The Fergusonian Impact: In Honor of Charles A. Ferguson on the Occasion of his 65th Birthday. Volume 1: From Phonology to Society. Volume 2: Sociolinguistics and the Sociology of Language [Reprint 2018 ed.] 9783110873641, 9783110104875

Citation preview

The Fergusonian Impact From Phonology to Society Volume 1

Contributions to the Sociology of Language 42

Editor Joshua A. Fishman

Mouton de Gruyter Berlin • New York • Amsterdam

The Fergusonian Impact In Honor of Charles A. Ferguson on the Occasion of His 65th Birthday

Volume 1

From Phonology to Society edited by Joshua A. Fishman Andrée Tabouret-Keller Michael Clyne Bh. Krishnamurti Mohamed Abdulaziz

Mouton de Gruyter Berlin • New York • Amsterdam

Mouton de Gruyter (formerly Mouton, The Hague) is a Division of Walter de Gruyter & Co., Berlin

Library of Congress Cataloging-in-Publication Data

The Fergusonian impact. (Contributions to the sociology of language ; 42) Includes bibliographies. Contents: v. 1. From phonology to society - v. 2. Sociolinguistics and the sociology of language. 1. Linguistics. 2. Sociolinguistics. 3. Ferguson, Charles A., 1 9 2 1 - . I. Ferguson, Charles A., 1921- . II. Fishman, Joshua A. III. Series. P26.F45F4 1986 410 86-12436 ISBN 0-89925-214-1 (alk. paper)

CIP- Kurztitelaufnahme der Deutschen Bibliothek

The Fergusonian impact : in honor of Charles A. Ferguson on the occasion of his 65th birthday / ed. by Joshua A. Fishman . . . — Berlin ; New York ; Amsterdam : Mouton de Gruyter (Contributions to the sociology of language ; 42) ISBN 3-11-010487-3 NE: Fishman, Joshua A. [Hrsg.]; GT; Ferguson, Charles A.: Festschrift Vol. 1. From phonology to society. — 1986 — 545 S.

Printed on acid free paper © Copyright 1986 by Walter de Gruyter & Co., Berlin. All rights reserved, including those of translation into foreign languages. No part of this book may be reproduced in any form — by photoprint, microfilm, or any other means — nor transmitted nor translated into a machine language without written permission from Mouton de Gruyter, a Division of Walter de Gruyter & Co., Berlin. Typesetting: Druckerei Appl, Wemding. — Printing: Ratzlow-Druck, Berlin. Binding: Dieter Mikolai, Berlin. Printed in Germany.

Honoring Charles A. Ferguson It took almost a month for Charles Ferguson and me to realize that we were living next door to each other during the Summer Linguistic Institute of 1964 at Indiana University. We had each come to Bloomington, with our respective families, in order to participate in the Seminar on Sociolinguistics that the Social Science Research Council was sponsoring at the Institute that summer. Ferguson was the chairman of the Seminar, primus inter pares, the participants all being faculty members selected from a wide variety of disciplines. That summer I also took my first linguistics course, attending Ferguson's "Introduction to Linguistics" offered through the Summer Linguistic Institute. Prior to that summer we had communicated only once in writing and once over the telephone (although I had read his diglossia article and "knew who he was"). In the summer of 1964 we became neighbors, colleagues, students (each acknowledging the other as teacher) and close friends, roles that we have enacted, either repeatedly or continuously, during the past two decades and more. I start off in this personal vein because Ferguson is reacted to precisely in this vein by a legion of neighbors, colleagues, students-teachers and close friends all over the globe. While it is true that his scholarship is both impeccable and seminal, it is even truer, and even more to the point in capturing the spirit of what Ferguson means to all those who have been privileged to know both the man and his work, that no one has ever felt about him (as we have all felt about so many other great scholars) that because his scholarship was unblemished it was justifiable to overlook his blemishes as an individual. Anecdotes about Ferguson as a fine human being, concerned for the welfare not only of his friends but of humankind, and not only for the welfare of humankind but for ilesh and blood individuals, streamed into the Editorial Group's office as soon as invitations were sent out inviting participation in this Festschrift. These anecdotes deal with acts of kindness and charity, hospitality and help. Taken together they provide testimony to a wonderful human being about whom it can be said without exaggeration that by honoring him we honor ourselves. The primary feeling toward Ferguson that we all share is not only one of admiration but of affection, the two together making for a very rare tribute. However, it is ultimately Ferguson the scholar to whom this Festschrift is addressed. If Ferguson's status as an exemplary human being led to the rare phenomenon of busy scholars pleading with the

VI

Honoring Charles A. Ferguson

Editorial Group to be permitted to participate in the Festschrift, then Ferguson's status as an exemplary, multifaceted scholar led to the fact that these volumes are, of necessity, subdivided into a large number of different substantive areas. Indeed, it is difficult to find rivals to Ferguson in connection with either scope of interest or depth of contribution. Accordingly, scholars from half a dozen different areas of specialization recognize Ferguson as a mentor and, indeed, as a force who unites them and who points out their relevance for each other and for the total language-related enterprise. Child language (as well as "baby talk") and Arabic, universals and sociolinguistics, phonology and applied linguistics, language of religion and sports announcer register, in Fergusons' hands and mind these all come together and enrich each other and pose questions to each other. Ferguson has not only contributed to all of these areas but he has impacted them, adding to their depth, widening them, interrelating them in his own work and introducing them to each other via his influence on the thought of countless students and colleagues. Although this integrative role can be noted utilizing any one of the foregoing fields or topics as point of departure, it is particularly from the perspective of sociolinguistics that his magic touch becomes crystal clear. Ferguson has the rare capacity of seeing all language-related phenomena in societal perspective, even the deepest structure and the most universal aspects manifesting sociocontextual, sociofunctional and sociohistorical influences in his work. From the very genesis of modern sociolinguistic efforts in the late 50's, Ferguson has served as exemplar, umpire, godfather (even fairy godmother), gentle critic and proud defender of the field. Both intellectually and intuitively he recognized its potential for more realistic, more valid, more thoughtful and more provocative insights into language related behavior. His contributions to this fledgling field can hardly be exaggerated. His espousal of the field provided it with a man of distinction, with a "name" that opened doors, prompted support, suggested confidence and reassured doubting Thomases, from the very outset. That sociolinguistics could not only hold its own but advance and grow strong, doing so initially during the very period when the Chomskian revolution was orienting linguistics in a non-sociolinguistic direction, is in no small part due to Ferguson's quiet but very evident identification with it. However, Ferguson's contribution to the sociolinguistic enterprise goes far beyond narrow sociolinguistics per se. He has always embraced sociology as well as linguistics, political science as well as anthropology, education as well as phonology, history as well as mor-

Honoring Chartes A. Ferguson

VII

phology, economics as well as syntax, in short: a complete multi-level, multi-method approach to language in society, in order to build the broader understanding which is the new discipline's ultimate quest. This, indeed, is the truest measure of a scholar: one who does not cut down a field to his own size, to his own areas of personal expertise or to his own methodological competencies, but, instead, recognizes, encourages and acquires the widest possible repertoire of methods and substantive emphases. Ferguson has elevated the sociolinguistic enterprise by infusing into it his own catholicity of interests, his own acceptance of diversity of methods, his own openness to intellectual stimulation regardless of disciplinary origin. He has provided the field with a receptivity for both data and theory, for both applied and academic emphases, which, hopefully, it will never "outgrow". Finally, the strong moral dimension that underlies all of Ferguson's life and work, his individual nature as well as his scholarly nurture, can only serve to guarantee that the broad-gauged sociolinguistic enterprise which he has helped launch will never be put to distinctive, self-aggrandizing, inhumane and non-humanistic purposes. Our trust in the moral soundness of the field mirrors our trust in his own moral and ethical example to us all. These volumes could not have been realized without the collaboration of scores of individuals and a cooperative publisher, united by a sense of admiration for a wonderful human being and a fine scholar. The Editorial Group (Mohamed Abdulaziz, Michael Clyne, Joshua A. Fishman, Bh. Krishnamurti and Andrée Taburet Keller) is grateful to them all, and particularly for their help and self-discipline in assuring the publication of these books in time for their presentation to Professor Ferguson on the occasion of the celebration of his 65th birthday. We pray that these volumes may provide the same stimulation to the further growth of the language-related disciplines and studies that Charles A. Ferguson himself has so admirably provided - and will yet continue to provide - for many, many years. Joshua A. Fishman, Chairman Editorial Group

Contents Volume 1 From Phonology to Society: The Fergusonian Impact Honoring Charles A. Ferguson Table of Contents (Vol. 1; Vol. 2)

V IX

I. Linguistics, Child Language, Language and the Child Borrowed Sounds Ian Maddieson The Voice Warp: A Phonological Universal Merrit Ruhlen Agreement, Rule Interaction and the Phonology-Syntax Interface Marlys A. Macken

1 17

21

Velars and Final Consonants in Early Words Marilyn May Vihman and Judith G. Hochberg

37

On the Acquisition of Final Voiced Stops Eve V. Clark and Melissa Bowerman

51

And They had to Speak Any Way . . . : Acquisition and Creolization of French Robert Chaudenson

69

Sociolinguistics, Linguistic Description, Language Change and Language Acquisition Norman Denison

83

The Interaction of Societal Factors in Language Acquisition . . . John Piatt Babytalk as a Stereotype and Register: Adult Reports of Children's Speech Patterns Lise Menn and Jean Berko Gleason

99

Ill

X

Contents

A Note on Ferguson's Proposed Baby-talk Universals Lilith M. Haynes and Robert L. Cooper Children's Dispute and Negotiation Strategies: A Naturalistic Approach Muriel Saville-Troike Development of Negation in Tamil-Telugu Bilingual Children B. Lakshmi Bai Italian and Vernacular Writing in Southern Regions of Italy: A Semantic Approach to Children's Language Behavior Maria Antonietta Pinto Some Social Traits of the Bilingual Child's Personality Renzo Titone

127

135 . 153

171 183

II. Arabic and Languages of Africa The Hal Construction and the Main Verb in the Sentence . . . . Peter Abboud

191

The Possessive Predicate in North African Vernacular Arabic . . 197 Bernard Comrie The Verb 'See' in Arabic Dialects Alan S. Kaye The Comparative Study of Medieval Arabic Grammatical Theory Jonathan Owens Arabic After Diglossia Youssef Mahmoud Language Mentions in Jordanian Newspapers (The English-Language Press) Björn H. Jernudd Were There Egyptian Koines? Joseph H. Greenberg

211

223 239

253 271

Lexical Retention in Ethio-Semitic: Checking up on a Myth . . . 291 M. Lionel Bender

Contents

XI

The High Central Vowel in Amharic: New Approaches to an Old Problem Dick Hayward

301

A Southwest Ethiopian Language Area and Its Cultural Background Hans-Jürgen Sasse

327

Language, Literacy and Society: The Case of Ethiopia Abraham Demoz

343

The Metaphorical Base of Grammatical Categories in Ewe (West Africa) Bernd Heine and Ulrike Claudi

367

Simplicity on Command: On Pidginization of Swahili in Shaba (Zaire) Johannes Fabian

377

Aspects of Language Contact in Africa Edgar C. Polome

387

III. Applied Linguistics Descriptive and Functional Adequacies of Explicit Grammars . 401 D.N.S.Bhat Authentic What, Who(m), and When? David Crystal Activity Types and the Structure of Talk in Second Language Learning Susan M. Ervin-Tripp Contrastive Sentence Perspective: English and German William G. Moulton Some Sociolinguistic Aspects of Error Analysis (EA) in Foreign Language (FL) Teaching (With Special Reference to Error Evaluation in the Federal Republic of Germany) Gerhard Nickel Sociolinguistics and Translating Eugene A. Nida

411

419 437

455 471

XII

Contents

Dialogue in One Language, Sub-titles in Another: Some Alternatives for Learning Languages Wallace E. Lambert

479

Applied Linguistics and Language Planning: A Case Study . . . 517 Albert Valdman The Center for Applied Linguistics, 1972-1977 Rudolph C. Troike

537

Volume 2 Sociolinguistics and the Sociology of Language Honoring Charles A. Ferguson Table of Contents (Vol. 2; Vol. 1)

V IX

IV. Microsociolinguistics: Acts, Actors and Events Good Data in a Bad Situation: Eliciting Vernacular Structures . Walt Wolfram

3

Deferential Speech Acts and the Pragmatics of Politeness in Tamil: from Case to Aspect Harold Schiffman

23

The Concept of Politeness and its Formulas in the Roumanian Language Tatiana Slama-Cazacu

35

Language and Society: The Case of Japanese Politeness J.V.Neustupny

59

Questions and Non-Answers J.B.Pride

73

Riddling and Lying Participation and Performance John R. Rickford

89

Stereotype Foreigner Talk in German Literature Peter Mühlhäusler

107

Contents

Role of Hypercorrection in French Linguistic Change Nicole Gueunier Naming in the Kahmiri Paptfit Community: Sociolinguistics and Anthroponymy Braj B. Kachru

XIII 121

139

V. Sociolinguistic Situation and Bilingual Variation Optimal Strategies in Linguistic Games Jonathan Pool

157

Secondary Responses, or Tertiary, to Language by Linguists . . . 173 Winfred P. Lehmann Variation in Australian Kriol John Sandefur and John Harris

179

Sociolinguistic Varieties of Malay Asmah Haji Omar

191

Oral Languages and Modernity: The New Emergence of Communication in Creole Alain Bentolila

207

Forms and Functions of Bilingual Speech in Pluricultural Migrant Communities in Switzerland Georges Lüdi

217

The Polyglossic Spectrum William F. Mackey

237

The Dynamics of Plurilingual Situations André Martinet

245

Multilingualism in Algeria Dalila Morsly

253

The Value and Problems of Census Data on Languages: An Evaluation of the Language Tables from the 1983 Population Census of Mauritius Peter Stein On Being and Becoming Bilingual in India D. P. Pattanayak

265 279

XIV

Contents

Trade Functions and Lingua Francas Louis-Jean Calvet

295

VI. Language Planning: Corpus and Status Neologisms in Bislama (Vanuatu) D.T.Tryon

305

How Does the Way Women are Referred to and Described Affect Their Participation in Development and Democracy? . . 315 Joan Rubin Nathan Birnbaum's Third Phase: The Activization of Jewish Sanctity Joshua A. Fishman

325

From Lingua Franca to National Language: A Re-examination of the Origins of Standard Swahili Joan Russell

337

Pilipino and the Filipino's Renewed Search for a Linguistic Symbol of Unity and Identity Bonifacio P. Sibayan

351

Implications of Canadian Research for Promoting a Language Competent American Society G.Richard Tucker

361

The World Balance of Languages William W.Gage

371

VII. Diglossia: Particular Cases and General Re-examination BilingualismandaSenseof'Peopleness" C.J. Dodson

387

Bilingualism and Diglossia in their Mutual Relationship Giuseppe Francescato

395

Diglossia and Code Switching Carol Myers Scotton

403

Contents

XV

Diglossia Applied: Vernacular Mixing and Functional Switching with Bangui Yakomas Paul Wald

417

Diglossia in Bangladesh and Language Planning Problems . . . 431 Udaya Narayana Singh Diglossia in Bangla: A Study of Shifts in the Verbal Repertoire of the Educated Classes in Dhaka, Bangladesh Afia Dil

4SI

VIII. Language Contact, Spread, Maintenance and Death Language Contact Versus Language Conflict Peter Hans Nelde

469

Towards a Systematization of Language Contact Dynamics . . . 483 Michael Clyne Social Factors in Language Maintenance and Language Shift . . 493 Christina Bratt Paulston Language Contact and the Life or Death of KuguMuminh . . . Ian Smith

