct entities all of which instantiate the same class and could be labeled
individually by the singular noun stem. That is not the case with pants,
binoculars, scissors, etc.; but clearly, the occurrence of the plural ending in
precisely these forms is not an accident — these nouns designate unitary ob-
jects that are nevertheless characterized by salient internal duality. I interpret
Langacker: Cognitive Grammar 25
such duality (and more generally, multiplicity) as constituting a secondary
meaning of the plural morpheme, a natural extension from the prototype.
A standard reason for subscribing to the autonomy thesis is that ex-
pressions often have to take a certain form, even though another form could
perfectly well convey the same meaning. As an example of such arbitrary
formal requirements, consider 'government', for instance the fact that certain
prepositions in German (among them gegen 'against', bis 'until', durch
'through', fur 'ior', urn 'around*, and ohne 'without') require that their object be
marked for accusative case, while others (including aus 'out of, von 'from', seit
'since', bei 'by', mit 'with', nach 'toward', and zu 'at') govern dative case. Now,
first of all I would argue (as a student of mine, Mike Smith, has done in great
detail (1987)) that these case inflections are actually meaningful. They appear
not to be, because the meanings are schematic (e.g., 'goal-directed path' is the
accusative prototype), each category is polysemous, and the meanings of the
case elements are subsumed by those of the governing prepositions. But let us
focus here on the fact that the case markings have to occur even though the
expressions would be semantically viable without them. Is this not a matter of a
certain form being required arbitrarily by grammatical convention?
Though I might quibble about how arbitrary it is, grammatical convention
certainly does impose a formal requirement that simply has to be stated,
learned, and adhered to. However, this does not establish the autonomy thesis,
as I have defined it, because it is perfectly possible to describe the situation in a
framework that posits only symbolic units. For example, the fact that gegen
occurs with accusative case would be specified by means of the constructional
schema that we can abbreviate here as [ [gegen [ACC + NML] ]. Abstracted
from instantiating expressions (e.g. gegen einen (ACC) Baum 'against a tree'),
this schema details the integration of the preposition gegen with a nominal (i.e.
noun phrase) bearing accusative case. Another constructional schema, abbre-
viated [ [aus [DAT + NML] ], describes a pattern wherein aus takes an object
marked with dative case (e.g. aus dem (DAT) Haus 'out of the house'). Granted
that the case markers themselves are symbolic structures, the patterns in ques-
tion are characterized by means of symbolic units alone. The patterns are list-
ed, not strictly predicted, but only symbolic structures figure in the listing.
What about the fact that these patterns are obligatory? That gegen, for
instance, governs accusative and does not tolerate a dative or caseless object?
All this implies is that no constructional schema other than the one that specifies
accusative case is available to sanction the integration of gegen with a nominal
object. No constructional schema allowing gegen with, say, a dative object is
extracted by the language learner because no expressions of that sort occur to
provide the basis for schematization. If such an expression were to be used, it
would thus be categorized as an intended instance of [ [gegen [ACC + NML] ],
whose specifications it violates.
26
Studies in the Linguistic Sciences 20:2 (FaU 1990)
This example also illustrates the approach taken to arbitrary distributional
classes, i.e., the fact that the elements occurring in a particular morphological or
syntactic construction are often less than fully predictable, if at all. To indicate
that an element does occur in a given construction, one does not tag it with a
diacritic or syntactic feature — that would violate the content requirement. In-
stead, the information is provided by a constructional schema which specifically
mentions that element, such as [ [gegen [ACC + Nfy/IL] ].
Like construction grammar, cognitive grammar treats general constructions
— for instance, the prepositional-object construction — as complex categories.
Such a construction takes the form of a network, where each node is itself a
constructional schema, as illustrated in Fig. 15. This network subsumes specific
expressions learned as fixed units; constructional subschemas that mention
particular lexical items, like those at the bottom level in the diagram; and more
abstract schemas representing higher-level generalizations. The nodes in such
a network differ both in specificity and in cognitive salience or entrenchment. I
assume a processing model in which the nodes in a network compete with one
another for the privilege of categorizing a novel expression. Other things being
equal, a lower-level structure wins out over a more abstract structure in this
competition, for it overlaps with the target expression in many points of specific
detail, each of which tends to activate it. As a consequence, a German prepo-
sitional phrase in which a dative follows gegen will judged a deviant instance of
the gfegen+accusative construction, not as a well-formed instance of the higher-
level schema which merely specifies the possibility of a preposition taking a
dative-marked object. This is admittedly quite sketchy; but it may at least in-
dicate that arbitrary distributional restrictions are not per se incompatible with
the symbolic alternative. Certainly they do not themselves establish the auto-
nomy of grammatical structure as a separate level or domain of structure. (Re-
call the type/predictability fallacy.)
Two more things have to be accounted for: our apparent ability to judge
grammaticality independently of meaning, and restrictions that evidently have to
be stated in purely formal terms. I will deal with them only very briefly.
[P [Case + NML] ]
[gegen [ACC [bis [ACC
+ NML] ] + NML]
von [ACC [mit[ACC
+ NML] ] + NML] ]
Figure 15
Langacker: Cognitive Grammar 27
The first point is exemplified by that well-known novel sentence Colorless
green ideas sleep furiously, which is supposedly grammatical though seman-
tically anomalous. Such examples pose no special problem in cognitive gram-
mar, which does recognize the existence of grammatical patterns and restric-
tions, but simply claims that they are fully describable by means of con-
structional schemas employing only symbolic units. Examples like Colorless
green ideas sleep furiously involve the proper use of constructional schemas,
with each schematic element instantiated by a lexical item belonging to the
appropriate class, but where certain specifications of these lexical items happen
to be mutually incompatible. Consider green idea. It represents one possible
instantiation of the constructional schema sketched in Fig. 13. This schema
however specifies that the trajector of the adjective corresponds to the profile of
the noun, with corresponding entities being superimposed to form the com-
posite structure. Now the adjective green characterizes its trajector as a phys-
ical entity of some sort, while idea profiles an abstract entity. Thus, when green
and idea are integrated in the manner dictated by the constructional schema
employed, entities with incompatible specifications are superimposed, and the
result is perceived as semantic anomaly. Still, the expression does instantiate a
grammatical pattern, characterized in terms of symbolic units alone.
Finally, what about restrictions that have to be stated in purely formal
terms? An example might be the coordinate structure constraint, exemplified in
(12).
(12) a. She likes the blouse but hates the skirt,
b. *What does she like but hates the skirt?
I will make only two brief observations. First, it is doubtful that such restrictions
can in fact be stated just in formal terms. On the basis of well-formed sentences
like (13), for instance, Lakoff (1986) has argued that so-called extraction is sen-
sitive to semantic factors that tend to correlate with certain structural con-
figurations but are in fact independent of them.
(13) a. What did she go to the store and buy?
' b. How much can you drink and still stay sober?
More generally, we have seen that cognitive grammar does recognize and ac-
commodate the various kinds of relationships depicted in standard syntactic
phrase trees — it simply interprets and handles these relationships in a different
manner, as distinct aspects of symbolic configurations. In principle, therefore,
any patterns and restrictions that do make reference to tree configurations are
susceptible to reformulation in symbolic terms.
Let me conclude by trying to put things in perspective. I believe it is true
that over the last three decades literally tens of thousands of human work years
28 Studies in the Linguistic Sciences 20:2 (FaU 1990)
have been devoted to investigating language from the standpoint of generative
theories that presuppose the autonomy of grammatical structure. The terms of
this inquiry are by now so familiar that they are easily taken for granted and ac-
cepted as self-evident. By contrast, at most a few tens of human work years
have been devoted to working things out from the standpoint of cognitive
grammar. One must be careful not to confuse the unfamiliarity of its concepts,
notations, and general outlook with the question of its potential viability and in-
sight. Given that language effects the phonological symbolization of conceptual
structure, cognitive grammar's view of linguistic organization is the most
straightforward, unified, natural, and intrinsically desirable one imaginable.
Despite its preliminary character, linguistic theorists ought to be vitally con-
cerned with trying to make it work if at all possible. In my own eyes, the matter is
clear: Language makes complete sense when viewed in this way, whereas
seen through the lenses of more traditional approaches, much of it is opaque
and mysterious.
REFERENCES
FILLMORE, Charles J. 1988. The mechanisms of "Construction Grammar".
BLS 14.35-55.
LAKOFF, George. 1986. Frame semantic control of the coordinate structure
constraint. Papers from the Parasession on Pragmatics and Grammatical
Theory, ed. by Anne M. Fariey et al. 152-167. Chicago: Chicago Lin-
guistic Society.
LANGACKER, Ronald W. 1982. 'Space Grammar, Analysability, and the
English Passive'. Language 58.22-80.
1987. 'Nouns and Verbs'. Language 63.53-94.
NEWMEYER, Frederick J. 1983. Grammatical theory: Its limits and its possib-
ilities. Chicago and London: University of Chicago Press.
SMITH, Michael B. 1987. The semantics of dative and accusative in German:
An investigation in Cognitive Grammar. San Diego: UCSD Ph.D.
dissertation in Linguistics.
APPENDIX: Selected references In Cognitive Grammar
CASAD, Eugene H., & Ronald W. Langacker. 1985. "Inside" and "outside" in
Cora Grammar. UAL 51 .247-281 .
COOK, Kenneth W. 1988. A cognitive analysis of grammatical relations, case,
and transitivity in Samoan. San Diego: UCSD Ph.D. dissertation in
Linguistics.
Langacker: Cognitive Grammar 29
FAUCONNIER, Gilles. 1985. Mental spaces: Aspects of meaning construction
in natural language. Cambridge, Mass. & London: MIT Press/Bradford.
FILLMORE, Charles J. 1982. Frame semantics. Linguistics in the Morning
Calm, ed. by the Linguistic Society of Korea, 11 1-137. Seoul: Hanshin.
. 1988. The mechanisms of "construction grammar". BLS 14.35-55.
HAIMAN, John. 1980. Dictionaries and encyclopedias. Lingua 50.329-357.
HAWKINS, Bruce W. 1984. The semantics of English spatial prepositions. San
Diego: UCSD Ph.D. dissertation in Linguistics.
JANDA, Laura A. 1984. A semantic snalysis of the Russian verbal prefixes za-,
pere-, do-, and ot-. Los Angeles: UCLA Ph.D. dissertation in Linguistics.
To Appear. A geography of case semantics: The Czech dative and the
Russian instrumental. Berlin: Mouton de Gruyter.
JOHNSON, Mark. 1987. The body in the mind: The bodily basis of meaning,
imagination, and reason. Chicago & London: University of Chicago
Press.
LAKOFF, George. 1986. Frame semantic control of the coordinate structure
constraint. Papers from the Parasession on Pragmatics and Grammatical
Theory, ed. by A. M. Farley et al., 152-167. Chicago: Chicago Linguistic
Society.
1987. Women, fire, and dangerous things: What categories reveal about
the mind. Chicago & London: University of Chicago Press.
, & Peter Norvig. 1987. Taking: A Study in Lexical Network Theory". BLS
13.195-206.
LANGACKER, Ronald W. 1982. Space grammar, analysability, and the English
passive. Language 58.22-80.
1984. Active zones. BLS 10.172-188.
1985. Observations and speculations on subjectivity. Iconicity in syntax,
ed. by John Haiman, 109-150. Amsterdam & Philadelphia: Benjamins.
1986a. Abstract motion. BLS 12.455-471.
1986b. An introduction to Cognitive Grammar. Cognitive Science 10.1-
40.
1986c. Settings, participants, and grammatical relations. Proceedings of
the Annual Meeting of the Pacific Linguistics Conference 2.1-31.
1987 (a). Foundations of Cognitive Grammar, vol. 1 : Theoretical prerequi-
sites. Stanford: Stanford University Press.
1987(b). Nouns and verbs. Language 63.53-94.
1987 (c). Grammatical ramifications of the setting/participant distinction'.
BLS 13.383-394.
1988. Autonomy, sgreement, and Cognitive Grammar. Agreement in
grammatical theory, ed. by D Brentan et al., 147-180. Chicago: Chicago
Linguistic Society.
1990. Subjectification. Cognitive Linguistics 1 .5-38.
In Press. Concept, image, and symbol: The cognitive basis of grammar.
Berlin: Mouton de Gruyter.
To Appear. Foundations of Cognitive Grammar, vol. 2: Descriptive ap-
plication. Stanford: Stanford University Press.
30 Studies in the Linguistic Sciences 20:2 (Fall 1990)
LINDNER, Susan. 1981. A lexico-semantic analysis of English verb-particle
constructions with up and out. San Diego: UCSD Ph.D. dissertation in
Linguistics.
1982. What goes up doesn't necessarily come down: The ins and outs of
opposites. CLS 18.305-323.
MALDONADO, Ricardo. 1988. Energetic Reflexives in Spanish. BLS 14.153-
165.
POTEET, Stephen. 1987. Paths through different domains: A Cognitive Gram-
mar analysis of f^andarin dad. BLS 13.408-421.
RICE, Sally. 1987a. Towards a transitive prototype: Evidence from some a-
typical English passives. BLS 13.422-434.
1987b. Towards a cognitive model of transitivity. San Diego: UCSD Ph.
D. dissertation in Linguistics.
1988. Unlikely lexical entries. BLS 14.202-212.
RUDZKA-OSTYN, Brygida (ed.) 1988. Topics in cognitive linguistics. Amster-
dam & Philadelphia: Benjamins.
SMITH, Michael B. 1987. The semantics of dative and accusative in German:
An investigation in Cognitive Grammar. San Diego: UCSD Ph.D.
dissertation in Linguistics.
SWEETSER, Eve E. 1984. Semantic structure and semantic change: A cog-
nitive linguistic study of modality, perception, speech acts, and logical
relations. Berkeley: University of California Ph.D. dissertation in Linguis-
tics.
1988. Grammaticalization and semantic bleaching. BLS 14.389-405.
TALMY, Leonard. 1978. Figure and ground in complex sentences. Universals
of Human Language, ed. by J. H. Greenberg, 4.625-649. Stanford:
Stanford University Press.
— . 1983. How language structures space. Spatial orientation: Theory, re-
search, and application, ed. by H. Pick & L. Acredolo, 225-282. New
York: Plenum Press.
1988. Force dynamics in language and cognition. Cognitive Science 12.
49-100.
TUGGY, David. 1981. The transitivity-related morphology of Tetelcingo NahuatI:
An exploration in space grammar. San Diego: UCSD Ph.D. dissertation
in Linguistics.
. 1986. Noun incorporations in NahuatI. Proceedings of the Annual Meet-
ing of the Pacific Linguistics Conference 2.455-469.
VANDELOISE, Claude. 1985. Au-dela des descriptions geometriques et logi-
ques de I'espace: Une description fonctionnelle. Lingvisticae Investiga-
tiones 9.109-129.
1986. Lespace en frangais. Paris: Editions du Seuil.
WIERZBICKA, Anna. 1988. The semantics of grammar. (Studies in Language
Companion Series, 18.) Amsterdam & Philadelphia: Benjamins.
Studies in the Linguistic Sciences
Volume 20, Number 2, Fall 1990
A TRIMODULAR ACCOUNT OF YIDDISH SYNTAX*
Jerrold M. Sadock
(University of Chicago)
During the first semester of the official existence of the Department of
Linguistics at the University of Illinois, I wrote a term paper on Yiddish for Lin-
guistics 401 , the introductory course in syntax. I find no grade on the typescript
that remains in my possession, and do not now recall what I got. My own
assessment of the paper, even allowing for the limitations of the theory of syn-
tax as it then existed, and allowing for the fact that it was my first course in syn-
tax, would be that it deserved no better than a B. One innovative change in the
treatment that I presented would have greatly improved my opinion of my fledg-
ling attempt at syntactic deschption. This change, while not suggested by the
descriptive traditions current in 1965, was also not forbidden by them, as we
shall see.
What I will do here is briefly summarize the ideas I had twenty-five years
ago, compare them to a state-of-the-art treatment of some of the same facts,
and then go on to present what I find to be a much more satisfying description
that is made available by the non-transformational view of grammar that I have
been developing for more than five years now.
Sadock 1965
To handle the basic fact that Yiddish is a rather hgidly verb-second langu-
age, as well as the fact that almost any constituent may be the first in the sen-
tence, I postulated deep structures along the lines of (1), and three movement
rules: rule (2) that fronted any constituent, substituting it for a dummy element
dominated by PRS; rule (3) that inverted the subject and finite element FIN; and
rule (4) that combined the main verb of the verb phrase with the element FIN,
the bearer of person and number information (NU).
The presentential element PRS in the deep structure of (1) was the posi-
tion under which sentence adverbs were generated. If an adverb such as ava-
de "certainly" was present, it would trigger inversion. If one of two phonetically
32
Studies in the Linguistic Sciences 20:2 (FaU 1990)
(1) Deep Structure:
S
(PRS) NP
PRED
FIN
I
NU
VP
(2) Topicalization:
PRS - X - C - Y
I
DUM
(3) Inversion:
PRS - NP - FIN - X
(4) Verb Raising:
PRED
PRS - X - Y
I
C
PRS-FIN-NP-X
PRED
FIN
NU + V
VP
I
X
empty elements was generated under PRS, viz. an interrogative marker or a
consecutive marker meaning, roughly, 'so', inversion would also take place,
but since these were assumed to lack phonetic content, the sentences contain-
ing them would appear to be verb-initial. This accounted for the fact that a
string like Geyt er aheym can mean either 'Does he go home?' or, roughly, 'So
he goes home', depending on intonation. If the presentential element domina-
ted a dummy element, the subject or some other constituent had to take its
place, in which case it would be interpreted as topic. ^ If there was no PRS ele-
ment, the subject would remain in initial position and would not be understood
as a topic.
What I failed to see was that one of these rules, namely inversion, could be
eliminated entirely if, non-standardly for the time, the finite element were gener-
ated above the subject, rather than below it, as in (5).
Sadock: A trimodular account of Yiddish syntax
33
(5)
PRS
Provided that the presentential element were obligatorily present and con-
tained either an overt sentence adverb or the dummy element, then verb-sec-
ond order could have been accounted for with two rules instead of three. The
verb would always move to FIN which is preceded either by a base-generated
adverbial, or an element (possibly the subject) that has moved to PRS to re-
place the dummy element.
Diesing 1990
Twenty-five years later, there are elements of this treatment that look rather
old-fashioned, but in other respects the treatment seems fairly up to date. Die-
sing (1990) has recently published a state-of-the-art account of the basic facts
of Yiddish syntax that posits base structures and derivations for main clauses
such as that schematized in (6) above.
(6)
34 Studies in the Linguistic Sciences 20:2 (Fall 1990)
Except for the modish labeling of nodes, the topology of the modern theory
is identical to what I should have given in 1965. There are. of course, other dif-
ferences, in particular the fact that the new theory is supposed to be grounded
in principles of universal grammar, and thus is supposed to explain, rather than
simply observe, the panoply of facts it covers. I cannot help but observe, how-
ever, that the modern theory is a lot looser than its proponents make it out to
be.2
Further facts
Let me now turn to some facts of Yiddish syntax that neither the portion of
my antique view sketched above nor Diesing's up-to-the-minute treatment deal
with, namely the position of pronominal arguments. These occur in a fixed or-
der immediately following the finite verb, an order which is different from the
order of full NP arguments.
(7) Ikh hob is im gegebn
1 have it him given
'I have given it to him.
(8) Ikh hob gegebn dem zeydn a matone
I have given the grandfather a present
'I have given a present to Grandfather.'
The ordering of pronominal elements turns out to be as follows: An invert-
ed subject, if present, is first, a reflexive pronoun next,3 followed by an accusa-
tive pronoun, and finally a dative pronoun. In my 1965 paper I dealt with the
accusative and dative pronouns by postulating two obligatory, ordered rules
that moved object pronouns from their original positions to a position immedi-
ately following the finite verb. The subject pronoun, and indeed any inverted
subject, would be positioned after the finite verb by the inversion rule. The or-
dering of the pronouns was assured by moving first the dative pronoun and
then the accusative pronoun, so that if both were present, the accusative would
stand closer to the verb, i.e., in the opposite order from what we find when both
are full noun phrases. Inversion would have to be ordered after both of these
rules so as to insure that an inverted subject stood immediately after the finite
verb and before any other pronouns. The reflexive was not dealt with at all.
Diesing (1990) only observes that the ordering of these pronouns is unex-
pected, referring the reader to a paper by den Besten and Moed-van Walraven
(1985). Ail that is to be found by way of explicit treatment of the phenomenon
in that paper, however, is the statement that 'the obligatory occurrence of weak
[sic] pronouns in preverbal position, however, should be treated as a case of A'
movement.' There is no indication of exactly what positions these pronouns
Sadock: A trimodular account of Yiddish syntax 35
move to, what accounts for their strict ordering with respect to one another, or
why the movement is obligatory.
Sadock 1990
I will now reconsider the facts of Yiddish word ordering in light of the sort of
theory originally presented in Sadock 1985 and developed more fully in Sa-
dock 1991. The hallmark of this view is the idea that different representations
of a linguistic expression are not to be related derivationally. Rather, they
occur as parallel structures governed by quite autonomous rules, but con-
strained with respect to each other by interface principles of various kinds.
The first of the levels that I will assume in my treatment of Yiddish is the
syntax per se, the level at which ordinary syntactic constituents are defined,
and at which syntactic relationships such as subject and complement are to be
located. Here Yiddish displays a quite unremarkable rule set. Sentences con-
sist of nominative (NOM) subject noun phrase and predicate verb phrase, in
that order, the person and number (PN) features of the subject and predicate
agreeing. Predicates, in turn, consist of a verb and its possible complements:
a dative NP (DAT), an accusative NP (ACC), and either another VP, or a sub-
ordinate clause, in that order. The verb subcategorizes for some combination
of these complements and for the feature of the complement clause that con-
trols what complementizer it contains. A monostratal account of this much of
English syntax would be identical, except that English does not distinguish
morphosyntactic dative and accusative cases.
(9) S -^ NP [NOM, PN] — VP [PN]
(10) VP -^ V — (NP[DAT]) — (NP[ACC]) — (VP/S')
(11) S' -* Comp S
The second dimension is where discourse notions like topic and focus are
represented. It is not syntax, but a separate and autonomous kind of represent-
ation with a different informational function, and therefore different categories.'*
There is only one rule here that we need to be concerned with for the time be-
ing, a rule that states that an utterance consists of a comment optionally pre-
ceded by a topic.
(12) U -^ (TOP) COM
Lumping all sentence-initial non-subjects together under the heading of
discourse topic is quite inaccurate, as noted by Ellen Prince (1981), but utter-
ance-initial non-subjects always have some special pragmatic function, and
this fact ought to be represented in a system separate from the syntax. The
coarse description contained in (12) will do for the purpose of describing the
36
Studies in the Linguistic Sciences 20:2 (Fall 1990)
word order of Yiddish, though the treatment of the discourse consequences of
word order clearly needs refinement.
Though (12) describes an independent discourse-functional level of or-
ganization, there is, of course, a connection between it and the ordinary syntax
defined by (9) - (11), namely the fact that the elements of (12) are also repre-
sented in the syntax of Yiddish. In particular, an utterance (U) has to be a
clause (S), a topic (TOP) has to be a phrase (X"), and the comment (COM) has
to be the rest of the sentence (i.e. an S/X" in the notation of Gazdar 1981).
Thus (12) could have been written as (12'), but that would give the impression
that it is a rule of syntax, which it is not.
(12-)
Xn _ S/Xn
I will therefore state the relation between the syntactic and discourse-func-
tional structures as the two interface rules (13) that make explicit the required
connection between the autonomous dimensions of analysis. This has the
added advantage of making a separate statement concerning the syntactic
nature of COM otiose, since it follows logically that it will have to be a clause
minus an X", that is, an S/X".
(13)
D-F Structure
Svntax
a.
U
s
b.
TOP
X"
Now, according to (12), any constituent may be the first element of a topic-
comment utterance (i.e. of a sentence, according to (13a)). This is a well-
known feature of Yiddish and numerous other languages alike. Various sen-
tence-initial constituents are illustrated by the following examples from the writ-
ings of Sholem Aleichem.^
(14) Dos shtiki shtol hob ikh rekht ongesharft ...
that piece steel have I properly sharpened
That piece of steel I have properly sharpened ...' (DM, 10.)
(15) In kheder hob ikh dos nit getort haltn.
in school have I that not dared hold
"In school I didn't dare to hold it.' (DM, 1 1 .)
(16) Plutsim khabt zikh oyf dertate vi funm shiof.
suddenly grabs self up the father as from sleep
'Suddenly Father jumped up as if from sleep.' (DM, 1 1 .)
(17) Morgn leg ikh anider tsurik dos meserl.
tomorrow lay I down back the knife
'Tomorrow I will put the knife back down.' (DM, 22.)
Sadcxrk: A trimodular account of Yiddish syntax 37
(18) Shraybn shraybt er zikh 'Albert',
write writes he self 'A'
'In writing, he calls himself "Albert".' (TsA, 157.)
When the subject is initial in the sentence, it may be a topic, as in (19), a
sentence parallel to (17) in being part of a discussion of the pocket knife, or it
may not be, as in (20), which has almost the same import as (17), but is not of
topic-comment form.
(19) Dos meserl zol zikh lign in keshene.
the knife should self lie in pocket
'The knife ought to lie in (my) pocket.' (DM, 9.)
(20) Ikh vel es nokhdem tsurik anider-legn.
I will it later back down-lay
'I'll put it back down later.' (DM, 16.)
Note that if there is a conflict in the ordering required by the syntax and
that required by the discourse-functional component, the latter wins out. It
couldn't be othenwise, since according to (12) the only expression of the dis-
course-functional organization of the utterance in Yiddish is in terms of word or-
der. In other words, if syntactic order prevailed, we would have no reason to
postulate rule (12) in the first place.
I turn now to the order of definite pronouns in Yiddish, which I propose to
handle by means of a template that requires them to occur in a certain fixed or-
der. The ordering template for the definite pronouns, the demonstrative dos
'that', as well as the indefinite nominative pronoun me(n) 'one' is;
(21 ) nominative — reflexive — accusative — dative
The following examples illustrate the ordering with various pairs and trip-
les of pronouns.
(22) Ikh hob dos mir take aleyn gemakht.
I have that (ACC) me(DAT) actually alone made
'I actually made it for myself by myself.' (DM, 9.)
(23) Ven ikh vil, zol ikh dos mir aroysnemen.
when I want shall I(NOM) it(ACC) me(DAT) out-take
'When I want to, I'll take me out that knife.' (DM, 9.)
(24) Es hobn zikh mir gekholemt ayzerne riter.
it have self me(DAT) dreamed iron rods
'Iron rods appeared to me in a dream.' (DM, 22.)
38 Studies in the Linguistic Sciences 20:2 (Fall 1990)
(25) Epes beyzert er zikh.
Something angers he(NOM) self
'He is angry over something.' (DM, 22.)
As mentioned, this cluster of pronouns immediately follows the finite verb.
The template, then, must also mention this fact:
(26) Vfin — ProNOM — ProREF — ProACC — ProoAT
Such a template strongly suggests the clitic-ordering templates of Ro-
mance languages that Perlmutter (1971) argued have to be handled by surface
constraints, or filters. Not only are the clitic elements of, say, French attracted to
a position next to the finite verb, indeed, they occur in exactly the same order
as they do in Yiddish. Other factors, such as the category of person, figure in
distributing French clitics, and the cluster of pronominal elements occurs to the
left, rather than the right, of the finite verb; but is is a remarkable fact that the
nominative, reflexive, accusative, and dative clitics of French come in that or-
der, just as they do in Yiddish. This is not to suggest, of course, that it is not
necessary to make a parochial statement of the order of small elements in Yid-
dish.
What Perlmutter demonstrated in an eloquent way was that any attempt to
constrain the individual transformations that moved elements around was
doomed to include ad-hoc restrictions and to repeat information in various
rules, just so as to conspire to produce the surface facts. Instead, he argued,
the transformations should be allowed to operate freely, but only those results
that conform to the template of clitic elements would be allowed to pass, the
others being filtered out.
In the present framework, which lacks transformations entirely, the tem-
plate simply defines an autonomous dimension of representation, separate
from, and sometimes in conflict with, the ordinary syntax. The basic principle is
that a proper expression must be well-formed with respect to all the compon-
ents, though in cases of conflict, certain relaxations of this requirement must be
allowed in order to get any descriptive mileage out of the separation of com-
ponents. Each representational level in this view acts as a filter on each of the
others.
For French and the other Romance, where the clitics are unstressed, can-
not be conjoined, and are phonologically quite distinct from independent pro-
nouns, this autonomous dimension can be identified with something like mor-
phology. ^ The Romance clitics, in other words, have some of the quality of in-
flectional affixes. But in Yiddish, things are quite different, for here the pro-
nouns ar^ identical to the independent pronouns, bear their own word stress
(and can even be contrastively stressed), can be conjoined, and so on.
Sadock: A trimcxiular account of Yiddish syntax 39
This, then, is a kind of alternative, simplified syntax, in which linear preced-
ence is the only structural relation. There is no pre-existing name for this level.
I will dub it 'surfotax'.
An alternative to assuming a template to handle the ordering of pronouns
in Yiddish would be a set of ordered mles that move the pronouns to their final
positions, as in my 1965 term paper. The dative pronoun would be moved first
to the position after the finite verb, and the accusative pronoun would be
moved second, thus insuring that it precedes the dative if both are present, and
so on. I will not go into the reasons why such a treatment is no longer accept-
able, but i should point out that in any case, it is untenable in a theory without
transformations such as the one I am investigating here.
Another idea, which suggests itself on the basis of much recent work in the
Barriers framework (Diesing's is an example), would be to build in landing sites
for the pronouns in the deep structures of Yiddish sentences, say. Spec of
ACC, Spec of DAT, and so on. Having done that, however, it is clear that the
stack of nodes from top to bottom is simply an ad-hoc means of ordering
elements from left to right, which is exactly what the template does in a more in-
genuous fashion.
This brings me to the last big step in my analysis, namely the addition to
the template in (26) of a single additional position, one in front of the finite verb,
reserved for constituents of unspecified type. The template now is (26') and ac-
counts for the position of the verb in Yiddish, as well as for the position of pro-
nouns.
(26') !Xn — IVfin — ProNOM — Ptoref — ProACC — ProDAT
This move calls to mind the proposal of Maling and Zaenen (1981) to
handle the verb-second property of Icelandic by means of a 'positive surface fil-
ter' along the lines of the clitic template of Perimutter's.'' Their arguments for
doing things this way rather than by constraining individual transformations are
quite similar to Perimutter's, showing, in the end, that powerful conditions on
transformations that would conspiratorially act to produce the desired results
could be eliminated by adopting the filter. Their one great worry was over the
additional power that the adoption of a system of surface filters would add to a
transformational grammar. This worry disappears entirely in the present con-
text, since the surfotax supplants the transformations altogether.
A few remarks on optionality and obligatoriness as they pertain to the tem-
plate are in order. The two initial positions are obligatorily present in every
clause, but the pronouns are present in their stipulated positions only if there
happen to be pronouns in the sentence. Rather than, say, parenthesize the
pronouns, which might suggest that they are only optionally governed by the
40 Studies in the Linguistic Sciences 20:2 (Fall 1990)
template, I have marked the obligatory elements of the clause template in (26')
with exclamation points. Secondly, pronominal arguments may be present
without occurring where they are indicated In the template, provided that they
are the first element, X". The template should therefore be interpreted as
meaning that any pronoun in a clause is obligatorily positioned in the template,
either In its specified slot, or in the catch-all Initial position.
Since the topic phrase in the discourse-functional sthng is optional,
clause-initial subjects will be ambiguous as to their discourse status, being first
either because they are subjects, positioned initially by the syntax, or because
they are topics, put first by the overriding demands of the discourse-functional
component. Other clause-initial elements will have to be topics. Note that
these remarks hold for both pronominal and non-pronominal constituents. If
we make the common assumption that the topic is stressed, then the only un-
stressed, sentence-initial pronouns will be subjects, a fact that has been ob-
served several times In the literature. (See Diesing 1990, §2, and the refer-
ences mentioned there.)
Some illustrative examples
Let us now see how this simple system, consisting of only five rules in
three autonomous components, accounts not only for the data that have been
introduced so far, but also for certain other facts that have not been dealt with In
the literature.
In a simple sentence with no topic, the subject will stand first in the sen-
tence, because the syntactic rule (9) puts the subject first. In initial position the
subject will also fulfill the requirement of the template that there be some single
element to the left of the finite verb: Di zhabes qvaken The frogs are croaking'
(DM, 9), Di zun zetst zikh 'The sun sets (itself)' (DM, 9), etc. If there is a prag-
matically emphasized item (a topic, in the inaccurate terminology that I employ
in common with many others), it must occur In utterance-initial position accord-
ing to the demands of the discourse-functional rule (12) which necessarily
overrides the syntactic demand for subject-initial clauses. Now the subject
cannot occur immediately before the verb phrase, because there would then
be two constituents to the left of the finite verb, in violation of the surfotactic tem-
plate. The subject will have to come later. Because the syntax places the sub-
ject first, it will come as close to the front of the sentence as it can without viol-
ating the template, the nearest possible spot being immediately after the finite
verb:
(27) Haynt est der zayde a joikh.
today eats the grandfather a soup
'Today Grandfather is eating soup.'
Sadock: A trimodular account of Yiddish syntax 4 1
Note that the template mentions the nominative pronoun as being the first
element after the finite verb. Ordinary non-pronominal subject phrases are not
mentioned in the surfotactic template, and indeed, such phrases regularly ap-
pear later in the sentence than one would expect from typical G-B treatments,
or from the primitive system of Sadock 1965. In both the early and latter-day
transformational theories of Yiddish, we should expect non-initial subjects to
come immediately after the finite verb. But when one or more of the objects is
pronominal and therefore controlled by the surfotactic template, by far the most
common order is with the subject following the objects:
(28) Hoybt mikh mayne^ on tsutraybn.
starts me(ACC) mine(FEM) PART to nag
'So my (wife) starts to nag me.' (DD, 144)
(29) ?? Hoybt mayne mikh on tsu traybn.
(30) Az..., hot uns di gantse velt mekane geven.
when has us(ACC) the whole world jealous been
'When ..., the whole world was jealous of us.' (DD. 133)
(31 ) ?? Az . . ., hot di gantse velt uns mekane geven.
Notice that in these examples the subject still goes as close to the begin-
ning of the sentence as it can without violating the template. The explanation
for this, as mentioned above, is that the subject is initial in the syntax, so if it
cannot be initial in the surface, it must be as close to initial as it can be. The
identical state of affairs obtains at the syntax-morphology boundary, as docu-
mented extensively in Sadock 1991. There it is shown that syntactic ordering
requirements are preserved to the extent that they do not violate morphological
ordering requirements.
Now, if the subject is a pronoun, its position is dictated by the template,
and it therefore obligatorily occurs before any other pronominal arguments.
This is illustrated by examples (15) and (23), repeated here as (32) and (34).
(32) In kheder hob ikh dos nit getort haltn.
in school have I that not dare hold
•In school I didn't dare to hold it.' (DM, 1 1 .)
(33) * In kheder hob dos ikh nit getort haltn.
(34) Ven ikh vil, zol ikh dos mir aroysnemen.
when I want shall I(NOM) it(ACC) me(DAT) out-take
"When I want to. I'll take me out that knife.' (DM, 9.)
(35) * Ven ikh vil, zol dos mir ikh aroysnemen.
42
Studies in the Linguistic Sciences 20:2 (Fall 1990)
For my account to go through, it must be the case that when the template
and the ordinary syntax are at odds, the template wins out. Once again, this is
the only logical assumption, since the template, like the discourse-functional
rule, has only ordehng as a means of expressing itself, and thus, if it yielded to
the word-order demands of the syntax, it would not exist.
As several of the examples above also show, the template has the effect of
interposing elements between the finite and non-finite parts of the verb. This is
automatically accounted for by the simple trimodular theory, as we can see by
considering example (32) more carefully. The syntax puts this sentence to-
gether as in (36), and in this order it would also be quite grammatical.
(36)
In (32), however, the prepositional phrase is not in its syntactic position,
but is initial in the sentence because it is a topic. Since the finite verb hob must
come next, according to the surfotactic template, the subject pronoun must also
be displaced. Both it and the accusative pronoun thus are positioned immedi-
ately following the finite verb, giving the sequence PP hob ikh dos. Following
this come the remaining pieces of the sentence, in the order in which the syn-
tax puts them, vix. nit getort haltn.
The same considerations explain the positioning of various pieces of com-
plex verbs in Yiddish. As shown by the pairs below, the non-verbal part of the
complex verb follows its verbal part if it is finite, but precedes it if it is not finite.
In (37) the non-verbal part is an adverbial particle, in (38) it is a nominal de-
Sadcxjk: A trimodular account of Yiddish syntax
43
rived from a verb (see Aronson 1985), and in (39) it is a particle of Hebrew ori-
gin.
(37) a.
b.
(38) a.
Ikh gey arayn.
I go in
Ikh vil arayn-geyn.
I want in-go
Ikh tu a kuk.
I do a look
Ikh hob a kuk
I have a look
geton.
done
(39)
Zey zenen
they are
Zey zenen
they are
maskem.
agreed
maskem geven.
agreed been
We must assume something about the way these verbal augments are
generated, and all that needs to be done here is to assume with others (e.g.
den Besten & Moed-van Walraven 1985) that they are under the V node. A
simple example like (37a) would have a complex verb in the syntax, but the
templatic requirements would break it up, placing the finite-verb word itself^ in
second position. This is illustrated by the dual tree in (40).
(40)
NP
VP[FIN]
V[FIN]
NP V[FIN]
A I
ikh a kuk tu
SYNTAX
ikh tu a kuk
SURFOTAX
XP V[FIN]
Other elements positioned by the template will precede the stranded part
of the complex verb, since it is not specifically mentioned in the template at all.
44 Studies in the Linguistic Sciences 20:2 (Fall 1990)
A non-initial, non-pronominal subject will also precede the verbal prefix, since
it is positioned earlier in the syntax than the verb, but will follow the pronominal
elements that are explicitly positioned by the template:
(41) Emitser tut mikh a tore,
someone do me(ACC) a jab
'Someone abruptly jabs me.'
(42) Es hot mikh emitser a tore getan fun hintn.
it has me(ACC) someone a jab done from behind
'Someone has abruptly jabbed me from behind.' (DD, 134)
Subordinate clauses
The most basic facts concerning the form of subordinate clauses of Yid-
dish fail out from the simple trimodular system with the addition of a very few,
quite straightforward assumptions. First, ordinary indicative clauses, which be-
have exactly like main clauses in having verb-second word order and in freely
allowing topicalization (both problems for the GB treatments mentioned above),
are simply treated like main clauses. The complementizer, which is absent in
main clauses, may be present in the subordinate clause, but at the level of S',
rather than S. The domain of the verb-second template, though, is 8, so the
complementizer is simply irrelevant. The possibility of topicalization is a matter
for the discourse-functional component, and here the question is not the syn-
tactic, but rather the pragmatic status of the subordinate clause. Those subor-
dinate clauses whose pragmatic status is such that topicalization within them
makes sense should freely allow it. (See Green 1976.)
For interrogative subordinate clauses, we need only assume that the inter-
rogative phrase, either a single interrogative pronoun or a prepositional phrase
whose object is an interrogative pronoun, is the complementizer. Otherwise,
the account is the same as for indicative complements. This assumption also
explains the appearance of a fronted element, either a topic or the empty pro-
noun es, when the subject postion is relativized, as in (43) and (44), respective-
ly. Since the interrogative word or phrase is not in the clause proper, either a
fronted element as in (43), or an epenthetic element as in (44), is required by
the template so that the finite verb will be second in the clause.
(43) (s'zol keyn ben-odem nit visn) ver do ligt.
(it should no human not know) who there lies
'(No one should know) who is lying there.' (TsA, 155)
(44) ...ver es ligt do
... who it lies there
"... who is lying there.'
Sadock: A trimodular account of Yiddish syntax 45
Main-clause interrogatives have no complementizer, and therefore the
fronted inten-ogative element counts in the template. It must be initial because
it is a topic in discourse-functional structure, and therefore no other element
may be fronted in such examples:
(45) Ver ligt do?
who lies there
'Who is lying there?'
(46) *Verdo ligt?
Antitopic postponement
Non-initial subjects often come later than the theory presented above pre-
dicts, namely at or near the end of the clause. Examples (16) and (24) above
are instances of this, as is the following, simpler example.
(47) Es iz ibergegangn ayor.
it is passed a year
'A year has passed.'
Like fronted elements, this postponed subject has special discourse prop-
erties, but these are in some ways the reverse of those of the topic, as demon-
strated in a detailed study of Ellen Prince's (1988). The displacement of the
'antitopic' from its syntactic position should therefore be relegated to the dis-
course-functional component, as roughly sketched in the multi-structural dia-
gram (48). Note that the expletive element es is not represented in the syntax
(48)
X" V
Syntax
ibergegangn] [^^^, ayor] D-Fstruc-
'^i^' ture
Surfotax
46 Studies in the Linguistic Sciences 20:2 (Fall 1990)
at all, nor is it a topic in discourse-functional structure. The phrase a yor, with
which the verb agrees, is the syntactic subject. With a displaced subject and
no topic present, es, the syntactically and semantically empty lexical item of
Yiddish, must be present to satifsy the verb-second requirement of the surfo-
tactic template. It would not be needed — or allowed — , if some other element,
say, a discourse-functional topic, were in initial position, a well-known feature
of Yiddish syntax exemplified by (49).
(49) Fun yener tsayt on \z ibergegangn a yor.
from that time on is passed a year
'A year has passed since that time.' (cf. DM, 1 3)
While both topics and anti-topics are displaced from their natural syntactic
positions by the discourse-functional component, 1° an unfair competitor in the
battle for the ordering of elements, these two positions are not the same. The
reason is that the topic position also counts in the surfotactic template as X",
whereas the final position of a clause is not mentioned by the template. There-
fore, pronominal elements may be 'hidden' in topic position, but not in anti-
topic position. Only full NPs may be displaced rightwards, and this is again the
correct result. (See Prince 1988 for further discussion.)
Matters for further study:
The five rules distributed among three components provide a considerable
coverage of the facts of Yiddish syntax, more in fact than in any other explicit
study of Yiddish that I am aware of. They do not, however, do everything. In
closing I would like to point out some further details of Yiddish syntax and sug-
gest how they might be accommodated in the present framework, without
working out the details.
Adverbials: Various sorts of adverbial expressions, particularly small
ones, tend to gravitate toward the zone between the finite and non-finite parts
of the verb phrase. Such adverbials may also occur elsewhere in the sen-
tence, either initially or finally, depending roughly on semantics. The variation
is nicely illustrated by the following near-minimal pair.
(50) ...er kon dos unz eybig nit fargesn.
... he can it us forever not forget
"... he could never forget it (on our behalf).' (DD, 143)
(51) ...er vet dos unz nit fargesn eybig.
... he will it us not forget forever
"... he would never forget it (on our behalf).' (DD, 143)
Note that as predicted, the intercalated element eybig in (50) follows the
specifically mentioned prominal elements, but precedes the negation^ ^ and the
Sadock: A tritncxlular account of Yiddish syntax 47
non-finite verb. But the stream of pronouns can be interrupted by small, paren-
thetical items, though the relative order of the pronouns remains the same.
Thus, the actual text of example (28) reads:
(52) Hoybt dokh mikh mistame mayne on tsu traybn ... (DD, 144)
with the words dokh 'in fact' and mistame 'probably' interrupting the sequence
of post-verbal elements. I suggest that these are simply too small to be seen by
the template, though an actual treatment of them remains for future research.
One argument in the template: As observed by den Besten and
Moed-van Walraven (1985), either the direct object or the indirect object, but
not both, may also optionally appear between the finite and non-finite parts of
the verb. This is particularly common with negative indefinite objects, but is by
no means restricted to them.
(53) Ikh kon keyn vetshere nit esn.
I can no dinner not eat
'I can't eat any dinner.' (DM, 16)
(54) Konst zikh shoyn gor keyn arbet nit op-zukhn?
can (2s) self already at all no work not out-seek
'Can't you find any work at all for yourself?' (DM, 20)
Note the position of the intercalated object in (54), between the little adver-
bials and the part of the sentence not under the control of the template. This
suggests the existence of a truly optional position in the template, making (55)
its final form:
(55) !Xn — IVfin — ProNOM — ProREF — Ptoacc — ProDAT — (NP)
NOTES
*A preliminary version of the thesis developed here is sketched in Sadock
Forthoming. While more jocular in tone, that piece is as serious in its intent as
the present one. Several people have given me help and encouragement with
this paper. I would especially like to thank Howard Aronson, Joan Maiing,
David Perlmutter, Ellen Phnce, and Elisa Steinberg for their help.
1 The obligatoriness of the rule was taken care of by assuming that any
sentence containing this dummy element in surface structure would be filtered
48 Studies in the Linguistic Sciences 20:2 (FaU 1990)
out, as in Chomsky's (1965) treatment of the obligatoriness of movement in
relativization.
2 Among the numerous 'principles' and 'paramaters' invoked in account-
ing for Yiddish word order are the direction of Case assignment, whether there
is one or more complementizer position, whether or not the complementizer
position can be doubly filled, whether the subject moves to Comp or Spec of IP,
whether the subject is generated under VP or S, whether movement to IP is A
movement or A' movement, whether there are empty complementizers, and
several other things as well. Putting all these parametric and theoretical op-
tions together, one sees that the theory allows literally thousands of possible
Yiddish-like languages. The claim that Yiddish syntax is explained by such a
theory seems greatly exaggerated.
3 I have a few examples in which the reflexive precedes the neuter nomin-
ative pronoun, suggesting that the theory developed below needs refinement.
One example of this is:
Dort ligt zikh es gants ruig
there lies self it very peaceful
'It lies there very peacefully.' (TsA, p. 1 54)
^ Many of the traditional difficulties that are encountered in determining the
position of topic phrases in syntax are eliminated if topics are not located in the
syntax at all, but rather in a component whose units all regard the discourse
status of the pieces of an expression.
5 The page references in these citations refer to the collection Sholem
Aleichem 1926. The abbreviations are as follows:
DM = Dos Meserl ('The pocket knife'), in volume 7: Mayses far
Yidishe Kinder;
DO = Der Daytsh ('The German') in volume 9: Oreme un Frey-
lekhe;
TsA = Tsvay Antisemitn ('Two antisemites") In volume 10: Oreme
un Freylekhe.
6 More specifically, this would seem to be the level of morphophonology
argued for by Woodbury (1989) and Baker (In Press).
'' The two ideas come together in the proposal of Hock's (1990) that verb-
second word order originated with the cliticization of the finite verb to the first
sentential element.
8 The word mayne is a possessive pronoun, but is not used pronominally.
Its declension shows that it is an adjective in this usage, presumably a modifier
of a phonetically empty noun. Hence it is not under the control of the template.
Sadock: A trimodular account of Yiddish syntax 49
9 In finding the finite verb, clearly the template looks for a morphological,
rather than a syntactic form. Thus, although arayn gey, a kuk tu, and maskem
zenen are the syntactic finite verbs in the (a) examples of (37) - (39), only the
morphological finite verbs gey, tu, and zenen are positioned by the surfotax.
This fact highlights the non-syntactic nature of the surfotactic template.
10 It is not clear to me whether the three elements that we have now recog-
nized in the discourse-functional pattern should be generated as a 'flat' struc-
ture or whether, perhaps, the comment and topic should form a constituent, or
the comment and antltopic should.
11 I assume that negation is generated in the syntax as a modifier of VPs
and is not handled specifically by the template.
REFERENCES
Aleichem, Sholem. 1926. Ale Verk fun Sholem Aleichem. New York: Sholem-
Aleichem Folskfond Oysgabe.
Aronson, Howard I. 1985. On aspect in Yiddish. General Linguistics 25. 171-
88.
Baker, Mark. In Press. On the syntax of pronominal agreement morphology.
CLS 26.
Chomsky, Noam. 1965. Aspects of the theory of syntax. Cambridge, MA: MIT
Press.
Den Besten, Hans, & Corretje Moed-van Walraven. 1985. The syntax of verbs
in Yiddish. Haider & Phnzhorn 1 985:1 1 1 -35.
DiESiNG, Molly. 1990. Verb movement and the subject position in Yiddish.
Natural Language and Linguistic Theory 8.41-80.
Gazdar, Gerald. 1981. Unbounded dependencies and coordinate structure.
Linguistic Inquiry 12.155-84.
GREEN, Georgia. 1976. Main clause phenomena in subordinate clauses. Lan-
guage 52.382-97.
Haider, H., & M. Phnzhorn (eds.). 1985. Verb second phenomena in Germanic
languages. Dordrecht: Foris.
HOCK, Hans Henrich. 1990. V2 in early Germanic: A 'heretical' view. Ninth
East Coast Indo-European Conference, University of Pennsylvania.
Maling, Joan, & Annie Zaenen. 1981. Germanic word order and the format of
surface filters. Binding and filtering, ed. by F. Heny, 255-78. London:
Croom-Helm.
Perlmutter, David M. 1971. Deep and surface structure constraints in syntax.
New York: Holt, Rinehart & Winston.
50 Studies in the Linguistic Sciences 20:2 (FaU 1990)
Prince, Ellen. 1981. Topicalization, focus movement, and Yiddish-movement:
A pragmatic differentiation. BLS 7.249-64.
— . 1988. The discourse function of Yiddish expletive es+subject-preposing.
Papers in Pragmatics 2.176-94.
Sadock, Jerrold M. 1965. Some aspects of Yiddish syntax. Unpublishable
term paper. University of Illinois at Urbana-Champaign.
1985. Autolexical syntax: A theory of noun incorporation and similar phe-
nomena. Natural Language and Linguistic Theory 3.379-440.
. 1991. Autolexical syntax: A theory of parallel grammatical representa-
tions. Chicago: University of Chicago Press.
Forthcoming. A paper on Yiddish for James D. McCawley. The joy of
grammar: A festschrift in honor of James D. McCawley, ed. by D. Brentari,
L. McC. Hand, and G. N. Larson. Amsterdam: Benjamins.
Woodbury. Anthony C. 1989. On restricting the role of morphology In auto-
lexical syntax. Workshop on Autolexical Syntax, University of Chicago, 1 6
April 1989.
Studies in t he Linguistic Sciences
Volume 20, Number 2. Fall 1990
SOME ISSUES IN LANGUAGE ORIGINS AND EVOLUTION
Frederick J. Newmeyer
(University of Washington)
Linguists have long considered discussion of the origins and evolution of
language to be disreputable. In 1866 the Societe Linguistique de Paris issued
an outright injunction against speculation on the topic at its conferences and in
its publications. Rumor has it that the Linguistic Society of America considered
the same ban upon its founding in 1924, but settled instead for a 'gentlemen's
agreement' (as such things were then known) prohibiting papers on language
origins. Whether this story is true or not, not a single article in Language has
ever addressed the topic.
It is not difficult to understand the reasons for the ill repute associated with
the question of the origins of language. There is probably no area of concern to
linguists that lends itself to so much uncontrolled speculation. And this follows
from the fact that there is so little in the way of hard evidence to fuel sensible
theohzation. There are no archeological digs turning up specimens of proto-
language. While fossil evidence has given us a clear picture of the evolution of
the vocal tract (Lieberman 1984), grammatical structure leaves behind no fos-
sils whatever. And, most seriously, a major tool of evolutionary biology, the
comparative method, is inapplicable to the study of the origins of language.
This method demands homologs in related species to the trait under exam-
ination, in which, from minimal differences between them one can build plaus-
ible stories about the trait's evolution. Yet the central aspects of language —
syntax and phonology — have no homologs, even for our most closely related
species. Language is thus an emergent trait (or 'key innovation') and therefore
poses, along with all such traits, particular problems for evolutionary biology.
As a result, much that has been written about the origins of language has
the flavor of 'just-so' stories, not much more advanced than the bow-wow,
heave-ho, and ding-dong theories reported in the introductory texts.
Nevertheless, I feel that it is time to put the question back on the theoretical
linguist's research agenda.'' Several factors now allow at least some of the
woolliness to be removed from the admittedly hirsute speculation that always
52 Studies in the Linguistic Sciences 20:2 (Fall 1990)
characterized it. First, several decades of work in generative grammar have led
to reasonable hypotheses about which aspects of language are innately de-
termined, and therefore germane to the question of the biological evolution of
language in a way that its more contingent properties are not. Before we can
know HOW language evolved, we need to be pretty sure about precisely what
evolved. Second, from attempts to teach signed language to chimpanzees and
gorillas, we have a fairly clear understanding of the linguistic capacities of high-
er apes, which, in turn, opens up the possibility of reasonable hypotheses about
the capacities of pre- and proto-hominids. And finally, new findings in paleo-
neurology have led to surprising discoveries about the evolution of the brain, in
particular those areas dedicated to language.
It is my feeling as well that generative grammarians have an obligation to
address the question of the evolution of language. A central tenet of the prin-
cipal approach within generative grammar, that associated with Chomsky and
his co-thinkers, is that our biological endowment embodies an innately determ-
ined universal grammar (UG) that accounts for the major grammatical properties
of the world's languages and helps to shape the acquisition by children of par-
ticular grammars. A persistent criticism of this UG position has centered on the
absence of any account of its phylogenesis. Why, it is often asked, would the
hypothesized universal properties of language, whether at the level of gram-
matical organization as a whole or at the level of particular UG principles, ever
have become incorporated into the human genome? One must concede that
the absence of even the rudiments of an answer to this question has conferred
a rhetorical advantage to those opposing the idea of an innate UG.
This criticism comes from many directions, but is especially vocal in the
'functionalist' wing of the field. While this wing is itself quite diverse, the majority
reject the very concept of an autonomous UG. Rather, they believe that gram-
matical patterning is grounded in what is seen as the most important 'function'
of language, namely communication. In place of an innate UG, they assume
that the child is endowed with general learning strategies that underlie both the
acquisition of grammar and the norms and conventions governing the ap-
propriate use of language in social interaction.
Not all functionalists, however, share such a view. A wing of functionalism
represented by Susumu Kuno, Ellen Prince, Georgia Green, and others does
not reject the UG perspective. Indeed, Kuno 'finds no conflict [in theory]
between functional syntax and, say, the government and binding theory of
generative grammar' (1987:1). Such linguists are functionalists, not because
they believe that functional principles invalidate the idea of an innate auton-
omous UG, but because their work is devoted to the discourse or processing
functions of syntactic forms.
I hope to demonstrate that the study of the origins of language lends
support to the Kuno-Prince-Green idea that there is no incompatibility between
Newmeyer: Some issues in language origins and evolution 53
taking a formalist perspective on language and taking a functionalist one. In
fact, I will go even farther than this and argue that the more of a functionalist one
is, the more one should be led to support the idea of an autonomous grammar
whose central principles are innate.
One might at first assume that the innate principles of UG are simply
immune to functional explanation. But this is not so. There exists a well-ac-
cepted (functional) mechanism for explaining the provenance of innate traits:
natural selection. It is logically plausible that the design of the grammatical
model as a whole or some particular grammatical principle might have become
encoded in our genes by virtue of its being so successful in facilitating com-
munication that the survival and reproductive possibilities of those possessing it
were enhanced. In this sense, a functional explanation would hold at the evo-
lutionary level.
Until very recently, formal linguists who have addressed the question at all
have appeared quite reluctant to point to natural selection as the evolutionary
force that shaped the language faculty. Chomsky, despite an earlier observa-
tion that 'language must surely confer enormous selective advantages' (1975:
252) and his speculation (with Lasnik) that if there is a functional explanation for
a particular filter, it might hold 'at the level of evolution of the species' (Chomsky
& Lasnik 1977:437), now takes the view that the nature of UG is perhaps be-
yond the reach of an adaptationist explanation, and points instead to 'physical
principles' being at work:
Evolutionary theory is informative about many things, but it has little to
say, as of now, of questions of [language evolution]. The answers
may well not lie so much in the theory of natural selection as in
molecular biology, in what kinds of physical systems can develop
under the conditions of life on earth and why, ultimately because of
physical principles. (Chomsky 1988a:167)
While Chomsky does not elaborate on the nature of these principles, he pre-
sumably has in mind the sort discussed so elegantly by D'Arcy Thompson in
1917 (cf. Thompson 1961). Thompson explains that many patterns and shapes
occurring repeatedly in nature are a natural consequence of such physical
properties as the ratio of an organism's length to its surface area, design pres-
sures for efficient utilization of space, and so on.
Elsewhere, Chomsky dismisses 'speculations about natural selection [as
being] no more plausible than many others; perhaps [properties of UG] are
simply emergent physical properties of a brain that reaches a certain level of
complexity under the specific conditions of human evolution' (Chomsky 1988b:
22). In this article, Chomsky goes so far as to claim that, far from conferring
selective advantage, some properties of UG are actually dysfunctional to the
species. For example, he considers the 'Last Resort' principle, which insures
54
Studies in the Linguistic Sciences 20:2 (Fall 1990)
that derivations be as economical as possible and contain no superfluous
steps, to be dysfunctional because it causes computational difficulties.
(Chomsky 1986a) The parser would seem to have to scan globally all possible
derivations before it came across the right one. He concludes that while lan-
guage might be 'beautiful', it is at the same time 'unusable', and must resort to a
number of 'computational tricks' to allow structure to be recovered at all. 2
What I believe to be Chomsky's current position is elaborated by Piattelli-
Palmarini (1989), who offers the opinion that 'the study of language has, in fact,
disclosed many instances of specificity and gratuity in the design of all natural
human languages, but hardly any instance of traits dictated by general com-
municative efficiency' (22). His two major examples are the Projection Principle
(Chomsky 1981) and the principle of Full Interpretation (Chomsky 1986a),
which 'adaptation cannot even begin to explain' (25). Piatelli-Palmerini's dis-
cussion of language evolution is embedded in a view popularized by Stephen
Jay Gould and others that extra-adaptive mechanisms vie with or perhaps even
eclipse natural selection as the prime mechanism of evolutionary change.^ To
Piatelli-Palmerini, language is a 'spandrel' (Gould & Lewontin 1979), es-
sentially an epiphenomenal byproduct of evolution.'*
It is logically possible that Chomsky and Piattelli-Palmarini are correct, that
the innate principles of UG arose as a chance byproduct, as it were, of forces
unrelated to their utility to the species. However, the remainder of this paper will
be devoted to arguing that there is no reason to reach such a conclusion.
Rather, I will defend the position that innate autonomous grammatical principles
were selected for because they alloted a greater evolutionary advantage to
populations that had them. In short, if the line of reasoning to be taken is cor-
rect, one can deduce the functional need for formal principles of grammar.
Let us begin with what linguists of all persuasions agree is the task of any
linguistic theory, namely to relate sounds and meanings (perhaps 'expressions'
would be a more appropriate term than 'sounds', so as not to exclude signed
languages). Since humans can conceptualize many thousands of distinct
meanings and can produce and recognize a great number of distinct sounds,
one's first thought might be that this relation could be expressed in large part by
a simple pairing of individual sounds with individual meanings, as in (1):
(1)
MEANING
SOUND
MEANING
SOUND,
MEANING
SOUND,
MEANING
SOUND,
At the domain of lexical meaning, no such one-to-one pairing exists, of course.
A vastly greater number of words can be stored, retrieved, and used efficiently if
Newmeyer: Some issues in language origins and evolution 55
sequences of a small number of distinctive sounds are paired with meanings
than by a direct mapping between individual meanings and Individual sounds.
But what about propositional meaning, where the question of a one-to-
one pairing Is rarely, if ever, raised? The Infinitude of possible messages that
can be conveyed cannot In and of itself be the explanation. While humans can
formulate an indefinite number of propositions, we can also produce and per-
ceive an Indefinite number of sound sequences. Thus a one-to-one pairing be-
tween them is at least within the realm of logical possibility.
The most plausible answer is that sound and meaning are too different
from each other for this to have ever been a practical possibility. Meanings,
whatever their ultimate nature, are first and foremost mental realities, with no
obvious physical instantiation. Sounds, physical realities par excellence, are
produced by a coordinated set of articulations In the vocal tract, under control of
a very different area of the brain from that responsible for meaning. Fur-
thermore, in the conceptual structures that represent meanings, temporality and
linearity play no role.^ Such structures do, however, contain diverse types of
hierarchies and structured relationships: predicate argument dependencies,
and relations of inclusion. Implication, cross-classification, and identity. More-
over, conceptual structures are discrete. In the representation of a sentence like
the girl threw the ball, for example, girl, threw, and ball do not grade continu-
ously into one another.
Phonetic representations, on the other hand, have almost none of these
properties. A phonetic representation is temporal and quantitative. While partly
hierarchical in nature, there is no direct relationship between the hierarchy of a
phonetic representation and that of a conceptual structure. Indeed, the articu-
latory gestures, formant frequencies, tone patterns, and so on relevant to pho-
netics have nothing in common with the properties of a conceptual structure.
And this mismatch is alleviated only slightly by appealing to phonological in-
stead of phonetic representations.
In other words, a major evolutionary step toward vocal communication was
the development of an intermediate level between sound and meaning, a
'switchboard', if you will, which had the effect of coordinating the two. Only at
that point could propositional meanings be conveyed with any degree of ef-
ficiency.
What properties might we deduce about this intermediate level? First, it
would have to contain a small number of basic units. No advantage would have
been conferred by the development of a third level with thousands of basic
entities. And second, this level would need to share some properties with
conceptual structures and some properties with phonetic representations, but
be constructed out of units common to neither. Communication (and its benefits
56 Studies in the Linguistic Sciences 20:2 (Fall 1990)
to the species) would not have been facilitated if this level had been skewed too
much either to the sound end or to the meaning end of the spectrnm.
What we have just done, of course, is to deduce the selective advantage of
autonomous syntax! This level contains a small number of basic units (no more
than a couple dozen syntactic categories are postulated for any given langu-
age), which are related to each other by the simple notions of 'dominate' and
'precede'. In this way, a syntactic representation contrasts markedly with the
complexity of a semantic or phonetic one. Further, a syntactic representation
shares some properties with the former (hierarchy, dependency) and some with
the latter (linear sequencing), yet is governed by a calculus neither semantic
nor phonetic.
Again, from the functional pressure favoring the development of a
workable system of communication (i.e. from pressure to pair sounds and
meanings efficiently) and with it the reproductive advantage that this ability to
communicate would confer, autonomous syntax arose in the course of langu-
age evolution.
Let us look more closely at the mapping between conceptual structures
and phonetic representations; cf. (2) next page.
Each level is linked by a set of rules to the level above or below it, which
carry a derivation a step closer to sound from meaning, or vice versa, and each
level is governed by its own autonomous principles of organization.
Subparts of conceptual (i.e. semantic) structures are replaceable by in-
dividual lexical items in lexical conceptual structure (Hale & Keyser 1986, 1987)
in accord with the lexicalization principle discussed in Jackendoff 1983. As a
result of the linking rules, predicate-argument structures are created, in which
the specific content of the thematic information present in lexical conceptual
structure is lost (Rappaport & Levin 1988). Linearization principles (the Prin-
ciple of Case Adjacency, directionality of Case and/or Q-role assignment, and
so on) transform predicate-argument structures into syntactic structures termina-
ting in phonologically specified lexical items. ^ The phonosyntactic rules are
sensitive only to a subset of syntactic constituent structure, namely that provided
by principles of X-bar theory, in building the phonological and intonational
phrases that define the level of prosodic structure (Selkirk 1986). All syntactic
information is lost by the time of the application of the phonological rules, and in
the phonetic realization rules, quantitative information enters the derivation for
the first time (Pierrehumbert & Beckman 1988). Thus, this autonomous-systems
view embodies a small set of manageable operations functioning in concert to
link the inherently disparate components of language.
Newmeyer: Some issues in language origins and evolution
57
(2)
CONCEPTUAL
STRUCTURE
LEXICAL
CONCEPTUAL
STRUCTURE
PREDICATE
ARGUMENT
STRUCTURE
SYNTACTIC
STRUCTURE
PROSODIC
STRUCTURE
SURFACE
PHONOLOGICAL
REPRESENTATION
PHONETIC
REPRESENTATION
LEXICALIZATION PRINCIPLES
LINKING RULES
LINEARIZATION PRINCIPLES
PHONOSYNTACTIC RULES
PHONOLOGICAL RULES
PHONETIC REALIZATION RULES
The reader who has been convinced of the functional utility of autononnous
syntax might wonder whether selective forces could have shaped a gram-
matical model with the intricacy of that depicted in (2). Before addressing this
question directly, I must say that I find Chomsky's Thompsonian explanation for
the design features of any significant aspect of the language faculty to be utterly
implausible. The hexagonal cell aggregates, the equiangular spirals, and so on
found repeatedly in nature, and determined by the same laws of physics that
suggest the optimal design for a bridge or arrangement of packing crates have
no counterpart in the language faculty. Indeed, perhaps the most salient (and,
at times, frustrating) aspect of UG is its lack of symmetry, the irregularity and
idiosyncracy it tolerates, the widely different principles of organization of its vari-
ous subcomponents and consequent wide variety of linking rules relating them.
And yet the entire package of properties is, without question, adaptive. In
this respect, UG is like other complex organs containing a multitude of subparts.
58 Studies in the Linguistic Sciences 20:2 (Fall 1990)
each performing its role in harmony with the others. And the only explanation
we have available for the origin of adaptive complexity is natural selection.
Richard Dawkins (1986:288) makes this point with reference to the human eye7
There is one particular property of living things, however, that I want to
single out as explicable only by Darwinian selection: adaptive com-
plexity ... Following Paley, I have used the example of the eye. Two
or three of the eye's well-'designed' features could, conceivably, have
come about in a single lucky accident. It is the sheer number of inter-
locking parts, all well adapted to seeing and well adapted to each
other, that demands a special kind of explanation beyond mere
chance. The Darwinian explanation, of course, involves chance too,
in the form of mutation. But the chance is filtered cumulatively by
selection, step by step, over many generations ... [T|his theory is cap-
able of providing a satisfying explanation for adaptive complexity.
UG, with its 'sheer number of interlocking parts, all well adapted to [langu-
age] and well adapted to each other', demands an explanation in terms of
Darwinian natural selection as well.
But by what means and through what steps could natural selection have
yielded such a model? Clearly, if there were no alternative to the conclusion
that it came into being full-blown as a result of a single monster mutation of
gigantic and miraculously salubrious proportions, then skepticism about an
evolutionary account would be more than justified. Fortunately, however, there
is no reason to appeal to the evolutionary equivalent of divine intervention.
Rather, language evolution is an example of 'mosaic' evolution, in which selec-
tive forces steered once separate and evolutionarily unrelated components to
become integrated over the passage of time. Each step fed the following one,
each resulting in a more efficient and utilitarian system of communication.
Jacques Monod observed that 'as soon as a system of symbolic communication
came into being, the individuals, or rather the groups best able to use it, ac-
quired an advantage over others incomparably greater than any that a similar
superiority of intelligence would have conferred on a species without language'
(1972:126). And each evolutionary improvement in this system bestowed a fur-
ther advantage to those possessing it.
To begin our story, studies of ape intelligence (Premack 1976, Premack &
Premack 1983) suggest that prehominids possessed a surprisingly sophistica-
ted level of mental representation. The conditions for the subsequent develop-
ment of language as a medium of communication were set by the evolution of
this level into a faculty able to integrate 'information from peripheral systems
such as vision, nonverbal audition, smell, kinesthesia, and so forth' (Jackendoff
1983:18), i.e. into the level of conceptual structure. (For discussion of how this
might have taken place, see Wilkins & Dumford 1990, and In Preparation). It is
here that we find the central evolutionary antecedents of language. As Bicker-
Newmeyer: Some issues in language origins and evolution 59
ton (To Appear) rightly stresses, the properties of human language have little in
common with any known system of animal communication, but much in com-
mon with animal, in particular primate, representational systems.
A first step toward the evolution of this system for communication was
undoubtedly the linking of individual bits of conceptual structure to individual
vocalizations, perhaps along the lines manifested by animals whose com-
municative repertoire consists of a series of fixed calls. Once this was in place,
the stage was set for the two crucial steps that would remove human com-
munication forever from the company of animal communication: first, the devel-
opment of the level of lexical conceptual structure whose central component is
an autonomous lexicon consisting of a set of bidirectional sound-meaning pair-
ings; and, second, the capacity to transmit an unbounded number of stimulus-
independent messages.
In an important paper, Hurford (1989) speculates on how the level of
lexical conceptual structure (which can be identified as the locus of the
Saussurean 'sign') might have originated. Since vocal communication can take
place without an autonomous lexicon (as in animal communication), he as-
sumes that at an early stage our ancestors were able to vocalize concepts and
to understand them when vocalized by others without having sound-meaning
pairings stored as such. Language acquisition and communication took place
simply as a result of generalizing observed verbal behavior. Employing mathe-
matical models, Hurford demonstrates that successful communication is greatly
facilitated if it is possible for the language learner to construct an autonomous
sign on the basis of the observation of these acts of linguistic transmission and
reception. Thus once the human brain had evolved to a level of complexity to
allow it, the level of lexical conceptual stmcture came into being.
Simultaneously, a phonological level distinct from phonetic representation
was in the process of evolving. Indeed, it seems likely that from the moment that
the vocal channel was employed for the expression of concepts, a primitive
phonology was in place. As Mattingly (1972) points out, the roots of phonology
(the imposition of structure on the continuous speech stream) lie in the ability to
perceive sign stimuli categorically, an ability shared by a wide variety of animal
species. Furthermore, we know that the human vocal tract underwent a rapid
evolution whose only function was seemingly to facilitate the production of an
ever greater diversity of sounds (Lieberman 1984). At a certain point in this
evolution, a workable system of phonotactics must have evolved, which allowed
for the possibility of different combinations of a set of basic sounds, each com-
bination linked with a different conceptual structure. In any event, the level of
phonological representation allowed for an efficient storage and retrieval of a
vastly greater number of elements than a phonetic level alone and must have
conferred an enormous evolutionary advantage to the populations that devel-
oped It.
60 Studies in the Linguistic Sciences 20:2 (Fall 1990)
The reproductive advantage of being able to convey an unbounded
number of stimulus-free messages can hardly be doubted, though the point at
which the emergent linguistic system was first able to achieve this is a mystery.
(Brandon & Hornstein (1986) suggest that evolutionary pressure for phenotypic
transmission of information, which demands a system with such properties, was
especially acute in the capricious and rapidly-changing environment in which
our ancestors lived.) What is clear, however, is that the communication of pro-
positions demands a syntactic level, the final link in the evolutionary chain lead-
ing to human language. This level, as suggested in the preceding section, has
design features that suggest that it arose as an interface to link preexisting com-
ponents, an idea stressed by Alvin Liberman (1974:44) in a paper not widely
known to grammatical theorists:
[T]he several components [of language] developed separately in evol-
ution and in connection with different biological activities. At the one
end of the system is long-term memory, as well as the nonlinguistic
aspects of meaning and thought ... At the other end of the system, the
components most directly concerned with transmission — the ear and
vocal tract — had also reached a high stage of development before
they were incorporated as terminals in linguistic communication ...
We might assume, then, following Mattingly (1972), that grammar de-
veloped as a special interface joining into a single system the several
components of transmission and intellect that were once quite sep-
arate.
The emergent syntactic level drew in particular on conceptual structure.
Indeed, if Jackendoff (1983) is correct that every major phrasal constituent in a
sentence corresponds to a conceptual constituent in the sentence's semantic
structure, then the influence of conceptual structure on syntactic representations
was profound. But the fact that syntax evolved to coordinate this former level
with the vocal output channel led to other, and sometimes conflicting pressures
on its design features. In particular, since concepts have to be expressed in
real time and by means of a vocal tract exapted from structures originally
evolved for respiration, olfaction, and digestion (and thus not in any sense 'per-
fected' for communication), a second set of forces contributed to the shaping of
syntax. In particular, there arose many conflicts between the demand that it 'fit'
well with semantics (which would favor a one-to-one match up between con-
cepts and syntactic categories) and the demand that it feed smoothly into the
expressive plane (which would favor structures designed for ease of production
and perception). The resulting level, as a consequence, came to mirror neither
perfectly, but rather developed its own distinct set of governing phnciples.
In the view of many formal linguists, it is not just the components of the
grammar that are innately specified, but also many specific principles operating
within and between components. One of these is Subjacency, which is stated
as follows (Van Riemsdijk & Williams 1986:62):8
Newmeyer: Some issues in language origins and evolution 61
(3) No rule can relate X, Y in the structure
...X...[a...[p ...Y... (or:...Y...]p ...]„-.. X...)
where a, p are bounding nodes.
Subjacency, in effect, keeps rules from relating elennents that are 'too far apart
from each other', where distance apart is defined in terms of the number of
designated nodes (bounding nodes) that there are between them.
Subjacency accounts for the violations of grammaticality in the English
sentences (4a-b):
(4) a. *Whatj do you wonder where John put |?
b. *Whatj do you believe the claim that John ate |?
In these sentences, two bounding nodes intervene between the gap and the
word what.
The strongest piece of evidence supporting the innateness of Subjacency
is based on the poverty of the stimulus presented to the child language learner.
Poverty of the stimulus arguments take the following form. One points to a hy-
pothesized principle of UG and reasons that given its abstractness, the limited
amount of relevant data made available to the child (in particular the fact that
children's syntactic errors are rarely corrected), and the speed of acquisition,
there is no way that it could have been learned inductively. Hence the principle
must be innate.
Hoekstra & Kooij (1988), for example, motivate the innateness of
Subjacency by pointing out that positive evidence alone could hardly suffice to
enable the child language learner to come to the conclusion that (5a) is am-
biguous as to the scope of where, while (5b) is not:
(5) a. Where did John say that we had to get off the bus?
b. Where did John ask whether we had to get off the bus?
They conclude, quite reasonably in my opinion, that knowledge of the per-
missible intervening structure between a Wh-phrase and its associated gap
must be prewired into the child.
Many linguists have attempted to provide functional grounding for UG
principles, in some cases arguing that such grounding invalidates their very ex-
istence. And to be sure, many functional explanations seem at first blush to be
highly plausible. Subjacency, in particular, has received attention as a principle
admitting to a functional explanation. Suggestions as to its functional basis
have pointed to the processing problems created for the hearer in matching the
62 Studies in the Linguistic Sciences 20:2 (Fall 1990)
displaced w/)-element with its coindexed gap (Givon 1979, Berwick & Weinberg
1984, Frazier 1985) or to cognition-based strategies of sentence interpretation
that disfavor Subjacency-violating structures (Deane 1988). However, it is rare-
ly pointed out that Subjacency performs no particular service for the speaker,
whose 'easiest' task would simply be to 'wh' any Noun Phrase regardless of its
subcategorized position in the structure. In other words, Subjacency exhibits a
functional asymmetry. This very asymmetry, it is worth pointing out, further
serves to bolster the case for its innateness, since, surely, children learning their
language could not be expected to refrain from uttering the relevant un-
grammatical structures because they had come to realize through experience
that their conversants might have trouble processing their utterances.
The same point can be made with respect to two other hypothesized
principles of UG, Principle A of the binding theory (Anaphor Binding) and the
ECP. While these principles may help the hearer more efficiently to pair ana-
phoric elements and their antecedents, they seemingly complicate matters for
the speaker, who, of course, is fully aware of the identity of the intended referent
and is thus forced to make a 'personally' unnecessary grammatical distinction.
Anaphor Binding and the ECP are therefore functionally asymmetrical as well.
In short, it seems to be the case that those grammatical phenomena whose
explanation is most convincingly attributed to some principle of UG tend to be
those whose functional grounding is asymmetrical between speaker and
hearer.
The tendency for innate constraints to exhibit a functional asymmetry is a
natural consequence, I believe, of evolutionary pressure for language to serve
as an ever more efficient medium of communication. In cases where ease for
the speaker and the requirements of the hearer were in direct conflict, an ob-
vious solution presented itself — to bypass directly the push-pull between
speakers' demands and hearers' demands by incorporating those constraints
necessary to the hearer directly into the innate language faculty itself. Thus the
principles of UG were selected for, allowing a stable innate core to language,
immune to the functional exigencies of the moment.
There was no evolutionary pressure, of course, to biologize what aided
speaker and hearer equally. There would hardly be any benefit in encoding in
our genes some linguistic principle that the path of least effort would lead both
participants in a discourse to follow anyway.
It may seem at first blush a bit ironic that function-based factors should lead
to an innate UG, but I feel that a moment's reflection will dispel any feelings of
irony. If we agree with the functionalist thesis that the ability to communicate by
spoken language is a paramount human attribute, and contributed more than
anything else to the survival and development of the species, then we would
EXPECT anything that facilitates this process to become biologized. If Sub-
Newmeyer: Some issues in language origins and evolution 63
jacency really does ease communicators* burdens, then is it not at least plaus-
ible that its biologization might have conferred an advantage to the species?