513

Remarks on a Case of Language Decay and Revival S. A. Wurm

533

Language Revival and Language Death Chaim Rabin

543

Gathering Language Data In Terminal Speech Communities . . 555 Nancy C. Dorian

IX. Charles A. Ferguson Bibliography, 1945-1986 . . . 591 X. Tabula Gratulatoria

599

L Linguistics, Child Language, Language and the Child

Borrowed Sounds Ian Maddieson University of California Los Angeles, California

1. Introduction and Preliminaries When one language is affected by another, one possible result is an addition to the phonemic inventory; new contrastive sounds may enter the affected language via the process of lexical borrowing. Of course, adaptation of loanwords to existing phonemic and phonotactic patterns is more common (Haugen 1950, Deroy 1956), and several linguists have suggested that this process provides a tool for investigating the nature of phonological representations (e.g. Hyman 1970, Kaye and Nykiel 1981). However, it is not unusual for borrowing to bring about some phonological realignment in the recipient language. Such a realignment might be one which affects only phonotactics, for example, introducing previously impermissible clusters (like /fm/ in English, introduced from Yiddish). More rarely, it is one which results in an extension to the phoneme inventory by the addition of previously unused phones or by the promotion to phonemic status of an allophonic variant of an existing phoneme. Language contact, including phonemic borrowing, has been discussed as an important factor in limiting the variability of phoneme inventory size (Maddieson 1984). Phonemic borrowing has also been put forward as the explanation for a suggested correlation between paucity of speakers and size of phoneme inventory (Haudricourt 1961). The question of how such borrowing proceeds is therefore of relevance to some quite general questions about language structure. Linguists seem to hold quite opposed views about which sounds are the ones that are likely to be borrowed. Some predict that the sounds that are the most distant from the phonetic resources of the recipient language are the ones most likely to be added to the inventory as

2

Ian Maddieson

a result of contact and borrowing. According to this view, exotic segments are only likely to be borrowed when they are rather markedly different from the sounds of the recipient language; otherwise, loanwords are adapted in shape so that the nearest matching existing segments in the language are used. When segment borrowing occurs, the motive might well be the prestige that is associated with the exotic pronunciation. Often cited as the prototype for this view is the borrowing of the voiced uvular trill / R / from French into German (e.g. Bach 1965). However, it is unclear if this is in fact a case where adoption of loanwords led the change (followed by innovation of the newly fashionable pronunciation for the existing German / r / phoneme), or whether a uvular / R / may have existed as a dialectal variant in German prior to the strong 17th century influence of French (see Brunt 1983 for a succinct summary of the issues). In discussion of a less ambiguous case, Deroy (1956) suggests a prestige motivation for the borrowing of / h / and / w / from Germanic into the Vulgar Latin of France. A different, but somewhat related hypothesis, has been put forward in a brief paper by David Cline (ms, 1984). Cline suggests that those segments which are more marked on a universal basis (i. e. less frequently found in the world's languages) are the ones more likely to be borrowed. What is common to these views is the opinion that borrowing is not influenced by the phonological structure of the recipient language, apparently also the view of Hagège and Haudricourt (1978). On the other side are those linguists who believe that the segments most likely to be borrowed are those which are easily fitted into existing oppositions. For example, a language with voiceless stops /p, t, k / and voiced stops /b, d / would be expected to easily borrow a voiced velar plosive / g / , since it already has a stop at the velar place of articulation and a voicing contrast in stops at other places. Jakobson (1931 [1964]) expressly proposed as a general principle of historical phonology that "les phonèmes étrangers que la langue s'approprie le plus aisément sont ceux qui s'incorporent dans les corrélations déjà existantes" (p. 323). He exemplified this principle with the case of the voiceless labio-dental fricatives in Russian. Given pre-existing voiceless fricatives /s, sj, x/ etc, and voiced fricatives /v, vj, z, zV etc, later nativizations of loanwords containing an original / f / introduce voiceless /{/ and / P / . (Which of these occurs in a given word depends on the vowel context in the donor language.) Obviously, no new features are introduced by the addition of these phonemes to Russian.

Borrowed Sounds

3

Along these lines, Maddieson (1984), assuming that a multi-lingual social setting is the normal situation for human languages, argues that an important factor in maintaining the typical phoneme inventory size of 30 or so segments (plus or minus, say, 5) is the limitation on inventory expansion imposed by the requirement that added segments must be capable of being "generated" by combinations of familiar features already in the recipient language. This process is seen as more explanatory than positing constraints on efficient communication as being responsible for the upper and lower limits on the size of a phoneme inventory. In order to decide how strongly the phonological structure of the recipient language affects which segments are borrowed, all of the consonant segments marked in the UPSID inventory as loan phonemes were examined in relation to the pattern of contrasts which previously existed in the languages. UPSID (the UCLA Phonological Segment Inventory Database) is a computer file containing the phonemes of 317 languages chosen to represent the genetic diversity of the world's extant languages (Maddieson 1984). It was inspired by the work of Ferguson, Greenberg and their colleagues on the Stanford Phonology Archive. In compiling UPSID, those segments in a language which only have phonemic status as a result of relatively recent borrowing were indicated by a special variable value. For the present paper, every language which has any consonantal segment marked with this value was examined. As is inevitable, some errors in the data file were detected during this process. These were corrected. Also, a few additional loan segments were identified - often those of an older layer of borrowing and a few cases from other languages outside the UPSID data base were also reviewed. One language, Zulu, was replaced by a close sister, Xhosa, for which more complete documentation on loan forms is available. The result is a set of 184 instances of borrowed segments to examine. Data are drawn from the references in Maddieson (1984) except where a reference is included here. Only consonants were examined because the greater number of parameters involved in consonants makes it easier to devise an intuitive metric for determining the relative distances of loan segments from the inventories they are entering. For this survey we will rely on a traditional phonetic framework of classification which divides consonant segments into series (voiceless plosives, voiced nasals, voiced fricatives, etc.) which have representatives at different places. Degree of similarity between segments will be assessed mainly by whether they share major manner and place fea-

4

Ian Maddieson

tures. Here "features" does not refer to phonological melody features in the technical sense (distinctive features) but rather to possession of articulatory or auditory phonetic properties. The framework is similar to that used for classifying segments in the UPSID database and described in Chapter 10 of Maddieson (1984). However, some phonetic distinctions will be disregarded. A labio-dental fricative will be considered to be matched in place by a bilabial stop, since bilabial fricatives are relatively rare (and labio-dental stops are impossible for many speakers and are not known to occur distinctively in any language). Palato-alveolar fricatives and affricates are considered matched in place by palatal segments of other types, such as nasals or approximants. Dental and alveolar places are not distinguished. Fricative and affricate series are not divided into sibilant and nonsibilant. Both / h / and / V are treated as sui generis, / h / being a voiceless vowel with consonantal function but a variable place of articulation, and / V a segment to which (most of) the usual series-generating contrasts cannot apply. On reviewing this data, it appears that the evidence is heavily in favor of the view that segments are most likely to be borrowed when there are already appropriate segments to promote the adoption of the new segment. In order to show that this result is not simply a reflection of the general frequency of segments, we will compare this data with a simulation. We have not examined the question of what social conditions favor segment borrowing, nor the issue of foreign influence on the direction of internal change in a language (e. g. as discussed by Petrovici 1957), and hence have no comments to make on these issues.

2. Classification of borrowed sounds We will classify the cases we have found in our data into groups ranging from those in-which the loan segment seems to be very close to the pre-existing resources of the languages, to those in which it is remote. Cases are sorted into six classes. Note that the form of the loan segment being compared is that which it has in the recipient language, not that of the donor language. We will return to this point briefly later. An appendix is attached in which the segments are listed, together with the segments which are considered to be the closest promotors of the borrowing in the given case.

5

Borrowed Sounds

(a) Close segments Many loan segments "fill a gap" in the inventory, that is, they add to an existing series at a place of articulation where an existing minimally different segment in another series already occurs. Such cases form class 1. A representative case is the Togo Remnant language Lelemi. This language had a stop inventory consisting of : b

t d

k g

Loanwords from European colonial languages have added a new voiceless stop segment / p / . This does not create a new segment type, since voiceless plosives already exist in the language; moreover a segment which is minimally different in voicing from / p / and shares its place of articulation is present, namely, / b / . A similar but in some ways more interesting case is found in Yoruba, which had the stop inventory: b

t d

k g

kp gb

Earlier loans containing / p / were borrowed with / k p / or / b / substituted, e.g. /kpotogi/ "Portuguese", /kpou/ "pound", /kôbà/ "penny" ( < "copper"). More recent loans, mostly from English (now a much more widely spoken language in the community) retain / p / , e.g. /pilo/ "pillow", /pid3Dt/ "Peugeot". These more recent loans fill the gap in the voiceless plosive series at the bilabial position. An example involving fricatives is the borrowing of /%/, mostly through loans from French, into German. German previously had three sibilant fricatives:

s

z

J

The previously existing voiceless palato-alveolar fricative acquired a voiced partner, parallel to the voiced alveolar fricative / z / , following the adoption of loan forms such as /3eni/ "genius" ( < French génie). Not all such cases involve simply a difference in voicing from an existing segment. Mazatec, for example, has borrowed / p / from Spanish. Although the language previously had no bilabial stops, it had a bilabial nasal / m / and fricative /(}/, and voiceless stops / t / and / k / . Tseshaht (Nootka) has borrowed uvular ejective stops / q ' / and / q w ' / from Makah and/or Nitinat (Jacobsen 1969). Previously the language had uvular plosives and velar ejectives, including labialized variants. Thus

6

Ian Maddieson

both place and manner are represented in existing phonemic segments. Approximately half of the loan phoneme cases surveyed consist of gap-filling moves of this kind ( 9 0 / 1 8 4 cases). In the appendix these cases are marked with the number 1. In class 1 cases, the newly added segment differs only minimally from two (or more) of the existing segments in the language, one with respect to place and one with respect to manner. We have placed in class 2 those cases where the borrowed segment is minimally different from (at least) two existing segments along two different dimensions of contrast, but where the new segment creates a new manner series. A relatively common type of example in class 2 is a language which has borrowed voiced stops when it had pre-existing voiceless stops and voiced nasals. An example is Tarascan, which had: p m

t n

k q

This language has borrowed / b , d, g / from Spanish. Other types of borrowings in this class include voiced fricatives taken into languages which already had voiceless fricatives and voiced stops at the same place, such as / z / taken into Hindi which already had / s / and / d / , as well as affricates taken into languages where none previously existed, but which had the separate "components" available. An example of the latter is Bashkir, which has / t / , / s / and / J 7 but no original sibilant affricates. Loanwords from Russian have introduced / t s / and / t J 7 , e.g. in /tsikl/ "cycle" ( < Russian /tsikl/), / t j a n / "tub, vat" ( < Russian / t j a n / ) . There are 19 examples of class 2. In classes 1 and 2 the segments borrowed are close to the existing phonetic resources of the language. The third class of cases where the borrowed segment seems to be appropriately considered to be close to the existing resources of the recipient language are those where borrowing results in the promotion of a well-defined allophonic variant to phonemic status. A well-known example of this type is the rise to phonemic status of the voiced allophone [v] of the fricative / { / in English. In Old and early Middle English, voiced fricatives were allophones of the voiceless fricatives in medial position between voiced segments. During the Middle English period the complementary distribution of [f] and [v] was disrupted because loanwords from Old French with initial / v / came into English with the initial voiced fricative maintained (e.g. vein, vine, veil); also words were incorporated into East Midland dialects from Southern dialects in which all fricatives had become voiced (e.g. vat, vixen). These

Borrowed Sounds

1

borrowings predate the loss of final vowels which caused / z / and / d 7 to become phonemic in pairs like "house (n.), house (v.)" and "breath, breathe". A similar allophony for stops in Central Zoque was disrupted by loanwords from Spanish with voiced stops in initial positions and in medial clusters (Wonderly 1946), and Russian loanwords in Chuvash have established both voiced stops and voiced fricatives as phonemic. A complex situation exists in the New Guinea language Telefol (Healey 1964). The phoneme / b / appears as [b] initially, as [b] or [fi] medially and as lightly aspirated [p] finally. Loanwords from Neo-Melanesian with initial or medial [p] are establishing a contrast between / p / and / b / . Since Telefol has / t / (which appears as [t] in all environments) the loan segment / p / can be regarded as belonging in class 1 as well as in class 3 by our definitions. A somewhat similar situation exists with regard to the loan segment /%/. Natively, [g] occurs as the result of coalescence of / k / and / l / at morpheme boundary; loans with initial / g / have introduced a contrast between / k / and / g / . Since both / k / and / b / exist, this case also belongs both to class 3 and class 1. The third loan segment in Telefol is regarded as being / d / . The earlier situation seems to have been complementary distribution of [d], [1] and [r], with [d] initial, [1] final and [r] medial. This phoneme has been treated as A / rather than / d / since the pattern of allophony is quite different from that of / b / ; for example, there is no final devoicing. Although different phonemic solutions are possible it is nonetheless true that it is the borrowing of items with / d / in medial position that leads to the establishment of contrast between / l / and / d / . Given pre-existing / b / and / t / , the borrowing of / d / belongs also to both class 1 and 3. In Hausa, / r / , and / V have become established as phonemes through borrowing from Arabic, Tuareg and elsewhere. The situation with respect to / r / is complex (Newman 1980), but essentially it seems that the occurrence of trilled / r / could be predicted from positional and morphological facts. Loans such as /bera/ "mouse" ( < Arabic fa^ra) have extended the range of / r / to positions where it has no predictability. As for P], it originally occurred as a predictable onset to an initial vowel. It could not be regarded as an allophone of any other consonant phoneme. However, it could appear medially as a result of reduplication of a vowel initial item, e.g. from /aikaa/ "to send" the iterative verb /a ? aikaa/ is formed, parallel to such forms as /gamu/ —• / gaggamu/ "to meet". Hence there was fertile ground for the adoption of / V as the phonemic segment, representing both Arabic / V and

8

Ian Maddieson

A / . This case is a little different from the other class 3 examples, but it is included with them since it also involves an existing non-phonemic segment being promoted to phonemic status. There are 14 examples in class 3 in our data. It seems likely that more might be found if fuller descriptions were available of some of the less well documented languages, or if more exhaustive research was conducted. Note that this would have the effect of adding to the group of examples in which a close segment is added to the recipient language.