There are, however, serious obstacles that an adaptationist account of UG
principles must overcome. To begin with, it is incumbent on one advocating
such an account to make a convincing case that the survival and reproductive
possibilities of constraint-bearing hominids are greater than those without them.
Is this plausible? Would an individual whose grammar was governed by Sub-
jacency be more likely to survive to adulthood than one without that constraint?
The answer, it seems clear to me is, all other things being equal, the very idea is
absurd. And so on for each individual constraint, taken one at a time, that has
been posited to be part of UG.
But what about all of the constraints taken together as a single package?
Here the answer is much less clear. If we consider the entire repertoire of UG
constraints, Subjacency, Anaphor Binding, the ECP, the Q-Criterion, and the
rest, then it is by no means far-fetched that their possession might have con-
ferred an evolutionary advantage. UG principles are involved in the derivation
of every sentence and therefore indirectly in every act of communicating. They
constrain the interpretation of quantifier scope and the antecedents of gaps,
they identify anaphoric elements, they help keep arguments distinct for partic-
ular predicates, and much more. In short, by allowing for an expressive rich-
ness that would be absent if they did not exist, they vastly aid the process of
communication.
It seems clear that to the extent that UG principles can be linked to each
other, i.e., one subsumed under the other, the less difficulty will be inherent in
an adaptationist account. What is needed for such an account to work is a very
small number of master principles arising in the course of evolution, and the
great bulk of principles following deductively from these. ^
A second problem involves the time frame. The conventional view links
the origins of language to the origins of the species Homo sapiens, and that in
turn to "art, the domestication of animals, agriculture, and the creative explosion
that produced the world we know today' (Bickerton 1991 :xxx). If such a scena-
rio is correct, then language must be very young indeed. According to recent
estimates, our species may be only 100,000 years old, and possibly as young
as 70,000 years. Australian aborigines, whose languages have essentially the
same linguistic devices as other languages, split off from the rest of the human
race about 40,000 years ago. As Bickerton notes, what may be as little as
30,000 years is far from enough time for ail of the principles of UG to have
arisen, each by a separate benificent mutation.
Again, we are led to the conclusion that only a very small number of master
principles could have arisen in the course of evolution.
64 Studies in the Linguistic Sciences 20:2 (Fall 1990)
Furthermore, other evidence suggests that it is perhaps unwise to tie lan-
guage to the origins of Homo sapiens. The time of the advent of civilization is
only part of what we have available that might help us date the origins of the
language faculty. Certain paleoneurological evidence suggests an evolution-
ary departure point coinciding with the emergence of the genus Homo, that is,
between 2.5 and 2 million years ago — more than enough time for natural
selection to run its course. For example, a specimen of Homo habilis yielded an
endocast which showed a folding pattern in the left frontal lobe similar to that
which is associated with Broca's area in living humans. This has led to the
conclusion that Habilis may have been capable of some form of speech (Tobias
1981, Dumford 1989). There is also evidence that the Habilis brain had a
Wernicke's area (Falk 1980, 1983), arguably indicating the possession of con-
ceptual structure, in my view the most important evolutionary antecedent to
language (see also Wilkins & Dumford 1990 and In Preparation). In short, it is
not implausible that the language faculty, in whatever form, greatly antedates
the cultural explosion to which Bickerton and many others appeal.
To conclude, focussing on language origins and evolution offers a new
perspective on the old debate between formalists and functionalists in lin-
guistics. Every formalist must recognize that many properties of the structural
systems whose workings he or she is devoted to elaborating, including the in-
nate principles which comprise them, arose for a good reason. More often than
most formalists have been willing to accept, external factors based in commun-
icative efficacy helped to steer grammar in the course of language evolution.
Functionalists, then, have been right in stressing the interest and importance of
identifying the external factors that have led grammar to take its present shape
and form.
On the other hand, functionalists should follow the lead of Kuno, Prince,
and Green and recognize that the existence of these factors, as profound as
they may have been, in no way threatens the fundamental formalist tenet,
namely that of the autonomy of grammar. Indeed, as we have seen, the func-
tionalist position, worked through to its logical conclusion, leads to the real-
ization that all linguists should accept the idea that central to language there
exists an autonomous grammar, shaped in part by natural selection.
NOTES
1 And I feel heartened by the fact that others have come to the same
conclusion (see for example Hurford 1989, Pinker & Bloom 1990, Bickerton
1990, Wilkins & Dumford 1990 and In Preparation). Many of the issues taken
up in the present paper are treated in more detail in Newmeyer 1990, 1991.
Newmeyer: Some issues in language origins and evolution 65
2 In one article, Chomsky (1976) casts aspersions on anyone even raising
the question of the evolutionary origins of language, suggesting that it is no
more or less interesting than those of any other organ, say, the heart. He implies
that the age-old quest for an answer to this question must reflect religious mo-
tives, rather than scientific ones. I find Bickerton's response (1981:315) to be
wholly compelling:
How we first got arms or a heart are questions so phylogenetically
remote and so unrelated to the mental life of our species that
Chomsky is right to dismiss them as not worth asking (except, pre-
sumably, for those whose professional specialism they are). But the
evolution of language is so recent that we may reasonably suppose
that its present nature is still conditioned by those origins, and its
crucial role in distinguishing between us and other species (while any
number of other species have arms and hearts) is such that it must
strongly influence, even if it does not wholly determine, all that we
think and do. Thus, to put the determination of its origins on a par with
the determination of the origins of physical organs seems to me a
piece of evasive perversity.
3 Gould himself (1987) has taken the position that language is not the
product of natural selection, but rather of nonadaptationist mechanisms.
^Chomsky's position is replete with irony. No one has campaigned more
vigorously than he for the autonomy of grammar. Yet his position that language
is, evolutionarily speaking, epiphenomenal, while not logically incompatible
with synchronic autonomy, does lead one to wonder how an epiphenomenon
managed over the course of time to develop its own internal set of principles.
5 In the approach taken in Langacker 1987, conceptual structures are
linearly ordered, and consequently 'fit' with phonetic representations better than
in the model I am describing. However, as I understand Langacker's theory, he
achieves this fit simply by building a great deal of syntactic structure directly into
his conceptual structures.
6 For discussion of syntactic principles, see Chomsky 1981, 1986a,b, Sells
1985. Van Riemsdijk & Williams 1986, and Newmeyer 1986. I leave open the
(for our purposes) irrelevant question of the number of levels of syntactic
structure, i.e., whether the model contains the levels of D-Structure and Logical
Form, as well as S-Structure.
''Pinker and Bloom (1990) discuss at length the adaptive complexity of
language and its consequent implications for evolution. They also address a
great many objections that have been raised to an evolutionary origin for UG. It
is worth summarizing their responses to the two most serious. First, Lieberman
(1984) claims that selection demands alletic variation, and none exists in syn-
66 Studies in the Linguistic Sciences 20:2 (Fall 1990)
tactic abilities. To this point, Pinker and Bloom note that enormous individual di-
fferences in such abilities exist, certain of which plausibly have a genetic basis.
Indeed, there have been demonstrated to be genetically-transmitted syndromes
of grammatical deficits.
Second, Geschwind (1980) argues that 'no hypothetical beneficial gram-
matical mutation could have benefitted its possessor, given that none of the
person's less evolved compatriots could have understood him or her' (Pinker &
Bloom 1990:xxx). To this they reply that comprehension abilities do not have to
be in perfect synchrony with production abilities, a point that they amply illus-
trate with examples of asymmetries between the two.
8 The principle of Subjacency dates from Chomsky 1973 and unifies sev-
eral of the extraction constraints proposed in Ross 1967. In different ways,
Kayne (1984) and Chomsky (1986b) attempt to unify Subjacency and the ECP,
a result which, if correct, has no bearing on the conclusions of this paper.
9 Bickerton (1990) makes the point that an adaptationist account demands
a small number of principles and sketches briefly how UG principles might have
arisen in the course of language evolution.
REFERENCES
BERWICK, Robert, & Amy Weinberg. 1984. The grammatical basis of linguistic
performance. Cambridge, MA: MIT Press.
Bickerton, Derek. 1981. Roots of language. Ann Arbor: Karoma.
1990. Language and species. Chicago: University of Chicago Press.
. 1991. Language origins and evolutionary plausibility. Language and
Communication.
Brandon, Robert, & Norbert Hornstein. 1986. From icons to symbols: some
speculations on the ohgins of language. Biology and Philosophy 1.169-
189.
Chomsky, Noam. 1973. Conditions on transformations. A festschrift for Morris
Halle, ed. by S. Anderson & P. Kiparsky, 81-162. New York: Holt.
1975. Reflections on language. New York: Pantheon.
1976. On the nature of language. Origin and evolution of language and
speech, ed. by S. Hamad, H. Stekelis, & J. Lancaster, 46-55. (= Annals of
the New York Academy of Sciences, 280.) New York: New York Academy
of Sciences, VIM.
. 1981. Lectures on government and binding. Dordrecht: Foris.
1986(a). Knowledge of language. New York: Praeger.
1986(b). Barriers. Cambridge, MA: MIT Press.
Newmeyer: Some issues in language origins and evolution 67
. 1988 (a). Language and problems of knowledge: the Managua lectures.
Cambridge, MA: MIT Press.
— . 1988 (b). Prospects for the study of language and mind. To appear in The
Chomskyan turn, ed. by A. Kasher. Oxford: Blackwell. [Citations are from
a pre-publication manuscript.]
, and Harold Lasnik. 1977. Filters and control. Linguistic Inquiry 8.425-504.
Dawkins, Richard. 1986. The blind watchmaker. New York: Norton.
Deane, Paul. 1988. Which NP's are there unusual possibilities for extraction
from? Papers from the 24th Annual Regional Meeting of the Chicago
Linguistic Society, 100-111.
DuMFORD, Jennie. 1989. Evolution of the brain and language. MS, Arizona
State University.
Falk, Dean. 1980. A reanalysis of the South African Australopithecine natural
endocasts. American Journal of Physical Anthropology 53.525-539.
1983. Cerebral cortices of East African early hominids. Science 221.
1072-1074.
Frazier, Lynn. 1985. Syntactic complexity. Natural language parsing, ed. by
D. Dowty, L. Karttunen, & A. Zwicky, 129-189. Cambridge: University
Press.
Geschwind, Norman. 1980. Some comments on the neurology of language.
Biological studies of mental processes, ed. by D. Caplan, 301-320. Cam-
bridge, MA: MIT Press.
GiVON, Talmy. 1979. On understanding grammar. New York: Academic Press.
Gould, Steven Jay. 1987. The limits of adaptation: Is language a spandrel of
the human brain? Paper presented to the Center for Cognitive Science,
MIT.
Gould, Steven Jay, & R. C. Lewontin. 1979. The spandrels of San Marco and
the Panglossian program: A critique of the adaptationist programme.
Proceedings of the Royal Society of London 205:281-288.
Hale, Kenneth, & Jay Keyser. 1986. Some transitivity alternations in English.
Lexicon working papers #7, Center for Cognitive Science, MIT.
— & — . 1987. A view from the middle. Lexicon working papers #10, Center
for Cognitive Science, MIT.
HOEKSTRA, Teun, & Jan Kooij. 1988. The innateness hypothesis. Explaining
language universals, ed. by J. Hawkins, 31- 55. Cambridge: University
Press.
HuRFORD, James. 1989. Biological evolution of the Saussurean sign as a com-
ponent of the language acquisition device. Lingua 77.187-222.
Jackendoff, Ray. 1983. Semantics and cognition. Cambridge, MA: MIT Press.
Kayne, Richard, 1984. Connectedness and binary branching. Dordrecht: Foris.
KUNO, Susumu. 1987. Functional syntax. Chicago: University of Chicago Press.
Langacker, Ronald. 1987. Foundations of cognitive grammar. Stanford:
Stanford University Press.
LiBERMAN, Alvin. 1974. The specialization of the language hemisphere. The
neurosciences: third study program, ed. by F. Schmitt & F. Worden, 43-56.
Cambridge, MA: MIT Press.
68 Studies in the Linguistic Sciences 20:2 (Fall 1990)
LlEBERMAN, Philip. 1984. The biology and evolution of language. Cambridge,
MA: Harvard University Press.
Mattingly, Ignatius. 1972. Speech cues and sign stimuli. American Scientist
60.327-337.
MONOD, Jacques. 1972. Chance and necessity. London: Collins.
Newmeyer, Frederick. 1986. Linguistic theory in America. 2nd ed. New York:
Academic Press.
1990. Speaker-hearer asymmetry as a factor in language acquisition: A
functional explanation for formal principles of grammar. Papers from the
Aixteenth Annual Meeting of the Berkeley Linguistics Society.
1991. Functional explanation in linguistics and the origins of language.
Language and Communication.
Piattelli-Palmarini, Massimo. 1989. Evolution, selection, and cognition: from
'learning' to parameter setting in biology and in the study of language.
Cognition 31.1-44.
PiERREHUMBERT, Janet, & Mary Beckman. 1988. Japanese tone structure.
Cambridge, MA: MIT Press.
Pinker, Steven, & Paul Bloom. 1990. Natural language and natural selection.
Behavioral and Brain Sciences.
Premack, David. 1976. Intelligence in ape and man. Hillsdale, NJ: Eribaum.
,& Ann Premack. 1983. The mind of an ape. New York: Norton.
Rappaport, Malka, & Beth Levin. 1988. What to do with Q-roles. Syntax and
semantics, 21 : Thematic relations, ed. by W. Wilkins, 7-36.
Ross, John. 1967. Constraints on variables in syntax. MIT Ph. D. dissertation
in Linguistics.
Selkirk, Elisabeth. 1986. On derived domains in sentence phonology. Pho-
nology Yearbook 3.371-405.
Sells, Peter. 1985. Lectures on contemporary syntactic theories. Chicago:
University of Chicago Press.
Thompson, D. W. 1961. On growth and form. [Abridged edition, ed. by J. T.
Bonner.] Cambridge: Cambridge University Press.
Tobias, P. V. 1981. The emergence of man in Africa and beyond. Philo-
sophical Transactions of the Royal Society of London, Series b, 292.43-56.
Van Riemsdijk, Henk, & Edwin Williams. 1986. Introduction to the theory of
grammar. Cambridge, MA: MIT Press.
Wilkins, Wendy, & Jennie Dumford. 1990. In defense of exaptation. Behavi-
oral and Brain Sciences.
& — . In Preparation. The emergence of language: perspectives on
evolution and acquisition.
Studies in the Linguistic Sciences
Volume 20, Number 2. Fall 1990
PHONOLOGY AND ITS INTERACTION WITH SYNTAX AND
MORPHOLOGY
David Odden
(Ohio State University)
1. INTRODUCTION
Linguistic research in the past thirty years has made significant progress
towards understanding the internal properties of the components of a grammar,
especially phonology, syntax, semantics, and to a lesser degree, morphology.
One of the remaining and more thorny problems in understanding the formal
properties of phonology is determining how it interacts with other parts of the
grammar. Starting with work by Mohanan and Kiparsky in the early 1980s, the
theory of Lexical Phonology has emerged as the dominant research paradigm
within which questions about the relationship between phonology, morphology,
and syntax are framed. This paper takes a closer look at the theory of com-
ponent interaction embodied in Lexical Phonology. Although the focus of my
inquiry is the interaction between phonology, syntax, and morphology, and in
particular the supposed ordering relation between morphology and phonology,
a further conclusion emerges from this investigation, namely that much work
remains to be done in order to answer the question 'What is morphology?' In
particular, we will see that most of the evidence cited in support of the in-
teractive view of phonology and morphology involves operations which are not
necessarily part of morphology, and which, given a certain restriction on the
generative power of morphology, is necessarily not part of morphology.
The principles of Lexical Phonology are generally summarized in the
standard graphic metaphor (2) [next page], which is adapted minimally from
Kiparsky 1982. The core principles of the theory appear to be those in (1).
(1) a. There is a construct 'level', common to phonology and
morphology.
b. Levels in phonology are the same as levels in morphology.
c. There are significant formal differences between lexical and
postlexical phonology.
70
Studies in the Linguistic Sciences 20:2 (Fall 1990)
Cyclic application of phonological rules derives from the inter-
action between lexical phonology and morphology as charac-
terized above.
Lexical phonology has no access to the output of the syntax.
Lexical phonology and morphology interact so that morphology
has access to phonological properties derived by applying
phonological rules on some earlier level
(2)
LEXICON
I level 1 morphology | t
level 2 morphology
level n morphology
underived lexical items
level 1 phonology
level 2 phonology
level n phonology
syntax
postlexical phonology~|
Claims (1a,b) recognize that in morphology and phonology, there are
domain restrictions on the application of rules, and (lb) in particular says that
the domain restrictions in morphology have the same basis as those in phono-
logy; that is, rules in morphology and phonology are assigned to levels, and in
so far as levels are properly a property of the whole lexicon (which includes
morphology and phonology), level ordering is supposedly uniform between
these components.
Claim (1c), that there is a distinction between lexical and postlexical
phonology, recognizes, and tries to make more systematic, the observation that
there seem to be two distinct kinds of phonological rules. A partial list of proper-
ties often cited as distinguishing the two kinds of rules is given in (3).
(3) Lexical Rules
May be sensitive to morphological or lexical properties
May apply word-internal cyclically
Obey the Strict Cycle condition
Cannot see the output of syntax
Are structure preserving
Precede all postlexical rules
Odden: Phonology and its interaction with syntax and morphology 71
Postlexical Rules
Follow all lexical rules
Have access to the output of syntax
Are Insensitive to morphological or lexical properties
Do not apply cyclically
Violate the Strict Cycle condition
Problems with these criteria for component assignment are not hard to find.
External sandhi rules can have exceptions or morphological conditions, be
structure preserving, and apply in a phrasal-cyclic manner; lexical rules can
violate structure preservation. I adopt the distinction between lexical and post-
lexical rules here, with no further commitment to the distinction. Claim (Id), that
cyclic behavior can be derived from a phonology ~ morphology interaction, is a
theory-internal claim, and other models can derive cyclicity without interaction.
Two other claims are implicit in the model (2), especially in the way that
components are placed in nested boxes connected by single or double headed
arrows. These claims, (1e) and (If) are the claims to be focused on, since they
represent the Lexical Phonology theory of component organization. The litera-
ture of Lexical Phonology has never given an explicit formal interpretation to
displays like (2), so it is difficult to know how they can be tested.
Consider the fact that in (2), the boxes called 'level 1 phonology' through
'level n phonology', and the boxes called 'level 1 morphology' through 'level n
morphology' are contained in a larger box called 'lexicon'. One interpretation of
such structures is that subcomponents contained in the same box have shared
characteristics, which those outside the box do not have. It is quite unclear what
the shared formal properties of morphology and lexical phonology are, except
that in the theory of Lexical Phonology, these components define the lexicon. In
fact, taking displays like (2) to be a claim about similarity in formal properties,
we would conclude that lexical phonology and morphology are more similar
than lexical and postlexical phonology. We might even conclude that lexical
and postlexical phonology have no shared properties, since they share no box.
This is clearly absurd, and other arrangements of the boxes have been pro-
posed, for instance in Kaisse & Shaw 1985, where boxes overlap, or Mohanan
1986, where phonology and morphology do not even share a box; cf. (4) below.
If display (2) has a meaning, it cannot be a claim about similarity of components.
The only plausible interpretation of these structures is as a claim about
ordering. Thus, lexical phonology precedes syntax, and postlexical phonology
follows syntax. What does it mean for one component to 'precede' another?
Empirically, we have no evidence for the real-time interpretation in milliseconds
of component ordering, so claims about ordering can only be tested if seen as
claims about the information available to a given module. If module M 'follows'
module L and 'precedes' module N, then M has access to the results of opera-
tions defined in L, but not to the results of operations defined in N. If lexical
72 Studies in the Linguistic Sciences 20:2 (Fall 1990)
(4) Kaisse and Shaw 1985
Mohanan 1986
LEXICON
PHONOLOGY
underlying
representation
lexical
representation
syntactico-
phonological
representation
morphemes
i
words
POSTLEXICAL MODULE
1
i
POST
phonological
phrases
SYNTACTIC MODULE
X ^
1
phonetic
representation
output of
phonology
Odden: Phonology and its interaction with syntax and morphology
73
phonology precedes syntax, information provided by syntax should not be avail-
able to lexical phonology. No rule of lexical phonology should have access to
properties of other words in the sentence, or to any fact about the syntactic
structure of the sentence which the word appears in. Assuming some criterion
for deciding whether a rule is lexical, this might entail that certain types of langu-
ages will never exist. If these predictions are correct, and providing that the pre-
dictions follow from the theory, we would have a reason to assume the Lexical
Phonology model. Similarly, interaction between phonology and morphology
predicts that we might find rules in the morphology which are sensitive to a pho-
nological property that results by applying an earlier phonological rule. It is
surprising for such a fundamental prediction that little evidence for such inter-
actions between phonology and morphology has been brought forth. I consider
such examples later.
(5)
SYNTAX
MORPHOLOGY
i
Lexicon of roots |
i
1 level 1 1
i
1 level 2 1
i
1 level n |
PHONOUOGY
Lexical Phonology
1 level
1
i 1
1 level
2
i 1
1 level
n
1
Postlexical Phonology
74 Studies in the Linguistic Sciences 20:2 (Fall 1990)
To bring into sharper focus the predictions of Lexical Phonology, an alter-
native organization of components is defended here, which can also be repre-
sented by a graphic metaphor, the one in (5) above.
What is most important in this theory, especially for distinguishing it from
Lexical Phonology, is the relative ordering and noninteraction of components.
The level-related claims (1a,b,c) have simply been retained from Lexical Pho-
nology, ^ though the mechanisms for generating the behavior are not the same.
The noninteractive theory also assumes, as does Lexical Phonology, the prim-
itive notion of levels, LI to Ln. The putative identity of morphological and pho-
nological levels results from an assumption within this theory about how levels
in phonology are defined.
Each morphological rule is encoded for the level where it applies. Word
construction starts by selecting a root, and morphological rules apply to this
structure, concatenating material with it. Morphological rules provide labeled
bracketing as well as the segmental content of its affix, hence take the form (6).
(6) [xY] ^[z[xY]W] [xY] -> [zW[xY]]
The labels attached to these structures indicate the level on which the operation
applies. In (7) we see the structure of Maltese Arabic hatfitkums 'she didn't
snatch you', composed of the Level 1 root hataf ar\6 the subject marker it, and
the Level 2 object suffix kum and the negative s.
(7) L1 Insertion of root
I
hataf
Suffixation of subject marker
Suffixation of object marker
LI
/ \
/
\
LI
\
\
hataf
it
L2
/ \
/ \
LI
\
/
\
\
/
\
\
LI
\
\
1
\
\
hataf
it
kum
Odden: Phonology and its interaction with syntax and morphology
L2 Suffixation of negative
/ \
/ \
L2 \
/ \ \
/ \ \
L1 \ \
/ \ \ \
/ \ \ \
LI \ \ \
hataf it kum s 'she didn't snatch you'
75
The only difference between this and standard lexical phonology is the in-
clusion of labels on the brackets which indicate distinctions of level.
In the phonological component, rules are encoded for their domain of
application, specifically the lowest and highest-numbered level where the rule
may apply. This is the same as specifying, for instance, that such and such a
rule is 'in' Level 1 phonology. The highest constituent dominated by LI be-
comes the initial input to the phonology, and phonological rules encoded for
application at LI apply to this string. Thus the boxed constituent in the first step
of (8) is the domain where Level 1 rules apply. After the last rule defined at LI
applies, in this case Apocope, the highest constituent dominated by L2 be-
comes the string subject to phonological rules, and rules encoded for ap-
plication at L2 apply to this string. This continues to the last lexical level and in-
to postlexical phonology. In this way, we keep cyclicity and level ordering, and
do so with essentially the same stipulations as are required for the Lexical
Phonology derivation of the cycle, without interleaving phonology and mor-
phology.
(8) L2
/ \
/ \
L2 \
/ \ \
LI phonQlpgy \ \
LI
/ \
/ \
LI
\
1
\
hataf
it
\ \
\ \
\ \
\ \
\ \^
kum s
Stress
76
Studies in the Linguistic Sciences 20:2 (Fall 1990)
L2
/ \
/ \
L2 \
/ \ \
L1 phonology \ \
LI
/ \
/ \
LI \
I \
\ \
\ \
\ \
\ \
\ \^
kum s
Apocope
1-2 phpnologY
L2
/ \
/ \
L2 \
/ \ \
/ \ \
LI \ \
/ \ \ \
/ \ \ \
LI \ \ \
I \ \ ^
hat! il kum i
Stress
With two competing models, we can now look to see where they make dif-
ferent predictions and check which predictions best match the facts. First, con-
sider syntax ~ phonology interaction. Lexical Phonology predicts that we will
never find a rule of the lexical phonology which can see the output of syntax,
whereas the noninteractive model claims that we should find such rules.
Deciding between these two models should be simple: We look lo see if rules
of lexical phonology ever can see the output of syntax. Now, suppose that after
a search for such rules, we fail to find such counterexamples. One could take
this as a refutation of the noninteractive theory, or we could fix the theory up a
bit, by stipulating that lexical rules cannot refer to material outside the word.
A proponent of Lexical Phonology could rightly object that one should not
handle with stipulation a fact that is explained in Lexical Phonology by the ar-
chitecture of the theory. This supposed gap in the set of attested rules, then,
seems to be the argument for ordering syntax after phonology. However, the
component ordering of (2) also predicts that information from lexical phonology
could be available to syntax, and such interactions are simply never found. The
architecture of the noninteractive model correctly makes this prediction, where-
as Lexical Phonology has to stipulate this as an independent principle. The box
model of Kaisse & Shaw 1985 happens to avoid this unfortunate conclusion by
Odden: Phonology and its interaction with syntax and morphology 77
running syntax in parallel with phonology and merging their outputs in the
postlexical phonology. What is interesting is that nobody, including Kaisse and
Shaw, seems to have noticed this bizarre prediction of Lexical Phonology, and
to the best of my knowledge, the Kaisse & Shaw model has not been explicitly
adopted by any researcher in Lexical Phonology in the past five years.
The models in (5) on the one hand, and (2) or the Kaisse & Shaw alterna-
tive on the other, thus part company over issues of access to information. The
noninteractive model claims that the rules of lexical phonology do have access
to information from syntax, and it also disallows morphology access to informa-
tion made available by phonological rules. These two issues are separable —
one COULD have an interactive model of phonology and morphology, and re-
order syntax relative to lexical phonology. Or one could retain the ordering of
phonology and morphology relative to syntax, but reject the interactive aspect of
Lexical Phonology. We will thus consider these claims separately.
In this paper, I will argue for two conclusions regarding what languages
DO. First, I show that systems exist with the properties which Lexical Phonology
predicts should not exist, namely lexical rules accessing the output of syntax.
Second, I argue that there are NO compelling cases of rules of morphology ap-
plying after phonology, so there remains a conspicuous lacuna in the evidence
for Lexical Phonology. Taken together, this argues for a noninteractive model
over model (2).
2. SYNTAX AND PHONOLOGY
The first problem I will look into is the ordering between lexical phonology
and syntax. The empirical problem is that some rules of Kimatuumbi phonology
must be lexical, but they also have access to syntactic structure and phono-
logical properties of surrounding words. The data and issues involved here are
also discussed in Odden 1990 and Hayes 1990. A few rules of Kimatuumbi
show the nature of the paradox. The first rule is the phrase-level rule Shorten-
ing, which shortens a long vowel if it is the head of a phrase and is followed by
material within its phrase.
(9) ki^koloombe 'cleaning shell'
ki^kolombe chaangu 'my cleaning shell'
ki^tuumbili 'monkey'
kijumbili ywaawi^i^le 'monkey who died'
naakijtweeti^ 'I took if
naaki^-tweti^ki^koloombe 'I took a cleaning shell'
Since this rule involves multiple words and syntactic structures, in Lexical Pho-
nology the rule has to be postlexical.
78 Studies in the Linguistic Sciences 20:2 (Fall 1990)
(10) Shortening
o
X\
(Y contains phonetic material)
^ / [x- [x x] Y x]
The second rule is Glide Formation, a lexical rule which desyliabifies a
prevocalic high vowel and compensatorily lengthens the following vowel.
(11) kijkalaango 'frying pan"
ky-gQIa 'frog' (cf. kaQla 'little frog')
ijkalaango 'frying pans'
y-gijla 'frogs*
mg-ki^kalaango 'in the frying pan'
mij-yijijla 'in the frogs'
mw-ij^kalaango 'in the frying pans'
tg-teli^^ke 'we cooked'
tw-eekije 'we laughed' (cf. eka 'to laugh')
We must restrict GF so that a prevocalic long vowel does not desyllabify, to ac-
count for (12). Vowel length which blocks GF may arise from applying GF, as
mwij-qte shows, indicating that the rule iterates from left to right.
(12) miji^-ate "in the banana hands'
(cf. mwaanjij 'in the firewood', from /mg-aanjQ/)
mwij^-ijte 'you should pull them (CI. 9)'
(cf. bayijtjte 'they pulled them', from /ba-i^-i^te/)
We will see that GF is a lexical rule, but paradoxically, applies after the
sandhi rule Shortening. The next step is to show that GF is lexical: The prob-
lem is that there is only one iron-clad principle that forces a rule into the lexical
component, as opposed to being postlexical, and that is a demonstration that
the rule is cyclic. The supposed correlation between lexicality and reference to
morphological or lexical features does not follow from the architecture of Lexical
Phonology, so a demonstration that it fails to hold does not refute the theory.
The only thing that follows from the nature of Lexical Phonology is the cor-
relation between cyclicity and lexicality, since the cycle itself is assumed to
derive from the phonology ~ morphology interaction in the lexicon. The Glide
Formation rule can be shown to apply cyclically, and in particular the domain of
cycling is the word-internal lexical level. Therefore GF must be a lexical rule.
One demonstration of the cyclicity of GF consists in the contrasting deriva-
tions of mwij-^te in (13a) with the Level 2 prefixes rrm and (vs. mi^yi^Qla in
(13b), with the Level 3 prefix mq and the Level 2 prefix (. We find the same
sequence of vocalic phonemes in both cases, but determining which vowel
Odden: Phonology and its interaction with syntax and morphology 79
becomes a glide requires knowing the level at which the morpheme is made
available.
(13)
[ my-i^-yte ]
my-i^-yte
Input to L2
mwi^i^ute
Glide Formation
'you should pull if
[mg[L-gia]]
t-Qla
Input to L2
yygia
Glide Formation
myyijgia
Input to L3
NA
Glide Formation
'in the frogs'
In mwix-ijte, the vowel sequence q+ (appears entirely in Level 2, so the first of
the prevocalic high vowels undergoes GF, giving the phonetic form. In mi^-
yijijla, mtj becomes available at Level 3, but /' is available at Level 2, so (
undergoes GF since it is the only vowel encountered at this stage. Therefore
GF must be lexical.
Now we consider the ordering of Shortening and GF. The supposedly
postlexical rule Shortening precedes GF, since Shortening does not apply to
the long vowel which arises as a result of applying GF, at least at Level 2.
(14) mi^-ake l[ -^ mwaake l[ 'you should not hunt'
(*mwake IQ
kijQIa chaangu -> kygQIa chaangu 'my frog'
The vowel combination undergoes GF, which lengthens the stem vowel, but the
derived long vowel is not shortened in the presence of a modifier. This requires
counterfeeding between GF and Shortening, which means that a postlexical
rule applies before a lexical rule.
This causes a serious problem for Lexical Phonology, but is allowed in the
theory proposed here, since there is no theoretical impediment to making Short-
ening a lexical rule. It is important to bear in mind that the previous data provide
examples of GF at Levels 2 and 3. GF also applies at Level 1 , the stem level, as
seen in (15), where the short vowel of the reciprocal suffix -an- lengthens as a
result of desyllabifying the suffix -(-.
(15) ak-a 'to net-hunt'
ak-an-a 'to net-hunt each other*
ak-y-aan-a 'to net-hunt for each other' (/ak-i^an-a/)
80 Studies in the Linguistic Sciences 20:2 (Fall 1990)
Interestingly, long vowels which arise by GF at Level 1 do undergo Short-
ening, in contrast to long vowels so derived at Levels 2 and 3. As can be seen
in (16), GF is applicable on two levels, namely in Level 1 to (+ a, and in Level 2
to V + a. As shown by the surface fornn twaakyana ijuumbili, the long vowel
derived by GF at Level 1 shortens, but the long vowel derived at Level 2 does
not.
(16) [tij-[ak-ijan-a]] ijuumbili
-> twaakyana ijuumbili "we net-hunt monkeys for each other*
This compounds the paradox for Lexical Phonology. On the one hand. Level 1
GF has to feed Shortening, and on the other hand Level 2 GF has to counter-
feed Shortening. The solution is simple if Shortening is in the lexical phono-
logy: As spelled out in (17), Shortening follows GF, and only applies at Level 1.
(17) ak-i^-an-a ijuumbili Input to Level 1
akyaana ijuumbili Glide Formation
akyana ijuumbili Shortening
ty-akyana ijuumbili Input to Level 2
twaakyana ijuumbili Glide Formation
For this to be possible. Shortening must be lexical, which means that rules of
the lexical component must have access to the output of syntax.
Other rules exhibit similar properties. One of these, Initial Tone Insertion
(ITI), will be considered since it is relevant to section 3. This rule, illustrated in
(18), assigns a H tone to the initial vowel of a lexical class of morphemes, as
long as the morpheme is preceded by a word bearing no stem H.
(18) ki^ndolo cha Mamboondo 'sweet potato of Mamboondo'
mpynga wa Mamboondo 'rice of Mamboondo'
mabigii^ga-bili 'two beer-brewing areas'
mijomondo y(.-bili 'two ntomoondo trees'
aat(.belekwa ky-Kij)o6i^ 'he was born in Kipooi'
abelekijwe kQ-Kij)o6i^ 'he was born in Kipooi'
An important condition on the triggering element seen in (19) is that, while
a stem H in the preceding noun blocks the rule, a prefixal H does not.
(19) ki^wikilyo gan^_ 'what type of cover?'
kijumbili gan(_ 'what type of monkey?'
k(^ng'ombe g^n^, 'what type of cows?*
Thus the stem H in kijumbili blocks assignment of H to ganl^, but the prefix H in
kijig'oombe does not.
Odden: Phonology and its interaction with syntax and morphology 81
The fact that the rule is lexically conditioned and is sensitive to the stem ~
prefix distinction suggests (though it does not prove) that it should be lexical.
An ordering argument cinches this conclusion. There is a restriction on GF in
Kimatuumbi, which is that a H-toned vowel cannot glide before a long vowel,
although a L-toned vowel can. Word-internal examples of this condition are
seen in (20), where the prevocalic vowels (and t^ do not glide before a long
vowel.
(20) chatgoondijte 'what we peeled'
pan[aandi,i^ke 'when I wrote'
Now consider the data in (21). with the prefix kij before a long vowel.
(21) ijtij[kgaanjij -^ ijtij(_ kwaanjQ 'you should run to the firewood'
utiji^ kgaanjQ -^ gtiji^ kgaanjQ 'you ran to the firewood'
A H can be assigned to ki^- by ITI, which affects whether GF can apply before a
long vowel: If the prefix has a derived H (the second example), it cannot
undergo GF. This shows that ITI applies before GF. Since GF is lexical, ITI must
also be lexical, which creates another paradox for Lexical Phonology.
The noninteractive model has no problem with this, since in that theory, the
output of the syntax is fully available to lexical phonology, and therefore ITI can
be lexical. While Lexical Phonology is not equipped to handle this problem, we
might consider whether the data could be handled in a theory that retains the
supposed ordering between syntax and lexical phonology. Hayes (1990), dis-
cussing similar problems, including this Shortening ~ Glide Formation paradox,
proposes a modification of phonology, namely precompilation theory. In pre-
compilation theory, a word may have multiple derivations in the lexical phono-
logy, hence multiple outputs will emerge from the lexicon for each word. Each
of these derivations is tagged for a diacritic referred to as a lexical instantiation
frame. Languages may define sets of instantiation frames which serve as con-
text for lexical phonological rules. Upon entering the postlexical phonology, the
frame definitions of the language are consulted, and the syntactic, morpho-
logical, and phonological properties of the word in its sentence are checked.
Out of the various derivations generated in the lexicon, the correct form is then
inserted into the sentence and the string is submitted to the postlexical phono-
logy. Precompiled Shortening is formulated as in (22).
(22) VV^V/[... ...] [Frame 11
The definition of Frame 1 is given as (23):
(23) Framel: U- ... [x ]Y] Y^0
82 Studies in the Linguistic Sciences 20:2 (Fall 1990)
The two derivations in (24) are then generated. The derivation where Frame 1
rules apply generates the Frame 1 allomorph, and the other one, where these
rules do not apply, generates the elsewhere form.
(24) [kyaandangyo chaangu] 'my forest farm'
/andaangyo/
[+F1] ^.,^^^^[-F1l
andangyo NA Shortening
ki^-andangyo ki^andaangyo Level 2 prefixing
kyaandangyo kyaandaangyo Output of lexical phonology
Later, the appropriate allomorph is selected. If the word in question appears in
the context defined in (23), the Frame 1 allomorph is inserted; otherwise the
'elsewhere' allomorph is inserted.
Frame definitions may include phonological information form neighboring
words. Kimatuumbi ITI could be stated to apply in the context Frame 2:
(25) Initial Tone Insertion
V ^ V / [ (Frame 2]
H
Frame 2 will then be defined as in (26).
(26) Frame 2: [stem ~ H ] X (X does not contain s] )
Precompilation thus allows lexical rules to indirectly see into surrounding
words, without DIRECTLY looking at surrounding words, since the rule itself only
refers to the mediating Frame diacritic. In other words, the rules freely generate
all sorts of forms, then later filter out the incorrect ones.
Although standard Lexical Phonology does not include precompilation,
this new machinery is not grossly inconsistent with the architecture of Lexical
Phonology, and since Lexical Phonology alone cannot handle Kimatuumbi, it
must adopt a subtheory of precompilation to achieve observational adequacy.
Both Extended Lexical Phonology and the noninteractive theory can generate
the correct forms, and since precompilation may include any information about
the syntax, morphology, and phonology of surrounding words, precompilation
generates the same forms as are generated by the theory which allows lexical
phonology direct reference to structures outside the word. Precompilation is
thus a virtual notational variant of the direct-access theory.
Odden: Phonology and its interaction with syntax and morphology 83
If the theories are weakly equivalent, we must turn to secondary considera-
tions such as computational complexity to evaluate the theories, where we find
an advantage to allowing lexical rules to directly see the output of syntax. In
precompilation theory, there can be multiple frames, and frames can overlap, so
each frame definition doubles the number of derivations necessary for a form.
For instance, if there are two frames defined in a language, then four derivations
are required for a word, one for Frame 1 , one for Frame 2, one for Frames 1 and
2, and one for the elsewhere form. In general, when there are n frames, we
need 2" derivations. On the other hand, in the theory ordering lexical phono-
logy after syntax, only a single derivation is required since the rules simply
inspect the word-external context to determine whether their conditions for
application are satisfied. Precompilation theory thus entails more complex com-
putations.
3. MORPHOLOGY AND PHONOLOGY
We now turn to the second issue of component interaction, namely the
fundamental question in morphology ~ phonology interaction. This is the ques-
tion whether, as predicted by the interactive theory, morphology can ever take
as its input the output of the phonological component. I will argue that morpho-
logy exclusively precedes phonology, but to illustrate the kind of evidence we
would look for to decide this issue, I begin by investigating a new case where
morphology seems to follow phonology.
3.1 ARABIC YA
The problem involves the phonology of glides in Classical Arabic (Brame
1970), and an allomorph of the first singular possessive suffix. A basic fact a-
bout glides in Classical Arabic is that they are phonologically 'weak' and often
elide. The important generalization for our purposes is that intervocalic glides
delete, and the resulting vowel cluster fuses into a single long vowel. Since
Classical Arabic does not allow long vowels in closed syllables, this long vowel
may then shorten. The left column of (27) gives a verb which suffers no alterna-
tions. The other two columns illustrate stems with final y and w which delete
intervocalically.
(27) Sound verb Final y Final w
qatalna ramayna da^awna 1p.
qatala rama da9a Ss.m.
qatalat ramat da^at Ss.f.
'kill' 'throw' 'call'
Three rules are responsible for these alternations, namely Glide Elision which
deletes intervocalic glides, Vowel Fusion which fuses vowel clusters into one
long vowel, and Closed Syllable Shortening.
Glide Elision
C
-con
. +son .
Vowel Fusion
CT CT
^
"-.==
V V
.'-^+
P
/ V
Closed Syllable Shortening
a
V V c
The form [rami] derives from /ramay-a/ via Glide Elision and Vowel Fusion, and
[ramat] derives from /ramayat/ by the further step of Shortening. The rules Glide
Elision and Vowel Fusion are the two crucial rules involved in this case of
phonology supposedly preceding morphology.
The relevant allomorphy is that of the first singular possessive suffix on
nouns. As we see in (29), nouns are composed of a stem plus a case ending, u,
i, or a:
(29) 7al-kitab-u 'the book (nominative)'
7al-kitab-i 'the book (genitive)'
7al-kitab-a 'the book (accusative)'
A possessive suffix such as ha, ka, or na may be added and stands after the
case ending, as (30) shows.
(30) kitab-u-ha 'her book (nom.)'
kitab-u-ka 'your (s.m.) book (nom.)'
kitab-i-ka 'your (s.m.) book (gen.)'
kitab-a-na 'our (s.m.) book (ace.)'
In (31) are nouns with the first singular suffix -/'-. This suffix combines with the
case marker in such a way that the case marker is completely lost.
Odden: Phonology and its interacrion with syntax and morphology 85
(31) "
kitab-r
•my book (nom)'
/kitab-u-F/
kitab-r
'my book (gen)'
/kitab-i-i/
kitab-i
'my book (ace)'
/kitab-a-F/
The first singular suffix is subject to phonologically conditioned allomorphy;
if it stands immediately after a long vowel or diphthong, it becomes ya.
(32) Noun Noun + 1 s. poss.
7ulam-a-ni 'slaves (n. dual)' 7ulam-a-ya
-yuiam-ay-ni 'slaves (a./g. dual)' yulam-ay-ya
mu9allim-u-na 'teachers (n. pi.)' muTallim-u-ya
mu^allim-F-na 'teachers (a./g. pi.)' mu9allim-F-ya
In these examples, the long vowel or diphthong which conditions ya is present
in underlying representation. Other examples, seen in (33), show that long
vowels which dehve by phonological rule, in particular Glide Elision and Vowel
Fusion, also trigger the ya allomorph of the first singular (1 s.) suffix.
(33) 7al-qahw-at-u 'the coffee (nom.)'
ma-qha-n 'a coffee house (nom.)' /ma-qhaw-un/
7al-ma-qha 'the coffee house (nom.)' /7al-ma-qhaw-u/
ma-qha-ya 'my coffee house (nom.)' /ma-qhaw-u-1 s./
7al-haway-at-u "the hobby (nom.)'
7al-hawa 'the desire (nom.)' /7al-haway-u/
hawa-ya 'my desire (nom.)' /haway-u-1 s./
qaday-tu 'I settled'
qadin "a judge (nom.)' /qadiy-un/
7al-qadF 'the judge (nom.)' /7al-qadiy-u/
qadF-ya 'my judge' /qadiy-u-1 s./
The stem for 'coffee' ends in the glide w, as shown by ?al-qahw-at-u. In-
tervocalically, in ?almaqha, the glide deletes before the case ending and the
vowel sequence fuses into a long vowel. This derived long vowel then con-
ditions the ya allomorph. As the derivation in (34) indicates, this seems to show
that the first singular allomorph is determined after Glide Elision and Syllable
Fusion.
(34) /maqhaw-u/ Underlying
maqhau Glide Elision
maqha Vowel Fusion
maqhaya Affixation of 1 s.
We must apparently first apply phonological rules, which gives us an inter-
mediate base for affixing a pronoun, and then the morphology selects the ap-
propriate allomorph for affixation to this phonologically derived form. Such a
86 Studies in the Linguistic Sciences 20:2 (Fall 1990)
derivation would seem to be a problem for the noninteractive model, which
requires all morphemes to be concatenated before any phonological rules ap-
ply. However, I will show that this case can be handled in the noninteractive
theory.
3.2 DISTINGUISHING PHONOLOGY AND MORPHOLOGY
It is appropriate to reconsider this and similar cases where morphology
supposedly follows phonology. A search of the literature reveals various ex-
amples which have been cited as cases of phonological rules applying before
morphology. The largest class is represented in (35).
(35) Overapplication of phonological rules under reduplication (Kihehe,
Tagalog, Javanese). In Kihehe, the stem reduplicates {kutele-
kateleka), excluding the prefixes. Rules of syllable fusion draw
prefix material into the stem, causing it to be reduplicated (kwiita-
kwiita).
The imperative in Danish is formed by deleting a -a suffix. Deletion
follows a vowel lengthening rule, so /baeSa/ becomes bae.da
(which is the infinitive), then the -a suffix is deleted in the impera-
tive, giving [bae:9]. Similar rules, deleting the agreement mor-
pheme y in Abkhaz, and the verb suffix a in Icelandic, have been
found.
The largest subclass includes overapplication of phonological rules under re-
duplication, as in Kihehe. The second largest subclass is typified by imperative
formation in Danish, which deletes the -a suffix of the infinitive. Imperative De-
letion has to be ordered after a phonological rule of open syllable vowel leng-
thening. This case has been cited in Hargus 1985 as exemplification of the pre-
dicted feeding from phonology into morphology.
There is little doubt that the phonological rules which supposedly precede
morphology in these cases are indeed phonological rules. What is not clear is
that the supposedly morphological operations are part of morphology. Many of
these cases are post-phonological subtractions, as in Danish Imperatives. In
light of the rule ordering facts, the only possible analysis in the noninteractive
model is phonological deletion.
(36) Danish Imperative Truncation
9^0/ w]
[IMPER]
I will adopt the analysis of Anderson (1975) that the imperative is based on a
form identical to the infinitive.
Odden: Phonology and its interaction with syntax and morphology 87
Nothing in the generally accepted and motivated theory of phonology, be it
Lexical Phonology or non-lexical theories, precludes (36) from the phonology.
Thus Danish Imperative Truncation could be a phonological rule, with a mor-
phological condition, as in (36). Or, it could be expressed as a morpheme-dele-
tion rule and be part of the morphological component, as in (37).
(37) [INFIN]->0 / w]
[IMPER]
If one allows deletion of phonological units in the morphology, not just deletion
of morphemes, one could even assume a rule with the formal statement (36),
but place the rule in the morphological component. With no further conditions
on phonology or morphology, three analyses are possible in Lexical Phono-
logy, and they cannot be distinguished empirically or on the basis of rule ele-
gance. If such rules are phonological rules with morphological conditions, and
not rules of morphology, then they do not show that phonology feeds into mor-
phology.
A basic stumbling block in resolving the relation between phonology and
morphology is this analytic ambiguity, and in particular the fact that many
theories allow one to consign phonological processes to the phonology or the
morphology rather willy-nilly. We must make clear what we mean by 'rule of
morphology'. My claim, and the claim which must be made in Lexical Phono-
logy if there is content to the claim for interleaving of phonology and mor-
phology, is that a 'mle of morphology' is a mle in the morphological component.
Putting Danish Imperative Truncation in the phonology but calling it a 'rule of
morphology' simply because it applies in a morphologically defined context tri-
vializes the difference between phonology and morphology. By analogous rea-
soning, we should call the Kimatuumbi rule Shortening a rule of syntax because
it applies in a syntactically-defined context.
The systematic uncertainty about what constitutes a possible mor-
phological rule surely needs a principled resolution; so to attack the problem
from the side of morphology, principle (38) is proposed.
(38) Principle of morphology ~ Phonology Segregation
The only operation allowed in morphology is concatenation
This segregation of morphological and phonological operations makes
strong claims about morphology. From the standpoint of 'pure' morphology, we
disallow operations such as deletion, movement, or fusion — in short, we deny
morphology the power of unrestricted rewrite rules. A further consequence of
this principle is to rule out on theoretical grounds supposed cases of 'process
morphology' (Matthews 1974, etc.), where morphology performs phonological
changes. Examples of process morphology must generally be reanalyzed as
two steps, namely purely morphological concatenation, plus a phonological
88 Studies in the Linguistic Sciences 20:2 (FaU 1990)
rule. For instance, the morphology of German Umlaut is affixation. It happens
that the phonological content of certain affixes contains a floating vowel feature
[-back], which the phonology is responsible for linking to the appropriate vowel.
Other processes such as 'subtractive morphology" are to be formalized as pho-
nological deletion rules, constrained to apply in specified morphological con-
texts.
Anderson (1988:150) argues that allomorphy cannot be reduced to oper-
ations in phonology, in opposition to the view set forth here. Three putative dif-
ferences are claimed to exist between 'phonological' and 'morpholexical' rules,
which might be taken as evidence for assigning process morphology and purely
concatenative morphology to one component, and strictly phonologically
conditioned rules to another. First, rules of feature exchange (aF -> -aF) are
claimed to be morpholexical. Second, variables in rules are supposedly inter-
preted disjunctively if the rule is phonological, but conjunctively if the rule is
morpholexical. Finally, disjunctive ordering in phonological rules is claimed to
be governed by the Elsewhere Condition, but disjunctive ordering among mor-
phological rules is governed by separate principles. These three arguments
dissolve on closer scrutiny.
Neither Anderson 1988 nor Anderson 1975 provide any evidence that vari-
ables in phonology receive different interpretations depending on whether rules
are morphologically conditioned or purely phonological, and the literature on
variables (Odden 1980, Jensen & Stong-Jensen 1979, inter alia) gives no sup-
port to the claim. More to the point, there is no evidence that variables are
needed in phonological rules. The complete elimination of variables has cer-
tainly been one of the primary goals, and most satisfying successes, of non-
linear phonological theories.
Similarly, there is little direct evidence that disjunctive application is need-
ed for phonological rules, whether governed by the Elsewhere Condition (EC)
or any other condition. The most compelling examples originally cited for the
EC are those found in Diola Fogny and Finnish, which have place-assimilation
rules that seem to precede but not feed consonant weakening processes. So,
in Diola Fogny, /bajum to/ surfaces as [bajunto], and does not undergo a rule
deleting preconsonantal nasals. However, multilinear phonological repre-
sentations render the EC explanation of failure of nasal deletion otiose, since
the nasal segment shares places features with the following stop, and, as a
partial geminate, cannot undergo nasal deletion. Most of the remaining evid-
ence involves rules of stress placement. As pointed out in Howard 1975, even
in the linear framework within which EC was offered, the facts of stress have an
independent explanation. Certainly, metrical accounts of stress render the EC
irrelevant.2
Finally, the evidence that exchange rules even exist, much less have a
correlation with morphological conditioning, is extremely weak. Anderson and
Odden: Phonology and its interaction with syntax and morphology 89
Browne (1973) mention four exchange processes. One is the interchange of a
and between perfective and imperfective in 'some forms of Arabic.'^ (457-8) A
second Is voicing exchange to pluralize nouns in Luo. The remaining three in-
volve exchange of vowel length in certain morphological categohes in Dinka,
Diegueno, and Czech. The latter three examples are irrelevant in the light of
the current understanding of vowel length as not being a feature, but a structural
relation. The fundamental failure in the argument is that there is no reason to
believe that the changes come about by phonological rules (of any variety), or
that a single rule is involved. A plausible reanalysis of such cases would com-
bine elements of the well-known autosegmental analysis of mutations in Fula,
where floating features are affixed, and the templatic approach to Arabic plurals,
where alternative prosodic templates are affixed.
Anderson and Browne (1973) admit there are many ways to form plurals in
Dinka: They state that There is a substantial class of nouns with long vowels in
the singular, which form their plurals by shortening this vowel,' and that 'there is
another class of nouns with short stem vowels which form their plurals by leng-
thening the vowel,' so 'there is a rule which is involved in plural formation which
exchanges long and short vowels.' (459-60) It is a total non sequitur to leap
from the fact that some nouns shorten vowels and some lengthen vowels to the
conclusion that there is a length-exchange rule. The Dinka facts are vastly
more complex: Denning (1987) shows that pluralization in Dinka can be ac-
complished by seven different processes, including total suppletion {weri cow' :
yok 'cows'), voice quality"* (kal 'fence' : kal 'fences'), vowel lengthening (gol
'clan' : g65l 'clans'), shortening (ibdd/< 'animal skin' : bo/c 'animal skins'), vowel
height change (dpk 'boy' : dak), diphthongization (d/f 'bird' : diet 'birds'), tone
{Iwet'We' : lwet_ 'lies'), etc. More commonly, plurals are formed by rather ran-
dom combinations of these processes, as in diir 'cricket' : d/r 'crickets', nar]
'crocodile' : n£rj 'crocodiles', alweet'War' : alwet'Wars', a77aiv'cat' : ari set' cats'.
There can be variation in which different processes are selected to signal the
plural, hence pvyor 'waterbuck' : plurals piar or pwoor. By selecting a subset of
the entire data, one might think that exchange of length is involved, but a com-
plete account of Dinka plurals does not support any exchange rules.
The evidence for an exchange rule in Dholuo is also weak. It is not poss-
ible to give a complete analysis of Dholuo here, but enough of a sketch can be
provided to show that a voicing exchange rule is unnecessary. The voicing-
exchange rule is supposedly motivated by the alternations got 'mountain' : gode
'mountains, and Iwedo 'hand' : Iwete 'hands'. Okoth-Okombo (1982) describes
Dholuo pluralization as involving, inter alia, deletion of any stem-final vowel,
change in the final consonant, and suffixation of eor /; cf. /(/c// 'stone' : plural kite,
and koti 'coat'S : plural kode. We assume that the final vowel is a 'theme vowel',
and the stem is kid. Both pluralization and the 'construct' case formation are
said to involve consonant changes and dropping of the final vowel, but only the
plural adds a suffix, so the construct form of kidi is kit, and the construct form of
koti is kod. Therefore, the consonant change is associated more generally with
90 Studies in the Linguistic Sciences 20:2 (Fall 1990)
'inflected' nouns, and the plural is further marked with a suffix. Inflected nouns
such as /kot/ with final voiceless consonants add a floating feature, [+voice], af-
ter the root, which docks to the final consonant, giving the base kod. In the case
of [kid], I apply a rule of (syllable)-final devoicing which is independently need-
ed in the verbal system (Okoth-Okombo 1982:38-41), giving kit. The reason that
the floating feature [+voice] is not subject to devoicing in [kod] is that devoicing
affects syllable-final [+voice], and a floating feature is not linked to any segment,
so is not syllable-final.
We conclude that, pace Anderson (1988), there is no evidence that phono-
logy should only contain 'purely phonological' rules. There is a very basic rea-
son to reject this view, namely that it entails a more powerful theory than the one
argued for here, since it allows uncontroversially morphological rules, namely
concatenations, to be ordered after phonological operations, which is imposs-
ible in the noninteractive theory.
3.3 REANALYSES
The division of process morphology into concatenation plus a phono-
logical rule removes reduplication from the pool of support for the interactive
model. The problem as exemplified in Kihehe is that the entity which re-
duplicates is the stem, which is the output of Level 1 morphology, and thus
generally excludes the object prefix or the infinitive prefix, which are at Level 2.
However, if a prefix segment fuses syllabically with the initial stem syllable, the
prefix segments are copied.
(39)
ku-teleka
'to cook"
ku-teleka-teleka
ku-lu-teleka
'to cook if
ku-lu-teleka-teleka
kw-iita
'to spill'
kw-iita-kw-iita (/ku-ita/)
ku-lw-iita
"to spill it'
ku-lw-iita-lw-iita (/ku-lu-ita/)
As numerous researchers have observed, reduplication is not a single
operation, but arises from multiple subparts. The morphology is affixation: A
degenerate element is added. I assume that the root plus following suffixes
form a higher prosodic unit such as 'word', so the empty affix is a word node.
The input to the phonological copy operation is thus (40).
(40) kuteleka kulwiita
WW v^v
a a a o a a a
\l/ V
Odden: Phonology and its interaction with syntax and morphology 91
Adding this prosodic affix constitutes the entire morphology of reduplication.
The ability to affix or the shape of the affix is not affected by phonological rules
in Kihehe, nor in any other language. The characteristic work of reduplication is
largely done by the phonology, which receives this degenerate representation
and fills in that template. In Kihehe and cases like it, this takes place after cer-
tain phonological rules. Thus reduplication falls out of the picture entirely.
Returning to cases like Danish, Kiparsky (1984:157) cites a supposed
word-formation process in Icelandic which forms nouns from verbs by truncating
-a; e.g. /c//frfrom klifra and sdtrirom sotra. This process follows the allophonic,
presumably postlexical, rule lengthening vowels in open syllables, and does
not feed into the lexical rules t;-epenthesis and y-deletion. Kiparsky observes
'that the morphological derivation that provides its input structure is also post-
lexical', and that 'we seem to have a prima facie case here of a word-formation
rule which applies lexically in one sets of words... and postlexically in another'.
(157) The supposed rule of word formation is indistinguishable from a morpho-
logically conditioned phonological rule (156):
(41) a-^0/ ]v]n
Since this rule is ordered after a very general and exceptionless postlexical rule
of allophony, by calling the process an operation in the morphological com-
ponent, the entire distinction between lexical and postlexical rules, and the very
integrity of the lexical component, are called into serious question.
Other cases of subtractive morphology have been brought out. Martin
(1988) shows that pluralization in Koasati may delete the stem-final rime.
(42) Singular Plural Gloss
pitaf-fi-n pft-li-n 'to slice up the middle'
tiwap-li-n tiw-wi-n 'to open something'
ataka:-li-n atak-li-n 'to hang something'
koyof-fi-n koy-li-n 'to cut something'
In the theory given here, this process must be the result of a morphologically
conditioned phonological rule, something like (43), since in general all de-
letions must be part of phonology.
(43) Koasati Rime Deletion
R ^ / steni]
[+PLURAL]
This case is similar to Danish, except that deletion affects a higher-level pro-
sodic unit, namely a rime, and therefore indirectly affects multiple segments.
92 Studies in the Linguistic Sciences 20:2 (Fall 1990)
There is a similar deletion process in Kinnatuumbi which deletes the
segmental material of one of the locative prefixes, ki^-, after a vowel. This rule,
illustrated in (44) is optional, so there are two variants for each sentence.
(44) nijenda kQkijjooi^ 'I am going to Kipooi'
nijendaa kijDooi^ (id.)
eendabijtgka kyKipati^mg 'he is running to Kipatimu'
eendabijtijkaa Kijjati^mij (id.)
It is apparent that the locative prefix kij- is present in underlying representations,
but is in part deleted. The evidence for its underlying presence even when
deleted is that its tone and mora are preserved. The rule deletes the segmental
material of the syllable /c^-, but preserves tone and morale structure.
(45) Locative Truncation (optional)
a [w o ^ {prosodic structure preserved}
The syllable preceding /cy- takes the tone and mora that are originally part
of the syllable of kt^-, so for this reason the rule affects only the segmental repre-
sentation. As seen in (46), the H tone on kij- comes from Initial Tone Insertion.
(46) ni^enda kg-kipo6i^ Underlying
ni^enda ki^-kipooi^ ITI
ni^enda ^i kipooi^ Locative Truncation
nijendaa kijDooi^ Reaffiliation of stranded mora
Finally, /((^-deletion is possible only if the prefix is monomoraic. Regular
syllable fusions can make the prefix bimoraic, thus blocking deletion.
(47) nijenda kw(^i^s(.wa 'I am going to the islands'
ni^enda kQgnkoongo 'I am going to Mkongo'
ni^enda kg-mij-koongo Underlying
nijenda ku-m-koongo U-deletion
ni^enda kQ-rp-koongo ITI
ni^enda kQg-nkoongo Nasal desyllabification
NA Locative Truncation
The second example shows lengthening of the locative prefix syllable as a
consequence of postlexical desyllabification of the syllabic nasal in nkoongo, so
Locative Truncation must be postlexical. Locative Truncation clearly applies
after phonological rules, so it must itself be a phonological rule, specifically the
dissociation of the segments of this morpheme, with retention of prosodic
structure.6
Odden: Phonology and its interaction with syntax and morphology 93
These deletions and prosodic restructurings bring us closer to the
apparent case of post-phonological allomorphy in Arabic which we started with.
Before getting to that case, consider another example which Is often treated as
phonologically conditioned allomorphy, but which has another Interpretation, as
a morphologically conditioned phonological rule, now with a similarity to
KImatuumbI Locative Truncation. This Is the case of Korean /' ~ ka allomorphy.
Given that we want to prohibit rules of -oses deletion and -iddletown Insertion In
the phonology, how can we treat the / ~ ka alternation as phonological?
Case markers in Korean vary according to whether they follow a con-
sonant-final base or a vowel-final base.
(48) Korean
citation param pori
nominative param-l pori-ka
accusative param-i-l pori-ri-l
topic param-i-m pori-ni-n
'wind' 'barley'
Writing njles to delete /or n after a consonant is simple, and handling the i-l
~ ri-l alternation with suppletlon-style allomorphy falls to capture the phono-
logical similarity between the allomorphs. The problem really Is in the nom-
inative: Can we handle this alternation by a phonological rule? The present
allomorphy can be handled by an operation entirely analogous to the Klma-
tuumbi Locative Truncation and Koasati Rime Deletion rules. Specifically, I as-
sume the underlying affix ka, and Involve a rule to delete the segmental content
of this syllable after a consonant.
(49) Korean Nominative Destructuring
o a ^ {prosodic structure preserved}
I [+NOM]
C
This will leave behind a segmentless mora. Default rules then assign the
necessary features, and we will end up with /; cf. (50). ^
As It happens, no phonological rules feed into (49), so Korean Is not
crucial for distinguishing the theories. Nevertheless, It suggests a direction for
reanalysis of other supposed cases of phonologically conditioned allomorphy:
Such rules might be slightly bizarre, morphologically conditioned rules In the
phonology.
94 Studies in the Linguistic Sciences 20:2 (Fall 1990)
(50)
Underlying
Nominative
Destructuring
para mi Defaults,
I I I I II Resyllabification
C V C V C V
This then brings us back to Arabic. How might we handle this problem with
a phonological rule? We can treat this as a kind of diphthongization arising
from prosodic restructuring. The syllable structure of underlying /i:/ is reduced to
a simple CV core syllable, with /assigned to the onset by rule (51).
(51) Classical Arabic First Singular Diphthongization
o o a
/ \ / \ I
V V -> C V / V
\ / I
The syllable peak lacks segmental material, so default rules result in a.
(52) maqha i Output of Glide Elision, Vowel Fusion
/\ /\
V V vv
Odden: Phonology and its interaction with syntax and morphology 95
maqha i Diphthongization
/\ I
V V cv
maqha i a Defaults
/\ I I
V V C V
Another case, cited by Hargus (1990) as an example of feeding from
phonology into morphology, is Elative Formation in spoken informal Javanese
(Dudas 1976), which changes the final vowel of a phmary adjective to a high
tense vowel, in order to create the intensive form of the adjective.
(53) Primary Elative Gloss
alus alus 'smooth'
resi? rasi? clean'
adoh aduh 'far'
rame rami 'noisy'
This process seems to interact crucially with two phonological rules in the
language. First, there is a rule laxing high vowels before a tautosyllabic con-
sonant, a rule which has applied to the first two non-Elative forms alus and
rdsi?:
(54) High Vowel Laxing
o
/ \
/ \
V C
I
[H-hi] ->[- tense]
In (53), the Elative has a tense vowel, not a lax vowel, and therefore Elative
Formation counterfeeds High Vowel Laxing.
Another interaction between Elative Formation and the phonology of Java-
nese involves stem-final /a/.
(55) Primary Elative Gloss
'easy'
'hard'
'strong'
In case the vowel a is followed by a consonant, as in the first two examples, the
Elative has the front vowel /'. In case a is word-final, as in the last example, it
shows up in the Elative with the back round vowel u. How do we explain this?
First, there is a rule in Javanese which rounds word final a to o:
(56) Low-Vowel Rounding
3^3/ ]
gampang
gamping
karas
karis
roso (/rosa/)
rosu
96
Studies in the Linguistic Sciences 20:2 (Fall 1990)
This rule has applied in the non-Elative form roso. The features [back] and
[round] are interdependent for high vowels, so all back vowels are round and all
front vowels are nonround. When Elative Formation raises the nonround vowel
a, its backness is determined on the basis of its roundness, resulting in a front
vowel. In rosD, the vowel has rounded by (56), so we find a back vowel. How-
ever, we must first apply (56), and then do Elative Formation.
As with the truncation examples, the real issue is precisely what Elative
Formation is. The Elative can, in fact, be explained without feeding from phono-
logy into morphology, using simple affixation. The Elative morpheme is a partial
feature specification containing only the vocalic features [+hi, +tense], hence is
analogous to floating tone suffixes as one finds in autosegmental analyses of
tone. Thus the underlying forms of [rami] and [rosu] are those given in (57).
(57) rame - f+hi 1 rosa - F+hi 1
L+tnsj L+tnsJ
The floating features of the Elative are mapped to the final vowel by a rule
analogous to the kinds of docking rules we find in tone languages with floating
tone affixes.
(58)
Elative Docking
V
\
\
[V HEIGHT]'
As it happens, this Docking rule applies after Low-Vowel Rounding and High
Vowel Laxing. Consider the derivation in (59) for [rosu]. Here, underlying /a/ is
followed by the floating features [+tense, +high], but these features do not inter-
fere with Low Vowel Rounding, which operates on the feature [round], a feature
lacking from the Elative morpheme. After rounding has applied, however, the
(59)
[r
E
a]
C V C V
/\
/\
/ \
/ \
[+rcl] \ ,
[-rd]^ \ .
+ tns
-tns
r+ tns
1+ hi .
.- hi
-hi
- lo .
+ lo .
[r s
0]
C V C
: V
/\
/\
/ \
/ \
[+rcl] \ ^
[+rcl]^ \ ,
+ tns
-tns
r+ tns'
L+hi .
-hi
-hi
-lo .
.+ lo .
Underlying
Low Vowel Rounding
Odden: Phonology and its interaction with syntax and morphology
97
[r
s
u]
C V C V
/\ /\
/ \ / \
[+r
?1 ^ n
+ tns
-hi
-lo
[+rd] \
+ tns
L+ hi
Eiative Docking
Elative features dock to the final vowel, raising It, and as noted earlier, the
feature [+back] then is assigned to the vowel on the basis of the feature
[+round].
The derivation of [aius], where docking the Elative features counterfeeds
High Vowel Laxing, proceeds similarly; cf. (60). The underlying high vowel is
followed by a consonant in its syllable, and therefore High Vowel Laxing ap-
plies. The presence of floating vowel height features does not affect the laxing
rule, since the rule is only concerned with whether there is a consonant in the
syllable, and floating features play no part in syllabification. Then Elative Dock-
ing applies and wipes out the effect of the High Vowel Laxing rule. In other
words, Javanese Elative Formation does not provide any support for the claim
that phonology can feed into morphology.
(60)
[a
1 u s]
a
1
1
1
o
/l\
/ 1 \
/ 1 \
V
C V C
1
-t-tns
.+ hi
â– - tns'
.+ hi .
[a
1 u s]
o
1
1
1
o
/l\
/ 1 \
/ 1 \
V
C V c
r 1 -1
-tns
.+ hi .
'+ tns'
.+ hi .
[a
1 u s]
o
1
1
1
o
/l\
/ 1 \
/ 1 \
V
C V c
^ 1 ^
Underlying
High Vowel Laxing
Elative Docking
r+ tns 1
L+hi J
98 Studies in the Linguistic Sciences 20:2 (Fall 1990)
Lack of space prevents actually reanalyzing all cases of feeding from
phonology to morphology, but one further example can be considered. Hyman
(1990) cites an assibilation rule in Shi and Luganda which could be interpreted
as a phonological rule applying before morphology. Certain suffixes with the
vowel /, notably the causative, trigger assibilation of a stem-final consonant,
hence the Shi causative form /sunik-i/ surfaces as [sunisi] 'cause to push';
similarly, Luganda /lamuk-i/ surfaces as [lamusi] 'greet'. In Shi, we encounter
the problem that this assibilation applies across the imperfective morpheme
-ag- (which itself does not undergo assibilation), viz. [sunisagi]. In Luganda,
assibilation applies to each morpheme-final consonant in a polymorphemic
stem, but not to morpheme-medial consonants, viz. [lamus-iriz-i] 'greet without
ceasing' from /lamuk-irir-i/. Hyman proposes an analysis of these data with
interleaving of phonology and morphology. In Shi, the assibilation applies to
the form /sunik-i/ giving /sunis-i/, and then the imperfective morpheme -ag- is
infixed to give [sunis-ag-i]. In Luganda, assibilation applies to the underlying
form /lamuk-i/ giving [lamusi], then -irir- is infixed in the morphology, and the
form is resubmitted to the phonology, giving [lamusirizi].
The motivation for the interleaving analysis is that it allows the segment
which undergoes assibilation to stand right before the triggering suffix. While it
is natural for the triggering element to stand immediately next to the element
undergoing the rule, phonological rules applying at a considerable distance are
not unheard of. Neutral vowels in vowel harmony systems are legendary, and
Odden (1989) discusses a considerable number of rules applying across un-
bounded strings. It is possible to construe the rule as spreading the relevant
feature — it is not trivially obvious under any view of morphology ~ phonology
interaction what feature is spreading — in an unbounded fashion to any mor-
pheme-final nonlabial consonant. No pnnciples of phonological theory have
been violated by construing the rule as unbounded spreading, therefore we
have no warrant to reject the noninteractive model on the basis of assibilation in
Shi and Luganda.
These examples show that deletion is needed in phonology, which is
hardly surprising, and that deletion can affect multiple segments, as long as
they form a higher-level constituent. Unsurprisingly, too, nodes in the feature
hierarchy may constitute morphemes, which explains away the Javanese
Elative. In this enterprise of reanalyzing feeding from phonology into mor-
phology, it would be useful to know what limits are to be imposed. The limits
are, of course, the limits imposed on phonological analysis: I do not propose
that ANY well-motivated constraints on phonology be relaxed for the sake of re-
analyzing supposed morphological operations. Consistent with the constraint
that phonological operations apply to prosodic or (sub-)segmental constituents,
I predict that there is no so-called subtractive morphology affecting morphemes
which are not phonological constituents. A specific case of this constraint arises
in Chimwiini.
'to sow'
ku-yal-o:w-a
'to be sown'
'to enter'
k-ingil-o:w-a
'to be entered'
Odden: Phonology and its interaction with syntax and morphology 99
Kisseberth and Abasheikh (1974) note that the passive suffix in Chimwiini
is generally -o:w, as we see in (61 ).
(61) ku-[um-o:w-a 'to be bitten'
na-kimb-il-o:w-a 'she is being sung to"
ku-dar-o:w-a 'to be touched'
Before the passive suffix, dental / becomes alveolar, as the forms of (62)
show.
(62) ku-ya:L-a
k-i:ngi|_-a
Kissseberth and Abasheikh suggest rule (63).
(63) L-^l/ + o:w
The perfective tense passive is somewhat problematic, since the passive
suffix o.w is not found on the surface. Furthermore, the perfective passive
irregularly selects the final vowel a rather than e, as (64) shows.
(64) l_um-i:j_-e 'he bit' [um-i:l-a 'he was bitten'
som-e:[-e 'he read' chi-som-e:l-a 'it was read'
These examples show that the rule changing dental / to alveolar applies, even
though the supposed conditioning factor, the passive morpheme, is lacking.
In an interactive model with morpheme deletions, we could handle this by
applying the dental-to-alveolar rule first, then deleting -o.w.
(65) l_um-i:l_-o:w-a Underlying
l_um-i:l-o:w-a / -^ I
l_um-i:l-a o.iv-deletion
This is not allowed in the model proposed here, since morpheme deletion is
barred. The rule cannot be phonological, since -o.w is not a constituent. If it is
unsyllabified, it is not a rime; if it IS syllabified, it straddles syllables.
Therefore, the only solution is to directly restrict the insertion of the passive
affix, so that it is not insertable in [+PERFECTIVE] verbs. Then how about the
dental-to-alveolar change? As Kisseberth and Abasheikh point out, this change
need not be triggered by the phoneme sequence -o.w, but could equally refer to
the feature [+passive]. Certainly the change from dental to alveolar does not
involve feature spreading from one of the suffix segments.^
(66) L -^ ' / + [PASSIVE]
100 Studies in the Linguistic Sciences 20:2 (Fall 1990)
Therefore, the noninteractive theory is forced to adopt the solution suggested
by Kisseberth and Abasheikh that the mutation of / is triggered not by the
passive suffix itself, but the morphological feature [passive].
It should therefore be clear that it will take much stranger ailomorphy than
is currently available to support the claim that phonology can feed into mor-
phology. Ailomorphy like English go ~ went, or be ~ am - is - were certainly
would qualify, but none of this variation is phonological conditioned, much less
conditioned by derived phonological information. Similarly, various examples
of phonologically conditioned ailomorphy cited in Carstairs 1989:70:1, such as
the alternation between preconsonantal a and prevocalic dz in Fang, appear for
the most part to be true cases of phonologicallly conditioned ailomorphy, but
none of these examples seem to involve derived phonological properties.
There are still a few untouched but well-known cases, where phonological
reanalysis is not the obvious solution. The best known case, cited inter alia in
Kiparsky 1982:33-34, is the verb-to-noun derivational suffix -a/ in English, which
seems to attach only to stems with final stress.
(67)
arrival
disposal
acquittal
refusal
reversal
survival
bestowal
withdrawal
betrothal
avowal
renewal
revival
approval
transferral
betrayal
appraisal
deferral
referral
perusal
upheaval
burial
denial
This stands as one of the strongest arguments available for phonology
preceding morphology, since the putatively derived phonological condition is
one of the factors determining whether the affix can be used at all; that is, we
are not dealing with variations in the shape of a morpheme or with phonological
deletion. This case is nevertheless not strong enough to unambiguously select
the interactive model. Given the nonproductivity of this affix and the small num-
ber of forms available, it is impossible to really test any hypothesis regarding
this affix. Supposing that position of stress were the correct generalization, it is
possible that stress in these words is present in underlying representation, so
this would not be a case of morphology being sensitive to derived phonological
information. Furthermore there is a different generalization which covers the
data, namely that -a/ only combines with Latinate bound prefix plus mono-
syllabic root.
Booij and Rubach (1987) cite two other cases from Dutch, but these submit
to reanalysis as well. The first case involves the interaction between a rule of
schwa-deletion and a rule of ailomorphy. Schwa-deletion supposedly applies
to the final vowel of the word "embassy", which is ambassade, when it stands
before the suffix eurin the word ambassadeur.
Odden: Phonology and its interaction with syntax and morphology 101
(68) 9^0/ V
The feminine form of ambassadeuris ambassadrice, which according to Booij
and Rubach arises by first generating ambassadeur from underlying am-
bassade + eur, and then replacing eurwith ris in the feminine.
(69) ambasada - or Affixation
ambasad - or Schwa-deletion
ambasad - or - is9 Affixation
ambasad - r - isa r/ce-allomorphy
Since the feminine allomorph is phonetically consonant-initial, it supposedly
could not trigger schwa-deletion, so if the ending rice were added directly to
ambassade, we would generate incorrect 'ambassaderice.
There are a number of possibilities for reanalysis, all of which are con-
sistent with the rest of Dutch phonology. One possibility is that schwa deletes
before rice by a morphologically conditioned phonological rule, as in the cases
we have seen earlier. Another possibility is to derive ambassadrice from the
stem ambassad, not the word ambassade. However, the best possibility seems
to be to not derive ambassadrice from anything, except French.
The second example, in many ways similar to English -a/, is the case of the
suffix -/ef which only productively attaches to nouns ending in unstressed /.
(70) psychologi'e 'psychology' psychologisch 'psychological'
hysterfe 'hysteria' hystehsch 'hysterical'
agressie 'aggression' agressief 'aggressive'
actie 'action' actief 'active'
Dutch stress is not transparent, so it is not obvious that we are dealing with
DERIVED phonological information. Gert Booij (p.c.) informs me that regular
stress is on the penult, which fits with our hypothesis. Words with irregular
stress, especially final stress, will be entered in the lexicon with stress pre-
assigned, whereas words with penultimate stress will have no underlying
stress. The condition for affixation of -iefls then simply, as Booij and Rubach
assume, that it only attaches to words ending in unstressed -/. Words such as
psychologie with final lexical stress do not satisfy this condition, so cannot take
the affix -ief.