(b) Remote sounds The remaining classes cover those loan segments which are relatively remote from the pre-existing resources of the recipient language. In class 4 are the remaining cases in which the recipient language has added a segment with a previously unrepresented manner but at a place where existing segments occur. In this class, the condition that there are two segments that are only minimally different is not satisfied. The largest number of cases in class 4 involve addition of stops which create new series. Tiwa and Tunica, for example, have both borrowed voiced stops / b , d, g/. These languages already have voiceless stops (aspirated in the case of Tunica) at these three places but they only have nasals at two of them, lacking a velar nasal. Hence the loan segment / g / is regarded as more distant than / b / and / d / , which fall into class 2, whereas / g / is in class 4. Po-Ai (a Northern Tai dialect from Yunnan) has only one native series of stops, which are voiceless unaspirated. It has borrowed a complete series of aspirated segments /p h , t h , tf h , k h / from Chinese. Somewhat similarly, new series of voiceless aspirated and breathy voiced stops have entered Telugu from Hindi and other Indo-Aryan languages or as part of learned Sanskritic vocabulary. Sources differ on the question of whether these two "aspirated" stop series are actually pronounced differently from the plain series in anything other than a rather artificial high style, and different conclusions might be reached about the extent to which they really represent established loan segments. However, they have been included in our counts in class 4. There is no question that the clicks in Xhosa and other Southern Bantu languages are fully integrated. Standard analyses of Xhosa suggest that there are 12 click phonemes (Davey 1975). These entered the language as part of the loan vocabulary from Khoisan languages

Borrowed Sounds

9

which, by some estimates, amounts to some 40 to 45% of the Xhosa lexicon (Louw 1977). Several other fricative and affricate segments that fall into class 1 have also been borrowed. The clicks fall in class 4. Most of the identified loanwords appear to come from Central Khoisan languages, especially Nama and !Ora (Korana), although this appearance may be partly an artifact of the much greater knowledge of these languages than of other possible donor languages (Louw 1974). In Xhosa, there are three types of clicks - dental, alveolar lateral, and palatal - in each of four series - plain voiceless, aspirated voiceless, breathy voiced, and nasalized voiced. Korana has 24 clicks, four types in six series, indicating that Xhosa has substantially reduced the complexity of the segmental inventory it was selecting from. Specifically, it has merged the separate dental and alveolar series (Xhosa has no dental/alveolar contrast) and eliminated distinctive glottalized and velaraffricated click releases, and made several other realignments (Lanham 1962, Davey 1975). This case is important in that it shows that even when contact between languages is extremely intimate (Ownby 1981), there is not a wholesale borrowing of unfamiliar segments. Several of the other cases in class 4 involve the borrowing of liquids. It is unusual for a language to have more than one lateral approximant or one rhotic segment and these are most commonly found with dental/alveolar place of articulation. Hence such segments are typically borrowed only if the recipient language lacks an original rhotic or lateral. Consequently, after borrowing has occurred, these segments form a series by themselves, although at a place of articulation which is already represented. Among languages which have borrowed a liquid in this fashion are Swahili, Hausa, Mazatec, Tiwa, Moxo, Guarani and Wappo. Wappo has created both plain and laryngealized / r / ' s from the loan process. Guarani had no original laterals and has borrowed both / l / and /X/ from Spanish. There are 54 examples of class 4 in total. The fifth class of cases are those in which a loan segment introduces a new place of articulation to an existing series where no pre-existing segment occurs at that place of articulation. Loans of this sort are quite rare. Only four cases are observed in the data file. These examples are the voiceless uvular stop and fricative segments / q / and / / / in Hindi-Urdu, a language which had no previous uvular segments, and the palato-alveolar fricative /J"/ borrowed into both Finnish and Ostyak. More will be said below about the Hindi case. The sixth class of cases are those where the borrowed segment has both a place and a manner which is different from any pre-existing

10

Ian Maddieson

segment in the recipient language. Hindi-Urdu again provides examples of this class, having / V and / k / . Note that / V and / h / have been defined as unique segment types, and hence whenever these are borrowed they fall into class 6, unless there are facts similar to those reported for Hausa above. The other language in our sample which has borrowed / V is Bashkir, and Georgian and Rumanian have borrowed / h / . In addition, Guarani has borrowed / t f / into a system which had neither affricate nor palatal segments. Thus there are 6 cases in class 6.

3. Summary of segment borrowing patterns The pattern in this data is quite clear. There are very many more instances where the segment borrowed fits closely in with the existing resources of the language. The number of cases in classes 1-3 is 123, whereas the number in classes 4-6 is 61. From this, it is appears that segments closer to the existing resources of the recipient language are more likely to be borrowed. Moreover, a relatively high proportion of the segments in classes 4-6 are somewhat dubiously integrated into the languages they have joined. The aspirated stops in Telugu have already been mentioned. Hindi-Urdu as described by Kelkar (1968) has 9 borrowed segments, of which six are limited to particular communal speech styles outside "common core" Hindi (Ohala 1983). These six are precisely the ones which fall into classes 4, 5 and 6. We have also noted a number of cases where it seems that a borrowed segment is itself modified in a way that brings it closer to existing sounds in the recipient language. The examples include Hausa / V corresponding to Arabic / c / , elimination of the dental/alveolar contrast between clicks entering Xhosa, and Sinhalese accepting // where the donor languages have /{/. We also note the relative rarity of the borrowing of sui generis segments such as / V and / h / . Both of these facts strengthen the impression that languages do not normally borrow segments that are far afield from the phonetic material that is already at hand.

4. Simulation of segment borrowing Despite the impression that emerges from the data we have surveyed, can we be sure that there is in fact any special selectivity going on in the process of borrowing loan segments? The numbers of loans of the

Borrowed Sounds

11

different types reported in this paper might well be a reflection not of preferential borrowing of sounds that fit into the recipient language's phonological pattern, but instead simply the result of the overall frequency across the world's languages of particular sounds and segment inventories. Frequency is necessarily one factor in the borrowing results since no sound can be borrowed from one language to another if the sound is already common to both languages. Segment borrowings can only occur if there are unmatched sounds. Obviously sounds have unequal chances of being borrowed because of this. Similarly, segments which are candidates for borrowing have unequal chances of being close or remote with respect to possible recipient languages because of the differing frequency of various patterns of inventory structure. It is hard to devise a real-life test that addresses the question of whether the observed patterns of sound borrowings could result from chance. This is partly because it is difficult to demonstrate that a segment had an opportunity of being borrowed but was in fact rejected. Examining the inventory of languages which frequently figure as donors, such as English, Spanish and Arabic provides some clues in that segments that are less likely to be matched in either place or manner such as the interdental fricatives of English, the palatal lateral of Spanish, and the pharyngeal and pharyngealized consonants of Arabic rarely appear as loan segments in these forms. However, in a very large number of the cases it is likely that the form of the donor language from which borrowings were being made was itself already modified in ways that may have eliminated these rarer types of speech sounds. Moreover, it is hard to be sure that there were appropriate lexical items with the sounds of interest that were good borrowing candidates. Because it is difficult to test the hypothesis that the results of our survey are not due to chance in the field, it was decided to perform a simulation of borrowing. We can show that the actual borrowing patterns are different from those which would be predicted by segment frequency and patterns of inventory structure. The simulation was done in the following way. First a subsample of all of the consonantal segments in UPSID was constructed by selecting every thirty-fifth entry in the file. This procedure was used in order to increase the likelihood of including in the sample consonants that are members of classes with a distribution that is disproportionately weighted in certain language families. The procedure produced a sample of 207 segments. These segments were then randomly ordered. Next the 317 language names were sorted into a random sequence and

12

Ian Maddieson

the first 200 language names in this sequence were matched one by one with the first 200 segments in the sample of 207. No language was matched with a segment which had originally been drawn from itself. In this way 200 confrontations of "foreign" segments and existing phoneme inventories were simulated and the resulting patterns examined. Of these 200,115 were exact matches, that is, the foreign segment (or a close enough relative) already existed in the elected recipient language. This provides us with a rough estimate of the proportion of cases where such matches might occur in real life situations. Three cases where a long consonant was involved in the comparison were counted as matched by their short counterparts. The remaining 85 cases were possible segment borrowings. In this simulation the greatest number of loan segments fall into class 4, that is, they would add a member of a new series at a place that is already represented in the language, but without two minimally different promoting segments being present. Recall that the most common class in the real data is the gap-filling type (class 1). Although class 1 is the next most frequent in the simulation, it is only about one third as frequent as the class 4 type, and is about equally frequent as the class 6 type, that is, cases where the segment has no close promoters in the pre-existing inventory. This class is one of the smallest in the real language data. No members of class 3 were identified in the simulation, although some undoubtedly exist. Overlooking their existence might distort our conclusions - we will describe a way of correcting for this below. The percentages of cases in each class in the simulation are tabulated in Table 1 below and compared with the percentages in the real data.

Table 1. Comparison of simulated and real loan segment data. Totals and percentages for classes 1 and 3 in the real data include 3 Telefol examples which are included in both classes. Real data

Simulated data

Class 1 Class 2 Class 3 Class 4 Class 5 Classò

n

%

n

%

16 7 0 44 4 14

18.8 8.2 0.0 51.8 4.7 16.5

91 19 14 54 4 6

48.9 10.3 7.6 29.3 2.2 3.3

Borrowed Sounds

13

The two sets of percentages were compared using the %2 test and found to be significantly different (p. < .001). Comparable levels of significance were also obtained for a comparison between real and simulated data grouping together classes 1-3 and 4-6. The comparison was repeated for both the six classes and the grouped data with loan segments in Lake Miwok (Callaghan 1964) included in the real language data. Callaghan suggests that there are 18 segments in Lake Miwok which might have been borrowed from Patwin (Wintun), Pomo or Wappo. Of these, 13 are classed as "remote", making this language comparable to Xhosa in the nature and degree of foreign effect on its inventory. Lake Miwok was not included in the main data because Callaghan seems tentative in making the claim that these segments are in fact all borrowed. With Lake Miwok included, significance levels were still above .001. Finally, the test for the grouped data was repeated with 7.9% added to the total for classes 1-3 and subtracted from the total for classes 4-6 in the simulated data. This represents an estimate of the correction that might be needed to allow for the overlooking of cases which should have been put in class 3 in the simulation. The correction factor is a little higher than the percentage of class 3 examples found in the real data, and assumes all cases would involve reclassification from groups 4-6. Even with this correction there is a significant difference at the .001 level between the simulated and the real data.

5. Summary The patterns observed in our survey of loan segments and the comparison between the real data and the simulation indicate that languages have a strong tendency to reject as loan segments those segments that are remote from their existing phonetic resources, or to modify them in such a way that they are no longer remote. In other words, languages do behave in the way that is hypothesized in Maddieson (1984) as contributing to maintaining the typical phoneme inventory size.

References Bach, A. 1965. Geschichte der deutschen Sprache (8th ed). Heidelberg. Brunt, R. J. 1983. The influence of the French Language on the German Vocabulary. De Gruyter, Berlin and New York.

14

Ian Maddieson

Callaghan, C. A. 1964. Phonemic borrowing in Lake Miwok. In W. Bright (ed.) Studies in Californian Linguistics, University of California Press, Berkeley and Los Angeles: 46-53. Cline, D. ms. 1984. What does a sound have to do to get borrowed into a language? Department of Linguistics, University of California, Los Angeles. Davey, A. S. 1975. Some Aspects of the Phonology of the Noun in Xhosa M. Litt, thesis, University of Edinburgh. Deroy, L. 1956. L'emprunt linguistique (Bibliothèque de la Faculté de Philosophie et Lettres de L'Université de Liège, Fascicule 141). Les Belles Lettres, Paris. Hagège, C. and Haudricourt, A-G. 1978. La phonologie panchronique. Presses Universitaires de France, Paris. Haudricourt, A-G. 1961. Richesse en phonèmes et richesse en locuteurs. L'Homme 1:5-10. Haugen, E. 1950. The analysis of linguistic borrowing. Language 26: 210-231. Healey, A. 1964. Telefol Phonology (= Pacific Linguistics, Series B, 3). Australian National University, Canberra. Hyman, L. M. 1970. The role of borrowing in the justification of phonological grammars. Studies in African Linguistics 1:1-48. Jacobsen, W. H. 1969. Origin of the Nootka pharyngeals. International Journal of American Linguistics 35:125-153. Jakobson, R. 1932 [1964]. Prinzipien der historischen Phonologie. Travaux du Cercle Linguistique de Prague A: 247-267. Translation into French by J. Cantineau printed as an appendix to Principes de Phonologie, N. S. Troubetszkoy, Klincksieck, Paris: 215-336. Kaye, J.D. and Nykiel, B. 1981. Loan words and abstract phonotactic constraints. Studia Anglica Posnaniensia 13: 21-42. Kelkar, A.R. 1968. Studies in Hindi-Urdu 1. Introduction and Word Phonology. Deccan College, Pune. Lanham, L. W. 1962. The proliferation and extension of Bantu phonemic systems influenced by Bushman and Hottentot. Preprints of papers for the Ninth International Congress of Linguists, Cambridge, Mass.: 264-270. Louw, J. A. 1977. The linguistic prehistory of the Xhosa. In W.J.G. Möhlig et al., eds. Zur Sprachgeschichte und Ethnohistorie in Afrika Reimer, Berlin: 127-151. Louw, J. A. 1974. The influence of Khoe on the Xhosa language. Limi 2/2: 45-62. Maddieson, I. 1984. Patterns of Sounds. Cambridge University Press, Cambridge. Newman, P. 1980. The two R's in Hausa. African Language Studies 17:77-87. Ohala, M. 1983. Aspects of Hindi Phonology. Motilal Banarsidass, New Delhi. Ownby, C.P. 1981. Early Nguni history: linguistic suggestions. Suid-Afrikaanse Tydskrift vir Afrikatale, Byblad\9%\ : 60-81. Petrovici, E. 1957. Kann das Phonemsystem einer Sprache durch fremden Einfluß umgestaltet werden ? - Zum slavischen Einfluß auf das rumänische Lautsystem. Mouton, The Hague. Wonderly, W. L. 1946. Phonemic acculturation in Zoque. International Journal of American Linguistics 12:92-95.

Borrowed Sounds

15

Appendix This appendix contains a listing of the 184 loan segments used as the database for this paper. Loan segments are listed by language, with languages grouped by family. Each entry consists of the following: (i) a number from 1 to 6, indicating the class of the borrowing accirding to the classification given in the text, (ii) the segment borrowed, in the form of the most representative allophone it has in the recipient language, and (iii) the closest segments) in the recipient language to the borrowed segment. A dash (-) indicates that no segment is considered to be close. A segment enclosed in [ ] is an allophone; if preceded by ~ it is an allophone of the segment before Entries for the same language are separated by a semi-colon. Irish 1 th (th, d) English 3 v ( f ~ [ v ] ) German 1 3 (z, /) Lithuanian 1 f(v, s);l P (v>, sJ);lx(k, s); 1 x>(k),s>); 1 y (g, z); 1 y j ( g J , 1 ts (tj, dz); 1 tsJ (tj, tJ) Russian 1 f (v, x); 1 P(vJ, sj) Romanian 1 z (s v); 1 3 (/, v); 6 h (-) Hindi-Urdu 5 q(k, -); 6 >(-); 1 f (p, s); 2 z (s, d); 5 %(s, -); 6 a (-); 1 f ( t f , s); 4 3(d3,-);4f(n,r) Bengali 1 s (t, 5) Panjabi 1 f (p, g); z Qj, d) Sinhalese 1 $ s); 1 J (tj, g); lq (n, g) E. Armenian 1 f (v, z) Ostyak 5 J (s, ts) Cheremis 4 pi (p, -); 41> (t, -); 2 ts (t, s); 1 f (p, s); 4 ^ (s, -) 1 x (y, s) Komi 1 ts (t, tf); 1 f (v, s); 1 x (k, 9) Finnish 1 b (p, $}); 1 g (k, d); 1 f (v, g); 5 J & - ) Hungarian 1 d3 (tf, dz) Azerbaijani 1 kh (k, ch) Chuvash3 b(p, ~[b]); 3 d(t ~[dD; 3 g ( k ~[g]);3z(s ~[z]); 3 3 ( f - f e l ) ; 1 ts (tj,s);lf(p,s) Yakut 1 p (p:, b); 1 f (p:, $); 4v(b, s); 2 (& d); 1 J (tj, 5); 4 3 (d 3 , a) Bashkir 1 f (p, 0); 4 v (b, 9); 2 z (s, d); 4 3 a , d); 2 ts (t, s); 2 t j (t, /); 6 > (-) Khalaj 1 f (p, s); 1 v (b, z); 4 d3 (tj, d); 1 x (k, s) Mongolian 1 ph (b, t h ); 1 kh (g, t h ); 1 $ (b, s); 4 p (b, s); 41 (1, s) Evenki 1 p (b, t); 2 z (s, d) Lelemi 1 p (b, t) Yoruba 1 p (b, t) Swahili 4 r (l,h (t); 4 (d); 4 f i Q); 4 >fi (1); 4 ID - (f, n); 4 tq (c', n); 4 ^ (1, n); 1 kx' (k', t f ) ; 1 x (kh, s); 1 y (g, z); 1 ts' (t\ tD Kanuri 1 x (k, s) Amharic 1 p (b, t); 1 p' (b, t')

16

Ian Maddieson

Hebrew 1 tj (ts, J); 4 d 3 (d, J); 1 3 (J, z); 1 g (n, g) Hausa 41 (d); 3 r (r ~[r]);3 >([>]) Po-Ai 4 ph (p, -); 4t h (t, -); 4 tj h (tj, -); 4 kh (k, -) Telefol 1,3 p (b ~ [p], t); 1, 3 d (t, b, [d]); 1,3 g (k, b, [g]) Sab'an 1 g (k, d) Zoque 3 b (p ~ [ b ] ) ; 3 d ( t ~[d]);3g(k ~[g]) Tzeltal 1 {} (t, b); 1 g (kh, b) Otomi 21/ (t, J); 41 (r) Mazatec 1 p (t, m); 4 b ("g, p); 1 ó (s, P); 1 y (k, P); 4 r (1) Tseshaht (Nootka) 1 q' (q, k'); 1 qw' (qw kw'); 1 % (q, x); 1 %w (qw, x*) Quileute 1 g (k, d) Yacquild(t>b);lg(k,b);lf(p,s) Tiwa 2 b (p, m); 2 d (t, n); 4 g (k, -); 4 r (1) Tarascan 2 b (p, m); 2 d (t, n); 2 g (k, q); 1 f (p, s); 1 ji (rj, 5); 11 (J, r) Tunica 2 b (p\ m); 2 d (t\ n); 4 g (kh, -); 1 f ( p \ s) Wappo 2 b (p, m); 2 d 0, n); 4 g (k, -); 1 f (p, s); 4 r (1); 4 r Q) Moxo 2 b (p, m); 2 d (t, n); 1 f (s, P); 1 D (k, n); 41 (r) Guaraní 6 t / k a ) ; 1 f (v,£); 1 d (£, v); 1 xw (x, kw); 1 yw (y, kw); 4 Í (f); 41 (r); 4 X

(0 Ticuna 4 f(p, -); 4 s (t, tj) Barasano 4 s (t, -) Telugu 4 ph (p, -); 4 th (t, -); 4 kh (k, -); 4 b (b, -); 4 d (d, -); 4 d (d, -); 4 g (g, -); 11§ (ts, s) Georgian 6 h (-) Kabardian 1 < (h, k) Burushaski 1 f (p, s); 1 % (q, s)

The Voice Warp: A Phonological Universal Merritt Ruhlen Palo Alto, California

Charles Ferguson once remarked that linguists may be divided into two groups, those who study a single language (or language family) and those who are concerned with human language as a general phenomenon. The kinds of problems investigated and the methodology employed by these two groups are usually quite distinct and there is often remarkably little cross-fertilization between the two camps. This is unfortunate because the two approaches are not in conflict and in fact complement each other. As I sought to show in Ruhlen (1979), an investigation of the origin and evolution of the French nasal vowels carried out in a universal context arrives at quite different conclusions from those reached by scholars who limit themselves to the French case alone. This is not really surprising because without a universal context it is impossible to tell which aspects of the development represent universal tendencies and which are idiosyncratic to French. I would like to discuss here an example of a phonological universal that has until recently gone all but unnoticed because, I suspect, it is not usually apparent if one restricts one's view to a single language. One of the most common phonological contrasts in the world's languages is that between voiceless and voiced stops. Such a contrast was found in 72% of the 693 language sample used in Ruhlen (1976), though like most linguistic traits it is more common in certain geographical areas and genetic families than in others. The usual pattern is for the contrast to be present at each place of articulation, as in English: (1)

p b

t d

k g

Merritt Ruhten

18

Gamkredlidze (1975) observed, however, that if there is a gap, or hole, in the system it is far more likely to be p or g than any of the other four. There thus appears to be a natural affinity between voicing and front articulation, on the one hand, and voicelessness and back articulation on the other. This asymmetry parallels Greenberg's (1970:126-27) observation that, of the two basic varieties of glottalic consonants, implosives are typically voiced with front articulation (e.g. 6, thunder (Lass, 1980). However, such a characterization seems misleading if the voiceless stop children produce under these circumstances is the outcome of articulatory difficulty. See also Macken and Barton (1980). This strategy should go hand in hand with the use of apparently voiceless stops for voiced targets. Whether it actually does is not well documented. For some solutions to difficulties in voicing initial stops, see Macken and Barton (1980), and Allen (1985).

On the Acquisition of Final Voiced Stops

65

References Allen, George D. (1985) How the young French child avoids the pre-voicing problem for word-initial voiced stops. Journal of Child Language 12,37-46. Clark, Eve V. (1982) Language change during language acquisition. In Advances in developmental psychology, ed. by Michael E.Lamb & Ann L. Brown, 171-195. Hillsdale, New Jersey: Lawrence Erlbaum Associates. Dodd, Barbara (1975) Children's understanding of their own phonological forms. Quarterly Journal of Experimental Psychology 27,165-172. Ferguson, Charles A., & Farwell, Carol B. (1975) Words and sounds in early language acquisition. Language 51,419-439. Fey, Marc E., & Gandour, Jack (1982) Rule discovery in phonological acquisition. Journal of Child Language 9,71-81. Hooper, Joan B. (1977) Substantive evidence for linearity: Vowel length and nasality in English. In Papers from the thirteenth regional meeting, ed. by Woodford A. Beach, Samuel E. Fox, & Shulamith Philosoph, 152-164. Chicago, Illinois: Chicago Linguistic Society. Ingram, David (1975) Surface contrast in children's speech. Journal of Child Language2,287-292. Kiparsky, Paul, & Menn, Lise (1977) On the acquisition of phonology. In Language learning and thought, ed. by John Macnamara, 47-78. New York: Academic Press. Lass, Roger (1980) On explaining language change. Cambridge: Cambridge University Press. Lass, Roger (1984) Phonology. Cambridge: Cambridge University Press. Maccoby, Eleanor E., & Bee, Helen L. (1965) Some speculations concerning the lag between perceiving and performing. Child Development 36,367-377. Macken, Marlys A. (1979) Developmental reorganization of phonology: A hierarchy of basic units of acquisition. Lingua 49,11-49. Macken, Marlys M., & Barton, David (1980) The acquisition of the voicing contrast in English: A study of voice onset time in word-initial stop consonants. Journal of Child Language 7, 41-74. Menn, Lise (1976) Pattern, control, and contrast in beginning speech. Unpublished PhD dissertation, University of Illinois. Slis, I.H. (1970) Articulatory measurements on voiced, voiceless, and nasal consonants. Phonetica 21,193-210. Smith, Bruce L. (1979) A phonetic analysis of consonantal devoicing in children's speech. Journal of Child Language 6,19-28. Smith, Bruce L., & Westbury, J. R. (1975) Temporal control of voicing during occlusion in plosives. Paper presented at the 89th meeting of the Acoustical Society of America, Austin, Texas. Smith, Neilson V. (1973) Acquisition of phonology: A case study. Cambridge: Cambridge University Press. Velten, H. (1943) The growth of phonemic and lexical patterns in infant language. Language 19,281-292.

66

Eve V. Clark and Melissa Bowerman

A p p e n d i x 1 : D a t a f r o m D a m o n (1 ;8,1—1 ; 1 0 , 1 4 ) Words are listed in order of production, with subsequent tokens and their forms noted according to the age at which they were observed. (All forms are listed with their original transcription from the diary record.) The emergence of adult-like production is marked with an asterisk (*) following the age (once stable, no subsequent instances are listed). Forms produced immediately after an adult production of the target word are marked by I (imitation). A capital R marks a spontaneous repair. rug [rArjk] 1 ;8,1(I)/1 ;8,8/l ;9,14/[rAg] 1 ¡10,17*/ pig [pirjk] 1 ;8,1/1;8,8/ bug [bAijk] 1 ;8,2/[bAg] 1 ¡10,16* ladybug [leidibAgk] 1 ;8,5/[beibibAi)k] l;8,12/l;10,16/[beibibAg] 1;10,19*/ 1;10,23*/ bag [bAijk] 1 ;8,2(I)/[baeijk] 1; 8,9/1; 8,27/1 ;9,1/1 ;9,29/[baeg] 1 ¡10,21*/ 1 ¡11,15*/ big (earlier [bi]) [biqk] 1 ¡8,4/1 ¡9,14/1 ;9,21/[big] 1 ;10,19*/1 ¡10,28*/ [birjk-R-big] 1 ;10,22*/[big] 1 ;10,28*/1 ¡11,4*/ [bigk-R-big] 1 ;ll,ll*/[big] 1 ;11,16*/1 ¡11,18*/ big stick [birj t°ik] 1 ;8,16(I)/big ball [biqk bol] 1 ¡9,25/ big car [biqk ka] 1 ;9,29/big door [big do] 1 ¡10,20*/ frog (earlier [ro]) [fwoqk] 1 ;8,7/[fwok-R-fwoqk] 1;8,8/ [fwoi)k/fwDr)] 1;8,26/ [f w og]l ¡10,19*/1 ¡10,22*/ dog (earlier and usually [do]) [doijk] 1 ¡8,11/1 ;8,22/[dog] 1 ¡10,14*/ leg (earlier 0e]) [lerjk] 1 ;8,12/[leg] 1 ¡11,17*/ flag [fwaerjk] 1 ;8,19/[raeij] 1 ;9,14/[fwaeg] 1 ¡10,28*/ plug [pworjk] 1 ¡8,19/1 ;8,26(I)/1 ¡9,2/1 ;9,16/[pog-R-pAg] 1 ¡10,19*/ Bob (earlier [bo]) [bom/bomp] 1 ;8,19/[bob] 1 ;10,19*/1 ¡10,21*/ bib (earlier [bi], [bi/bib/bep] 1;7,24) [bim] 1 ¡8,21/1 ¡8,22/1 ¡9,21/1 ¡10,16/[bib] 1 ¡10,23*/l ¡10,24*/ scab [k°asm] 1 ¡8,21/1 ¡8,25/[k°aemp] 1 ;9,14/[k°aem/k°aeb] 1 ;10,28*/[skasb] 2 ¡0,9*/ God [gDm]l ¡8,22(1)/ lid (earlier [ji]) [jim] 1 ;8,22/[jin] 1 ¡8,25/1 ;9,14/[lid] 1 ¡11,8*/ crab [k°aem] 1 ¡8,22/1 ¡8,25/[k°aeb] 1 ;9,5*/[k°aemp-R-k°aeb] 1 ;9,5*/[k°aeb] 1 ;10,25*/[k°aem/k°aeb] 1 ¡10,28*/ sled [sjem/sem] 1 ¡8,22/ bed (earlier [be]) [bem] 1 ;8,23/[bed] 1 ;10,23*/1 ¡11,10*/ rhubarb [ruba:m] 1 ¡8,23/1 ¡9,28/ cupboard [kaegum] 1 ¡8,26/ tub [tAm]l ¡8,26/ Rod [mm] 1 ¡8,29/ red (earlier [re]) [rem/ren] 1 ¡8,29/1 ;10,12/[ren/red] 1 ;ll,ll*/[red] 1 ¡11,23*/ road (earlier [row]) [rowm] 1 ;9,0/[rowd] 1 ¡11,17*/ knob (earlier [no]) [nom] 1 ¡9,0/ read (earlier [ri]; once [rid] (I)) [rim] 1 ;9,5/[ri] 1 ;10,9/[rid] 1 ¡11,5*/

On the Acquisition of Final Voiced Stops

67

bird (earlier [b3]) [b3n] 1 ;9,13/[b3d] 1 ;10,27*/ fig [nrjk]l;9,14(1)/ word [w3n] 1 ;9,14/1 ;9,28(I)/ wood [wun] 1 ;9,21/ ride (earlier [wai]) [waid] 1 ;9,21*/ Herb (earlier [h3], always major form) [h3m] 1 ;9,23 (very emphatic)/l ;9,25/ 1 ;9,28/[h3b] 1 ;10,13 * / l ;10,14*/ head (earlier [he]) [hen] 1 ;10,15/[hed] 1 ;11,6V showerhead [sawshed] 1 ;11,16*/ cloud (new word) [kawdj 1 ;10,21*/ outside (earlier [awtsai]) [awtsaid] 1 ;11,4*/1 ;11,5*/ bread (earlier [bwe]) [bwed] 1 ;11,13 */ shutted [sAtid] 1 ;11,13*/ sad [saed] 1;11,16*/

Appendix 2: Data from Eva (1;5,14-1;10,7, with occasional forms to 2;0,5) As in Appendix 1, the words produced are listed in order, by age, together with subsequent observations noted from daily records or from summaries made every two weeks, in their original transcriptions. Nasalized forms produced immediately after an adult utterance of the same word are marked by I (imitation). Although few nasalized forms are noted after 1 ;10,7, the summaries suggest they persisted in some words at least to age 2;0. egg [aeijk] 1 ;5,14/1 ;6,0-l ;6,14/1 ;6,15-1 ;7,0/1 ;7,0-l ;7,14/1 ;7,15-1 ;8,0/[aeq] 1;8,28/1;10,7/ eggs [aerjks] 1 ;6,12/1 ;6,13/[yaeks-R-aer)]) 1 ;8,28/ seed (earlier [di] 1;5,6) [dint] l;5,15/[din] 1;6,9/1;6,15-1;7,0/1;7,0-1;7,14/ 1;7,15-1;8,0/ dog (earlier [das/da]) [daq] 1 ;8,1/1 ;8,4/ hot-dog [-darjk] 1 ;5,16/1 ;5,20/[-dasq] 1 ;6,28/ hot-dogs [darjs] 1 ;6,28/ frog [orj] 1;5,18 (I)/[baij] 1 ;6,4(I)/[bai]k] 1 ;6,0-1 ;6,14/1 ;6,17/1 ;7,0-1 ;7,15/ l;7,15-l;8,0/[farj]2;0,5/ frogs [barjks] 1 ;6,11/1 ;6,13/1 ;7,0-l ;7,14/ bag (earlier [baey] 1;5,23) [baerjk] 1 ;5,20/[baek-R-baeq] 1 ;6,4(I)/[b8egk/baeij] 1 ;6,15-1 ;7,0/[baen] 1 ;7,11(I)/ bean bag [binbaen] 1 ;6,15-1 ;7,0/[binbaeqk] 1 ;7,0-l ;7,14/ bib (earlier [baba] 1 ;4,27/[biba] 1;5,8) [bim] 1 ;5,29(I)/[bimp] l;6,10/[bim] 1 ;6,18/1 ;6,19/[bimp] 1 ;7,4/l ;7,21/ bibs [bimps] 1 ;6,19(I)/1 ;6,19/1 ;10,7/ bed (earlier [bae]) [baen] 1 ;5,30/l ;6,4/ good [gun] 1;5,30(I)/1;6,0-1 ;6,14/1 ;6,10/1 ;7,0-1 ;7,14/[gunt] 1 ;7,23(I)/[gun] 1 ;10,7/