3.4 PRECYCLICITY
So far we have only considered proposals that morphology may follow
phonology. One question to consider is whether it is possible to directly refute
lexical phonology in terms of phonology ~ morphology interactions. There is a
kind of interaction between phonology and morphology which is amenable to
102 Studies in the Linguistic Sciences 20:2 (Fall 1990)
description in the non-interactive theory but which cannot be handled in Lexical
Phonology, and that is precyclic rules. I conclude with a look at such a case in
Maltese Arabic, where Level 1 phonology needs to access Level 2 information.
This should refute the interactive theory, since Level 2 morphology has not even
been done at the stage of Level 1 phonology. However, there is a way for
Lexical Phonology to wiggle out of this problem, by appeal to precompilation.
The cyclic Stress rule of Maltese is involved, and for this Brame (1974) is
useful, who formulates the stress rules as in (71 ).
(71) Stress
V ^ [+stress] / Co { (VC) VC^ ) ]
There is an Apocope rule deleting an unstressed vowel in open syllables.
(72) Apocope
^^0/ CV
These rules interact to account for the paradigm of hataf, in (73).
(73)
htaft
â– 1 snatched'
/hataf-t/
htafna
'we snatched*
/hataf- na/
hataf
'he snatched'
/hataf/
hatfu
'they snatched'
/hataf-u/
hatfet
'she snatched'
/hataf-it/
Stress and Apocope apply to underlying hataf+u to yield hatfu, and apply to
hataf+na to yield htafna.
(74)
hataf
hataf-u
hataf- n a
Underlying
hataf
hataf-u
hataf-na
Stress
NA
hatf-u
htaf-na
Apocope
Other morphemes which are Level 2 suffixes, seen in (75), can be added
to the verb after affixation of subject agreement, including the object suffixes
-kum 'you (pi.)' and -ik 'you (sg.)', and the negative suffix -s.
(75) hataf-s 'he didn't snatch'
hatf-it-kom 'she snatched you (pi.)'
hatf-ek 'he snatched you (sg.)'
The argument that these affixes are at a different level is the cyclic pattern of
stress assignment. One fact which cyclic stress explains is the surface contrast
in (76) between bistratal [hatafna] 'he snatched us' and monostratal [htafna] 'we
snatched'. On Level 1, Stress and Apocope apply to the form [hataf] 'he
snatched" and to [hataf+na] "we snatched'. The second form is directly mapped
Odden: Phonology and its interaction with syntax and morphology 103
onto the phonetic output [htafna]. The first form rtafa/ contains a Level 2 suffix,
so Stress and Apocope reapply on the L2 cycle. The stress is reassigned to the
penult, but due to the previously assigned stress on the first syllable, Apocope is
blocked and the phonetic form is [hatafna].
(76) [hatafna] [htafna]
'he snatched us' 'we snatched'
hataf hataf-na Input to LI
hataf hataf-na Stress
NA htaf-na Apocope
hataf-na Input to L2
hataf-na Stress
NA Apocope
Therefore the object suffixes and negative s must only be available on Level 2,
and Stress and Apocope are cyclic rules.
Now we turn to the other rule in the paradox. The Level 2 affixes also
lengthen an immediately preceding vowel. Thus htaftu+na becomes htaftuuna
and htafna+kom becomes htafnfekom. The negative suffix -s also induces leng-
thening, so ma hatfu+s becomes ma hatfuus.
(77) htaftuuna 'you (pi.) snatched us' (htaftu)
htafnfekom 'we snatched you (pi.)' (htafna)
ma hatfuus 'they didn't snatch' (hatfu)
A straightforward formulation of this lengthening is possible within any theory:
Any Level 2 suffix induces Lengthening.
(78) Boundary Lengthening
V-^VV/ ]X
Note that this derived length attracts stress, so we get hatfuus, not *hatfuus.
Therefore, Boundary Lengthening precedes Stress on Level 2.
Now we come to the paradox in (79). The problems is that verb stems
ending with a vowel, such as ?ara, lengthen their final vowel before a Level 2
suffix as predicted, but they must do so on Level 1 before stress is assigned.
(79)
7ara
'he read'
/7ara-0/
7raana
'he read us'
/7ara-0 + na/
jara
'it happened'
/jara-0/
jraa-li
'it happened to me'
/jara-0 + l-i/
104 Studies in the Linguistic Sciences 20:2 (Fall 1990)
In Lexical Phonology, Lengthening would have to apply to Level 2, since it is
triggered only by Level 2 suffixes. Therefore assignment of Stress on Level 1
should precede Lengthening, and should not be sensitive to the output of
Lengthening. But this prediction is incorrect, as seen in the derivation (80).
(80)
7ara+0
7ara+0
Subject affixing
Stress
NA
Apocope
[L2]
Object affixing
Lengthening
Stress
7ara+na
7araa+na
7araa+na
NA
Apocope
*[7araana]
On the Level 1 cycle, stress should be assigned to the first vowel, just as it is in
the unsuffixed form. On Level 2, the final vowel is lengthened and that vowel
then gets the stress, but the subordinated stress on the initial vowel would
incorrectly block Apocope. The correct derivation requires that Lengthening
apply prior to Level 1 Stress, giving 7araa+na as the input to Stress, so that the
penultimate vowel is stressed, and the initial vowel is never stressed.
(81 ) 7ara - na Output of morphology
7araa - na (Precyclic) Boundary Lengthening
7araa - Level 1 Stress
7raa - Level 1 Apocope
7raana Level 2 (nothing applies)
This can be accommodated in any theory where all morphemes are
concatenated before phonological rules apply, provided some rules, and in
particular. Boundary Lengthening, apply precyclically, so that Level 2 suffixes
can be seen and trigger lengthening before Level 1 phonological rules apply.
As noted earlier, there is a way out of this problem for Lexical Phonology,
and that is via the theory of rule precompilation. We can treat Lengthening as a
precompiled rule, which would be written as (82):
(82) Lengthening (precompiled)
V->V:/ ] [FRAME 1]
The Level 1 phonology will generate both 7ara and 7araa. The former leaves
Level 1 as 7ara and the latter leaves Level 1 as 7raa.
Odden: Phonology and its interaction with syntax and morphology 105
(83) 7ara
[+F1]
7araa
NA
Lengthening
7araa
7ara
Stress
7raa
NA
Apocope
Frame 1 is defined as in (84), so when a negative or object affix appears in
Level 2, 7raa is selected.
(84) Frame 1 : [verb ] { ^^f.^ )
Negative
4. CONCLUSION
I have considered two questions of how phonology interacts with syntax
and phonology, and using as my diagnostic the class of information available to
each component in the grammar, I have considered how Lexical Phonology
and the noninteractive model explain the behavior of linguistic systems. The
constrained version of Lexical Phonology cannot explain the behavior of sandhi
rules in Kimatuumbi or precyclic Boundary Lengthening in Maltese, but extend-
ing the theory with precompilation makes up for these descriptive shortcomings
in Lexical Phonology. However, this is accomplished only at the expense of
adding machinery to Lexical Phonology which renders it empirically indis-
tinguishable from the non-interactive theory. A very basic prediction of the inter-
active model, that phonology can feed morphology, has no empirical support.
These conclusions do not refute the theory of Lexical Phonology, since model
(2) might still be right, even if there is no evidence for it in terms of the kinds of
information available to each component. At this stage, it is not obvious where
else we could derive support for the interactive model.
NOTES
^ There is little evidence from morphology that the construct 'level' is use-
ful, and some evidence in the form of bracketing paradoxes that it is coun-
terproductive. It is quite possible that levels are purely phonological, and that
the structures corresponding to levels in this model are actually constructed at
the interface between phonology and morphology. It is also unclear what dis-
tinguishes 'postlexical' from 'lexical' rules. Odden (1990) suggests a division
where rules refering to nonphonological properties are lexical; however, evid-
ence exists, some of which is discussed here, that purely automatic rules can be
ordered before rules with morphological and syntactic conditions. This calls
106 Studies in the Linguistic Sciences 20:2 (Fall 1990)
into question the view that lexical and postlexical phonology form separate
connponents.
2 As originally stated in Kiparsky 1973, the EC requires comparing the set
of imput strings which 'fit' rule A to the set of input strings which 'fit' rule B, and
testing if the A set is a subset of the B set. Such an extensional characterization
of the conditions for disjunction runs into the insurmountable problem that pho-
nological rules may involve multiple words in a sentence. Therefore, the set of
input strings to a rule is potentially infinite, so enumerating that set and com-
paring it to any other set of strings in the language is impossible. Intensional
characterizations of EC have been given, such as that in Kiparsky 1982, which
is stated in terms of the structural deschptions of rules, not the class of strings
undergoing the rules — this is of course just the Proper Inclusion Precedence
Principle. The problem with this statement of EC is that it gives tremendous
latitude to the theory of disjunction, since a rule can quite often be stated
formally in many different ways with no effect on the set of sthngs which under-
go the rule.
3 The source of this example is SPE, which actually cites a height-ex-
change rule in a different language, Hebrew.
'' Vowels with subscript double dots (or "umlauts') have breathy voicing,
and vowels with subscript tilde have 'hard' voice.
5 Most examples of the change from voiceless to voiced consonant appear
in stems which lack a final vowel, viz. got. The only two stems with voiceless
consonant I have located in Okoth-Okombo 1982 which have a theme vowel
are loan words. Similarly, nouns with stem-final voiced consonant all appear to
select a theme vowel. This distribution is a consequence of the fact that the
voicing alternation arises out of two processes, final devoicing and intervocalic
voicing.
6 This rule poses a problem for the claim that lexical and postlexical rules
reside in separate components. Locative truncation is clearly not a 'clean' post-
lexical rule, but it follows nasal-resyllabification rules which are quite excep-
tionless and purely phonological. One could nevertheless assign the resyllabi-
fication rules to the lexical phonology, but this again raises the question
whether there is any principle determining whether a rule is lexical or post-
lexical.
7 It is often assumed that /-not /, is the vowel which arises from default rules
in Korean. However, ^has a restricted distribution in Korean: No morphemes
end in /-except for the demonstratives /c^and nk The relevant generalization is
that ^ cannot be prepausal — the demonstratives can never be prepausal. A
similar constraint on i- appears in Tigrinya: Epenthesis inserts i- in word-final
position, but the vowel is realized phonetically as / in that position.
Odden: Phonology and its interaction with syntax and morphology 107
8 It is in fact not clear what feature distinguistics the two kinds of / in
Chimwiini — Kisseberth and Abasheikh do not voice an opinion about the fea-
tural basis for distinguishing these segments.
REFERENCES
AQUILINA, J. 1965. Maltese. London: English Universities Press.
, & B. Isserlin. 1981. A survey of contemporary dialectal Maltese. Leeds: B.
S. J. Isserlin.
Anderson, S. 1975. On the interaction of phonological rules of various types.
Journal of Linguistics 11.39-62.
1988. Morphological theory. Linguistics: the Cambridge survey, vol. I,
Linguistic theory: Foundations, ed. by F. J. Newmeyer, 146-191. Cam-
bridge: University Press.
, & W. Browne. 1973. On keeping exchange rules in Czech. Papers in
Linguistics 6.445-82.
BOOIJ, G., & J. Rubach. 1987. Postcyclic versus postlexical rules in lexical pho-
nology. Linguistic Inquiry 18.1-44.
Brame, M. 1970. Arabic phonology. MIT Ph.D. dissertation in Linguistics.
1974. The cycle in phonolgoy: Stress in Palestinian, Maltese, and Span-
ish. Linguistic Inquiry 5.39-60.
Carstairs, a. 1988. Some implications of phonologically conditioned sup-
pletion. Yearbook of Morphology 1988.67-94.
Denning, K. 1987. Nominal number and morphological class in Dinka: Dia-
chronic and synchronic views. MS.
DUDAS, K. 1976. The phonology and morphology of modern Javanese.
University of Illinois Ph.D. dissertation in Linguistics.
Hargus, S. 1985. The lexical phonology of Sekani. UCLA Ph.D. dissertation
in Linguistics.
1990. A comparison of theories of phonology-morphology interaction.
Lexical Phonology Workshop, June 8, 1990, University of Washington.
Hayes, B. 1990. Precompiled phrasal phonology. The phonology-syntax con-
nection, ed. by S. Inkelas & D. Zee, 85-108. Chicago: University of
Chicago Press.
Howard, I. 1975. Can the 'elsewhere condition' get anywhere? Language 51.
109-127.
Hyman, L. 1990. Conceptual issues in the comparative study of the Bantu verb
stem. MS. (21st Conference on African Linguistics, April 1990, University
of Georgia.)
Jensen, J., & M. Stong-Jensen. 1979. The relevancy condition and variables in
phonology. Linguistic Analysis 5. 1 25-1 60.
108 Studies in the Linguistic Sciences 20:2 (Fall 1990)
Kaisse, E., & P. Shaw. 1985. On the theory of Lexical Phonology. Phonology
Yearbook 2.1-30.
KiPARSKY, P. 1973. 'Elsewhere' in phonology. A Festschnft for Morris Halle,
ed. by S. R. Anderson & P. Kiparsky, 93-106. New York: Holt, Rinehart &
Winston.
1982. Lexical phonology and morphology. Linguistics in the Morning
Calm, ed. by I. S. Yange, 3-91.
1984. On the lexical phonology of Icelandic. Nordic Prosody III, ed. by S.
S. Elert, I. Johansson, & E. Strangert, 135-164. Stockholm: Almquist &
Wiksell.
KISSEBERTH, C, & M. Abasheikh. 1976. On the interaction of phonology and
morphology: a Chi-mwi:ni example. Studies in African Linguistics 7:1.31-
110.
Martin, J. 1988. Subtractive morphology as dissociation. WCCFL 8.
Matthews, P. 1974. Morphology: An introduction to the theory of word forma-
tion. Cambridge: University Press.
MOHANAN, K.P. 1986. The theory of lexical phonology. Dordrecht: Reidel.
Odden, D. 1980. The irrelevancy of the relevancy condition: Evidence for the
feature specification constraint. Linguistic Analysis 6.261-304.
1989. Adjacency parameters in phonology. MS, Ohio State University.
1990. Syntax, lexical mles, and postlexical rules in Kimatuumbi. The pho-
nology-syntax connection, ed. by S. Inkelas & D. Zee, 259-277. Chicago:
University of Chicago Press.
, & M. Odden. Ordered reduplication in Kihehe. Linguistic Inquiry 16.497-
503.
Okoth-Okombo, D. 1982. Dholuo morphophonemics in a generative frame-
work. Berlin: Reimer.
studies in the Linguistic Sciences
Volume 20. Number 2, Fall 1990
APHASIC LANGUAGE UNDER DISCOURSE PRESSURE:
FUNCTIONAL SYNTAX VS. PSYCHOLINGUISTIC FUNCTION'
Use Menn
(University of Colorado, Boulder)
This paper is a preliminary report on work in progress and should be cited
only with that caution in mind. Nevertheless, I feel that this research represents
a new departure of potential importance, and that the early results indeed sug-
gest that we are on a promising track. If they are borne out by further data, we
will have solved an aphasiological problem of about ten years' standing, and in
the process we will have shown how some current and some classical work in
functional syntax may bear on the question of on-line sentence production in a-
phasic patients and perhaps in normals as well.
The present approach began with contemplation of the false starts made
by some Japanese aphasic patients in trying to narrate the events in a simple-
looking little cartoon strip (Figure 1).
Here are some samples of their speech from the Japanese 'hat' story and from
conversations, with the errors explained. Hesitations are indicated by dots (...).
Note the difficulty that patients have in deciding which noun to use as subject of
the events in the second panel, and, for Patient F, in getting the right verb
voice.
Examples (1) - (2) are from Patient F (fluent aphasia, female, age 54).
(1) otoko-no ko-ga... booshi-o ura-ni ton-de-tta.
male-of child-SUBJ ... hat-OBJ back.N-into fly-CONJ-go-PERF
'The boy ... went flying the hat into the back(yard).'
There is an error in the intransitive verb phrase 'go flying': tobu 'fly' + Iru 'go'
should not be used with a direct object 'hat' (marked with -o ). Second, ura-ni
does not mean 'backwards' or the like. An animate noun is used in (1) as
subject, and an inanimate is marked as object, but is used with an intransitive
verb form.
no
Studies in the Linguistic Sciences 20:2 (Fall 1990)
Figure 1: The hat story
(2) otoko-no ko-ga... booshi-ga ... eeto ike-ni
male-of child-SUBJ ... hat-SUBJ ... well pool-into
koro-n-de-isoo da-tta.
fall (onto solid surlace)-coNF-seem aux.perf
The boy ... The hat ... well, looks like it's hit into the puddle.'
Menn: Aphasic language under discourse pressure 1 1 1
Example (2) contains a semantic error: The verb korobu cannot be used for
falling into water, only for falling onto a hard surface. (The effect of this error
can be compared with that in Engl. The hat splashed onto the floor.) A correct
verb choice would be ochiru 'fall', in the form ochi-te-isoo.
Examples (3) - (4) are from Patient M ('mixed' aphasia, male, age 34):
(3) booshi-ga booshi-ga ka-kaze-ga f-fui-te-iru.
hat-SUBJ hat-SUBJ wi-wind-SUBJ b-blow-CONJ.AUX
'The hat (2x), the wind is blowing.'
Patient M here starts with 'The hat' {booshi-ga), then switches to "The wind is
blowing.' The patient's second attempt at this frame is:
(4) kodomo-o booshi-ga booshi-ga booshi-ga
child-OBJ hat-SUBJ hat-SUBJ hat-SUBJ
booshi-ga kaze-o n kaze-no koroga-tte umi-ni
hat-SUBJ wind-OBJ mm wind-POSS roll-CONJ ocean-in
ochiru.
falls
'The child, the hat (4x), the wind, mm, the wind's, it rolls and falls into
the ocean.'
This starts with 'child' as object {kodomo-o), then again with 'hat' {booshi-ga )
as subject and 'wind' as object {kaze-o) — a canonical word order, but
semantically impossible. Patient M attempts to correct the particle on 'wind' but
only comes up with the possessive marker {no) rather than the instrument
marker {kaze-de 'with the wind'). The final complex verb phrase, including the
goal (um/'-n/ 'into the ocean'), is correct.
Various experimental studies of comprehension and production by agram-
matic aphasics in different languages have been reaching the consensus that
sentences in 'canonical form' were easier for patients than others (Bates,
Friederici, Wulfeck & Juarez 1988; Caplan & Hildebrandt 1988; Kudo, Tateishi,
Kashiwaga & Segawa 1982; Menn & Obler 1988; Saffran, Schwartz & Marin
1980; Sasanuma, Kamio, & Kubota 1990). However, canonical form is a
problematic notion if looked at theoretically. A clause 'in canonical form' is tak-
en to be an active declarative main clause whose grammatical subject is an
agent. Therefore, this notion is stated in terms of both syntactic and semantic
constraints.
In practice, there is a further unstated restriction. Given that the usual
question is 'Can patients process the syntax of (a given) string', experimenters
must present them with sentences in which lexical semantics and real-world
knowledge are not sufficient to determine the interpretation. Sentences like
The boy ran, The girl has a cold, A woman is eating some sushi are never
1 1 2 Studies in the Linguistic Sciences 20:2 (Fall 1990)
tested, since there is nothing syntactic to test at the comprehension response
level of pointing to pictures or acting out a scene with toy figures. Almost the
only sentence types that are studied for comprehension are 'reversible' active
and passive sentences like The cat chased the dog or The girl was kissed by
the boy— elaborated, perhaps, with indirect objects or relative clauses. For no
obvious reason except symmetry, the same types of sentences dominate in
syntactic production studies, although many other types have been used when
the focus is on the production of grammatical morphemes.
Another result in the literature has been without adequate explanation for
a decade. Saffran, Schwartz & Marin (1980) showed that some non-fluent
patients had a very difficult time when confronted with the task of describing
pictures in which an animate was acted on by an inanimate — say, a child was
hit by a ball (no thrower visible) — and also pictures in which two animates or
two inanimates were involved. The patients sometimes produced the argu-
ments in the wrong slot, getting the meanings entirely reversed. Saffran et al.
suggested that, when these patients were grappling with two-argument claus-
es, they were relying on an animacy hierarchy rather than on syntax. This find-
ing did not seem to apply to narrative speech, however, and for this and other
reasons it remained problematic. (A full discussion of subsequent treatments
of their findings is beyond the scope of this paper.)
Given such a background, it seemed to me that the patients' problems with
narrating the 'hat' story could be related to the animacy and canonical form
findings, but not without some further work. Yes, an inanimate, the wind, is
acting on another inanimate, the hat. But why try to start the sentence with the
man (or boy)? Why would speakers of Japanese and English, which do not
have grammaticized animacy hierarchies, seem to show the influence of ani-
macy considerations when they become aphasic?
To begin to answer this, we must ask why some languages have animacy
hierarchies in the first place — in other words, why might a semantic factor
become part of the grammar? This takes us on an excursion into functional
syntax, and a general consideration of both semantic factors and discourse
factors. Semantic factors might include inherent properties of the referent (ani-
macy, humanness, ...), and also objective and psychological aspects of the
roles that the referent plays in a particular situation (agency, patiency,
volitionality, sentience, ...). Discourse factors are treated by several different
schools of thought. DuBois (1987), Hopper & Thompson (1984), Durie (1985),
DeLancey (1987), and Fox (1987) emphasize the relations of transitivity and
discourse structure as reflected in morphosyntax. They also include informa-
tion status (given, inferred, new; recency of mention). Kuno (e.g. 1973) has
worked on empathic status: Tthis has to do with who we care about, or whose
point of view we take, in a given situation. (The definitions of all of these terms
are problematic, but that does not mean they are useless; the core notions are
readily exemplified, and clear cases are easy to recognize.)
Menn: Aphasic language under discourse pressure 1 13
in normal discourse, these factors tend to vary together. We tend to empa-
thize with the prototypical protagonists of narrative, namely volitional given
human agents and sentient given human undergoers. Although Du Bois, Du-
rie, and Fox do not work with the empathy variable, they among others suggest
that grammatical patterns of semantic factors, such as grammaticized animacy
hierarchies, grow out of such probabilistic distributions in language use.
I hypothesized that the problem with non-canonical sentences might be
that the patients attempted to start their sentences with the empathic focus —
not as a deliberate strategy, but as the effect of an automatic orientation. How-
ever, when the empathic focus was on the undergoer, perhaps the patients
tended to get into trouble because of being unable to deploy the syntax need-
ed to make the undergoer the first noun phrase in the sentence — for example,
in English, they would be unable to access the passive. Topicality might have
a parallel (but conceptually distinct) effect. So Barbara Fox and I designed
some more pictures and sequences of pictures to test these ideas. First we
commissioned a set of pictured narratives loaded with non-volitional actions
and animate undergoers (e.g. a ball seen breaking a house window, knocking
over a lamp inside, and startling a man sitting next to it). Then, to get tighter
control over our variables, we commissioned two types of controlled-picture
sets.
The first of these varied only in the animacy and/or empathic attrac-
tiveness of the undergoer. Everything else was held constant. The examiner
presented the pictures (in quasi-random order, with other type of test items
interspersed) with the question 'What is happening?' There were two animacy/
empathy series:
(5) Brick falls on parked truck; on wagon pulled by child; on teddy bear
in wagon pulled by child; on woman's arm.
(6) Sled being pulled by a child; snowball hits empty sled; hits bag of
groceries on sled; hits teddy bear on sled; hits child on sled.
Some examples of these are given in Figures 2a-e.
The second type of controlled-picture set varied the context in which a
pictured event was presented, aiming deliberately to build up the empathy/
TOPICALITY of the undergoer. The test pictures were each first presented as a
single panel (no-context condition) with the undergoer-biased question, 'What
is happening to (undergoer)?' Then, in a later test block, each of these pictures
was also presented as the last panel of series of two or more pictures which
provided a context condition for it. In the latter, each series was accompan-
ied by a narrative, until the last panel was reached — when the same question.
114
Studies in the Linguistic Sciences 20:2 (Fall 1990)
Figure 2a: Brick falling on teddy bear
Figure 2b: Brick falling on woman's arm
Menn: Aphasic language under discourse pressure
115
Figure 2c: Snowball hitting empty sled
Figure 2d: Snowball hitting grocery bag on sled
Figure 2e: Snowball hitting child on sled
116
Studies in the Linguistic Sciences 20:2 (Fall 1990)
'What is happening to (undergoer)?', was presented. The events in the test
panels may be briefly entitled 'Boy saves dog", 'Dog saves boy'. 'Teacher
catches kids smoking', 'Girl catches teacher sneaking a drink", 'Boy injured in
gangster crossfire', 'Baby hit by fly ball'. Samples of these pictures are given in
Figures 3a-b.
The patients were asked to respond to the questions about the test
pictures — orally if possible, and also by arranging (previously shuffled) cards
with relevant phrases printed on them. Examples of the phrases provided on
the cards are:
(7) falls on — the teddy bear — the brick — hits. (4 cards)
(8) the girl — the teacher — catches — gets caught — is caught — by.
(6 cards).
c^-^
> >^'>< >
^t
.a'^-.r
Figure 3a: Dog saves boy
Menn: Aphasic language under discourse pressure
117
Figure 3b: Baby hit by fly ball.
We also tested to see if the patients could comprehend passive (and active)
sentences describing the second controlled set better in the context condition
than in conditions without context; but our results so far are insufficient for ana-
lysis, as the six patients who have done this task to date mostly performed
either at chance or at ceiling.
Data from four patients have been analyzed to date, but two of them were
unable to give oral responses, and another one has not yet done the card task.
1 1 8 Studies in the Linguistic Sciences 20:2 (Fall 1990)
No normal controls have been taped yet, either, so these results must be
considered as very preliminary. However, our predictions about the effects of
animacy/empathy and topicality/empathy seem to be borne out.
On the animacy/empathy card arrangement tasks, one patient, Mrs. K, al-
ways put the 'brick' card before the card naming the person or object that it fell
on, but she showed a clear animacy/empathy effect with the 'snowball' set. Her
card-arrangement responses were:
(9) The snowball ... hits ... the sled.
(10) The snowball ... hits ... the bag.
(11) The teddy ... gets hit ... by ... the snowball.
(12) The child ... gets hit ... by ... the snowball.
In the latter two cases, the undergoers are fronted, but in the first two cases
they are expressed as grammatical objects.
In her oral responses, Mrs. K showed the animacy/empathy effect in the
'brick' set:
(13) Brick dent truck.
(14) The kid's ... the brick fall down on the (?)box ... lucky.
(15) The bear is dizzy ... the brick fall down and bear get dizzy.
(16) She get hurt.
However, in the 'snowball' set, which we will examine shortly, there was no
gradation In her responses.
The predicted effect of varying the amount of context was also found in the
card responses of all three patients whose responses have been analyzed so
far: For all of them, more undergoers were fronted in context condition than in
the no-context condition.
We now have to examine the question of whether the animacy effect is the
result of a 'strategy', as Saffran et al. suggested, or of what neuropsychologists
call a 'pull'. The data — mostly from Mrs. K at the moment — suggest that both
these notions may apply. What, first, is the difference between them? If we call
a response pattern an effect of a 'strategy', this implies that the patient had a
voluntary choice among responses. If we say that it was due to a 'pull', on the
Menn: Aphasic language under discourse pressure 1 1 9
Other hand, we imply that the response was less than fully voluntary. Often we
can't tell which was the case, but sometimes the evidence is quite suggestive.
Consider Mrs. K's oral reponses to the 'snowball' set (remember that these
pictures were presented in quasi-random order, and with other test items
interspersed):
(17) Somebody throw the snowball. (E: And?) The snowball fell on the
sled.
(18) Somebody throw the snowball on the sled ... bag ... the snowball hits
the carton of milk.
(19) Somebody throw the snowball right in the bear's face.
(20) The snowball is ... Somebody throw a snowball on the girl right
(here/head).
It seems very likely that this invocation of the off-stage agent 'somebody' is a
strategy. Mrs. K produces a relatively successful sentence, and there is no
straightforward way to account for the 'somebody' in terms of empathy or topic-
ality, i.e. in terms of discourse pragmatics.
However, consider these other oral responses:
(21 ) The kid's ... the brick fall down on the (?)box.
(22) The baby ... no ... the baseball hits the baby.
(23) He ... the boy ... shot ... the bu ... the bullet ... the arm ...
(24) The kiddie ... the girl ... the baseball (gesture) ... the baseball hits the
baby.
Here the floundering starts suggest that Mrs. K was 'pulled' to the em-
pathic focus, the undergoer, but was unable to proceed either with a get or an
is passive. She had to switch to an active structure with an inanimate subject
— or else, as in 23, fail to produce a well-formed clause at all.
We have plans to develop controlled elicitation materials for other non-
canonical structures, since this will be necessary before a psycholinguistic ac-
count of the above response patterns can be defined and defended. Here also
is where the cross-linguistic approach will become central again. Consider the
following generalizations, denved from informal clinical observations and ex-
aminations of less-controlled narratives: Dative structures are hard in English,
but apparently easy in Japanese; locatives, with the same number of
120 Studies in the Linguistic Sciences 20:2 (Fall 1990)
arguments, have no reported difficulty in either language. Why might this be
the case? The Instrumental, also basically a three-argument structure, is hard
for both languages. If we go back to the 'hat' story, we find, for Mrs. K:
(25) The cane ... the men hook a hat ... o-on the hat ... the men ... the cane
... no ... The men ... pick up ... the hat ... with the cane.
Similar efforts were put forth by the Japanese patients recorded by Dr. Fujita
and transcribed by Morishima.
As a speculative conclusion, let me give the 'psycholingulstic account' of
the above response patterns to which I am very much attracted. The extensive
series of experiments on syntactic priming by Kay Bock (e.g. Bock 1986) has
indicated that, whatever else a clause structure is, it is a primable object
independent of the lexical items it contains. And of course, lexical items are
primable, too. Perhaps the high accessibility of the active clause in English is
to be modeled by its having a low threshold of activation. Patients find the
active to be available whether or not it is the structure that will fit with their prag-
matic preferences; meanwhile the word most highly primed is the empathic
focus/topic. (I am assuming, with Bock, a Garrett-type sentence production
model; see Garrett 1980 or 1982.) If the empathic focus/topic happens to be
undergoer, the resulting sentence will be incorrect: For example. The baby hits
the bat instead of The ball hits the baby. When the patient's self-monitoring is
adequate, s/he will reject the incorrect sentences that result. (Even Mrs. K, a
superb self-monitor, probably produced one such error, though there are arti-
culation problems which make that particular example uncertain.) If an altern-
ative strategy is found quickly, we may never see any errors; but if the patient is
unable to switch to a less-highly-aroused noun phrase which fits the aroused
syntax (e.g. the instrument or the agent) and is also unable to access a non-
active syntactic structure, then we get ill-formed utterances like Mrs. K's He ...
the boy ... shot ... the bu ... the bullet ... the arm ...
A final note: I have some hope that these results, if further work sustains
them, may have clinical applications. Patients with good self-monitoring might
be taught to reject the undergoer as the first word if they can't come up with the
right structure, and to try to begin sentences with the agent or instrument,
regardless of empathy or topicality. It might not work, but some of my clinician
friends and I think that it's worth a try.
Menn: Aphasic language under discourse pressure 121
NOTES
*The work reported here is being carried out with the help of Prof. Barbara
Fox and research assistant Yasunori Morishima in the U.S., and Prof. Akio
Kamio, Dr. Sumiko Sasanuma, and Dr. Ikuyo Fujita in Japan. It is supported by
NIH grant 1R01-DC00730-01.
Figure 1 is from the Japanese Standard Language Test of Aphasia; it was
provided by Dr. I. Fujita and is copyrighted. Figures 2 and 3 were drawn by
Kuniko Tada, who was awarded the M.A. in linguistics at the University of
Colorado in 1990 and is currently a doctoral student in the Applied Linguistics
Department at UCLA. Her sophisticated appreciation of the linguistic and
psycholinguistic aspects of our project was crucial to her execution of these
drawings.
REFERENCES
Bates, Elizabeth, Suzanne Hamby, & Edgar Zurif. 1983. The effects of focal
brain damage on pragmatic expression. Canadian Journal of Psycho-
logy 37.59-84.
, Angela Friederici, Beverly Wulfeck, & L A. Juarez. 1988. On the preser-
vation of word order in aphasia. Brain and Language 35.323-64.
Bock, J. Kathryn. 1986. Syntactic persistence in language production. Cog-
nitive Psychology 18.355-87.
Caplan, David. 1985. Syntactic and semantic structures in agrammatism.
Agrammatism, ed. by M.-L. Kean, 125-52. New York: Academic Press.
1988. The biological basis for language. Linguistics: The Cambridge
Survey, ed. Fr. J. Newmeyer, 3.237-55. Cambridge: University Press.
, & Nancy Hildebrandt. 1988. Disorders affecting comprehension of
syntactic form: Preliminary results and their implications for theories of
syntax and parsing. Canadian Journal of Linguistics 33.1-29.
DeLancey, Scott. 1987. Transitivity in grammar and cognition. Coherence
and grounding in discourse, ed. by R. S. Tomlin, 53-68. Amsterdam:
Benjamins.
Du BoiS, John W. 1987. The discourse basis of ergativity. Language 63. 805-
55.
DURIE, Mark. 1985. A grammar of Acehnese on the basis of a dialect of North
Aceh. Dordrecht: Foris.
Fox, Barbara. 1987. The Noun Phrase Accessibility Hierarchy revisited. Lan-
guage 63.856-70.
122 Studies in the Linguistic Sciences 20:2 (Fall 1990)
Garrett, Merrill F. 1980. Levels of processing in sentence production.
Language production, ed. by B. Butterworth, 1.177-220. New Yorl<:
Academic Press.
1982. Production of speech: Observation from normal and pathological
language use. Normality and pathology in cognitive functions, ed. by A.
Ellis, 19-76. London: Academic Press.
GivoN, Talmy. 1979. On understanding grammar. New York: Academic Press.
Hopper, Paul, & Sandra A. Thompson. 1980. Transitivity in grammar and dis-
course. Language 56.251-99.
& 1984. The discourse basis for lexical categories in universal
grammar. Language 60.702-52.
Jarema, Gonia, & Eva Kehayia. 1987. Does there exist a hierarchy of difficulty
along the active/passive dichotomy? Evidence from two Greek-speaking
non-fluent aphasics. (Tapuscrits, CHCN working papers, 13.) Montreal:
Centre de Recherche, Centre Hospitaller Cote-des-Neiges.
, Danuta Kadzielawa, & J. Waite. 1987. On comprehension of active/
passive sentences and language processing in a Polish agrammatic
aphasic. Brain and Language 32.215-32.
Kudo, T., M. Tateishi, T. Kashiwaga, & N. Segawa. 1982. Sensitivity to func-
tors in Japanese aphasics. Neuropsychologia 20.641-51 .
KUNO, Susumu. 1973. The structure of the Japanese language. Cambridge,
MA: MIT Press.
Menn, Use, & Loraine K. Obler. 1988. Findings of the Cross-Language Apha-
sia Study, phase I: Agrammatic narrative. Aphasiology 2.347-50.
Obler, Loraine K., & Use Menn. 1988. Agrammatism: The current issues.
Journal of Neurolinguistics 3.63-76.
Saffran, Elinor, Myrna F. Schwartz, & Oscar S. M. Marin. 1980. The word
order problem in agrammatism, II: Production. Brain and Language
10.263-80.
Sasanuma, Sumiko, Akio Kamio, & Masahito Kubota. 1990. Agrammatism in
Japanese. Agrammatic aphasia: A cross-language narrative source-
book, ed. by L. Menn & L. K. Obler, 1225-307. Amsterdam: Benjamins.
WULFECK, Beverly. 1988. Grammaticality judgements and sentence compre-
hension in agrammatic aphasia. Journal of Speech and Hearing Re-
search 31.72-81.
Studies in the Linguistic Sciences
Volume 20, Number 2, Fall 1990
STANDARDS AND NORMS FOR WORLD ENGLISHES:
ISSUES AND ATTITUDES
Peter H. Lowe n berg
Georgetown University
In the by now extensive research on institutionalized nonnative varieties of
English, such as Indian English or Nigerian English, 1 a crucial distinguishing
characteristic of such varieties has been posited as nativization (e.g. Kachru
1986, Sridhar & Sridhar 1986). In this context, nativization refers to systematic
changes in the forms and functions of English at all linguistic levels, resulting
f^m the extensive use of the language among nonnative speakers in the
absence of native speakers, in non-Western sociocultural contexts, and in
constant contact with other languages in multilingual speech communities. 2
Nativized features have sociolinguistic status as stable modifications in the
forms and functions of fully elaborated varieties of English that have developed
Fn nonnative sociolinguistic contexts.
Most nativized features would be considered deviant if transplanted to
countries where 'native-speaker' varieties of English (e.g. American or British
English) are used. However, jn the settings of their use, many of these linguistic
innovations and modifications of English are so widespread that they have
become de facto local norms for usage along the entire style range of English.
Tor example, attitudinal studies reported in Kachru 1976 and Shaw 1981
indicate that in at least two countries, India and Singapore, between forty and
fifty per cent of college-educated English users feel that nativized features /
should be included in local norms for English teaching, that is, in Standard /
English. 3
While such studies indicate that nativized features, in general, are
becoming accepted as pedagogical standards for nonnative varieties of
English, little research to date has focused on determining exactly which
innovations are nativized features of Standard English in particular nonnative
varieties. This paper, in proposing an approach for beginning to identify these
nativized norms, focuses on morphosyntactic deviations from the norms of Stan-
dard American English in sample written texts taken from domains of Standard
English in several nonnative varieties.^ Particular attention is devoted to the
124 Studies in the Linguistic Sciences 20:2 (Fall 1990)
nonnative variety of English that has developed in Malaysia. The analysis dem-
onstrates that in many cases, norms for Standard English cannot be identified
on linguistic grounds alone. Rather, such a determination frequently depends
far more on attitudinal variables, particularly on the relative sociolinguistic status
of the sources of an innovation. The paper concludes with a discussion of the
significance of these attitudinal factors in assessing the English proficiency of
speakers of nonnative varieties.
Extension of productive morphosyntactic processes of Standard
English
A major source of nativization in nonnative varieties is the extension of in-
novative morphosyntactic processes that are also very productive in, and
frequently cause differences between, the native-speaker varieties of English.
One such process by which British and American English often diverge is the
I conversion to countability of noncount nouns which semantically include an
. I aggregate of countable units, as in examples (1) and (2) from reputable sources
^ of British English.
(1 ) Some small initial fall-off in attendances is unavoidable.
(Times of London, 10/27/86:17, in Algeo 1988:7)
(2) ... iceberg lettuces are down in price and should be selling for
between 35p and 55p, depending on size.
(Daily Telegraph, 8/9/85:6, in Algeo 1988:7)
This process, which is restricted to specific lexical items in each variety of
English, likewise results in innovations in nonnative varieties, as in (3), from
both Singapore and Malaysian English, (4), from Nigerian English, and (5), from
Ghanaian English. 5
(3) Pick up your chalks.
(Tongue 1974:43)
(4) I lost all my furnitures and many valuable properties.
(Bokamba 1982:82).
(5) I was in charge of all CORRESPONDENCES.
(Bokamba 1982:82).
Another process that frequently produces differences between British and
\ American English is the creation of divergent verb-phrase collocations, as in (6),
\i from British English. 6
Lowenberg: Standards and norms for World Englishes 125
(6) This envisaged 16 to 20 'technology schools' in big cities, each
CATERING FOR 1000 selected pupils...
(Times of London, 9/15/86:1. in Algeo 1988:25)
Extensions of this process in nonnative varieties are illustrated in (7), from
Indian English, and (8). from the English of both Singapore and Malaysia.
(7) Everyone is dismissing off my career.
(Mehrotra 1982:161)
(8) It is a bit difficult to COPE up with all the work they give us.
(Tongue 1979:56)
A third productive process common to both the native-speaker and non-
native varieties is the coining of neologisms through morphological derivation,
especially prefixation, as in (9), from Indian English, and (10), from the English
of several nonnative varieties that have developed in East Africa.
(9) If a passenger on a preponed flight shows up at the time written on
his ticket and finds the plane has already left, he is not entitled to a
refund.
(Coll 1990:A13)
(10) He OVERLISTENED to the boys' conversation.
(Hancock & Angogo 1982:318)
In (9), prepone is "to decide to do something earlier than expected' (Verma
1982:180); overlisten in (10) means to 'eavesdrop*.
Oi A fourth productive process frequently leading to innovations in both the
native-speaker and nonnative varieties is expansion of the lexicon through
compounding, as illustrated by (11), from Philippine English, and (12), from
Ghanaian English.
(1 1 ) Most of the students here are bed-spacers.
(Gonzalez 1983:158)
(1 2) You have to be careful with these been-to boys.
(Bokamba 1982:89).
Bed-spacers in (11) refers to students who rent a bed in a boardinghouse or
dormitory without eating their meals there. In (12), been-to boys (and girls) are
young Ghanaians who have recently returned from studies and/or employment
in Great Britain and frequently have difficulty readjusting to life in Ghana. In In-
dian English, such people are often called England-returned (Kachru 1982:
363).
126 Studies in the Linguistic Sciences 20:2 (Fall 1990)
Identification of nativized features in primary sources
Of course, these very morphosyntactic processes frequently also underlie
interlanguage features in the second-language acquisition of English (see, for
example, papers in Richards 1974). Relying on secondary sources, as in the
above examples from nonnative varieties, provides no way of ascertaining first-
hand if these examples do indeed reflect nativized features of Standard English
in their respective speech communities, or if they are markers of individual
learners' interlanguages.
On the other hand, the researcher wishing to supplement these reports
with primary data encounters a methodological problem of determining exactly
which linguistic divergences from native-speaker English are nativized features.
Since the majority of users of nonnative varieties still do learn English as a
second language, and since many nativized features arise from a subset of the
linguistic processes which also underlie interlanguage, it is important to dis-
tinguish nativized features that create differences between varieties, such as
the differences between British and American English, from interlanguage de-
ficiencies in the acquisition of nativized Standard English by learners of non-
native varieties.
In some cases, making this distinction is quite easy, as in (13) and (14).
(13) Asia's longest bridge and rank third in the world. The $850 million
concrete bridge link Penang Island to Peninsula Malaysia.
(14) Citizen makes your office calculation more easier.
(New Straits Times, 9/16/87:12)
(13) is the caption on a Malaysian postcard, while (14) Is an advertisement that
appeared in a leading Malaysian English-language newspaper. Neither of
these deviations from native-speaker Standard English arises from the produc-
tive processes underlying (1) through (12) above. Rather, both reflect either
aberrant mistakes or some stage in their authors' acquisition of English as a
nonnative language.
At the other extreme, certain deviations from native-speaker English can
be confidently proposed as being nativized norms in that they have been
institutionally codified by the same types of authorities who make such de-
cisions in the native-speaker varieties. Examples of this occur in the highlighted
constructions in (15) and (16), from Malaysian ESL textbooks published by
Oxford University Press and by the Malaysian Ministry of Education, res-
pectively.
Lowenberg: Standards and norms for World Englishes 127
(15) Give your book IN.
(Howe 1974:125)
(1 6) A CONSIDERATION for Others is most important.
(Koh & Leong 1976:238)
In other cases, newspaper style sheets reveal evidence of stabilized
constructions, as in (17) and (18), from the style sheet of The Straits Times,
Singapore's leading English-language newspaper.
(17) She lives in 6th Avenue.
(Straits Times Press 1985:4)
(18) I live in an apartment at Belmont Road.
(Straits Times Press 1985:177)
In fact, (17) and (18) may not be nativized at all, but instead reflect British
English, as in (19).
(19) Entrance in Sherwood Street.
(Algeo 1988:13)
Between these extremes of codification and clear deficiency comes the
more problematic area of identifying nativized norms and distinguishing them
from interlanguage. In many cases, extensions of productive processes in Eng-
lish may not yet be codified, but their acceptability is enhanced through use by
writers whose scholarship is highly regarded. This also occurs in the native-
sp'eaker varieties, as illustrated by the construction knowledges in (20) and (21),
which would be considered unacceptable by many speakers of Standard Am-
erican English.
(20) Equally certainly, twenty-five authors and two editors do not know
enough to write this book, and by virtue of knowledges and view-
points they may not provide as cohesive a book as a single author.
(21) In the cultural and academic spheres, one finds national knowl-
edges and discourses coexisting with Continentalist constructs ...
However, upon learning that (20) was written by Charles Ferguson and
Shirley Brice Heath (1981 :xxxviii) and that (21) appears in a paper by their
Stanford colleague Mary Louise Pratt (1986:34), readers familiar with these
scholars' work would be slower to reject this plural form as ungrammatical,
especially when used in the registers of these writers' domains of expertise.
The spread of the form knowledges through the writing of other scholars in the
language sciences could lead to a change in the norms of this register of Stan-
dard American English.
128 Studies in the Linguistic Sciences 20:2 (FaU 1990)
Such acceptance of knowledges on the basis of the stature of its authors
motivates a similar response to switchings in (22) and (23), both of which were
written by the prominent Malaysian linguist Asmah Haji Omar (1985:20,22),
whose status among Southeast Asian language specialists is equivalent to that
of the American authors of (20) and (21 ).
(22) In this context, there were variations such as (code) switchings
between English and their own language.
(23) Intrasentential code-switching may take place in a formal or semi-
formal situation, like at official meetings, seminars or conferences.
Most SWITCHINGS at these levels take place between standard Malay
and formal Malaysian English.
As with l<nowledges above, if other Malaysian linguists likewise begin to use
switchings, this construction could become a nativized feature in this register of
Standard Malaysian English.''
Even when the sociolinguistic status of the individual authors is unknown,
identification of registers within specific domains can be useful in determining
the acceptability of particular constructions. For example, (24) appears several
times on my children's box of Crayola chalk, manufactured and packaged in the
United States.
(24) 1 2 Crayola colored chalks.
(Binney and Smith, Inc., Easton, Pennsylvania)
Notice that chalks here is identical to the Singapore/Malaysian nativization in
(3) above.
This criterion of domain of use can likewise be applied in distinguishing
possible nativized features from acquisitional deficiencies in the nonnative vari-
eties. Examples (25) and (26) are taken from the front news sections in two of
Malaysia's leading English-language newspapers.
(25) Complaints of threats and intimidations have surfaced and these
could affect the security situation in the State.
(New Straits Times, 5/1/86:1)
(26) That way the forms would be filled and processed within minutes,
rather than have the passengers fill up all the details while at the
checkpoint.
(The Sunday Star. 3/31/85:2)
Lx)wenberg: Standards and norms for World Englishes 129
Appearing where they did in these prestigious newspapers, these passages
were probably written, or at least edited, by highly proficient Malaysian users of
English. This further suggests that many of the educated Malaysian speakers of
English who read these newspapers might not object to the constructions intim-
idations and fill up. On this basis, intimidations and fill up might be considered
possible nativized features of Malaysian English. ^
However, intimidations and fill up might just as well be mistakes. In the
absence of repeated occurrences of an item by an author, as in switchings in
(22) and (23) above, it is important, when possible, to consult with authors as to
whether they intended to produce such innovations. For example, Ferguson
and Heath (p.c.) did indeed use knowledges intentionally in (20). On the other
hand, this was not the case with correspondences in the Georgetown University
employment advertisement in (26).
(27) Editor/Writer II - Incumbent will research, write, edit, and supervise
the production of direct mail projects to include brochures and
CORRESPONDENCES of Alumni Annual Fund, the Hospital Annual
Fund, and phone/mail.
(Chronicle of Higher Education, 10/24/90:645)
Though correspondences appears in a conjoined nominal phrase that is struc-
turally similar to those in (20), (21), and (25), and is identical to the Ghanaian
nativization in (5), the author of this advertisement reports (p.c.) that he had no
intention of using this plural form, and that it was most likely a typographical
error that occurred during the publication process.
Nevertheless, despite such methodological problems in identifying nativ-
ized features, from this analysis jimerge several heuristics for identifying nativ-
ized norms for Standard English in nonnative varieties. First, many nativized
features result from a limited number of the productive linguistic processes that
also produce differences among the native-speaker varieties of English. 6ec-^
ond, in some instances, nativized features have been codified by institutions
having authority over domains of Standard English, such as government-au-
thorized textbooks and newspaper style sheets. Third, deviations from native-
speaker norms may be considered as possible nativized features when pro-
duced by English speakers with high status in the relevant speech community
and/or when appearing in texts likely to have been written and edited by
speakers who are highly proficient in English.
A fourth heuristic, illustrated by knowledges \ri (21) and (22) and by switch-
ings in (23) and (24), is the frequency of use of particular innovations by educ-
ated speakers of English. For example, a developing nativized norm in Stan-
dard Malaysian English may be hotting up in (28), which I've now seen three
times in the New Straits Times.
130 Studies in the Linguistic Sciences 20:2 (Fall 1990)
(28) With three days to go before acceptance, the battle for Umno Youth
Exco seats is hotting up.
(New Straits Times 6/1 4/80:1 )
In fact, hotting up appears to occur across nonnative varieties, as suggested by
its use in (29), from a major Indian English-language newspaper.
(29) Gujarat scene hots up again.
(Hindustan Times, 7/13/89:1)
Implications for assessing nonnative English proficiency
These heuristics for distinguishing nativized features from errors and mis-
takes have important implications for the assessment of nonnative speakers'
proficiency in English. Most obviously, such analyses suggest limits on how far
it can be assumed that norms of Standard English in any variejy extend to other
varieties, native or nonnative.
One practical implication of this observation concerns the international val-
idity of certain items in the Test of English for International Communication
(TOEIC), which the Educational Testing Service (ETS) has been administering
since 1979. In its Bulletin of Information for the TOEIC, the Educational Testing
Service (1987:2) describes the TOEIC as 'designed to test the English lan-
guage as it is used internationally in business, commerce, and industry'.
Further, ETS points out that the TOEIC differs from the Test of English as a
Foreign Language (TOEFL), which 'is designed to determine how well a
candidate can use English in colleges and universities in the United States'
(ETS 1986:3). Thus, whereas the TOEFL is based on the norms of Standard
American English, the TOEIC is implicitly unbiased toward any variety of Stan-
dard English.
ETS publications concerning the TOEIC do not mention whether the
norms considered acceptable for international use are only those from native-
speaker varieties or include those from nonnative varieties as well. However,
through personal correspondence with ETS, I have been advised that only
norms from the native-speaker varieties are accepted.
Apparently working within this bias, a commercially published practice g
book for the TOEIC considers item (30) incorrect, though it would be acceptable I
in Standard Singapore English, as in (17) and (18), and even in Standard
British English, as in (19) above.
(30) He lives in Main Street.
(Lougheed 1986:13)
Lxjwenberg: Standards and norms for World Englishes 131
Cross-varietal differences likewise become a factor in (31), a practice
problem in the official ETS bulletin for the TOEIC. The student's task here is to
identify which underlined item is ungrammatical.
(31 ) Please fill out the enclosed form to tell us how you think about
our service.
(ETS 1987:18)
The ungrammatical item here is considered to be how. However, the con-
struction fill out might well also be unacceptable to a candidate accustomed to
filling up a form, as in (26) above or in Standard British English (see Note 8).
For such a candidate, (31) could be a very troublesome problem to answer, as it
would contain two errors.
Examples (32) and (33), similar in examinee task to (31), are actual test
items from a retired 1980 version of the TOEIC now available from ETS (1980:
27-28).
(32) HIS proposal met with a lot of resistances.
(33) The new equipments shipped from Hong Kong will be the only
ITEMS ON sale this week.
Resistances and equipments, which result from the same productive process
that yield (1) through (5) above, may well be acceptable to educated speakers
of particular nonnative varieties of Standard English.
Beyond the realm of standardized tests, educators charged with evalua-
ting the English proficiency of speakers of nonnative varieties can attempt to
distinguish deficiencies in the acquisition of English by these students from
varietal differences in the students' usage resulting from their having previ-
ously learned nativized English. For example, among such students in the
United States, possible nativized features would include systematic deviations
from Standard American English which result from morphosyntactic processes
which are also productive in native-speaker varieties, such as the ones
discussed above, and which highly educated English users in the students'
home countries might therefore use in domains of Standard English.
An illustration of how this distinction might be made comes from analysis
of (34) through (37), from papers written by Malaysian graduate students in
linguistics at Georgetown University.
(34) For example, when the first time I came here, I did not have enough
vocabularies...
(35) In the past, several interesting research had been conducted ...
1 32 Studies in the Linguistic Sciences 20:2 (Fall 1990)
(36) Forty college-educated MEBs studying in the Washington, DC, and
Northern Virginia area were the subject of a research entitled ...
(37) I would like to RECALL back the process I went through ...
All of the highlighted items would be considered incorrect by most speakers of
Standard American English. In (34) and (35), when the first time I came here
and several interesting research do not result from the productive processes
discussed above and are therefore most likely either performance mistakes or
interlanguage errors in all varieties of Standard English. However, vocabula-
ries in (34), a research in (36), and recall back in (37) do result from the
productive processes discussed earlier. These constructions could be nativized
features that the students were taught in Malaysia, they could be mistakes, or
they could be acquisitional deficiencies. Therefore, in assessing the students'
English proficiency, vocabularies, a research, and recall back cannot be evalu-
ated as quickly and clearly as can when the first time I came here and several
interesting research.
Another, deeper implication of becoming sensitive to nativized features
arises in (38).
(38) We often exchange our knowledges.
The highlighted construction here appears to be identical to (20) and (21)
above, from Ferguson & Heath and from Pratt. However, the author of (38) is a
Georgetown linguistics graduate student from Japan, where no nonnative vari-
ety has yet been identified. Is (38), therefore, a mistake or a marker of inter-
language? If so, do (39) and (40), written by Georgetown linguistics graduate
students from the United States, also reflect interlanguage?
(39) As A HOMEWORK, Students chose ten words or phrases to write in
sentences.
(40) A slash (/) between terms indicates both were present in the data, an
EVIDENCE of variable assimilation ...
Or does (40) seem preferable to (39) once we know that evidence has similarly
been made countable by the eminent American psycholinguist Jean Berko
Gleason in (41).
(41) Parents' eagerness to teach their 6-month-old children the pre-
linguistic routine 'bye bye' is one evidence of their desire to show
that their baby is on its way to being a socialized person.
(Berko Gleason 1988:276)
Lowenberg: Standards and norms for World Englishes 133
Apparently, as indicated by many of the data examined above, the basis
for distinguishing between differences and innovations, on the one hand, and
DEFICIENCIES in Standard English, on the other, can be extremely attitudinal a^
well linguistic. " — ~°-T^;::r^!^5sr=-^''-^"""
Conclusion
C>
The present analysis and discussion, though suggesting heuristics for
identifying nativized features in nonnative varieties of English, also illustrate
how little can actually be determined on the basis of such limited data as are
presented here. Research on these varieties has not yet advanced to the point
of being able to identify all, or even most, of the nativized features in any variety.
Many conceptual and methodological problems remain. For one thing, a much
broader data base will be necessary. ^ Equally important for the identification of
nativized features in a particular vahety will be judgments of the acceptability of
specific innovations in that variety by that variety's most highly educated speak-
ers.
In addition, the scope of analysis of nonnative varieties must be extended
beyond morphology and syntax to other linguistic levels. For example, research
by B. Kachru (1986) and Y. Kachru (1988) has revealed significant register-
specific stylistic differences in Standard English between nonnative varieties
and the native-speaker varieties.
Nevertheless, even with the great deal that remains to be learned about
nonnative varieties of English and the dynamics of nativization, insights from
research to date can already have practical applications for irhproving tests
measuring nonnative proficiency in English, and for analyzing nonnative
English speakers' deviations from native-speaker norms. On a more theoretical
level, such research will be crucial in writing a truly comprehensive grammar of
English as a world language, and should yield valuable insights on basic pro-
cesses of language vahation and change.
NOTES
I The term 'institutionalized nonnative varieties' originates with Braj
Kachru (e.g. 1986). These varieties have developed in countries formerly
colonized by Britain or the United States where English continues to be used by
substantial numbers of nonnative speakers as a second, often official, language
in a broad range of iNTRAnational domains. Among these countries are Bangla-
desh, Botswana, Brunei, Cameroon, Ethiopia, Fiji, The Gambia, Ghana, India,
Israel, Kenya, Lesotho, Liberia, Malawi, Malta, Mauritius, Myanmar, Namibia,
Nauru, Pakistan, Seychelles, Sierra Leone, Singapore, South Africa, Sri Lanka,
134 Studies in the Linguistic Sciences 20:2 (Fall 1990)
Sudan, Swaziland, Tanzania, Tonga, Uganda, Western Samoa, Zambia, and
Zimbabwe (Encyclopedia Britannica 1986:838-41; McCallen 1989:7-9). In
many of these countries, English is still used for some of the legislative, admin-
instrative, and judicial functions of government and is the principal medium of
instruction, especially in secondary and postsecondary education.
2 A source of confusion in recent discussions of nativization has been the
use of the term by other linguists to describe various aspects of language acqui-
sition and use. For Sankoff (1980) nativization refers to the first-language ac-
quisition of a pidgin by children, a development traditionally considered to co-
incide with creolization. Also related to pidgins is Todd's (1984:15) use of na-
tivization in reference to the 'period of expansion and stabilization' of a pidgin
when it is used as a lingua franca, 'a period when the local people [make] the
pdigin serve their purposes.' Andersen (1979, 1980, 1981) expands the do-
mains of nativization to include all language acquisition: the 'acquisition to-
wards an internal norm' (1980:273) of any target language — first or second, in-
cluding, but not restricted to, pidgins and Creoles — by individuals or groups.
3 Following Trudgill (1983) and Tay & Gupta (1983), the standard model of
a variety of English — native or nonnative — is here defined as the linguistic
forms of that variety that are normally used in formal speaking and writing by
speakers who have received the highest level of education available in that
variety. Standard English is the accepted model for official, journalistic, and
academic writing and for public speaking before an audience or on radio or
television.
'^As in Lowenberg 1989, this analysis is of morphosyntactic features in
Standard English because these can be easily identified and classified for
cross-varietal comparison; because they have already been well described in
native-speaker varieties; because authoritative prescriptive norms are frequent-
ly available in school textbooks and newspaper style sheets; and because
these forms are addressed in most assessments of English profiency. The
focus is on written, as opposed to spoken texts, since regional phonetic and
phonological processes can often mask the realization of morphosyntactic
standards, and since written language has a greater likelihood of being suc-
cessfully monitored or edited, making possible a distinction between mistakes
and acquisitional errors.
For discussions of nativization processes at other linguistic levels in non-
native varieties, see Piatt, Weber, & Ho 1984, Braj Kachru 1986, Lowenberg
1986a, 1986b, 1991, Smith 1987, and Cheshire 1991.
5 As Henry Widdowson notes (p.c), many of these noncountable-to-count-
able conversions are register-specific. For example, the register of real estate
in Standard American English includes real estate properties, a construction
identical to that in (4). Additional examples of count/noncount differences be-
Lx)wenberg: Standards and norms for World Englishes 135
tween British and American English are given in Schur 1987, Algeo 1988, and
Lowenberg 1989.
6 In American English, the more usual collocation would be catering to.
See Svejcer 1978, Trudgill & Hannah 1985, Schur 1987, Algeo 1988, and
Lowenberg 1989 for additional examples of differences between British and
American English in verb-phrase collocations.
7 I am grateful to Chin-W. Kim and Hans Henrich Hock for pointing out that
knowledges in both (20) and (21) occurs as part of conjoined noun phrases in
which the other noun is the plural form of a count noun: knowledges and view-
points in (20) and knowledges and discourses in (21 ). This is not the case with
switchings in (322) and (23), suggesting that the processes underlying these
mass-to-count changes may not be the same in (22) and (23) as they are in (20)
and (21). Note however that (25) below, from Malaysia, appears in the same
type of construction as do (20) and (21).
8 Fill up also frequently appears in Standard British English (Schur 1987:
135).
9 Sidney Greenbaum of the University of London is currently compiling
such a data base of nonnative varieties of English, a data base which will even-
tually include data from fifteen such varieites around the world (Braj Kachru,
p.c).
REFERENCES
Algeo, John. 1988. British and American grammatical differences. Inter-
national Journal of Lexicography 1:1.1-31.
Andersen, Roger. 1979. Expanding Schumann's pidginization hypothesis.
Language Learning 29.105-119.
1980. Creolization as the acquisition of a second language as a first
language. Valdman and Highfield 1980:273-295.
1981. Two perspectives on pidginization as second language acquisition.
New dimensions in second language acquisition research, ed. by R.
Anderson, 165-195. Rowley, MA: Newbury House.
ASMAH Haji Omar. 1985. Patterns of language communication in Malaysia.
Southeast Asian Journal of Social Science 13:1.19-28.
Bailey, Richard, & Manfred Gorlach (eds.) 1982. English as a world language.
Ann Arbor, Ml: University of Michigan Press.
Berko Gleason, Jean. 1988. Language and socialization. The development
of language and language researchers: Essays in honor of Roger Brown,
ed. by F. K. Kessel, 269-80. Hillsdale, NJ: Lawrence Eribaum Associates.
1 36 Studies in the Linguistic Sciences 20:2 (Fall 1990)
BOKAMBA, Eyamba. 1982. The Africanization of English. The other tongue, ed.
by B. B. Kachm, 77-98. Urbana, IL: University of Illinois Press.
Cheshire, Jenny (ed.) 1991. English around the world: Socioiinguistic per-
spectives. Cambridge: Cambridge University Press.
Coll, Steve. 1990. India's terminal frustration. The Washington Post, January
15:A13.
Educational Testing Service. 1980. Test of English for International Com-
munication [retired version]. Phnceton: Educational Testing Service.
1986. Guide for TOE IC users. Phnceton: Educational Testing Service.
1987. Bulletin of information. Test of English for International Com-
munication. Princeton: Educational Testing Service.
Encyclopedia Britannica. 1986. Bhtannica Book of the Year, 1986. Chicago:
Encyclopedia Britannica, Inc.
Ferguson, Charles A., & Shirley Brice Heath. 1981. Introduction. Language in
the USA, ed. by C. A. Ferguson & S. B. Heath, xxv-xxxviii. Cambridge:
Cambridge University Press.
Gonzalez, Andrew. 1983. When does an error become a feature of Philippine
English? In Noss 1983:150-172.
Hancock, Ian, & Rachel Angogo. 1982. English in East Africa. Bailey &
Gorlach 1982:306-323.
Howe, D.H. 1974. New guided English, Book 1. Kuala Lumpur: Oxford Uni-
versity Press. [Repr., Penerbit Fajar Bakti Sdn. Bhd., Kuala Lumpur, 1986.]
Kachru, Braj B. 1976. Models of English for the Third World: White man's
linguistic burden or language pragmatics? TESOL Quarterly 10:2.221-239.
1982. South Asian English. Bailey & Gorlach 1982:353-383.
1986. The alchemy of English. Oxford: Pergamon Press.
Kachru, Yamuna. 1988. Interpreting Indian English expository prose. IDEAL
3.39-50.
KOH, Judy, & Leong Suat Chin. 1976. Dewan's communicative English, Book
4. Kuala Lumpur: Dewan Bahasa dan Pustaka, Ministry of Education.
LouGHEED, Lin. 1986. English, the international language. Tokyo: Addison-
Wesley.
LOWENBERG, Peter H. 1986 (a). Non-native varieties of English: Nativization,
norms, and implications. Studies in Second Language Acquisition 8:1.1-
18.
1986 (b). Sociocultural context and second language acquisition: Ac-
culturation and creativity in Malaysian English. World Englishes 5:1.71-83.
1989. Testing English as a world language: Issues in assessing
nonnative proficiency. Language teaching, testing, and technology:
Lessons from the past with a view to the future, ed. by J. A. Alatis, 216-227.
Washington, DC: Georgetown University Press.
— . 1991. The nativization of English in Malaysia: A case study in language
pragmatics. Cheshire 1991:364-375.
McCallen, Brian. 1989. English: A world commodity. London: The Econo-
mist Intelligence Unit.
Lowenberg: Standards and norms for World Englishes 1 37
Mehrotra, R R.. 1982. Indian English: A sociolinguistic profile. Pride 1982:
150-173.
NOSS, Richard B. (ed.) 1983. Varieties of English in Southeast Asia. Singa-
pore: SEAMEO Regional Language Centre. [Anthology Series, 11.]
Platt, John, Heidi Weber, & Ho Mian Lian. 1984. The new Englishes. Lon-
don: Routledge & Kegan Paul.
Pratt, Mary Louise. 1986. Comparative literature as a cultural practice.
Profession 86.33-35. New York: Modern Language Association of Ameri-
ca.
Pride, John (ed.) 1982. New Englishes. Rowley, MA: Newbury House.
Richards, Jack C. (ed.) 1974. Error analysis: Perspectives on second langu-
age acquisition. London: Longman.
Sankoff, G. 1980. Variation, pidgins, and Creoles. Valdman and Highfield
1980:139-164.
Schur, Norman W. 1987. British English, A to Zed. New York: Facts on File
Publications.
Shaw, Willard. 1981. Asian student attitudes toward English. English for cross-
cultural communication, ed. by L. E. Smith, 108-122. New York: St. Mar-
tin's Press.
Smith, Larry E. (ed.) 1987. Discourse across cultures: Strategies in World
Englishes. New York: Prentice Hall.
Sridhar, Kamal K., & S. N. Sridhar. 1986. Bridging the paradigm gap: Second
language acquisition theory and indigenized varieties of English. World
Englishes 5:1.3-14.
Straits Times Press. 1985. Our style: The way we use words in The Straits
Times. Singapore: Straits Times Press, Ltd.
Svejcer, Aleksandr D. 1978. Standard English in the United States and Eng-
land. The Hague: Mouton.
Tay, Mary W. J., & Anthea Fraser Gupta. 1983. Towards a description of
Standard Singapore English. Noss 1983:173-189.
Todd, Loreto. 1984. Modern Englishes: Pidgins and Creoles. Oxford: Basil
Blackwell.
Tongue, R. K. 1979. The English of Singapore and Malaysia. Singapore:
Eastern Universities Press.
Trudgill, Peter. 1983. Sociolinguistics (Revised edition). Harmondsmith,
England: Penguin Books.
, & Jean Hannah, 1985. International English: A guide to varieties of
Standard English (Second edition). London: Edward Arnold.
Valdman, A., & A. Highfield (eds.) 1980. Theoretical orientations in Creole
studies. New York: Academic Press.
Verma, S. 1982. Swadeshi English: Form and function. Pride 1982:174-187.
studies in the Linguistic Sciences
Volume 20, Number 2, Fall 1990
SYNTACTIC THEORY AND LANGUAGE ACQUISITION:
A CASE AGAINST PARAMETERS*
Gabriella Hermon
(University of Delaware)
1.0 Introduction
Syntactic theory over the last ten years has developed from a theory of
rules, filters, and constraints to a theory of phnciples and parameters. The
change from a rule-based to a principles-based theory is a very significant one
and has triggered changes in the theory of language acquisition. It has re-
shaped the way psychologists and linguists have come to think about the
problems facing the child who is acquiring a particular language. While in the
rule-based system, the question centered around the issue of how the child
selects a rule from a space of infinitely many rules of some rule writing system,
in the phnciples-andparameters-based approach the task facing the child
seems to be simpler. In this model, the principles of grammar are part of the
innate knowledge the child is born with, part of a segregated mental language
faculty (a language 'organ', to use Chomsky's term) which functions autono-
mously, like the liver or any other organ. Why, then, are all languages not the
same? In order to account for the variation observed crosslinguistically, we
must assume that the principles are slightly under-specified, namely that certain
parameters are left open to be filled in by the child, depending on the particular
language the child is exposed to. Thus the child's job is to calibrate endo-
genous rule schemata (now called 'phnciples') by fixing parameter values that
the innate endowment leaves unspecified.
These universal principles are like templates which limit the choice of wild
guesses that children might make and gives them advance specific knowledge
of certain properties of language. For example, children are born with the
knowledge that the order of head vs. complement has to be fixed and that it is
usually the same across all head + complement relations, across categories.
Thus, for example, discovering that complements precede heads in the VP will
force a child learning Quechua or Japanese to set the head parameter one way,
while the English learning child will set it a different way. Setting the head di-
rectionality parameter should have a sweeping effect on the whole grammar for
140 Studies in the Linguistic Sciences 20:2 (Fall 1990)
the child since, given the principles of X'-theory, this affects not just the order of
verb and object, but also the order of head nouns and their complements, and
of adpositions and their complements (i.e. all heads and complements in the X'-
schema). This is obviously an oversimplification, but illustrates the 'power' of
parameters. A somewhat more complex version of the head-directionality para-
meter, which subdivides the parameter to take into account the directionality of
Case and theta marking, is given in Travis 1987:
(1) The Headedness parameter (branching direction) (Travis 1987)
Case theta headedness
a. PP2 V NP PP1 Chinese — right final
b. V NP PP1 PP2 English — right initial
c. NP V PP1 PP2 Kpelle left — initial
d. PP1 PP2 NP V Japanese left — final
As mentioned above, as far as the child is concerned, finding the correct
setting for a parameter is considered an easy task. Parameter setting is typic-
ally viewed as triggering rather than learning; i.e., particular settings are
selected (from a list of possible options) rather than learned. This view is per-
haps best summarized in Lightfoot 1989. The general view adopted by Light-
foot and other researchers in this paradigm is that it is not that the environment
shapes the grammar (the language organ) but that the organ selects certain op-
tions which are specified to begin with (head initial or final in the simple ex-
ample given above). There is no real learning involved, just selection of rele-
vant stimuli from the environment.
In the last five years, researchers have begun to address a number of
problems related to the notion of parameters. This research addresses the is-
sue of identifying the relevant parameters, separating the parameters from one
another and from the universal principles of grammar, and, crucially, giving a
plausible account of how these parameters are fixed in the course of language
acquisition by the child. For an overview of this program of research see the
articles in Roeper & Williams 1987 and Frazier & De Villiers 1990.
1.1 P-parameters vs. R-parameters
In this paper I would like to concentrate on one particular type of para-
meter, and ask the question of whether there is good evidence from syntactic
theory and language acquisition for the existence of this type of parameter.
Note that the word order example given above is an instance of a so-called
'open' parameter: Head-directionality is fixed by early exposure to the lan-
guage. This is an example of what Freidin and Quicoli (1989) called r-para-
meters. R-parameters are the concrete values assigned to category variables in
rule schemata (the X' schema, for example). An additional example of an r-
parameter is the set of parameters which assign the possible category values to
alpha in the rule schema 'move-alpha* (adjunction or substitution).
Hermon: Syntactic theory and language acquisition 141
Another important class are parameters associated with the subsystems of
principles (p-parameters). Here the argument goes: No data can fix the values
since this would involve negative input, which is unavailable to the child. For
example, if one of the values of the definition of Governing category in Binding
Theory is 'subject' or 'root clause' (as discussed below), this cannot be left op-
en, since it must be assumed that these settings develop in the absence of
linguistic experience. If the value is left open, the child learning English would
require negative evidence to set the parameter. Since the absence of negative
evidence in child language is a common assumption (see Morgan & Travis
1988 and Lightfoot 1989 for good summaries), it has been claimed that these
parameters must come preset to a certain value (actually, the most restrictive
value on the list). The child's setting (the preset value) may then differ from the
adult setting if the adult language happens to have a different setting from the
preset value. The issue here then is: What triggers parameter resetting for the
child and how does the child move from one setting to another with only positive
data? This is discussed more fully below.
Another example of a p-parameter is the pro-drop parameter suggested in
Hyams 1986. Hyams claims, based on evidence from English child language,
that child grammar can differ from adult grammar since this particular parameter
comes fixed with an initial setting (that is, the value of the parameter assumed in
advance of experience with a particular language), which happens (in English)
not to be the correct parameter for the adult language. It is important to discuss
why this situation could arise. This parameter dictates whether the language in
question allows null subjects in tensed clauses, as in Italian, or prohibits them
as in English, as illustrated in examples (2-5) below:
(2) a. I am going to the movies,
b. * am going to the movies.
(3) (lo) vado al cinema.
(I) am going to the movies.
(4) a. It seems that John loves Mary,
b. * seems that John loves Mary.
(5) sembra que Gianni ama Maria,
seems that G. loves M.
The English speaking child, who has the parameter preset as +pro-drop,
has no negative input from English to convince her that she is in the wrong
setting. As claimed by Hyams, however, English has other triggers (such as the
existence of overt expletives and the placement of modals) to convince the child
that the initial setting is wrong. Clearly, then, there are sehous issues here of
what the triggering data are, of what initial settings should look like, and of what
determines initial settings. These have been addressed in the literature to
some extent, especially in the works cited above. ^
142 Studies in the Linguistic Sciences 20:2 (Fall 1990)
I would like to take a different approach here and entertain the hypothesis
that the parameters related to principles described in the literature are not de-
sirable and have arisen only as a result of the principles being poorly worked
out. I would like to claim that postulating such parameters (namely p-para-
meters) leads to a number of conceptual problems, such as the question of how
to determine the initial setting of a parameter and the related question of what
triggers parameter resetting (the issue of triggering experience, discussed in
detail in Lightfoot 1989). My claim is that as the principles are better worked
out, the need for such preset parameters will vanish, obviating the need for both
parameter presetting or resetting.
In what follows I will take a look at the principles and parameters proposed
to account for the distribution of anaphors across languages. In section 2, I
claim that the parametric approach to binding theory, as developed in Wexler &
Manzini 1987, is conceptually undesirable on at least two grounds: (a) It forces
us to invent principles such as the Subset Principle, which assist the child in
determining the preset value of the parameter; and (b) it leads to atomization by
forcing us to have separate parameters for various properties of reflexives
crosslinguistically. This is in direct contradiction to the spirit of the principles
and parameters model, a model which has as its explicit aim to shift away from
individual njles which each describe a certain construction or property, to a sys-
tem of principles accounting for more than one property.
In 2.2, I propose an alternative approach, which does not involve para-
meters, and thus avoids some of the problems inherent in the parameterized
approach. In this alternative approach, the child comes equipped with general
principles with no associated parameters. What the child has to figure out is
which lexical items are associated with the principles. What the child is 'learn-
ing' is the lexicon and the lexical and morphological structures of the language.
The grammar will change as the child acquires particular lexical items and their
associated properties. Thus the child may be discovering what properties INFL
has in his language (given that he knows what kind of properties INFL could
have). This could take some time, and, since the structure of INFL interacts with
various principles (as illustrated below), it may be reflected as restructuring of
the grammar. The claim is that children do not differ from adults as far as the
principles involving binding are concerned. They only differ in having an in-
complete knowledge of the lexical and morphological structures of their lan-
guage, and this in turn affects what their grammars look like. Finally, in section
3, I review the evidence from first language acquisition and conclude that there
is no strong support for parameter resetting or for the Subset Principle in the
studies conducted so far.
2.0 Binding properties across languages
Let me start by reviewing some of the facts to be accounted for by a theory
of binding for reflexives. As described in Cole, Hermon, & Sung 1990, one of
Hermon: Syntactic theory and language acquisition 143
the most striking differences between Chinese and English is the fact that while
English reflexives have to obey a clause-mate condition (as illustrated in ex-
ample (6)), Chinese reflexives (and reflexives in other East Asian languages
like Korean) can be indefinitely far away from their antecedents, as shown in (7)
below:
(6) John thinks [Tom knows [Bill likes himself]].
(7) Zhangsan renwei [Lisi zhidao [Wangwu xihuan ziji]].
Zhangsan thinks Lisi knows Wangwu like self.
'Zhangsan thinks that Lisi knows that Wangwu likes himself.'
In (6) the reflexive himself can only refer to 8/7/ in English. In Chinese, however,
the reflexive form ziji can refer to any of the bold-faced NPs. Following standard
terminology, we will call the Chinese type reflexives long-distance (LD) re-
flexives.
Other languages, such as Icelandic, allow LD binding of reflexives with
certain clause types, but not others. Thus in Icelandic, with subjunctive and in-
finitival clauses reflexives can be bound from outside, but with indicative claus-
es only local binding is allowed, as illustrated in (8).
(8) Icelandic (Hyams & Sigurjonsdottir 1990)
a. Jonj segir a9 Peturj raki sigj j
'John says that Peter shaves (SUBJ) himself.'
b. Jonj skipafii Petrij a9 raka sigj ;
'John ordered Peter to shave (INF) himself.'
c. Jonj veit ab Peturj rakar sig-j j
'John knows that Peter shaves (IND) himself.'
In Italian, the bare reflexive se and the possessive reflexive propria can
refer to a subject outside its clause, as long as the embedded sentence is sub-
junctive or infinitival. If, however, these reflexive forms occur inside an indic-
ative clause, they must have an antecedent inside that clause. This is des-
cribed in Giorgi 1984, and illustrated below:
(9) Clauses in subjunctive mood (Giorgi 1984)
a. Credo che [|p Marioj sostenga (SUB) che [|p tu
abbia (SUB) parlato di sej e della sua famiglia in TV]].