68

Eve V. Clark and Melissa Bowerman

bead (earlier [bi], [bi:t] 1 ;4,23/l ;5,8/l ;5,9) [bin] 1;6,0-1;6,14/1;6,15-1;7,0/ 1 ;7,0-l ;7,14/[bi:n] 1 ;8,27/[bi:t] 1 ;8,27/[bin] 1 ;10,7/ beads [bins] 1 ;6,11/1 ;6,19/1 ;8,16/1 ;8,27/l ;8,28/ pig [birjk] 1 ;6,4(I)/[bin] 1 ;8,25/ food [bun] 1 ;6,5(I)/[fun] 1 ;10,7/ slide (earlier [daey] 1;5,1/1;6,0-1;6,15) [dayn] l;6,15-l;7,0/[daynts] 1;7,0/ [dayn] 1 ;8,4/ mad [mae:n]l;6,18(1)/ did [din] 1 ;6,21(I))/[dint] 1 ;7,8(I)/[din] 1 ;10,22/[dint] 1 ;10,27/ kids [gins] 1 ;6,27/l ;7,0-l ;7,15/1 ;7,15-1 ;8,0/[gits] 1 ;8,28/ Brad [baen] 1 ;6,28/l ;7,29/l ;8,27/ outside [dayn] 1 ;6,29/l ;7,17/[awtdayn] 1 ;8,27/[awtsayn] 1 ;10,27/ dead [daent]l;7,0(1)/ big ([bik] 1;5,14/1;5,15) [birjk] 1 ;7,2(I)/[biq] 1 ;7,5/[birjk] l;7,9/[birj] 1;7,12/ [bin/big k] 1 ;7,29(I)/[biij] 1 ;8,0/[birjk] 1 ;8,17/[bik] 1 ;9,5/ Big Bird [bin bun] 1 ;8,7/[bin bun] 1 ;8,17/[bin bun] 1 ;8,28/ big horsie [big owsi] 1 ;8,16/big bear [birj be] 1 ;8,17/ washrag [abaen] 1 ;7,6(I)/ leg [laerj] 1 ;7,6(I)/ legs [aerjks] 1 ;8,18(I)/1 ;8,19/1 ;10,17/ bread (earlier [byae] 1 ;4,19) [baen] 1 ;7,6(I)/1 ;7,8/l ;7,15-1 ;8,0/ ride [wayn] 1 ;7,15/1 ;7,23/l ;8,1/1 ;8,16/1 ;8,23/ Bob [bamp] 1 ;7,16/ hide [ayn] 1 ;7,18(I)/1 ;8,19/ read [min] 1 ;7,18/[win] 1 ;8,18/1 ;8,28/l ;9,15/ hard [ayn] 1 ;7,23(I)/1 ;7,24/ need [nin] 1 ;8,l/[nint] 1 ;10,7/ hold [own] 1;8,4/1 ;8,7/[ayn] 1 ;8,21(I)/ bird (earlier [bi] 1 ;3,2/l ;4,15/1 ;4,19) (Big) Bird [bun] 1 ;8,7/l ;8,17/1 ;8,28/ Hoag (name [hoig]) [oirjk] 1 ;9,27(I)/ head [aen] 1 ;10,6/ bad [baen] 1;10,7/ crib [kimp] 1 ;10,7/ hood [u:n] 1 ;10,7/ red [rent] 1;10,22(1)/ Richard [witsun] 1 ;10,26/

And They had to Speak Any Way... : Acquisition and Creolization of French. Robert Chaudenson University of Provence Aix-en-Provence

"A language in the normal process of diachronic change may be modified in ways similar to pidginization. The term pidginization however seems best limited to those situations in which the structural change is relatively rapid and results from the communicative interaction of native and non-native speakers of the common source target language" (Ch. Ferguson and Ch. De Bose, 1977, p. III). This remark is true indeed, but could be made more accurate by substituting the word "variation" to "diachronic change" because in a certain number of cases, homologies are observed between pidginization (also creolization) and variation, without any of the temporary "solutions" reached at in the course of synchronic variation being ever integrated for good into the diachronic changes of the language. This could be illustrated by the fact that French Creoles have "institutionalised", in their own systems, changes affecting the very same structures which French speakers have been in vain trying, for many centuries, to alter in French. Without specifically devoting to these questions (though he often alludes to them), Ch. Ferguson has obviously perceived the link existing between variation and linguistic changes, or between wilful simplifications of any language (baby talk, foreigner talk), whether stemming from child language or broken language, and the facts highlighted by pidginization and creolization. Although my own research-work has almost exclusively dealt with the Indian Ocean French Creoles (and more specifically, Reunion Creole), it seems possible to generalize, up to a certain extent, the conclusions which have been drawn in that particular instance, at least to the languages which have developed in identical or similar conditions (and that could be the situation of most of the other French Creoles).

70

Robert Chaudenson

To avoid entering into too much detail, I will summarize my hypothesis about the genesis of Reunion Creole by saying that it is the result of the acceleration and radicalisation of the "autoregulative propensities" which have been evidenced by the general study of variation in French (this has been observed particularly in all cases where normative "pressure" weakens or completely disappears). This process is conditioned by specific historic, social and cultural parameters, as described in Chaudenson, 1977, and elaborated in Chaudenson, 1984. This hypothesis, on which my thesis is based (Chaudenson, 1974), has been extended later on by the observation of similar propensities existing in the successive grammatical systems constituted by child learners of French as a maternal tongue (Chaudenson, 1978) and by learners of French as a foreign language in a "natural" context (Chaudenson, 1985). One of the major problems is to find how and why Creole systems have gained autonomy from French. This question can be divided into 2 parts: the first one, which I shall not tackle here, concerns the determination of the "frontiers" which separate the two systems somewhere on the continuum running between them. In my opinion, one of the appropriate tools to perform this task consists in the elaboration of a model designed to account for variation in French, and which would integrate all the forms and modalities of that language. This "zero French" would make it possible to discriminate between what is specifically French and what is not, and therefore to set the farthest limits of French (Chaudenson, 1984,1985). The second problem consists in identifying the conditions and sociolinguistic processes bringing about the "autonomisation" of Creole languages by means of an evolution actualizing the potential tendencies present in many varieties of French. I would like to tackle the second question through a close study of the linguistic situation in Bourbon (which was later to become Reunion), by confronting the facts observed from a sociolinguistic point of view with the definition given for pidginisation by Ch. Ferguson and Ch. De Bose (1977), taking into account the "conditions" laid down by D. Bickerton (1981) concerning the use of the term "creole". The period we are dealing with (1665-1725) suggests a first question. D. Bickerton proposes to restrict the use of the term "creole" to cases where the following conditions are satisfied: 1. The pidgin (the original one) must not have existed for more than the time span of one generation. 2. A creole must develop in a population composed of a maximum of 20% of narrative speakers of the dominant language, and the remain-

71

And They had to Speak Any Way.

ing 80% must consist of linguistically heterogeneous speakers (Bickerton, 1981, p. 4). The demographic evolution of Bourbon during the first fifty years of colonisation is as follows (see table I below). (1665: French settlement in Bourbon island) Table I

1686 1704 1709 1713

Whites

Negroes

Halfcastes

Total

82 423 507 633

82 311 387 538

105 This category no longer appears

269 734 894 1171

It is clear from these figures that the second condition is not fulfilled (it is not either in the case of the other French Creoles, particularly in Martinique and Haiti). Moreover, the fact that certain Creoles are issued from other (prior) Creoles, spread by diasporas, must be taken into consideration (Chaudenson, 1981). The first condition has a greater bearing on the matter under discussion, because, as far as the second is concerned, it is obvious why Bickerton, who is so anxious to see the "genetic bioprogram" at work, insists on defining a set of conditions in which no linguistic norm is prevalent (However, these conditions do not seem to be indispensable, at all the more so as the application of these restrictions eliminates some of the languages on which Bickerton himself, up to a certain extent, grounds his analysis) (Chaudenson, 1982, p. 83-84). The notion of "generation" is itself ambiguous and misleading. Why not say x years (20 or 25)? The use of this term contributes towards distorting the already very vague picture one has of the situation, by giving the impression that the speakers of any given generation quietly wait on some starting line until the moment comes for them to bound forward and move on in step. That the children could possibly have played a part in creolization is doubtless, but how can we imagine a "first" generation of children "inventing Creole" in that linguistically heterogeneous society, simply because there is no norm strong enough to prevent the resurgence of the genetic bioprogram? The documents available in Reunion are quite accurate, and they make it possible to find out what the birth-rates were since the very beginning of this colonial society (mainly, as one would expect, by means

72

Robert Chaudenson

of civil-status registers). The first child was born in the island two years after the landing of the first settlers: Estienne Pau (7.8.1667). The second one is Estienne Bellon (12.10.1667), the third, Jeanne Mirebeau (4.7. 1968). The latter is the first halfcaste born in Bourbon; if both parents of the first two children were French, Jeanne Mirebeaud's mother, Anne Randranar was Madagascan. From 1678 to 1688, the number of christenings recorded increases sensibly, (see Table II). Table II Year

Births

1688 1689 1690 1691 1692 1693 1694 1695 1696 1697 1698

8 16 16 20 10 18 18 23 21 22 23

Total population

324

488

(source: Barassin, p.233)

It must be noted that in 1689 (year in which an interesting census was made from our point of view), a certain number of mothers were themselves born in the island, and therefore were "creoles": 10 out of a total of 45: six were born of French parents, two were halfcastes ("French Indo-Portuguese women), and three halfcastes who were "French-Madagascan". This is caused by the small number of women, resulting in early marriage in the island. In "one generation" (less than 25 years) there already are children whose parents are "creoles" themselves.

This rapid growth of the "creole" population (people born in the island, irrespective of colour) is an important factor: in 1704, one third of the total number of slaves is "creole" (i. e. born in the island). A second factor to be taken into consideration is the large fraction of the servile population introduced into the island, which consisted of children or adolescents. This policy is in no way uncommon, as during the whole period of the slave-trade, young slaves have always been appreciated for their docility and greater adaptability to their new French "culture".

73

And They had to Speak Any Way.

In the case of Bourbon, the age of immigrants (mostly Madagascan) is unknown for the period 1669-70 (22 newcomers in all), but more accurate data are available from 1696 to 1715: (see table III). Table III Year 1696 1699-1701 1702 1705 1711 1713-1714 total

Slaves aged less than fifteen

Slaves aged more than fifteen

20 19

1 17

-

-

5

-

59

18 (see note 2) 26

103 (see note 3)

62

-

Slaves whose age is unknown -

7 15 (see note 1) 8 -

30

Note 1: These slaves are aged ten to twenty-seven, but the detailed age-distribution is not specified. Note 2: These are entered under the heading "adult", but one is considered an adult at an early age in this society and a certain number of them are very likely to be less than fifteen. Note 3: Eight children are born of mothers who were already pregnant when they landed on the island.

If only the slaves whose age is known for certain are considered, it can be seen that between 1696 and 1714, more than 60% of them are less than fifteen, the youngest being one and a half years old, and we will not mention those who arrived, while being still in their mothers' wombs! This is a factor which must undoubtedly be taken into account in any sociolinguistic investigation of the situation prevailing in the new colony. If we refer to the period 1686-1689 (allowing for "one generation" to lapse by, as the French settlement takes place in 1665), we can have a relatively accurate picture of the situation of the various families, as well as of the organisation of the society as a whole. In 1690, the white population (which comprehends, as already seen halfcastes) comprises 55 men, 35 women and 113 children, while the negro population consists of 63 men, 26 women and 24 children, summing up to a total of 316 people (Barassin, p. 196). According to J. Barassin, the family distribution can be established as in Table IV. If we now focus our attention on the families as units, the situation is as follows (the linguistic homogeneity or heterogeneity of these families being taken into consideration). (See Table V)

74

Robert Chaudenson

Table IV Family

Number of children

Total

7 French

34

48

2

1

5

French (born in Bourbon) 3

3

9

Dutch 1

French 1

0

2

French

"French-Indo Portuguese" (Halfcaste) 12

34

68

"French-Indo Portuguese" (Halfcaste) 1

1

3

"French-Indo Portuguese" (Halfcaste) 1

1

3

French 14

Madagascan 14

44

72

Dutch

French-Indo Portuguese (Halfcaste) born in Madagascar 1

0

2

French-Indo Portuguese (Halfcaste) born in Bourbon 1

0

2

Franco-Madagascan (Halfcaste) born in Bourbon 1

0

Father

Mother

French

French (from France)

7 French 2 French 3

12 Venitian 1 Dutch 1

1 Dutch 1 French-Indo Portuguese 1

(from Bourbon)

Bachelors (French) Madagascan 8

2 12

Madagascan 8

24

40

And They had to Speak Any Way...

75

Table V Number of families

Total number of members

15 8 12

62 40 58

14

72

A - "French speaking" families B - "Malagasy speaking" families C - French/French-Indo-Portuguese families D - French/Franco-Madagascan families

A rough typology of the communication situations and languages in use can now be arrived at, as shown in Table VI. Table VI Communication Intra-family Inter-family

Situation

A

B

C

D

Par./Par. Par./Child. Child./Child. A

F F F/F

M M/F F F

F/F F/F F F/F

F/F F/F F F/F

XXX

F

F/F'/M

XXX

F/F

XXX XXX

B

F

XXX

C

F/F

F

XXX

D

F/F

F/M

F/F

XXX XXX

F=French; F = French approximations; Af=Malagasy; ID= Indo-Portuguese (?) Par= Parent; Child. = Children; A, B, C, D= as in table V above.

In the case of group B, I have extrapolated, as far as intrafamily communication is concerned, the data collected in Europe on the linguistic behaviour of immigrant families. It appears from this source that a large majority of children, from the very first generation, no longer use the language of their parents. The remainder of the population (bachelor Europeans or slaves), fits very well into the structure of this "société d'habitation", which, as a matter of fact, is characterised by the existence of small-scale economic units (the "habitation" is an agricultural concern established on a concession allotted to an "habitant"). The "habitation" generally comprises one family (of the A, B, C, type as above) and a very limited

76

Robert Chaudenson

number of slaves, often less numerous than the members of the family). To illustrate this, here are two of the forty-one cases found in Firelin's census, both of which have been taken at random: "Guy Roger, married to a white woman, a Creole from this island [born in Bourbon], has four Children, two boys and two girls, and possesses a negro slave [... ]". "Mr.Cané-Talhoit [... ] married to a Frenchwoman has four Children [... ] and also two negroes, of which one is married and has one Child, the other being an old negress from India" (Quoted from J. Barassin, p. 198). Table VI illustrates the marked centripetal character (towards French) of this situation where French and approximate forms of French overwhelmingly dominate the scene (these forms would be called "broken French'V'foreigner French" in Ferguson's terminology)This fact is not astonishing in itself, and is confirmed to a lesser extent in many works (the "subtractive bilingualism" put forward by W.Lambert, which characterises minority communities in situations where they are socially dominated: in these cases the second language is learnt to the detriment of the first language (1975), J. Bloom and J. Gumperz also make it obvious that in a multi-lingual society, the dominant language is exclusively used for communication between groups while the "intra-group" relations are characterised by codeswitching, etc ). Taking into consideration the sociolinguistic factors is essential as it makes it possible, right from the start, to distinguish between a pidgin (in the traditional sense), which is a medium of communication between two different social and/or ethnic groups (this is the case, for example, of the "baragouin" in the Antilles) and those attempts at approximating French, which are specific to these colonial "sociétés d'habitation". From this, one can draw the conclusion that terms such as "pidginisation" and "decreolisation" are much too vague in meaning, and result in subsequent confusions. A more attentive scrutiny confirms the very dominant position occupied by French as a target-language. Without entering into an excessively detailed study, the two main aspects of this "société d'habitation" can be emphasised: 1 - There is an utter supremacy of the French speaking group, as they dominate ethnically (whites/negroes), socially (master/slave), numerically and sexually (man/woman). Sexual domination is an aspect which has been overlooked but which is undoubtedly, fundamental, as

And They had to Speak Any Way.