'I believe that Mario claims that you spoke about
self and his family on TV.'
b. Giannij suppone che [|p tu creda (SUB) che [|p io
sia (SUB) innamorato della propriaj moglie]].
'Gianni supposes that you believe that I am in love
with self's wife.'
144 Studies in the Linguistic Sciences 20:2 (FaU 1990)
(10) Clauses in Indicative Mood
a. *Credo che [|p Marioj sostenga (SUB) che [|p tu
hai/IND parlato di sej e della sua famiglia in TV]].
'I believe that Mario claims that you spoke about
self and his family on TV.'
b. ?*Giannij mi ha detto che [|p tu sei (IND) innamorato
della propriaj moglie].
'Gianni told me that you were in love with self's wife.'
2.1 A parameter-setting account
To account for this variation, Wexler & Manzini 1987 (henceforth W&M)
have proposed a parameter, called the Governing Category Parameter
(GOP):
(11) Governing Category Parameter (Wexler & Manzini 1987)
Y is a governing category for a, iff
Y is the minimal category which contains a and
a. has a subject, or
b. has an INFL, or
c. has a TNS, or
d. has an indicative TNS, or
e. has a root TNS
Note that English corresponds to the first value (subject means Accessible
Subject), Italian and Icelandic assume value (d), and Chinese reflexives are
value (e). Crucially, W&M argue that values are associated not with languages
but with lexical items in given languages. This is stated in W&M as the lexical
PARAMETERIZATION HYPOTHESIS:
(12) Lexical Parameterization Hypothesis:
Values of a parameter are associated not with particular languages,
but with particular lexical items in a language.
Thus, in Chinese the reflexive form ziji has value (e), but the local reflexive
taziji has value (a).
In Icelandic, the pronoun hann has value (c) (since pronouns have to be
free within Tensed S), but the reflexive sig has value (d), while the local re-
flexive sjalfen sig has value (a). See Hyams & Sigurjonsdottir 1990 for a dis-
cussion of the Icelandic facts for reflexives.
Note the predictions of this for acquisition: The child has to learn the set-
ting of a given parameter not for a language as a whole but for each lexical
item. There is nothing principled about this, and items with the same setting are
not predicted to be learned at the same time. As noted in Safir 1987 in
Hermon: Syntactic theory and language acquisition 145
response to W&M, this leads to the child being very conservative and setting
parameters item by item (the undergeneralization problem).
The question also arises on how a child will eventually determine the
correct setting for a certain parameter in her language. W&M note that the
values stand in a subset relation to each other and the languages fall into
nested, subset relations. Thus a language which chooses value (a) for ana-
phors will be properly contained in a language choosing value (e). For pro-
nouns, the order is reversed: (e) is the most restrictive and (a) the least re-
strictive value. W&M then suggest that the relationship among the possible
values of a parameter be regulated by the subset condition. Informally, the
Subset Condition is a restriction requiring that the languages generated by two
values of any given parameter are a subset of one another, for every given
parameter and every two values of it. The formal definition is given in (13):
(1 3) For every parameter p and every two values /, yof p, the languages
generated under the two values of the parameter are one a subset of
the other, that is, L(p(i)) c L(p(j)). or L(p(j)) c L(p(i)).
W&M then suggest that the theory of learnability includes a restriction to this
effect (the Subset Principle), ensuring that a learner selects the value of a
parameter which generates the smallest language compatible with the data:
(14) The Subset Principle:
The learning function maps the input data to that value of a para-
meter which generates a language:
a. compatible with the input data; and
b. smallest among the languages compatible with the input data.
The Subset Principle can then be used as a principle of markedness in determ-
ining default settings for parameters: The value children should start with re-
gardless of the target language will be the smallest set. If children were to start
with a value which is too large, no recovery is possible, given the no-negative
evidence hypothesis. W&M's statement is given in (15) below:
(15) A given ordering of the values of a parameter is a markedness
hierarchy if and only if the language generated by each value is a
subset of the language generated by the immediately following value
in the ordering.
Note the implications of this for the acquisition of reflexives in various lan-
guages: In actual development the child will pass through a stage in which her
grammar allows only local binding of anaphors. Strictly speaking, this stage
may be very short if the evidence for resetting is very robust, leading the child to
reject the unmarked setting very early. Leaving aside the developmental evid-
146 Studies in the Linguistic Sciences 20:2 (Fall 1990)
ence for this, which I will review shortly, let me point out some basic problems
with the parameterized approach to binding domains.
2.2 Problems with the parameter setting approach to Binding
Theory
The most important problem with this approach is the atomization issue
discussed in Safir 1987. Safir shows that an unwanted consequence of the
subset principle is that parameters have been atomized, i.e., limited to one
feature per parameter, in complete contradiction to the spirit of the theory which
requires parameters to have sweeping effects on various parts of the grammar
once they are set. Let me exemplify this with respect to reflexives.
Looking at the languages in question, the binding domain seems to cor-
relate with other properties of anaphors. (I will limit my discussion to anaphors
from here on.) As is widely observed in the literature, LD reflexives are subject-
oriented; that is, only subjects can be proper antecedents for LD reflexives, as
illustrated in the Icelandic example below.
(16) Subject orientation for LD reflexives (Hyams & Sigurjonsdottir 1990)
a. *Eg sagfli Jonij a3 Marfa hef9i bo9i9 sen
'I told John that Maria had (SUBJ) invited himself.'
b. *Eg lofafli Haraldij a9 raka sigi
'I promised Harold to shave (INF) himself.'
Similar facts from Chinese are discussed in Cole, Hermon, & Sung 1990 and
Sung 1990.2
Another property of LD anaphors (in some but not all languages, see the
discussion below) is the so-called blocking effect. In Chinese, the presence
of a first or second person subject in a clause intervening between the LD ana-
phor ziji and its antecedent has the effect of blocking the LD interpretation of the
reflexive. This is illustrated by (17). In this example, ziji can only refer to Wang-
wu, since the subject of the next clause up, wo '!', prevents ziji from referring to
Zhangsan.
(17) Blocking Effect in Chinese (contrast with (7) above)
Zhangsan renwei [wo zhidao [Wangwu xihuan ziji]].
Zhangsan thinks I know Wangwu like self.
'Zhangsan thinks that I know that Wangwu likes himself.'
Crucially, not all languages with LD reflexives exhibit blocking effects. Thus in
Italian, LD anaphors are not subject to this effect, as discussed below.
Hermon: Syntactic theory and language acquisition 147
In the parameter-based approach to reflexives there is no way to link any
of these properties. W&M therefore propose that subject orientation be cap-
tured as a separate parameter, their proper antecedent parameter (PAP):
(18) The proper Antecedent Parameter
A Proper Antecedent for a is
a. a subject p; or
b. any element (i.
The PAP also obeys the Subset Condition, but note that given the way the Sub-
set Condition is stated, there is no possible connection between these two
parameters. If the two parameters are made into a single parameter, the values
of the new parameter define languages which are not subsets of each other,
thus violating the Subset Condition. For instance, Chinese is value (e) for the
GCP (the least restrictive value) and value (a) for Proper Antecedents (the most
restrictive value), while English is value (a) for the GCP and (a) for antecedents.
It is exactly to avoid any such conflict that W&M introduce the independence
principle:
(19) Independence Principle:
The subset relations between languages generated under different
values of a parameter remain constant whatever the values of the
other parameters are taken to be.
So, if these two parameters were one single parameter (in the grammar),
the new parameter would violate the principles and conditions of the theory of
learnability. In effect then, the W&M approach leads to atomization, and para-
meters are viewed as each addressing one particular feature of the grammar.^
To further illustrate this point, note that the restriction on intervening pot-
ential antecedents (illustrated in (17) above) will have to be stated as a para-
meter which some languages (such as Chinese) instantiate as + and others
(like Italian) instantiate as -, with the + setting being the unmarked or initial set-
ting for the parameter:
(20) Blocking Effect Parameter
All potential antecedents for a must agree in phi features with a.
Given the Independence Principle, there is no way to draw the connection
between these facts. In reality, however, a language cannot choose to have an
anaphor which has setting (e) but no subject orientation. Furthermore, the
Blocking Effect parameter may have a positive value only in languages which
choose a value larger than (b) for the GCP. The parameters stand in an im-
plicational relation with each other, in violation of the Independence Principle:
In order to have a positive value for the Blocking Effect parameter, a given lan-
guage needs to choose settings larger than (b) on the GCP. Moreover, if a Ian-
148 Studies in the Linguistic Sciences 20:2 (Fail 1990)
guage chooses the (e) setting for the GCP it must choose the subject orientation
setting on the PAP, while either setting on the PAP may be chosen for any other
setting but the (e) setting on the GCP. This is very complicated and is exactly
the situation W&M want to exclude. In summary, not only does the parameter- |
setting approach to binding domains lead to atomization, but there is no way to
state the fact that the setting of certain parameters depends on the setting of
others.
2.3 An alternative proposal
I would like to suggest an alternative analysis, which avoids these prob-
lems and is not based on parameters at all. This approach, I claim, is more ap-
pealing conceptually than an approach which involves parameters. Moreover,
the acquisition data reviewed below does not support the parameterized ap-
proach, while it does not contradict the proposed principles-based approach.
The approach adopted here views LD reflexives as a subcase of normal
local reflexives, by analyzing LD anaphors as a chain of local relations. In this
analysis, it is assumed that all non-local reflexives undergo head-to-head
movement at LF (a local rule), in the spirit of Lebeaux 1983 and Chomsky 1986.
This is discussed in detail for Chinese and Korean in Cole, Hermon, & Sung
1990, and refined and extended to other languages in Cole & Sung 1991 and
Sung 1990. Let us assume then that Chinese LD reflexives move at LF in the
manner illustrated in Figure 1 (next page).
Note that no barriers prevent this movement (VP is L-marked by I in each
sentence) and that both the Head Movement Constraint and Relativized Mini-
mality are observed. The movement is optional, giving the various options for
the antecedent, varying from local (movement to lowest I) to LD (movement to
the highest I). Next, the question must be addressed of why this movement is
blocked in English. This is due to the fact that Chinese reflexives are xo heads,
undergoing head to head movement. English himself (and the Chinese local
reflexive form taziji "he-self"), on the other hand, are complex forms (NPs), and
are therefore blocked from undergoing head movement, by general principles
of the grammar, spelled out in Chomsky 1986:
(21) Only xo elements undergo adjunction to xo, X-max elements can
only adjoin to X-max.
The options for LF movement in English are illustrated in Figure 2. I
Hermon: Syntactic theory and language acquisition
149
A
Spec C
CI IPl
r
Zhangsan
VPl not a bamer
I A
nwei / \
renwei
'ihmk' Spec
NP
I
Lisi
12 yP2 /I
V A
V
I
zhidao
dao / \
'know' Spec
not a barrier
CP3
s
C
IP3
T
Wangwu
13
VPS
xihuan
'like'
Figure 1: Movement of reflexives at LF (LF of ex. (7) above, from Cole.
Hermon, & Sung 1990)
150
Studies in the Linguistic Sciences 20:2 (Fall 1990)
Figure 2: LF structure for English (from Sung 1990)
The question, then, is why X-max movement cannot succeed in moving the
English reflexive at LF out of its clause. A phrasal reflexive like himself can
adjoin to IP, but then no further movement is allowed, just as for QR, adjunction
to IP cannot be an escape hatch for further movement. The phrasal form could
adjoin to VP, but it cannot move from VP to Spec of CP2, since it is not an
operator. There are too many intervening barriers to adjoin to the highest VP
directly (CP2 inherits barrierhood from IP in this case).'* At least for Chinese
and English, then, no parameters are needed for governing category. Crucially,
this analysis makes the prediction that only XO-type reflexives can be LD. Given
the constraint on movement cited in (21), X-max reflexives can never be LD.
This seems to be universally the case. The question now becomes one of what
counts as an XO form versus an X-max form.s
In addition, the head-movement analysis for reflexives avoids the problem
encountered by the parameter-setting approach of not being able to connect
among various features like LD reflexives, Blocking Effects, and Subject Orien-
tation. This follows from the fact, that in this approach we need not view these
as separate parameters. By adopting the ideas about feature percolation (de-
veloped in Sung & Cole 1990), the Blocking Effect and Subject Orientation for
LD anaphors are explained. Thus, assuming the principles stated in (22)
(principles which are independently motivated in UG), we can explain the
Blocking Effects in Chinese and their lack in equivalent sentences in Italian.
Hermon: Syntactic theory and language acquisition
151
(22) Feature Percolation Principles [FPP] (Cole & Sung 1990)
a. The features of ail the daughters of a head node will percolate
upward to the mother. Thus, the features of the mother will be
the union of the features of the daughters.
b. In cases of feature conflict among the daughter nodes, the
mother node will have the features of the head.
Assuming that Chinese INFL has no phi-features (a common assumption,
see Hermon & Yoon 1989), while Italian has INFL with fully specified phi-fea-
tures, the FPP predicts that in Chinese the features of the reflexive (an adjoined
position) will percolate to the mother node. Assuming that phi-features for ziji
are base-generated randomly (or picked up via spec-head agreement on the
first cycle from the local subject), the FPP together with the assumption of spec-
head agreement will block sentences in which a reflexive in I disagrees with the
features of the NP in spec of IP. Let us see how this happens. In Figure 3, ziji
(which is arbitrarily marked 3rd person) moves from VPa to l3. In l3 the [+3]
feature is percolated up to the mother node. Since the head I has no features of
Zhangsan
II'
VI
renwei
think'
IP2
VP2
ziji[+31 lv2
â– seir I
zhidao SPEC
"know'
Figure 3: Blocking effects in Chinese (Sung 1990)
its own, no feature conflict can arise. Spec-head agreement applies and the
sentence is grammatical, since 13 and the subject agree in features. Next ziji
152
Studies in the Linguistic Sciences 20:2 (Fall 1990)
adjoins to l2. Again, the [+3] feature percolates up to l2, since I itself has no fea-
tures. This time, there is a conflict between the [+1] feature on the subject and
the [+3] feature on I, and the sentence is ruled out.
In Italian, unlike in Chinese, INFL has phi-features of its own. Given the
FPP, the features of the reflexive are never percolated up to I (the mother node).
This is illustrated in (23), whose tree diagram is given in Figure 4.
(23) Lack of Blocking Effects in Italian
Giannij suppone che tu sia innamorato della
Gianni supposes that you are (SUB) in love with
propria] moglie (Giorgi 1984)
self's wife
'Gianni supposes that you are in love with self's wife.'
Figure 4: Tree diagram of (23)
In (23). propria (base-generated with a [+3] feature) adjoins at LF to l2. In
Italian, however. I has phi-features ([+2] in this example). Hence, given (22b),
only the features of the head percolate up to the mother node, and Spec-head
agreement marks the sentence as grammatical. In (23), propria can then adjoin
Hermon: Syntactic theory and language acquisition 153
to li (which is marked [+3] in Italian). Since the features of the reflexive are nev-
er percolated up, they do not play a role in Italian, and no blocking effects are
found. The prediction is that any language in which INFL can be shown to have
phi-features will not exhibit Blocking Effect. This is not a separate parameter,
but follows from the stmcture of I in a given language and the FPP.
To explain subject orientation, no additional mechanism is needed: Spec-
head agreement explains why at LF it is only the subject which is a potential
antecedent for LD reflexives: Since at LF the reflexive has been moved to I, on-
ly the subject can c-command the reflexive. Note that local (phrasal) reflexives
which exhibit subject orientation (such as Chinese taziji) will have to adjoin to
IP. Once in IP, no further movement is possible, a stipulation necessary in the
grammar anyway. In languages in which local reflexives appear as l-clitics
(such as Quechua -ri- as in riku-ri-n 'he see-self-3pr'), the reflexives are base-
generated in I and then adjoined to the V by V-to-l movement. Again only
subjects are potential antecedents, as discussed in Hermon 1985. Thus noth-
ing special needs to be said about this, and no parameters need to be added to
the grammar. All LD reflexives are predicted to be subject oriented. In addition,
local reflexives may be subject oriented if, in the particular language, the re-
flexive form is base generated in I (as in Quechua) or adjoined to IP.^
2.4 Major differences between the two approaches
While there is no necessity to state any parameters associated with
binding domains, the principles-based approach has a larger number of prin-
ciples, viz. the FPP, Spec-head agreement, the principles determining xo vs. XP
movement, the ban on further movement after adjunction to IP, the ban on
moving through Spec of CP except for wh-movement. These are general prin-
ciples, in the sense that no variation across languages is permitted. The claim
then is that languages can differ with regard to the morphological structure of
anaphors (X° or XP) and the structure of INFL. These are differences tied to
overt morphological features of the language in question; that is, the child has
robust evidence for these differences.^ Crucially, various features of LD and
local reflexives are clustered together and fall out of an interaction between the
principles and the choice of lexical items in the language.
In contrast, the parameter-setting approach does not draw the connection
between the various properties of LD reflexives. In addition to assuming the
parameters discussed in 2.1 (the GCP, the PAP, and the Blocking Effect
parameter), this approach may also have to assume all the principles discussed
in this section, since some of the principles are general principles, not limited to
anaphors. Thus the FPP can be shown to play a role in other parts of the
grammar (see Cole & Sung 1991) and Spec-head agreement is independently
motivated. However, the task for the child, in this approach, is not just to learn
the lexical items, but also to determine for each item an associated parameter
setting.
154 Studies in the Linguistic Sciences 20:2 (Fall 1990)
3.0 Evaluation of the two approaches from an acquisition point of
view
When evaluating two competing syntactic theories one can look at
language acquisition as a guiding principle in choosing between alternative ap-
proaches. It is therefore important to describe what acquisition picture is pre-
dicted by each approach. Contrary to the claims in the literature, I do not think
the parameterized approach is supported by the data.
The parameter-setting approach to reflexives is being used in the literature
as the prime example of how theory fits with acquisition. Given the W&M
approach, with the Subset Principle as a guiding principle in acquisition, the
child is supposed to start out with the most limited setting of each parameter and
then, in face of positive evidence, relax the settings. The subset relation defined
by the parameter translates into a markedness hierarchy with the smallest
language generated as the 'default' or 'unmarked' case, i.e., the value the child
should start with. It is then predicted that children in all languages will pass
through a stage in which reflexives will be local (the (a) setting). As discussed
in Hyams & Sigurjonsdottir 1990, even though as a learning principle the
Subset Principle does not require that the child actually pass through a local
binding stage (since the evidence for a larger setting may be so overwhelming
that he immediately rejects the more limited setting), there has been an implicit
(and often explicit) assumption that this stage will be actually evidenced.
In contrast, the principles-based approach makes no such prediction.
What the child comes 'equipped' with in this approach is the basic principles of
UG, such as the constraints on head movement (the ECP), the FPP, Spec-head
agreement (which is another wrinkle on feature percolation, stating that features
of heads and Specs cannot disagree), and other more ad-hoc principles (such
as no further movement from IP adjunction, no movement from VP to Spec of
CP). What the child has to learn from the input is which lexical items belong to
which class. For example, children have to learn that in Chinese zip is an xo
form, and they need to be able to analyze taziji and himself as XPs. In addition,
children need to know whether INFL in the language has phi-features (overt as
in Italian, or covert as in Danish, as discussed in Sung 1990). For more com-
plicated cases, children will need to learn that certain COMPs have certain
features (such as indicative COMPs in Icelandic and Italian, which block head
movement of reflexives out of certain clause types, as discussed in Pica 1987
and Sung 1990). Thus, children are not predicted to go through any discreet
stage in which all binding is local. Moreover, once they exhibit LD binding, the
Blocking-Effect and Subject-Orientation questions should be resolved too, since
these properties depend on the proper application of head movement and its in-
teraction with features of INFL in the language.
I would like to claim that not only is the parameter-setting approach less
appealing conceptually, it also is not supported from the acquisition studies
Hermon: Syntactic theory and language acquisition 155
done so far. The parameter-setting approach predicts that children will go
through a stage which is unlike the adult language. The principles-based ap-
proach does not predict such a stage; i.e., children (once they determine
whether forms are simple or phrasal) should have the same options for LD
reflexives as adults. Of course, even within this approach there could be
possible delays: Children could have a hard time figuring out which items are
anaphors vs. pronouns, or what features INFL has, or whether a form marked for
third person and accusative is phrasal or xo.
The acquisition studies done so far for English seem to accommodate
either hypothesis. For example, Chien & Wexler (1991) report that by age 3.6
English speaking children prefer a local antecedent for the reflexive. ^ The para-
meters-based theory claims that this is due to the fact that the GC for anaphors
comes preset to domain (a), the smallest domain. The only thing the child
needs to learn is what constitutes an anaphor in English. The principles-based
theory makes the same prediction for English: Once the child learns that Eng-
lish has a phrasal reflexive, local binding is the only option.
The crucial test, then, comes from languages in which the adult setting is
larger than the English subset. Chien & Wexler (1987) and Li & Wexler (1987)
make the claim that children go through a distinct stage in which they strongly
prefer the local anaphor. Let us review these experiments. I would like to claim
that they are inconclusive, since they do not really address the issue of whether
young children allow LD reflexives from the very start. First, let's take a look at
the data from Chinese. 150 children in Taipei were tested on their knowledge
of LD versus local reflexives and pronouns. Note, however, that the test was a
preference test: Children were tested using the 'party game' in which the child
has a party with two puppets and lots of objects and one puppet tells the child to
do something or give something to zip, as in the following example:
(24) The Party Game (Wexler & Chien 1987):
xiao-huozi suo Xiaohua gei ziji yi-zhang tiezi
'The little monkey says that Xiaohua gives SELF a sticker.'
This experiment induces a forced preference: The child is instructed to perform
the action which best interprets the utterance. Chien and Wexler found that
children have a strong bias for local binding. This in itself is not conclusive,
since it only tests for preference rather than for grammaticality. It could be that
children can get the LD interpretation, but that for some reason it is not the
preferred interpretation. Crucially, note that Chien and Wexler found the same
bias with adults, as illustrated in Figure 5 (next page). The data from Chinese is
therefore inconclusive, since it does not show that young children do not have
LD reflexives in their grammar. At best, it shows that both children and adults
have a preference for the local interpretation of ziji. Moreover, we know that
adults find LD reflexives grammatical, even if their use is pragmatically marked
(see Zubin et al. 1989 for a description of when LD reflexives can be used in
156
Studies in the Linguistic Sciences 20:2 (Fall 1990)
Chinese and Korean). Hence, the above experiments may show something
about speakers' preferences rather than grammaticality.
00
90
80.
J
70
y
60
50
â–
40
â–
30
\
20
\
10
L- * — "^
Gl G2
2.06 - 3.00
-h- local antecedent
-â– - long-distance antecedent
-JK- sentence-external referent
G4 G5 G6
— AGE GROUP
6.06 - 7.00
Figure 5: LD Reflexives in Chinese (Chien and Wexler 1987)
The Korean facts suffer from another problem. Here adults choose the
local antecedent only about 38% of the time, preferring the LD interpretation.
Children go from a 60% preference for local at age 3.06 to 100% preference for
local at age 4.06, and stay there up to age 6.06 (the oldest age group in the
study). At this late age, Korean children still have not broadened the GC for the
reflexive. This is not in line with either the parameter-setting or the principles-
based approach. The problem is, I think, that Lee and Wexler are using cak\^ a
form which for many speakers is simply a pronoun (with forced disjoint refer-
ence, rather than an anaphor). For other speakers caki may indeed be both an
anaphor and a pronoun (or possibly a pronominal anaphor). The form to be
tested is really casin, which is a pure anaphor, but not available for some
speakers. Again, the test was a forced choice, in the form of an act-out. (The
puppet was telling the child to do something to caki, which could be the child or
the puppet.^)
In conclusion, neither Chinese nor Korean child language provides sup-
port for the Subset Principle, and it remains to be seem what explains children's
preferences in these languages. Moreover, in Icelandic, children perform like
adults from quite early on; cf. (25) with Figure 6.
Hermon: Syntactic theory and language acquisition
157
(25) Kermitj segir a9 Joni gefi (SUBJ) serj/j bil
'Kermit says that John gives SIG a car'
Gl
2.06
UU"
90-
80
70
â– ^ Local Antecedent
60
^^^^"""^
â– â– - Long Distance Antecedant
50
/
43- Outside NP
40
NV^
-X- No Response
30'
/\.
20-
K. \^ / ^' — -^^^ ^
10.
n.
r^t^S^^ ... *
G2
G3
G4 G5 G6 G7
^ AGE GROUP â–º 6.06
Choice of antecedent for SIG in infinitive sentences.
Figure 6: Choice of antecedent for SIG in infinitive sentences
(Hyams & Sigurjonsdottir 1990)
Thus there is no support for the Subset Principle from these data, and they
are also neutral with respect to the phnciples-based theory. ''o In the principles-
based theory, the child needs to decide whether a certain reflexive form in her
language is Xo or XP. It is therefore important to ask the question of what kind of
evidence the child might consider in making this decision. Clearly, surface mor-
phology is one chterion. For bi-morphemic reflexives the child can safely as-
sume that this form is an XP (as for himself or taziji). Moreover, it seems that this
is not the only evidence available to the child. In a recent paper, Ahid Hestvik
has suggested that modification provides evidence in Norwegian for certain
forms (pronouns in this case) being heads rather than maximal projections. The
idea here is that an NP cannot be modified, whereas an N can. See Hestvik
1990 for details on Norwegian and English. It is sinking that similar evidence is
available to the child in Chinese, as far as reflexive forms are concerned. Thus,
in Chinese, but not in English, it is possible to modify ziji, as illustrated in (25)
and (26).
(25) Wangwu, renwei Zhangshanj bu xihuan zuotian de zijij j
Wangwu think Zhangshan not like yesterday poss self
'Wangwu thinks that Zhangshan does not like yesterday's self.'
158 Studies in the Linguistic Sciences 20:2 (Fall 1990)
(26) Wangwuj renwei Zhangsharij bu xihuan nage cuenxixi
Wangwu think Zhangshan not like that stupid
de zijiji
pOSS SELF
'Wangwu thinks that Zhangshan does not like that stupid self.'
In the above sentences z/y/'is modified by a prenominal modifier, an adverbial in
(25) and a determiner+adjective in (26). Note that the reflexive form is LD since
it can refer either to Zhangshan or Wangwu in the above sentences. This type
of data can show the child that ziji is an N rather than an NP in Chinese. In con-
trast, when complex reflexives are used (like taziji 'he-self') modification is not
allowed, not even with a local reading for the reflexive:
(27) "Wangwu renwei Zhangshan; bu xihuan zuotian de tazljij
Wangwu thinks Zhangshan not like yesterday poss he-SELF
('Wangwu thinks that Zhangshan does not like yesterday's self.')
(28) "Wangwu renwei Zhangshanj bu xihuan nage cuenxixi
Wangwu think Zhangshan not like that stupid
de tazijij
poss HE-SELF
('Wangwu thinks that Zhangshan does not like that stupid self.')
Thus, by the time the child has access to such data, it is easy for him to figure
out whether a certain reflexive form is N or NP. Note, that it is claimed in the
acquisition literature that children in English have access to such information
very early. For example, Bloom (1990a) claims that children, even in their very
first word combinations, almost never say things like big Fred or big he, while
they freely allow adjectival predicates which follow proper names or pronouns.
Bloom claims that this is evidence that children distinguish between N and NP
from very early on, properly restricting modification to N. He then proposes a
theory as to how children could use semantic information to draw the noun/NP
distinction. In discussing crosslinguistic variation. Bloom argues that in langua-
ges like Japanese and Korean, in which pronouns and proper names can ap-
pear with modifiers, children must start out by encoding proper names and pro-
nouns as NPs. Given robust evidence from input, the child then recategorizes
these words as nouns. ''^ Given the data from Chinese, it is conceivable that the
child has robust evidence (from modification) early on, indicating that certain
reflexive forms are Ns while others are NPs (as illustrated in (25) - (28) above).
3.1 Conclusions
I have argued that there may be a way to reduce the number of para-
meters suggested in the literature by reanalyzing the data and allowing them to
fall out from general principles rather than from parameter setting. If one can
make a case that parameters like the GCP and Subject Orientation are epi-
Hermon: Syntactic theory and language acquisition 159
phenomena, stemming from the fact that we did not have a sound analysis for
the data, we could eliminate some of the problems discussed in the intro-
duction. Specifically, I would like to argue that UG does not consult a learn-
ability module since there is no direct evidence that learners use the Subset
Principle in figuring out which setting of a parameter to adopt. A similar point,
based on rather different arguments from Binding Theory is made in Kapur et al.
1990. Kapur et al. claim that the Subset Principle does not sufficiently restrict
the theory of Binding Domains, while a strong theory of UG does. This leads the
authors to argue against the W&M proposal which regards the Subset Principle
as constraining both first language acquisition of anaphora and the formal lin-
guistic theory of anaphora.
Clearly, I have not shown that all parameters are dispensable. I have
said nothing about 'open' parameters such as head directionality. It seems that
what the child knows in this case is some principles (such as 'fix the order of
head and complement') with a number of possible options. Given positive data,
the options are quickly chosen. Other problems, such as figuring out what type
of INFL one's language has or what the list of head governors is, can be solved
given positive data. This is not parameter setting in the strict sense.
Moreover, I have not attempted to find alternative explanations for all pos-
sible p-parameters in this paper. Specifically, one should reconsider the ques-
tion of the pro-drop parameter and of subjacency.''^ in principle, it would be
desirable to investigate what principles could replace all p-parameters. It would
make the child's job easier, since there would be no independent measures
(such as markedness) to keep in mind. Let me again emphasize that the need
for preset parameters and for resetting arise only in a theory in which there is no
explanation for a certain set of facts. Take LD reflexives again. What forced
W&M to assume the need for a preset unmarked value? Given that all the
values were rather arbitrary (not connected to other features of the language)
the child would be in an impossible situation if the parameter were not preset
the unmarked (most restricted) way. If she were to assume LD reflexives as an
option from the beginning, she would need negative evidence to reset in
English to the more limited option. Since I have claimed that there is simple
positive evidence in all languages (stemming from the categorial status of the
reflexive) and that this interacts with general principles governing the behavior
of anaphors at LF, there is no need for preset parameters. Each child can
determine which language she belongs to by examining the lexicon and the
morphology of that language and by letting forms interact with universal
principles. The notion of parameter becomes obsolete in these cases. In
conclusion, the more explanatory our grammars become, the fewer parameters
we will need.
160 Studies in the Linguistic Sciences 20:2 (Fall 1990)
NOTES
* I would like to thank the following people for their helpful comments and
discussions of earlier versions of this paper: Peter Cole, Katrin Christie, Bob
Freidin, Roberta Golinkoff, Hans Henrich Hock, James Lantolf, Li-mei Sung, and
James Yoon. Special thanks go to Cheng-chi Wang for providing me with the
data from Chinese.
1 I will not discuss the pro-drop facts in this paper. Clearly, if we are trying
to dispense with p-parameters altogether, the pro-drop facts in early English
need to be reinterpreted. See also note 12 for further discussion.
2 Similar facts are discussed for Italian (in Giorgi 1984) and for Danish (in
Vikner1985) .
3 In Manzini & Wexler 1987, it is argued that there is an additional mech-
anism of markedness linking the two parameters. They state that a language
may choose only one marked setting for either the GCP or the PAP. Thus, lan-
guages with LD reflexives (a marked setting for the GCP) will have to choose an
unmarked setting for the PAP (+ Subject Ohentation). In this way M&W can link
parameters in direct contradiction to the Independence Principle. The reasons
for linking two parameters this way seem rather arbitrary. Now the child has to
compute not just markedness (from using the Subset Principle) but must store
the marked/unmarked features (for various parameters) and compare them. For
additional criticism of this approach see also Kapur et al. 1 990.
-* As far as Icelandic and Italian are concerned, I will follow Pica 1987 in
assuming that certain types of Comps (indicative Comp in this case) create an
extra barrier, making head movement impossible out of indicative clauses. In
what follows, I will limit the discussion to Chinese.
5 It seems intuitively clear that a monomorphemic form like Chinese z/y/is
an xo form and that a bimorphemic form like English himself shou\6 count as an
XP. These are the easy cases. It is less clear why Icelandic sig (which is
marked as third person) should count as X°. For a discussion of the possible in-
dications in the grammar for whether something is or is not an XO element see
Hestvik 1990 and the discussion in 3.0 below.
6 Reflexives in Imbabura Quechua appear as verbal clitics and are Subject
Oriented (Hermon 1985). In Chinese, according to Sung 1990, local reflexives
like taziji "he-self" are also subject oriented. We must assume then that the
local forms also undergo LF movement (possibly adjunction to IP in Chinese).
For the reasons discussed in the text, no further movement (out of the clause) is
possible for X-max forms.
Hermon: Syntactic theory and language acquisition 161
'' See also the discussion below about other evidence the child may use in
determining that forms like z/y/are NO heads.
8Chien & Wexler (1991) cite the following figures using the Party Game
strategy: Children's performance of the locality property of reflexives increases
from 56.75% at age 2.6 (chance level) to 80.75% (for say- reflexives) by G3 (3.6)
goes up to 84.25% by G4 (4.0-4.6), and is almost perfect at 6.6.
9 Hyams & Sigurjonsdottir (1990) have suggested that locality restrictions
on some LD reflexives in child language may be related to the fact that in some
languages, early on pronouns are also restricted to local domains (in violation
of disjoint reference). Thus, Korean children may be treating caki as a pro-
nominal anaphor, forcing a local interpretation, since they also allow a local
interpretation for pure pronominals. Li & Wexler claim that children's tendency
to interpret the pronoun with a non-local antecedent is less than chance in the
youngest group (35%) and actually decreases slightly as age level increases
(to 20% at age 6.07).
10 This is not to say that one could not attempt to find more evidence for the
Subset Principle in Chinese or Korean child language. Thus, one could try to
test whether given a certain action or picture, children accept LD reflexives (with
a modified grammaticality judgment task). In such a test, even if the number of
LD reflexives accepted is smaller than the number of local reflexives, the claim
cannot be made that young children do not have LD reflexives in their gram-
mars.
11 Bloom claims that this is similar to the NP-to-noun shift children make in
English when acquiring the syntactic categorization of the word one. Children
may first encode one as an NP, since it is semantically like a pronoun. Only
after receiving positive evidence from input like this one will children categorize
it as a noun. Bloom relies here on the notion that children have innate prin-
ciples (semantic in nature) which initially lead them to categorize all names of
objects and substances as nouns and all pronouns and names for individuals
as NPs (a version of the semantic bootstrapping hypothesis). Only given evid-
ence to the contrary will children in a language like Japanese or Chinese ana-
lyze pronouns as nouns. The same claim can be made with respect to reflexive
anaphors: In English these are NPs (just like pronouns), while in languages
like Chinese anaphors could be either Ns or NPs, depending on whether they
are simple or complex forms.
12 Recently, a number of people have argued that there is no grammatical
pro-drop in early child English. See Bloom 1990b for arguments that missing
subjects in early English are due to performance conditions. Similarly, Valian
(1989) argues that early English differs from early Italian in having many more
subjects. Crucially, in these analyses missing subjects are not due to a preset
parameter which needs to be reset later. As far as subjacency is concerned.
162 Studies in the Linguistic Sciences 20:2 (Fall 1990)
there may be some questions of whether this is an independent phenomenon.
(See the remarks in Rizzi 1989 on this matter.).
REFERENCES
Bloom, Paul. 1990 (a). Syntactic distinctions in child language. Journal of
Child Language 17.343-355.
1990 (b). Subjectless sentences in child language. Linguistic Inquiry
21.491-504.
Chien, Y. C, & Kenneth Wexler. 1987. A comparison between Chinese-speak-
ing and English-speaking children's acquisition of reflexives and pro-
nouns. Paper presented at the 12th Annual Boston University Conference
on Language Development.
1991. Children's knowledge of locality conditions in binding as evidence
for the modularity of syntax and pragmatics. To appear in Language Ac-
quisition.
Chomsky, Noam. 1986. Knowledge of language: It's nature, origin, and use.
New York: Praeger.
Cole, Peter, Gabriella Hermon, & Li-mei Sung. 1990. Principles and para-
meters of long-distance reflexives. Linguistic Inquiry 21 :1.1-22.
, & Li-mei Sung. 1991. Feature percolation in government and binding
theory. Manuscript submitted for publication. Newark, DE: University of
Delaware.
Frazier, Lyn, & Jill DeVilliers. 1990. Language processing and language
acquisition. Dordrecht: Kluwer Academic Publishers.
Freidin, Robert, & Carlos A. Quicoli. 1989. Zero stimulation for parameter
setting. Commentary on Lightfoot. Behavioral and Brain Sciences 12:2.
GiORGi, Alessandra. 1984. Toward a theory of long distance anaphora: A GB
approach. The Linguistic Review 3.307-359.
Hermon, Gabriella. 1985. Syntactic modularity. Dordrecht: Foris.
, & James Yoon. 1989. On the identification of null subjects. Papers from
the Twenty-Fifth Regional Meeting. Chicago: Chicago Linguistic Society.
Hestvik, Arild. 1990. LF-movement of pronouns. Proceedings of NELS 20.
Amherst, MA: GLSA, University of Massachusetts.
Hyams, Nina M. 1986. Language acquisition and the theory of parameters.
Dordrecht: Reidel.
, & Sigridur Sigurjonsdottir. 1990. The development of 'long-distance
anaphora': A cross-linguistic comparison. Language Acquisition 57-93.
Kapur, Shyam, B. Lust, W. Harbert, & G. Martohardjono. 1990. Universal
grammar and learnability theory. MS. Ithaca, New York: Cornell Univer-
sity.
Lebeaux, David. 1983. A distributional difference between reciprocals and
reflexives. Linguistic Inquiry 16.723-730.
Hermon: Syntactic theory and language acquisition 163
Li, Hyeonjin, & Kenneth Wexler. 1987. The acquisition of reflexive and pronoun
in Korean. Paper presented at the 12th Annual Boston University Con-
ference on Language Development, Boston, MA.
LiGHTFOOT, David. 1989. The child's trigger experience: Degree 0-iearnability.
Behavioral and Brain Sciences 12.321-375.
Manzini, Rita M., & Kenneth Wexler. 1987. Parameters, binding theory, and
learnability. Linguistic Inquiry 18.413-444.
Morgan, James L., & Lisa Travis. 1988. Limits on negative information in
language input. To appear in Journal of Child Language.
Pica, Pierre. 1987. On the nature of the reflexivization cycle. NELS 17.483-
499.
Rizzi, Luigi. 1989. On the format of parameters. Commentary on Lightfoot.
Behavioral and Brain Sciences 12:2.
ROEPER, Thomas, & Edwin Williams. 1987. Parameter Setting. Dordrecht:
Reidel.
Safir, Ken. 1987. Comments on Wexler and Manzini. Parameter setting, ed.
by T. Roeper & E. Williams, 77-90. Dordrecht: Reidel.
Sung, Li-mey. 1990. Universals of reflexives. University of Illinois at Urbana-
Champaign Ph.D. dissertation in Linguistics.
, & Peter Cole. 1990. Non-c-commanding antecedents of bound anaphors
and feature percolation in government and binding theory. Proceedings of
the 9th West Coast Conference on Formal Linguistics.
Travis, Lisa. 1987. Parameters of phrase structure. McGill Working Papers in
Linguistics 4.2.
Valian, Virginia. 1989. Children's production of subjects: Competence, per-
formance, and the null subject parameter. Papers and Reports on Child
Language Development, 28. Stanford, California: Stanford University.
ViKNER, Sten. 1985. Parameters of binder and binding category in Danish.
Working Papers on Scandinanvian Syntax, 23.
Wexler, Kenneth, & Rita M. Manzini. 1987. Parameters and learnability in
binding theory. Parameter setting, ed. by T. Roeper and E. Williams, 41-
76. Dordrecht: Reidel.
ZUBIN, David A., S. A. Chun, & N. Li. 1989. Misbehaving reflexives in Korean
and Mandarin. Manuscript. Buffalo: University of Buffalo.
Stuijig? in thg Lingijigtic Scignggg
Volume 20, Number 2, Fall 1990
WHAT ARE APPLIED LINGUISTICS?
S. N. Sridhar
(State University of New York, Stony Brook)
1. Introduction
It's a great honor to be invited to join in celebrating the twenty-fifth anni-
versary of the Department of Linguistics at the University of Illinois. I had the
pnvilege to be a student in this Department at a period when the research done
here was an active contributor to linguistic theory in many ways. To name only
a few examples: Chuck Kisseberth had just contributed to the controversy on
abstractness in phonology; Relational Grammar was being outlined under Jerry
Morgan's supervision; Peter Cole was testing the predictive powers of Relation-
al Grammar; Georgia Green was showing that semantics could account for part
of syntactic irregularity; Braj Kachru was writing his influential papers on code-
mixing and World Englishes; Hans Henrich Hock's book on historical linguistics
was our class handout; Yamuna Kachru's students were the single most active
group of students at conferences on South Asian linguistics; Charles Osgood
was writing his Outline of Abstract Performance Grammar and inviting everyone
around to critique it; Howard Maclay had heralded the coming of the cognitive
revolution; and Ladislav Zgusta kept everything in perspective in his tour de
force History of Linguistics course. Indeed they were good times, though
harassed by overdue term papers, one didn't realize just how good it was to be
a student. I got a sound, broad-based education in linguistics in the Department
and, I am proud and happy to say, that training has stood me in good stead.
I just said, 'broad-based education in linguistics', and that brings me to the
topic of my paper, 'What are applied linguistics?' The plural is not intended
merely to be cute, of course, nor is it merely a terminological question, either.
As I hope to show, it underscores the coexistence of several different concep-
tualizations of this diverse field, reflecting substantive differences that go to the
very heart of our enterprise, namely the conception of what language is, and
therefore, what linguistics is, or ought to be. In addition, the question has far-
reaching institutional dimensions. It affects decisions about what courses we
should offer and require of our students, what questions shall be criterial in our
exams, what kind of faculty we should hire, and what types of research
166 Studies in the Linguistic Sciences 20:2 (Fall 1990)
programs shall be funded. It also has a bearing on the question of the linguist's
social responsibility. Whether we explicitly answer it or not, we presuppose a
particular conception of applied linguistics in our institutional decision-nnaking.
There is also a more practical urgency to the question. The saturation of
the academic linguistics departments in the U.S. has forced graduates to look
for jobs in fields other than theoretical syntax and phonology. The majority of
new job openings now require additional expertise in non-core areas. This
point is proven time and again, including in the job list published in the latest
LSA Bulletin. On the one hand, there are pursuits such as the cognitive science
enterprise; on the other, there are the more 'practical' areas such as, for exam-
ple, TESOL, Bilingual Education, rhetoric, and literacy programs, as well as the
'applied linguistics' component of foreign language programs, such as Spanish
or Italian.
Yet, there is a serious mismatch between the kinds of jobs that seem to be
in demand and the type of training our graduate students receive. As Raskin
(1986:xiii) points out.
The best linguistics programs are having trouble placing even their
best graduates in academic positions in linguistics. Problem: the
nature of these graduates' training, which was exclusively 'pure
linguistics', makes them virtually unemployable in any alternative
professional capacity.
While many departments have instituted courses that address the societal
demands, many others have chosen a variety of 'creative' responses: The term
applied linguistics has been interpreted with a great deal of imagination. In
some departments, courses such as Phonetics, Structure of an Uncommonly
Taught Language, Theoretical Approaches to Language Acquisition, and even
Descriptive Linguistics are put under the rubric of applied linguistics (whether or
not they have any 'applied' content). This window-dressing is apparently
deemed sufficient to claim social relevance, and more pertinently, to justify
program expansion (more faculty lines and resources). We could call this the
taxonomic response. And then there is a new generation of linguists who would
like to be considered applied linguists because they study the hypotheses of
formal linguistic theory with reference to 'real-life' data from, say, second lan-
guage learners. In this approach, the focus is not necessarily on whether the
phenomenon in question is better understood as a result of being studied by a
linguist, but on the fact that linguistic theory gets a new testing ground. The
problems the linguist chooses to work on, the section of the data that is focused
on, and so on, are all determined by the linguist's theoretical preoccupations,
rather than the consumer's priorities. On the other side of the coin are some
applied linguists who feel that they are mislabeled because what they do is as
much linguistic theory as what the formal linguists do.
Sridhar: What are applied linguistics? 167
Additionally, there are some linguists who would like to maintain a dis-
tinction between applied linguistics (which they consider largely synonymous
with language teaching and regard as basically uninteresting, unworthy of seri-
ous scholarly attention) and applications of linguistic theory (understood to
refer, basically, to formal linguistics). Finally, the old, unidirectional, cozy
relationship between theoreticians and applicationists has given way to a more
chaotic state as more and more applied linguists question the relevance of lin-
guistics as it is fashionably practised now and have gone on to invent a parallel,
complementary linguistics more responsive to their applied concerns. Clearly, it
is time to take a fresh look at the scope and foundations of applied linguistics.
2.0 Two views of applied linguistics
2.1 The showpiece
So, what, then, are applied linguistics? Let me start with a familiar dicho-
tomy: the so-called 'narrow' and 'broad' conceptions of the term. In the narrow
conception, applied linguistics is equated with language teaching — mainly
second or foreign, but also first language teaching. S. Pit Corder, writing from
what was once one of the major centers of applied linguistics in the world, the
University of Edinburgh, defined it as follows:
Applied linguistics refers to a set of related activities or techniques
mediating between the various theoretical accounts of human lan-
guage on the one hand and the practical activites of language
teaching on the other (1975:5).
The use of applied linguistics (AL, hereafter) as a synonym for language
teaching seems to have come into vogue in the 1940*s in the U.S., when some
language teaching professionals wished to upgrade their status by association
with the more 'scientific' field of linguistics. (I will return to the irony of this wish
later.) Of course, that was the heyday of the prestige and influence of linguistics
as the most rigorous of the social sciences. Had it not been for this early
identification of AL with language teaching, the whole debate on the scope of
AL may never have materialized.
This synecdochic use of the term for language teaching is a result of the
fact that the most extensive application of linguistics has been in the field of
language teaching. As Peter Strevens (1980:32) put it,
the intellectual support for the massive array of published materials,
teaching techniques, and of professional support for the teacher and
the learner ... is supplied primarily through applied linguistics.
Although language teaching has long been regarded as AL's showpiece
for demonstrating the potential for applications of linguistics, and the con-
168 Studies in the Linguistic Sciences 20:2 (Fall 1990)
tribution of areas like phonetics and grammar are still acknowledged, it is im-
portant to note that the claim to the relevance of formal linguistic theory (which is
the main thrust of graduate student training and research in many established
and aspiring linguistics programs) is not undisputed. For example, in the 1940s
through the 1960s, language teaching subscribed to the notion of the 'centrality
of grammar" (as defined in the structuralist sense). Much of the intellectual
support from linguistics that Strevens refers to in the statement just cited came
in the form of descriptions of language (either of the target language or a com-
parison of the target and native languages, see Sridhar 1980). Halliday et al.
(1964), Corder (1975), and others have claimed that this is the main conthbu-
tion linguists can make to language teaching. However, language description
long ceased to be the central activity in linguistics.
In the '70s, the centrality of grammar was challenged in language teaching
as it was in, e.g., first language acquisition and psycholinguistics. The linguist's
preoccupation with the most elegant formal characterization of the structural
regularities of the language was felt to be (i) too limiting (in so far as it ignored
contextual, communicative parameters), and (ii) too arcane (too abstract to be
psychologically real). This parting of ways was characterized, in kinder and
gentler formulations, as the difference between competence and performance
orientations. While the move toward a theoretically motivated approach to
language needs no defense, it has brought in its wake an altered approach, a
new conceptualization of what language is: The focus is on discovering the
most general, abstract formal patterns, and the prized criterion is internal con-
sistency and compatibility of assumptions and arguments, even if that leads the
models ever further from ordinary language. This is because the goals of the
theoretical linguist and the applied linguist are different: The applied linguist
finds that the most general, rigorous, formally elegant models are often not the
most useful ones in field applications. This situation is succinctly outlined in the
following remarks by Henry Widdowson (1980:74-75), one of the major advoc-
ates of AL:
The relevance of linguistics cannot be taken for granted because it is
not obvious that the way linguists conceive of language is the most
appropriate for language teaching purposes ...
It is possible — even likely — that linguistics, as customarily con-
ceived, may not be the most suitable source for a practical teaching
model of language ...
The main business of applied linguistics should be the establishment
of appropriate concepts or models of language in the pedagogical
domain, without prejudicing the issue by supposing that a relevant
model must inevitably derive from a model of linguistic description in
the technical sense.
Sridhar: What are applied linguistics? 169
Instead of grammar, language teaching professionals have over the last decade
or more, come to regard notions such as communicative competence and prag-
matics as the most useful contribution of 'linguistics' to language teaching and
testing. The irony of this situation is that neither approach originated within the
field of linguistics — the honor goes to anthropological linguists in one case
(Dell Hymes, Gumperz, and others), and to philosophers of language, in the
other (Austin, Searle, and others). More important, neither area has been in-
tegrated into the so-called mainstream of linguistic theory. We therefore have a
situation in which the most important and influential application of linguistics,
language teaching, involves, in part, application of notions disowned or at least
ignored by linguistic theory.
2.2 A basket of many goodies
As opposed to the 'narrow' view, which identifies AL with language
teaching, there is the 'broad' view, in which the term refers to a wide range of
areas outside of phonology, morphology, syntax, and semantics.
In this view, the term includes, in its various uses, two broad categories of
subjects. One category includes sub-fields such as the following: socio-
linguistics, pragmatics, discourse analysis, bilingualism, diglossia, psycho-
linguistics, first and second language acquisition, sociology of language and
ethnography of communication, lexicography, historical linguistics, and dialect-
ology, among others. The second category includes fields such as language
teaching, translation, literary sylistics, language for special purposes, poetics,
language planning, lexicography, speech pathology, reading research, con-
trastive rhetoric, neurolinguistics, computational linguistics, and so on, ranging
all the way to such industrial, 'very, very applied' uses as speaking dolls, voice
typewriters, and sophisticated editing and printing devices.
A moment's reflection reveals that, however flattering, there is something
strange about this laundry list (a sort of everything you always wanted to know
about language but were denied by theoretical linguists). The problem comes
into focus when we ask: What exactly is being applied?
One might suppose that the fields listed above follow from the common-
sense definition of applied linguistics (cf. Kaplan 1980, Ferguson 1987, and
many others), cited here in Bjarkman and Raskin's formulation (1986:xiv):
By linguistic applications we will mean here the use of data, methods
and/or theories accumulated or developed in linguistics to solve the
problem from a different field of study which might need linguistic
expertise.
In the case of fields in the second category above, such as contrastive
analysis, neurolinguistics, or speech pathology, the definition seems ap-
170 Studies in the Linguistic Sciences 20:2 (Fall 1990)
propriate: Phonetics, phonology, morphology, syntax, or semantics is used to
describe, diagnose, or predict situations involving language use or acquisition.
The only problem with this characterization is that it doesn't apply to most
of the fields listed in the first category above, and that includes much of what
passes for, and what theoretical linguists consider, applied linguistics. Let me
elaborate this point. Linguistic theory, as it is conceived and practised in certain
influential journals (e.g., Linguistic Inquiry, Natural Language and Linguistic
Theory) and major textbooks, etc., has little to say about matters such as bi-
lingualism, speech acts, social meaning of language, or functional motivations
for grammatical and lexical choices, to name but a few examples. This is be-
cause linguistic theory has come to be too closely identified with grammatical
theory in the formal sense. These fields seek to go beyond grammar — without
denying the wonderful intricacy and regularity of grammar — to achieve an
understanding of language in a broad sense, that is, as it is situated in time,
place, society, and culture, and interacting with other cognitive processes, and
used for communication, rather than only as a formal object. Fields such as
pragmatics, interactional and variational sociolinguistics, psycholinguistics, and
so on, are therefore not applications of linguistics as much as they are
extensions of linguistics. I will, therefore, call them examples of 'extended lin-
guistics' or 'complementary linguistics', since they complement the study of the
formal aspects of language focused on by what is now regarded as theoretical
linguistics. They contribute toward a broader theoretical model that would ad-
dress significant aspects of language use in real-life situations. We thus have a
situation where many examples of applied linguistics are really not 'applied' at
all, but just linguistics.
3. Better still: What Is linguistics?
Much of what is commonly understood to be AL is, thus, merely an
alternative paradigm of doing linguistics. In this conceptualization, applied lin-
guistics is what linguistics should have been all along. It is in this sense that the
dichotomy between theoretical and applied linguistics is spurious and un-
tenable. A better distinction might be between linguistics (of language in the
broad sense) and formal linguistics (or grammar). These two types of language
study have gone their separate routes because formal linguistics is perceived
as working within a paradigm that leaves much of what is ordinarily understood
to be language out of account.
The study of the formal properties of language has, no doubt, registered
many impressive gains. However, these gains have depended upon a number
of self-imposed limitations or exclusions. While formal linguists consider these
limitations necessary or unavoidable, extended linguists consider them crip-
pling, if not almost fatal. These include:
Sridhar: What are applied linguistics? 171
a. an exclusion of function;
b. an exclusion of performance;
c. an exclusion of context.
The extended linguists' primary complaint against formal linguistics is that
it identifies language with grammar, and linguistic theory with grammatical the-
ory, leading to an exclusive preoccupation with form and disregard of or scep-
ticism toward language use or function. If linguistics is defined as the scientific
study of language, why should it be limited to the study of (basically) syntax,
semantics, morphology, and phonology? Recall Wittgenstein's observation,
'We must plough over the whole of language.* Although the scope of linguistics
began to expand in the 1960's with Chomsky's challenge to American
Structuralism, and extended quite a bit in the hands of the Generative Seman-
ticists, it shrunk again with the introduction of Extended Standard Theory and its
subsequent revisions. And Chomsky has steadfastly asserted the autonomy of
grammar and its independence from considerations of language use and func-
tions. He has even stated that 'language is not a task-oriented device' (1980:
53).
The attempt to provide functional explanations of linguistic structures is
very old. It is found in Panini's work itself, and it has continued through the
ages, in traditional grammar's explanations of structures such the Passive, the
Prague School's Functional Sentence Perspective, Firth's model of a 'socially
realistic linguistics' (see Kachru 1981) and it is witnessing a revival in recent
years (in the works of Dik, Givon, Halliday, Hopper, MacWhinney, Slobin, and
others). It is true that the functionalist approach has as yet achieved only limited
and sporadic success. It is, however, necessary to keep in mind that (i) not too
many linguists are working in this paradigm, and (ii) much of the criticism of
functionalism stems from specialized and by no means uncontroversial as-
sumptions (e.g. grammar as given) and caricatures of the positions being criti-
cized (e.g. that all structures can or must be explained functionally, or that the
inability to functionally explain a single structure will doom the whole enterprise
to failure). As Hopper (1988:132) has pointed out.
The supposed dispute over functionalism is held to consist in the
choice between two positions: either grammar is redundant, being
wholly derivative of function, or function is irrelevant, being a
separate system only partly isomorphic with structure. But the terms
of the debate are skewed ...
The precise formulation of the form-function correspondence, its possibilities
and limits, are central to the issue. There have been a number of promising
hypotheses put forward in this area in recent years (see, e.g.. Hopper's Emerg-
ence-of-Grammar model (1988), Bates & Macwhinney's (1986) Competition
Model, among others). The relationship between form and function is too cen-
172 Studies in the Linguistic Sciences 20:2 (FaU 1990)
trai to the understanding of language to be dismissed on the basis of very
preliminary results.
The exclusion of performance — and the concomitant abstraction and
idealization of data — has engendered considerable scepticism regarding the
relevance of the theory to 'real-life' language. The exclusion of context is bound
up with the exclusion of performance. On the one hand, the claim that grammar
is independent of context is disingenuous: As Gumperz (1981) and many,
many others have pointed out, many empirical findings of formal linguistics, e.g.
the grammatical judgments which furnish the data for syntactic analysis, depend
on speakers' ability to imagine a context in which the sentence could occur. On
the other hand, it leaves linguistic theory tongue-tied in the face of some of the
most pervasive and profound linguistic phenomena, such as bilingual code-
switching and code-mixing, intra-linguistic style shifting, language variation in
general, or even conditions for the successful performance of speech acts. As
Leech has correctly observed (1983:3), 'one result of this limitation of genera-
tive grammar to a strict formalism has been that, since about 1 970, it has been
progressively losing its position as the dominant paradigm in linguistics.' A
number of alternate paradigms have arisen to deal with linguistic phenomena
that linguistic theory would not consider. Leech goes on to list the challengers:
socioiinguistics (with its rejection of Chomsky's construct of 'the ideal native
speaker/hearer in a completely homogeneous speech community'), psycho-
linguistics (with its process orientation), conversational analysis (with its stress
on the primacy of the social dimension of language study), and pragmatics (with
its attention to meaning in use, rather than meaning in the abstract), to name
only some of the more influential ones. Cumulatively, these approaches have
led to a remarkable shift of direction away from 'competence' and toward
•performance' (Leech 1983:4).
Thus, while the successes of formal linguistics in discovering structural
regularities are impressive, they have come at a price: It is arguable that lin-
guistic theory may have become a science at the expense of its subject matter,
namely language as an instrument of communication in real-life situations. The
emergence of the alternate paradigms represents an attempt to fill the void left
by linguistic theory, but 'no comprehensive paradigm has yet emerged as a
successor to generative grammar. A unified account of what language is has
been lost.' (Leech 1983:4)
4. Applied linguistics as complementary linguistics
One point that becomes clear from this discussion is that the current use of
the terms theoretical and applied linguistics is thoroughly misleading. Cur-
rently, the label theoretical linguistics is used to refer to syntax, phonology (and
morphology), and semantics, and everything else is relegated to the domain of
AL. However, as noted above, the other areas (especially the ones discussed
under the category of extended linguistics) are not 'applied' areas, because
Sridhar: What are applied linguistics? 173
what is currently considered theoretical linguistics has nothing to say about
them. Areas such as sociolinguistics, psycholinguistics, and pragmatics are,
therefore, best regarded as complementary linguistics, because they com-
plement the partial account of language currently provided by formal linguistics.
As Dell Hymes (1984) has observed, areas such as sociolinguistics are in fact
basic research. 'As linguists concerned with communication in human groups
we need to go beyond mere description of language usage patterns to con-
centrate on aspects of shared knowledge and cognitive abilities which are
every bit as abstract and general as the knowledge glossed by Chomsky's more
narrowly defined notion of linguistic competence.'
One can go further and point out that the 'applied linguistics' orientation
ought to be the central concern of all linguistics. AL is not a single, unified con-
cept, but there is a common thread that runs through the various applications: a
commitment to empirical data, a contextualized view of language, a functionalist
emphasis, and an interdisciplinary openness. AL is concerned with 'language
in its total human and environmental context'. This context-sensitive, problem-
oriented, performance-friendly approach characteristic of the so-called applied
linguistics has already proved its worth and calls into question the limiting as-
sumptions of formal linguistics. As Gumperz has noted, sociolinguistic research
[of the 1970s] has demonstrated not only that all existing human communities
are diverse at all levels of linguistic structure, but also that grammatical diversity,
multifocality of linguistic symbols, and context dependence are essential com-
ponents of the signalling resources that members rely on to accomplish their
goals in everyday life (1981:324).
Gumperz also points out that the theoretical linguists' insistence on main-
taining a strict separation between linguistic and extralinguistic phenomena has
... become untenable in many key areas of linguistic research (ibid. 325). This
suggests that the boundary between linguistics and other allied disciplines, e.g.,
sociology, anthropology, and psychology may not be clear-cut. A unified ac-
count of language calls for an interdisciplinary approach. This is only to be
expected, for, as Schegloff (1988:155) puts it,
The fabric of the social world does not seem to be woven with seams
at the disciplinary boundaries ... The use of language as a vehicle
for social action binds the features of language and the features of
action and interaction together, at least in part. This requires a theo-
retical stance toward language different from some others which are
current. It implies certain forms of inquiry.
The famous neurologist Oliver Sacks's characterization of an analogous situ-
ation in medicine applies to linguistics with equal force. He notes (1987:40-41)
that advances in modern medicine resulted in 'real gains in knowledge but a
real loss in understanding' because of compartmentalization into motor, intel-
lectual, and affective domains and excessive abstraction associated with 'nar-
174 Studies in the Linguistic Sciences 20:2 (Fall 1990)
row formulations of theories.' A unified, interdisciplinary, account of language
would lead, in contrast, to William James's 'the light of the world's concrete
fulness.'
In this paradigm, the relation between theory and data is different. Lan-
guages would not simply be used as sources of data (isolated sentences,
decontextualized) for testing theoretical notions, but they would be studied in all
their complexity and uniqueness. Saville-Troike (1988:249) puts it this way
I am firmly in the camp of those who believe that theory should be
'grounded' in data. I believe collecting data only to confirm or
disprove a priori hypotheses is likely to exclude crucial evidence for
phenomena which occur in the process of [language acquisition]
which are merely not salient to the investigator.
Labov (1988) makes the same point, but he goes further, and because of
his point's centrality to the issue of the goals of applied linguistics, I give below
some extended excerpts:
When we contrast linguistic theory with linguistic practice, we
usually conjure up a theory that builds models out of introspective
judgments, extracting principles that are remote from observation
and experiment. This is not the sort of theory I have in mind when I
search for a way to establish the facts of a matter I am involved in. It
is hard to imagine that a concept like subjacency or ECP would be
used in court to decide a question of fact ...
Do we gather facts to serve the theory, or do we create theories to
resolve questions about the real world? I would challenge the com-
mon understanding of our academic linguistics that we are in the
business of producing theories, that linguistic theories are our major
product. I find such a notion utterly wrong. (182)
General theory is useful, and the more general the theory the more
useful it is ... But it is still the application of the theory that deter-
mines its value. (182)
5. Conclusions
In this paper, I have tried to outline and analyze a number of different
conceptualizations of applied linguistics and their relation to linguistic theory. I
have also tried to show that many of the instances of so-called applied lin-
guistics do not really involve applications of notions developed in linguistic
theory. In fact, linguistic theory has not had as much impact in the solution of
practical language problems as it could be expected to. I have tried to analyse
the reasons for this lack of impact. Specifically, the identification of linguistic
Sridhar: What are applied linguistics? 175
theory with formal linguistics, and the exclusion of considerations of function,
performance, and context may be regarded as factors responsible for the alien-
ation of linguistics from real-life language concerns. The result has been the
development of parallel disciplines that perform the applied-linguistics func-
tions, though they are not really applications of linguistic theory at all. I have
tried to show that a number of fields are now included under the applied cate-
gory because of the rather restrictive notion of language adopted in current
theoretical work. These are really alternative paradigms for conducting basic
research on language, and are best regarded extensions of linguistic theory
interpreted in a suitably broad sense.
Finally, I have suggested that, if linguistics is to become anything more
than an esoteric footnote to the real study of language, linguists would do well
to adopt some features of the 'applied* orientation: It offers a more realistic and
insightful approach to the study of language as a communicative device.
Until then, the term 'applied linguistics' will remain a misnomer. After all,
as the redoubtable Dwight Bolinger remarked in his Aspects of Language
(1975:550), in his usual solidly common-sensical way, 'the practical work of
describing languages goes forward — to make dictionaries, assimilate minor-
ities, provide bilingual instruction, train translators — with or without linguistic
theory.'
REFERENCES
Bates, Elizabeth, & Brian MacWhinney. 1982. Functionalist approaches to
grammar. Language acquisition: The state of the art, ed. by E. Wanner &
L. Gleitman. Cambridge: Cambridge University Press.
Bjarkman, Peter C, & Victor Raskin (eds.) 1986. The real-world linguist:
Linguistic applications in the 1980s. Norwood, NJ: Ablex.
Bolinger, Dwight. 1975. Aspects of language. Second edition. New York:
Harcourt, Brace, Jovanovich.
CORDER, S. P. 1975. Applied linguistics and language teaching. Readings for
Applied Linguistics, vol. 1, ed. by J. P. B. Allen & S. P. Corder. London:
Oxford University Press.
Ferguson, Charles A. 1987. Applied Linguistics. Linguistic Society of America
Annual Meeting (MS).
GUMPERZ, John J. 1981. The linguistic bases of communicative competence.
Analyzing discourse: Text and talk, ed. by D. Tannen, 323-334. Washing-
ton, DC: Georgetown University Press.
Halliday, Michael A. K., et al. 1964. The linguistic sciences and language
teaching. London: Longmans.
176 Studies in the Linguistic Sciences 20:2 (Fall 1990)
HOPPER, Paul. 1988. Emergent grammar and the a priori grammar postulate.
Tannen 1988.
Hymes, Dell. 1984. Sociolinguistics: Stability and consolidation. International
Journal of the Sociology of Language 45.
Jakobson, Roman. 1990. Linguistics in relation to other sciences. On lan-
guage, ed. by Linda Waugh & Monvllle-Burston. Cambridge: Harvard
University Press.
Kachru, Braj B. 1981. Socially-realistic linguistics; The Firthian tradition. So-
ciolinguistic theory. (= Special Issue of International Journal of the Soci-
ology of Language.) [An earlier version in Studies in the Linguistic Scien-
ces 10:1.85-111 (1980).]
Kaplan, Robert (ed.) 1980. On the scope of applied linguistics. Rowley, MA:
Newbury House.
Labov, William. 1988. The judicial testing of linguistic theory. Tannen
1988:159-182.
LEECH, Geoffrey N. 1983. Principles of pragmatics. London: Longmans.
Raskin, Victor. 1986. Preface: Linguistic applications in the 1980s. Bjarkman
& Raskin 1986:xiii-xix.
Sacks, Oliver. 1987. The man who mistook his wife for a hat and other clinical
tales. New York: Summit Books.
Saville-Troike, Muriel. 1988. From context to communication: Paths to sec-
ond language acquisition. Tannen 1988:249-268.
SCHEGLOFF, Emanuel A. 1988. Discourse as an interactional achievement.
Tannen 1988:135-158.
Sridhar, S. N. 1980. Contrastive analysis, error analysis, and interlanguage:
Three phases of one goal. Readings on English as a Second Language
for teachers and teacher trainers, ed. by K. Croft, 91-119. Boston: Win-
throp.
Strevens, Peter. 1980. Who are applied linguists and what do they do? A
British point of view. Kaplan 1980:28-36.
Tannen, Deborah (ed.) 1988. Linguistics in context: Connecting observation
and understanding. Norwood, NJ: Ablex.
WiDDOWSON, Henry G. 1980. Applied linguistics: The pursuit of relevance.
Kaplan 1980:74-87.
Studies in the Linguistic Sciences
Volume 20, Number 2, Fall 1990
INDEX TO STUDIES IN THE LINGUISTICS SCIENCES
VOLUMES 1 - 19
A. Author index
Abasheikh, Mohammad Imam. Reflexivization in Chimwi:ni. 6:2.1-22 (Fall
1976)
See also: Kisseberth, Charles W. & Mohammed Imam Abasheikh.
Abbi, Anvita, & Mithilesh Kumar Mishra. Aspectual elements of simultaneity and
interaction in Indian languages: A case for an areal universal. 17:1.1-14
(Spring 1987)
Abd-el-Jawad, Hassan R. — See: Abu-Salim, Issam M., & Hassan R. Abd-el-
Jawad.
Abdul-Karim, Kamal. See: Kenstowicz, Michael, & Kamal Abdul-Karin.
Abu-Salim, Issam M. Epenthesis and geminate consonants in Palestinian
Arabic. 10:2.1-11 (Fall 1980)
Syllable structure in Palestinian Arabic. 12:1.1-28 (Spring 1982)
, & Hassan R. Abd-el-Jawad. Syllable patterns in Levantine Arabic. 18:2.1-
22 (Fall 1988)
Aggarwal, Narindar K. Reference material in Hindi: State of the art. 1 1 :2.209-
219 (Fall 1981)
Agnihotri, Rama Kant. Crisis of identity: Sikhs in England. Review by Jean
Aitchison. 19:1.169-171 (Spring 1989)
Ahn, Sang-Cheol. On the nature of /7 in Korean. 16:2.1-13 (Fall 1986)
Aitchison, Jean. Review of Agnihotri, Crisis of identity: Sil<hs in England.
19:1.169-171 (Spring 1987)
Akatsuka, Noriko. Emotive verbs in English and Japanese. 2:1.1-15 (Spring
1972)
Ali, Mohammed. Trends in Oromo lexicon and lexicography. 19:2.155-168
(Fall 1989)
Archangeli, Diana. CV-skeleton or X-skeleton: The Turkish evidence. 15:1.1-
10 (Spnng 1985)
Arora, Harbir, & K. V. Subbarao. Convergence and syntactic reanalysis: The
case of so in Dakkhini. 19:1.1-18 (Spring 1989)
Awoyale, Yiwola. Yoruba gerundive structures and the notion of 'Target
Structures'. 4:1 ,1-31, (Spnng 1974)
Bader, Yousef. Vowel sandhi and syllable structure in Kabyle Berber. 13:1.1-
17 (Spnng 1983)
178 Studies in the Linguistic Sciences 20:2 (Fall 1990)
Barnitz, John G. Bloom-p-field, Chom-p-sky, and phonetic epen-t-thesis. 4:2.1-
13 (Fall 1974)
Baumgardner, Robert — See: Chamberlain, Dick, & Robert Baunngardner
(eds.).
Becker, Lee A. On the representation of contour tones in Generative
Phonology. 7:1.8-20 (Spring 1977)
, & David P. B. Massamba. CiRuri tonology (A preliminary view). 10:1.1-13
(Spring 1980)
, & Farid Mohamed Onn. The rise and fall of a transderivational constraint:
The case of Malay. 7:2.106-114 (Fall 1977)
Bennett, J. Fraser. Consonant merger in Navajo: An underspecified analysis.
17:2.1-19 (Fall 1987)
Bentur, Esther. Orthography and the formulation of phonological rules. 8:1.1-
25 (Spring 1978)
Bereiter, Marilyn. A study of duration in speech production. 4:2.14-24 (Fall
1974)
Bethin, Christine Yurkiw. The development of the Slavic mid vowels in newly
checked syllables in the Northwestern Ukrainian dialects. 5:2.1-11 (Fall
1975)
Bhat, D. N. S. Physical identification in Kannada. 11 :2. 1-8 (Fall 1981)
Bhatia, Tej K. A computational investigation on the perception of aspirated con-
sonants in Hindi. 3:1.63-80 (Sphng 1973)
The evolution of tones in Punjabi. 5:2.12-24 (Fall 1975)
A history of the Hindi grammatical tradition. Review by Rajeshwari Pan-
dharipande. 19:1.173-179 (Spring 1989)
On the scope of negation in Hindi. 3:2.1-27 (Fall 1973)
A study of aspirated consonants as spoken and recognized by Hindi
speakers. 4:2.25-39 (Fall 1974)
Transplanted South Asian languages: An overview. 11:2.129-139 (Fall
1981)
The treatment of transitivity in the Hindi grammatical tradition. 1 1 :2.195-
208 (Fall 1981)
— . Trinidad Hindi: Three generations of a transplanted variety. 11:2.135-150
(Fall 1981)
Vahation in Hindi: Problems and prospects. 13:2.1-19 (Fall 1983)
— See also: Kachru, Yamuna, & Tej K. Bhatia; Subbarao, K. V., & Tej K.
Bhatia..
Bhatt, Rakesh Mohan. Good mixes and odd mixes: Implications for the
bilingual's grammar. (Squib.) 19:1.165-168 (Spring 1989)
Biloa, Edmond. Tuki gaps: Null resumptive pronouns or vanables? 19:2.43-54
(Fall 1989)
Blomeyer, Charlotte, & Tamara Valentine. 'We makin' some cookies': A child
language case study of the effects of situational variation on pragmatic
function and syntax. 13:1.19-42 (Spring 1983)
Index to Volumes 1-19 — Author index 179
Bokamba, Eyamba G. African linguistic research and publications from the
University of Illinois at Urbana-Channpaign, 1970-1979. (Appendix.)
9:2.201-206 (Fall 1979)
Authenticity and the choice of a national language: The case of Zaire.
6:2.23-64 (Fall 1976)
(ed.). The contribution of African linguistics to linguistic theory:
Proceedings of the 20th Annual Conference on African Linguistics (Vol. 1).
19:2 (Fall 1989)
French colonial language policy in Africa and its legacies (Part I) 14:2.1-
35 (Fall 1984)
Inversions as grammatical relation changing rules in Bantu languages.
9:2.1-24 (Fall 1979)
. Language and national development m sub-Saharan Africa. (A progress
report). 11:1.1-25 (Spring 1981)
(ed.). Language in African culture and society. 14:2 (Fall 1984)
On the syntax and semantics of Wh-questions in Kikongo and Kiswahili.
6:2.65-88 (Fall 1976)
, & Nkonko M. Kamwangamalu. The significance of code-mixing to linguistic
theory: Evidence from Bantu languages. 17:2.21-43 (Fall 1987)
, & Charles W. Kisseberth (eds). Papers on African linguistics. 6:2 (Fall
1976)
Botne, Robert. Reconstruction of a grammaticalized auxiliary in Bantu.
19:2.169-186 (Fall 1989)
Britti, Anthony. A history of right dislocation in certain Levantine Arabic dialects.
10:2.121-139 (Fall 1980)
Quantifier Repositioning. 9:1.1-16 (Spring 1979)
Broselow, Ellen. Syllable structure in two Arabic dialects. 10:2.13-24 (Fall
1980)
Bundrick, Camille. An inference-based account of restrictive relative which and
that. 19:1.19-31 (Spring 1989)
A lexical phonology approach to Hindi schwa deletion. 17:1.15-23
(Spring 1987)
Burt, Susan Meredith. Another look at nara conditionals. 9:2.25-38 (Fall 1979)
Remarks on German nominalization. 9:1.17-30 (Spring 1979)
Busnel, Rene-Guy, & Andre Classe. Whistled languages. Review by Chin-W.
Kim. 7:2.196-199 (Fall 1977)
Cardona, George. Relations between causatives and passives in Indo-lranian.
8:2.1-42 (Fall 1978)
Carreira, Maria. The representation of diphthongs in Spanish. 18:1.1-24
(Spnng 1988)
Cassimjee, Farida An autosegmental analysis of Venda nominal tonology.
13:1.43-72 (Spnng 1983)
,& Charles W Kisseberth. Downstep in Venda. 14:1.1-29 (Spnng 1984)
, & Shingazidja nominal accent. 19:1.33-61 (Spring 1989)
Castillo, Rafael — See: Green, Georgia M., & Rafael Castillo.
1 80 Studies in the Linguistic Sciences 20:2 (Fall 1990)
Cerron-Palomino, Rodolfo. Morphologically conditioned changes in Wanka-
Quechua. 4:2.40-75 (Fall 1974)
Cervin, Richard. On the notion of 'second position' in Greek. 18:2.23-39 (Fall
1988)
Chamberlain, Dick, & Robert Baumgardner (eds.). ESP in the classroom:
Practice and evaluation. Review by Numa Markee. 1 9:1 . 1 81 -1 85 (Spring
1989)
Chang, Suk-Jin. Tag questions in Korean: Form and function. 16:2.15-26 (Fall
1986)
Cheng, Chin-Chuan. The Esperanto of El Popola Cinio. 12:1.49-62 (Spring
1982)
A quantification of Chinese dialect affinity. 12:1 .29-47 (Spring 1982)
Teaching Chinese numeration on computer. 7:2.165-177 (Fall 1977)
Tonal correlations in Chinese dialects: A quantitative study. 7:2.1 15-128
(Fall 1977)
& Chin-W. Kim (eds.). Studies in East Asian linguistics. 7:2 (Fall 1977)
& Charles W. Kisseberth. Ikorovere Makua Tonology (Part I). 9:1.31-64
(Spring 1979)
& Ikoravere Makua Tonology (Part 2). 10:1.15-44 (Spring 1980)
& Ikorovere Makua Tonology (Part 3). 11:1.181-202 (Spring 1981)
& . Tone-bearing Nasals in Makua. 12:1.123-139 (Spring 1982)
See also: Lu, Zhiji, & Chin-Chuan Cheng; Sherwood, Bruce Arne, & Chin-
Chuan Cheng.
Childs, G. Tucker. Where do ideophones come from? 19:2.55-76 (Fall 1989)
Cho, Euiyon. On the morphology of morphological causative verbs in Korean:
An argument against Lieber's morpheme-based lexicon. 16:2.27-43 (Fall
1986)
Cho, Jae Ohk, & Jerry Morgan. Some problems on NP coordination in Korean.
16:2.45-65 (Fall 1986)
Choi, Yeon Hee. A study of coherence in Korean speakers' argumentative
writing in English. 16:2.67-94 (Fall 1986)
Chung, Raung-fu. On the representation of Kejia diphthongs. 19:1.63-80
(Spring 1989)
Chvany, Catherine V. Ergative and argative (nee ergative too). 15:2.1-2 (Fall
1985)
Clamons, Cynthia Robb. Modification of the gender system in the Wollegan
dialect of Oromo. 19:2.187-195 (Fall 1989)
Classe, Andre — See: Busnel, Rene-Guy, & Andre Classe.
Clements, G(eorge) N. African linguistics and its contributions to linguistic
theory. 19:2.3-39 (Fall 1989)
Binding domains in Kikuyu. 14:2.37-56 (Fall 1984)
Coats, Herbert S. Palatalization in Russian. 15:2.3-8 (Fall 1985)
Cole, Peter. An apparent asymmetry in the formation of relative clauses in
Modern Hebrew. 5:1.1-35 (Spring 1975)
Global grammar versus index grammar: A question of power. 3:1.45-53
(Spring 1973)
Index to Volumes 1-19 — Author index 181
A matter of scope: McCawley versus Postal on the origin of noun phrases.
7:1.117-146 (Spring 1977)
— . Noun phrases as quantifiers. 2:1.16-41 (Spring 1972)
Syntactic analogy and backward pronominalization. 3:1.33-44 (Spring
1973)
, Wayne Harbert, & Gabriella Hermon. Headless relative clauses in
Quechua. 8:1.26-41 (Spring 1978)
, , , & Shikaripur Sndhar. On the acquisition of subjecthood. 8:1.42-
71 (Spring 1978)
, Wayne Harbert, Shikaripur Sridhar, Sachiko Hashimoto, Cecil Nelson, &
Diane Smietana. Noun phrase accessibility and island constraints.
6:1.170-185 (Spring 1976)
, & Gabriella Hermon. Subject to Object Raising in an EST framework:
Evidence from Quechua. 9:1.65-90 (Spring 1979)
, & Janice Jake. Accusative subjects in Imbabura Quechua. 8:1.72-96
(Spring 1978)
, & S. N. Sridhar. Clause union and Relational Grammar: Evidence from
Hebrew and Kannada. 6:1.216-227 (Spring 1976)
See also: Morgan, Jerry, Georgia Green, & Peter Cole (eds.).
Comrie, Bernard. Linguistics is about languages. 8:2.221-236 (Fall 1978)
The sun letters in Maltese: Between morpho-phonemics and phonetics.
10:2.25-37 (Fall 1980)
Creider, Chet A. Language differences in strategies for the interactional
management of conversation. 14:2.57-65 (Fall 1984)
Cureton, Richard D. The exceptions to passive in English: A pragmatic
hypothesis. 9:2.39-54 (Fall 1979)
The inclusion constraint: Description and explanation. 9:1.91-104 (Spring
1979)
Dabair-Moghaddam, Mohammad. Passive in Persian. 12:1.63-90 (Spring
1982)
Dalgish, Gerry (Gerard M.). Arguments for a unified treatment of y-initial and
vowel-initial roots in Olutsootso. 4:2.76-90 (Fall 1974)
The diachronic development of nasal deletion in Olutsootso. 5:. 2 25-40
(Fall 1975)
Passivizing locatives in Olutsootso. 6:1.57-68 (Spring 1976)
, & Gloria Shemtuch. On the justification for language-specific sub-
grammatical relations. 6:2.89-107 (Fall 1976)
Davison, Alice. Case and control in Hindi-Urdu. 15:2.9-23 (FalM 985)
Contextual effects on 'genenc' indefinites: Cross-linguistic arguments for
pragmatic factors. 9:2.55-66 (Fall 1979)
Some mysteries of subordination. 9:1.105-128 (Spring 1979)
lA/'/i-movement in Hindi-Urdu relative clauses. 17:1.25-33 (Spring 1987)
Dickerson, Wayne B. Variable rules in the language community: A study of lax
[u] in English. 5:2.41-68 (Fall 1975)
Dil, Afia — See: Ferguson, Charles, & Afia Dii.
1 82 Studies in the Linguistic Sciences 20:2 (FaU 1990)
Domingue, Nicole. Internal change in a transplanted language. 11:2.151-159
(Fall 1981)
Donaldson, Susan Kay. Faulty referents and their relationship to tense.
3:1.155-171 (Spring 1973)
Movement in restrictive relative clauses in Hindi. 1:2.1-74 (Fall 1971)
On the (possibly) presuppositional nature of when-clauses in Hindi.
3:2.28-42 (Fall 1973)
Downing, Laura J. Tone in Jita questions. 19:2.91-1 13 (Fall 1989)
— . Tonology of noun-modifier phrases in Jita. 18:1.25-60 (Spring 1988)
Drame, Mallafe. Aspects of Mandingo complementation. 9:2.67-90 (Fall 1979)
D'souza, Jean. Codification of non-native English: Is it necessary/possible?
16:1.1-11 (Spring 1986)
Schwa syncope and vowel nasalization in Hindi-Urdu: A non-linear
approach. 15:1.11-30 (Spring 1985)
Du, Tsai-Chwun. A computer tool in the study of Taiwanese tones. 17:2.45-62
(Fall 1987)
Dudas, Karen. The accentuation of Lithuanian derived nominals. 2:2.108-136
(Fall 1972)
A case of functional phonological opacity: Javanese elative formations.
4:2.91-111 (Fall 1974)
, & Margie O'Bryan. Lithuanian verbal accentuation. 2:2.86-107 (Fall 1972)
Dunn, Andrea S. Swahili policy implementation in Tanzania: The role of the
National Swahili Council (BAKITA). 15:1.31-47 (Spring 1985)
Dunn, Christian— See: Timmons, Claude, & Christian Dunn.
Ervin-Tripp, Susan. Whatever happened to communicative competence?
8:2.237-258 (Fall 1978)
Faraclas, Nicholas. Rivers Pidgin English: Tone, stress, or pitch-accent lan-
guage? 14:2.67-76 (Fall 1984)
Farina, Donna M. The morphological rule of learned backing and lexical
phonology. 14:1.31-56 (Spring 1984)
Multiword lexical units in French. 17:2.63-76 (Fall 1987)
Ferguson, Charles A. Multilingualism as object of linguistic description. 8:2.97-
106 (Fall 1978)
, & Afia Dil. The sociolinguistic variable (s) in Bengali: A sound change in
progress? 9:1.129-138 (Spring 1979)
Foster, F. Joseph. Phmitiveness, naturalness, and cultural fit. 15:2.25-37 (Fall
1985)
Fox, Samuel E. Problems of the dual in Soqotri. 5:2.69-75 (Fall 1975)
Garapati, U. Rao. The development of personal pronouns in modern Gondi.
17:1.35-50 (Spnng 1987)
Garber, Anne. Word order change and the Senufo languages. 10:1.45-57
(Spring 1980)
Gerdemann, Dale. Restriction as a means of optimizing unification parsing.
19:1.81-92 (Spring 1989)
, & Erhard W. Hinrichs. UNICORN: A unification parser for attribute-value
grammars. 18:2.41-86 (Fall 1988)
Index to Volumes 1-19 — Author index 183
Gladney, Frank Y. On glides following vocalic verbs in Russian. 15:2.39-59
(Fall 1985)
Green, Georgia M. Competence for implicit text analysis: Literary style
discrimination in five-year-olds. 11:1.39-56 (Spring 1981)
. The derivation of a relative infinitive construction. 3:1.1-32 (Spring 1973)
Do inversions in English change grammatical relations? 7:1.157-181
(Spring 1977)
Governed-rule change and Universal Grammar. 6:1.152-169 (Spring
1976)
Notes on clefts and pseudo-clefts and other related matters. 1:1.1-7
(Spring 1971)
(ed.). Papers on syntax and semantics. 2:1 (Spring 1972)
Pragmatics and syntactic description. 11:1 .27-37 (Spring 1 981 )
. Some remarks on why there is implicature. 17:2.77-92 (Fall 1987)
, & Rafael Castillo. A selected bibliography of semantics-based generative
grammar. 2:1.123-140 (Spring 1972)
, & J.L. Morgan. Notes toward an understanding of rule government.
6:1.228-248 (Spring 1976)
See also: Morgan, Jerry, Georgia Green, & Peter Cole (eds.).
Habick, Timothy. Suppletive verb phrase deletion. 3:1.172-182 (Spring 1973)
Haddad, Ghassan F. Epenthesis and sonohty in Lebanese Arabic. 14:1.57-88
(Spring 1984)
, & Michael Kenstowicz. A note on the parallels between the definite article
and the relative clause marker in Arabic. 10:2.141-147 (Fall 1980)
Halle, Morris. Formal vs. functional considerations in phonology. 8:2.123-134
(Fall 1978)
Remarks on the scientific revolution in linguistics 1926-1929. 15:2.61-77
(Fall 1985)
Haller, John A. Enhanced miniature artificial languages. 13:1.73-81 (Spring
1983)
Halpern, Richard Neil. The bivalence of NEG raising predicates. 6:1.69-81
(Spring 1976)
A note on seem. 7:1.79-87 (Spnng 1977)
— . Notes on the origin of quantifier floating. 7:1.41-45 (Spnng 1977)
. Time travel or the futuristic use of to gro. 5:1.36-41 (Spring 1975)
Harbert, Wayne — See: Cole, Peter, Wayne Harbert, & Gabriella Hermon;
Cole, Peter, Wayne Harbert, Gabriella Hermon, & Shikaripur Sridhar;
Cole, Peter, Wayne Harbert, Shikaripur Shdhar, Sachiko Hashimoto, Cecil
Nelson, & Diane Smietana.
Hashimoto, Sachiko — See: Cole, Peter, Wayne Harbert, Shikahpur Sridhar,
Sachiko Hashimoto, Cecil Nelson, & Diane Smietana.
Helmreich, Stephen C. DevanagarF word-processing on the IBM-PC. 17:1.51-
61 (Spnng 1987)
Hermon, Gabriella. Rule ordehng versus globality: Evidence from the inversion
construction. 9:1.139-146 (Spring 1979)
1 84 Studies in the Linguistic Sciences 20:2 (FaU 1990)
See also: Cole, Peter, & Gabriella Hermon; Cole, Peter Wayne Harbert, &
Gabrlella Hermon; Cole, Peter, Wayne Harbert, Gabriella Hernnon, &
Shikaripur Sridar.
Hinrichs, Erhard W. — See; Gerdemann, Dale, & Erhard W. Hinrichs.
Hock, Hans Henrlch. Archaisms, morphophonemic metrics, or variable rules in
the Rig-Veda? 10:1.59-69 (Spring 1980)
Aux-cliticization as a motivation for word order change. 12:1.91-101
(Spring 1982)
The Baltic e-preterit: An older a-preterit? 2:2.137-164 (Fall 1972)
Clitic verbs in PIE or discourse-based verb fronting? Sanskrit sa hovaca
gargyahand congeners in Avestan and Homeric. 12:2.1-38 (Fall 1982)
Conjoined we stand: Theoretical implications of Sanskrit relative
structures. 19:1.93-126 (Spring 1989)
Exceptions and synchronic analogy in Sanskrit. 3:1.81-101 (Spring 1973)
. Finiteness in Dravidian (Review article): Sanford B. Steever (1988): The
serial verb formation in the Dravidian languages. 18:2.21 1-233 (Fall 1988)
— -. Is there an a-epenthesis in Sanskrit? 3:2.43-58 (Fall 1973)
Language-death phenomena in Sanskrit: Grammatical evidence for
attrition in contemporary spoken Sanskrit. 13:2.21-35 (Fall 1983)
— (ed.). Papers from the 1986 South Asian Languages Analysis Roundtable.
17:1 (Spring 1987)
(ed.). Papers on diachronic syntax: Six case studies. 1 2:2 (Fall 1 982)
(Pre-)Rig-Vedic convergence of Indo-Aryan with Dravidian? Another look
at the evidence. 14:1 .89-1 08 (Spring 1 984)
Problems in the synchronic derivation of the Lithuanian e-formations.
2:2.165-203 (Fall 1972)
Sanskrit causative syntax: A diachronic study. 1 1 :2. 9-33 (Fall 1981)
The Sanskrit quotative: A histoncal and comparative study. 12:2.39-85
(Fall 1982)
Substratum influence on (Rig-Vedic) Sanskrit? 5:2.76-125 (Fall 1975)
Yes, Virginia, syntactic reconstruction is possible. 15:1.49-67 (Spring
1985)
, & Michael J. Kenstowicz (eds.). Studies in Baltic linguistics. 2:2 (Fall
1972)
See also: Zgusta, Ladislav, & Hans H[enrich] Hock (eds.).
Hodges, Kathryn Speed. Object relations in Kimeru causatives. 6:2.108-141
(Fall 1976)
, & Susan U. Stucky. On the inadequacy of a grammatical relation referring
rule in Bantu. 9:2.91-100 (Fall 1979)
Holisky, Dee Ann, & Nancy Yanofsky. On the pragmatic motivation for perhaps.
9:2.101-108 (Fall 1979)
Hook, Peter Edwin. Poguli syntax in the light of Kashmiri: A preliminary report.
17:1.63-71 (Spring 1987)
Houston, John R., Jr. Dari relative clauses. 4:1.32-58 (Spring 1974)
Hyman, Larry M. Accent in Bantu: An appraisal. 19:2.115-134 (Fall 1989)
Index to Volumes 1-19 — Author index 185
Irshied, Omar, & Michael Kenstowicz. Some phonological rules of Bani-Hassan
Arabic: A Bedouin dialect. 14:1.109-147 (Spring 1984)
, & Peter Whelan. Exploring the dictionary: On teaching foreign learners of
Arabic to use the Arabic-English dictionary. 18:1.61-75 (Spring 1988)
Jake, Janice L. Object verb agreement in Tigre. 10:1 .71 -84 (Spring 1 980)
Some remarks on relativization in Imbabura Quechua. 9:2.109-130 (Fall
1979)
Why Dyirbal isn't ergative at all. 8:1.97-110 (Spring 1978)
,& David Odden. Raising in Kipsigis. 9:2.131-156 (FalM 979)
See also: Cole, Peter, & Janice Jake.
Jenkins, Fred M. Cinema-verite, golf-bijou, and sandwich beurre in
contemporary French. 6:1.1-21 (Spring 1976)
Review of Martinet, Studies in functional syntax/Etudes de syntaxe fonc-
tionnelle. 7:2.193-195 (Fall 1977)
Johnson, David E. Adjective flipping and the notion of target structure. 4:1 .59-
79 (Sphng 1974)
Why delete tense? 3:1.54-62 (Spring 1973)
Ka, Omar. Syllable structure and suffixation in Wolof. 15:1.61-90 (Spring 1985)
Kachru, Braj B. The bilingual's creativity: Discoursal and stylistic strategies in
contact literatures in English. 13:2.37-55 (Fall 1983)
ESP and non-native varieties of English; Toward a shift in paradigm.
16:1.1 3-34 (Spring 1986)
General linguistic studies in Hindi: A review of resources. 3:2.59-86 (Fail
1973)
(ed.). Linquistics in the seventies: Directions and prospects. 8:2 (Fall
1978)
(ed.). Papers on South Asian linguistics. 3:2 (Fall 1973)
Socially realistic linguistics: The Firthian tradition. 10:1.85-111 (Spring
1980)
(ed.). Studies in language variation: Nonwestern case studies. 13:2 (Fall
1983)
Toward structuring the form and function of code-mixing: An Indian per-
spective. 5:1.74-92 (Spring 1975)
World Englishes and applied linguistics. 19:1.127-152 (Spring 1989)
Kachru, Yamuna. Applied linguistics and foreign language teaching: A non-
Western perspective. 15:1.91-107 (Spring 1985)
Applied linguistics and foreign language teaching: a non-Western
perspective. 16:1.35-51 (Spring 1986)
Causative sentences in Hindi revisited. 1:2.75-103 (Fall 1971)
Corpus planning for modernization: Sanskntization and Englishization of
Hindi. 19:1.153-164 (Spnng 1989)
Cross-cultural texts and interpretation. 13:2.57-72 (Fall 1983)
— (ed.). Dimensions of South Asian linguistics. 1 1 ;2 (Fail 1 981 )
impact of expanding domains of use on a standard language:
Contemporary Hindi in India. 17:1.73-90 (Spring 1987)
1 86 Studies in the Linguistic Sciences 20:2 (Fall 1990)
On the syntax, semantics, and pragmatics of the conjunctive participle in
Hindi-Urdu. 11:2.35-49 (Fall 1981)
— (ed.). Papers on Hindi syntax. 1 :2 (Fall 1 971 )
Pragmatics and verb serialization in Hindi-Urdu. 9:2.157-170 (Fall 1979)
Some aspects of pronominalization and relative clause construction in
Hindi-Urdu. 3:2.87-103 (Fall 1973)
Syntactic variation and language change: Eastern and Western Hindi.
12:2.87-96 (Fall 1982)
Transitivity and volitionality in Hindi-Urdu. 1 1 :2. 181 -193 (Fall 1981)
& Tej K. Bhatia. Evidence for global constraints: The case of reflexivization
in Hindi-Urdu. 5:1.42-73 (Spring 1975)
& Rajeshwari Pandharipande. On ergativity in selected South Asian
languages. 8:1.111-126 (Spring 1978)
& Toward a typology of compound verbs in South Asian languages.
10:1.113-124 (Spring 1980)
— See also: Pandharipande, Rajeshwari, & Yamuna Kachru.
Kamwangamalu, Nkonko Mudipanu. 'C-command' and the phonology-syntax
interface in Ciluba. 18:2.87-109 (Fall 1988)
. Passivization in Bantu languages: Implications for relational grammar.
15:1. 109-133 (Spring 1985)
See also: Bokamba, Eyamba G., & Nkonko M. Kamwangamalu.
Katre, Sumitra M. Astadhyayfof Panini. Review by Ladislav Zgusta. 19:1.187-
193 (Spring 1989)
Kaye, Jonathan. Functionalism and functional explanations in phonology.
8:2.135-148 (Fall 1978)
Keenan, Edward. On surface form and logical form. 8:2.163-204 (Fall 1978)
Kendall, Sue Ann, & James Hye-Suk Yoon. Sentence particles as evidence for
morphosyntactic interaction with pragmatics. 16:1.53-77 (Spring 1986)
Kenstowicz, Michael (J.). Gemination and spirantization in Tigrinya. 12:1.103-
122 (Spring 1982)
Inflectional accent of the Serbo-Croatian noun. 4:1.80-106 (Sphng 1974)
(ed.). Linguistic studies in memory of Theodore M. Lightner. 15:2 (Fall
1985)
Lithuanian phonology. 2:2.1-85 (Fall 1972)
Notes on Cairene Arabic syncope. 10:2.39-53 (Fall 1980)
— . The phonology and syntax of wh-expressions in Tangale. 15:2.79-91 (Fall
1985)
The phonology of Chukchee consonants. 16:1.79-96 (Spring 1986)
Some rules of Koryak phonology. 6:1.22-37 (Spring 1976)
— - (ed.). Studies in Arabic linguistics. 10:2 (Fall 1980)
, & Kamal Abdul-Karim. Cyclic stress in Levantine Arabic. 10:2.55-76 (Fall
1980)
, & Charles W. Kisseberth. Unmarked bleeding orders. 1:1.8-28 (Spring
1971)
, Emmanuel Nikiema, & Meterwa Ourso. Tonal polarity in two Gur
languages. 18:1.77-103 (Spnng 1988)
Index to Volumes ! - 19 — Author index 187
, & Wafaa Wahba. Clitics and the double object construction in Cairene
Arabic. 10:2.149-163 (Fall 1980)
See also: Haddad, Ghassan, & Michael Kenstowicz; Hock, Hans Henrich,
& Michael J. Kenstowicz (eds.); Irshied, Omar, & Michael Kenstowicz.
Khan, Baber S. A. The ergative case in Hindi-Urdu. 17:1.91-101 (Spring 1987)
Killean, Carolyn G. Demonstrative variation in Oral Media Arabic in Egypt.
10:2.165-178 (Fall 1980)
Kim, Chin-W. Epenthesis and elision in metrical phonology. 11:1.57-71
(Spring 1981)
(ed.). Illinois studies in Korean linguistics. 16:2 (Spring 1986)
. Neutralization in Korean revisited. 9:1.147-156 (Spring 1979)
— -. A note on tonal conjunction in Efik. 4:2.112-122 (Fall 1974)
— -. Phonology on the C-string7 15:2.93-100 (Fall 1985)
— -. Review of Busnel & Classe, Whistled languages. 7:2.196-199 (Fall 1977)
Review of Singh, Distinctive features: Theory and validation. 7:2.200-204
(Fall 1977)
Vowel length in Korean. 7:2.184-190 (Fall 1977)
, & Han Sohn. A phonetic model for reading: Evidence from Korean.
16:2.95-105 (Fall 1986)
See also: Cheng, Chin-chuan, & Chin-W. Kim (eds.); Kisseberth, Charles
W., & Chin-W. Kim (eds.); Sohn, Han, & C-W. Kim.
Kiparsky, Paul. Analogical change as a problem for linguistic theory. 8:2.72-96
(Fall 1978)
Kisseberth, Charles W. Displaced tones in Digo (Part 2). 11:1.73-120 (Spring
1981)
, & Mohammad Imam Abasheikh. A case of systematic avoidance of homo-
nyms. 4:1.107-124 (Spring 1974)
& Chimwi:ni prefix morphophonemics. 6:2.142-173 (Fall 1976)
& The 'object' relationship in Chi-Mwi:ni, a Bantu language. 6:1.100-
129 (Spring 1976)
& On the interaction of phonology and morphology: A Chi-Mwi:ni
example. 4:2.139-147 (Fall 1974)
& The perfect stem in Chi-Mwi:ni. 4:2.123-148 (Fall 1974)
& Chin-W. Kim (eds.). Papers on phonetics and phonology. 4:2 (Fall
1974)
, & David Odden. Aspects of tone assignment in Kimatuumbi. 10:1.125-
140 (Spring 1980)
& Winifred J. Wood. Displaced tones in Digo (Part I). 10:1.141-177
(Spring 1980)
See also: Bokamba, Eyamba G., & Charles W. Kisseberth (eds.);
Cassimjee, Fanda, & Charles W. Kisseberth; Cheng, Chin-Chuan, &
Charles W. Kisseberth; Kenstowicz, Michael, & Charles W. Kisseberth.
Kleiman, Angela B. Some aspects of the causative construction in Hindi.
1:2.104-135 (Fall 1971)
Kong, Yong-ll. The Korean writing system: A linguistic examination. 16:2.107-
119 (Fall 1986)
188 Studies in the Linguistic Sciences 20:2 (Fall 1990)
Krause, Scott R. — See: Silver, Pamela S., & Scott R. Krause.
Kumar, Suresh. Discourse structure in a Hindi short story. 11:2.51-66 (Fall
1981)
Lara, Luis Fernando. El concepto de norma en linguistica. Review by Ladislav
Zgusta. 7:2.191-192 (Fall 1977)
Lederman, Shiomo. Problems in a prosodic analysis of Hebrew morphology.
12:1.141-163 (Spring 1982)
— . Relativization and pronoun deletion in Hebrew. 13:1.83-88 (Spring 1983)
Lee, Cher-leng. Review of Hakuta, Mirror of language: The debate on
bilingualism. 1 7:2. 1 37-1 42 (Fall 1 987)
Lee, Sang Oak. Conspiracy in Korean phonology revisited: As applied to
historical data. 7:2.1-23 (Fall 1977)
An explanation of syllable structure change in Korean: With special
reference to Vennemann's preference laws. 16:2.121-133 (Fall 1986)
Lee, Yen Ling. Correlation among attitudinal factors, speed, and tone sandhi in
Chinese. 7:2.129-141 (Fall 1977)
A study on code-switching in Taiwan. 11:1.121-136 (Spring 1981)
Lehman, Frederic K. A brief note on the reconstruction of MA? in Tibeto-
Burman. 7:2.24-38 (Fall 1977)
. Prefixing, voicing, and syllable reduction in Burmese: Juncture and
syllable structure. 3:2.104-120 (Fall 1973)
, with Namtip Pingkarawat. Missing nominals, non-specificity, and related
matters, with especial reference to Thai and Burmese. 15:2.101-121 (Fall
1985)
Lehmann, Winfred P. The persistence of pattern in language. 15:2.123-126
(Fall 1985)
Leskosky, Richard J. Further comments on instrumentals. 2:1.66-83 (Spring
1972)
— -. Intensive reflectives. 2:1.42-65 (Spring 1972)
— . Not your usual use of you. 4:1.125-131 (Spring 1974)
Levine, James. The relative pronoun and the long form adjective in Russian.
8:1. 127-136 (Spring 1978)
Livnat, Michal Alien. The indicator particle baa in Somali. 13:1.89-132 (Spring
1983)
Lowenberg, Peter H. Lexical modernization in Bahasa Indonesia: Functional
allocation and variation in borrowing. 13:2.73-86 (Fall 1983)
Lu, Zhiji. Shanghai tones: A nonlinear analysis. 17:2.93-1 13 (Fall 1987)
Tonal changes: Interplay between tone and tone sandhi: A case study of
the Shanghai dialect. 16:1.97-111 (Sphng 1986)
, & Chin-Chuan Cheng. Chinese dialect affinity based on syllable initials.
15:2.127-148 (Fall 1985)
Lunt, Horace G. On the progressive palatalization of early Slavic: Synchrony
versus history. 15:2.149-169 (Fall 1985)
Lutz, Richard. Hindi verbs of judging: An application of Fillmore's system of
semantic description. 10:1.179-188 (Spring 1980)
Index to Volumes 1-19 — Author index 189
Magura, Benjamin J. Language variation and language standardization: The
case of Shona in Zimbabwe. 13:2.87-98 (Fall 1983)
Makino, Seiichi. A note on the intransitive nature of the Japanese raising verb
omo'u and its implications. 7:2.39-48 (Fall 1977)
Markee, Numa. Review of Chamberlain & Baumgardner (eds.), ESP in the
classroom: Practice and evaluation. 1 9 : 1 . 1 8 1 - 1 85 (Spri ng 1 989)
Martinet, Andre. Studies in functional syntax/Etudes de syntaxe fonctionelle.
Review by Frederic M. Jenkins. 7:2.193-195 (Fall 1977)
Massamba, David P. B. — See: Becker, Lee A., & David P. B. Massamba.
McCarthy, John J. A note on the accentuation of Damascene Arabic. 10:2.77-
98 (Fall 1980)
McCawley, James. Language universals in linguistic argumentation. 8:2.205-
220 (Fall 1978)
McClanahan, Virginia K. Negation m Korean and pragmatic ambiguity.
16:2.135-145 (Fall 1986)
Menn, Lise. Assertions not made by the main clause of a sentence. 4:1.132-
143 (Spring 1974)
Micklesen, Lew R., Sally R. Pitluck, & Edward J. Vajda. Derived imperfectives in
Slavic: A study in denvational morphology. 15:2.171-186 (Fall 1985)
Mishra, Mithilesh Kumar — See: Abbi, Anvita, & Mithilesh Kumar Mishra.
Mohanan, K. P. Pronouns in Malayalam. 1 1 :2. 67-75 (Fall 1 981 )
Morgan, Jerry (L.). Introduction. 6:1.47-48 (Spnng 1976)
. Prelimmaries to the reconstruction of verbal endings in Yuk. 7:1.1-7
(Spnng 1977)
(ed.). Relational Grammar and semantics. 9:2 (Fall 1979)
Some observations on discourse and sentence grammar. 11:1.137-144
(Spring 1981)
Some problems of verb agreement. 2:1.84-96 (Spring 1972)
, Georgia Green, & Peter Cole (eds). Topics in Relational Grammar.
Special section of 6:1 (Spring 1976)
See also: Cho, Jae Ohk, & Jerry Morgan: Green, G(eorgia) M., & J. L.
Morgan.
Moshi, Lioba. A functional typology of ni in Kivunjo (Chaga). 18:1.105-134
(Spnng 1988)
Nakazawa, Tsuneko. How do tense and aspect Interact In determination of verb
forms? Verb past forms and non-past forms in Japanese 'when'-clauses.
15:1.135-146 (Spnng 1985)
, & Laura Neher. Rule expansion on the fly: A GPSG parser for Japanese/
English using a bit vector representation of features and rule schemas.
17:2.115-124 (Fall 1987)
Neher, Laura — See: Nakazawa, Tsuneko, & Laura Neher.
Nelson, Cecil — See: Cole, Peter, Wayne Harbert, Shikanpur Sridhar, Sachiko
Hashimoto, Cecil Nelson, & Diane Smietana.
Nihalani, Paroo. In defence of implosives. 16:1.1 13-122 (Spring 1986)
Nikiema, Emmanuel — See: Kenstowicz, Michael, Emmanuel Nikiema, &
Meterwa Ourso.
190 Studies in the Linguistic Sciences 20:2 (Fail 1990)
Obeidat, Hussein Ali. Relative clauses in Standard Arabic revisited. 14:2.77-96
(Fall 1984)
O'Bryan, Margie. Opacity and rule loss. 4:2.148-160 (Fall 1974)
Opacity and the loss of a morphological process. 3:1.102-120 (Spring
1973)
Restructuring in the verbal system of Pali. 3:2.121-133 (Fall 1973)
The role of analogy in non-derived formations in Zulu. 4:1.144-178
(Spring 1974)
Some problems with /-insertion in Pali. 1:1.29-51 (Spring 1971)
See also: Dudas, Karen, & Margie O'Bryan.
Odden, David. Aspects of Iraqi Arabic verbal phonology. 8:1.137-152 (Spring
1978)
. Evidence for the Elsewhere Condition in Shona. 11:1.145-162 (Spring
1981)
Principles of stress assignment: A crosslinguistic view. 9:1.157-176
(Spring 1979)
See also: Jake, Janice, & David Odden; Kisseberth, Charles W., & David
Odden.
Ogura, Masako. On the function of righthand NPs in Japanese. 7:2.49-64 (Fall
1977)
Onn, Farid M. Speech chain as an analysis-by-synthesis model: A review.
4:2.161-171 (Fall 1974)
See also: Becker, Lee A., & Farid Mohamed Onn.
Ortiz de Urbina, Jon. Partitive constructions, unaccusativity and ergativity. 15:1.
147-155 (Spring 1985)
Ortiz de Urbina, Juan M. Empty categories and focus in Basque. 13:1.133-156
(Spring 1983)
Osgood, Charles. Conservative words and radical sentences in the semantics
of international politics. 8:2.43-62 (Fall 1978)
Ottenheimer, Harriet, & Heather Primrose. Current research on ShiNzwani
ideophones. 19:2.77-87 (Fall 1989)
Ourso, Meterwa A. Root control, underspecification, and ATR harmony.
18:2.111-127 (Fall 1988)
See also: Kenstowicz, Michael, Emmanuel Nikiema, & Meterwa Ourso.
Pandharipande, Rajeshwari. Counteracting forces in language change:
Convergence vs. maintenance. 12:2.97-116 (Fall 1982)
Exceptions and rule government: The case of passive rule in Hindi.
8:1.153-173 (Spring 1978)
Interface of lexicon and grammar: Some problems in Hindi grammar.
11:2.77-100 (Fall 1981)
— . Mixing and creativity in multilingual India. 13:2.99-113 (Fall 1983)
. On the semantics of Hindi-Urdu calna^ 5:1.93-124 (Spring 1975)
Postpositions in passive sentences in Hindi. 9:2.171-188 (Fall 1979)
Review of Bhatia, A history of the Hindi grammatical traditior). 19:1.173-
179 (Spring 1989)
- — . Transitivity in Hindi. 11:2.161-179 (Fall 1981)
Index to Volumes 1-19 — Author index 191
, & Yamuna Kachru. Relational grammar, ergativity, and Hindi-Urdu.
6:1.82-99 (Spring 1976)
See also: Kachru, Yamuna, & Rajeshwari Pandharipande.
Patterson, Trudi A. Some morphological and phonological interactions in
Lakhota. 18:1.135-149 (Spring 1988)
Pearce, Elizabeth. Infinitival complements in Old French and diachronic
change. 12:2.1 17-145 (Fall 1982)
. Variation in case marking with infinitival and clausal complements in Old
French. 14:1.149-166 (Spring 1984)
Pingkarawat, Namtip — See: Lehman, F(rederic) K., with Namtip Pingkarawat.
Pitluck, Sally R. — See: Micklesen, Lew R., Sally R. Pitluck, & Edward J.Vajda.
Primrose, Heather — See: Ottenheimer, Harnet, & Heather Pnmrose.
Radanovic-Kocic, Vesna. Synonym split in the dialect of Bosnia and
Hercegovina: A study of a change in progress. 16:1.123-131 (Spring
1986)
Rauch, Irmengard. Language-likeness. 8:1.174-180 (Spring 1978)
. Semantic features inducing the Germanic dental preterit stem. 5:2.125-
138 (Fall 1975)
Riddle, Elizabeth. What they say about say. 5:1.1 13-1 12 (Spring 1975)
— , G. Sheintuch, & Y. Ziv. Pseudo-passivization: On the role of pragmatics in
determining rule unity. 7:1.147-156 (Spring 1977)
Rohrbach, Paula Chen. The acquisition of Chinese by adult English speakers:
An error analysis. 9:1.177-192 (Spring 1979)
Two notes on negation in Japanese. 13:1.157-174 (Spring 1983)
Rohsenow, John S. Perfect le: Temporal specification in Mandarin Chinese.
7:2.142-164 (Fall 1977)
Rosenberg, Marc. Another prefend paper. 4:1.179-197 (Spring 1974)
Rubach, Jerzy. Does the obligatory contour principle operate in Polish?
16:1.133-147 (Spring 1986)
Satyanarayana, Pulavarthi, & Karumuh V. Subbarao. Are rightward movement
rules upward bounded? 3:1.183-192 (Spring 1973)
Saxena, Anju — See: Subbarao, Karumuh V., & Anju Saxena.
Schaufele, Steven. Where's my NP? Non-transformational analyses of Vedic
pronominal fronting. 18:2.129-162 (Fall 1988)
Schmerling, Susan F. Apparent counterexamples to the coordinate structure
constraint: A canonical conspiracy. 2:1.91-104 (Spnng 1972)
A stress mess. 1:1.52-66 (Spring 1971)
Schwarte, Barbara. Intuitions of grammaticality and the 'law of contrast": A pilot
study. 4:1.198-216 (Spring 1974)
Searle, John. Intentionality and the use of language. 8:2.149-162 (Fall 1978)
Sheintuch, Gloria. On the gradation of grammatical relations. 6:1.186-194
(Spring 1976)
On the syntactic motivation for a category 'chomeur' in Relational
Grammar. 6:1.49-56 (Spnng 1976)
Periphrastic verb formation in Persian. 5:2.139-156 (Fall 1975)
Subject-raising — A unitary rule? 5:1.125-153 (Spring 1975)
192 Studies in the Linguistic Sciences 20:2 (Fall 1990)
See also: Dalgish, Gerard M., & Gloria Sheintuch; Riddle, E., G.
Sheintuch, & Y. Ziv.
Sherwood, Bruce Arne. Statistical analysis of conversational Esperanto, with
discussion of the accusative. 1 2:1 . 1 65-1 82 (Spring 1 982)
. Variation in Esperanto. 12:1.183-196 (Spring 1982)
, & Chin-Chuan Cheng. A linguistics course on international communication
and constructed languages. 10:1.189-201 (Spring 1980)
Shim, Seok-Ran. Umlaut in Korean. 16:2.147-155 (Fall 1986)
Shuy, Roger W. Multilingualism as a goal of educational policy. 8:2.107-122
(Fall 1978)
Siddiqui, Ahmad H. Notes on queclaratives and tag questions in Hindi-Urdu.
3:2.134-148 (Fall 1973)
Silver, Pamela S., & Scott R. Krause. A reanalysis of the class 5 prefix in
Shona. 8:1.181-196 (Spnng 1978)
Singh, Sadanand. Distinctive features: Theory and validation. Review by
Chin-W. Kim. 7:2.200-204 (Fall 1977)
Singh, V. D. Bazaar varieties of Hindi. 13:2.115-141 (Fall 1983)
Skousen, Royal. Consonant gradation in Finnish. 1:1.67-91 (Spring 1971)
Smietana, Diane — See: Cole, Peter, Wayne Harbert, Shikaripur Sndhar,
Sachiko Hashimoto, Cecil Nelson, & Diane Smietana.
Smith, Carlota S. Sentence topic in texts. 15:2.187-203 (Fall 1985)
Smith, N. v. Lexical representation and the acquisition of phonology. 8:2.259-
273 (Fall 1978)
Soheili-lsfahani, A. A comparative study of dialect variations in Iran. 6:1.38-46
(Spring 1976)
Sohn, Han, & C-W. Kim. Phonetic research at the University of Illinois. 4:2.172-
177 (Fall 1974)
See also: Kim, Chin-W., & Han Sohn.
Sohn, Hyang-Sook. Korean irregular verbs and nonlinear phonology.
15:1.157-193 (Spring 1985)
. Toward an integrated theory of morphophonology: Vowel harmony in
Korean. 16:2.157-184 (Fall 1986)
Sridhar, Kamal K. — See: Sndhar, S. N., & Kamal K. Sridhar.
Sridhar, S. N. Dative subjects, rule government, and Relational Grammar. 6.1.
130-151 (Spnng 1976)
Linguistic convergence: Indo-Aryanization of Dravidian languages.
8:1.197-216 (Spring 1978)
, & Kamal K. Sridhar. The syntax and psycholinguistics of bilingual code
mixing. 10:1.203-215 (Spring 1980)
See also: Cole, Peter, & S. N. Sridhar; Cole, Peter, Wayne Harbert,
Gabriella Hermon, & Shirakaripur Sridhar; Cole, Peter, Wayne Harbert,
Shikaripur Sridhar, Sachiko Hashimoto, Cecil Nelson, & Diane Smietana.
Stahike, Herbert. Derivational conditions on morpheme structure in Ewe.
14:2.97-113 (Fall 1984)
Ross' constraints and related problems in Yoruba. 3:1.193-230 (Spring
1973)
Index to Volumes 1-19 — Author index 193
Steever, Sanford B. Remarks on Dravidian complementation. 17:1.103,119
(Spring 1987)
. The serial verb formation in the Dravidian languages. Review article by
Hans Henrich Hock. 18:2.211-233 (Fall 1988)
Steffensen, Margaret. A deverbal analysis of adverbials in Hindi. 1:2.136-179
(Fall 1971)
Steinbergs, Aleksandra. Loanword incorporation processes: Examples from
Tshiluba. 14:2.115-125 (Fall 1984)
— . Some problems concerning the origin of the Latvian broken tone. 5:2.157-
185 (Fall 1975)
Stock, Roberta. On recently and lately. 3:1.231-248 (Spring 1973)
Stucky, Susan U. How a noun class system may be lost: Evidence from Kituba
(Lingua Franca Kikongo). 8:1.217-233 (Spring 1978)
Locatives as objects in Tshiluba: A function of transitivity. 6:2.174-179
(Fall 1976)
See also; Hodges, Kathryn Speed, & Susan U. Stucky.
Subbarao, Karumuri V. Notes on reflexivization in Hindi. 1:2.180-217 (Fall
1971)
, & T. K. Bhatia. A bibliography of research done on South Asian linguistics
and languages in the Department of Linguistics, University of Illinois at
Urbana-Champaign. 3:2.149-154 (Fall 1973)
, & Anju Saxena. Reflexives and reciprocals in Dravidian. 17:1.121-135
(Spring 1987)
See also: Arora, Harbir, & K. V. Subbarao; Satyanarayana, Pulavarthi, &
Karumuri V. Subbarao.
Surintramont, Aporn. Some aspects of underlying syllable structure in Thai:
Evidence from Khamphuan— a Thai word game. 3:1.121-142 (Spring
1973)
Taylor, Daniel J. Aspects of negation in classical Greek. 2:1.105-122 (Spring
1972)
Tegey, Habibullah. The relevance of morphological structure and of stress to
clitic placement rule-l in Pashto. 7:1.88-116 (Spring 1977)
A study of Pashto clitics and implications for linguistic theory. 5:1.154-190
(Spring 1975)
Teoh, Boon Seong. Geminates and inalterability in Malay. 17:2.125-136 (Fall
1987)
Timmons, Claude, & Christian Dunn. La selection morphophonologique des
classes en kpokolo. 19:2.135-151 (Fall 1989)
Toh, Soo-Hee. Glide y in Korean historical phonology. 7:2.178-183 (Fall 1977)
• On the relationship between the early Paekche language and the Kara
language in Korea. 16:2.185-201 (Fall 1986)
Treece, Rick. What is a Bantu noun class? 16:1.149-166 (Spring 1986)
Tsiang, Sarah. The discourse function of the absolutive in the Pancatantra.
18:2. 163-181 (Fall 1988)
194 Studies in the Linguistic Sciences 20:2 (Fall 1990)
, & Albert Watanabe. The Pancatantra and Aesop's Fables: A comparison
of rhetorical structure in classical Indian and western literature. 17:1.137-
146 (Spring 1987)
Tsutsui, Michio. Topic marker ellipsis in Japanese. 11:1.163-179 (Spring
1981)
Valentine, Tamara M. Interactional sociolinguistics and gender differentiation in
North Indian speech. 17:1.147-162 (Spring 1987)
— . Sex, power and linguistic strategies in the Hindi language. 15:1.195-211
(Spring 1985)
— . Sexism in Hindi: form, function, and variation. 13:2.143-158 (Fall 1983)
— See also: Blomeyer, Charlotte, & Tamara Valentine.
Vajda, Edward J. — See: Micklesen, Lew R., Sally R. Pitluck, & Edward J.
Vajda.
Vijaykrishnan, K. G. The syllable in phonological theory: Arguments from
Tamil. 11:2.101-105 (Fall 1981)
Vine, Brent. African 'shadow vowels': A descriptive survey. 14:2.127-137 (Fall
1984)
Wahba, Wafaa — See: Kenstowicz, Michael, & Wafaa Wahba.
Wallace, William D. The evolution of ergative syntax in Nepali. 12:2.147-211
(Fall 1982)
The government and binding analysis of Nepali EQUi and subject-raising
clauses. 17:1.163-179 (Spring 1987)
How registers register: A study in the language of news and sports.
7:1.46-78 (Spring 1977)
— . The interaction of word order and pragmatics in a Sanskrit text. 14:1.167-
188 (Spring 1984)
Object- marking in the history of Nepali: A case of syntactic diffusion. 1 1 :2.
107-128 (Fall 1981)
Wang, William S-Y. The three scales of diachrony. 8:2.63-76 (Fall 1978)
Wanner, Dieter. A note on diphthongization. 5:2.186-202 (Fail 1975)
Warie, Pairat. Some aspects of code-mixing in Thai. 7:1.21-40 (Spring 1977)
Watanabe, Albert — See: Tsiang, Sarah, & Albert Watanabe.
Welden, Ann. Stress in Cairo Arabic. 10:2.99-120 (Fall 1980)
Whelan, Peter — See: Irshied, Omar, & Peter Whelan.
White, Gregory Thomas. The derivation of each other. 3:1.249-258 (Spring
1973)
Wilbur, Ronnie. The identity constraint: An explanation of the irregular behavior
of some exceptional reduplicated forms. 3:1.143-154 (Spring 1973)
Wong, Maurice K. S. Origin of the high rising changed tone in Cantonese. 9:1.
193-206 (Spnng 1979)
The use of the high rising changed tone in Cantonese: A sociolinguistic
study. 7:2.65-81 (Fall 1977)
Wong-opasi, Uthalwan. On deriving specifiers in Spanish: Morpho-phono-
syntax interactions. 18:1.151-177 (Spring 1988)
Wongbiasaj, Soranee. On the passive in Thai. 9:1.207-216 (Spring 1979)
Index to Volumes 1-19 — Author index 195
Quantifier floating in Thai and the notions cardinality/ordinality. 9:2.189-
200 (Fall 1979)
Wood, Winifred J. — See: Kisseberth, Charles W., & Winifred J. Wood.
Yanco, Jennifer J. Modifiers in Bantu: Evidence from spoken Lingala.
14:2.139-147 (Fall 1984)
Yanofsky, Nancy — See: Holisky, Dee Ann, & Nancy Yanofsky.
Yeo, Sang-Pil. Fortition of loanwords in Korean. 16:2.203-214 (Fall 1986)
Yeoh, Chiang Kee. Restrictive relative clauses in Bahasa Malaysia. 7:2.82-105
(Fall 1977)
Yokwe, Eluzai M. Arabicization and language policy in the Sudan. 14:2.149-
170 (Fall 1984)
Yoon, James Hye-Suk. Some queries concerning the syntax of multiple subject
constructions in Korean. 16:2.215-236 (Fall 1986)
See also: Kendall, Sue Ann, & James Hye-Suk Yoon.
Zamir, Jan. Two social varieties of Farsi: 'Jaheli' and 'Armenian Persian' 13:2.
159-187 (Fall 1983)
Zgusta, Ladislav. Review of Katre, Astadhyayfof Panini. 19:1.187-193 (Spnng
1 989)
Review of Lara, El concepto de norma en linguistica. 7:2.191-192 (Fall
1977)
, & Hans H[enrich] Hock (eds.). Papers on historical linguistics: Theory and
method. 5:2 (Fall 1975)
Zhou, Xinping. On the head movement constraint. 18:2.183-210 (Fall 1988)
Ziv, Yael. On the diachronic relevance of the promotion to subject hierarchy.
6:1. 195-215 (Spring 1976)
Restrictive relatives with generic heads — are they 'its'? 3:1.259-273
(Spring 1973)
Riddle, E., G. Sheintuch, & Y. Ziv
Zwicky, Arnold M. The case against plain vanilla syntax. 15:2.205-225 (Fall
1985)
B. Title index
Accent in Bantu: An appraisal. Larry M. Hyman. 19:2.115-134 (Fall 1989)
The accentuation of Lithuanian derived nominals. Karen Dudas. 2:2.108-136
(Fall 1972)
Accusative subjects in Imbabura Quechua. Peter Cole & Janice Jake. 8:1.72-
96 (Spring 1978)
The acquisition of Chinese by adult English speakers: An error analysis. Paula
Chen Rohrbach. 9:1.177-192 (Spring 1979)
Adjective flipping and the notion of target structure. David E. Johnson. 4:1.59-
79 (Spring 1974)
African linguistic research and publications from the University of Illinois at
Urbana-Champaign, 1970-1979. Eyamba G. Bokamba. 9:2.201-206 (Fall
1979)
African linguistics and its conthbutions to linguistic theory. G. N. Clements.
19:2.3-39 (Fall 1989)
African 'shadow vowels': A descriptive survey. Brent Vine. 14:2.127-137 (Fall
1984)
Analogical change as a problem for linguistic theory. Paul Kiparsky. 8:2.72-96
(Fall 1978)
Another look at nara conditionals. Susan Meredith Burt. 9:2.25-38 (Fall 1979)
Another prefenc/ paper. Marc Rosenberg. 4:1.179-197 (Spring 1974)
An apparent asymmetry in the formation of relative clauses in Modern Hebrew.
Peter Cole. 5:1.1-35 (Spring 1975)
Apparent counterexamples to the coordinate structure constraint: A canonical
conspiracy. Susan F. Schmerling. 2:1.91-104 (Spring 1972)
Applied linguistics and foreign language teaching: A non-Western perspective.
Yamuna Kachru. 15:1.91-107 (Spnng 1985)
Applied linguistics and foreign language teaching: A non-Western perspective.
Yamuna Kachru. 16:1.35-51 (Spring 1986)
Arabicization and language policy in the Sudan. Eluzai M. Yokwe. 14:2.149-
170 (Fall 1984)
Archaisms, morphophonemic metrics, or vahable rules in the Rig-Veda? Hans
Henrich Hock. 10:1.59-69 (Spring 1980)
Are rightward movement rules upward bounded? Karumuri V. Subbarao &
Pulavarthi Satyanarayana. 3:1.183-192 (Spring 1973)
Arguments for a unified treatment of y-mitial and vowel-mitial roots in Olu-
tsootso. Gerry Dalgish. 4:2.76-90 (Fall 1974)
Aspects of Iraqi Arabic verbal phonology. David Odden. 8:1.137-152 (Spring
1978)
Aspects of Mandingo complementation. Mallafe Drame. 9:2.67-90 (Fall 1979)
Aspects of negation in classical Greek. Daniel J. Taylor. 2:1.105-122 (Spring
1972)
Aspects of tone assignment in Kimatuumbi. Charles W. Kisseberth & David
Odden. 10:1.125-140 (Spring 1980)
198 Studies in the Linguistic Sciences 20:2 (FaU 1990)
Aspectual elements of simultaneity and interaction in Indian languages: A case
for an area! universal. Anvita Abbi & Mithilesh Kumar Mishra. 17:1.1-14,
(Spring 1987)
Assertions not made by the main clause of a sentence. Use Menn. 4:1.132-
143 (Spring 1974)
AstadhyayFof Panini, by Sumitra M. Katre. Review by Ladislav Zgusta. 19:1.
187-193 (Spring 1989)
Authenticity and the choice of a national language: The case of Zaire. Eyamba
G. Bokamba. 6:2.23-64 (Fall 1976)
An autosegmental analysis of Venda nominal tonology. Farida Cassimjee.
13:1.43-72 (Spring 1983)
Aux-cliticization as a motivation for word order change. Hans Henrich Hock.
12:1.91-101 (Spring 1982)
The Baltic e-preterit: An older a-preterit? Hans Henrich Hock. 2:2.137-164
(Fall 1972)
Bazaar varieties of Hindi. V.D.Singh. 13:2.1 15-141 (Fall 1983)
A bibliography of research done on South Asian linguistics and languages in
the Department of Linguistics, University of Illinois at Urbana-Champaign.
K. V. Subbarao & Tej. K. Bhatia. 3:2.149-154 (Fall 1973)
The bilingual's creativity: Discoursal and stylistic strategies in contact literatures
in English. Braj B. Kachru. 13:2.37-55 (Fall 1983)
Binding domains in Kikuyu. George N. Clements. 14:2.37-56 (Fall 1984)
The bivalence of NEG raising predicates. Richard Neal Halpern. 6:1.69-81
(Spring 1976)
Bloom-p-field, Chom-p-sky, and phonetic epen-t-thesis. John G. Barnitz. 4:2.1-
13 (Fall 1974)
A brief note on the reconstruction of MA? in Tibeto-Burman. Frederic K.
Lehman. 7:2.24-38 (Fall 1977)
'C-command' and the phonology-syntax interface in Ciluba. Nkonko Mudipanu
Kamwangamalu. 18:2.87-109 (Fall 1988)
The case against plain vanilla syntax. Arnold M. Zwicky. 15:2.205-225 (Fall
1985)
Case and control in Hindi-Urdu. Alice Davison. 15:2.9-23 (Fall 1985)
A case of systematic avoidance of homonyms. Charles W. Kisseberth &
Mohammad Imam Abasheikh. 4:1.107-124 (Spring 1974)
Causative sentences in Hindi revisited. Yamuna Kachru. 1:2.75-103 (Fall
1971)
Chimwi:ni prefix morphophonemics. Charles W. Kisseberth & Mohammad
Imam Abasheikh. 6:2.142-173 (Fall 1976)
Chinese dialect affinity based on syllable initials. Zhiji Lu & Chin-Chuan
Cheng. 15:2.127-148 (Fall 1985)
Cinema-verite, golf-bijou, and sandwich beurre in contemporary French. Fred
M.Jenkins. 6:1.1-21 (Spring 1976)
CiRuri tonology (A preliminary view). Lee A. Becker & David P. B. Massamba.
10:1.1-13 (Spring 1980)
Index to Volumes 1 - 19 — Title index 199
Clause union and Relational Gramnnar: Evidence from Hebrew and Kannada.
Peter Cole & S. N. Sridhar. 6:1.216-227 (Spring 1976)
Clitic verbs in PIE or discourse-based verb fronting? Sanskrit sa hovaca gir-
gyah and congeners m Avestan and Homeric. Hans Henrich Hock. 12:2.
1-38 (Fall 1982)
Clitics and the double object construction in Cairene Arabic. Michael
Kenstowicz & Wafaa Wahba. 10:2.149-163 (Fall 1980)
Codification of non-native English: Is it necessary/possible? Jean D'souza. 16:
1.1-11 (Spring 1986)
A comparative study of dialect variations in Iran. A. Soheili-lsfahani. 6:1.38-46
(Spring 1976)
Competence for implicit text analysis: Literary style discrimination in five-year-
olds. Georgia M. Green, 11:1.39-56 (Spring 1981)
A computational investigation on the perception of aspirated consonants in
Hindi. Tej K. Bhatia. 3:1.63-80 (Spring 1973)
A computer tool in the study of Taiwanese tones. Tsai-Chwun Du. 17:2.45-62
(Fall 1987)
El concepto de norma en linguistica, by Luis Fernando Lara. Review by
Ladislav Zgusta. 7:2,191-192 (Fall 1977)
Conjoined we stand: Theoretical implications of Sanskrit relative structures.
Hans Hennch Hock. 19:1.93-126 (Spnng 1989)
Conservative words and radical sentences in the semantics of international
politics. Charles Osgood 8:2.43-62 (Fall 1978)
Consonant gradation in Finnish. Royal Skousen. 1:1.67-91 (Spring 1971)
Consonant merger in Navajo: An underspecified analysis. J. Fraser Bennett.
17:2.1-19 (Fall 1987)
Conspiracy in Korean phonology revisited: As applied to historical data. Sang
Oak Lee. 7:2.1-23 (Fall 1977)
Contextual effects on 'generic' indefinites: Cross-linguistic arguments for
pragmatic factors. Alice Davison. 9:2.55-66 (Fall 1979)
The contribution of African linguistics to linguistic theory: Proceedings of the
20th Annual Conference on African Linguistics, Vol. 1, ed. by Eyamba G.
Bokamba. 19:2 (Fall 1989)
Convergence and syntactic reanalysis: The case of so in Dakkhini. Harbir
Arora & K, V, Subbarao. 19:1.1-18 (Spring 1989)
Corpus planning for modernization: Sanskntization and Englishization of Hindi.
Yamuna Kachru. 19:1.153-164 (Spring 1989)
Correlation among attitudinal factors, speed, and tone sandhi in Chinese. Yen
Ling Lee. 7:2.129-141 (Fall 1977)
Counteracting forces in language change: Convergence vs. maintenance.
Rajeshwari Pandhahpande. 12:2.97-116 (Fall 1982)
Crisis of identity: Sikhs In England, by Rama Kant Agnihotn. Review by Jean
Aitchison. 19:1,169-171 (Spring 1987)
Cross-cultural texts and interpretation. Yamuna Kachru. 13:2.57-72 (Fall 1983)
Current research on ShiNzwani ideophones. Harriet Ottenheimer & Heather
Primrose. 19:2.77-87 (Fall 1989)
200 Studies in the Linguistic Sciences 20:2 (Fall 1990)
CV-skeleton or X-skeleton: The Turkish evidence. Diana Archangeli. 15:1.1-10
(Spring 1985)
Cyclic stress in Levantine Arabic. Michael Kenstowicz & Kama! Abdul-Karim.
10:2.55-76 (Fall 1980)
Dari relative clauses. John R. Houston, Jr. 4:1.32-58 (Spring 1974)
Dative subjects, rule government, and Relational Grammar. S. N. Sridhar. 6:1.
130-151 (Spring 1976)
Demonstrative variation in Oral Media Arabic in Egypt. Carolyn G. Killean. 10:
2.165-178 (Fall 1980)
The derivation of a relative infinitive construction. Georgia Green. 3:1.1-32
(Spring 1973)
The derivation of each other. Gregory Thomas White. 3:1.249-258 (Spring
1973)
Derivational conditions on morpheme structure in Ewe. Herbert Stahlke.
14:2.97-113 (Fall 1984)
Derived imperfectives in Slavic: A study in derivational morphology. Lew R.
Micklesen, Sally R. Pitluck, & Edward J. Vajda. 15:2.171-186 (Fall 1985)
Devanagarf word-processing on the IBM-PC. Stephen C. Helmreich. 17:1.51-
61 (Spring 1987)
The development of personal pronouns in modern Gondi. U. Rao Garapati.
17:1.35-50 (Spring 1987)
The development of the Slavic mid vowels in newly checked syllables in the
Northwestern Ukrainian dialects. Christine Yurkiw Bethin. 5:2.1-11 (Fall
1975)
A deverbal analysis of adverbials in Hindi. Margaret Steffensen. 1:2.136-179
(Fall 1971)
The diachronic development of nasal deletion in Olutsootso. Gerry Dalgish.
5:2.25-40 (Fall 1975)
Dimensions of South Asian linguistics, ed. by Yamuna Kachru. 1 1 :2 (Fall 1 981 )
The discourse function of the absolutive in the Pancatantra. Sarah Tsiang. 18:
2.163-181 (Fall 1988)
Discourse structure in a Hindi short story. Suresh Kumar. 11:2.51-66 (Fall
1981)
Displaced tones in Digo (Part I). Charles W. Kisseberth & Winifred J. Wood.
10:1.141-177 (Spring 1980)
Displaced tones in Digo (Part 2). Charles W. Kisseberth. 11 :1. 73-1 20 (Spring
1981)
Distinctive features: Theory and validation, by Sadanand Singh. Review by
Chin W. Kim. 7:2.200-204 (Fall 1977)
Do inversions in English change grammatical relations? Georgia M. Green. 7:
1.157-181 (Spring 1977)
Does the obligatory contour principle operate in Polish? Jerzy Rubach. 16:1.
133-147 (Spring 1986)
Downstep in Venda. Farida Cassimjee & Charles W. Kisseberth. 14:1.1-29
(Spnng 1984)
Index to Volumes 1-19 — Title index 201
Emotive verbs in English and Japanese. Noriko Akatsuka. 2:1.1-15 (Spring
1972)
Empty categories and focus in Basque. Juan M. Ortiz de Urbina. 13:1.133-156
(Spring 1983)
Enhanced miniature artificial languages. John A. Haller. 13:1.73-81 (Spring
1983)
Epenthesis and elision in metrical phonology. Chin-W. Kim. 11:1.57-71
(Spring 1981)
Epenthesis and geminate consonants in Palestinian Arabic. Issam M. Abu-
Salim. 10:2.1-11 (Fall 1980)
Epenthesis and sonority in Lebanese Arabic. Ghassan F. Haddad. 14:1.57-88
(Spring 1984)
Ergative and argative (nee ergatlve too). Catherine V. Chvany. 15:2.1-2 (Fall
1985)
The ergative case in Hindi-Urdu. Baber S. A. Khan. 17:1.91-101 (Spring 1987)
ESP and non-native varieties of English: Toward a shift in paradigm. Braj B.
Kachru. 16:1.13-34 (Spnng 1986)
ESP in the classroom: Practice and evaluation. ELT Documents 128, ed. by
Dick Chamberlain & Robert Baumgardner. Review by Numa Markee. 19:
1.181-185 (Spnng 1989)
The Esperanto of El Popola Cinio. Chin-Chuan Cheng. 12:1.49-62 (Spring
1982)
Evidence for global constraints: The case of reflexivization in Hindi-Urdu.
Yamuna Kachru & Tej K. Bhatia. 5:1.42-73 (Spring 1975)
Evidence for the Elsewhere Condition in Shona. David Odden. 11:1.145-162
(Spring 1981)
The evolution of ergative syntax in Nepali. William D. Wallace. 12:2.147-211
(Fall 1982)
The evolution of tones in Punjabi. Tej K. Bhatia. 5:2.12-24 (Fall 1975)
Exceptions and rule government: The case of passive rule in Hindi. Rajeshwari
Pandhanpande. 8:1.153-173 (Spnng 1978)
Exceptions and synchronic analogy in Sanskrit. Hans Henrich Hock. 3:1.81-
101 (Spring 1973)
The exceptions to passive in English: A pragmatic hypothesis. Richard D.
Cureton. 9:2.39-54 (Fall 1979)
An explanation of syllable structure change m Korean: With special reference
to Vennemann's preference laws. Sang-Oak Lee. 16:2.121-133 (Fall
1986)
Exploring the dictionary: On teaching foreign learners of Arabic to use the
Arabic-English dictionary. Omar Irshied & Peter Whelan. 18:1.61-75
(Spring 1988)
Faulty referents and their relationship to tense. Susan Kay Donaldson.
3:1.155-171 (Spnng 1973)
Finiteness in Dravidian. The serial verb formation in the Dravidian languages,
by Sanford B. Steever. Review article by Hans Henrich Hock. 18:2.211-
233 (Fall 1988)
202 Studies in the Linguistic Sciences 20:2 (Fall 1990)
Formal vs. functional considerations in phonology. Morris Halle. 8:2.123-134
(Fall 1978)
Fortition of loanwords in Korean. Sang-Pil Yeo. 16:2.203-214 (Fall 1986)
French colonial language policy in Africa and its legacies (Part I). Eyamba G.
Bokamba. 14:2.1-35 (Fall 1984)
A functional typology of ni in Kivunjo (Chaga). Lioba Moshi. 18:1.105-134
(Spring 1988)
Functionalism and functional explanations in phonology. Jonathan Kaye. 8:2.
135-148 (Fall 1978)
Further comments on instrumentals. Richard J. Leskosky. 2:1.66-83 (Spring
1972)
Geminates and inalterability in Malay. Boon Seong Teoh. 17:2.125-136 (Fall
1987)
Gemination and spirantization in Tighnya. Michael Kenstowicz, 12:1.103-122
(Sphng 1982)
General linguistic studies in Hindi: A review of resources. Braj B. Kachru.
3:2.59-86 (Fall 1973)
Glide y in Korean historical phonology. Soo-Hee Toh. 7:2.178-183 (Fall 1977)
Global grammar versus index grammar: A question of power. Peter Cole.
3:1.45-53 (Spring 1973)
Good mixes and odd mixes: Implications for the bilingual's grammar. (Squib.)
Rakesh Mohan Bhatt. 19:1.165-168 (Spring 1989)
Governed-rule change and Universal Grammar. Georgia M. Green. 6:1.152-
169 (Spring 1976)
The government and binding analysis of Nepali equi and subject-raising
clauses. William D. Wallace. 17:1.163-179 (Spring 1987)
Headless relative clauses in Quechua. Peter Cole, Wayne Harbert, & Gabriella
Hermon. 8:1.26-41 (Spring 1978)
Hindi verbs of judging: An application of Fillmore's system of semantic des-
cription. Richard Lutz. 10:1.179-188 (Spring 1980)
A history of the Hindi grammatical tradition, by Tej. K. Bhatia. Review by
Rajeshwari Pandharipande. 19:1.173-179 (Spring 1989)
A history of hght dislocation in certain Levantine Arabic dialects. Anthony Britti.
10:2.121-139 (Fall 1980)
How a noun class system may be lost: Evidence from Kituba (Lingua Franca
Kikongo). Susan U. Stucky. 8:1.217-233 (Spring 1978)
How do tense and aspect interact in determination of verb forms; Verb past
foms and non-past forms in Japanese 'when'-clauses. Tsuneko
Nakazawa. 15:1.135-146 (Spring 1985)
How registers register: A study in the language of news and sports. William D.
Wallace. 7:1.46-78 (Spring 1977)
The identity constraint: An explanation of the irregular behavior of some
exceptional reduplicated forms. Ronnie Wilbur. 3:1.143-154 (Sphng
1973)
Ikorovere Makua tonology (Part 1). Chin-Chuan Cheng & Charles W.
Kisseberth. 9:1.31-64 (Spring 1979)
Index to Volumes 1 - 19 — Title index 203
Ikoravere Makua tonology (Part 2). Chin-Chuan Cheng & Charles W.
Kisseberth. 10:1.15-44 (Spring 1980)
Ikorovere Makua tonology (Part 3). Chin-Chuan Cheng & Charles W.
Kisseberth. 11:1.181-202 (Spnng 1981)
Illinois studies in Korean linguistics, ed. by Chin-W. Kim. 16:2 (Spring 1986)
Impact of expanding domains of use on a standard language: Contemporary
Hindi in India. Yamuna Kachru. 17:1.73-90 (Spring 1987)
In defence of implosives. Paroo Nihalani. 16:1.113-122 (Spring 1986)
The inclusion constraint: Description and explanation. Richard D. Cureton.
9:1.91-104 (Spring 1979)
The indicator particle baa in Somali. Michal Allon Llvnat. 13:1.89-132 (Spnng
1983)
An inference-based account of restrictive relative whicti and ttiat. Camille
Bundrick. 19:1.19-31 (Spring 1989)
Infinitival complements in Old French and diachronic change. Elizabeth
Pearce. 12:2.117-145 (Fall 1982)
Inflectional accent of the Serbo-Croatian noun. Michael Kenstowicz. 4:1.80-106
(Spring 1974)
Intensive reflectives. Richard J. Leskosky. 2:1.42-65 (Spring 1972)
Intentionality and the use of language. John Searle. 8:2.149-162 (Fall 1978)
The Interaction of word order and pragmatics In a Sanskrit text. William D.
Wallace. 14:1.167-188 (Spring 1984)
Interactional sociolinguistics and gender differentiation in North Indian speech.
Tamara M. Valentine. 17:1.147-162 (Spring 1987)
Interface of lexicon and grammar: Some problems in Hindi grammar.
Rajeshwari Pandharipande. 11:2.77-100 (Fall 1981)
Internal change in a transplanted language. Nicole Domingue. 11:2.151-159
(Fall 1981)
Introduction. Jerry Morgan. 6:1.47-48 (Spring 1976)
Intuitions of grammaticality and the 'law of contrast': A pilot study. Barbara
Schwarte. 4:1.198-216 (Spring 1974)
Inversions as grammatical relation changing rules in Bantu languages.
EyambaG. Bokamba. 9:2.1-24 (Fall 1979)
Is there an a-epenthesis in Sanskrit? Hans Henrich Hock. 3:2.43-58 (Fall
1973)
Korean irregular verbs and nonlinear phonology. Hyang-Sook Sohn. 15:1.
157-193 (Spring 1985)
The Korean writing system: A linguistic examination. Yong-ll Kong. 16:2.107-
119 (Fall 1986)
La selection morphophonologique des classes en kpokolo. Claude Timmons &
Christian Dunn. 19:2.135-151 (Fall 1989)
Language and national development in sub-Saharan Africa (A progress report).
Eyamba Bokamba. 11 :1. 1-25 (Spring 1981)
Language differences in strategies for the interactional management of
conversation. Chet A. Creider. 14:2.57-65 (Fall 1984)
204 Studies in the Linguistic Sciences 20:2 (Fall 1990)
Language universals in linguistic argunnentation. James McCawley. 8:2.205-
220 (Fall 1978)
Language variation and language standardization: The case of Shona In
Zimbabwe. Benjamin J. Magura. 13:2.87-98 (Fall 1983)
Language-death phenomena in Sanskht: Grammatical evidence for attrition in
contemporary spoken Sanskrit. Hans Henrich Hock. 13:2.21-35 (Fall
1983)
Language in African culture and society, ed. by Eyamba G. Bokamba. 14:2 (Fall
1984)
Language-likeness. Irmengard Rauch. 8:1.174-180 (Spring 1978)
Lexical modernization in Bahasa Indonesia: Functional allocation and variation
in borrowing. Peter H. Lowenberg. 13:2.73-86 (Fall 1983)
A lexical phonology approach to Hindi schwa. Camille Bundrick. 17:1.15-23
(Spring 1987)
Lexical representation and the acquisition of phonology. N. V. Smith. 8:2.259-
273 (Fall 1978)
Linguistic convergence: Indo-Aryanization of Dravidian languages. S. N.
Sridhar. 8:1.197-216 (Spring 1978)
Linguistic studies in memory of Theodore M. Ligfitner, ed. by Michael J. Kensto-
wicz. 15:2 (Fall 1985)
A linguistics course on international communication and constructed langu-
ages. Bruce Arne Shenwood & Chin-Chuan Cheng. 10:1.189-201 (Spring
1980)
Linguistics in the seventies: Directions and prospects, ed. by Braj B. Kachru.
8:2 (Fall 1978)
Linguistics is about languages. Bernard Comne. 8:2.221-236 (Fall 1978)
Lithuanian phonology. Michael J. Kenstowicz. 2:2.1-85 (Fall 1972)
Lithuanian verbal accentuation. Karen Dudas & Margie O'Bryan. 2:2.86-107
(Fall 1972)
Loanword incorporation processes: Examples from Tshiluba. Aleksandra
Steinbergs. 14:2.115-125 (Fall 1984)
Locatives as objects in Tshiluba: A function of transitivity. Susan U. Stucky.
6:2.174-179 (Fall 1976)
A matter of scope: McCawley versus Postal on the origin of noun phrases.
Peter Cole. 7:1.117-146 (Spnng 1977)
Mirror of language: The debate on bilingualism, by Kenji Hakuta. Review by
Cher-leng Lee. 17:2.137-142 (Fall 1987)
Missing nominals, non-specificity and related matters, with especial reference to
Thai and Burmese. F. K. Lehman with Namtip Pingkarawat. 15:2.101-121
(Fall 1985)
Mixing and creativity in multilingual India. Rajeshwari Pandhahpande. 13:2.99-
113 (Fall 1983)
Modification of the gender system in the Wollegan dialect of Oromo. Cynthia
Robb Clamons. 19:2.187-195 (Fall 1989)
Modifiers in Bantu: Evidence from spoken Lingala. Jennifer J Yanco.
14:2.139-147 (Fall 1984)
Index to Volumes 1 - 19 — Title index 205
The morphological rule of learned backing and lexical phonology. Donna M.
Farina. 14:1.31-56 (Spring 1984)
Morphologically conditioned changes in Wanka-Quechua. Rodolfo Cerron-
Palomino. 4:2.40-75 (Fall 1974)
Movement in restrictive relative clauses in Hindi. Susan K. Donaldson. 1:2.1-
74 (Fall 1971)
Multilingualism as a goal of educational policy. Roger W. Shuy. 8:2.107-122
(Fall 1978)
Multilingualism as object of linguistic deschption. Charles A. Ferguson. 8:2.97-
106 (Fall 1978)
Multiword lexical units in French. Donna M. Farina. 17:2.63-76 (Fall 1987)
Negation in Korean and pragmatic ambiguity. Virginia K. McClanahan. 16:2.
135-145 (Fall 1986)
Neutralization in Korean revisited. Chin-W. Kim. 9:1.147-156 (Spring 1979)
Not your usual use of you. Richard J. Leskosky. 4:1.125-131 (Spring 1974)
A note on the accentuation of Damascene Arabic. John J. McCarthy. 10:2.77-
98 (Fall 1980)
A note on diphthongization. Dieter Wanner. 5:2.186-202 (Fall 1975)
A note on the intransitive nature of the Japanese raising verb omo'u and its
implications. Seiichi Makino. 7:2.39-48 (Fall 1977)
A note on the parallels between the definite article and the relative clause
marker in Arabic. Ghassan Haddad & Michael Kenstowicz. 10:2.141-147
(Fall 1980)
A note on seem. Richard Neil Halpern. 7:1.79-87 (Spring 1977)
A note on tonal conjunction in Efik. Chin-W. Kim. 4:2.112-122 (Fall 1974)
Notes on Cairene Arabic syncope. Michael Kenstowicz. 10:2.39-53 (Fall 1980)
Notes on clefts and pseudo-clefts and other related matters. Georgia M. Green.
1:1.1-7 (Spring 1971)
Notes on queclaratives and tag questions in Hindi-Urdu. Ahmad H. Siddiqui.
3:2.134-148 (Fall 1973)
Notes on the origin of quantifier floating. Richard Neil Halpern. 7:1.41-45
(Spring 1977)
Notes on reflexivization in Hindi. Karumuri V. Subbarao. 1:2.180-217 (Fall
1971)
Notes toward an understanding of rule government. G. M. Green & J. L.
Morgan. 6:1.228-248 (Spring 1976)
Noun phrase accessibility and island constraints. Peter Cole, Wayne Harbert,
Shikaripur Shdhar, Sachiko Hashimoto, Cecil Nelson, & Diane Smietana.
6:1.170-185 (Spring 1976)
Noun phrases as quantifiers. Peter Cole. 2:1.16-41 (Spnng 1972)
Object-marking in the history of Nepali: A case of syntactic diffusion. William D.
Wallace. 1 1 :2. 107-128 (Fall 1981)
Object relations in Kimeru causatives. Kathryn Speed Hodges. 6:2.108-141
(Fall 1976)
The 'object' relationship in Chii-Mwi:ni, a Bantu language. Charles W.
Kisseberth & Mohamed Imam Abasheikh. 6:1.100-129 (Spring 1976)
206 Studies in the Linguistic Sciences 20:2 (Fall 1990)
Object verb agreement in Tigre. Janice L. Jake. 10:1.71-84 (Spring 1980)
On the acquisition of subjecthood. Peter Cole, Wayne Harbert, Gabrieiia Her-
mon, & Shikaripur Sridhar. 8:1.42-71 (Spring 1978)
On deriving specifiers in Spanish: Morpho-phono-syntax interactions. Uthal-
wan Wong-opasi. 18:1.151-177 (Spring 1988)
On the diachronic relevance of the promotion to subject hierarchy. Yael Ziv.
6:1.195-215 (Spring 1976)
On ergativity in selected South Asian languages. Yamuna Kachru & Rajesh-
wari Pandharipande. 8:1.111-126 (Spring 1978)
On the function of righthand NPs in Japanese. Masako Ogura. 7:2.49-64 (Fall
1977)
On glides following vocalic verbs in Russian. Frank Y. Gladney. 15:2.39-59
(Fall 1985)
On the gradation of grammatical relations. Gloria Sheintuch. 6:1.186-194
(Spring 1976)
On the Head Movement Constraint. Xinping Zhou. 18:2.183-210 (Fall 1988)
On the inadequacy of a grammatical relation referring rule in Bantu. Kathryn
Speed Hodges & Susan U. Stucky. 9:2.91-100 (Fall 1979)
On the interaction of phonology and morphology: A Chi-Mwi:ni example.
Charles W. Kissebertha & fy/lohammad Imam Abasheikh. 4:2.139-147 (Fall
1974)
On the justification for language-specific sub-grammatical relations. Gerard 1^.
Dalgish & Gloria Sheintuch. 6:2.89-107 (Fall 1976)
On the morphology of morphological causative verbs in Korean: An argument
against Lieber's fvlorpheme-based lexicon. Euiyon Cho. 16:2.27-43 (Fall
1986)
On the nature of /) in Korean. Sang-Cheol Ahn. 16:2.1-13, (Fall 1986)
On the notion of 'second position" in Greek. Richard Cervin. 18:2.23-39 (Fall
1988)
On the passive in Thai. Soranee Wongbiasaj. 9:1.207-216 (Spring 1979)
On the (possibly) presuppositional nature of when-clauses in Hindi. Susan Kay
Donaldson. 3:2.28-42 (Fall 1973)
On the pragmatic motivation for perhaps. Dee Ann Holisky & Nancy Yanofsky.
9:2.101-108 (Fall 1979)
On the progressive palatalization of early Slavic: Synchrony versus history.
Horace G. Lunt. 15:2.149-169 (Fall 1985)
On recently and lately. Roberta Stock. 3:1.231-248 (Spring 1973)
On the relationship between the early Paekche language and the Kara langu-
age in Korea. Soo-Hee Toh. 16:2.185-201 (Fall 1986)
On the representation of contour tones in Generative Phonology. Lee A.
Becker. 7:1.8-20 (Spring 1977)
On the representation of Kejia diphthongs. Raung-fu Chung. 19:1.63-80
(Spring 1989)
On the scope of negation in Hindi. Tej K. Bhatia. 3:2.1-27 (Fall 1973)
On the semantics of Hindi-Urdu cdlna. Rajeshwari Pandharipande. 5:1.93-124
(Spring 1975)
Index to Volumes 1 - 19 — Title index 207
On surface form and logical form. Edward Keenan. 8:2.163-204 (Fall 1978)
On the syntactic motivation for a category 'chomeur' in Relational Grammar.
Gloria Sheintuch. 6:1.49-56 (Spring 1976)
On the syntax and semantics of wh-questions in Kikongo and Kiswahili.
Eyamba G. Bokamba. 6:2.65-88 (Fall 1976)
On the syntax, semantics, and pragmatics of the conjunctive participle in Hindi-
Urdu. Yamuna Kachru. 11:2.35-49 (Fall 1981)
Opacity and rule loss. Margie O'Bryan. 4:2.148-160 (Fall 1974)
Opacity and the loss of a morphological process. Margie O'Bryan. 3.1.102-120
(Spring 1973)
Origin of the high rising changed tone in Cantonese. Maurice K. S. Wong. 9:1.