77

the Madagascan or Indo-Portuguese wives of the settlers were bound to aim at becoming culturally and linguistically "assimilated" to their husbands' status, in order to gain, from then on, a higher position, clearly distinct from that of their fellow countrywomen who were still enslaved. 2 - Close and constant interaction takes place between whites and negroes, who, in spite of the very strong domination of the former, share nearly the same everyday living conditions which were near to those known by Robinson Crusoe and Man-Friday. This permanent interaction is set into a complementary and asymetric relation which can be formalised, as shown below, in Table VII, which is inspired by Bateson (1977, p. 115). Table VII A - Whites

B - Negroes

Domination Assistance Exhibitionism

Submission Dependence Voyeurism

Obviously, the third point particularly concerns verbal communication (the whole table is as relevant to the relationship between a Frenchman and either a Madagascan woman or an Indo-Portuguese woman in a mixed family, as it is to the relationship between the master's family and the slaves living on the "habitation"). As a result of the development of colonial agriculture on an industrial scale (sugar-cane and/or coffee) profound changes are going to affect the original situation characterising the "société d'habitation". It will be replaced by the "société de plantation", the latter being often mistaken for the former. The most striking change (which, however, is no more than a consequence of the evolution described above) is the large scale increase in the negro population, owing to the huge masses of slaves arriving in the island because a lot of labour was needed to work in the new cultures. In Bourbon, the slave population, in 1735, is seventeen times larger than it was in 1708 (Scherer, 1966, p. 26). The aggravating imbalance between the negro and white populations must not be considered simply as regards its quantitative impact, but much more as a factor modifying the binary interaction described above. In Bateson's terminology, the situation turned into a ternary interaction system, where A stands for the whites (inclusive of halfcastes assimilated to that group), B for the Creole or "creolised" slaves (present in the island for a period long enough for them to have integrated the creole

78

Robert Chaudenson

group), and C for the freshly landed slaves ("bozals"). The system now •ends to shape itself into a hierarchy of the serial type, because the mainstream of the relations between A and C flows through B.G.Bateson illustrates this by the relation model between parents, nurse and 'hllvlren, in Great-Britain: "The functional role of the intermediary consists mainly in educating an inculcating to the third group the proper standards of behaviour, so that, when in contact with the first group, it adopts the adequate comportment" (1977, p. III). In the "société de plantation", the role of the intermediary is more complex and extensive, but it is a relationship of the same type. The "noirs de bande", or "d'atelier", or "de pioche" (slaves working in the fields) are no longer, as during the first stage of colonisation, in direct contact with the whites (reputedly Frenchspeaking). These slaves are now living in a sort of "peripheric" linguistic zone, in which the medium of communication is constituted by those "approximations of French" which will become the target-language of the new slaves. In turn, this transitory state of the language will be the input in a new process in which foreigner talk and broken language will be at work, resulting in an acceleration and radicalisation of the evolution of that medium away from French, which is at this stage no more than a distant linguistic model for many of the speakers of the group; see Table VIII for a diagrammatic representation of that situation. Table VIII A B C

Groups

Medium(s) used

A: Whites (inclusive of those who are assimilated to this category) B: Creole and creolised slaves C: New slaves (bozals)

-F: French -F': Approximations of F -F -F": Approximations of F'. Language(s) spoken in their country(ies) of origin.

This has many consequences, concerning either the history of certain Creoles, or the theorisation of the creolization process.

And They had to Speak Any Way.

79

In the Indian Ocean, the genetic relation between Reunion and Mauritian creóles can thus be explained (in spite of the obvious structural differences between them). As a matter of fact, when the beginning of the colonisation of Ile de France (which was eventually to become Mauritius) takes place, the Whites and slaves from Bourbon who take part in it, are assigned the duty of acting as intermediaries in a ternary system in which the element A is constituted by the newcomers (Frenchmen intending to settle in the new colony), the element C consisting of slaves freshly introduced in the island. This method is not at all uncommon, as it had already been put into practice when Martinique had been colonised, as noted down by Du Tertre (1667): "[en juillet 1635] Monsieur d'Esnambuc résolut de ne plus différer de prendre possession de la Martinique [...]. Pour réussir dans cette entreprise, il prit environ 100 hommes de vieux habitants de l'Isle Saint Christophe, tous gens de main accoutumés à l'air, au travail, à la fatigue du pays qui en un mot n'ignoraient rien de tout ce qu'il faut faire pour défricher la terre, la bien cultiver, y planter des vivres et y entretenir des habitations". In the case of Ile de France, the Whites and slaves brought in from Bourbon undoubtedly introduced the language they had in common, that is French in a variable degree of approximation (1720 being at the end of the "société d'habitation" stage). As a result of this, these approximations of French, already characterised by certain of the features which will later on become those of Reunion creóle, are introduced in Ile de France, and accordingly will be observed in Mauritian creóle. This hypothesis is the only one making it possible to explain both the presence of very original and specific linguistic characteristics shared by these two creóles, as well as the existence of structural differences presently opposing these two languages in other respects. This second aspect of the question can be accounted for by the fact that the second stage ("société de plantation") takes place after 1720 (therefore each of the languages undergoes changes separately from the other), and also by the differences in the ulterior evolution of each of the two, which will cause them to diverge more and more from each other (for additional detail on this, Chaudenson, 1981). From a theoretical point of view, it can be seen that the conditions set down by D. Bickerton for a language to be classified as a "creóle" (see supra) are not at all justified. Most of the creóles have appeared as a consequence of European colonisation in the sixteenth and seventeenth centuries, and most must have had an evolution pattern comparable to that of Reunion creóle (this is true in any case for the Lesser

80

Robert Chaudenson

Antilles and Haiti). Besides it appears that these conditions are, to a large extent, pointless: they have been laid down, as it had already been mentioned, to make it possible for a first generation of children to put freely their genetic bioprogram to work, in such a highly heterogeneous linguistic situation that the dominant group, speaking European languages, cannot at all exert its influence. Right from the start, this is a thoroughly theoretical point of view. How can this'Tirst generation" be accounted for, and how can its "model" spread in a society which is mostly rural, and where the "habitations" are nearly self-sufficient? Why postulate a linguistic "coup" on the part of these young speakers, when the same results can be explained by a much more progressive evolution. This is rendered credible as it is based on the remarkably convergent restructurations operated on French, which are found in its ancient and/or popular forms, in baby talk, in foreigner talk, as well as in the transitory grammars of native children or of non native learners in natural language-acquisition situations. It would be difficult to make, on that question, an overall assessment which would be more synthetic and to the point than it is in A. Tabouret-Keller's article (1979). My present aim is simply to confront a few theories with a historical situation which is known with quite a considerable degree of precision, so that it becomes apparent that the use of the term pidgin is rather inadequate for the situations where, in theory, a "preocreolization" stage took place, while at the same time, there was a more or less convergent evolution of the forms of French which were already marginal varieties. This evolution was due to the French speakers adapting their performance to their interlocutors (foreigner talk) and to the attempts at approximating French on the part of non-native speakers (broken French), as well as to the restructurations operated on French by children during their acquisition of the speech faculty, in a situation characterised by a linguistic continuum running from a popular and/ or regional variety of French to quite distant approximations of that system. It can be said in conclusion that the conditions set by D. Bickerton to the use of the term "creole" are excessively restrictive, as they result in the rejection of languages belonging to that category - this is evidenced by the fact that Bickerton himself partly bases his analysis on languages which, according to his own definition, are not Creoles. These conditions are also needless, because the part played in creolization by children born in this type of sociolinguistic situation can be accounted for without making of them the "inventors", ex-nihilo, of a new system. Besides, it seems that the pidginization-creolization mod-

And They had to Speak Any Way .

81

el, considered as having been true since R. Hall, should be questioned. The hypothesis just presented causes us to consider, in the creolization process, less the nativisation of the approximations of French (this process can start almost immediately after colonisation begins, as children are born in mixed families), than the "shift" in the target-language label from French to its approximated forms - (I am on purpose using this term rather than "pidgin", which I would prefer to see restricted to communication situations of a different nature, sociolinguistically speaking). Consequently, the creolisation process implies the acquisition of those approximations, at the same time, as a first language by the children bom in groups B and C (possibly in A also, relayed by the "nénènes", i.e. nurses), and as second language by the newcomers to the colony, who, during the second stage of creolization, no longer have a constant and direct contact with French (former target-language in the "société d'habitation"). I would like to thank Didier de Robillard who kindly accepted to translate this article; quotations from works translated into French have sometimes been translated back into English, as they were not readily available in the original text.

References Barassin, Jean, 1953, Naissance d'une chrétienté. Bourbon des origines jusqu'en 1714, Saint-Denis de la Réunion, Cazal. Bateson, Georges, 1977, Vers une écologie de l'esprit, Paris, Seuil. Bickerton, Derek, 1981, Roots of language, New-York, Karoma. Bloom, J.P. and Gumperz, John, 1972 (rep.) "Social meaning in linguistic structures : Code-switching in Norway" in J. Gumpers and D. Hymes, Directions in Sociolinguistics. Chaudenson, Robert, 1974, Le lexique du parler créole de la Réunion, Paris, Champion. Chaudenson, Robert, "Toward the Reconstruction of the Social Matrix of Creole Language" in Valdman, Albert, 1977, Pidgin and Creole linguistics, Bloomington, 259-276. Chaudenson, Robert, "Créole et langage enfantin: phylogénèse et ontogénèse" in R. Chaudenson, éd. Langue française, 1978, n° 37, Paris, Larousse, 76-88. Chaudenson, Robert, 1981, Textes créoles anciens (La Réunion et Ile Maurice). Comparaison et essai d'analyse, Hamburg, H. Buske. Chaudenson, Robert, "Continuum intralinguistique et interlinguistique" in Etudes créoles, 1981, IV, 1,19-46, Montréal, CIEC.

82

Robert Chaudenson

Chaudenson, Robert, Article Review, Roots of Language, in Studies in Second Language Acquisition, 1982, 5,1, 82-102. Chaudenson, Robert, "Diglossie créole, diglossie coloniale", in Mélanges Willy Bal, Cahiers de l'Institut de Linguistique de Louvain, 1984,9,3-4,19-29. Chaudenson, Robert, "Français avancé, français zéro, créoles" in Actes du XVIle Congrès International de Linguistique et de Philologie Romanes, 1984, vol. 5, Aix-en-Provence, 165-180. Chaudenson, Robert, 1985 "Pidginisation, créolisation, acquisition d'une langue seconde", Actes du Colloque dAix-en-Provence, 1984, GRAL. Ferguson, Charles et De Bose, Charles, "Simplified Registers, Broken language, and Pidginization" in Valdman, Albert, 1977, Pidginand creole Linguistics, Bloomington, 99-128. Lambert, Wallace, "Culture and Language as factors in learning and education" in A. Wolgang, Education of Immigrant Students. Scherer, André, 1966, Histoire de la Réunion, Paris, PUF. Tabouret-Keller, Andrée, "Origine et simplicité: des langues créoles au langage des enfants" in Enfance, 1979,3-4,269-292. Du Tertre (R.P.), 1667, Histoire générale de l'établissement de colonies françaises dans les Ant Isles de l'Amérique.

Sociolinguistics, Linguistic Description, Language Change and Language Acquisition Norman Denison University of Graz Graz

Charles Ferguson's "Diglossia", 1959, was a landmark in the study and recognition of language varieties and their functions. It is remarkable, however, that whilst it provided an important impulse in the development of sociolinguistic awareness, the main direction in linguistic theory at that time (and up to the present time) was axiomatically based on an extremely homogeneous view of linguistic structure and linguistic competence. The following observations arise from the conviction that we have still not effectively integrated into linguistic theory the consequences, both synchronic and diachronic, of work on language variety from Schuchardt onwards. I will begin by making a claim and then attempt to enlarge upon it and finally to substantiate it by exemplification: The status of linguistic structures and linguistic changes, the question as to whether distinctiveness or opposition is or is not present for a native speaker or hearer, whether a given utterance is an acceptable ("correct") sequence in "his language", can in some cases not be settled unequivocally without including concrete sociolinguistic and psycholinguistic data as part of the evidence. In this connection, not only so-called "free" variation, not only sociolectally or regionally identifiable variants, are relevant but also situationally determined variation and its relative stability or instability. The psychological status of structures and developments (e. g. sound changes) in language depends on whether the variety (e.g. a "language" or "dialect") to be described is spatially and socially contiguous with related varieties or not, and whether or not any such linguistic

84

Norman Denison

and social embedding is in any sense part of the psychological background (or foreground !) of the speaker or hearer(s) at the time and in the circumstances of the utterance(s) to be analysed and described. Awareness of the inherent relativity, blurredness and interpenetrability of boundaries in language analysis can hardly fail to emerge from any realistic preoccupation with a range of seemingly fairly distinct sectors of linguistic research. The facts here are not in dispute: we know for instance how difficult it is to delimit in time (and therefore in structure and substance) an état de langue. The difficulty emerges nowhere more clearly than in the field of semantics, where an intrusive short-term diachronic dimension is introduced by the claim to descriptive relevance of the competence and usage of co-existent and interacting (but different) age-groups. This complicates the establishment of a fictive "present" or "cross-section" in semantic descriptive statements (and similarly, if less obviously, at other levels). Among the basic problems of synchronic dialectology, on the other hand, we can list the spatial and social (and hence the structural and substantial) delimitation of the objects of study (the language varieties) and this holds regardless of whether interest centres on languages, dialects, sociolects, registers or other varieties having parts of their structure and/or substance in common. From work on contact linguistics we know that interference phenomena are not restricted to sporadic cases, but can extend to sweeping re-structuring of whole sections of the grammar. Research since Sandfeld 1936, Weinreich 1953 and Haugen 1953 has focussed both on the synchronic mechanisms of interference and, from the diachronic point of view, on the systematised sediment resulting from the re-structuring and integration of such interference. These and similar considerations make it a problematic undertaking to attempt to establish the limits of a specific "language" or a specific "dialect" on the basis of linguistic criteria alone. "Problematic" is here intended not as a milder alternative to "impossible" but primarily in the sense that the usefulness of such attempts is questionable: axiomatic definitions are always possible and for some purposes legitimate, but they part company with linguistic reality. However, the main objection to the intuitive abstraction from usage by linguists (including lexicographers and the compilers of "standard" grammars) of the structures and substances which constitute the language or "the" dialect, "the" style, "the" register etc. to be described, in other words, the main objection to an intuitive and a priori projection of langue on to parole, is that it obscures the complexity

Sociolinguistics, Linguistic Description, Language Change

85

of the issues involved and prevents us from tackling certain important questions. Among these one might single out just two: (1) how do we (as linguists and) as language-users acquire the ability to extract (if we do !) and handle such homogeneous sub-systems from the heterogeneous totality of the linguistic environment and (2) just how strong is the intersubjective agreement among members of a speech-community concerning the results arrived at? At this point it will be as well to stress that no challenge is here intended to the basic (and compelling) assumption that, during acquisition, structures and systems are indeed in some way abstracted from (or projected on to) speech and that there must indeed be a quite remarkable degree of intersubjective agreement achieved amongst speakers for communication to function as well as it does. The question is: how on earth? The standard answer has apparently become: through the innate linguistic competence of the human infant. Leaving aside the question of just how detailed and specific any L. A. D. might be - we may or may not wish to accept the necessity for an innate blueprint in the shape of an overall underlying structure - we (and the child) are still left with the problem of the steps necessary to arrive at the acquisition of competence in the specific "language" (better: rule-and-substance répertoire) of the adult speech community concerned. One of the aims of this contribution is to suggest that at any rate in some communities this task is even more complex than commonly assumed and can scarcely be a matter of a linear sequence of transformations. Primarily, however, we address ourselves to the elucidation and substantiation of the claim made at the outset; or: when is a soundchange not a sound-change? Or: when is an opposition not an opposition? The fortunate circumstance that some members of a speech community could in the past become virtually totally separated (through migration or otherwise) from the main body of speakers and remain so for many hundreds of years enables us to compare the status of corresponding linguistic phenomena in the isolated group with the status of - in substance - the "same" phenomena in the original body of speakers (or rather their linguistic heirs). If separate development (in this case in the phonology) is not to have given rise to the degree of divergence we are inclined to associate with distinct language codes, we need a period of much less than the fifteen hundred years separating English from German (even Low German!) but long enough for awareness of the original unity to have disappeared. The case is nicely