193-206 (Spring 1979)
Orthography and the formulation of phonological rules. Esther Bentur. 8:1.1-25
(Spring 1978)
Palatalization in Russian. Herbert S. Coats. 15:2.3-8 (Fall 1985)
The Pancatantra and Aesop's Fables: A comparison of rhetorical structure in
classical Indian and western literature. Sarah Tsiang & Albert Watanabe.
17:1.137-146 (Spring 1987)
Papers from the 1986 South Asian Languages Analysis Roundtable, ed. by
Hans Henrich Hock. 17:1 (Spring 1987)
Papers on African linguistics, ed. by Eyamba G. Bokamba & Charles W.
Kisseberth. 6:2 (Fall 1976)
Papers on diachronic syntax: Six case studies, ed. by Hans Henrich Hock.
12:2 (Fall 1982)
Papers on Hindi syntax, ed. by Yamuna Kachru. 1 :2 (Fall 1 971 )
Papers on historical linguistics: Theory and method, ed. by Ladislav Zgusta &
Hans H[enrich] Hock. 5:2 (Fall 1975)
Papers on phonetics and phonology, ed. by Charles W. Kisseberth & Chin W.
Kim. 4:2 (Fall 1974)
Papers on South Asian linguistics, ed. by Braj B. Kachru. 3:2 (Fall 1973)
Papers on syntax and semantics, ed. by Georgia M. Green. 2:1 (Spring 1972)
Partitive constructions, unaccusativity and ergativity. Jon Ortiz de Urbina. 15:1.
147-155 (Spring 1985)
Passive in Persian. Mohammad Dabair-Moghaddam. 12:1.63-90 (Spring
1982)
Passivization in Bantu languages: Implications for relational grammar. Nkonko
Mudipanu Kamwangamalu. 15:1.109-133 (Spring 1985)
Passivizing locatives in Olutsootso. Gerard M. Dalgish. 6:1.57-68 (Spring
1976)
Perfect le: Temporal specification in Mandarin Chinese. John S. Rohsenow.
7:2.142-164 (Fall 1977)
The perfect stem in Chi-Mwi:ni. Charies W. Kisseberth & Mohammad Imam
Abasheikh. 4:2.123-148 (Fall 1974)
Periphrastic verb formation in Persian. Gloria Sheintuch. 5:2.139-156 (Fall
1975)
208 Studies in the Linguistic Sciences 20:2 (Fall 1990)
The persistence of pattern in language. Winfred P. Lehmann. 15:2.123-126
(Fall 1985)
A phonetic model for reading: Evidence from Korean. Chin-W. Kim & Han
Sohn. 16:2.95-105 (Fall 1986)
Phonetic research at the University of Illinois. Han Sohn & C-W. Kim. 4:2.172-
177 (Fall 1974)
The phonology and syntax of wh-expressions in Tangale. Michael Kenstowicz.
15:2.79-91 (Fall 1985)
The phonology of Chukchee consonants. Michael Kenstowicz. 16:1.79-96
(Spring 1986)
Phonology on the C-string? Chin W. Kim. 15:2.93-100 (Fall 1985)
Physical identification in Kannada. D. N. S. Bhat. 11:2.1-8 (Fall 1981)
Poguli syntax in the light of Kashmiri: A preliminary report. Peter Edwin Hook.
17:1.63-71 (Spring 1987)
Postpositions in passive sentences in Hindi. Rajeshwari Pandharipande. 9:2.
171-188 (Fall 1979)
Pragmatics and syntactic description. Georgia M. Green. 11:1.27-37 (Spring
1981)
Pragmatics and verb serialization in Hindi-Urdu. Yamuna Kachm. 9:2.157-170
(Fall 1979)
(Pre-)Rig-Vedic convergence of Indo-Aryan with Dravidian? Another look at the
evidence. Hans Henrich Hock. 14:1.89-108 (Spring 1984)
Prefixing, voicing, and syllable reduction in Burmese: Juncture and syllable
structure. F. K. Lehman. 3:2.104-120 (Fall 1973)
Preliminaries to the reconstruction of verbal endings in Yuk. J. L. Morgan.
7:1.1-7 (Spring 1977)
Primitiveness, naturalness, and cultural fit. F. Joseph Foster. 15:2.25-37 (Fall
1985)
Principles of stress assignment: A crosslinguistic view. David Odden. 9:1.157-
176 (Spring 1979)
Problems in a prosodic analysis of Hebrew morphology. Shiomo Lederman.
12:1.141-163 (Spring 1982)
Problems in the synchronic derivation of the Lithuanian e-formations. Hans
Henrich Hock. 2:2.165-203 (Fall 1972)
Problems of the dual in Soqotri. Samuel E. Fox. 5:2.69-75 (Fall 1975)
Pronouns in Malayalam. K. P. Mohanan. 11:2.67-75 (Fall 1981)
Pseudo-passivization: On the role of pragmatics in determining rule unity. E.
Riddle, G. Sheintuch, & Y. Ziv. 7:1.147-156 (Spring 1977)
Publications of Theodore M. Lightner (1939-1984). 15:2.iv-vii (Fall 1985)
A quantification of Chinese dialect affinity. Chin-Chuan Cheng. 12:1.29-47
(Spring 1982)
Quantifier floating in Thai and the notions cardinallty/ordlnality. Soranee Wong-
biasaj. 9:2.189-200 (Fall 1979)
Quantifier repositioning. Anthony Britti. 9:1.1-16 (Spring 1979)
Raising in Kipsigis. Janice Jake & David Odden. 9:2.131-156 (Fall 1979)
Index to Volumes 1-19 — Title index 209
A reanalysis of the class 5 prefix in Shona. Pamela S. Silver & Scott R. Krause.
8:1.181-196 (Spring 1978)
Reconstruction of a grammaticalized auxiliary in Bantu. Robert Botne. 19:2.
169-186 (Fall 1989)
Reference material in Hindi: State of the art. Narindar K. Aggarwal. 11 :2.209-
219 (Fall 1981)
Reflexives and reciprocals in Dravidian. Karumuri V. Subbarao & Anju Saxena.
17:1.121-135 (Spring 1987)
Reflexivization in Chimwi:ni. Mohammad Imam Abasheikh. 6:2.1-22 (Fall
1976)
Relational Grammar and semantics, ed. by Jerry L. f^organ. 9:2 (Fall 1979)
Relational Grammar, ergativity, and Hindi-Urdu. Rajeshwari Pandharipande &
Yamuna Kachru. 6:1.82-99 (Spring 1976)
Relations between causatives and passives in Indo-lranian. George Cardona.
8:2.1-42 (Fall 1978)
Relative clauses in Standard Arabic revisited. Hussein AN Obeidat. 14:2.77-96
(Fall 1984)
The relative pronoun and the long form adjective in Russian. James Levine.
8:1.127-136 (Spring 1978)
Relativization and pronoun deletion in Hebrew. Shiomo Lederman. 13:1.83-88
(Spring 1983)
The relevance of morphological structure and of stress to clitic placement rule-l
in Pashto. Habibullah Tegey. 7:1.88-116 (Sphng 1977)
Remarks on Dravidian complementation. Sanford B. Steever. 17:1.103,119
(Spring 1987)
Remarks on German nominalization. Susan Meredith Burt. 9:1.17-30 (Spring
1979)
Remarks on the scientific revolution in linguistics 1926-1929. Morris Halle.
15:2.61-77 (Fall 1985)
The representation of diphthongs in Spanish. Maria Carreira. 18:1.1-24
(Sphng 1988)
Restriction as a means of optimizing unification parsing. Dale Gerdemann.
19:1.81-92 (Spring 1989)
Resthctive relative clauses in Bahasa Malaysia. Chiang Kee Yeoh 7:2.82-105
(Fall 1977)
Restrictive relatives with generic heads — are they 'its'? Yael Ziv. 3:1.259-273
(Spring 1973)
Restructuring in the verbal system of Pali. Margie O'Bryan. 3:2.121-133 (Fall
1973)
The hse and fall of a transdehvational constraint: The case of Malay. Lee A.
Becker & Fand Mohamed Onn. 7:2.106-1 14 (Fall 1977)
Rivers Pidgin English: Tone, stress, or pitch-accent language? Nicholas
Faraclas. 14:2.67-76 (Fall 1984)
The role of analogy in non-dehved formations in Zulu. Margie O'Bryan. 4:1.
144-178 (Spring 1974)
210 Studies in the Linguistic Sciences 20:2 (Fall 1990)
Root control, underspecification, and ATR harmony. Meterwa A. Ourso. 18:2.
111-127 (Fall 1988)
Ross' constraints and related problems in Yoruba. Herbert Stahlke. 3:1.193-
230 (Spring 1973)
Rule expansion on the fly: A GPSG parser for Japanese/English using a bit
vector representation of features and rule schemas. Tsuneko Nakazawa &
Laura Neher. 17:2.115-124 (Fall 1987)
Rule ordering versus globality: Evidence from the inversion construction.
Gabriella Hermon. 9:1.139-146 (Spring 1979)
Sanskrit causative syntax: A diachronic study. Hans Henrich Hock. 11:2.9-33
(Fall 1981)
The Sanskrit quotative: A histohcal and comparative study. Hans Henrich
Hock. 12:2.39-85 (Fall 1982)
Schwa syncope and vowel nasalization in Hindi-Urdu: A non-linear approach.
Jean D'souza. 15:1.11-30 (Spring 1985)
A selected bibliography of semantics-based generative grammar. Georgia M.
Green & Rafael Castillo. 2:1.123-140 (Spring 1972)
Semantic features inducing the Germanic dental preterit stem. Irmengard
Rauch. 5:2.125-138 (Fall 1975)
Sentence particles as evidence for morphosyntactic interaction with pragmatics.
Sue Ann Kendall & James Hye-Suk Yoon. 16:1.53-77 (Spring 1986)
Sentence topic in texts. Garlota S. Smith. 15:2.187-203 (Fall 1985)
The serial verb formation in the Dravidian languages, by Sanford B. Steever.
Review article by Hans Henrich Hock. 18:2.211-233 (Fall 1988)
Sex, power, and linguistic strategies in the Hindi language. Tamara Valentine.
15:1.195-211 (Spring 1985)
Sexism in Hindi: Form, function, and variation. Tamara Valentine. 13:2.143-
158 (Fall 1983)
Shanghai tones: A nonlinear analysis. Zhiji Lu. 17:2.93-113 (Fall 1987)
Shingazidja nominal accent. Farida Cassimjee & Charles W. Kisseberth. 19:1.
33-61 (Spring 1989)
The significance of code-mixing to linguistic theory: Evidence from Bantu lan-
guages. Eyamba G. Bokamba & Nkonko M. Kamwangamalu. 17:2.21-43
(Fall 1987)
Socially realistic linguistics: The Firthian tradition. Braj B. Kachru. 10:1.85-111
(Spring 1980)
The sociolinguistic variable (s) in Bengali: A sound change in progress?
Charles A. Ferguson & Afia Dil. 9:1.129-138 (Spring 1979)
Some aspects of code-mixing in Thai. Pairat Wane. 7:1.21-40 (Spring 1977)
Some aspects of pronominalization and relative clause construction in Hindi-
Urdu. Yamuna Kachm. 3:2.87-103 (Fall 1973)
Some aspects of the causative construction in Hindi. Angela B. Kleiman. 1 :2.
104-135 (Fall 1971)
Some aspects of underlying syllable structure in Thai: Evidence from Kham-
phuan — a Thai word game. Aporn Surintramont. 3:1.121-142 (Spring
1973)
Index to Volumes 1 - 19 — Title index 21 1
Some morphological and phonological interactions in Lakhota. Trudi A.
Patterson. 18:1.135-149 (Spring 1988)
Some mystenes of subordination. Alice Davison. 9:1.105-128 (Spring 1979)
Some observations on discourse and sentence grammar. Jerry L. Morgan. 1 1 :
1.137-144 (Sphng 1981)
Some phonological rules of Bani-Hassan Arabic: A Bedouin dialect. Omar
Irshied & Michael Kenstowicz. 14:1.109-147 (Sphng 1984)
Some problems concerning the ohgin of the Latvian broken tone. Aleks Stein-
bergs. 5:2.157-185 (Fall 1975)
Some problems of NP coordination in Korean. Jae Ohk Cho & Jerry Morgan.
16:2.45-65 (Fall 1986)
Some problems of verb agreement. Jerry L. Morgan. 2:1.84-96 (Spring 1972)
Some problems with /-insertion in Pali. Margie O'Bryan. 1:1.29-51 (Spring
1971)
Some queries concerning the syntax of multiple subject constructions in
Korean. James Hye-Suk Yoon. 16:2.215-236 (Fall 1986)
Some remarks on relativization in Imbabura Quechua. Janice Jake. 9:2.109-
130 (Fall 1979)
Some remarks on why there is implicature. Georgia M. Green. 17:2.77-92 (Fall
1987)
Some rules of Koryak phonology. Michael Kenstowicz. 6:1.22-37 (Spring
1976)
Speech chain as an analysis-by-synthesis model: A review. Farid M. Onn.
4:2.161-171 (Fall 1974)
Statistical analysis of conversational Esperanto, with discussion of the
accusative. Bruce Arne Sherwood. 12:1.165-182 (Spnng 1982)
Stress in Cairo Arabic. Ann Welden. 10:2.99-120 (Fall 1980)
A stress mess. Susan F. Schmerling. 1:1.52-66 (Spring 1971)
Studies in Arabic linguistics, ed. by Michael Kenstowicz. 10:2 (Fall 1980)
Studies in Baltic linguistics, ed. by Hans Henrich Hock & Michael J. Kenstowicz.
2:2 (Fall 1972)
Studies in East Asian linguistics, ed. by Chin-chuan Cheng & Chin-W. Kim. 7:2
(Fall 1977)
Studies In functional syntax/Etudes de syntaxe fonctionnelle, by Andre Martinet.
Review by F. M. Jenkins. 7:2.193-195 (Fall 1977)
Studies in language vanation: Nonwestern case studies, ed. by Braj B. Kachru.
13:2 (Fall 1983)
A study of aspirated consonants as spoken and recognized by Hindi speakers.
Tej K. Bhatia. 4:2.25-39 (Fall 1974)
A study of coherence in Korean speakers' argumentative whting in English.
Yeon Hee Choi. 16:2.67-94 (Fall 1986)
A study of duration in speech production. Mahlyn Bereiter. 4:2.14-24 (Fall
1974)
A study of Pashto clitics and implications for linguistic theory. Habibullah
Tegey. 5:1.154-190 (Sphng 1975)
212 Studies in the Linguistic Sciences 20:2 (Fall 1990)
A study on code-switching in Taiwan. Yen Ling Lee. 11:1.121-136 (Spring
1981)
Subject-raising — A unitary rule? Gloria Sheintuch. 5:1.125-153 (Spring 1975)
Subject to object raising in an EST framework: Evidence from Quechua. Peter
Cole & Gabriella Hermon. 9:1.65-90 (Spring 1979)
Substratum influence on (Rig-Vedic) Sanskrit? Hans Henrich Hock. 5:2.76-125
(Fall 1975)
The sun letters in Maltese: Between morpho-phonemics and phonetics.
Bernard Comrie. 10:2.25-37 (Fall 1980)
Suppletive verb phrase deletion. Timothy Habick. 3:1.172-182 (Spring 1973)
Swahili policy implementation in Tanzania: The role of the National Swahill
Council (BAKITA). Andrea S. Dunn. 15:1.31-47 (Sphng 1985)
The syllable in phonological theory: Arguments from Tamil. K. G. Vijaykrishnan.
11:2.101-105 (Fall 1981)
Syllable patterns in Levantine Arabic. Issam M. Abu-Salim & Hassan R. Abd-el-
Jawad. 18:2.1-22 (Fall 1988)
Syllable structure in Palestinian Arabic. Issam M. Abu-Salim. 12:1.1-28
(Spring 1982)
Syllable structure and suffixation in Wolof. Omar Ka. 15:1.61-90 (Spring 1985)
Syllable structure in two Arabic dialects. Ellen Broselow. 10:2.13-24 (Fall
1980)
Synonym split in the dialect of Bosnia and Hercegovina: A study of a change in
progress. Vesna Radanovic-Kocic. 16:1.123-131 (Spring 1986)
Syntactic analogy and backward pronominalization. Peter Cole. 3:1.33-44
(Spring 1973)
Syntactic variation and language change: Eastern and Western Hindi.
Yamuna Kachru. 12:2.87-96 (Fall 1982)
The syntax and psycholinguistics of bilingual code mixing. S. N. Sridhar &
Kamal K. Sridhar. 10:1.203-215 (Spring 1980)
Tag questions in Korean: Form and function. Suk-Jin Chang. 16:2.15-26 (Fall
1986)
Teaching Chinese numeration on computer. Chin-Chuan Cheng. 7:2.165-177
(Fall 1977)
The three scales of diachrony. William S-Y. Wang. 8:2.63-76 (Fall 1978)
Time travel or the futuristic use of to go. Richard Neil Halpern. 5:1.36-41
(Spring 1975)
Tonal changes: Interplay between tone and tone sandhi: A case study of the
Shanghai dialect. Zhiji Lu. 16:1.97-111 (Spring 1986)
Tonal correlations in Chinese dialects: A quantitative study. Chin-Chuan
Cheng. 7:2.115-128 (Fall 1977)
Tonal polarity in two Gur languages. Michael Kenstowicz, Emmanuel Niklema,
& Meterwa Ourso. 18:1.77-103 (Spring 1988)
Tone in Jita questions. Laura J. Downing. 1 9:2.91 -1 1 3 (Fall 1 989)
Tone-bearing nasals in Makua. Chin-Chuan Cheng & Charies W. Kisseberth.
12:1.123-139 (Spring 1982)
Index to Volumes 1 - 19 — Title index 213
Tonology of noun-modifier phrases in Jita. Laura J. Downing. 18:1.25-60
(Spring 1988)
Topic marker ellipsis in Japanese. Michio Tsutsui. 11 :1. 163-179 (Spring 1981)
Topics in Relational Grammar, ed. by Jerry Morgan, Georgia Green, & Peter
Cole. Special section of 6:1 (Fall 1976)
Toward an integrated theory of morphophonology: Vowel harmony in Korean.
Hyang-Sook Sohn. 16:2.157-184 (Fall 1986)
Toward structuring the form and function of code-mixing: An Indian perspective.
Braj B. Kachru. 5:1.74-92 (Spring 1975)
Toward a typology of compound verbs in South Asian languages. Yamuna
Kachru & Rajeshwari Pandharipande. 10:1.113-124 (Spring 1980)
Transitivity and volitionality in Hindi-Urdu. Yamuna Kachru. 11:2.181-193 (Fall
1981)
Transitivity in Hindi. Rajeshwari Pandharipande. 11:2.161-179 (Fall 1981)
Transplanted South Asian languages: An overview. Tej K. Bhatia. 11:2.129-
139 (Fall 1981)
The treatment of transitivity in the Hindi grammatical tradition. Tej K. Bhatia.
11:2.195-208 (Fall 1981)
Trends in Oromo lexicon and lexicography. Mohammed Ali. 19:2.155-168 (Fall
1989)
Trinidad Hindi: Three generations of a transplanted variety. Tej K. Bhatia.
11:2.135-150 (Fall 1981)
Tuki gaps: Null resumptive pronouns or variables? Edmond Biloa. 19:2.43-54
(Fall 1989)
Two notes on negation in Japanese. Paula Chen Rohrbach. 13:1.157-174
(Spring 1983)
Two social varieties of Farsi: 'Jaheli' and 'Armenian Persian'. Jan Zamir. 13:2.
159-187 (Fall 1983)
Umlaut in Korean. Seok-Ran Shim. 16:2.147-155 (Fall 1986)
UNICORN: A unification parser for attnbute-value grammars. Dale Gerdemann
& Erhard W. Hinrichs. 18:2.41-86 (Fall 1988)
Unmarked bleeding orders. Michael J. Kenstowicz & Charles W. Kisseberth.
1:1.8-28 (Spring 1971)
The use of the high rising changed tone in Cantonese: A sociolinguistic study.
Maurice K. S. Wong. 7:2.65-81 (Fall 1977)
Variable rules in the language community: A study of lax [u] in English. Wayne
B. Dickerson. 5:2 41-68 (Fall 1975)
Variation m case marking with infinitival and clausal complements in Old
French. Elizabeth Pearce. 14:1 .149-166 (Spnng 1984)
Vahation in Esperanto. Bruce Arne Sherwood. 12:1.183-196 (Spring 1982)
Variation in Hindi: Problems and prospects. Tej K. Bhatia. 13:2.1-19 (Fall
1983)
Vowel length in Korean. Chin-W.Kim. 7:2.184-190 (Fall 1977)
Vowel sandhi and syllable structure in Kabyle Berber. Yousef Bader. 13:1.1-17
(Spring 1983)
214 Studies in the Linguistic Sciences 20:2 (Fall 1990)
'We makin" some cookies': A child language case study of the effects of
situational variation on pragmatic function and syntax. Charlotte Blomeyer
& Tamara Valentine. 13:1.19-42 (Spring 1983)
lVA)-movement in Hindi-Urdu relative clauses. Alice Davison. 17:1.25-33,
(Spring 1987)
Whatever happened to communicative competence? Susan Ervin-Tripp.
8:2.237-258 (Fall 1978)
What is a Bantu noun class? Rick Treece. 16:1.149-166 (Spring 1986)
What they say about say. Elizabeth Riddle. 5:1.113-112 (Spring 1975)
Where do Ideophones come from? G. Tucker Childs. 19:2.55-76 (Fall 1989)
Where's my NP? Non-transformational analyses of Vedic pronominal fronting.
Steven Schaufele. 18:2.129-162 (Fall 1988)
Whistled Languages, by Rene-Guy Busnel & Andre Classe. Review by C-W.
Kim. 7:2.196-199 (Fall 1977)
Why delete tense? David E. Johnson. 3:1.54-62 (Spring 1973)
Why Dyirbal isn't ergative at all. Janice Jake. 8:1.97-110 (Spring 1978)
Word order change and the Senufo languages. Anne Garber. 10:1.45-57
(Spring 1980)
World Englishes and applied linguistics. Braj B. Kachru. 19:1.127-152 (Spring
1989)
Yes, Virginia, syntactic reconstruction is possible. Hans Henrich Hock. 15:1.49-
67 (Spring 1985)
Yoruba gerundive structures and the notion of 'target structures'. Yiwola
Awoyale. 4:1.1-31 (Spring 1974)
All checks and money orders must be in U.S. Dollars, drawn on a U.S. Bank,
and made payable to University of Illinois . Sales Tax must be included as follows (ex-
cept tax exempt organizations: IL 7V4%, IN 5%, MI 4%, MN 6%, OH 5%, WI 5%.
While quantities last, Volumes 7:2 through 16:2 are behig ofTered at half-price.
Vol. 7:2
Fall 1977
Vol. 8:1
Spring 1978
Vol. 8:2
Fall 1978
Vol. 9:1
Spring 1979
Vol. 9:2
Fall 1979
Vol. 10.1
Spring 1980
Vol. 10:2
Fall 1980
Vol. 11:1
Spring 1981
Vol. 11:2
Fall 1981
Vol. 12:1
Spring 1982
Vol. 12:2
Fall 1982
Vol. 13:1
Spring 1983
Vol. 13:2
Fall 1983
Vol. 14:1
Spring 1984
Vol. 14:2
Fall 1984
Vo.. 15:1
Spring 1985
Vol. 15:2
Fall 1985
Vol. 16:1
Spring 1986
Papers in General Linguistics
Papers in General Linguistics
Papers in General Linguistics
Papers in General Linguistics
Relational Grammar and Semantics
(Editor: Jerry L. Morgan)
Papers in General Linguistics
Studies in Arabic Linguistics
(Editor: Michael J. Kenstowicz)
Papers in General Linguistics
Dimensions of South Asian Linguistics
(Editor: Yamuna Kachru)
Papers in General Linguistics
Papers on Diachronic Syntax:
Six Case Studies
(Editor: Hans Henrich Hock)
Papers in General Linguistics
Studies in Language Variation:
Nonwestem Case Studies
(Editor: Braj B. Kachru)
Papers in General Linguistics
Language in African Culture and Society
(Editor: Eyamba G. Bokamba)
Papers in General Linguistics
Linguistic Studies in Memory of
Theodore M. Lighiner
(Editor: Michael J. Kenstowicz)
Papers in General Linguistics
$2.50 $5.00
$2.50 ^00
$2.50 $5^
$2.50 $^tO0
$2.50 $&m
$2.50 $&m
$2.50 $sm
$2.50 um
$2.50 $5.00
$2.50 $5t00
$2.50 $5.00
$2.50 $-5.00
$2.50 $5.00
$2.50 $5.00
$2.50 $5.00
$2.50 $5.00
$3.00 $6.00
STUDIES IN THE LINGUISTIC SCIENCES
The following issues are available:
Vol. 16:2 Fall 1986 Illinois Studies in Korean Linguistics
(Editor: Chin-W. Kim)
Vol. 17: 1 Spring 1987 Papers from the 1986 South Asian
Languages Analysis Roundtable
(Editor: Hans Henrich Hock)
Vol. 17:2 Fall 1987 Papers in General Linguistics
Vol. 18:1 Spring 1988 Papers in General Linguistics
Vol. 18:2 Fall 1988 Papers in General Linguistics
Vol. 19:1 Spring 1989 Papers in General Linguistics
Vol. 19:2 Fall 1989 The Contribution of African Linguistics
to Linguistic Theory, Vol. 1
(Editor: Eyamba G. Bokamba)
Vol. 20:2 Fall 1990 Linguistics for the Nineties:
Papers from a Lecture Series in Celebration
of the Department's Twenty-Fifth Anniversary
(Editor: Hans Henrich Hock)
Vol. 20:3 Spring 1991 Meeting Handbook: Thirteenth South Asian
Languages Analysis Roundtable, 25 - 27 May
1991, University of Illinois at Urbana-Champaign
IN PREPARATION:
Vol. 20:1 Spring 1990 The Contribution of African Linguistics
to Linguistic Theory, Vol. 2
(Editor: Eyamba G. Bokamba)
FOR EARLIER ISSUES AND ORDER INFORMATION SEE INSIDE COVER
Orders should be sent to:
SLS Subscriptions, Department of Linguistics
University of Illinois
4088 Foreign Languages Building
707 S. Mathews
Urbana, Illinois 61801
Studies in
rhe I inoiiistir Sciences
20, No. 3
ring 1991
Aug 'L ^ i^'^i
UtiilOiS
410 STX
St92
20:3 SPR 1991 COPY 2
MEETING HANDBOOK
THIRTEENTH SOUTH ASIAN LANGUAGES ANALYSIS
ROUNDTABLE
25 -27 May 1991
University of Illinois at Urbana-Champaign
Preface vii
Program Overview 1
Program 3
General Abstracts 15
Panel Abstracts 121
Conventions of Politeness in South Asian Languages
(organized by Manindra K. Verma) 123
Discourse in the Other Tongue
(organized by Yamuna Kachru) 129
Language Identity and Conflict in South Asia
(organized by Rakesh M. Bhatt) 137
Language of Religion
(organized by Rajeshwari Pandharipande) 145
Literary Responses to India
(organized by Girdhari Tikku) 151
Index of Participants 157
Department of Linguistics
University of Illinois
STUDIES IN THE LINGUISTIC SCIENCES
PUBLICATION OF THE DEPARTMENT OF LINGUISTICS
COLLEGE OF LIBERAL ARTS AND SCIENCES
UNIVERSITY OF ILLINOIS AT URBANA-CHAMPAIBN
EDITOR: Hans Henrich Hock
EDITORIAL BOARD: Eyamba G. Bokamba, Chin-chuan Cheng, Georgia M.
Green, Erhard W. Hinrichs, Hans Henrich Hock, Braj B. Kachru, Yamuna
Kachru, Chin-W. Kim, Charles W. Kisseberth, Howard Maclay, Jerry L.
Morgan, Rajeshwari Pandharipande, and Ladislav Zgusta.
AIM: SLS is intended as a forum for the presentation of the latest original
research by the faculty and especially students of the Department of Linguistics,
University of Illinois, Urbana-Champaign. Invited papers by scholars not as-
sociated with the University of Illinois will also be included.
SPECIAL ISSUES: Since its inception SLS has devoted one issue each year to
restricted, specialized topics. A complete list of such special issues is given on the
back cover.
BOOKS FOR REVIEW: Review copies of books may be sent to the Editor,
Studies in the Linguistics Sciences, Department of Linguistics, University of
Illinois, 707 S. Mathews, Urbana, Dlinois 60801.
SUBSCRIPTION: There will be two issues during the academic year. Requests
for subscriptions should be addressed to SLS Subscriptions, Department of
Linguistics, 4088 Foreign Languages Building, 707 S. Mathews, University of
Illinois, Urbana, Illinois 60801.
UPCOMING ISSUES: Vol. 21:1: Papers in General Linguistics.
Price: $7.50 (per issue)
SfiLfiXIII
25-27 May 1991
Uniuersity of Illinois
at Urbana-Champaign
THIRTEENTH SOUTH ASIAN
LANGUAGES ANALYSIS
ROUNDTABLE
25 - 27 May 1991
Sponsored by
The Department of Linguistics
University of Illinois at Urbana-Champaign
in cooperation with
The Programs in Comparative Literature,
Religious Studies,
and South and West Asian Studies,
Intemational Programs and Studies,
and the Division of English as an International Language
Meeting Handbook
Editor
Hans Henrich Hock
EDITORIAL ASSISTANT
Lynne Murphy
(Special issue of Studies in the Linguistic Sciences,
Volume 20, Number 3, Spring 1991)
The Thirteenth South Asian Languages Analysis
Roundtable
is part of the
Twenty-Fifth Anniversary Celebrations
of the Department of Linguistics,
University of Illinois at Urbana-Champaign
and honors
Professors Bh. Krishnamurti and Ladislav Zgusta
for their outstanding contributions to South Asian
linguistics
Organizing Committees:
Local: Jennifer Cole National: Tej K. Bhatia (Syracuse)
Hans Henrich Hcx;k, Chair Bruce Pray (UC Berkeley)
Braj B. Kachru S. N. Sridhar (SUNY, Stony Brook)
Yamuna Kachru Sally J. Sutherland (UC Berkeley)
Rajeshwari Pandharipande plus the Local Committee
Girdhari Tikku
Local Support Committee:
Rakesh M. Bhatt, Patrick Marlow, Mithilesh K. Mishra, Mariam Ahmed, Sarah Tsiang
Organizing Assistants: Lieve van de Walle (Fall 1990), Lynne Murphy (Spring 1991)
CONTENTS
Preface vii
Program Overview 1
Program 3
General Abstracts 1 5
Panel Abstracts 1 2 1
Conventions of Politeness in South Asian Languages
(organized by Manindra K. Verma) 123
Discourse in the Other Tongue
(organized by Yamuna Kachru) 129
Language Identity and Conflict in South Asia
(organized by Rakesh M. Bhatt) 137
Language of Religion
(organized by Rajeshwari Pandharipande) 145
Literary Responses to India
(organized by Girdhari Tikku) 151
Index of Participants 157
PREFACE
The present volume contains the program and abstracts for the
Thirteenth South Asian Languages Analysis (SALA) Roundtable, 25 -
27 May 1991, University of Illinois at Urbana-Champaign. Just as
the SALA Roundtable, it celebrates the twenty-fifth anniversary of
the Department of Linguistics and honors Professors Bh. Krishna-
murti and Ladislav Zgusta for their outstanding contributions to
South Asian linguistics. In addition, jointly with the other parts of
Volume 20 of Studies in the Linguistic Sciences, it celebrates the
twentieth anniversary of the Department's journal.
1965/66 to 1990/91:
Twenty-five years of Linguistics at Illinois
Just as at many other American universities. Linguistics at
Illinois had a 'prehistory', preceding its official institutionalization as
a Department. As early as 1910-1921, Leonard Bloomfield taught at
the University of Illinois. Although a member of the German Depart-
ment, Bloomfield offered general linguistics courses (Introduction to
the Study of Language, and Comparative Philology of the Indo-Euro-
pean Languages), as well as a topic of specific interest to South
Asianists, Sanskrit. (Although justly famous for later becoming the
founder of Structuralism, Bloomfield had an excellent background in
Indo-European and Sanskrit studies, was thoroughly familiar with
the Sanskrit grammarians, and applied that familiarity to the des-
cription of Indigenous American languages.)
By the forties, the University had acquired a number of faculty
members, coming from Psychology, Speech, and the language de-
partments, who had an active interest in linguistics. Among these
were Lee Hultzen, Henry Kahane, and Charles Osgood. These were
joined in the early sixties by John Casagrande and Kenneth L. Hale in
Anthropology, Robert B. Lees in English and Communications, and
Mary Sleator Temperly in English.
By 1961, the number of faculty members interested in linguis-
tics had reached a 'critical mass', and a Program in Linguistics was
established, directed by Henry Kahane. By 1963, the Program em-
braced eleven core faculty members plus six associate members. A
Linguistics Committee was established under the leadership of
Robert B. Lees, with the task of preparing for the establishment of a
viii Studies in the Linguistic Sciences 20:3 (Spring 1991)
Department of Linguistics. The task of this Department was to be
instruction and research not only in linguistics, but also in non-
western languages, a fact which turned out to be highly significant i
for the development of South Asian linguistics at the University of
Illinois.
South Asian linguistics received more specific — and special —
attention in 1963-64, when Robert B. Lees offered courses on Guja-
rati, and especially in 1964 when Braj B. Kachru, joining the Uni-
versity as the first full-time faculty member in Linguistics, initiated
a program in Hindi and Kashmiri, assisted by Tara Singhwi.
When established in 1965, the Department of Linguistics made
several additional faculty appointments with South Asian interest:
Gargi Guha-Thakurta, who taught Sanskrit and Hindi, and Arnold
Zwicky, who took a research interest inSanskrit. Moreover, Yamuna
Kachru was associated with the Department, becoming a regular fac-
ulty member in 1967. After Guha-Thakurta's departure, the Sanskrit
program received further attention when Hans Henrich Hock was
hired in 1967. In 1968, the arrival of Giri Tikku at the University of
Illinois added another component, the study of South Asian literature
from a comparative perspective.
Beside the Kachrus, Hock, and Tikku, who have remained at the
University of Illinois to the present, other faculty members tem-
porarily helped in the instruction and development of the South
Asian language, linguistics, and literature programs. These include
Karine Schomer for Hindi literature, and Stefan Anacker and Daniel
Bisgaard for Sanskrit. In addition, Alice Davison was associated with
our Department for several years during the seventies and eighties.
The Hindi program further benefited from some excellent teach-
ing assistants, including Ahmad H. Siddiqui and especially K. V. Sub-
barao and Tej K. Bhatia, both of whom received Ph.D.s in Linguistics
from the University of Illinois and have since made names for them-
selves in South Asian linguistics. The Department is pleased and
honored that both of them are planning to participate in this year's
SALA.
Another former assistant in Hindi-Urdu, Rajeshwari Pandhari-
pande, an excellent teacher just like her colleagues Subbarao and
Bhatia and like them receiving a Ph.D. in Linguistics from the Univer-
sity of Illinois, became a member of the Illinois faculty in 1984.
Although her major appointment is in the Religious Studies Program,
where she teaches enormously popular courses on Asian Mythology
and Hinduism, she has maintained a foothold in Hindi instruction and
Preface
is currently offering a linguistics course on the Language of Religion.
Her major research area continues to be South Asian linguistics.
Moreover, having received an earlier Ph.D. in Sanskrit, she is able to
provide support in supervising dissertations in Sanskrit, together
with Ladislav Zgusta, who joined us in 1970.
The most recent addition to our faculty with South Asian
interest is Jennifer Cole, a phonologist who has done work on Sindhi,
and who is currently at the University of Illionis on a visiting ap-
pointment.
The founding of the Department coincided with the initiation of
two significant research projects, both of them funded by the US
Office of Education. One of these, a joint project by Yamuna Kachru
and Robert B. Lees, was concerned with Hindi syntax and resulted in
Y. Kachru 1966. (For this and the following publications, see the list
below.) The other, undertaken by Braj B. Kachru, dealt with Kashmiri
and resulted in B. B. Kachru 1969. A further (1970) grant from the
same agency funded work toward B. B. Kachru 1973. These grants,
and the publications produced under them, set the pattern for a long
tradition of research and publication by the faculty. An incomplete
list of the major publications dealing with South Asia is presented
below. (See also the list of special issues of Studies in the Linguistic
Sciences p. xii below.)
Hans Henrich Hock, ed. 1991. Studies in Sanskrit syntax. Delhi:
Motilal Banarsidass.
Braj B. Kachru. 1969. A reference grammar of Kashmiri. Urbana:
University of Illinois.
. 1973. An introduction to spoken Kashmiri, two parts.
Urbana: University of Illinois.
. 1981. Kashmiri literature. Wiesbaden: Harrassowitz.
. 1983. The Indianization of English: The English language in
India. Delhi: Oxford University Press.
, & Yamuna Kachru. 1968. Studies in Hindi linguistics. New
Delhi: American Institute of Indian Studies.
Yamuna Kachru. 1966. An introduction to Hindi syntax. Urbana:
University of Illinois.
. 1968. Studies in a transformational grammar of Hindi.
Dhanbad: East West Books.
. 1974. Topics in a transformational grammar of Hindi.
Agra: Central Hindi Institute.
. 1980. Aspects of Hindi grammar. New Delhi: Manohar.
, & D. P. Pattanayak. 1982. Relational Grammar: An intro-
duction. Mysore: Central Institute of Indian Languages.
Studies in the Linguistic Sciences 20:3 (Spring 1991)
, & Rajeshwari Pandharipande. 1983. Intermediate Hindi.
Delhi: Motilal Banarsidass. (Third edition 1990.)
Rajeshwari Pandharipande. 1990. The eternal self and the cycle i
of sarhsara: Introduction to Asian mythology and religion.
Needham Heights: Ginn Press.
During the Department's twenty-five years of existence, fifteen
Ph.D. dissertations dealing with South Asian linguistic issues have
been produced. These are (in chronological order):
Robert P. Fox. 1968. A transformational treatment of Indian
English syntax.
Margie O'Bryan. 1973. The history and development of the
verbal system of early Middle Indie.
K. V. Subbarao. 1974. NP complementation in Hindi.
Geoffrey Hackman. 1976. An integrated analysis of the Hindi
tense and aspect system.
Habibullah Tegey. 1977. The grammar of clitics: Evidence from
Pashto (Afghani) and other languages.
Tej K. Bhatia. 1978. A syntactic and semantic description of
negation in South Asian languages.
S. N. Sridhar. 1980. Cognitive determinants of linguistic struc-
tures: A cross-linguistc experimental study of sentence pro-
duction.
Rajeshwari V. Pandharipande. 1981. Syntax and semantics of the
passive construction in selected South Asian languages.
Cecil L. Nelson. 1984. Intelligibility: The case of non-native vari-
eties of English.
William D. Wallace. 1985. Subjects and subjecthood in Nepali: An
analysis of Nepali clause structure and its challenges to
Relational Grammar and Government and Binding.
Richard D. Lutz. 1985. The effect of pronoun type on first and
second language perceptual strategies in Hindi.
Tamara M. Valentine. 1986. Linguistic interaction and women:
A South Asian case study.
Jean D'Souza. 1987. South Asia as a sociolinguistic area.
Baber S. A. Khan. 1989. The Urdu case system.
Steven W. Schaufele. 1991. Free word-order syntax: The chall-
enge from Vedic Sanskrit to contemporary formal syntactic ,
theory. '
In addition to providing instruction and research in — and about
— South Asian languages, the Department of Linguistics has hosted
several institutes concerned totally or in part with South Asian
languages and linguistics. The first among these and the one which
established a model for everything that followed was the Summer
Preface
1967 CIC South Asian Languages Institute (Academic Director: Braj B.
Kachru; Administrative Director: Solomon Levine). Visiting faculty
included Asiiok Aklujkar, Ronald Asher, George Cardona, Gordon
Fairbanks, J. F. Staal, and Girdhari Tikku. Among a number of special
events, the Institute featured a South Asian Language Conference.
South Asian languages and linguistics received further attention
at the 1968, 1969, and 1978 Linguistic Institutes of the Linguistic
Society of America, hosted by the Department of Linguistics.
South Asian Languages Analysis Roundtables
The 1978 Linguistic Institute, directed by Braj B. Kachru, fea-
tured a Conference on South Asian Languages and Linguistics. The
conference, organized by Hans Henrich Hock and Braj B. Kachru, was
to become the 'Mother of all SALAs': It was so successful in bringing
together South Asian linguists and language pedagogues that the De-
partment followed up this Conference with the first South Asian Lan-
guages Analysis Roundtable in 1979, as a forum for South Asian lin-
guists, language teachers, and scholars of literature. The success of
this undertaking can be measured by the fact that since then, SALA
Roundtables have been taking place every year. The 'varhsV is as
follows:
SALA I: University of Illinois at Urbana-Champaign (1979)
SALA II: University of Texas at Austin (1980)
SALA III: State University of New York at Stony Brook (1981)
SALA IV: University of Syracuse (1982)
SALA V: University of Illinois at Urbana-Champaign (1983)
SALA VI: University of Illinois at Urbana-Champaign (1984)
SALA VII: University of Michigan (1985)
SALA VIII: University of Illinois at Urbana-Champaign (1986)
SALA IX: Cornell University and University of Syracuse (1987)
SALA X: University of Washington at Seattle (1988)
SALA XI: University of Wisconsin at Madison (1989)
SALA XII: University of California at Berkeley (1990)
The present Rountable, then, is the thirteenth. But those who con-
sider the number thirteen unlucky may look at this year's SALA as
the fourteenth North American conference on South Asian languages,
linguistics, and literatures, counting the 1978 'Mother of all SALAs'
as number one.
Members of the Department of Linguistics at Stanford University
have expressed their interest and willingness to organize next year's
SALA.
Studies in the Linguistic Sciences 20:3 (Spring 1991)
A second offspring of the 1978 Conference was an intended
series of International Conferences on South Asian Languages and
Linguistics, to be held in India. Two very successful conferences of
this type were organized, one in Hyderabad (1980) by Bh. Krishna-
murti, the other in Mysore (1982) by D. P. Pattanayak. Unfortunate-
ly, later installments of this tradition did not materialize. SALA,
originally intended as a North American forum, therefore has in-
creasingly drawn participants from other parts of the world. This
year is no exception, although the number of Indian participants has
been severely reduced by the devastating effects of the recent Gulf
Crisis on India's economy.
Studies in the Linguistic Sciences
This year also marks the twentieth anniversary of Studies in the
Linguistic Sciences, the Department's journal established in 1971 at
the instigation Braj B. Kachru who had just taken over as Head of
Linguistics. Several special issues of the journal have been dedicated
to, or dominated by. South Asian linguistics:
Papers on Hindi syntax, ed. by Yamuna Kachru. (SLS 1:2, 1971)
Papers on South Asian linguistics, ed. by Braj B. Kachru. (SLS
3:2, 1973)
Dimensions of South Asian linguistics, ed. by Yamuna Kachru.
(SLS 11:2, 1981)
Papers on diachronic syntax: Six case studies, ed. by Hans Hen-
rich Hock. (SLS 12:2, 1982)
Studies in language variation: Nonwestern case studies, ed. by
Braj B. Kachru. (SLS 13:2, 1983)
Papers from the 1986 South Asian Languages Analysis Round-
table, ed. by Hans Henrich Hock. (SLS 17:1, 1987)
Volume 20 of Studies in the Linguistic Sciences celebrates itself
by coming out in three numbers. Number 1, edited by Eyamba G.
Bokamba, constitutes the second of two proceedings volumes from
the Twentieth Annual Conference on African Linguistics. (The tradi-
tion of African linguistics conferences, just like SALA, was started by
our Department, during the time of Braj B. Kachru's Headship and
with his instigation and strong support. And like SALA, the Confer-
ence has been hosted in turn by a large number of universities.)
Number 2, edited by Hans Henrich Hock, is entitled Linguistics for
the Nineties and contains papers presented in a special lecture series,
fall 1990, by former students of the Department who have made a
name for themselves in the field. In addition, the volume contains a
Preface
cumulative index of SLS. The present SALA Meeting Handbook
constitutes the third part of this special celebratory volume.
Bh. Krishnamurti and Ladislav Zgusta
Last, but certainly not least, this year's SALA honors two schol-
ars for their outstanding contributions to South Asian languages and
linguistics. The following presents only a few highlights of their
careers. More will be said during the course of the Roundtable.
Bh. Krishnamurti is without any doubt the major, most pro-
ductive linguist in India. His areas of specialization cover compar-
ative and historical linguistics with special reference to Dravidian,
lexicography, language typology, sociolinguistics, literacy, and langu-
age planning. His publications include Telugu verbal bases: A com-
parative and descriptive study (1961), Studies in Indian linguistics, a
volume edited in honor of M. B. Emeneau's sixtieth birthday (1968),
A basic course in modern Telugu (1968), A grammar of modern
Telugu (1985), and South Asian languages: Structure, convergence,
and diglossia, a co-edited proceedings volume from the International
Conference on South Asian Languages and Linguistics held 1980 at
Hyderabad (1985). In spite of being Vice-Chancellor of the Univer-
sity of Hyderabad, he is keeping up a steady stream of publications.
One of the most recent among these, 'The emergence of the syllable
types of stems (C)VCC(V) and (C)VC(V) in Indo-Aryan and Dravidian:
Conspiracy or convergence?' is to appear in a volume Studies in the
phonologies of Asian languages, edited by Michael C. Shapiro et al.
Ladislav Zgusta holds two doctor's degrees from Prague Uni-
versity, one in classical philology and Indology, the other in philology
of Asia Minor. He has been at the University of Illinois since 1970,
permanent member of the University's Center for Advanced Study
since 1974, and Director of the same center since 1988. Much of his
work has centered around Asia Minor, Iranian languages, lexi-
cography, and onomastics. But he has retained an active interest in
Indology as well, especially in the indigenous Indian tradition of
grammar and linguistic philosphy. (See e.g. his 1969 article on 'Pa-
nini: descriptivist or transformationalist?') Under a 1975 Ford Foun-
dation grant, he offered a seminar for Indian scholars in his major
area of specialization, lexicography. For the past several years he has
undertaken lexicographical research on Sanskrit, examining both
Sanskrit dictionaries produced by westerners (or in the western
tradition) and the traditional thesauruses (kos'as) of the indigenous
Indian tradition.
Program Overview
Saturday, 25 May
Sunday, 26 May
Monday, 27 May
8:30-9:15 flUni Union
Registration
9:15- 10:00 mini Union
Inaugural Session
10:00- 12:00 Illini Union
Session I
Syntax & Semantics I
Language of Religion
9: 15 -12:00 mini Union
Session III
Phonology
Sanskrit Syntax
Literary Responses to
India
9:15-12:00 Dlini Union
Session V
Syntax & Semantics II
General Session II
Conventions of
Politeness
12:00-2:00
Lunch Break
12:00-2:00
Lunch Break
12:00- 1:45
Lunch Break
2:00-4:45 Illini Union
Session II
Dravidian Linguistics I
Historical Perspectives
Discourse in the Other
Tongue
2:00-4:45 mini Union
Session IV
Dravidian Linguistics n
Language Identity &
Conffict
General Session I
1:45-5:00 Illini Union
Session VI
Hindi-Urdu Syntax/
Semantics
Sociolinguistics
Sanskrit Indigenous
Grammar
5:00-8:00
Dinner Break
5:00-8:00
Dinner Break
8:00 - 9:30 Levis Center
Plenary Session
Bh. Krishnamurti
Shift of authority in
written and oral texts in
India with special
reference to Telugu
8:00 - 9:30 Levis Center
Plenary Session
Ladislav Zgusta
Dvaikosyam: The
controversy over the
Sanskrit-English
dictionary ofMonier-
Williams
Program
Saturday, 25 May 1991
REGISTRATION (8:30 - 9: 15 am, 4th floor Illini Union)
INAUGURAL SESSION (9:15 - 10:00 am, 405 Illini Union)
Welcoming remarks:
Professor Larry R. Faulkner, Dean, College of Liberal Arts and Sciences
Professor Elmer H. Antonsen, Acting Head, Department of Linguistics
VandanS
Ceremony to honor Professors Bh. Krishnamurti and Ladislav Zgusta
SESSION I (10:00 am - 12:00 noon)
SYntax/Sg^^antitS I 405 Illini Union
Chair: James Yoon (University of Illinois at Urbana-Champaign)
10:00- 10:30: Finiteness and case in Hindi-Urdu complements (Alice Davison,
University of Iowa)
10:30-11:00: Kashmiri causals: Evidence for a transformational cqyproach (Peter E.
Hook, University of Michigan & Omkar Nath Koul, National
Academy of Administration)
1 1 :00 - 1 1 :30: The INFL nodes in non-finite clauses in Dravidian and Tibeto-Burman
languages (K. V. Subbarao & Lalitha M., Delhi University)
1 1 :30 - 12:00: New dimensions of word-order freedom in verb-final languages (Asha
Tickoo, University of Pennsylvania)
Lang ua ge of Relig ion 404 Illini Union
Organizer and Chair: Rajeshwari Pandharipande
(University of Illinois atUrbana-Champaign)
10:(X) - 10:30: The question of defining the language of religion (Rajeshwari
Pandharipande, University of Illinois at Urbana-Champaign)
10:30 - 11 :00: Vasat, s'rausat, and other Vedic ritual particles: Their origin and use in
Vedic ritualistic literature (Hans Henrich Hock, Univerisity of
Illinois at Urbana-Champaign)
1 1:00 - 1 1 :30: The role ofdeixis in defining ordinary vs. religious language (Mithilesh
K. Mishra, University of Illinois at Urbana-Champaign)
11:30-12:00: DISCUSSION
[4]
SESSION II (2:00 - 4:45 pm)
Dravidian Linguistics I 405 Illini Union
Chair: Bh. Krishnamurti (University of Hyderabad)
2:00 - 2:30: Case particles and postpositions in Kannada (S. Chandrashekar & S. N.
Sridhar, SUNY at Stony Brook)
2:30 - 3:00: Tamil verb stem formation (A. G. Menon, University of Leiden)
3:00-3:15: COFFEE BREAK
3:15 - 3:45: Pronouns in Kannada: Sociolinguistic implications (Jayashree Nadahalli,
New York University)
3:45-4:15: Malayalam syllabification (SuchitraSadanandan, University of Southern
California)
4: 15 - 4:45: On change and variation of[ I } in Kannada (T. S. Satyanath, University
of Delhi)
Historical Perspectives 404 Illini Union
Chair: Elmer H. Antonsen (University of Illinois at Urbana-Champaign)
Some observations on the development of West Indo- Aryan pronominal
systems from Apabhramsa (Vit M. Bubenik & C. Paranjape,
Memorial University of Newfoundland)
The sources of a passive (Eric P. Hamp, University of Chicago)
COFFEE BREAK
Sibilant confusion in Early Indie: Sanskrit pradiir (Brian Joseph, Ohio
State University)
Meet me in the Bazaar: A historical perspective on the origin of a North
Indian Koine (Patrick E. Marlow, University of Illinois at Urbana-
Champaign)
Convergence and syntactic change: The case of the negative participles in
Dakkhini (K. V. Subbarao & Harbir Arora, Delhi University)
2:00
2:30
2:30
3:00
3:00
3:15
3:15
3:45
3:45-4:15:
4:15-4:45:
Discourse in the Other Tongue 402 Illini Union
Organizer and Chair: Yamuna Kachru (University of Illinois at Urbana-
Champaign)
2:00-2:30: Recreating South Asian speech acts in English: A study in linguistic
transfer (Jean D'Souza, National University of Singapore)
2:30 - 3:00: Speech acts in the mother tongue and the other tongue (Yamuna Kachru,
University of Ilhnois at Urbana-Champaign)
3:00-3:15: COFFEE BREAK
3:15 - 3:45: On creating speech acts: The creativity of Indian English writers (Cecil
Nelson, Indiana State University)
3:45 - 4: 15: Language and female identity in India (Tamara Valentine, University of
South Carolina at Spartanburg)
4:15-4:45: DISCUSSION
[5]
KEYNOTE LECTURE
(8:00 - 9:30pm, Levis Faculty Center, third floor)
Bh. Krishnamurti
(Vice Chancellor, University of Hyderabad)
Shift of authority in written and oral texts in India,
with special reference to Telugu
(Chair: Hans Henrich Hock, University of LUnois at Urbana-Champaign)
(The talk is followed by a reception in the Levis Faculty Center)
[7]
Sunday, 26 May 1991
SESSION III (9:15 am - 12:00 noon)
Phonology 405 Illini Union
Chair: S. N. Sridhar (State University of New York at Stony Brook)
9:15- 9:45: An essay on Kashmiri stress (Rakesh M. Bhatt, University of Illinois at
Urbana- Champaign)
9:45- 10:15: Alliteration in Sindhi poetry: Evidence for phonological structure
(Jennifer Cole, University of Illinois at Urbana-Champaign)
10:15 - 10:30: COFFEE BREAK
10:30 - 1 1 :00: Sanskrit reduplication: A templatic approach (Gyanam Mahajan,
Brandeis University)
1 1 :(X) - 1 1 :30: Articulator^ and acoustic characteristics of apical and laminal stop
consonants: A cross-language study (Paroo Nihalani, National
University of Singapore)
1 1:30 - 12:(X): The pure vowels of Punjabi (Kamlesh Sadanand, University of
Hyderabad)
SansHrit Syntax SvmPQSiuni 404 Illini Union
Chair: Madhav Deshpande (University of Michigan)
9: 15 - 9:45: Demonstratives with non-third persons in Vedic Sanskrit (Stephanie W.
Jamison, Harvard University)
9:45 - 10:15: Reflexive pronouns in Vedic (Madelyn J. Kissock, Harvard University)
1 0: 1 5 - 1 0: 30: COFFEE BREAK
10:30 - 1 1 :00: The Vedic clause-initial string and Universal Grammar (Steven
Schaufele, University of Illinois at Urbana-Champaign)
1 1 :00 - 1 1 :30: Clausal vs. non-clausal subordination in Sanskrit narratives (Sarah
Tsiang, University of Dlinois ai Urbana-Champaign)
11:30- 12:00: Syntax or phonological form? Reconsidering some allegedly syntactic
phenomena of Vedic Sanskrit (Hans Henrich Hock, University of
Illinois at Urbana-Champaign)
Literary Responses tQ India 402 mini Union
Organizer and Chair: Girdhari Tikku (University of Illinois at Urbana-
Champaign)
9:15-9:45: RabindranathTagore's Nobel Prize: What does it mean? (Ali
Anushiravani, University of Illinois at Urbana-Champaign)
9:45-10:15: Chinese response to Tagore: Pin Hsin's poetry (Yongan Wu, University
of Illinois at Urbana-Champaign)
10:15-10:30: COFFEE BREAK
10:30 - 1 1:00: Aldous Huxley's The Island (Girdhari Tikku, University of Illinois at
Urbana-Champaign)
1 1 :00 - 1 1 :30: Coleridge and Basho: The legacy of Indian monism (Hiroko Harada,
University of Illinois at Urbana-Champaign)
11:30-12:00: DISCUSSION
[8]
SESSION IV (2:00 - 4:45 pm)
Dravidian Linguistics 11 405 mini Union
Chair: Alice Davison (University of Iowa)
2:00 - 2:30: A lexicalist analysis of participle compounds in Kannada (S. N. Sridhar
and Mark Aronoff, SUNY at Stony Brook)
Verbal compounds in Malayalam (Shuichi Yatabe, Stanford University)
COFFEE BREAK
2:30 - 3:00:
3:00-3:15
3:15-3:45
From aspect to tense in Old Tamil: Evidence from narrative discourse
(Susan Herring, California State University at San Bemadino)
3:45 - 4: 15: Coordination and word order (Nalini Rau, University of Illinois at
Urbana-Champaign)
4: 1 5 - 4:45: The associative case in Malayalam: Making sense of a catch-all category
(Rodney Moag, University of Texas at Austin)
General Session I 404 Illini Union
Chair: Norman Zide (University of Chicago)
2:00 - 2:30: Drastic modernization of the curricula of the teacher training courses (M.
V. Sreedhar, Institute for Socially Disadvantaged Groups)
2:30 - 3:00: The aesthetics of play in Punjabi folkloric tradition (Atamjit Singh, Guru
Nanak Dev University)
3:00-3:15: COFFEE BREAK
3:00-3:30: The web of the spider: Language and politics in Sri Lanka (Wilfrid
Jayasuriya, Southern Illinois University)
3:45 - 4: 1 5: A socio-cognitive approach to designing a self-instructional multi-media
course in English communicative skills (P. N. Pandit, Indira
Gandhi National Open University)
4: 15 - 4:45: Script Manager software for Indie scripts on the Macintosh (Lloyd B.
Anderson, Ecological Linguistics)
Language Identity and Conflict in South Asia 402 Illini Union
Organizer and Chair: Rakesh M. Bhatt, University of Illinois at Urbana-
Champaign
2:00 - 2:30: A house divided: Conflict and rivalry in two varieties of a language
(Mariam Ahmed, University of Illinois at Urbana-Champaign)
2:30 - 3:00: Transplanted languages and ethnic minorities (Tej K. Bhatia, Syracuse
University)
3:00-3:15: COFFEE BREAK
3:15 - 3:45: Language minorities: Issues of identity in a global perspective (Kamal K.
Sridhar, SUNY at Stony Brook)
3:45 - 4:15: Identity, conflict, and convergence: South Asia as a sociolinguistic area
(Rakesh M. Bhatt, University of Illinois at Urbana-Champaign)
4:15 - 4:45: Discussion (Eyamba Bokamba, University of Illinois at Urbana-
Champaign)
[9]
KEYNOTE LECTURE
(8:00 - 9:30pm, Levis Faculty Center, third floor)
Ladislav Zgusta
(Director, Center for Advanced Study, University of Illinois at Urbana-
Champaign)
Dvaikosyam:
The controversy over the Sanskrit-English dictionary of Monier-Williams
(Chair: Braj B. Kachru, University of Illinois at Urbana-Champaign)
(The talk is followed by a reception in the Levis Faculty Center)
[11]
Monday, 27 May 1991
SESSION V (9:15 am - 12:00 noon)
Svntax/SgmailtitS II 405 mini Union
Chair: Mangesh Nadkami (National University of Singapore)
9: 1 5 - 9:45: Conditionals and emphasizers in Bangla: Some effects of their interaction
(Tista Bagchi, University of Chicago)
9:45 - 10: 15: Advancement in some Asian and African languages (Nkonko M.
Kamwangamalu, National University of Singapore)
10:15-10:30: COFFEE BREAK
10:30-11:00: Compound verbs in Oriya (Gopabandhu Mohanty, Deccan College)
1 1 :00 - 1 1 : 30: Telugu negatives and non-capabilitatives: Morphological structure and
syntactic structure (Rosanne Pelletier, Yale University)
1 1:30 - 12:00: The category ofnominals in Bangla (Gillian Ramchand, Stanford
University)
General SgSSiQn 11 404 mini Union
Chair: Jennifer Cole (University of DUnois at Urbana-Champaign)
9: 15 - 9:45: Issues in translating Tamil Puranas (John A. Loud, University of
Wisconsin at Madison)
9:45 - 10: 15: The mental dictionary: Its role in linguistic theory (K. G. Vijaya-
krishnan, University of Hyderabad)
1 0: 1 5 - 1 0: 30: COFFEE BREAK
10:30 - 11 :(X): The Harappan Script: The most ancient form ofDravidian
(Clyde A. Winters, Uthman Dan Fodio Institute)
1 1:00 - 1 1:30: A sketchy history of cliticization and verb stem noun incorporation in
Munda (Norman Zide, University of Chicago)
Conventions of Politeness in South Asian Languages 406 lllini Union
Organizer and Chair: Manindra K. Verma (University of Wisconsin at
Madison)
A grammar of politeness in Marathi (Rajeshwari Pandharipande,
University of Illinois at Urbana-Champaign)
Towards an ethnography of politeness in Maithili (Mithilesh K. Mishra,
University of Illinois at Urbana-Champaign)
COFFEE BREAK
Directives in Punjabi andLahanda (Tej K. Bhatia, Syracuse University)
Linguistic strategies of politeness in Bhopuri and Magahi (Manindra K.
Verma, University of Wisconsin at Madison)
11:30-12:00: DISCUSSION
9:15-
9:45:
9:45-
10:15:
10:15
- 10:30:
10:30
-11:00:
11:00
-11:30:
1:45
-2:15:
2:15-
- 2:45:
2:45-
-3:00:
3:00
- 3:30:
[12]
SESSION VI (1:45 - 5:00 pm)
Hindi-Urdu Syntax and Semantics 405 lUini Union
Chair: K. V. Subbarao (University of Delhi)
Focus movement in Hindi-Urdu (Gurprit Bains, Long Island University)
Semantic case in Urdu (Miriam J. Butt and Tracy Holloway King, Stan-
ford University)
COFFEE BREAK
Perfectivity and the resultative state in Hindi (Jayshree Chakraborty, Agra
University)
3:30 - 4:00: Against wh-movement in Hindi (Anoop Mahajan, University of
Wisconsin at Madison)
Sanskrit Indigenous Grammar and Philosophy 404 lllini Union
Chair: Tej K. Bhatia (University of Syracuse)
1 :45 - 2: 1 5: Sociolinguistic parameters ofPdnini's Sanskrit (Madhav Deshpande,
University of Michigan)
2:15- 2:45: Assessing Sahara's arguments for the conclusion that a generic term
denotes just a class property (Peter M. Scharf, University of
Pennsylvania)
2:45 - 3:00: COFFEE BREAK
3:00 - 3:30: Semio-linguistic aspect ofDhvani Siddhdnta (Krishna K. Sharma,
University of Hyderabad)
3:30 - 4:00: Naming and expressing objects in Pdnini (Rama Natha Sharma,
University of Hawaii at Manoa)
Sociolinguistics 406 lllini Union
Chair: Rodney Moag (University of Texas at Austin)
1 :45 - 2: 1 5: Convent English: Structure and attitudes (Mariam Ahmed, University of
Illinois at Urbana-Champaign)
2: 1 5 - 2:45: Multilingualism and social identity: The case of Singapore (Nkonko M.
Kamwangamalu, National University of Singapore)
2:45 - 3:00: COFFEE BREAK
3:00 - 3:30: On liberating English to be a world language: An Indian perspective
(Mangesh V. Nadkami, National University of Singapore)
3:30 - 4:00: Functional articulation: Analyzing diglossic variation (John C. Paolillo,
Stanford University)
4:00 - 4:30: Sammelani Hindi and Malviya Hindi: Language and politics in India
between 1875 and 1930 (Alok Rai, University of Allahabad)
4:30 - 5:00: Aspects of the syntax of spoken Indian English (S. N. Sridhar & Indira
Ayyar, SUNY at Stony Brook)
[13]
Read by Title
Analogy as argument in Adi Samkara (Kapil Kapoor, Jawaharlal Nehru University)
On binding of reflexives in Sindhi (Pushpa Boolchandani, Kendriya Hindi Sansthan)
The development ofergativity in Indo-European languages of Western India in the 15th to
20th centuries (Boris A. Zakharin & L. V. Khokhlova, Moscow
State University)
The historical roots and development ofergativity in Indo-Aryan and Dardic (Boris A.
Zakharyin, Moscow State University)
On the loss of gender distinctions in Nepali (Carol Genetti, University of California at
Santa Barbara)
Pair-list answers in Hindi indirect questions (Veneeta Srivastav, Rutgers University)
Religious cum linguistic problems in modern India (M. Gnanam, Central Institute of
Hindi)
A situation-type analysis of compound verbs (Mona Singh, University of Texas at Austin)
Sociolinguistics of verbal abuse in Hindi (Raja Ram Mehrotra, Banaras Hindu University)
PART I: GENERAL ABSTRACTS
General abstracts. Thirteenth South Asian Languages Analysis Roundtable
CONVENT ENGLISH:
STRUCTURE AND ATTITUDES
Mariam Ahmed
(University of Illinois at Urbana-Champaign)
In South Asia, the term 'Convent English' has been used for a
variety of English which has developed in India and Pakistan in
schools originally initiated by British and American missionaries. It
is claimed that this variety of English has features which mark it as
distinct from the other varieties of South Asian English.
This paper focuses on three aspects of this variety: (a) linguistic
features at various levels; (b) the sociolinguistic context of Convent
English; (c) attitudes toward Convent English and its users and
attitudes of Convent English users toward other varieties of South
Asian English. This paper also presents a critical discussion of earlier
observations on Convent English.
18 Studies in the Linguistic Sciences 20:3 (Spring 1991)
SCRIPT MANAGER SOFTWARE FOR INDIC SCRIPTS
ON THE MACINTOSH
Lloyd B. Anderson
(Ecological Linguistics)
Script Manager software for Indie scripts can handle the auto-
matic choices of virama-marked consonants, half-consonants, sub-
scripts, and conjuncts when these are available in the fonts, as well
as irregular consonant-plus-vowel combinations such as /ru, ruu/.
These context-determined choices should not burden the user. In
the current version of the software, it is possible to have screen and
print show standard Devanagari while typing nearly normal trans-
literation. This brings much greater convenience for the user at
input and editing time, and nearly automatic transliteration between
different Indie alphabets and between them and Roman.
General abstracts. Thirteenth South Asian Languages Analysis Roundtable
CONDITIONALS AND EMPHASIZERS IN BANGLA:
SOME PRAGMATIC EFFECTS OF THEIR INTERACTION
Tista Bagchi
(University of Chicago)
As scholars such as Ferguson and Comrie have noted (Ferguson,
in Traugott et al. 1986: Overview; Comrie 1986), Bangla has two
principal conditional constructions. One involves a nonfinite condi-
tional participle (marked by the suffix /-le/, /-ile/ in the archaic
'High' form of written Bangla) formed from the verb of the protasis
(while the apodosis is a finite clause). The other is characterized by
the form /jodi/ 'if signalling the protasis (which, in this kind of con-
struction, is a finite clause) and optionally by one of the following
forms immediately preceding the apodosis: /tObe/ 'then', /ta hole/
'that being (so)' (itself containing the conditional participial form of
the verb /hO-/ 'be, occur'), and the discourse particle /to/ (which is
extremely difficult to translate, but in its most typical function
roughly denotes 'as we both (i.e., speaker and addressee) know or
accept as being true, ...'). It is with the second of these two construc-
tion types that this paper is concerned.
The problem addressed is the occurrence of the enclitic em-
phasizers (Dasgupta, 1984; treated as quantifiers by Ramchand 1990)
I'll (roughly, 'only', 'alone') and /o/ ('also, even') in the protasis of a
conditional sentence in Bangla. First, if one considers a sentence such
as (la), uttering this sentence carries the Gricean conventional im-
plicature (lb) (more generally, uttering a 'central' conditional, i.e., a
conditional construction of the logical structure p -> q, carries with it
the implicature -p -> -q, which is not present with a 'peripheral'
conditional — a conditional construction not logically representable as
a material-implication proposition p -* q, e.g. when q is a non-truth-
functional proposition such as a question or a request — as Eilfort
1987 demonstrates):
(1) a. golap phul-er rON jodi holde hOY
rose flower-Gen color if yellow be-3rd ord.-Pres.
20 Studies in the Linguistic Sciences 20:3 (Spring 1991)
(tObe/ta-hole) ta-te gOndho thake na
then/in that case it-Loc scent stay-3rd ord.-Pres. Neg
'If the color of a rose is yellow there is no scent in it.'
b. golap phul-er rON jodi holde na hOY
Neg be-3rd ord.-Pres.
(tObe/ta-hole) ta-te gOndho thake
stay-3rd ord.-Pres.
'If the color of a rose is not yellow there is (soine)
scent in it.'
The implicature is cancellable by the addition of the emphasizer
l-ol (here 'even') to /tObe/ta-hole/. However, its distribution in the
protasis (where it occurs optionally) is determined by what is pre-
supposed regarding the protasis, as will be shown in this paper. The
emphasizer /-i/, interestingly, does not readily occur in the protasis
of a 'central' conditional such as (la) or (lb), but it can occur in the
protasis of a rhetorical 'peripheral' conditional such as (2):
(2) tumi jodi na-(i) ele/aSbe
you-sg.ord. if Neg(-Emp) come-3rd ord. -Past/Future
(tObe/ta-hole/to) amar ar eSe ki labh?
then my anymore come-ConjPpl. what good
'If you aren't going to come what good will my coming be
(anymore)?'
I show that, in addition to the fact that certain tense co-
occurrence restrictions between protasis and apodosis hold for
'central' but not 'peripheral' Bangla conditionals (even of a rhetorical
sort), the non-occurrence of the emphasizer /-i/ with 'central' con-
ditionals has an explanation in the framework of data semantics as
spelled out by Veltman (1986) — in particular, in terms of the
modality of the protasis as a datum in the information state pre-
supposed in uttering the conditional. This may have wider im-
plications also for an understanding of the possible pragmatic con-
straints on the ways in which conditional constructions have evolved
in the history of Indie, from Sanskrit yadi and cet constructions
onward.
General abstracts, Thirteenth South Asian Languages Analysis Roundtable 2 1
FOCUS MOVEMENT IN HINDI-URDU
Gurprit Bains
(Long Island University)
In Hindi-Urdu the following contrast holds:
(1) a. *aap kahte hairii ki raam ne kaunsii kitaab paRhii
you say are that Ram-erg which book read
'You say Ram has read which book?'
b. kaunsii kitaab aap kahte hairh ki raam ne t paRhii
which book you say are that Ram-erg read
'Which book do you say Ram has read?'
Example (lb) appears to show that wh-movement is permitted in
Hindi-Urdu, in spite of the fact that Hindi-Urdu is a wh-in-situ
language (see Davison 1985).
In this paper I show that what looks like wh-movement in
Hindi-Urdu, as in example (lb), is an instance of the more general
phenomenon of Focus-movement (see Gair 1986 for a similar pheno-
menon in Sinhala), which in Hindi-Urdu may take place in syntax,
rather than at the level of Logical Form, as in English (see Chomsky
1976, May 1985). I provide evidence for the hypothesis that such a
movement must result in S or IP-adjunction (contra Chomsky 1986)
and that it is not an instance of movement of Spec of CP. 1 show that
once adjunction to IP is allowed, then an explanation can be found
for what appear to be violations of much of the subjacency condition
in its classical version (see Chomsky 1973) in the case of Hindi-Urdu,
for Hindi-Urdu permits extraction of focused wh-elements out of NP,
PostP, and in general out of wh-islands, with neither S nor S-bar
being observed as a bounding node. 1 show that, under the assump-
tion that IP-adjunction is allowed, almost all cases of extraction out
of bounding nodes result in 0-subjacency.
22 Studies in the Linguistic Sciences 20:3 (Spring 1991)
AN ESSAY ON KASHMIRI STRESS
Rakesh M. Bhatt
(University of Illinois at Urbana-Champaign)
There is a type of stress system that is not represented in the
typology of Hayes 1981. These are stress systems that have more
than two degrees of syllable weight, where stress is assigned to the
heaviest syllable of the word. Kashmiri is a language with such a
system.
In this paper, I propose an algorithm for assignment of word
stress in Kashmiri using the theoretical framework of Halle &
Vergnaud 1987. I argue that Kashmiri is similar to Khalkha Mangoli-
an (lacking a head marking rule on line constituent(s)) — with one
additional wrinkle: Khalkha Mangolian contrasts two degrees of syl-
lable weight (heavy vs. light) while Kashmiri has several degrees of
syllable weight, and the assignment of stress appears to be sensitive
to a syllable weight hierarchy, a distinction not easily encoded EITHER
in a theory which considers stress a binary feature OR a theory as-
signing stress without the use of metrical constituents.
The stress patterns of Kashmiri can be succinctly described in
the form of the following generalizations:
a. Only heads of syllable are stress-bearing.
b. The assignment of stress in Kashmiri is sensitive to a
syllable-weight hierarchy. The hierarchy is the following:
cwc > cw > cvc > cv
c. The heaviest syllable of the word receives the stress.
d. If two or more syllables qualify as the 'heaviest' because
they each have the same degree of syllable weight, then the
stress appears on the leftmost one.
e. The word-final syllable never appears stressed.
The analysis proposed in this paper captures the generalization
that long syllables are heavier than closed syllables which, in turn,
are heavier than (open) light syllables. Gupta (1987), Pandey (1989),
General abstracts, Thirteenth South Asian Languages Analysis Roundtable 23
and Davis (1989) have shown that the assignment of primary stress
for Sooke and for Hindi is sensitive to a syllable weight hierarchy. I
argue that Kashmiri, though superficially seeming to require a 'grad-
ual' or non-binary syllable weight distinction, does not in fact serve
as a counterexample to the claim that only binary syllable weight
distinctions exist in Universal Grammar. The syllable weight hier-
archy of Kashmiri falls out as a consequence of the combination of
rules needed for stress assignment. The special mechanisms devel-
oped in Halle & Vergnaud's metrical theory allow Kashmiri to be
analyzed using a combination of devices independently motivated for
other, less superficially complex, languages.
24 Studies in the Linguistic Sciences 20:3 (Spring 1991)
ON BINDING OF REFLEXIVES IN SINDHI
Pushpa Boolchandani
(Kendriya Hindi Sansthan)
In this paper I analyze reflexives in Sindhi in order to find out
whether the notion of c-Command & the principle of Binding are
adequate to formulate a theory of reflexives for the language. In
Sindhi, we have sentence constructions like:
(1) gi:ta: Je Pi:u Pa:na Khe baca:yo.
Gita Gen father self ace protected
'Gita's father protected himself.
(2) ghar Pa:na hi: Kiri: Pyo:.
house self emph. fell
'The house fell by itself.'
(3) ma: Pa:hiye ghar vana: tho.
I my house going (pres. cont.)
'I am going to my home.'
In the above sentences, the reflexive element Pd.na, the em-
phatic reflexive Pd.na hi:, and the reflexive-possessive Pa.hiye are
bound to their respective antecedent NPs. I analyze the rules and
conditions binding these reflexives with their coreferential elements.
Chomsky (1981) has formulated the Binding theory as follows:
a. A bound anaphor must be bound in its governing category.
b. A pronoun must be free in its governing category.
c. A lexical NP must be free.
A Governing Category has been defined as: 'a is the governing
category for X if and only if a is the minimal category containing X,
and a subject accessible to X.' The governing categories are S and NP.
General abstracts. Thirteenth South Asian Languages Analysis Roundtable 25
I examine whether the rule of Binding and the notion Governing
Category can account for the coreferentiality in the following types of
sentences of Sindhi:
(4) nia:i huna kha: 2 Pa:hi:jei/2 ghara jo pato puchyo.
I him/her from my/his home gen address asked
'I asked him the address of my/his home.'
(5) ma:ui dhi:ukhe2 Pa:hi:jai/2 gaha {pa:ra:e chadya:}
mother daughter ace her jewels made to wear
'The mother asked the daughter to wear her jewels.'
(6) ma:ui dhi:ukhe2 Pa:hi:jai/2 gaha pa:ina la:e cayo.
mother daughter ace her jewels wear asked
'The mother asked the daughter to wear her jewels.'
(7) ma:i hunakhe2 cayo ki pahrf: Pa:hijo*i/2 kam kar.
I he/she ace. asked that first self work do
'I told him/her to do his/her work first.'
(8) tu:i mukhe2 Pa: hi:joi/*2 kita:bu de.
you me ace self book give
'You give me your book.'
26 Studies in the Linguistic Sciences 20:3 (Spring 1991)
SOME OBSERVATIONS ON THE DEVELOPMENT OF WEST INDO-
ARYAN PRONOMINAL SYSTEMS FROM APABHRAM^A.
Vit M. Bubenik and C. Paranjape
(Memorial University of Newfoundland)
In this paper we address pertinent issues which have not
received satisfactory explanation and due attention in older and
more recent Indological literature. This will be done on the basis of
primary data taken from representative texts (in Apabhrarhsa, Old
Marathi, and Old Hindi) covering the period of the tenth to sixteenth
centuries A.D. In addition. Old Western Rajasthani and Old Gujarati
pronouns are considered for a complete picture of the system.
The pronominal inflection of Apabhrarhsa is in a transitional
stage with respect to that of Old Indo-Aryan. Our purpose is to
investigate the nature of this change in Apabhrarhsa and the extent
to which it is carried over in Old Hindi and Old Marathi. Some of the
issues to be discussed in terms of diatopy and diachrony are:
(1) The declining use of pronominal clitics of Middle Indo-
Aryan (with only a few relics observable in late Apa-
bhrarhsa and Old Hindi);
(2) The emerging oblique forms for pronouns to host postposi-
tions on a par with the nouns.
(3) The loss of the double-oblique system in late Apabhrarhsa
(maiM T Acc/Instr) with some relics found in Old West
Rajasthani and Old Gujarati, and the use of new syncretic
systems operating on some pronouns in Old Hindi and Old
Marathi.
(4) The nature and direction of paradigmatic leveling observ-
able in the system of pronouns in late Apabhrarhsa.
(5) The syntactic behaviour of personal pronouns of Old Hindi
and Old Marathi in the context of ergativity.
General abstracts. Thirteenth South Asian Languages Analysis Roundtable 27
SEMANTIC CASE IN URDU
Miriam J. Butt and Tracy Holloway King
(Stanford University)
An analysis of subject and object case marking in Urdu involves
both grammatical (functional) and semantic information. Lexical
Functional Grammar (LFG) has provided accounts for both lexical
(idiosyncratic) and grammatical case marking in various languages
(Bresnan et. al. 1982), but as of yet has not concentrated on an ac-
count of semantic case. My analysis of subject and object case
marking in Urdu modifies current implementations of Lexical Func-
tional Grammar to allow both grammatical and semantic information
to influence case marking.
In Urdu, grammatical subjects are not restricted to the nomina-
tive case but can appear with almost any case (Mohanan 1990). (1)
is an example of a dative subject construction, while (2) has a geni-
tive subject.
(1) naila-ko kitaab milii.
naila-Dat book-Nom received
Naila received a book.'
(2) naila-kii ek bahin hai
nail-Gen one sister-Nom is
'Naila has one sister.'
It might appear that the case of the subject must be designated
individually (idiosyncratically) for each verb. However, Mohanan
(1990) showed that the case of Hindi subjects and objects is predict-
able given both semantic and syntactic information. Although some
verbs require either only semantic or only syntactic information in
determining subject case marking, others depend on the interaction
of syntactic facts with semantic information like volition, definite-
ness, or aspect. Examples (3) and (4) demonstrate how semantic in-
formation interacts with grammatical information to determine case
marking in Urdu. Only subjects in Urdu can be in the ergative case,
28 Studies in the Linguistic Sciences 20:3 (Spring 1991)
but as (3) shows, whether the subject is marked with the ergative
also depends on volitionality. (4) focuses on the alternation of ac-
cusative and nominative case objects. When an inanimate object is
definite, it appears in the accusative, as in (4a).
(3) a. mohan-ne cillayaa.
mohan-Erg screamed
'Mohan screamed (on purpose).'
b. mohan cillayaa
mohan-Nom screamed
'Mohan screamed (involuntarily).'
(4) a. naila axbaar-ko parhtii hai
naila-Nom newspaper- Ace read be-Pres
'Naila reads the newspaper.'
b. naila axbaar parhtii hai
naila-Nom newspaper-Nom read be-Pres
'Naila reads newspapers.'
Previously, the assignment of non-nominative case to gram-
matical subjects involved the stipulation of idiosyncratic (lexical)
case. In LFG this consists of specifying in the lexical entry for each
verb the case of the subject, unless it appears in the nominative.
This technique was used by Andrews (1982) to explain the case
marking of non-nominative subjects in Icelandic. However, the stip-
ulation of subject or object case in each lexical entry is redundant for
a language like Urdu where this case-marking is predictable on the
basis of semantic information. To capture this regularity, I make use
of the relation between functional-structure and argument-structure,
which has been incorporated into LFG to account for the lexical
mapping of arguments and complex predicates (Isoda & Sells 1990).
General abstracts, Thirteenth South Asian Languages Analysis Roundtable 29
PERFECTIVITY AND THE RESULTATIVE STATE IN HINDI
Jayshree Chakraborty
(Agra University)
Aspectual contrast of the imperfective versus perfective verbal
forms in Hindi has already been discussed by a number of scholars.
Perfectivity in Hindi is expressed by compound verb expressions
consisting of a main verb and a modifying verb in which the latter,
being grammaticalized, loses its conceptual meaning and adds an
additional semantic dimension to the main verb. For example:
usne am kha liya/
He+ case mango eat take + asp+tense
'He has eaten the mango.'
The verb le 'to take' here has been grammaticalized and, along with
other senses, it gives the sense of perfectivity. In Hindi, however,
there is another verbal element cuk which expresses perfectivity and
which just like a modifying verb is attached to the main verb (which
appears in the bare root form). For example:
Vah am kha cuka
He mango eat asp+tense
'He has eaten the mango.'
Although in a number of contexts, such as the ones given above,
the two constructions, i.e., the compound verb and the verb + cuk
constructions, appear to have the same interpretation, closer examin-
ation of the two brings about a number of differences. Scholars of
Hindi working in this field have, however, not paid much attention to
the essential differences between these two constructions, or, they
have not paid attention to a closer examination of the status of cuk in
the language.
This paper brings out the syntactic and semantic differences be-
tween the two types of perfective constructions in Hindi. In order to
do so, I first define the conceptual status of the two different verbal
30 Studies in the Linguistic Sciences 20:3 (Spring 1991)
elements: operators (such as le-) and cuk. The following two as-
sertions are made in this respect:
(i) While an operator essentially emphasizes the termination of
an act, cuk clearly indicates the resultative state;
(ii) While an operator does not necessarily indicate an expected
act, cuk essentially does imply an expected or presupposed
act.
The paper provides the scope for testing the validity of cuk as
an indicator of resultative state cross-linguistically. Data from Ben-
gali showing resultative state are examined, and it is found that the
syntactic and semantic tests valid for testing cuk as an indicator of
the resultative state in Hindi are valid for them as well.
General abstracts, Thirteenth South Asian Languages Analysis Roundtable 31
CASE MARKERS AND POSTPOSITIONS IN KANNADA
S. Chandrashekar
(University of Washington and California State University at Fresno)
and
S. N. Sridhar
(State University of New York at Stony Brook)
Traditional grammars of Kannada maintain a clear distinction
between morphological case markers and postpositions. These
accounts list -u (nom.), -annu (ace), -ge (dat.), -inda (instr.), -alii
(loc), and -a (gen.) as case markers, and me.le 'on', oLage 'in/inside',
hattira 'near', etc., as postpositions.
Recently, Sridhar (1989:160), although maintaining the same po-
sition regarding these categories, has raised some doubts about such
a treatment without going into further details.
In this paper, we argue that Instrumentals and Locatives are not
really case markers but true postpositions and what was considered
as (true) postpositions in earlier accounts are really pseudo-post-
positions sharing some [+N] features along with other nominal ele-
ments in this language. Our analysis crucially hinges on the case as-
signment property of pseudo and real postpositions. Whereas true
postpositions assign 'oblique' case to their arguments under govern-
ment, pseudo-postpositions assign genitive case to their arguments
just like other nominal elements. But, there is a clear distinction
between pseudo-postpositions and nominal elements, in the sense
that pseudo postpositions cannot be preceded by nominal modifiers
(such as determiners and adjectives) and cannot occur in the subject
or object positions. The case of Datives is slightly different and
somewhat controversial. We want to maintain that the Dative -ge is
a case marker but behaves differently from the Nominative and Ac-
cusative case markers.
32 Studies in the Linguistic Sciences 20:3 (Spring 1991)
ALLITERATION IN SINDHI POETRY:
EVIDENCE FOR PHONOLOGICAL STRUCTURE
Jennifer Cole
(University of Illinois at Urbana-Champaign)
A linguist analyzing poetry looks for patterns and constructs
that reveal the structure of linguistic representation. Following
Kiparsky (1973), we may ask what are the syntactic or phonological
building blocks which are subject to patterned recurrence, and how
is parallel structure defined? The present paper attempts to answer
these questions with respect to the pattern of alliteration found in a
popular form of Sindhi poetry (particularly in the work of the seven-
teenth-century poet Shah Latif). The study of alliteration in Sindhi
poetry presents clear evidence for a theory of phonological segments
which attributes a hierarchical, articulatory-based organization for
distinctive features (Clements 1985, Sagey 1986), and entails some
surprising results concerning the relationship between dental, retro-
flex, and palatal consonants in Sindhi.
The pattern of alliteration found in Shah Latif s poetry is illus-
trated in the following verse:
kaadiru pahanje kudarata siin
kaaimu aahi kadiimu
In this example, we find a pattern in which a sequence of consonants
is repeated twice in each line: /k, d, r/ in the first line and /k, m/ in
the second. Other verses show patterns built on a single consonant, a
single vowel, or sequences of consonant and vowel. In the example
shown here the pattern involves a repetition of the exact same se-
quence of consonants, but in other examples the pattern may involve
consonants which are only partially identical, such as /t, t /, /d, dh/,
/j, j7. These patterns demonstrate that the following sounds are
identical for the purposes of alliteration: dentals and retroflex, plain
and aspirate stops, plain and implosive stops. In addition, we find
vowel patterns which include long and short vowels /a, aa/, and
even patterns with distinct vowels /a, o/ (note: [a] = schwa).
General abstracts, Thineenth South Asian Languages Analysis Roundtable 33
In this paper, I explore the interpretation and analysis of these
and other facts, which pose several challenges for phonological theo-
ry. In current models, identity between two segments is calculated
on the basis of the hierarchical representation of distinctive features.
Two segments may be identical if they share Place of Articulation
features, or Place + Manner features, or simply Laryngeal features.
The feature geometry allows only certain groups of features to
function in defining a natural class. In Sindhi alliteration, we find
that the Laryngeal features of aspiration and glottalization may be
ignored in determining identity (e.g. /d, dh), but segments in the
same identity class must bear consistent [voice] specifications. The
identity class including retroflex and dentals (/t, t/) seems to in-
dicate that Place of Articulation is relevant and different kinds of
coronals count as identical. Yet [c] never occurs in the same class as a
dental or retroflex.
Another interesting characterization of this system is that pat-
terns can consist of sequences of sounds which are not adjacent, as in
the example seen above. The non-linear framework of current pho-
nological theory explains why this pattern is possible, and predicts
that alliterative patterns involving non-adjacent segments will ne-
cessarily define identity in terms of Place of Articulation.
34 Studies in the Linguistic Sciences 20:3 (Spring 1991)
FINITENESS AND CASE IN HINDI-URDU COMPLEMENTS
Alice Davison
(University of Iowa)
Finite clauses in Hindi-Urdu, as in many other South-Asian
languages, do not occur phrase-internally except under highly res-
tricted conditions. Hindi-Urdu, lacking a quotative or a disjunctive
suffix, allows finite clauses only in external positions. For example,
the complement clause in (la) may not be inside V as an argument
receiving case directly from V:
(1) a. *ham [[ki woo nahii aaeegaa] sooctee haf]
we that 3p not come-fut-sg think-impf are
b. ham [(yah) sooctee hai] [ki woo nahii aaeegaa]
'We think [that he won't come]' (cf. Subbarao 1984)
The finite clause is left or right adjoined to the whole main clause, as
in (lb), or is right adjoined to a noun such as baat 'matter', khabar
'news', etc. The position to the left of a lexical category like N, V, A,
or P assigns Case, and only non-finite clauses may occur there.
This restriction holds for all types of finite clauses, including
complements, relative clauses, and adverbials. It does not
distinguish between clauses which are arguments and must get a
thematic role like theme, and modifiers which do not receive a
thematic role. Relative clauses, for example, are exactly like comple-
ments in this respect, but do not get thematic roles:
(2) a. *[joo kal aayaa hai [laRkee-nee]] mujhee bataayaa.
which yest. came boy-erg I-dat. told
b. [PRO kal aaee huee [laRkee-nee]] mujhee bataayaa
c. is-laRkee-nee mujhee bataayaa [joo kal aayaa hai]
'The boy [who came yesterday] told me (this).'
(cf. Kachru 1978).
General abstracts. Thirteenth South Asian Languages Analysis Roundtable 35
Stowell's (1981) account of the 'Case Resistance' of finite clauses
ascribes the ungrammaticality of (la) to two factors: (i) the finite
complement cannot receive Case, and (ii) phrases lacking Case are
not 'visible' for Theta role assignment. The Hindi-Urdu data show
that failure to receive a theta role is not the cause of ungram-
maticality, since (2a) is also ill-formed, and no theta role is involved
in a modifier clause. Rather, ill-formedness results from the incom-
patible combination of finite inflection and morphological features
imposed by a governor governing to the left. These generalized
'Case' features include verbal affixes and the requirement of a null
subject, as in (2b). Government and Case assignment to the left
always places these features on embedded verbs in head-final
languages. The consequences of this are escaped in other SOV langu-
ages by the presence of a quotative or disjunctive suffix in final
complementizer position which is inert to Case (cf. Steever 1987).
36 Studies in the Linguistic Sciences 20:3 (Spring 1991)
SOCIOLINGUISTIC PARAMETERS OF PANINI'S SANSKRIT
Madhav Deshpande
(University of Michigan)
This paper reviews the latest research in relation to the status of
Panini's Sanskrit. Was it a spoken language? Was it a living lan-
guage? Was it a first or a second language? Who spoke this langu-
age, and to whom? These questions, while they have been discussed
before, need to be looked at again. The extreme views in this regard
range from considering Panini's Sanskrit as being his mother tongue,
to believing that Sanskrit in general including Panini's Sanskrit, was
never a truly colloquial spoken language, and that the speakers of
Sanskrit always had a Prakrit or some other non-Sanskrit language
as their mother tongue. Both extremes have their pitfalls, and one
needs to have a more exact characterization. The present paper
attempts to arrive at such a characterization.
General abstracts, Thirteenth South Asian Languages Analysis Roundtable 37
ON THE LOSS OF GENDER DISTINCTIONS IN NEPALI
Carol Genetti
(University of California, Santa Barbara)
The Indo-Aryan language Nepali is closely related to Hindi and
is quite similar grammatically in many respects. However, one
aspect of Nepali grammar that differs strikingly from Hindi is the
gender system. While Hindi has lexical gender and a full system of
gender based verb agreement, in Nepali the system is quite different.
The loss of lexical gender distinctions in Nepali has been noted
previously. While there are still a few feminine nouns in Nepali,
they are etymologically derived from forms with feminine
morphology (i.e. nokarlnokarni 'servant'); the vast majority of
Nepali nouns are not distinguished by gender. Grammatical
morphology, on the other hand, still shows some agreement with
semantically feminine animate nouns. The feminine forms however,
while recognized and used in formal registers in Nepali, are
relatively rare in the spoken language.
The current study will be an examination of the distribution of
gender agreement in several genres of Nepali discourse. The
prescribed system of gender agreement will be compared with the
actual distribution of forms in order to ascertain which morphological
environments are most likely to conserve gender agreement, and
which have lost it entirely. Written genres of the study include
newspaper and scholarly articles and a woman's magazine about
current media celebrities. Spoken genres will include narrative and
conversation.
The paper will conclude with consideration of the motivation for
the apparently ongoing loss of gender distinctions. Bloch suggests
that gender loss is due to substratum influence; a hypothesis which
seems very likely. In this particular case, the substratum group
would be speakers of Newari, a language with no gender morphology
whatsoever. The hypothesis will be tested by comparing the
38 Studies in the Linguistic Sciences 20:3 (Spring 1991)
percentages of gender agreement found in the speech of bilingual
Newars and monolingual Nepalis.
General abstracts. Thirteenth South Asian Languages Analysis Roundtable 39
RELIGIOUS CUM LINGUISTIC PROBLEMS IN MODERN INDIA
M. Gnanam
(Central Institute of Hindi)
Religion and language are two major factors of social identity
and cultural barriers in India. These two social institutions indivi-
dually are capable of generating emotional integration, as well as
disintegration as its essential counterpart, and when they join to-
gether as often seen in India it becomes a matter of serious study for
the sociologists of language as far as the discipline of linguistics is
concerned.
Right before the time of Buddha, religions and languages went
together fully or partially and caused problems in India. With the
advent of time the religious and language groups have often, if not
always, grown large in number and size and problematic in nature.
At the time of liberation from colonial rule, India underwent a
partition after considerable bloodshed on account of religious and
linguistic attitudes.
In modern India (i.e., India after independence) we find the
following religious and linguistic ties.
Hinduism + Sanskrit / Sanskritized Hindi
Islam + Urdu / Persianized Hindi
Christianity + English or French or Portuguese
Sikh religion + Punjabi
Atheism / Anti-Hinduism + Tamil / pure Tamil
Besides, there are hundreds of tribes together with their
traditional religions. But these religions are less institutionalized and
their languages have less communal affiliations.
There are a lot of separate religious or linguistic problems also
in India. But this paper discusses only those situations when they
mix up and result in sociolinguistic problems.
40 Studies in the Linguistic Sciences 20:3 (Spring 1991)
THE SOURCES OF A PASSIVE
Eric P. Hamp
(University of Chicago)
It is not clear how Proto-Indo-European expressed a predication
in passive diathesis, nor to what degree such an argument structure
was resorted to in Indo-Iranian, to judge by the earliest documented
attestation, employed for the passive, an idiosyncratic and defective
paradigm of forms. The background of these inflexions and syntaxes
has not been adequately traced. It seems to have escaped notice
that, using different inherited morphs and combinations of cate-
gories. Old Irish (and, in basically like fashion, British Celtic) shows
just the same suppletive and defective configuration of paradigmatic
inflexions. From this complex correspondence we may infer a PIE
ancestor set of morpho-syntaxes and lines of descent to the two
daughter branches.
The facts: Indo-Iranian passive in -yd- (badh-yd-te, uh-yd-te,
s'lr-yd-te, kr-iyd-te: Avest. kir-ye-Hi) is restricted to the presential
system, requires zero grade, and is distinguished from the fourth
class and the denominatives. Avestan evidences not only early ac-
tive inflexion but also a near-restriction to the 3rd person. Outside
the distinctive presential the only separate inflexion is the lone 3sg.
aorist, gunated with Brugmann's Law + -/ {d-kar-i, (d)s'dv-i, (d)jdn-i <
*k^6r-i, k'lou-i, g6n?-i), to which there was the added possibility of a
periphrastic phrase with the perf. pass. pple. in -td- (tatdm me dpas
'my work is done' (RV 1:110:1); yuktds te astu ddksinah 'let thy
right (horse) be yoked' (RV 1.82.5)).
The Old Irish and Mediaeval Welsh passive (and Breton imper-
sonal) with endings in -r (Olr. carthair carthar, leicthir leicther,
berair berar, subjunctive berthair berthar, gessair gessar, relative
liicther, berar, berthar, gessar, plural cartair cartar, moraitir mortair
mdr(a)tar, miditir, lecetar, bertair bertar, gessatar, rel. mdr(a)tar,
leic(e)ter leicetar, bertar, and correspondingly in future formations:
Welsh cerir, subj. car(h)er; Bret, laser 'is killed', guillir 'one can',
future caror 'will be loved') applied this *-ri only to the Celtic prima-
General abstracts. Thirteenth South Asian Languages Analysis Roundtable 41
ry tenses, and the inflexion was formed only for the third person,
first and second being treated as if morphological objects (Isg. no-
mcharthar, Ipl. no-ncarthar 'I am, we are loved'; y-m gelwir 'I am
called', ny-n cerir 'we are not loved'). The third singular preterite
was suppletive (Olr. carad, Uced, breth brethae, pi. cart ha,
Micthea, bretha, W. carat, erchit; Br. croeat 'was created'; Olr. slass
'was struck' = W. lias 'was slain'; Olr. fess 'was known' = W. gwys =
Br. gous) < PIE *-ro-participle with or without auxiliary 'be'.
The inferred source structures are:
*intransitive-existential -ie/o- {- nonpersonal)
(with verb base in complementation with 'denomina-
tives' — On the restricted Indo-Iranian evidence see
now E. Tucker (1988), TPS 86.93-114)
*impersonal -r-
*-rd-participle predication ~ *perfect ('pret-pres.') + hic
The descents are:
*John loves/d Mary
*There is love, 'one loves'; *Somebody loves M. it is Mary
*It was loved; it is Mary
CELTIC
> One loves Mary > Somebody loves M., me, us;
Somebody loves
Mary was loved Mary was loved
INDO-IRANIAN INDO-IRANIAN AND CELTIC
> Mary is loved > Mary is, I am, we are,
sg. is, pi. are loved
Mary was loved Mary was loved
(instrum. agent) > preterite expanded to match
primary range of arguments
42 Studies in the Linguistic Sciences 20:3 (Spring 1991)
FROM ASPECT TO TENSE IN OLD TAMIL:
EVIDENCE FROM NARRATIVE DISCOURSE
Susan Herring
(California State University at San Bernadino)
Despite Tolkappiar's assertion around the first century A.D. that
old Tamil had three 'times' or tenses, modern scholars are consistent
in reconstructing only two morphological tenses — a past, and a non-
past — for the ancient language, and indeed for Proto-Dravidian as a
whole. Zvelebil (1962), while supporting a two-way distinction,
suggests however that the opposition in Early Old Tamil might better
be viewed as 'perfective past' vs. 'imperfective future', indicating 'a
primary aspectual (and not temporal) dichotomy' (15). To anyone
who has worked with Old Tamil texts, this view has intuitive appeal,
yet surprisingly, no attempt has been made to verify it with textual
evidence.
In this paper I propose to examine portions of two epic narrative
texts — Cilappatikaram (5th c.) and Kamparamayanam (12th c.) — in
an effort to provide support for Zvelebil's claim. Narrative discourse
is ideally suited to the study of tense/aspect, since it crucially
involves the notions of sequence and simultaneity. There is a
growing body of research which shows a correlation between
tense/aspect forms and basic narrative functions, such as event line
vs. background information, and primary event line vs. secondary
event line. Much of this work focuses on the discourse functions of
aspect: In numerous languages, perfective aspect 'foregrounds' or
highlights the main narrative event line, while imperfective aspect
encodes 'background' situations and events (Hopper 1979).
My preliminary analysis of the distribution of 'tense' forms in
the two Tamil narrative texts reveals significant functional patterns.
Cilappatikaram contains relatively few tensed forms. However, what
past forms appear consistently encode events which are dynamic,
crucial to the plot development, or otherwise contextually fore-
grounded, while non-past forms (used with past time reference) are
found in static, descriptive passages, and with verbs of saying which
General abstracts. Thirteenth South Asian Languages Analysis Roundtable 43
introduce quotes. The Kamparamayanam, composed seven centuries
later, after the modern three-way tense system was already in place,
exhibits a different set of patterns from Cilappatikaram. While pre-
serving the stylized use of non-past forms to introduce quoted ma-
terial, the later text uses past forms for narrative events and situ-
ations of all types, the one exception being the use of non-past to
refer to habitual activities in past time (a sense still conveyed by the
future tense in Modern Tamil).
There is thus evidence in support of the view that in Old Tamil,
the two-way distinction was a (functionally) aspectual one — i.e. fore-
ground/background — whereas by the twelfth century, tenses had
essentially taken on their modern values, leaving the fore-
ground/background distinction to be signalled, presumably, by other
In concluding, I consider the implications of these findings for
the historical development of aspectual auxiliary verbs in Tamil, sug-
gesting that aspectual auxiliaries were innovated after the original
perfective/imperfective distinction was lost, i.e., in an attempt to
regain the functional utility of the earlier system. I cite examples of
'grounding' in Modern Tamil narrative via perfective and imper-
fective auxiliaries and discuss possible motivations for the morpho-
syntactic reorganization of the verbal system as a whole.
REFERENCES
Hopper, Paul. 1979. Aspect and foregrounding in discourse.
Discourse and syntax, ed. by T. Givon. New York: Academic
Press.
Zvelebil, Kamil. 1962. Lectures on historical grammar of Tamil.
Madras: University of Madras.
44 Studies in the Linguistic Sciences 20:3 (Spring 1991)
SYNTAX OR PHONOLOGICAL FORM?
RECONSIDERING SOME ALLEGEDLY SYNTACTIC PHENOMENA
OF VEDIC SANSKRIT
Hans Henrich Hock
(University of Illinois at Urbana-Champaign)
A number of Vedic Sanskrit phenomena have traditionally been,
and continue to be, considered syntactic. These are (i) the fact that
finite verbs are said to be accented in dependent clauses (DCs) but
unaccented in main clauses (MCs), unless initial; (ii) the fact that
certain constructions (viz., deliberative questions) are characterized
be clause-final pluti (+ trimoric vowel in the final syllable); and (iii)
the fact that particles, discourse-deictics, clitic pronouns, and topic-
alized elements tend to stack up at the beginning of the clause in
what I have referred to (Hock 1982) as 'initial strings'. Cf. the illus-
trations in (1) - (3).
I show that none of these phenomena is purely syntactic and
that significant aspects of their behavior must be accounted for in
Phonological Form (PF).
This is most easily done for types (1) and (2), since their overt
manifestations (accent and trimoricity) are clearly phonological.
Further, their conditioning is not exclusively syntactic but involves
other, mainly semantic/pragmatic, factors. In addition, non-initial
verb accentuation is not limited to DCs, but occurs in certain MCs, as
in (4). Klein (In Press) shows that in some cases, non-initial MC verb
accentuation is conditioned by poetic structure (such as caesuras).
Moreover, MC verbs not initial in their clauses, but initial in the
poetic line, always are accented; cf. (5). Sanskrit poetic structure,
however, is clearly to be accounted for in PF, not in the syntax.
For the type (3), the argument is more subtle. Here again, pho-
nological criteria play a role, since the relative order of particles and
clitic pronouns is determined by their underlying accent. Secondly,
while there are conceivable syntactic accounts that can get topic-
alized elements and discourse-deictics to stack up in the proper
General abstracts, Thirteenth South Asian Languages Analysis Roundtable 45
order (by moving then to different 'landing sites'), the relative order
of particles can be given a syntactic account only be a highly partic-
ularized proliferation of additional, 'adjoined' landing sites. A PF
account, which can be simultaneously sensitive to syntactic in-
formation and phonological criteria, provides a much more elegant —
and motivated — alternative. (Something similar could be accomp-
lished by letting the syntax wildly generate any conceivable ordering
and letting a 'filter' eliminate the ones that do not occur. But that
filter would have to contain the same information that the PF account
would, except as a negative, rather than a positive conditioning, and
without any plausible motivation.)
I conclude the discussion by considering the implications of
some of these findings for linguistic theory in general.
(1) a. [tasmai visah syayam eva namante]Mc
[yasmin brahma purva eti]Rc (RV 4.50.8)
'Before him even the common people bow, for whom
the brahmin is preeminent.'
b. [syama ^ id indrasya sarmani]Mc (RV 1.4.6c)
'May we be under Indra's protection.'
(2) adhah svid asfsd upari svid asfst (RV 10.129.5)
'Was it above, or was it below?'
(3) (NEXUS) 12 3 4 5
atho X/t> P P E t)
(Where NEXUS =- clause connector; P = unaccented
particle, P - accented particle; E = enclitic pronominal;
D = pronominal {tad, etdd, ydd, kim); X = accented
fronted word other than particles or t).)
Cf. pra ha va enam pas'avo vis'anti (MS 1.8.2)
(X P f> E ...)
'The cattle indeed turn toward him.'
(4) para ca yanti piinar a ca yanti (RV 1.132.12)
'They go away and they come back.'
(5) surupakrtnum utaye sudiigham iva goduhe I juhumasi
dvavi-dvavi (RV 1.4.1)
'Day by day we invoke for support the one who puts on
a good shape just as (we approach) the good milk cow
for the milker.'
46 Studies in the Linguistic Sciences 20:3 (Spring 1991)
KASHMIRI CAUSALS:
EVIDENCE FOR A TRANSFORMATIONAL APPROACH
Peter E. Hook
(University of Michigan)
and
Omkar Nath Koul
(National Academy of Administration, India)
In his recent monograph on the morphological causative in
Kashmiri, S. Syeed presents evidence for a lexical rather than a
syntactic ('transformational') account. His primary arguments are
two: (i) There are unpredictable gaps and idiosyncrasies in the forms
and meanings of members of morphologically related causal sets, (ii)
The number of iterations of the causal morpheme -inaay in causal
stems cannot be correlated with the number of causees in the PAS's
(predicate argument structures) of such stems.
We accept the first of these arguments as sufficient to justify the
separate listing in the lexicon of individual members of causal sets.
However, we show that a closer examination of the second argument
leads to a conclusion opposite to that drawn by Syeed. In particular,
instances in which the number of causees is greater than the number
of iterations of the causal suffix force us to posit a syntactic source
for much of the Kashmiri causal data. In (1) below, the correct
interpretation of the reduplicated conjunctive participle preery can-
not be obtained except by allowing the finite verb to have scope over
it. That scope includes the causal agent of hokhir-aav 'make dry;
exasperate' which has to be inherited by the non-causal predicate
praar 'wait'. To account for this inheritance in the lexicon would
force the Lexicalist to set up an extra (empty) slot for a causal agent
in the PAS of every predicate in Kashmiri. Further, it would render
the principle of Full Interpretation (Projection Principle) vacuous
('Thematic roles are maintained throughout a lexical derivation'
Roeper 1987:274), since the filling of the extra slot would depend on
whether or not a predicate fell under the scope of a causal in a tree.
General abstracts. Thirteenth South Asian Languages Analysis Roundtable 47
(1) temy hokhir-eev zanaani preer-y preer-y
he-ER dry-CAUSed wife wait-CPM wait-CPM
'He exasperated his wife by making her wait and wait.'
(Cf. Syeed 1985:48)
We adduce data from Kashmiri to show that inheritance of a
causal agent must be allowed even if the bequeathing predicate is a
simple monomorphemic lexical transitive:
(2) razyi thaph ker-yith khooru-n-as bi paS-as pyeTH
rope-DT grip do-CPM lifted-3s-ls I roof-DT onto
'He made me hold onto the rope and climb onto the roof.'
In (2), the form khoor-n-as (from khaar 'raise; send up') is in a sup-
pletive relation to the intransitive khas 'climb'. Even so, the tran-
sitive thaph kar 'grab, hold' has to inherit khaar's agent as its own
causal agent in order to yield the causative interpretation that it has
in (2). Therefore, we must posit not only a syntactic rule for causa-
tive but also allow generative semantic decomposition if we are to
obtain the inputs such a syntactic rule would need.
Thus, the complete set of semantic and syntactic facts requires
both a lexical and a syntactic treatment of causal phenomena in
Kashmiri as well as a (de)compositional theory of lexical semantics.
REFERENCES
Roeper, T. 1987. Implicit arguments and the head-complement re-
lation. Linguistic Inquiry 18:267-310.
Syeed, S. M. 1985. Morphological causatives and the problems of the
transformational approach. Bloomington: Indiana University
Linguistics Club.
48 Studies in the Linguistic Sciences 20:3 (Spring 1991)
DEMONSTRATIVES WITH NON-THIRD PERSONS
IN VEDIC SANSKRIT
Stephanie W. Jamison
(Harvard University)
The demonstratives sa/tad and, less commonly, esaletad can be
employed with non-third person pronouns and verb forms (type so
'ham, sa tvam; sa bodhi, sa naydmi, etc.). Though this phenomenon is
well-known and frequently mentioned in the secondary literature
(e.g. Delbruck, MS §138; Speyer/Speijer, Skt. Syn. §445; Ved. u. Skt.
Syn. §266; Renou, Gr. 1. ved. §400; Gr. 1 sans. §260), it has not, to my
knowledge, been the subject of a thorough study. This paper makes
such a study, utilizing a complete collection of the examples in the
Rg-Veda and extensive collections from the Atharva-Veda, Sarhhita
prose, and Brahmana prose, as well as (rarer) examples from the
Mahabharata and Classical Sanskrit texts.
The study focuses especially on asymmetries in the distribution
of the examples: distributional variation between 1st and 2nd
persons, doubling of enclitics and of accented pronouns, of nom-
inatives and other cases, of pronouns and verb forms, of indicatives
and other moods of the verb, as well as differences due to chronology
and/or text type. The context(s) of the examples are considered, and
the function(s) of the doubling, when discernible, discussed.
The treatment of salesa doubling complements one recently com-
pleted on ayamlasau doubling (Jamison, World Sanskrit Conference,
August 1990) and should contribute to the growing literature on the
functions and distribution of demonstratives in Vedic Sanskrit.
General abstracts, Thirteenth South Asian Languages Analysis Roundtable 49
THE WEB OF THE SPIDER:
LANGUAGE AND POWER IN SRI LANKA
Wilfrid Jayasuriya
(Southern Illinois University at Carbondale)
Language and politics became almost inseparable from the time
of the granting of independence to Sri Lanka. The theme of this
paper is the interplay of political concepts such as sovereignty with
the nature of language as the unique characteristic of the human in
the external world.
I trace the changes in the standing of the three languages,
English, Tamil, and Sinhalese, from 1830 to the present day and
indicate the social and political forces behind these changes. I in-
dicate the points of comparison and contrast between the 'Aryan'
Sinhalese language and the 'Dravidian' Tamil language in Sri Lanka
and dwell on the lack of bilingualism in pre-colonial and post-
colonial times. I indicate the nature of bilingualism in colonial times
and discuss abortive efforts at bilingualism in the eighties. How
would linguistics be able to help promote bilingualism if the state
sponsors it?
This raises the issue of the role of English in colonial times and
its changed role in the post-colonial era. Does recognizing English as
a 'link language' in the constitution make it more effective as a civil-
izing agency? I discuss the ideas that have been canvassed about the
role of English in a divided polity.
50 Studies in the Linguistic Sciences 20:3 (Spring 1991)
SIBILANT CONFUSION IN EARLY INDIC:
SANSKRIT prddiir
Brian Joseph
(Ohio State University)
The Sanskrit adverb prddiir 'forth to view, forth, in sight'
(Atharva Veda and later), restricted to co-occurrence with the roots
as- or bhu- in the meaning 'become manifest; be visible' or kr- in the
sense 'make manifest; reveal', is without a satisfactory etymology,
despite some discussion in the literature. Two suggestions have been
made that are worthy of serious consideration, that of Monier-
Williams and MacDonnell connecting prddiir with dvdr- 'door', and
that of Mayrhofer connecting it ultimately with prdtar 'in the early
morning; tomorrow'.
Both etymologies, however, are a bit problematic. In Monier-
Williams' account, prddiir, as an isolated derivative of 'door', would
have to show the same specifically Indo-Aryan — and thus relatively
late — alteration of the initial *dh- of the Indo-European proto-form
(cf. Greek thiird 'door', English door, Latin fores, all pointing to *dh-)
to an initial d-, a reshaping usually attributed to influence from dvd
'two' (so Pokorny and Mayrhofer), that is shown by the base noun
'door', even though isolated extraparadigmatic forms typically retain
older shapes of morphs. In Mayrhofer's view, on the other hand,
prddiir is to be connected with Pali pdtu(r)- 'in sight; evident', a form
which can presuppose an early Indie prdtur, and ultimately therefore
with Sanskrit prdtar. For this analysis to work, it must be assumed
that prddiir entered Sanskrit as a dialect borrowing, from a vernacu-
lar form in which the original *-t- was weakened to -d-, even though
the evidence regarding the outcome of intervocalic t at an early stage
in the Prakrits points to a variety of treatments (cf. Pischel for de-
tails) and the development of intervocalic t is far from uniform in all
the Prakrits. Moreover, the semantic connection with prdtar is not so
strong as to be completely convincing.
Thus although both of the proposed etymologies have some
attractive aspects, neither is wholly satisfactory. An alternative
General abstracts, Thirteenth South Asian Languages Analysis Roundtable 51
etymology is given here, one that takes on significance when viewed
against the backdrop of sibilant 'confusion' (i.e. interchanges), especi-
ally involving / and s and especially adjacent to sonorant, as in later
Sanskrit root-variants such as sriv-ls'riv- 'fail' and srams-l s'rams-
'loosen, let hang', or by-forms such as syala-lsydla- 'wife's brother',
and — more important — as in Vedic sruvat- 'dissolving, melting', a
variant of sruvat-, in RV 1.127.3.
In particular, I propose that the well-attested Indo-European
root *derlc- 'see' is the source for prddiir, from a preform *prd-drlc.
The semantics of the preverb pra- with the Sanskrit root drs'-,
'become visible; be seen; appear', argues for this etymology, as does
the existence of other old adverbs from the same root combined with
a preverb (e.g. Greek hupodra '(looking) sternly, grimly' from *upo-
drlc). This etymology requires only the assumption that word-final
*-rlc# somehow yielded Sanskrit -ur#. The phonetic similarity of the
Sanskrit sibilants (shown by sporadic assimilations involving sibilant
sequences, e.g. sasa- 'hare' from earlier *sas-a- or suska- 'dry' from
earlier *sus-ka) and the sibilant confusion phenomenon together pro-
vide a mans by which a final sequence *-rs in *prddrs could have
become *'rs; -ur would have developed regularly from *-rs, as in the
genitive singular of kinship terms, e.g. mdtur 'mother's' from *mdtr-s.
This account, if correct, has important consequences for the
question of the sociolinguistics of the Old Indie dialects in the Vedic
period, since it provides another early example (along with sruvat/
sruvat-) of sibilant interchanges of the sort which are found later in
Sanskrit in great numbers (e.g. srams-l s'rams-) and which are almost
surely sociolinguistically induced.
52 Studies in the Linguistic Sciences 20:3 (Spring 1991)
ADVANCEMENT IN SOME ASIAN AND AFRICAN LANGUAGES
Nkonko M. Kamwangamalu
(National University of Singapore)
Advancement, a rule which promotes a nominal bearing a given
grammatical relation in a clause to a higher relation in the same
clause (Perlmutter 1983), has been one of the central themes in
Relational Grammar (RG) for the past twenty or so years. In RG,
examples of advancement include such traditional rules as dative
movement, raising, and passive. This paper discusses advancement
of dative nominals and locatives in passive constructions in some
South Asian and African languages, with a focus on Hindi and Ciluba.
In particular, the paper is concerned with the RG claim that 'the
Relational Network of every passive clause in any human language
has a nominal bearing the 2-relation and the 1 -relation in successive
strata' (Perlmutter & Postal 1983). Data are presented which not
only challenge this claim but also have far-reaching implications for
relational laws resulting therefrom, viz. the Agreement Law, the
Chomeur Law, and the Stratal Uniqueness Law. The implications of
the data for relational concepts such as 'Terms' and 'Chomeur' are
also discussed. It will be suggested that the theory modify its claim,
laws and concepts so as to accommodate the data presented here and
elsewhere in the literature on South Asian (e.g. Mohanan 1990) and
African languages (e.g. Bresnan and Mchombo 1989).
General abstracts, Thirteenth South Asian Languages Analysis Roundtable 53
MULTILINGUALISM AND SOCIAL IDENTITY:
THE CASE OF SINGAPORE
Nkonko M. Kamwangamalu
(National University of Singapore)
Within the Acts of Identity framework (Le Page & Tabouret-
Keller 1985), speech acts are seen as acts of projection. Through
language speaker projects their identity, their inner universe, and
shape it according to the behavioral patterns of the groups with
which they wish to identify. Drawing on this framework a
questionnaire was designed to determine how multilingual speakers
in the Singapore context express their social identity through
language; how they relate to their languages; and how they perceive
various English accents to which they are exposed. In this paper the
results of this questionnaire are discussed. It is found that speakers
in the Singapore context express their social identities through a
number of linguistic means, ranging from Singapore English continua
(acrolect, mesolect, basilect), through ethnic languages (e.g. Malay,
Chinese, Tamil), to language contact phenomena such as code-mixing
and code-switching. Also, it is found that despite the high prestige
associated with British and American English accents, multilingual
Singaporians relate more to local Singapore English accent than to the
former accents, since the local accent projects their identity as Singa-
poreans.
54 Studies in the Linguistic Sciences 20:3 (Spring 1991)
ANALOGY AS ARGUMENT IN ADI ^ANKARA
Kapil Kapoor
(Jawaharlal Nehru University)
Adi Sarikara comes at the end of a long tradition of text-
interpretation (artha-nirdharana) that formally began with Yaska
(9th century B.C.) and culminated in Kumarila Bhatta (6th century
A.D.), and in the course of which a rigorous system of interpretation
(sastra paddhati) developed which was common to the three major
contending Schools (sampradaya) — the Brahmin, the Buddhist, and
the Jaina — and which came to define India as an interpretive
community. This paper, while examining Adi Sahkara's exegetical
practice as a superb representation of this shared system, centers
chiefly on one of the major modes of interpretation in Sarikara — the
analogy as metarule. The textual indices where Sarikara invokes
analogy are first identified and then followed up with a listing, a
typology, and a structural analysis of Sarikara's similes. Their role
and place in the total argument, relative to the other nine instru-
ments of interpretation in the Paddhati — verbal testimony, meta-
assumption, coherence, metarules, grammar, etymology, four levels
of meaning, and verbal symbolism — is next examined and defined.
The object text for this analysis will be Sarikara's celebrated
Vivekacudamani, though supportive materials will also be drawn
from the Sarikara-Bhasya and some of his upanisad-commentaries.
Finally, Sarikara's use of analogy as argument will be placed in the
context of India's traditionally preeminent epistemological status of
analogical reasoning since the Rg-Veda itself.
General abstracts, Thirteenth South Asian Languages Analysis Roundtable 55
REFLEXIVE PRONOUNS IN VEDIC
Madelyn J. Kissock
(Harvard University)
In this paper, I explore the syntactic and semantic features of
reflexive pronouns in Vedic Sanskrit. The focus will be on the pos-
sessive reflexive svd and the indeclinable svaydm. Using data col-
lected from the Rg-Veda, as well as from earlier prose (Taittiriya and
MaitrayanT Saihhitas), I examine the antecedence relationships of
these anaphors in light of current binding theory, while keeping in
mind the traditional accounts of Delbriick, Wackernagel, and others.
(Delbriick (1888), for example, notes that both svd and svaydm may
be coreferent with substantives other than those which occur in sub-
ject position.) The paper also focuses on the particular properties of
svaydm which appears to have two possible functions — a simple
emphatic or a true reflexive substantive which can occupy an
argument position. For these functions, questions of word order and
movement (that of extraposition site) are addressed.
An analysis of this type should provide enough comparative
data to support my claim from an earlier paper that tanii is not a
reflexive, contrary to the assertion made in most, if not all, scholarly
works including Delbriick, Mayrhofer, Grassmann, and Wackernagel.
As a related topic, the functional purpose of reflexives with regard to
active/middle voice distinctions is discussed. In addition, possible
argument structure differences between 'benefactive' and 'passive'
middles are described.
I hope that the conclusions drawn in this paper will provide
some further insights into Vedic sentence structure and the function
of reflexives.
56 Studies in the Linguistic Sciences 20:3 (Spring 1991)
ISSUES IN TRANSLATING TAMIL PURANAS
John A. Loud
(University of Wisconsin at Madison)
In this paper, I present some of the problems I have en-
countered in translating Medieval Tamil Puranas as well as the
solutions for them that I have come up with. I focus my discussion
both on general aspects of form, style, and tone, and on specific
problems that are particular to the Tamil language and the genre of
the Purana. The following verse from the Koyil Puranam of Umapati
Civacariyar illustrates some of these issues.
The great young sage walked through the forest
where the ceaseless recitation of the four ancient Vedas
resounded like the ocean. He passed by marshes
of kuvalai flowers whose scent is spread by young bees
that eat the dripping honey. He saw a tank,
which removes all bondages, full of golden lotus flowers.
He praised the tank with his mind refreshed
as tears poured forth from his eyes.
One problem is that of format. The original text is in verse and
full of elaborate descriptions. The above translation is in strict prose
order with normal punctuation. However, I have set it in the form of
an eight line stanza to break up the long and flowery sentences into
shorter units that are easier to follow. Also, the use of the stanza
form lends a rhythmic structure to the translation that makes it
seem more like verse, while maintaining the flow of the narrative.
In this way I hope to combine the artistry of poetry with the clarity
of prose.
Another consideration is to represent the philosophical orienta-
tion of the author, who in this case was a major theologian of the
Saiva Siddhanta school.
I discuss these and other problems in detail in the paper, such
as knowing who did what and with which to whom.
General abstracts, Thirteenth South Asian Languages Analysis Roundtable 57
AGAINST WH-MOVEMENT IN HINDI
Anoop Mahajan
(University of Wisconsin at Madison)
It is well known that Hindi does not have S-Structure wh-
movement (cf. Davison 1986; Mahajan 1987; Srivastava 1989). The
question then arises whether Hindi has LF wh-movement of the sort
suggested for Chinese and Japanese (Huang 1982; Lasnik & Saito
1984). In this paper, I suggest that Hindi does not have wh-
movement to SPEC of CP at any level of representation. I argue that
wh-in-situ in Hindi (and possibly in general) is treated like a clause-
bound quantifier at LF and simply undergoes quantifier raising. I
further argue that QR and LF is clause-bound. This general property
of QR along with the assumption that Hindi does not have any move-
ment to SPEC of CP yields a variety of consequences for the syntax of
wh-questions in Hindi.
One of the consequences of the theory that is outlined above is
that it automatically explains the ungrammaticality of (1):
(1) *raam-ne socaa ki mohan-ne kis-ko dekhaa
Ram-erg thought that Mohan-erg who saw
Intended as: 'Who did Ram think that Mohan saw?'
If QR is local then at LF the wh-phrase in the embedded clause will
be trapped in the lower clause. This will lead to a selectional vio-
lation since the embedded comp which is -wh will govern a wh-
phrase. However, if the wh-phrase is moved out of the embedded
clause at S-structure, the resulting sentence is no longer ill-formed
(cf. Gurtu 1985):
(2) kis-ko raam-ne socaa ki mohan-ne dekhaa?
This result is yielded under my analysis because S-Structure
scrambling of the wh-phrase (cf. Bains 1988; Mahajan 1990) has
moved it to a position from which it can QR to adjoin to the matrix IP
at LF. This theory then does not require wh-movement at S-struc-
58 Studies in the Linguistic Sciences 20:3 (Spring 1991)
ture for Hindi as suggested by Gurtu (1985). I discuss a number of
other factors concerning multiple extractions that support this view.
I also suggest a new approach to /:>'aa-questions explored in
earlier studies such as Davison 1986, Mahajan 1987, and Srivastava
1989. I suggest that the subordinate clause in a Hindi ^yaa-question
is base-generated adjoined to IP. The kyaa- expletive (associated
with this base-generated adjoined clause) is generated as a comple-
ment to the verb. This particle undergoes QR at LF to adjoin to IP. I
suggest that this expletive is replaced at LF by the associated clause.
This gives us a configuration in which the wh-phrase is governed by
the matrix -t-wh COMP. This approach to kyaa- questions also yields a
number of interesting consequences in Hindi syntax, some which are
explored in this paper.
On the basis of the framework developed in this paper, I explore
the question of whether this approach can be extended to other
languages with wh-in-situ. I provide some evidence that wh- in-situ
must in fact be treated in a unified manner in all languages.
General abstracts. Thirteenth South Asian Languages Analysis Roundtable 59
SANSKRIT REDUPLICATION:
A TEMPLATIC APPROACH
Gyanam Mahajan
(Brandeis University)
Recent studies of reduplication processes suggest that reduplica-
tion involves the affixation of a templatic prosodic unit to the stem as
a reduplicant. This reduplicant derives its segmental melody from
the stem through a process of directional templatic association. The
Sanskrit intensive reduplication provides an interesting case of what
seems to look like discontinuous association. If directional associa-
tion to a template is assumed to be free of specific rules and condi-
tions, then this instance of discontinuous association poses serious
problems for a templatic approach to reduplication. It would appear
that there is no way of predicting what the reduplicant would be.
However, this paper argues that it should not be analyzed as a case
of discontinuous association but rather another instance of Edge-In
directionality at work, not very different from other instances where
Edge-In directionality is used (for example in the templatic morpho-
logy of Arabic, cf. Yip 1988).
Sanskrit intensives are formed by adding a prefix to the root.
The prefix seems to be a reduplicating affix since it is based on the
root it attaches to and derives its segmental melody based on it. This
reduplicant is monosyllabic and is minimally heavy and thus bi-
moraic in my terms.
The intensive form of a verb like krand 'cry' in Sanskrit is kan-
i-krand and not *kran-i-krand. Similarly, the reduplicated intensive
form of a verb like kriid is kai-kriid and not *kar-kriid. Steriade
(1988) proposes an account for these kind of facts for partial redup-
lication followed by processes of pruning of sorts to derive the re-
duced form of the reduplicant intensive prefix. Full reduplication
copies both the segmental information and the prosodic structure
that goes along with it. Assuming all the other phonological rules to
be the way Steriade proposes them, my proposal is to derive these
60 Studies in the Linguistic Sciences 20:3 (Spring 1991)
forms within a templatic association approach using Edge-In while
satisfying the template.
Specifically, I suggest the use of a bimoraic template for the
intensive reduplicant in Sanskrit. Association to the template takes
place with Edge-In directionality and is assumed to be free. The
kind of Edge-In that I use is left to right at the left edge and right to
left at the right edge and then left dominant. The aim is to claim that
Sanskrit reduplication can be handled elegantly within a templatic
association approach and that a full reduplication approach is
unnecessary and in fact undesirable, since it has to resort to specific
rules and conditions to first create a form that is then severely
pruned.
General abstracts, Thirteenth South Asian Languages Analysis Roundtable 61
MEET ME IN THE BAZAAR:
A HISTORICAL PERSPECTIVE
ON THE ORIGIN OF A NORTH INDIAN KOINE
Patrick E. Marlow
(University of Illinois at Urbana-Champaign)
Bazaar Hindustani is a contact language in wide currency
throughout North India. It has been described by various authors as
a pidgin or pidginized variety of Hindi-Urdu. This claim seems to be
based on the fact that Bazaar Hindustani shows a noticeable degree
of structural 'simplification' and that it is a language of interlingual
contact and trade. The term pidgin, however, carries with it certain
connotations of power relations that these same authors recognize as
inapplicable to BH. The result is a call for the re-definition of pidgin
and a neglect of koine.
Previous studies have been unable to properly 'label' Bazaar
Hindustani as a koine because they have been too narrow in their
scope. Earlier authors have been largely concerned with the task of
describing individual geographical varieties, thus neglecting both the
supra-regional character of Bazaar Hindustani and its historical
depth.
In this paper I attempt to widen the scope of the study, both
geographically and historically. I begin by briefly comparing the
varieties of Bazaar Hindustani spoken in Calcutta and Bombay to
show that a single-origin account is necessary for these (and
presumably all) varieties. Then, looking at British accounts of
language and language use in North India from the late 18th and
early 19th centuries, I show that Bazaar Hindustani most likely
originated as a koine of the Moghul Empire and competing kingdoms.
62 Studies in the Linguistic Sciences 20:3 (Spring 1991)
SOCIOLINGUISTICS OF VERBAL ABUSE IN HINDI
Raja Ram Mehrotra
(Banaras Hindu University)
This paper attempts a sociolinguistic description and analysis of
the language of abuse in Hindi, which has not received adequate
scholarly attention to date. The institutionalized verbal behavior in
the context of abuse has been analyzed both formally and func-
tionally. It is examined with reference to both use and user. Among
various types of popular abuse in Hindi, sex abuse, kin term abuse,
animal abuse (particularly in the light of Leach's formulations), and
metaphorical abuse receive special attention. The words of abuse are
employed for a variety of purposes, such as insult, curse, contempt,
challenge, and endearment. They are used to give vent to one's
anger, irritation, annoyance — a kind of safety valve. The ritualistic
uses of verbal abuse during wedding feasts and festivals like Holi are
examined in the context of culture. Terms of abuse have also been
classified according to usef, such as those restricted to men, women,
children, and members of specific occupations and trades. Register
mixing and code-mixing in the context of personal abuse assume
special sociolinguistic significance. Certain proverbs centering a-
round words of abuse are recalled in the context of culture. It is pos-
tulated that the range of Hindi verbal abuse is wide and its pattern
complex and of considerable sociolinguistic relevance. It is marked
by a remarkable degree of convergence between culturally regulated
patterns of behavior — both linguistic and non-linguistic.
General abstracts, Thirteenth South Asian Languages Analysis Roundtable 63
TAMIL VERB STEM FORMATION
A. G. Menon
(University of Leiden)
The suffixes -kk-, -pp-, -k-, and -p- found in infinitives such as
mara-kk-a, mara-pp-a, nir-p-a, and nir-k-a have led to various ex-
planations regarding their origin and description. When compared
with infinitives such as Jr-a, var-a, kdn-a, and pec-a, in which the
above-mentioned suffixes are absent, the presence of such suffixes,
synchronically, does not add any extra meaning to these forms.
Therefore they are described as stem formatives', 'extension suf-
fixes', and 'augments'. Earlier discussions which were limited to Ta-
mil, concentrated on the question of whether stems with these suf-
fixes (e.g. poku 'to go') are earlier than the stems without these
suffixes (e.g. po 'to go'). Those who believed that the extended stems
are original, explained that the stem dk- 'to become' has changed to
a- by the loss of k. This change is correlated to a historic trans-
formation during which an uriccol developed into a vinaiccol. As far
as the suffixes are concerned, it is said that the suffix -pp- is earlier
than -kk-.
With the advent of comparative Dravidian and the diachronic
study of Tamil, a new dimension is added to the solution of the above
problem. Emeneau's Studies in Dravidian verb stem formation (JAGS
95, 1975) and P. S. Subrahmanyam's Dravidian Verb Morphology
(1971) provided more empirical data and brilliant analyses of the
verb stems and the correlations between suffixes and their mean-
ings. Emeneau's study concentrated on the Central and North Dra-
vidian languages.
The present paper attempts to make a synchronic, diachronic,
and comparative analysis of the so-called stem formative suffixes in
Tamil and Malayalam. The verbal forms in which such suffixes are
found are analyzed in their syntactic contexts. The aim of the ana-
lysis is to find out whether there is a correlation between these
suffixes and the meaning of the stems.
64 Studies in the Linguistic Sciences 20:3 (Spring 1991)
As a member of an agglutinative language family, Tamil is rich
in stem formation. However, we have only a few studies on this
important aspect of the verb morphology. Emeneau's work (1975)
which correlates verbal suffixes with voice derivation, plural action,
motion, and personal object, sets an examples for future research in
this area of Dravidian linguistics.
General abstracts, Thirteenth South Asian Languages Analysis Roundtable 65
THE ASSOCIATIVE CASE IN MALAYALAM:
MAKING SENSE OF A CATCH-ALL CATEGORY
Rodney F. Moag
(University of Texas at Austin)
The paper begins by reviewing both the variety of nomenclature
for the grammatical case in Malayalam herein referred to as
'associative', and the inadequacy of the descriptions attempted by
both traditional and modern scholars. Those from the Paninian
(Sanskrit) tradition limited their descriptions to semantics, ignoring
syntactic features completely, while those trained in modern
linguistics focussed on syntactic features with little attention to
semantics.
I next present an organized description of the associative de-
rived from the analysis of a considerable corpus of data drawn from
both spoken and written Malayalam collected over a two-year peri-
od, both in India and in the U.S. It will be first shown that uses
divide syntactically into three categories: (i) those required by a
postposition, (ii) those selected by a full verb, and (iii) those con-
ditioned by the related noun cooccurring with the existive verb. Sec-
ondly, examples are set forth illustrating that semanto-syntactically
uses (ii) and (iii) break down into two classes: (a) a small group of
verbs which require their patient to appear in the associative in all
constructions, and (b) a larger class of verbs which select accusative
case for their patient but whose derived noun forces the patient to
appear in the associative case in a sentence in a dative construction.
I show that this demotion of case of patient fits the principles of
Comrie's 'case hierarchy', though one counterexample is shown per-
mitting either accusative or associative. I next show how class (b)
breaks down into subclasses along semantic lines.
Several semantic minimal pairs are next presented showing con-
trasts between the use of the associative and other cases — contrasts
heretofore unreported in either the scholarly or pedagogical litera-
tures. It is then pointed out that the analysis presented here leads to
the conclusion that the traditional scholars were more correct than
66 Studies in the Linguistic Sciences 20:3 (Spring 199 1 )
has been acknowledged in taking semantics as the primary organ-
izing feature of the associative cases in Malayalam, since it also plays
a role in defining subclasses. On the other hand, a careful examina-
tion of the syntactic constraints is equally essential to a fuller under-
standing of this case and its place in the overall case system of
Malayalam and other Dravidian languages.
Finally, the areal aspects of this case are explored by pointing
out similarities between the uses outlined for the associative case in
Malayalam, representing the Dravidian family, and comparable ex-
amples from Hindi-Urdu, representing the Indo-Aryan language
family. It is suggested that, though there are peripheral usages of
the postposition se in Hindi-Urdu and the associative ending -ooTu in
Malayalam which differ from each other, there is a large core of com-
mon or shared usages which render this seemingly unorganized body
of interacting semantic and syntactic features a required addition to
the features defining South Asia as a linguistic area.
General abstracts, Thirteenth South Asian Languages Analysis Roundtable 67
COMPOUND VERBS IN ORIYA
Gopabandhu Mohanty
(Deccan College)
Although a lot has been brought to our attention in the last two
decades that the normally available inflectional and derivational
categories are just not enough to accommodate the wide range of
grammatical meaning that the phenomenon of verb concatenation in
Indian languages encodes, a little is really available as refined
material to be stored and used for the purpose of a data base. As a
matter of fact, most analyses do not exploit this proliferation of the
morpho-syntactic categories which generally serve as a vehicle for
the extension of the grammatical and lexical resources. On the
contrary, the ad-hoc taxonomy sometimes employed for document-
ation and enumeration of such facts (e.g. Hook 1974 and many of his
followers) show minimal application for cross-linguistic data. One
such example (appearing in the literature) is the motivation 'to do
away with lexical semantics and properties of secondary verbs' (also
known as auxiliary verbs and normally occurring at V2 position), by
assuming that semantic delimitation is an impossible task (counter to
the early attempts of Kachru 1965, Cardona 1965, and many others,
for example).
Another such example is the biased convention of accepting the
participle-V2 sequence (the non-conjunct type) as the only member
for true compounding, an analysis which has neither contributed to
the understanding of the function of the invariable semantics of the
participle, nor the syntax and semantics of the other type of se-
quencing and their roles in compound formation.
The lack of exploration of the lexical semantic has sometimes
forced the analysts to draw upon pragmatic explanations when
different circumstantial meanings are in competition due to the
connective functions of a single secondary verb. Questions of lexical
characteristics of V2 and their 'Infl-ish' character are raised again
and again — for example by Steever (1983) and Dasgupta (1989, an
updated version of 1977 work). This paper aims at reviewing the
68 Studies in the Linguistic Sciences 20:3 (Spring 1 99 1 )
available literature with a view to correct the old account, in addition
to reviewing and mapping the old analysis of compound verbs in
Oriya (an Indie language of Eastern India). Although this list-re-
ported language does not have a large set of verbs occurring in V2
position, the wide range of compound types that it possesses are of
great interest to typologists. Attempts are also made to relate the
quotative (finite verb sequences, also labeled Serial Verbs in Steever
1989) to the prototypical compound verbs by reexamining the
sequences 'V-infinitive/gerund -i- verb', "V-participle/Advp -i- verb',
and other types of sequencing in the light of grammaticalization and
degrammaticalization of V2. Along with contributing to the meaning
of the infinitive/gerund or participle construction this also is expect-
ed to set up the constituent classes for Oriya compound verbs.
General abstracts. Thirteenth South Asian Languages Analysis Roundtable 69
PRONOUNS IN KANNADA:
SOCIOLINGUISTIC IMPLICATIONS
Jayashree Nadahalli
(New York University)
Since the study of pronouns by R. Brown and A. Oilman, many
sociolinguists have analyzed deixis in different societies in various
contexts. The purpose of this paper is to make a comparative and
contrastive study of pronouns in two dialects of Kannada, a major
Dravidian language spoken in South India. The dialects chosen are:
the dialect spoken in Bangalore, the capital city of Karnataka State
(hereafter Dialect B), and the Havyaka dialect (hereafter Dialect H)
mainly spoken in Shimoga District. Dialect B represents highly
educated society and people in different occupations. On the other
hand. Dialect H represents a simple phase of the society, the main
occupation of whose speakers is cultivation.
The present study shows that sex predicts form of address in
Dialect H, whereas this is not true for Dialect B. For example. Dialect
H lacks a third person feminine pronoun. Moreover, the analysis
reveals sociolinguistic changes of pronouns in Dialect B, relating them
to the fact that in this society, solidarity and informality have been
giving way to power and formality outside the family, while the
reverse holds true within the family. For example, youngsters are no
longer addressed in the second person singular by elders if they are
not familiar to them. In the past, wives never used the second
person singular to address their husbands, but now the picture is
comparatively relaxed. Dialect H has retained its simple stratification
even though speakers are exposed to urbanism.
70 Studies in the Linguistic Sciences 20:3 (Spring 1991)
ON LIBERATING ENGLISH TO BE A WORLD LANGUAGE:
AN INDIAN PERSPECTIVE
Mangesh V. Nadkarni
(National University of Singapore)
The emergence of English as a world language is in many ways a
new phenomenon in human history. We therefore often find a tend-
ency to restrict the form and notion of world-language English to cer-
tain outmoded concepts about a world language. English needs to be
'liberated' from these restrictions. In this paper the problem of lib-
erating English will be discussed at two levels — at the level of the
cultural resonance of the language and then at the most surface level
of its expression, namely the spoken form.
It has been generally accepted that the English language is
valuable to India because it is a means of promoting political, eco-
nomic, and technological interests in the modern world by making
available the magnificent centuries of the mind. I argue in this
paper that English is valuable to India also because it renders possib-
le the most magnificent expression of the soul of India, which is as
multifaceted because of its synthetical and assimilative genius as the
English language is in its capacity to express multifarious types of
consciousness. But for this to happen effectively, the English lan-
guage will have to be Indianized not so much in its purely linguistic
apparatus but in its inner vibration. It will have to be capable of
achieving what ancient Sanskrit was able to achieve as the language
of the Vedas, the Upanishads, and the Gita, by acquiring the potenti-
alities appropriate to the magic, mystery, the depth, and the sudden
and revelatory reach of the spiritual consciousness.
Secondly, it will be argued that if English is to be a world lan-
guage in the true sense of this term, we will have to let it develop
the way Chinese has — to be accessible to all educated speakers of
the language in the written form but not necessarily in the spoken
form. The need for internationally intelligible spoken models of
English for countries in South Asia is often exaggerated. English is
needed more for intranational than for international communication
General abstracts, Thirteenth South Asian Languages Analysis Roundtable 7 1
in these countries. Since English is being learned through formal
schooling by speakers of many diverse languages and often in very
difficult circumstances all over the world today, it is unrealistic to
insist on internationally intelligible spoken English as a feature of
adequate competence in English. This is a level of attainment best
left to a small group of 'brokers'. Furthermore, English should prove
hospitable to the cultural aspirations of the people using it so that
their need to feel a certain closeness to it even in the spoken form is
not frustrated.
72 Studies in the Linguistic Sciences 20:3 (Spring 1991)
ARTICULATORY AND ACOUSTIC PROPERTIES OF APICAL AND
LAMINAL STOP CONSONANTS:
A CROSS-LANGUAGE STUDY
Paroo Nihalani
(National University of Singapore)
Chomsky and Halle (1968) have hypothesized that laminal artic-
ulations have a larger constriction than apical consonants. This paper
explores the question of what exactly is meant articulatorily by
apical and laminal articulation in Indian languages and then exam-
ines the corresponding acoustic differences. In the second part, an
attempt is made to determine whether these differing articulatory
strategies are consistent within a language or vary according to
speaker-specific idiosyncrasies.
General abstracts. Thirteenth South Asian Languages Analysis Roundtable 73
A SOCIO-COGNITIVE APPROACH TO DESIGNING A SELF-
INSTRUCTIONAL MULTI-MEDIA COURSE IN ENGLISH
COMMUNICATIVE SKILLS
P. N. Pandit
(Indira Gandhi National Open University)
Indira Gandhi National Open University, New Delhi, disseminates
education by the distance mode, i.e. by correspondence, printed
materials, and audio-video cassettes. This paper studies the various
considerations that were made by the team of course writers in
designing the course in English communicative skills. Indians need
to learn English as a second language to fulfill certain socio-cognitive
needs — for communication as a link in a multi-lingual background,
as a language of advanced studies, research, and library reference,
and for professional purposes in science and technology and in
business. The recent advent of information technology in the form of
teletext and compunication (computer communication) has made it
necessary for Indians in urban settings to be properly cognizant of
this.
74 Studies in the Linguistic Sciences 20:3 (Spring 1991)
FUNCTIONAL ARTICULATION:
ANALYZING DIGLOSSIC VARIATION
John C. Paolillo
(Stanford University)
Diglossia is a widespread characteristic of South Asian languages:
Many South Asian languages, both past and present, show either
strong diglossic tendencies, or long histories of diglossia. South Asia
is also a rich example of a multilingual nation, a sociolinguistic 'type'
of situation often compared with diglossia. Thus, South Asia provides
a perfect opportunity to explore the nature of diglossia, and its rela-
tion to multilingualism.
Grammatical description of diglossic (and multi-lingual) situa-
tions has been limited to separately describing the varieties in-
volved, or cataloguing a list of differences between H and L. Some
recent work, notably Britto 1985, recognizes that diglossic situations
are usually much more complex than is suggested by the H and L
labels: Individual grammatical features may be manipulated separ-
ately to produce particular communicative effects, particularly in
functionally intermediate contexts. Britto proposes a notion of
'mutability' to account for such features, an H grammatical feature
that is used in some L contexts is 'mutable to L', while an H feature
used only in H contexts is 'immutable'.
Britto's approach, however, fails to characterize correctly the
language of functionally intermediate contexts, particularly in situa-
tions such as Calcutta Bengali (Chatterji 1986), where a minimum of
two dimensions of functional and grammatical variation can be ob-
served. In Bengali, lexical choice (Sanskritic vs. non-Sanskritic) cor-
responds to a dimension of formal vs. informal contexts, and mor-
phophonemics (Sadhu vs. Cholit morphophonemics) corresponds to a
dimension of educational vs. non-educational contexts, yielding a
total of four grammatically distinct functional ranges. Britto's muta-
bility, with only one dimension of variation between H and L, is in-
capable of characterizing this situation.
General abstracts. Thirteenth South Asian Languages Analysis Roundtable 75
If we resort to the traditional approach of independently
characterizing the grammar of each recognizable variety, we are
faced with another problem: The number of varieties to be charac-
terized increases exponentially with each feature to be considered. A
hypothetical diglossia with eight mutable features would require 64
different grammars.
I propose here an alternative to both the traditional approach
and Britto's approach to grammatical description in diglossias,
illustrated with examples from two South Asian languages, Bengali
and Sinhala. This approach, which I call 'functional articulation', re-
quires all grammatical dimensions of variation in a diglossia to be
encompassed within a single grammar. The individual grammatical
features that participate in the diglossic variation are associated with
distinct functional ranges (they are 'functionally articulated') by
associating them with different 'communicative attitudes', compon-
ents of meaning which refer to contextual parameters such as speak-
er, hearer, audience, message, etc. Examples of communicative
attitudes are 'public' (intended to reflect immediate personal in-
volvement of the speaker), and 'edited' (intended to reflect prior
consideration of the speaker). Different contexts select for different
attitudes according to social norms so that, for example, in a public
speaking situation, grammatical features possessing 'public' and
'interactive' communicative attitudes would be selected.
Thus, functional and grammatical description are integrated in
this approach for a richer understanding of diglossic variation than is
possible with either the traditional approach, or with Britto's notion
of 'mutability'. Using communicative attitudes that refer to group
membership allows one to make an account of multilingual situations
where the different languages share a great deal of structure (e.g.
Hindi and Panjabi). Thus, with functional articulation, it is possible to
cast an account of the differences between diglossia and multi-
lingualism, two important types of language situations in the
linguistic environment of South Asia.
REFERENCES
Britto, Francis. 1986. Diglossia: a study of the theory with ap-
plication to Tamil. Washington, DC: Georgetown University Press.
76 Studies in the Linguistic Sciences 20:3 (Spring 1991)
Chatterji, Suhas. 1986. Diglossia in Bengali. South Asian languages:
Structure, convergence, and diglossia, ed. by Bh. Krishnamurti et
al. Delhi: Motilal Banarsidas.
General abstracts. Thirteenth South Asian Languages Analysis Roundtable 77
TELUGU NEGATIVES AND NON-CAPABILITATIVES:
MORPHOLOGICAL STRUCTURE AND SYNTACTIC STRUCTURE
Rosanne Pelletier
(Yale University)
This paper analyzes the syntactic structure properties of two
negative constructions in the Dravidian language Telugu. I demon-
strate the non-concatenative fit between the levels of morphological
structure and syntactic structure and give a historical account of the
rather quirky syntactic structure facts. In (A) we see that the nega-
tive morpheme lee- is used in both the non-capabilitative and the
past negative constructions. Morphologically (1) and (2) differ in
that while the non-capabilitative lee- agrees with the subject nuw-
wu, the past negative lee- does not. Both negative constructions are
mono-clausal; the difference in agreement stems from the different
morphological structures of the two negative constructions. Since the
target of agreement is the first predicative word to the right, it is
clear that the components of the complex predicate ammalee- 'cannot
sell' in (1) are treated as a single unit in terms of morphological
structure. On the other hand, amma and lee- in (2) do not constitute
such a predicative word; the agreement target in (2) is amma, which
as an infinitive, cannot carry agreement features; leedu carries de-
fault features.
While in terms of morphological structure, the non-capabilita-
tive and the past negative constructions differ from each other, in
terms of syntactic structure the two constructions behave exactly
alike. For example, the emphatic clitic -ee can be attached only to
the highest projection of a lexical category, as we see in (3a) and
(3b). This cliticization pattern, as well as the lack of a VP node, are
two syntactic-structure features which Telugu shares with its sister
language Malayalam (Mohanan 1982). In (4a) and (4b), then, we see
that -ee can be attached to both amma and leedu, which is hardly
surprising in view of their behavior as separate predicate units with
respect to agreement.
78 Studies in the Linguistic Sciences 20:3 (Spring 1991)
The facts in (5) show that the components of the morphologi-
cally complex non-capabilitative amma-lee- 'cannot sell' can also
both be cliticized, indicating their status as separate syntactic con-
stituents. Thus the elements of the non-capabilitative construction,
while unified in morphological structure, are discrete in syntactic
structure. The disparate structures of the non-capabilitative con-
struction, while striking, are not problematic under the assumption
that morphology does not 'feed into' syntax (see e.g. Sadock 1991).
Its dual structure IS puzzling, however, when compared with the
capabilitative structure. In (6a) and (6b) we see that amma-gala-
'can sell' functions as a complex predicate with respect to both levels
of structure. We see then that the disparate structures of amma-lee
do not constitute a random mismatch between morphological struc-
ture and syntactic structure; rather, this mismatch indicates prop-
erties specific to negative structures.
In fact, these syntactic negative constructions are an innovation
of Modern Telugu (Mahadeva Sastri 1969). The innovated negative
forms were grafted onto the Serial Verb pattern which, although ex-
tremely restricted in Modern Telugu, is shown by Steever (1987) to
be a prominent pattern in the proto-language. Based on historical
and comparative evidence, Steever gives precisely the same analysis
of negative constructions in Dravidian that syntactic structure tests
induce us to give for the Telugu forms: These are morphologically
defective Serial Verbs. This analysis predicts exactly those con-
stituent-structure facts that we find: the syntactic-structure/mor-
phological-structure divergence in the non-capabilitative, as well as
the status as discrete syntactic constituents in the non-capabilitative
but not the capabilitative.
A. NEGATIVE FORMS
(1) Non-capabilitative:
nuwwu aa pedda illu amma lee-wu
you-sg that big house sell-INF neg-2sg
'You CANNOT sell that big house.'
(2) Past negative:
nuwwu aa pedda illu amma lee-du
you-sg that big house sell-INF neg-3sg neut
'You DID NOT sell that big house.'
B. CLITICIZATION
(3) Non-syntactic constituent:
a. [aa pedda ill ]-ee
[that big house ]-EMPH
General abstracts. Thirteenth South Asian Languages Analysis Roundtable 79
amma
sell-INF
b. *[aa pedd-ee illu]
[that big-EMPH house]
(4) Past negative:
a. nuwwu aa pedda illu
you-sg that big house
'You did NOT sell that big house.'
b. nuwwu aa pedda illu amm-ee
you-sg that big house sell-INF-EMPH
'You did not SELL that big house.'
(5) Non-capabilitative:
a. nuwwu aa pedda illu
you-sg that big house
'You CANNOT sell that big house.
b. nuwwu aa pedda illu
you-sg that big house
'You cannot SELL that big house
(6) Capabilitative:
a. nuwwu aa pedda illu
you-sg that big house
'You CAN sell that big house.'
b. *nuwwu aa pedda illu
you-sg that big house
'You can SELL that big house.'
lee-d-ee
neg-3sg neut-EMPH
lee-du
neg-3sg neut
amma
sell-INF
lee-w-ee
neg-2sg-EMPH
amm-ee lee-wu
sell-INF-EMPH neg-2sg
amma gala-w-ee
sell-INF be able-2sg-EMPH
amm-ee gala-wu
sell-INF-EMPH neg-2sg
REFERENCES
Bresnan, Joan (ed.). 1982. The mental representation of grammatical
relations. Cambridge: MIT Press.
Mahadeva Sastri, Korada. 1969. Historical grammar of Telugu. Tiru-
pati: Sri Venkateswara University.
Mohanan, K. P. 1982. Grammatical relations and clause structure in
Malayalam. Bresnan 1982.
Sadock, Jerrold. 1991. Autolexical syntax: A theory of parallel gram-
matical representations. Chicago: University of Chicago Press.
Steever, Sanford. 1987. The serial verb formation in the Dravidian
Languages. Delhi: Motilal Banarsidass.
Studies in the Linguistic Sciences 20:3 (Spring 1991)
SAMMELANI HINDI AND MALVIYA HINDI:
LANGUAGE AND POLITICS IN INDIA
BETWEEN 1875 AND 1930
Alok Rai
(University of Allahabad)
The struggle for Hindi from about 1870 to 1930. Our story
starts from the 'Court Character' or Nagari controversy of the last
decades of the nineteenth century and culminates in the contentious
acceptance of Hindi as 'the national language'. This is a long and
extremely complex story. In microcosmic form, it is the story of the
transformation of the national movement — the transformation of the
Congress Party from a bunch of Anglicized memorialists waiting upon
the pleasure of the colonial masters, into the mass instrument of
Gandhi. This is a stirring history, but we are now becoming increas-
ingly sensitized to the ambivalences and the complexities of these
developments. Thus, historical phenomena are multivalent, and dif-
ferent aspects become salient in different historical periods.
Initiatives that appear bold and creative in an earlier period may be
found to be the carriers also of some of the seeds of our present
problems. Thus, at some point, through processes that will bear
thinking about, the 'democratic' struggle for Hindi modulates into the
formation/assertion of a Hindu nationality — i.e. the communalization
of politics. And this troublesome communalization is itself only a
subset of that process of the vernacularization of politics of which the
struggle for Hindi is an integral part. Further, in all this process, it
isn't only the external context of Hindi that is changing, or even
Hindi's own angle of intervention — the svarupa of the language is
itself being redefined, so that the language that Malaviya is agitating
for in the closing decades of the nineteenth century has little resem-
blance to — or important differences from — the Sammelani Hindi
that becomes triumphant in the 1930s.
General abstracts, Thirteenth South Asian Languages Analysis Roundtable 8 1
THE CATEGORY OF NOMINALS IN BANGLA
Gillian Ramchand
(Stanford University)
Bangla nominals are interesting for a variety of reasons. The
ways in which common nouns interact with plurals, definiteness and
case-marking seem to indicate a split in type between animates and
nonanimates. (I will use this terminology even though the split isn't
strictly inanimate vs. animate. It seems to be a difference between
one privileged group of animates versus all other nominals.) More-
over, it is quite difficult to articulate the effect of these morphologi-
cal affixes in the language in a way that would be coherent for both
types. The aim of the paper is to lay out some of the problems in
defining their denotations.
I discuss the properties of the animate and inanimate nouns
with respect to three main classes of phenomena: semantic interpre-
tation (i.e. possibilities of definite and/or generic meanings and
scopal interactions), the interaction with the determiner system with
respect to the above semantic properties, and case marking.
Readings for inanimate nouns cannot be definite unless they
have a 'definiteness' clitic attached to the end of the noun. In gen-
eral, bare inanimate nouns can have both indefinite and generic
meanings.
(1) a. Ami am-ta kheyechi
I-NOM mango-DEF eat-PERF PAST 1st
'I have eaten the mango.'
b. Ami am kheyeche.
I-NOM mango eat-PRES PERF 3rd
'I have eaten mangos (one or more).'
Animate nouns cannot in general take singular definitive en-
clitics. In bare form, they can have definite or indefinite readings.
82 Studies in the Linguistic Sciences 20:3 (Spring 1991)
In object position, a definite reading must be accompanied by an
accusative case marker.
(2) a. Se meye daekho
s/he-NOM girl see-PRES 3rd.