86

Norman Denison

met by the tiny German linguistic island of Sauris (Ger. die Zahre) in N. E. Italy, where some seven hundred years have elapsed since separation from the main body of German speakers - specifically from the varieties of S. Bavarian spoken in the mid-13th C. around the southern - most point of the present frontier between E.Tyrol and Carinthia in Austria. The sound-system of the still "German" part of present-day Sauris German, then, if we undertake a traditional dialectological identification and description of this part of the Sauris linguistic répertoire (which includes two further "languages", Friulian and Italian), is virtually identical, according to traditional dialectological criteria, with that of certain dialects still spoken (alongside - or, nowadays, intermingled with - other, higher prestige varieties of present-day German) in the original homeland of (many of) the forefathers of (most) Sauris German speakers. Indeed, the correspondence is striking enough for it to be one of the arguments for supposing the original home of the migrants to have been in the region outlined above and not elsewhere. A number of sound changes were initiated in Bavarian and more specifically in S. Bavarian in the period leading up to the Sauris migration, especially in the stressed vowels. Some of these were completed1 before the migration, others apparently shortly afterwards. In the case of those we are about to discuss, the end-point was in any case the same in Sauris as in some parts of the restricted area from which the migration occurred, and these changes in fact took place over much of the Austrian dialect area, some of them far more widely still. Representative changes with exemplification (Sauris forms are quoted) are: (l)a) i:>ai b)

(2)a) b)

(3) a)

b)

mi:n>mai(n)"my"di:n>dai(n)"thy" ri:de > raide "pot-gallows" riiidelin > raidie "scab" ü: >i: > ai mti:ze > maize "mice" hii:zer > haizar "houses" ros > rös "horse", got > got "God' o>ö hoffen > höufn "hope", o: (where it was a lengthened allophone holaere > hôular "scraper", "hollower" of o) > ou latte > lote "lat", a>o khatse > khotse "cat", napf > nopf "dish" Ja:f>Jo:f "sheep", ma:ne>mo:ne a:>o: "moon", ha:ze [ho:ze "hare"

Sociolinguistics, Linguistic Description, Language Change

(4)

ae:, ai, a u > a :

(5) a) i > i, i: (lengthened allophone),

b) u>i,i:,u/(o) 2

87

kae:ze > kha:ze "cheese", vasilen > vailn "err"; anvailen>onva:ln "offer", zaife>za:fe "soap"; laufen>la:fn "run", gelaube > galaibe "faith" u/(o) 2 (before rin closed syllables): mit(e)>mit(e) "with", gehilbe> gehilbe "hazy, cloudy"; wize>bi:ze "meadow", derzider > darzi ¡dar3 "since"; hirte>hurte (hôrte)2 "cowherd", wirt > burt (bôrt)2 "landlord" J"tiikhelin > Jtikhle "piece", hiitte > hite ; miil(e)>mi:le "mill", vïille>vi:le "sausage filling"; vtirhten » vurtn (vôrtn)2 "fear", kiirtser > khurtsar (khortsar)2 "shorter".

It is important to remember that in their essential features the developments outlined above (and many more; for a provisional list valid for Graz see Denison et al. 1978) were complete in S.Bavarian by about the year 1300. In the linguistic island of Sauris they all realised their full status as sound changes, with all the structural consequences implied. Elsewhere, however, where S. Bavarian remained territorially contiguous with and communicationally and psychologically increasingly embedded in the totality of the German speech community, sharing in the emergence of regional, national and supranational varieties, with all that these have implied for social and linguistic acculturation, only those changes established and retained full distinctive and psychological status which were also shared by the standard language when it emerged. We should mention that most speakers of Sauris German have no experience either active or passive, with written representations of the non-Italian part of their linguistic répertoire, and none with standard German or indeed any other varieties of German, so that they are uninfluenced by any implicit structuring which visual forms might assign to the lexis. Hence, in Sauris, the phonemic status of the outcome of the processes exemplified under 1 a) and 1 b) is identical. There are no variants of the diphthongs developed under 1 b), with or without sociolinguistic implications. Likewise, developments 5 a) and 5 b) have in Sauris provided unified input to a series of phonemes i, i: and u (in S. Sopra I, i: and o).

88

Norman Denison

There is no longer any u (or ii:) phoneme in Sauris, and so the regular Umlaut relationship displayed by u (and u:) is with i (and i:) and in S.Sopra, over a restricted range, with o (the more general source of which throughout Sauris, unrelated to Umlaut, is mhg. o, further input coming from mhg. e in certain environments): zuin "son", pi. zi:ne, vuas "foot", pi. viase, Jlunt "swig, swallow", dim. /lintle, tsikalan "zucchini", lu:J" "artificial light" (from Friulian), pi. liije. A combination of pre- and post-migration developments (including borrowing from Romance) have led in Sauris to separate phonemic status for both / o f and / o u / (2 a, 2 b): Jtrotsn "drag", Jtroutsn "spill". Neither / o / nor / o u / have any synchronic structural or sociolinguistic connection with the Sauris phonemes / o / and / o : / (3 a, 3 b): votse "face (pej.)", votse "piece, shred", ho:ze "hare", houze "stocking". On the other hand the regular Umlaut of ou is ei (k h roupf "goitre", pi. k h reipfe, v o u j "voice" ( ai, aei b) ü: > ai, ai, 01 (OY) 6 (2) o > ö u , ou, o, o: (3) a) a > o, a(u), a b) a: > o:(u), oi, a:(u), a: (4) a) ae: > a:, e:, (e:)6 b) an > (oa)7, ai, ai, asi c) au > a:, au (5) a) i>i,i:,id, ii(0 (before 1) b)

Ü>I,Ü(I)

mai(n), maein "my" mais, mais, moize (moYze)6 "mice" röus, rous, [pfert] "horse" göut, gout, got "God" houl, ho:l "hollow" khots katse "cat" md u —• man "man" bo:utvon —• ba:devane "bath-tub" Jo:uf —>- Jaif "sheep" hois, ho:s —»• ha:ze "hare" kha:s, khe:s, khe:ze, (ke:ze)6 "cheese" (soaf)7, sa:fm, saife, saoife "soap" pa:m, baum "tree", laufm "run", glaubm "believe" mit "with", vi:zn, vi:ze "meadow", Jriat, Jnt "step", Ju(li)r), Jilirj "Schilling" hitn, hüte "hut", J"tikhl, Jtükhl [ftük9en] "piece".

90

Norman Denison

Lexical frames admitting only one vocalic realisation are the exception rather than the rule in Graz (and in the rest of urban Austria). They are characteristic of vocabulary items for the most part of relatively recent introduction via the standard language, which retain their typical phonetic-phonological shape even when embedded in L macro-structure: Park "park", parken "park (a car)"; or very occasionally it is a question of one of the few words of Bavarian origin accepted with indigenous phonetics/phonology into the standard language strand of the lexical répertoire: aper "snow-free", Rahm [ra:m] "(sour)cream" or the place-name Graz (older Grätz, Graetz) for instance. Otherwise, the selection of one rather than another realisation of a word typically involves the crossing of phonemic boundaries. Relatively few of the L realisations involve vowels which are in and of themselves phonetically stigmatised amongst Austrian speakers; the dialectal diphthongs öu and ei are good candidates for this status. However, it is in general the differing lexical distribution of phonetically similar or identical vowel phonemes which provides the substance for sociosemantic function: o: is H in Boot "boat" but L in Bad "bath", likewise a: will pass for H in lad' (ein) "invite!" but not in (tut mir) leid "sorry!". The fact that a less fronted variant of the vowel in e.g. lad' is prescribed as standard than the dialectal [a;] in leid is here irrelevant: any vowel quality in lad' less rounded than [o:] must be seen diachronically as an imported substitute for the quality of the indigenous phoneme, which was in clear opposition to the phoneme / a : / . mais is the H (and only) realisation of Mais "maize" but L as a realisation of Mäuse "mice"; the sociosemantic status of the sequence [va:s] diminishes according to whether it is a realisation of the lexicosemantics of (i) was? "what (emphatic)?" (ii) war ei? "was it?", (iii) weiß "know", or (iv) wäre es?"would it be?". The alternatives include (i) vos? (ii) va:r es?, voas?, (iii) vais, vaeis, voas (if in citation of rural speech) (iv) veos?, veore es? If we attempt a phonemicisation of the vowels occurring in a representative sample of Graz speech without a prior sociolinguistic analysis, we are forced to admit multiple phonemic realisations of one and the same lexical item. The answer to the question "when is an opposition not an opposition?" would then be "when it refers to the same lexical item or another lexical item with the same constellation of alternative phonemic realisations". A corollary would then be "When is a homonym not a homonym?" Answer: "When not all the alternative phonemic realisations of the lexical items in question coincide". However, if we try to handle the evidence along the traditional lines of an

Sociolinguistics, Linguistic Description, Language

Change

91

unequivocal phonemic representation for each lexical item (with "allophonemic" variants possible), then we must admit that many "alio-" variants (a) cross phonemic boundaries and (b) are not always phonetically plausible in relation to alternative "allophones". Cutting out the phonemic level of analysis as such and proceeding directly from formatives to bundles of distinctive features is of no great help either, since the feature composition of the alternative realisations of formatives is not explainable in a coherent way if we have no prior sociolinguistic analysis. The descriptive problem we have to face is of course not confined to the speech of Austrian urban dwellers. It is everywhere present to some extent where the outcomes of divergent linguistic developments, originally characteristic of different networks of speakers separated fairly completely by geographical or social barriers, come together as parts of the total répertoire of a wider speech community. The various strands tend thereupon to become associated primarily with sociolinguistic functions (though they may in addition retain regional and/or social group connotations). A not uncommon strategy for coping psychologically with this kind of situation appears to be to pretend that the historical changes in question never happened, unless they also happened in the variety which has come to carry the highest prestige and form the basis of the standard. This applies not only to phonological changes, of course, but it is most striking in this connection. School teachers in Austria (and many other parts of the German speaking area) have for many generations been doing battle against historical vowel changes such as u > i, ô > e(i), a > o, ai > a:. Evidence of such long completed changes in the speech of schoolchildren is treated as individual deviation from the unshifted standard system. In 18th centuiy England there seems to have been a sociolectally based development, initiated in the London area, eliminating / h / from the phonology, and this spread to most parts of England in the speech of the same social class. Here too the reaction of society at large has been to pretend that no such change has taken place and to treat evidence of it in school-children and the rest of the population as (stigmatised) "hdropping". Hence, we may now tentatively answer the question "when is a sound-change not a sound-change" in the following way: "when it is not shared by the standard variety" (and one might add: "and even where it is, it may be resisted for quite some time on the basis of a written standard").

92

Norman Denison

This kind of discrepancy between linguistic data and psycho-social attitude to the structural facts leads in the wider speech community to efforts to bridge the gaps where possible, to compromise between competing phonemic realisations of the same underlying semantic units, to re-constitute homogeneity in place of heterogeneity and in effect to roll back the sound-changes to the extent that this is possible. So far as the phonology is concerned, there are several possible solutions open to a speaker, once the variants are associated with differing sociolinguistic connotations for the same set of speakers. (1) He can stick to one set of variants, regardless of the undesirable sociosemantics he may be transmitting thereby. (2) In clearly defined sociolinguistic domains, the speaker can use now one variant, now another, as sociolinguistically appropriate. (3) In the many marginal and ill-defined areas of linguistic interaction, he can mix variants in what his Sprachgefühl tells him is the right proportion. Work on spontaneous speech in Graz suggests that this is a solution frequently adopted. It is also presupposed by Labov's work (e.g. Labov 1966) in America and Trudgill's comparable investigations in Britain (e.g. Trudgill 1974) involving the statistics of use of sociolinguistic indicators by various social groups under various conditions. (In areas of variation outside phonetics/phonology, for instance with syntactical and lexical alternatives, this would seem to be the only strategy available. However, not all speech communities permit mixing of comparable devices at a given point in time. For instance, in the English of Shakespeare, the same partner could be addressed first as thou, then as you, then as thou again (Macintosh and Halliday 1966,70-82) in the same interaction; this is virtually impossible in modern standard German with du and Sie).

(4) Where the phonetic substance makes it possible, the speech community can a) introduce punctual intermediate phonetic values between extremes or b) connect fixed phonemic points by a phonetic continuum. Let us now consider some examples of variation in the light of these possibilities. A-forms alongside h-less variants in English clearly do not lend themselves phonetically to the creation of a compromise continuum. Perhaps a glottal hiatus or a lack of liaison, such as has replaced "Germanic" (but not the lost original Latin) h- in standard French, could have developed as an intermediate compromise point

Sociolinguistics, Linguistic Description, Language Change

93

between forms with and forms without etymological h-, but this solution was not the one adopted by the South Humberside basilect of my early schooldays, of which the frequency dosage solution (lowest frequency of A-forms in the least formal situations) was characteristic, as I recall. Interestingly, we considered the occasional hyper-correct hforms produced by others a joke and our own as slips of the tongue. We handled the variation between -in and -iq in the suffix -ing and doubtless most of the substantial amount of further sociolinguistic variation we had to cope with - on the same frequency distributional basis. In the matter of R. P. English /A/, which was part of our passive répertoire, versus the / u / of our active northern basilect in words like but, bus, cut, efforts to induce us to pronounce such words less "broadly" or more "nicely", were in effect an invitation, had we known it, not to pretend that a sound change had not taken place, but rather to pretend that, for us too, it had. Here we responded with a compromise continuum, which ranged from [u] to [a] but stopped there, short of overlapping (as [A] would have done) with realisations of our own basilectal phoneme / a / in bat, bass, cat. It had escaped our attention that a range of words, including butcher, sugar, put, had / u / in R. R, and we extended our compromise continuum to these also. As a young adult I felt able to add [A] to my [U —*• A] continuum only after I had learnt to use [ae] in my H realisation of bat, bass, cat, etc. Clearly I was earlier under some kind of phonemic boundary constraint, and equally clearly I was learning to disentangle register strands in my phonology (phonologies?) before going on to infringe such constraints in the totality of my still diversifying active répertoire. If we now turn once more to the Austrian urban phenomena, it seems that in Graz the vowel series listed as nos. (3) a, (3) b and (1) b are realised as continua, with an increasing degree of backness and lip rounding in (3) from [a] to [o] and from [a:] to [o:], corresponding to an increasing degree of sociolinguistic informality. An increasing degree of unrounding in (1) b of the diphthong (OY) OI, until ai is reached, also constitutes a phonetic continuum, but the sociolinguistic implications appear to be distributed unevenly in this case. It seems that in Graz (and more generally in Austria) even a slight degree of lip-rounding of the first element of the diphthong is sufficient to avoid the L connotations of the extreme unrounding represented by the complete identity with / a i / < / i : / . The same remarks apply to the continuum which functions in Graz between u and i (no. (5) b): Any degree of rounding suffices to separate this phoneme from (5) a) and assure it sociolinguis-