'S/he sees girls (one or more).'
b. Se meye-ke daekho
s/he-NOM girl-ACC see-PRES 3rd.
'S/he sees the girl.'
Just as the definite clitics may not appear on animates, the accusative
case-marker cannot appear on inanimates in object position.
Besides patterning differently with respect to definite markers
and accusative case, these two noun classes also have different
morphological requirements in numerical quantification, and with
respect to plural marking. But in addition to this 'morphological
split', animates and inanimates have different scopal properties with
respect to other quantifiers in a sentence. Thus, it appears that with
inanimates in the bare unmarked form, only the narrow-scope
reading is possible. With the animates, the situation is rather
different. It is possible to get both the wide-scope and the narrow-
scope readings for this kind of nominal.
Thus, there are striking differences between the meanings of
inanimates and the meanings of animates in the same contexts,
which parallel the split in morphological selection processes. The
claim one would like to make is that these differences are not
arbitrary, but reflect a fundamental difference in the denotation of
these two classes of nominal.
The problem this paper addresses is twofold: (i) How do we
understand the denotation of the nominal category in Bangla, so that
the possible semantic readings and their interaction with the so-
called determiner system can be systematically derived? (ii) How do
we represent the split in behavior between the human/animate and
nonhuman nominals in Bangla within a denotational system of this
kind?
The analysis this paper offers is to say that the animate/human
nominals are actually different from the inanimates in that they can
represent atoms in a 'count' domain (in the sense of Link 1983),
whereas the latter must be seen as representing atoms in a 'mass'
domain lattice. I present evidence for this position, and consider its
General abstracts. Thirteenth South Asian Languages Analysis Roundtable 83
implications for argument structure and case-marking in this and
related languages.
84 Studies in the Linguistic Sciences 20:3 (Spring 1991)
COORDINATION AND WORD ORDER
Nalini Rau
(University of Illinois at Urbana-Champaign)
An interesting facet of coordination in Kannada is the degree of
manifestations of agreement features on the verb. This paper
focuses on the relation between word order and agreement in
coordination in Kannada. Agreement in terms of number and person
is unaffected by word order. However, agreement in terms of neuter
and ± human features is dependent on word order. (1) and (2) are
typical of examples where the verb agrees with the noun phrase
closest to it with respect to the human feature, but where agreement
in terms of person and number are not affected by proximity to the
verb.
(1) banDiyu aaLugaLu bandaru
cart workers came
3sg neut 3pl 3pl nonneut
'Cart and workers came.'
(2) aaLugaLu banDiyu bandavu
3pl 3sg neut 3pl neut
(T. N. Srikentiah 'Kannada Madyama Vyadarana')
A GPSG analysis of the above facts is provided.
General abstracts. Thirteenth South Asian Languages Analysis Roundtable 85
THE PURE VOWELS OF PUNJABI
Kamlesh Sadanand
(University of Hyderabad)
In standard colloquial Punjabi spoken in different parts of the
State of Punjab there are ten pure oral vowel phonemes. Some of
these have the potentiality of occurring with heavy nasalization even
if there are no nasal consonants in the syllables in which they occur.
The aim of the paper is twofold:
(i) to ascertain the precise phonetic quality of each of these
vowels with the help of spectography. The KAY Elemetrics Sona-
Graph, 606 IB, is used to study the formant frequencies, particularly
the first and second formants of each of these ten pure oral vowels.
With the help of the logarithmic graph suggested by Joos (1945) and
Ladefoged (1975), the tongue positions of the vowels are ascertained.
(ii) to examine the formant frequencies of the nasalized vowels
to see if the phenomenon of nasalization has any considerable
influence on the formant frequencies of the vowels.
I also examine the degree of nasalization in the case of phonem-
ically distinct nasalized vowels and compare it with the degree of
nasalization in the case of oral vowels which are accidentally nas-
alized because of their occurrence in proximity with nasal con-
sonants. This experiment is done with the help of an electro-
aerometer connected to a mingograph.
86 Studies in the Linguistic Sciences 20:3 (Spring 1991)
MALAYALAM SYLLABIFICATION
Suchitra Sadanandan
(University of Southern California)
Syllabification in Malayalam presents an interesting problem.
On the one hand, Malayalam seems to prohibit codas but, on the
other hand, generalizations in distribution can only be captured if
codas are allowed (Tara Mohanan 1989). Tara Mohanan presents one
possible explanation for the puzzle. She argues that in a lexical
model of phonology the paradox is the result of the difference in
wellformedness conditions holding at the different levels. Such an
analysis, while being descriptively adequate, opens the door to a po-
tentially unrestrained exploitation of constraints which might result
in explanatorily inadequate grammars.
In this paper, 1 present an alternative way of analyzing syllable
structure in Malayalam, an analysis which rejects the unconstrained
interpretation of wellformedness conditions that Tara Mohanan
adopts. My analysis uses as its basis some of the insights into
syllabification that a Government Phonology approach (e.g. Jonathan
Kaye 1988) offers. The proposed analysis shows that the paradox in
Malayalam syllabification is only an apparent paradox, one that does
not exist if we distinguish between word-internal codas and word-
final ones. This distinction is not a mere stipulation, but is a con-
sequence of adopting the principle of coda licensing, a crucial part of
a government approach to syllable structure (Kaye 1989). This ap-
proach to Malayalam syllabification not only resolves the above-
mentioned paradox but also helps to advance a unified account of
diverse phonological processes in Malayalam, namely epenthesis,
stem-initial gemination, stem-final gemination, and lenition.
General abstracts. Thirteenth South Asian Languages Analysis Roundtable 87
ON CHANGE AND VARIATION OF (l) IN KANNADA
T. S. Satyanath
(University of Delhi)
This paper attempts to study the change and variation of (1) and
other related sounds in Kannada inscriptions during the period A.D.
450 to 1100. It has been claimed by Narasimiah (1941) and Gai
(1946) that intervocalically / -^ I, and in the case of consonant clust-
ers IC -> rC -> CC, eventually leading to the loss of /. However, a
closer look suggests that a series of splits and mergers took place in
the sounds /, /, r, and d during the period under consideration. The
details of the various splits and mergers are outlined in the present
study.
The data for the analysis have been drawn from a computer
database of Kannada inscriptions created on the Macintosh. The
database has been scanned for all potential and possible occurrences
of the variable. A multivariate analysis has been done using Gold
Varb 2.0 to determine the role of different linguistic environments
and social-geographical factors in the use of the variable.
Studies in the Linguistic Sciences 20:3 (Spring 1991)
ASSESSING ^ABARA'S ARGUMENTS FOR THE CONCLUSION
THAT A GENERIC TERM DENOTES JUST A CLASS PROPERTY
Peter M. Scharf
(University of Pennsylvania)
In the Akrtyadhikarana, Sahara argues that a generic term
denotes just a class property and not an individual. He argues vigor-
ously to refute the view upheld by Paninians and many Naiyayakas
that a generic term denotes both. The first of two arguments is by
concomitant presence and absence. It depends on the doubtful
example of the term syenacit (falcon altar) adduced as evidence of a
generic term whose use is followed by cognition of a class property
and not of an individual. Hence, the argument fails. The second
argument is based on the limitation composed by virtue of adopting
presumption (arthdpatti) as part of the means to determine what a
word denotes. That limitation is that if one can account for cognition
of the individual otherwise, one cannot presume that the word de-
notes it. According to Sahara, however, the manner by which a
listener arrives at cognition of an individual, in each instance of
understanding a sentence using a generic term, requires the same
condition. Hence comprehension of each sentence and determining
that a word has the capacity to denote are equally based on pre-
sumption. Hence this argument too proves faulty.
General abstracts. Thirteenth South Asian Languages Analysis Roundtable 89
THE VEDIC CLAUSE-INITIAL STRING
AND UNIVERSAL GRAMMAR
Steven Schaufele
(University of Illinois at Urbana-Champaign)
One of the distinguishing characteristics of Vedic as opposed to
later periods of Sanskrit is what Hock (1982 and elsewhere) has
called the 'clause-initial string'. This string constitutes a 'landing site'
for a variety of syntactic phenomena, including topicalization, the
placing of 'sentential particles', and the fronting of various pro-
nominals, including but not limited to clitics and 'wh-' words.
Attempts have been made (e.g. Hale 1987) to describe some of these
phenomena within an 'orthodox' version of the Revised Extended
Standard Theory based primarily on English and to some extent on
other modern Western-European languages. But it has become in-
creasingly clear that such attempts create more problems than they
solve.
In this paper I discuss these problems and show that they are
susceptible of plausible solutions if we allow our theory of Universal
Grammar to be adjusted by the consideration of data from languages
originating within the Indo-European stock but outside of modern
Western Europe. In the process, I argue that COMP is less a linguistic
universal than a coalescence of several features which some lan-
guages keep distinct; that particle placement is best treated not as a
syntactic phenomenon at all but as a local transformation at PF; and
that both lexical topicalization and lexical adjunction must be
allowed. And I show that these adjustments do much less damage to
the theory than more 'orthodox' attempts do to the data.
90 Studies in the Linguistic Sciences 20:3 (Spring 1991)
SEMIO-LINGUISTIC ASPECT OF DHVANI SIDDHANTA
Krishna K. Sharma
(University of Hyderabad)
Literature is semiotic transposition of poetic impulse grounded
in and through the language. Anandvardhana, in his celebrated work
Dhvanyaloka, deals with the methods, both linguistic and semiotic,
underlying the transfer of the linguistic structure from the
grammatical level to the semiotic level conveying the poetic message.
Anandvardhana for the first time categorically stated that the
poetic message imbued in the linguistic structure is derived with the
help of the features rendering it a semiotic importance, and this
thinking leads to the semiolinguistic perspective of the theory. This
is perhaps the least attended to, yet most significant issue of Dhvani
Siddhanta.
Anandvardhana concerned himself with poetic structure at the
level of discourse, a level much higher than that of the word or that
of the sentence, although these are vital ingredients.
The analysis of the grammatical level of the discourse is not
sufficient to obtain the poetic message. It has to be considered in the
light of features which illuminate it with semiotic value.
There are a number of notable commentaries on Dhvanyaloka,
but none has strived to bring out this significant aspect of this
masterpiece. The problem of the transfer of discourse from the low-
er level of meaning to the higher level of significance, which is the
major principle involved in the development of this theory, has not
been addressed.
The different contexts, giving rise to different grammatical
structures, color them with specific semiotic values. The choice of
words and their placing in the structure is directed and controlled by
the semiotic features. Each of the linguistic items functions in the
specific contextual milieu. In bringing out the poetic message,
General abstracts, Thirteenth South Asian Languages Analysis Roundtable 9 1
Anandvardhana lays necessary balanced weightage on both of the
dimensions, i.e. linguistic and semiotic.
In this paper, I bring out this aspect of Dhvani Siddhanta, which
has not received due consideration from the scholars of this work of
the great master.
92 Studies in the Linguistic Sciences 20:3 (Spring 1991)
NAMING AND EXPRESSING OBJECTS IN PANINI
Rama Natha Sharma
(University of Hawaii at Manoa)
The Astadhyayl of Panini, as a graminatical device, manipulates
naming and expressing as tools to denote grammatical and notional
relations between nominals on the one hand, and the action on the
other. This paper briefly outlines the main characteristics of this
device, preparatory to a detailed analysis of naming and expressing
the object (karman). An attempt is then made to explain various
types of objects as they have been discussed by the tradition. But
the basic purpose of this paper remains to deal with the nature, and
problems of representation, with possible resolutions, especially as
they relate to karman named and expressed. While no attempt is
made to compare the Paninian system of naming and expressing with
similar techniques in other formal systems, my findings may very
well form a basis for such a comparison.
General abstracts, Thirteenth South Asian Languages Analysis Roundtable 93
THE AESTHETICS OF PLAY IN PUNJABI FOLKLORIC
TRADITION
Atamjit Singh
(Guru Nanak Dev University)
The religious culture of India revels an exceptionally rich
diversity of cultural activities permeated by the spirit of play. The
play is usually associated with childhood, but a line of development
can be traced from the infantile phase, though childhood and
adolescence, to maturity. It begins as spontaneous pleasure seeking
activity of the infant. The root Ids in Sanskrit would appear to refer
to playing in its primal form; it combines the meaning of shining,
sudden appearance, sudden noise, blazing up, moving to and fro with
irrepressible playfulness, — Blakes' delight. But it has been often
noticed by the psychologists after observing young children that play
is not all pure pleasure, not the equivalent of the adult's recreation
or recovery of the repressed unconscious.
As the child grows up, the spontaneous manifestations of play
become increasingly structured. The play activity as such can be
placed within the polarity of turbulence and order or spontaneity
and contest. This metamorphosis of impulsive play into a cultural
institution, with rules, conventions, and highly trained personnel is a
FORMATIVE PROCESS. In Sanskrit, the usual term for play among
children is krida. It also refers to gaiety, hopping, skipping, or
dancing, and it approximates to the root nrt which is applied to the
whole field of dance and drama. Looked at philologically, Indian
play terms are linked with one another since ancient times.
A salient feature of Punjabi life and culture is the play element
transformed into different aesthetic and cultural configurations and
manifests itself in diverse forms in the literary and cultural life of
the people of Punjab. Some basic patterns of play in their aesthetic
forms is discussed in this paper.
94 Studies in the Linguistic Sciences 20:3 (Spring 1991)
A SITUATION-TYPE ANALYSIS OF COMPOUND VERBS
Mona Singh
(University of Texas at Austin)
Compound Verbs (CVs) have been said to have aspectual content
(Porizka 1969), but most research on them has been into their non-
aspectual functions. Recent literature on the aspectual function of
CVs has been restricted to the emergence of CVs and the statistical
analyses of CVs in CV-rich and CV-poor languages (Hook 1988).
Attempts have also been made to provide percentages of the oc-
currences of various CVs in Hindi and their correlation with their
position in the paragraph (Hook 1989).
According to extant analyses the function of CVs is merely to
EMPHASIZE the perfective aspect (Hook 1989). But this view does not
provide a convincing explanation of the distinction between
sentences (1) and (2) below, both of which have the perfective
aspect. A complete and exhaustive analysis of the aspectual function
of CVs must provide general rules that can account for the Compound
Verb : Simple Verb dichotomy.
(1) hamne kitaab paRhi
we-ERG book read-PERF
'We read the book.'
(2) hamne kitaab paRh li
we-ERG book read take-PERF
'We read the book entirely.'
This paper is an extension of work reported in Singh 1990. It pre-
sents an analysis of CVs as markers of telic situation types and of
their interaction with the perfective aspect. Sentences describing
telic events (achievements and accomplishments) obligatorily have
CVs focusing on one of the following: INITIAL ENDPOINT, FINAL END-
POINT, RESULTANT STATE and ENTIRE EVENT. Sentences describing atelic
events (activities) and states, however, can only have a simple verb.
Of primary importance here is the aspectual class or situation type of
General abstracts. Thirteenth South Asian Languages Analysis Roundtable 95
the event. Sentences with the same verb can fall into either the telic
or the atelic class. For example, sentences (1) and (2) both have the
same verb 'to read'. However, the situation type of (1) is atelic, since
it is an activity without any endpoint, while the situation type of
sentence (2) is telic, since it has a natural final endpoint. Compound
verbs are a mechanism for capturing the telic nature of the event.
In this paper, I present a simple explanation of why CVs must
be used in some sentences and not in others. The situation-type
analysis of CVs also provides a very natural explanation for the non-
acceptability of CVs in sentences expressing negation and pro-
gression. I also discuss the implication of this analysis for the prag-
matic factors associated with CVs. Briefly, the situation-type analysis
not only provides a simple rule for the semantics of CVs, it also
makes the pragmatics of CVs understandable.
REFERENCES
Hook, Peter. 1988. On the emergence of perfect aspect in Indo-
Aryan. 10th South Asian Languages and Analysis Roundtable
Madison.
1989. Determining thresholds for the emergence of perfective
aspect in Indo- Aryan languages. CLS 11:3.239-261.
Porizka, Vincec. 1969. On the perfective verbal aspect in Hindi
(concluded). Archiv Orientalni 37.345-364.
Singh, Mona. 1990. The aspectual content of compound verbs. Eas-
tern States Conference on Linguistics.
96 Studies in the Linguistic Sciences 20:3 (Spring 1991)
DRASTIC MODERNIZATION OF THE CURRICULA OF THE
TEACHER TRAINING COURSES
M. V. Sreedhar
(Institute for Socially Disadvantaged Groups)
A sizable population of the deprived children throughout the
world, including the immigrant children from third-world countries
to industrialized countries like Germany, the USA, etc. are 'out of
school' or poor scorers at the schools, the result of which is an ever-
increasing number of adult illiterates producing an inefficient work
force.
The research being conducted by this institute through three
experimental primary schools with children belonging to the nomadic
tribes, denotified tribes, and scheduled castes (formerly known as
untouchables) reveal that the teacher, who is the pivot around which
the entire education revolves, is primarily responsible for this path-
etic state-of-affairs because of deficiencies in the teacher training
curricula. The middle-class teachers have no knowledge of the social
heritage of the deprived children, which in turn prevents them from
understanding the issues pertaining to the unpreparedness of the
deprived children to meet the challenges of the middle-class biased
school when they enter at the age of 6+ and to take remedial steps.
The lack of preparedness arises out of the fact that the deprived
children, on entering the class at the age of 6+, find a discontinuity
between their home subculture and home language on one hand and
the variety used at the school on the other hand. The middle-class
biased teachers blame the children for their so-called 'deficiency' and
condemn them as uneducable. The negative attitude of the teachers
towards the deprived children induces/encourages the middle-class
children in the class to ridicule the deprived children and segregate
them. This pathetic situation can only be remedied through a
thorough overhaul of teacher training curricula and by providing in-
service training programs to existing teachers. The three major areas
where this needs to be attended to are: (i) language use in different
contexts and teaching the standard variety as an alternative variety
and not as a replacive one, (ii) taking advantage of the strong points
General abstracts. Thirteenth South Asian Languages Analysis Roundtable 97
of the deprived children, and (iii) providing the teacher trainers and
the teachers with detailed information regarding the concept of 'Self-
fulfilling prophecy' which would enable them to improve the quality
of the learning process. The three principal hypotheses of any such
attempt should be:
(i) no child is uneducable;
(ii) If the children from deprived groups fail to benefit from
the formal educational system, the fault lies with the sys-
tem and not with the children and hence the system must
change; and
(iii) the child is never at fault
The Institute is working on these hypotheses, and the result so far
has been quite encouraging.
98 Studies in the Linguistic Sciences 20:3 (Spring 1991)
A LEXICALIST ANALYSIS OF PARTICIPLE COMPOUNDS
IN KANNADA
S. N. Sridhar and Mark Aronoff
(State University of New York at Stony Brook)
Based on a detailed analysis of Modern Literary Kannada par-
ticipial constructions, we defend a classic abstract version of lexical
theory according to which syntactic phenomena (in the broadest
sense of the word syntactic) are divided between distinct domains,
with the dividing line being drawn at the level of the major lexical
category, X, rather than the surface word. On this view, all pheno-
mena that can be described entirely within a single member of a
major lexical category are lexical, while those that involve reference
to phrasal or sentential information are syntactic.
Because the division between lexical and syntactic is one of do-
mains rather than rules, there is no fundamental difference between
lexical and syntactic constructions. However, if a construction is
found in both domains, it will show distinct properties in each do-
main, due to conditions imposed by the domains themselves.
The possibility of a single construction having manifestations in
two domains permits the use of a powerful analytic technique whose
locus classicus is Wasow's 1977 article on the English passive, where-
in lexical and syntactic versions of a single rule are compared. The
advantage of this technique is that it allows the investigators to fac-
tor out the construction itself, thus allowing for a better under-
standing of the differences between the domains.
In our presentation, we use this technique to analyze sentential
and lexical versions of the participial construction in Kannada. This
construction consists of a verb participle followed by a noun,
analogous to English forms like living forest or long-departed soul.
In its sentential use in Kannada, this participial construction forms
relative clauses and is the most common type of relative clause
construction in the language. In its lexical use, it forms participial
compound nouns, a type of construction whose proper description
General abstracts. Thirteenth South Asian Languages Analysis Roundtable 99
has occupied traditional grammarians for centuries (e.g. Kesiraja
1260).
Despite their basic identity, there are numerous morphological,
syntactic and semantic differences between the lexical and syntactic
uses of this participial construction. We discuss these differences in
detail and show how they follow quite directly from the single
theoretical distinction that we have drawn. We also discuss the
comparable English construction and show that it is neither lexical
nor sentential, but rather phrasal. Finally, we discuss the relation
between the notions 'lexical' and 'lexicon'.
100 Studies in the Linguistic Sciences 20:3 (Spring 1991)
ASPECTS OF THE SYNTAX OF SPOKEN INDIAN ENGLISH
S. N. Sridhar and Indira Ayyar
(State University of New York at Stony Brook)
While there have been a few studies of Indian English based on
written material, literary and non-literary, there do not seem to be
many systematic (as opposed to impressionistic) studies of the spok-
en language. Yet, there is reason to believe that typically non-native
features might occur more in spoken than in the (monitored) written
language. In this paper, we analyze six spoken narratives and con-
versations involving educated speakers of Indian English and
identify a number of syntactic patterns characteristic of Indian Eng-
lish. We then go on to show that these patterns are more charac-
teristic of speakers educated through one of the regional languages
than of speakers who studied through English. We suggest that this
difference is attributable to what might be called the 'permeability of
syntax' in bilingual usage, i.e., the bilingual's tendency to freely
extrapolate from the syntax of one language in using the other while
addressing fellow bilinguals.
General abstracts. Thirteenth South Asian Languages Analysis Roundtable 1 1
PAIR-LIST ANSWERS IN HINDI INDIRECT QUESTIONS
Veneeta Srivastav
(Rutgers University)
Under standard accounts the possibility of (2) as an answer to
(1) is taken to involve LF movement of what, the wh in-situ, to
matrix spec, as shown in (3):
(1) Who knows where Mary bought what?
(2) Joe knows where Mary bought the book and Bill knows
where she bought the pen.
(3) [CP [spec whati whoj] [IP tj knows [CP [spec ti wherek]
[IP Mary bought ti tk ]]]]
The facts of Hindi, however, are a problem for this analysis of pair-
list answers.
Hindi has wh in-situ but its scope properties are very different from
those of Chinese wh in-situ, as noted by Davison (1984), Mahajan
(1987), and Srivastav (1989). In particular, when it occurs inside a
finite complement it necessarily takes narrow scope. (4), for exam-
ple, can only be interpreted as an indirect question. An LF like (5)
with the embedded wh in-situ in matrix spec is clearly ruled out:
(4) ravii jaantaa hai merii ne kyaa kiyaa
Ravi knows Mary what did
'Ravi knows what Mary did.'
(NOT: What does Ravi know Mary did?)
(5) *[ CP [ spec whati ] [ IP ravi knows [ CP [ spec ti ]
[ IP mary ti did ]]]]
Now consider (6), the Hindi counterpart of (1):
(6) kaun jaantaa hai ki merii ne kahaa kyaa kharidaa
who knows that Mary where what bought
'Who knows where Mary bought what?'
102 Studies in the Linguistic Sciences 20:3 (Spring 1991)
(6) readily allows for a pair-list answer like (2). This answer ob-
viously cannot be derived from an LF like (3) in which kyaa 'what'
has moved to matrix spec, given (4) - (5). The Hindi facts thus call
for an account of pair-list answers in these contexts which is not
dependent on movement of wh in-situ to matrix spec.
The alternative pursued here therefore takes (7) as the only LF
of (6):
(7) [CP [spec whoj] [IP tj knows [CP [spec whati wherek]
[IP Mary tk ti bought ]]]]
Following Karttunen 1977 it is assumed that this question denotes a
set of true propositions of the form 3x know' (x, P), where P is the set
of true propositions of the form 3y 3z bought' (mary, y at z).
Schematically put, the meaning of the propositions P, the meaning of
the questions depends on a relation between a set of individuals X
and a set of propositions P, where both S and P may contain one or
more members. Though it is not specified whether every member of
X knows every member of P, there is a conventional implicature that
Vx IX and Vp IP, x knows p. Thus, if (6) were answered with John
and Bill it would be taken to imply that they both know the two
propositions linking the pen and the book to the places Mary bought
them at.
Note that a pair-list answer is given in situations where neither
X nor P are singletons, but it is not the case that Vx IX and Vp IP, x
knows p. That is, the relation between X and P does not distribute
down to the members of the two sets. The pair-list answer, we
might say, cancels the conventional implicature that every member
of X knows every member of P by making explicit that the members
of X JOINTLY know P. The pair-list answer thus involves a cumulative
reading of (7), rather than the pure distributive reading, in the sense
of Scha 1981.
While the primary motivation for moving from an account based
on movement to one based on a semantic distinction between dis-
tributive, and cumulative readings to a single representation comes
from Hindi, it is shown that there are advantages in adopting this
approach even for English. As noted by Hirschbuhler (1978), the
movement account does not predict the absence of a pair-list answer
for (8):
(8) Which girl knows where Mary bought which book?
General abstracts, Thirteenth South Asian Languages Analysis Roundtable 103
Under the movement account the wh in-situ which book should
be able to move to matrix spec on a par with (1). But if the matrix
spec contains which girl, and which book has not moved to matrix
spec, there is a principled reason why (1) should allow such move-
ment but not (8).
Under the present account the absence of a pair-list answer for
(8) is predicted. Since which girl carries a uniqueness implication,
the set of individuals who know the indirect question will be a sin-
gleton, but cumulative readings require both sets in know' (X,P) to
have more than one member. In the case of (1), the cumulative
reading is possible since who does not carry a uniqueness implication
and the set X can have more than one member.
104 Studies in the Linguistic Sciences 20:3 (Spring 1991)
CONVERGENCE AND SYNTACTIC CHANGE:
THE CASE OF THE NEGATIVE PARTICIPLES IN DAKKHINI
V. Subbarao and Harbir Arora
(Delhi University)
The purpose of this paper is to discuss negative participles
(conjunctive, infinitival, and relative) in Dakkhini with a view to
arrive at the principles governing syntactic change in contact situ-
ations. We show how far Dakkhini negative participles have retained
the original Hindi-Urdu pattern and how far they have converged
with Telugu.
Dakkhini amalgamates the syntax of both Telugu and Hindi-
Urdu in negative conjunctive participles functioning as adverbial
clauses. In Telugu conjunctive participial constructions a negative
conjunctive participle can be used in place of an affirmative con-
junctive participle. The negative in such cases has no semantic im-
port.
(1) T: ne:nu WELLAKAMUNDU kamala pustakam kone:sindi
I go NEG before Kamala book bought
'Kamala bought the book before I went.'
(2) T: ne:nu WELLE MUNDU Kamla pustakam kone:sindi
I go before Kamla book bought
Dakkhini in such cases uses an infinitival construction, and the
negative here too has no semantic import.
(3) D. mai nai: ja:ne ke pail kamla: kita:b khari:d li:
I NEG go INF GEN before Kamla book bought
'Kamla bought the book before I went.'
(4) D. mafjaine ke pail kamla: kita:b khari:d li:
I go INF GEN before Kamla book bought
General abstracts, Thirteenth South Asian Languages Analysis Roundtable 105
However, there are negative participles with so which do not
have an affirmative counterpart. This, we argue, is due to the
principle of Avoidance of Constructional Homonymity which operates
in syntactic change.
We also show that when convergence takes place, certain hybrid
constructions emerge which retain the morphological features of the
parent language but converge with the syntax of the contact
language.
106 Studies in the Linguistic Sciences 20:3 (Spring 1991)
THE INFL NODES IN NON-FINITE CLAUSES
IN DRAVIDIAN AND TIBETO-BURMAN LANGUAGES
K. V. Subbarao and Lalitha M.
(Delhi University)
The purpose of this paper is to provide an explanation for the
occurrence of lexical subjects with ergative or nominative case
marking in non-finite clauses.
According to the standard GB framework, the subject position of
a non-finite clause is always occupied by PRO since it should occur in
an ungoverned and unease-marked position. However, 'languages
use other marked options to permit phonetically realized subjects of
infinitives to surface; e.g. nominative subjects in Portuguese with
infinitives with AGR, dative subjects of indirect question in Russian,
prepositional phrase subjects in Hebrew' (Chomsky LGB 1981:142).
We present data from Dravidian and Tibeto-Burman languages
which clearly show that
(i) Non-Finite clauses (adjectival, adverbial, and conjunctive
participial clauses) permit lexical subjects; and
(ii) the Subject Identity constraint is not obeyed in conjunctive
participial clauses.
To account for these phenomena, we attempt to provide an
explanation in terms of the marked options which the languages of
these two families permit. We show that the contents of the INFL
node and case theory play a crucial role in such an explanation.
General abstracts. Thirteenth South Asian Languages Analysis Roundtable 107
NEW DIMENSIONS OF WORD ORDER FREEDOM
IN VERB-FINAL LANGUAGES
Asha Tickoo
(University of Pennsylvania)
SOV languages are said to have less rigid word order than SVO
languages because they allow scrambling. This paper attempts to
demonstrate that the 'freedom' of SOV languages is also expressed in
the fact that there are weaker functional constraints on preposing in
these languages compared to the constraints on preposing in verb-
medial languages. A comparison is made of preposing in verb-final
Hindi and Kashmiri and verb-medial English. The analytical frame-
work adopted in the evaluation of the constraints on Kashmiri and
Hindi preposing is the approach of Prince (1981, 1984) and Ward
(1985), in which (i) the referent of the preposed constituent of Eng-
lish preposing marks a salient scalar relationship to another
discourse entity, itself already evoked or saliently inferrable from
the discourse, and (ii) the preposing is 'presuppositional' (cf.
Jackendoff 1972) in that it marks an open proposition (OP) as salient
in the discourse (cf. Prince 1981).
The study demonstrates that while preposing in verb-final lan-
guages is more functionally constrained than fronting by scrambling,
in that it occurs only in a clause that is temporally asequenced to its
preceding clause and fronting by scrambling is not constrained to do
this, it is less functionally constrained than preposing in verb-medial
English. The OP of English preposing must be salient given. Hindi
and Kashmiri preposing, on the other hand, are felicitous when the
OP is shared knowledge given (cf. Prince 1981).
108 Studies in the Linguistic Sciences 20:3 (Spring 1991)
CLAUSAL VS. NON-CLAUSAL SUBORDINATION
IN SANSKRIT NARRATIVES
Sarah Tsiang
(University of Illinois at Urbana-Champaign)
As is well known, Sanskrit has two major means of subordina-
ting one proposition to another. One employs clausal structures with
finite verbs (mainly relative-clause constructions), and the other
employs non-clausal structures with non-finite verbs (mainly ab-
solutive, participial, infinitival, and verbal-noun constructions).
What is not so well known is the functional difference between the
two methods of subordination. In this paper, I try to give an account
of what motivates the choice between clausal and non-clausal sub-
ordination in particular contexts by considering the consequences of
each choice. For example, finite verbs are able to express modality
(indicative, optative, imperative), while non-finite verbs cannot un-
ambiguously indicate modality. On the other hand, non-finite verbs
can be used to clearly indicate relative tense, while finite verbs can-
not. I further illustrate the role that these considerations and others
play in Sanskrit narrative texts.
For instance, in the Vetalapancavirh s'ati there are two des-
criptions of the same procedure for obtaining magical power. The
first is a brief summary of the rite given in order to illustrate to the
king how he will meet his death. All but the last of the verbs in the
description appear in the absolutive:
tvam ito mahasattvamaharajasrlvikramadityasya raja-
dhanirh gatva tasya rajna upadhaukanaya bilvaphala-
bhyantararatnani vinives'ya dinakatipayarh yojayitva ma-
hasattvarh sahayarh krtva pretadhisthitajfianinarh raja-
sahayena daksinasmasane nTtva tarn savarh rajanam ca
mandalam racayitva tatra nitva mandalapujarh vidhaya
savasya mastake padam dattva rajanarh devyai balirh
dattva savadvayamastake padadvayarh dattva nigadisyasi
devi ...
General abstracts, Thirteenth South Asian Languages Analysis Roundtable 109
Here, then, in what essentially is a narrative passage, the absolutive
is sufficient to indicate the temporal sequencing of the actions.
The second description consists of the instructions for
performing the rite as given by the vetala to the king. To ensure
proper execution, the sequence of actions and their relationship to
each other must be clearly expressed. In this version we find a
variety of methods of subordination, including absolutive (Abs.),
locative absolute (LA), participial (Pple.), and relative-clause (RC)
constructions, as well as serialization of sentences, each containing a
finite verb (FV) or other main verb (MV). (Sentence breaks are
indicated by I.)
tvam idanirii mam evarii nitva (Abs.) tatsakasarh gaccha
(FV) I gate tvayi (LA) tada sa tu tvam drstva harsam atularii
prapya (Abs) nigadisyati (FV) I bho maharaja .../ tatra
drutarfi snanaiii kuru (FV) I devataradhanarh samacara (FV)
I tvaya saha devatapujam vidhaya (Abs.) avayoh eva
yathavaca siddhih bhavati (FV) I tada kartavyam (MV) I
tarhi tvayi api tad vacanena devyah kunde payasi snatva
(Abs.) tatra upasthite (LA) tada tvarh vadisyati (FV) I
devataiTi pujaya (FV) ... I tada devatarh pujayisyasi (FV) I
krte devataradhane (LA) tada vadisyati (FV) I devatarh
pradaksinlkrtya (Abs.) ... I tada tvam vadisyasi (FV) I ...
narapatir aham ... I tava vacanam idam avagamya (Abs.)
[yada sa tu ksantisilah devataya dandapranamam kurvan
(Pple.) kayaprapatam kurute (FV)]rc tatksanat tvam api
khadgena s'irah chittva (Abs.) tanmastake mrtakasya ma-
stake ca padadvayarh dattva (Abs.) mukulakararh s'irasi
anjaliih nivesya (Abs.) vadi§yasi (FV) I devi ... I varam imam
prarthayisyasi (FV)
110 Studies in the Linguistic Sciences 20:3 (Spring 1991)
THE MENTAL DICTIONARY:
ITS ROLE IN LINGUISTIC THEORY
K. G. Vijayakrishnan
(University of Hyderabad)
In this paper, it is argued that apart from the Permanent
Lexicon, defined in (i) below, which is quite non-controversial, the
construct 'Mental Dictionary', defined in (ii) below, is also necessary
in linguistic theory.
(i) The Permanent Lexicon is a list of all the underived entries
of the language. Idiosyncratic properties of non-derived
entries like their phonological structure, meaning, sub-
categorization, case-marking, selectional restriction, theta
frame, etc., are part of the lexical entry.
(ii) The Mental Dictionary is a paradigmatic list of lexical
entries of the language.
(i) is a proper subset of (ii) (henceforth MD).
The paper is organized in two parts. The first part examines
some aspects of the verb morphology of Tamil — a diglossic,
Dravidian language — and argues that some of the differences bet-
ween the high and the low varieties must be explained with
reference to the MD. The seemingly accidental gaps in the depleted
paradigms of the low variety can be accounted for in a principled
manner with reference to the related paradigms in the high variety.
In other words, these differences can be captured as sub-regularities
in the MD. A theory lacking the MD will not be descriptively
adequate.
The second part of the paper is concerned with the definition of
MD with respect to the recent proposal regarding the Lexicon as an
artifact of the Listing Principle (Borer MS). A modified version of the
Listing Principle is proposed in the light of a discussion of some
aspects of verb formation in Tamil. It is suggested that incorporating
General abstracts. Thirteenth South Asian Languages Analysis Roundtable 1 1 1
the construct 'Mental Dictionary' in linguistic theory would require
redefining the notion of lexical insertion at D-Structure. I
demonstrate with data from slip-of-the-tongue phenomena that this
is a move in the right direction which would bridge the gap between
theoretical linguistics and psycholinguistics.
12 Studies in the Linguistic Sciences 20:3 (Spring 1991)
THE HARAPPAN SCRIPT:
THE MOST ANCIENT FORM OF DRAVIDIAN
Clyde A. Winters
(Uthman Dan Fodio Institute)
There has been much speculation concerning the identity of the
language of the Harappan script. Recently the writing was
deciphered using the Dravidian languages. In this paper I discuss
the Dravidian character of the Harappan language and the historical
linkages between the Harappan language and contemporary
Dravidian languages, especially Tamil.
General abstracts. Thirteenth South Asian Languages Analysis Roundtable 1 1 3
VERBAL COMPOUNDS IN MALAYALAM
Shuichi Yatabe
(Stanford University)
Mohanan (1986) shows that compounds in Malayalam can be
grouped into two categories, subcompounds (compounds that involve
one head and one non-head) and cocompounds (compounds that in-
volve coordinate, and hence multi-headed structure), and that the
various phonological differences between the two can be explained
by assuming that they are formed in different strata (or levels) in
the lexicon. In this paper I argue that subcompounds in Malayalam
should be further divided into two subcategories, verbal compounds
(compounds in which there is a predicate-argument relationship
between the immediate constituents) and root compounds
(compounds in which there is no predicate-argument relationship
between the immediate constituents). I claim that this
subclassification makes it possible to eliminate several residual
stipulations found in Mohanan's (ibid.) account.
Mohanan, in the course of demonstrating that subcompounds in
Malayalam undergo Onset Fusion, Sonorant Degemination, Stem-
Initial Gemination, Stem-Final Gemination, and Nasal Deletion, and
that both subcompounds and cocompounds undergo Vowel Leng-
thening and Vowel Sandhi, points out in passing that verbal
compounds, which he takes to be a type of subcompound, do not
undergo Stem-Final Gemination or Nasal Deletion; but Onset Fusion,
Sonorant Degemination, Stem-Initial Gemination, Vowel Lengthening,
and Vowel Sandhi are likewise inapplicable to verbal compounds.
This bifurcation between two classes of subcompounds justifies
postulation of a separate stratum for verbal compounding.
The behavior of verbal compounds in Japanese and English also
supports the proposed analysis. Root compounds, but not verbal
compounds in Japanese undergo a phonological rule called Rendaku
(Sequential Voicing) (see Vance 1987 for references). Likewise, root
compounds, but not verbal compounds in English undergo the
Rhythm Rule (Roeper & Siegel 1978). A universal generalization
seems to be that verbal compounds undergo fewer phonological rules
than root compounds. The proposed analysis accounts at least for
114 Studies in the Linguistic Sciences 20:3 (Spring 1991)
why root compounds and verbal compounds undergo different sets
of phonological rules, if not necessarily for why one of the two sets is
smaller than the other.
Alternatives to the stratum-ordering account of the phenomena
in question are discussed and ultimately rejected. For instance, the
attempt to reduce the difference between verbal compounds and
root compounds to the Right Branch Condition, which has been
proposed for the application of Rendaku in Japanese, and the attempt
to resort to the categorial difference between N and V to explain the
different behavior of the two classes of compounds are both shown
to be empirically inadequate. The former proposal fails to capture
the Malayalam facts and the latter to capture some Japanese facts.
Provided that the putative universal generalization mentioned in the
last paragraph is correct, the stratum-ordering account or something
analogous to it seems necessary.
Thus, the assumption that verbal compounding constitutes a
separate stratum in the lexicon not only allows us to capture some
properties of verbal compounds in Malayalam but offers a way to
make a universal characterization of verbal compounds.
General abstracts. Thirteenth South Asian Languages Analysis Roundtable 1 15
ERGATIVITY IN THE INDO-EUROPEAN LANGUAGES
OF SOUTH ASIA:
DIACHRONIC AND SYNCHRONIC PROCESSES
Boris A. Zakharyin
(Moscow State University)
The modern Indo-European languages of South Asia should be
divided typologically not into two (Eastern — non-Eastern) but into
three groups: Northern — non-Northern and a further differentiation
of the last one into Eastern — non-Eastern. Synchronically only the
languages of the Eastern group are devoid of ergativity which no
doubt (compare the Old Bengali text of Charya) diachronically were
characterized by the same type of (late) ergativity that now the
languages of the Northern group clearly demonstrate; see Zakharyin
1987. Not purely historical but historico-sociolinguistic factors must
be taken into account, as the evolutionary accusativization of Dakhini
and of Literary (but not Colloquial) Singhalese show. The degree of
conscious subjective interference into functioning of a certain lan-
guage (as evolutionary processes in such pairs as, say, Punjabi-Hindi,
Shina-Kashmiri, Assamese-Bengali demonstrate) is also playing an
important role in the historico-typological processes in South Asia.
In his search for diachronic antecedents of the ergativity in NIA
and Dardic H. H. Hock (1986) has correctly given up the traditional
term 'past participle' in connection with -tal-na forms and has justly
rejected as 'dubious' the suggestions by Pray (1976), Anderson
(1977), and Klaiman (1978). But his own analysis of examples like
Old Indo-Aryan 'to be born' ('intransitive but not passive', as he
says) is not quite sound, as he does not take into account the dif-
ference between the tense stems jan-a- 'to give birth' - jd-ya- 'to be
born'; see S. W. Jamison 1979. Hock's statement that in Vedic
'passives are barred from intransitives' could also be correlated, as in
Vedic personal passives (especially Aorist passives) there existed a
typically ergative opposition 'A - S/P'. Compare, for example,
...nabhd ny asddi hold 'The hotar got placed at the hub ...' (RV 3.4.4.)
and agnir hold ny asidad...upasthe mdtuh 'Agni-hotar ... placed
himself in his mother's lap' (RV 5.1.6); see T. Elizarenkova 1982.
116 Studies in the Linguistic Sciences 20:3 (Spring 1991)
Thus we could state that in Vedic an 'A - S/P' ergative device oper-
ates with all the three types of P-oriented constructions (finite
passive, participial, and gerundivial), as in all the three the
morphological marking of A-NP differs from the marking of either P-
NP or S-NP.
The history of ergativity in Indo-Aryan and Dardic also shows
that there has been a continuous process of semo-syntactic con-
vergence of the stative and active verbal forms in which the active
(and finite) forms have always been playing the role of a 'standard
model' for their stative counterparts. In the evolution of ergativity a
direct line can be established between Vedic proper and later
Modern Indo-Aryan and Dardic, standardized Sanskrit and the
Prakrits being 'gaps' in this evolution.
General abstracts, Thirteenth South Asian Languages Analysis Roundtable 1 17
THE DEVELOPMENT OF ERGATIVITY IN INDO-EUROPEAN
LANGUAGES OF WESTERN INDIA IN THE FIFTEENTH
THROUGH TWENTIETH CENTURIES
Boris A. Zakharyin and L. V. Khokhlova
(Moscow State University)
This paper aims at evaluating trends in the development of
ergativity (morphological and semantic) in some Indo-Aryan and
Dardic languages during the fifteenth through twentieth centuries;
data obtained from informants and extracted out of the texts of the
fifteenth through twentieth centuries were analyzed. Morphological
ergativity was treated in accordance with the codified definition by
R. M. W. Dixon (1979), and in the syntactic sphere, B. Comrie's
differentiation (1979) between accusative-nonaccusative types of
verbal concord was taken into account.
The starting point for Old Punjabi, Gujarati, and Rajasthani of the
fifteenth century was the maximal development of morphological
ergativity and preservation of the accusative type of syntax. Later in
Punjabi and Rajasthani, and still later in Gujarati, morphological
ergativity started decreasing and syntactic ergativity, increasing cor-
respondingly. Modern Rajasthani and Gujarati have retained
syntactic ergativity. but Punjabi has returned to accusative syntax in
noun functioning — with the exclusion of the personal pronouns, as
the latter have developed the accusative type of morphology and the
ergative type of syntax.
The diachronic analysis permits us to take a non-trivial look at
synchronic ergativity. While remaining on purely synchronic soil we
could agree that Rajasthani is the only exception to Anderson's
(1977) and Comrie's (1978) universal (stressing that combination
"ergative verbal concord + non-accusative marking' does not occur in
one and the same language) and to back Magier in noting this. Of
course, by manipulating synchronic data it is always possible to con-
vert this 'exception' into 'non-exception', following, for example,
Klaiman (1987) who supposed that as in Marwari, A, S and also P
might have the same unmarked case. (We deal here with 'neutral
1 1 8 Studies in the Linguistic Sciences 20:3 (Spring 1 99 1 )
case marking'. NP marking in Marwari thus does no harm to the An-
derson-Comrie's universal.) Diachronically, it is quite clear that the
identical marking of A, S, and P in Rajasthani is not a unique
phenomenon, that it was some historical stage in the development of
split ergativity in different Indo-European languages of the area.
The split in P-marking in Kashmiri noted by P. Hook (1984) and
thought to be a 'unique feature' of the language does not look like
this in a broad historical perspective.
General abstracts. Thirteenth South Asian Languages Analysis Roundtable 119
A SKETCHY HISTORY OF CLITICIZATION AND VERB STEM
NOUN INCORPORATION IN MUNDA
Norman Zide
(University of Chicago)
Noun incorporation in verb stems is reconstructable for South
Munda (SM), but not for Proto-Munda (PM). Subject-marking pro-
nominal prefixes are reconstructable for SM and, it seems, likely for
PM. Intransitivity/transitivity is marked by distinct tense suffixes
in SM (but not Northern Munda), and is perhaps to be reconstructed
for PM. In the (ca. 10) modern languages. North and South, enclitic
pronouns marking subject, and, less commonly, direct and indirect
object are common. I present and analyze the data, and look at the
observations and analyses of others (B. P. Mahapatra, J. Sadock, D.
Stampe, and S. Starosta) on cliticization in various Munda languages.
I propose a tentative reconstruction of clitics in Munda, and offer a
hypothesis about the development of noun incorporation in verb
stems in South Munda.
i
PART IT: PANKrS
i
CONVENTIONS OF POLITENESS
IN SOUTH ASIAN LANGUAGES
Organized by Manindra K. Verma
(University of Wisconsin at Madison)
i
Abstracts, Conventions of Politeness, Thirteenth S ALA Roundtable 1 25
A GRAMMAR OF POLITENESS IN MARATHI
Rajeshwari Pandharipande
(University of Illinois at Urbana-Champaign)
In this paper I discuss theoretical as well as empirical issues
related to defining the notion of politeness in Marathi. I argue that
the notion of politeness cannot be defined purely in terms of struc-
tural configurations and that pragmatic as well as discourse consider-
ations must be incorporated for an adequate analysis of politeness.
The following major points are in focus: (a) the structures (mor-
phological and syntactic) acknowledged as correlates of politeness in
traditional grammars of Marathi (e.g., morphological markers of
number and person, verb agreement markers, etc.) and their in-
adequacy in defining politeness; (b) pragmatic and discourse con-
ditions which are relevant for defining politeness; and (c) finally, the
necessity to incorporate the 'interactional' dimension of the discourse
in adequately characterizing politeness in Marathi.
Additionally, this paper compares and contrasts the notion of
politeness in Marathi and Hindi and demonstrates the relevance of
the approach proposed in the paper for crosslinguistic/typological
generalizations.
126 Studies in the Linguistic Sciences 20:3 (Spring 1991)
TOWARDS AN ETHNOGRAPHY OF POLITENESS IN MAITHILI
Mithilesh K. Mishra
(University of Illinois at Urbana-Champaign)
This paper discusses some of the salient semantic and pragmatic
aspects of Maithili (verb) agreement markings which are very com-
monly used to signal politeness in this language. On the basis of the
evidence drawn from Maithili, the paper argues that POLITENESS, like
ILLOCUTIONARY FORCE, should be treated as a derivative property of
utterances/speech acts and not as some kind of structurally prim-
itive concept or property of utterances or speech acts. The paper
also argues that by analyzing the mechanisms that relate SENSE of a
sentence with its FORCE, it is indeed possible to map the grammar of a
culture which the speakers (belonging to that particular culture) ac-
quire and command as a very vital part of their communicative com-
petence. Finally, the paper compares and contrasts the strategies
used by Maithili and Hindi speakers for conveying POLITENESS and/or
INDIRECTNESS.
(
Abstracts, Conventions of Politeness, Thirteenth SALA Roundtable 127
DIRECTIVES IN PANJABI AND LAHANDA
Tej K. Bhatia
(Syracuse University)
The aim of this paper is to focus attention on the notion of
directives and how they are coded and actually used in Punjabi and
Lahanda. Directives constitute a universal feature of human langu-
ages, and this paper will examine spoken as well as written direc-
tives with reference to the degree of politeness, power, and authori-
ty. In addition to exploring the various formal and functional aspects
of directives, the paper will also examine the claims of theories such
as Brown and Levinson's 'face-saving' theory (1987).
REFERENCE
Brown, Penelope, & Stephen C. Levinson. 1987. Politeness: Some
universals in language use. Cambridge: Cambridge University
Press.
128 Studies in the Linguistic Sciences 20:3 (Spring 1991)
LINGUISTIC CONVENTIONS OF POLITENESS
IN BHOJPURI AND MAGAHI
Manindra K. Verma
(University of Wisconsin at Madison)
This paper discusses the elaborate pronominal system and the
agreement markings used in Bhojpuri and Magahi to signal polite-
ness. Bhojpuri seems particularly rich in pronominal distinctions of
politeness, whereas Magahi seems to bring in agreement features for
finer distinctions. Beyond these, both languages seem to involve uses
of certain syntactic constructions in preference to others for their
expression of politeness. The role of the passive — both morpho-
logically and syntactically — seems to be particularly important.
DISCOURSE IN THE OTHER TONGUE
Organized by Yamuna Kachru
(University of Illinois at Urbana-Champaign)
Abstracts, Discourse in the Other Tongue, Thirteenth SALA Roundtable 131
RECREATING SOUTH ASIAN SPEECH ACTS IN ENGLISH:
A STUDY IN LINGUISTIC TRANSFER
Jean D'Souza
(National University of Singapore)
This paper is concerned with pragmatic aspects of language use.
Studies by Apte 1974, D'Souza 1988, Loveday 1982, Olstain & Cohen
1981, Sridhar & Sridhar 1986, among others have shown that there
are important differences, both cross-cultural and cross-linguistic, in
the way a given speech act (e.g. directive, commissive, expressive) is
performed in a given language. Speech acts in Indian English are
examined in an attempt to discover how, if at all, the realization of
speech acts in this variety differs from their realization in the native
varieties of English. Data will be taken from creative writing in
English, and the 'high' vs. 'colloquial' use of English (as manifest in
text and dialogue) will be contrasted to discover whether the 'context
of situation' (Kachru 1980) affects the realization of the speech act.
132 Studies in the Linguistic Sciences 20:3 (Spring 1991)
SPEECH ACT IN THE MOTHER TONGUE AND THE OTHER
TONGUE
Yamuna Kachru
(University of Illinois at Urbana-Champaign)
Levinson, toward the end of a discussion of speech act theory,
mentions two major traditions of research on language use 'under-
mining' speech act theory: one, ethnography of speaking, and the
other, language acquisition (1983:278-283). Later, in a brief section
on the interrelationship of pragmatics, sociolinguistics, and psy-
cholinguistics, Levinson again points to the common interests shared
by both pragmatics and sociolinguistics and suggests potential con-
tributions each could make to the other (374-375). The present
study makes an attempt to show that it is essential to draw on all
three subdisciplines — pragmatics, sociolinguistics, and ethnography
of communications — to account for speech act data from South Asian
languages and Indian English. The data are drawn from literary
texts in select South Asian languages and Indian English. It is hoped
that insights gained from this study will be helpful in designing a
more comprehensive study on speech acts in South Asian languages
and in Indian English.
REFERENCE
Levinson, Stephen C. 1983. Pragmatics. Cambridge: Cambridge
University Press.
Abstracts, Discourse in the Other Tongue, Thirteenth S ALA Roundtable 1 33
ON CREATING SPEECH ACTS:
THE CREATIVITY OF INDIAN ENGLISH WRITERS
Cecil L. Nelson
(Indiana State University)
As B. Kachru has pointed out (1988:584), Raja Rao and, by ex-
tension, other Indian English creative writers expand the expressive
repertoire of English in their 'extensive use of native similes and
metaphors...; the transfer of rhetorical devices; the transcreation of
proverbs, idioms, and speech acts; ... and an extensive employment of
code-mixing ... and code-switching." This paper examines and cate-
gorizes examples of created speech acts in the works of some Indian
English authors, including 'classics' such as Raja Rao and 'moderns'
such as Bharati Mukherjee, 'in a transplanted language in a speech
community that does not share the native cultural contexts of the
transplanted language', as Y. Kachru (1987:87) puts it. Through ex-
amination of such fictional language performance, we may see into
the dual — or, if it is not partitioned, greatly expanded — competence
of the practiced bilingual. A given speech act may serve, for ex-
ample, to let the reader know what the relevant appropriateness
conditions in a scene are. Use of such devices tests authors' creativi-
ty, since writers (most often) try to let monolingual readers in on the
meaning/significance of such elements. In the novel Jasmine, for
example, Mukherjee (1989:49) writes (bracketed insertions mine):
[Dida, the protagonist's paternal grandmother,] spoke only to
Pitaji [Father].... 'Some women think they own the world
because their husbands are too lazy to beat them,' but
Mataji [Mother] just went about her cooking with her mouth
zipped and her veiled head down.
The passage is densely packed with clues about the relationship
of the mother- and daughter-in-law in this traditional Hindu family.
The indirection of the speech act in using a proverbial style and ad-
dressing the comment to someone other than the real target gives
the pronouncement more force than a simple and specific criticism
directed at the son's wife would have. Interestingly, Mukherjee also
shows in this novel how speech-act bilingualism develops in both
134 Studies in the Linguistic Sciences 20:3 (Spring 1991)
directions. Jasmine's Indian protagonist says of her American
husband (p. 26): 'Bud calls me Jane. Me Bud, you Jane. I didn't get
it at first. He kids.'
This examination of the speech act creativity of Indian English
authors will support the 'justification' of a 'method of expression ... as
distinctive and colorful as the Irish or the American' (Raja Rao, cited
by B. Kachru 1988:586).
REFERENCES
Kachru, Braj B. 1988. Toward expanding the English canon: Raja
Rao's 1938 credo for creativity. World Literature Today 62:4.582-
586.
Kachru, Yamuna. 1987. Cross-cultural texts, discourse strategies and
discourse interpretation. Discourse across cultures: Strategies in
World Englishes, ed. by L. E. Smith, 87-100. New York: Prentice-
Hall.
Mukherjee, Bharati. 1989. Jasmine. New York: Grove Weidenfeld.
Nelson, Cecil L. 1988. The pragmatic dimension of creativity in the
other tongue. World Englishes 7:2.173-182.
Raja Rao. 1963 [1938]. Kanthapura. New York: New Directions.
Abstracts, Discourse in the Other Tongue, Thirteenth SALA Roundtable 135
LANGUAGE AND FEMALE IDENTITY IN INDIA
Tamara Valentine
(University of South Carolina at Spartanburg)
That linguistic variation correlates with the social variable
'gender' is a universal feature of all speech communities. But, are
there common life experiences or gender-specific circumstances
particular to women which affect cultures of the world? Is it that
male and female language differences are directly related to the
universal oppression felt by all women and girls of the world? Or
are language styles a reflection of subcultural differences? I address
interactions of female Hindi and English speakers from actual spoken
data recently collected in the urban and rural areas of North India. 1
examine the major theme emerging on 'women's style of talking';
that is, women's interactive style is based on solidarity, support, and
cooperation. I examine the recurring patterns in Indian female con-
versations, consider how the patterns are similar or different to the
major perspectives explored in Western studies, and discuss the im-
portance of studying the role of gender in cross-cultural com-
munication and the implications for the understanding of language
use across cultures.
LANGUAGE IDENTITY AND CONFLICT IN SOUTH ASIA
Organized by Rakesh M. Bhatt
(University of Illinois at Urbana-Champaign)
Abstracts, Language Identity and Conflict, Thirteenth SALA Roundtable 139
A HOUSE DIVIDED:
CONFLICT AND RIVALRY IN TWO VARIETIES OF A
LANGUAGE
Mariam Ahmed
(University of Illinois at Urbana-Champaign)
This paper addresses the issue of conflict and rivalry concerning
the history, functions, and status of Urdu in pre-1947 India and
post-1947 South Asia. The paper aims at (1) the genesis of the
conflict, (2) the attitudes toward Urdu, (3) the political, social,
religious, and educational implications of such attitudes.
I also discuss briefly the diversity which has developed between
Indian Urdu and Pakistani Urdu and the underlying motivations for
such diversity. I would also like to add a comment about the
diversity which has developed between the Urdu spoken in the north
of India and the Urdu spoken in other regions.
140 Studies in the Linguistic Sciences 20:3 (Spring 1 99 1 )
TRANSPLANTED LANGUAGES AND ETHNIC IDENTITY
Tej K. Bhatia
(Syracuse University)
Several South Asian languages have been transplanted around
the world as a result of emigration from Asia. Although these
languages have not provoked any serious nationalism or linguistic
rivalry or conflict in their nonnative context, nevertheless they are
emerging as powerful markers and rally-points for group identity.
The aim of this paper is to examine the Hindi diaspora in this
context and to shed light on the question of language identity and
other related phenomena. The paper is centered around notions such
as language, ethnic identity, group solidarity, etc. The interaction of
these notions is examined primarily from two view-points: social
and scholarly.
The data are drawn from a wide variety of sources, including
field data, interviews and the related research literature.
i
Abstracts, Language Identity and Conflict, Thirteenth SALA Roundtable 141
LANGUAGE MINORITIES:
ISSUES OF IDENTITY IN A GLOBAL PERSPECTIVE
Kama! K. Sridhar
(State University of New York at Stony Brook)
Theoretical models that accommodate issues of linguistic iden-
tity, contact, and conflict are ideally an area of research for linguists,
particularly sociolinguists. Lack of such models in linguistics forces
us to look to available models in other disciplines. There are several
existing models in sociology, philosophy, and political science. The
present paper draws on models used in sociology and philosophy.
Special reference is made to Allard's (1979) work, which is based on
an analysis of forty-six minority groups in Western Europe. The
applicability of his approach to the study of minority languages in
South Asia is evaluated.
142 Studies in the Linguistic Sciences 20:3 (Spring 1991)
IDENTITY, CONFLICT AND CONVERGENCE:
SOUTH ASIA AS A SOCIOLINGUISTIC AREA
Rakesh M. Bhatt
(University of Illinois at Urbana-Champaign)
The past few years have seen a global conflict in terms of
movements and uprisings which seem to have revolved around a
common theme: IDENTITY. In these few years, the United States has
witnessed the rise of a powerful lobby that demands that English be
established as the national language of the United States, while in the
Soviet Union the Baltic states have started asserting their ethno-
linguistic identity to claim autonomy from the Union. Similar
expressions of identity and solidarity resulted in upheavals in the
other Eastern bloc countries. The outcome was inevitable: the fall of
the Iron Curtain and the union of the two Germanics.
This paper focuses on South Asia (mainly India) which provides
a fascinating sociolinguistic laboratory for the study of the
interaction of (1) language contact and language identity; (2)
language identity and language conflict; and (3) language conflict
and language change. Diverse groups of people belonging to different
caste, religious, or ethnic groups are integrated within a single
political structure. The socio-political and sociolinguistic
consequences of such social settings can be seen in the conflict that
arises in almost all domains of interpersonal interactions. Examples
include the violent situation in Jaffna, Sri Lanka, where Tamil and
Sinhalese speakers are engaged in a seemingly unending
ethnolinguistic war which has resulted in the loss of thousands of
lives so far; the creation of Pakistan out of British India in 1947 on
primarily religious grounds and later the creation of Bangladesh out
of Pakistan in 1971 on a mainly linguistic basis; the political
restructuring of Indian society after the creation of Hindi-speaking
Haryana out of the earlier state of Punjab and of Telugu-speaking
Andhra Pradesh from Madras; and recently, the terrorist attacks in
Abstracts, Language Identity and Conflict, Thirteenth SALA Roundtable 143
the states of Kashmir, Punjab, and Assam in support of regional
autonomy.
At the bottom of all these movements, conflicts, and uprisings is the
single issue of identity. This raises one important question: How do
different social groups in multilingual, multiethnic, and multicultural
countries become politically mobilized? A related question that is of
tremendous sociolinguistic importance is: How do different linguistic
groups maintain their linguistic identity under pressures of language
shift and convergence? These and other questions relating to issues
of language policies and language planning are discussed.
4
LANGUAGE OF RELIGION
Organized by Rajeshwari Pandharipande
(University of Illinois at Urbana-Champaign)
I
Abstracts, Language of Religion, Thirteenth SALA Roundtable 147
THE QUESTION OF DEFINING THE LANGUAGE OF RELIGION
Rajeshwari Pandharipande
(University of Illinois at Urbana-Champaign)
Although it is a well-known fact that religious language signi-
ficantly differs from its non-religious counterpart in terms of its
linguistic structure and sociolinguistic function, there is no definition
available at present which adequately describes the language of
religion. Language of religion has been studied from diverse per-
spectives by scholars in disciplines such as sociology, philosophy,
theology, psychology, and linguistics. The paper discusses the defini-
tions provided within the frameworks of the above disciplines and
points out that the shortcomings of these definitions stem from the
fact that they are based either exclusively on the content, or the
structure, or the function of the language of religion.
The paper further argues for a unified approach toward defining
the language of religion, an approach which takes into account its
structural, content-related, and functional features, and their inter-
dependency.
148 Studies in the Linguistic Sciences 20:3 (Spring 1991)
Vasat, srausat, AND OTHER RITUAL PARTICLES:
THEIR ORIGIN AND THEIR USE IN VEDIC RITUALISTIC
LITERATURE
Hans Henrich Hock
(University of Illinois at Urbana-Champaign)
Sanskrit has a long tradition of using a variety of particles in
ritualistic and religious/philosophical literature. Many of these, e.g.
om, him (hum), ho differ markedly in structure and connotations
from ordinary words of the language. In late or post-Vedic literature
they are therefore able to acquire very special and 'deep' religious/
philosophical connotations. This is especially true for om. Another
consequence of their differences from ordinary lexical items is that
the question of their etymologies has given rise to a number of
conflicting views.
In a forthcoming paper I show that om, the particle gaining the
most significant connotations in late and post-Vedic, arose from an
interjection comparable to Engl. o(h), ho(h), and the like. The route
through which it gained special connotations lay in Vedic recitation
practices. After briefly rehearsing the history of om, I argue that a
similar explanation is likely for him, attested as early as the Rig-
Veda, whose prehistoric development can only be inferred.
Explanations of this sort are not available for two other ritualis-
tic expressions, attested as early as the Rig-Veda, viz. vasat and s'rau-
sat: These are related to 'normal' words of the language, viz. the
roots vah- 'convey' and s'ru- 'hear, listen', respectively. The latter
relation is perfectly obvious in ritualistic literature, where the
s'rausat call is prompted by the formulas in (1) which contain a
causative of s'ru-.
(1) a. 6 sravaya (MS 1.4.11)
b. 6m sravaya (MS 4.1.11)
c. = a sravaya (KS 31.13) etc.
'Call for the s'rausat.'
Abstracts, Language of Religion, Thirteenth SALA Round table 149
Now, in some of its versions, viz. (la, b), the prompting formula
exhibits the particle (o)m. Combined with other evidence this sug-
gests that srausat functions as a ritualistic particle in the Veda.
Further, srausat and especially vasat share with the early history of
om the fact that they undergo various phonological deformations and
mystical interpretations. Thus, in the Satapatha-Brahmana (1.7.2.21),
vasat appears as vaiit and, via vauk, is then equated with vdk
'(sacred) speech'. In the final part of this paper 1 show that these
'ritualistic' deformations make it possible in principle to explain the
prehistoric phonetic developments leading to vasat and srausat.
150 Studies in the Linguistic Sciences 20:3 (Spring 1991 )
THE ROLE OF DEIXIS
IN DEFINING ORDINARY VS. RELIGIOUS LANGUAGE
Mithilesh K. Mishra
(University of Illinois at Urbana-Champaign)
This paper argues that the nature and function of religious
language can be adequately defined and explained by properly un-
derstanding the nature of deixis in language. The paper argues that
though Buhler's (1934) seminal work on deixis (i.e. the notion of
three coordinates of time, space, and ego-center or T) is a valid
starting point and is useful in certain contexts, his postulation of the
distinction between deixis at phantasma (imaginary deixis) and
real/ordinary deixis is arbitrary, unnecessary, and untenable. I
argue that by taking into account the ontology of Buhler's coordin-
ates, it is possible to 'derive' all kinds of deictic fields (including the
one referred to and used for cognition in the religious language) and
their concomitant discoursal semantic properties. This exercise is in-
dependently needed to account for the unbridled creativity of one of
the coordinates, T. Lastly, the paper discusses the functions of vari-
ous religious symbols and artifacts for both modifying the overall
nature of deixis in ordinary language and for 'collapsing' the (above-
mentioned) tripartite distinctions of the coordinates.
LITERARY RESPONSES TO INDIA
Organized by Girdhari Tikku
(University of Illinois at Urbana-Champaign)
Abstracts, Literary Responses to India, Thirteenth SALA Roundtable 153
RABINDRANATH TAGORE'S NOBEL PRIZE:
WHAT DOES IT MEAN?
Ali Anushiravani
(University of Illinois at Urbana-Champaign)
This paper is a study of the presentation of the Swedish
Academy's Nobel Committee. The purpose of the paper is to examine
the Western response to Tagore. In the Nobel Committee's presenta-
tion Tagore is called an 'Anglo-Indian poet'. Tagore is praised for his
'never-failing concomitant of the expansion of British civilization' and
his Christian message. Even Bengal is considered 'the oldest Anglo-
Indian province'. The Nobel Committee alienates Tagore from his
own Indian heritage and attempts to place his thoughts and ideas
within a Christian framework as though Tagore were celebrating
Christianity in his works. What the Nobel Committee of the Swedish
Academy sees in Tagore is not Tagore as he really is; it is Tagore
made in their own image. In other words, what they see in Tagore is
not 'the Other' but 'Us'. It seems what Vasco da Gama could not ac-
complish five hundred years ago, namely to Christianize India, has
been achieved by the Swedish Academy.
The question addressed in this paper is: If this is the case, then what
does Tagore's Nobel Prize mean? The point is that Tagore deserves
the Nobel Prize for bringing East and West nearer. This is what the
Nobel Prize is all about: to bring 'the Other' nearer, to create a
respect for 'the Other' as he is, not as we want him to be. The Nobel
Committee's presentation seems to fail to appreciate Tagore's
universal message for which he really deserves the Prize.
154 Studies in the Linguistic Sciences 20:3 (Spring 1991)
CHINESE RESPONSE TO TAGORE: PIN HSIN'S POETRY
Yongan Wu
(University of Illinois at Urbana-Champaign)
On a windy, moon-lit autumn night last year, you
suddenly came to me in the form of a book. On reading
your poems and biography, I had nothing in my heart but a
feeling of deep, crystal-like beauty.
Thus wrote Pin Hsin to Tagore on her first discovery of the Indian
poet and philosopher. This was a moment of contact between two
hearts, one in China and the other in India. The two writers had so
much in common that when Pin Hsin first read Tagore, there emitted
a ready echo from her heart.
This paper deals with the theme of love (divine love, universal
love, maternal love, brotherly love ...) in the works of Tagore and of
Pin Hsin through a study of their different philosophical and religious
backgrounds — the points where they started, the similar messages
they sent to the world in their writings — the point where they met,
and other differences there are in their treatment of the same theme
— the point where they varied. It is a comparative study of Tagore
and Pin Hsin yet it goes beyond their literary works to their
philosophical and religious beliefs that played a part in the pro-
duction of their literary works.
i
Abstracts, Literary Responses to India, Thirteenth SALA Roundtable 155
ALDOUS HUXLEY'S THE ISLAND
Girdhari Tikku
(University of Illinois at Urbana-Champaign)
The Island of Aldous Huxley is a Utopian novel, some say a
continuation of his Brave New World. The novel, one can argue, is an
attempt to use the facilities of modern medicine and psychology to
create an ideal society. The philosophical setting of this novel draws
heavily from Hindu and Buddhistic philosophic thought and mytho-
logy. Huxley was very close to Christopher Isherwood and both
collaborated with Swami Prabhavananda of the Vedanta Center at
Los Angeles in publication of the Gita texts. This paper expands on
these points to see the novel as a western response to Indian
literature and thought.
156 Studies in the Linguistic Sciences 20:3 (Spring 1991)
COLERIDGE AND BASHO:
THE LEGACY OF INDIAN MONISM
Hiroko Harada
(University of Illinois at Urbana-Champaign)
This comparative study analyses two outstanding poets of Japan
and England, Matsua Basho (1644-1694) and Samual Taylor Coler-
idge (1772-1834), in the light of early Indian philosophy.
A gap of time, space, and culture lies between Coleridge and
Basho, and numerous differences found in their poetic styles and
subjects naturally separate the two poets. Despite their extrinsic dif-
ferences there is undoubtedly a sort of internal and intrinsic common
ground where the two artists meet in terms of their goal of creation,
namely an organic poetry, achieved by the amalgamation of Self (the
poet) and Other (nature).
The concept of the organic nature of poetry is founded, in Coler-
idge, on his knowledge of Hellenism, especially Plotinus, and He-
braism, and in Basho, on his close attachment to Zen Buddhism. One
of the common aspects found in the oriental and occidental religious
and philosophical thoughts is the notion of monism which, according
to R. H. Blyth, directly or indirectly originates in early Indian
thought.
On this basis I would like to propose the thesis that both Coler-
idge and Basho present their individual theories of poetry standing
upon one common ground, India. I would like to closely examine
their theories and point out the comparable elements, primarily their
ideas of monism which establishes organic unity, and finally seek the
identity of their monism in Indian monism, in order to visualize the
invisible thread that tightly connect the English poet and the Japan-
ese poet.
INDEX OF PARTICIPANTS
Ahmed, Mariam
Anderson, Lloyd B.
Antonsen, Elmer H.
Anushiravani, Ali
Aronoff, Mark
Arora, Harbir
Ayyar, Indira
Bagchi, Tista
Bains, Gurprit
Bhatia, Tej K. 8, 1
Bhatt, Rakesh M. 7,
Bokamba, Eyamba
Boolchandani, Pushpa
Bubenik, Vit M.
Butt, Miriam
Chakraborty, Jayshree
Chandrashekar, S.
Cole, Jennifer
Davison, Alice
Deshpande, Madhav
D'Souza, Jean
Faulkner, Larry R.
Genetti, Carol
Gnanam, M.
Hamp, Eric P.
Harada, Hiroko
Herring, Susan
Hock, Hans Henrich
Hook, Peter E.
Jamison, Stephanie W.
Jayasuriya, Wilfrid
Joseph, Brian
Kachru, Braj B.
Kachru, Yamuna
Kamwangamalu, Nkonko
Kapoor, Kapil
Khokhlova, L. V.
King, Tracy Holloway
Kissock, Madelyn J.
Koul, Omkar Nath
Krishnamurti, Bh.
Loud, John A.
M., Lalitha
Mahajan, Anoop
Mahajan, Gyanam
8, 12, 17, 139
Marlow, Patrick E.
4, 61
8. 18
Mehrotra, Raja Ram
13, 62
3,4
Menon, A. G.
4. 63
7. 153
Mishra, Mithilesh K.
3. 11,
126. 150
8.98
Moag. Rodney P.
8, 12, 65
4. 104
Mohanty, Gopabandhu
11. 67
12, 100
Nadahalli. Jayashree
4.69
11, 19
Nadkami. Mangesh V.
11, 12, 70
12, 21
Nelson, Cecil L.
4. 133
1. 12. 127, 140
Nihalani, Paroo
7. 72
8, 22. 137, 142
Pandharipande, Rajeshwari
8
3. 11
.. 125.
145. 147
13. 24
Pandit. P. N.
8.73
4.26
Paolillo. John C.
12. 74
12. 27
Paranjape, C.
4. 26
12. 29
Pelletier. Rosanne
11. 77
4. 31
Rai, Alok
12, 80
7. 11. 32
Ramchand, Gillian
11. 81
3. 8, 34
Rau, Nalini
8, 84
7, 12, 36
Sadanand, Kamlesh
7, 85
4, 131
Sadanandan, Suchitra
4. 86
3
Satyanalh, T. S.
4. 87
13, 37
Scharf, Peter M.
12, 88
13, 39
Schaufele, Steven
7. 89
4. 40
Sharma, Krishna K.
12. 90
7, 156
Sharma, Rama Natha
12. 92
8, 42
Singh, Atamjit
8,93
3, 5, 7, 44, 148
Singh, Mona
13. 94
3,46
Sreedhar, M. V.
8. 96
7,48
Sridhar, Kamal K.
8, 141
8, 49
Sridhar, S. N. 4. 7. 8.
12. 31.98. 100
4. 50
Srivastav. Veneeta
13, 101
9
Subbarao. K. V. 3.
4. 12.
104, 106
4, 129, 132
Tickoo, Asha
3, 107
M.
Tikku, Girdhari
7.
151, 155
11, 12, 52, 53
Tsiang, Sarah
7. 108
13, 54
Valentine, Tamara
4, 135
13, 117
Verma, Manindra K.
11.
123, 128
12, 27
Vijayakrishnan, K. G.
11, 110
7, 55
Winters, Clyde A.
11, 112
3,46
Wu, Yongan
7, 154
3,4,5
Yatabe, Shuichi
8, 113
11, 56
Yoon, James
3
3, 106
Zakharyin, Boris A.
13,
115. 117
12. 57
Zgusta, Ladislav
3, 9
7. 59
Zidc, Norman
!
^, 11. 119
"te'".. i
Vol. 4:1
Spring 1974
Vol. 5:1
Spring 1975
Vol. 7:1
Spring 1977
Vol. 7:2
Fall 1977
Vol. 8:1
Spring 1978
Vol. 8:2
Fall 1978
Vol. 9:1
Spring 1979
Vol. 9:2
Fall 1979
Vol. 10.1
Spring 1980
Vol. 10:2
Fall 1980
Vol. 11:1
Spring 1981
Vol. 11:2
Fall 1981
Vol. 12:1
Spring 1982
Vol. 12:2
Fall 1982
Vol. 13:1
Spring 1983
Vol. 13:2
Fall 1983
Papers in General Linguistics $3.00
Papers in General Linguistics $3.00
Papers in General Linguistics $3.50
Papers in General Linguistics $5.00
Papers in General Linguistics $5.00
Papers in General Linguistics $5.00
Papers in General Linguistics $5.00
Relational Grammar and Semantics $5.00
(Editor: Jerry L. Morgan)
Papers in General Linguistics $5.00
Studies in Arabic Linguistics $5.00
(Editor: Michael J. Kenstowicz)
Papers in General Linguistics $5.00
Dimensions of South Asian Linguistics
(Editor: Yamuna Kachru) $5.00
Papers in General Linguistics $5.00
Papers on Diachronic Syntax: $5.00
Six Ca.se Studies
(Editor: Hans Henrich Hock)
Papers in General Linguistics S5.00
Studies in Language Variation: $5.00
Non western Case Studies
(Editor: Braj B. Kachru)
Vol. 14:1
Spring 1984
Vol. 14:2
Fall 1984
Vo.. 15:1
Spring 1985
Vol. 15:2
Fall 1985
Vol. 16:1
Spring 1986
Vol. 16:2
Fall 1986
STUDIES IN THE LINGUISTIC SCIENCS
The following issues are available:
Papers in General Linguistics $5.00
Language in African Culture and Society $5.00
(Editor: Eyamba G. Bokamba)
Papers in General Linguistics $5.00
Linguistic Studies in Memory of $5.00
Theodore M. Lightner
(Editor Michael J. Kenstowicz)
Papers in General Linguistics $6.00
Illinois Studies in Korean Linguistics $6.00
(Editor: Chin-W. Kim)
Vol. 17:1 Spring 1987 Papers from the 1986 South Asian $6.00
Languages Analysis Roundtable
(Editor: Hans Henrich Hock)
Papers in General Linguistics
Papers in General Linguistics
Papers in General Linguistics
Papers in General Linguistics
The Contribution of African Linguistics
to Linguistic Theory, Vol. 1
(Editor Eyamba G. Bokamba)
The Contribution of African Linguistics
to Linguistic Theory, Vol. 2
(Editor: Eyamba G. Bokamba)
Vol. 20:2 Fall 1990 Linguistics for the Nineties:
Papers from a Lecture Series in Celebration
of the Department's Twenty-Fifth Anniversary
(Editor: Hans Henrich Hock)
For earlier issues see inside cover
Orders should be sent to:
SLS Subscriptions, Deptartment of Linguistics
4088 Foreign Languages Building
707 S. Mathews
University of Illinois
Urbana, Illinois 61801
Vol. 17:2
Fall 1987
Vol. 18:1
Spring 1988
Vol. 18:2
Fall 1988
Vol. 19:1
Spring 1989
Vol. 19:2
Fall 1989
IN PREPARATION:
Vol. 20:1
Spring 1990