94

Norman Denison

tic respectability. These judgements are based on informal observations and direct informal questioning of Graz colleagues, but research is necessary to substantiate these impressions with more objective findings. The vowel changes we have described under nos.(4) a, b, c and (5) seem to admit of no social bridge-building, whether in the form of fixed point intermediate values or by means of a sliding phonetic scale. Compromise integration can hence here be achieved only by the statistical frequency of the competing realisations in different sociolinguistic environments. Where intermediate realisations (punctual or shifting scale) have been developed between the extremes produced by sound changes, speakers' strategies in ill-defined or marginal situations may of course involve the use both of compromise phonetic forms and of statistically relevant frequencies of more extreme forms. Both strategies seem in fact to be used in the case of a —• o and a: —• oi, but again, research is required to determine the precise extent and the precise circumstances of such alternative compromise strategies. I have pointed out elsewhere (Denison 1984) that a monolingually biased model of acquisition with its homogeneous structural implications is inappropriate as a basis for studies of acquisition in plurilingual environments. In principle the same observation applies to plurilectal environments (in other words fairly generally if not universally!). The inadequacy consists in the failure to account for the specific linguistic structures selected by the child from the apparent chaos on offer, the order in which, and the mechanisms by which, the various strands are ultimately disentangled (to the extent that they are disentangled - certainly never totally) by each new generation, in the sense of the adult solution, and structurally and sociolinguistically graded. It would also seem that too much reliance on a strong innateness hypothesis, in a way which over-emphasises the child's own (admittedly remarkable) acquisition achievement, under-rates the role of implicit teaching - especially in the sense of filtering and pre-selection - by parents and other contact persons. It could also divert attention from the need for more research into speech addressed to children, as something different from (in the earlier stages perhaps crucially distinct from) the totality of the child's linguistic environment. Without wishing to oversimplify or distort Bernstein's notion of code (which has been neither very stable nor linguistically very precise), one may see work under his direction on the interaction between working-class London mothers and their children as research (inter alia) into the im-

Sociolinguistics, Linguistic Description, Language Change

95

parting of a sociolect (Bernstein 1971-75). Some idea of the way in which German-speaking mothers in Graz, Austria, in their regulatory and didactic function, serve as sociolinguistic models for their children - not least when urgency or intensity intrudes to modify their more self-consciously adopted linguistic stances - may be gained by consulting Penzinger 1985. On the other hand, a recent collection of essays on the state of the art in acquisition studies, Wanner and Gleitman 1982, whilst it shows signs of a growing recognition that the structure of the specific language to be acquired may in part determine the timing, order, and detailed strategies of acquisition - Slobin 1982,128-172 - and that some of the crucial determinants come from "outside" the child - Shatz 1982,102-127 - stops short of identifying structural heterogeneity and overlapping, of the kind we are here considering, as among the chief hurdles the children in most speech communities have to take. I do not wish here to enter the debate on the importance or otherwise of "motherese" in language acquisition tasks as these are usually seen in the psycholinguistic literature. However, if we concentrate on the task of selecting and disentangling, which faces the child in plurilectal or plurilingual environments, then the role played by speech actually addressed to children in shaping their active linguistic competence in the early stages, as distinct from speech merely present in the immediate environment, is dramatically underscored by language shift such as that which has been taking place in Sauris over the past forty to fifty years. Up to the second world war, virtually the whole adult population of Sauris was actively trilingual in Sauris German, Friulian and Italian. In 1985 most speakers there under 30 are actively competent in Italian only. In 1935 the chief initial language of infant socialisation in the home was still Sauris German, by 1955 it had become Italian and is now almost exclusively so. Moreover, Italian has become virtually the only language used by the under-30 age group for group-internal communication. Over the period concerned there has been a quantitative as well as a functional shift in the amounts of the languages concerned, Italian, Friulian and Sauris German, present in the linguistic environment of children growing up in Sauris - i. e. actually reaching their ears (and in the case primarily of Italian, their eyes), and this shift has been away from German first (in order of time), next, from Friulian, always and ever increasingly in the direction of Italian. However, even now, in 1985, most children are exposed to quite large amounts of Friulian and German from birth, both inside and (in the case of Friulian especially)

96

Norman Denison

outside the home. Grandparents are often present in the home, and these regularly use German together, and sometimes to address younger adults, in the presence of children. Friulian is often used between older villagers from Sauris in conversation with visitors from surrounding villages or even with each other. Most Sauris children therefore have varying amounts of passive understanding of Sauris German and Friulian, but nowadays little or no active competence. Fifty years ago, when children grew up trilingual in Sauris, it is doubtful if they heard more Italian outside school (three years on average) than their grandchildren now hear Sauris German, but much of what they did hear was addressed to them, and as a result they acquired active competence in Italian, as in Friulian and Sauris German. Nowadays children do not usually acquire active competence in Sauris German or Friulian, neither of which is regularly addressed to them. They do acquire active competence in Italian, which is addressed to them from birth in the home, at kindergarten, at school, and inside and outside the village by most contemporaries. The conclusion seems compelling: children apparently pay less attention (not nil attention, as all parents know to their occasional embarrassment) to speech not directed at them, which seems hardly surprising. The suggestion follows automatically that it is primarily to such speech that acquisition research should inter alia also pay attention. What applies to shifts from plurilingual to monolingual competence is also valid mutatis mutandis for shifts from (active) plurilectal to (active) monolectal competence such as is evidenced in growing measure in present-day urban environments. Of course, considerations of relative prestige have their impact on acquisition. But it may be that the earliest parameter the child can seize upon is the particular selection of rule and substance répertoire directed at it for preference by its elders and betters. Other selections (e. g. from peer group interaction) may follow, and for some, temporarily or permanently predominate. In either case, whether these selections coincide with the wider community's overt prestige selection or not, they are each probably more homogeneous in structure than the total répertoire, and hence of help to the child in its task of disentangling the structural and sociolinguistic strands of the whole. Linguists who carry out research on language description, or engage in acquisition studies, on the basis of a strong structural homogeneity hypothesis, will probably say that they take for granted that children sort out the strands of social variation in the total répertoire of their environment (and, where necessaiy, different languages which

Sociolinguistics, Linguistic Description, Language Change

97

may be simultaneously represented). As, indeed, children do, more or less, in due course. But some of us are going to have to continue exercising our minds on not taking it for granted.

References Bernstein, Basil. 1971-1975. Class, Codes and Control, Vol. 1-3, London-Boston: Routledge & Kegan. Denison, Norman et al. 1978. Linguistic Behaviour in Graz - a Preliminary Report. Proceedings of the Twelfth International Congress of Linguists. Vienna, August 28-September 2, 1977, ed. by Wolfgang U. Dressier and Wolfgang Meid, 257-261, Innsbruck: Innsbrucker Beiträge zur Sprachwissenschaft. Denison, Norman. 1984. Spracherwerb in mehrsprachiger Umgebung. Spracherwerb - Sprachkontakt - Sprachkonflikt, ed. by Els Oksaar, Berlin-New York: de Gruyter. Dressler, Wolfgang U., and Wolfgang Meid (eds.). 1978. Proceedings of the Twelfth International Congress of Linguists. Innsbruck: Innsbrucker Beiträge zur Sprachwissenschaft. Ferguson, Charles A. 1959. Diglossia. Word 15. 325-340. Haugen, Einar. 1953. Norwegian Language in America, 2 vols., Philadelphia. Labov, William. 1966. The Social Stratification of English in New York City. Washington, DC: Center for Applied Linguistics. Mcintosh, Angus, and M.A.K.Halliday. 1966. Patterns of Language. Papers in General, Descriptive and Applied Linguistics, London: Longmans. Oskaar, Els (ed.). 1984. Spracherwerb - Sprachkontakt - Sprachkonflikt. BerlinNew York: de Gruyter. Penzinger, Christine. 1985. Soziostilistische Sprachstrukturen in der MutterKind-Interaktion. An Beispielen aus der Grazer Umgangssprache, Frankfurt/Main: Lang. Sandfeld, Kristian. 1930. Linguistique Balkanique, problèmes et résultats, Paris. Shatz, Marilyn. 1982. On mechanisms of language acquisition: Can features of the communicative environment account for development? Language acquisition: the state of the art, ed. by Eric Wanner and Lila R.Gleitman, 102-127, New York: C. U. P. Slobin, Dan 1.1982. Universal and particular in the acquisition of language. Language acquisition: the state of the art, ed. by Eric Wanner and Lila R.Gleitman, 128-170, New York: C.U.P. Trudgill, Peter. 1974. The Social Differentiation of English in Norwich, Cambridge : C. U. P. Wanner, Eric, and Lila R. Gleitman (eds.). 1982. Language acquisition: the state of art, New York: C. U. P. Weinreich, Uriel. 1953. Languages in contact, New York: Publications of the Linguistic Circle of New York No. 1.

98

Norman Denison

Footnotes 1 For our purposes a sound change involving a vowel can be considered "completed" when a shift in the point of articulation in a given direction comes to a provisional halt (say, for a hundred years at least!). 2 This variant is typical of speakers from the fractions of Plotsn (Sauris di Sopra) and Velt. 3 alongside darzaidar ( < *derzi:der). 4 In some speakers, especially in Sauris di Sopra, di is heard for ei regardless of the origins of the phoneme. 5 except where a: < ae:, an Umlaut of mhg. a, in (scarce) examples like [ha:ran] "shave (a pig)"/[ho:r] "hair." 6 Very high style, rare in Austria. 7 Strong rural connotations in Graz.

The Interaction of Societal Factors in Language Acquisition John Piatt Monash University, Clayton, Victoria

The acquisition of the mother tongue can be considered a normal process in any individual's life. There are, of course, considerable differences in the total lexicon and in the syntactic structures acquired, even among speakers of the same language in any one region. Phonetic differences are also apparent, as are some differences in pragmatics and discourse style. However, every normal individual, unless deprived of an appropriate environment for first language acquisition, acquires the ability to communicate adequately within the range of topics appropriate to his or her environment and interests. The acquisition of a further language, however, is often a very different matter. The environment for acquisition may range from one which is as natural as that for first language acquisition to one which is highly artificial. Children urgently need to acquire the first language in order to communicate their needs and their emotions, to make full contact with the world around them. This is sometimes the case with a further language, as when children migrate with their parents to another country or when, having first acquired a minority language, a language spoken by only a relatively small group in the neighbourhood, they need to communicate with the peer group and the wider community. This latter situation exists in nations such as Singapore and Malaysia where children whose first language is a minority Chinese dialect such as Hakka (Kheh) or Hainanese will acquire the dominant Chinese dialect of the area, such as Hokkien or Cantonese. Children may, of course, acquire three or more languages, at least to the degree they find necessary. This would be quite possible in Singapore where, for example, a child with a Hainanese mother might

100

John Platt

have a Cantonese nurse or paternal relatives and pick up Hokkien, the dominant Chinese dialect, from other children, or if the family lived in a predominantly Malay area, the child might acquire Malay well before attending an English-medium kindergarten or primary school. The acquisition stage on which I shall concentrate is the acquisition of a further language (or further languages) through formal education. The acquisition of the further language(s) may be commenced at primary school (or even kindergarten), at secondary school, at a tertiary institution or even through government sponsored or privately run evening classes. The degree to which an individual successfully acquires a further language depends, to a great extent, on the interaction of a number of factors. In many nations, government policies seek to change language use in the community. In these situations, there is an interplay between government policies on the one hand and individual and societal factors on the other. The type of situation with which I am particularly concerned is where the further language is a language of wider communication, the typical second language (L2) situation. In what follows I shall consider some of the factors which may determine the degree to which the individual may be successful in acquiring a further language through the medium of education.

Government Policies In bilingual or multilingual nations where these so-called second languages play an important part in the life of the community, governments have often designed quite elaborate policies with regard to these languages. Language policy implementation may be considered in the wider sense throughout the community or in the narrower sense of implementation through the education system. In the wider sense, implementation may range from what may be called an administrative approach to a purely emotive approach or a mixture of the two. An administrative approach may include government directives for the use of a language in the media (combined with non-use of some other language or languages), its compulsory use in signs, e.g. shop signs and road signs, and possibly official sanctions against another language or other languages in the community and rewards for the use of this particular language, e.g. automatic promotion for government employees when a high proficiency in the language has been reached, compulsory tests in it for certain occupations.

The Interaction of Societal Factors in Language Acquisition

101

An emotive approach may include appeals and exhortations by political leaders at public rallies or in the media and language campaigns, including stickers and badges, e.g. 'Speak more X (and less Y)' or 'I can speak X'. As far as educational policies in particular are concerned, the desired language may be made the sole medium of education and other languages may be relegated to a secondary status or not available at all. An examination for entry into secondary or tertiary educational institutions may be conducted entirely in the language or it may be mandatory to pass an examination in the language and even obtain fairly high grades. Obviously, government policies are extremely important influences in second language acquisition. However, government implemented language policies may be less than fully successful if they are not complemented by individual motivation and societal approval. In bi- or multilingual nations there are a number of factors which may either assist or hinder the acquisition of a second language or languages favoured by the government. Some of the main factors are indicated in Figure 1.

Individual and Societal Factors A division has often been made between instrumental and integrative motivation (e.g. Gardner and Lambert 1972). Often, of course, instrumental and integrative factors operate together. The instrumental factors may favour the acquisition of a particular second language or may work against it because the language is either seen or not seen as being of material benefit. This is referred to as -I- or — I in Figure 1. The integrative factors may favour the acquisition because the individual wishes to identify with a particular group which is associated in some way with the language or they may work against it because the individual rejects the values of the group which is associated with this language. This is referred to as + or — i in Figure 1. As will be discussed later, although some motivative factors may be more instrumental than integrative, they may also have some integrative elements in them, e. g. Individual's Status in Society. Similarly, mainly integrative factors may also contain instrumental elements.

102

John

>

±1

Commi Value of 12

i c

03

o>

1u o I sf *

60

u a

«

t & fi * UO

t> •o .a g I s u u SP'S eg | £ c w (5/9)

Initial Clusters str 72% (8/11) Gr 61% (14/23)

r,7/8; s , l / 8 0,11/14; r,10/14

dr,kr, gr fl pi, bl pr, br kl, si sp, st, sk

r,42/42 1,12/12 1,12/12 r, 11/12; b, 1/12 1,10/10 s, 6 / 6

61% 52% 50% 52% 42% 8%

(42/69) (12/23) (12/24) (12/23) (10/24) (6/72)

r > w , 5/7 >tw, 8/14; >tr, 4/14 > w , 38/42 scatter >w,6/12; >0,5/12 r > w , 10/11 > w, 9/10 >0,4/6

(note: the data on medial and final clusters are rather meager, so numerical values are not reported here.)

This is to be contrasted with an absence of changes in singleton stop and nasal consonants (except in 'animal'), and fewer than 10% changes in initial singleton / s / or in initial / s C / clusters (except in 'spaghetti', where the initial cluster was broken up in a variety of ways). The remaining singleton of interest, / l / , was changed 20% of the time in medial position, except in 'yellow', where it was altered half of the time. We did not find changes of velar stops to alveolars in the students' renditions of 'cat' etc., in spite of the presence of this change in the cartoon character's speech. The hypothesis that 'spaghetti', 'yellow', and 'animal' are believed to be especially difficult is clearly borne out by the exceptionally large number of non-adult-like renditions of these marker words. Our list unfortunately did not contain any medial / k c / cluster except for the one in 'picture', but more than half of our respondents did alter that cluster, although not necessarily to the expected / c / . The only real sur-

116

Lise Menti and Jean Berko Gleason

prises in the data are l ) / s C / clusters were so rarely reduced to the stop consonant: 'store' was rarely rendered as "tore"; and 2) as mentioned, we did not find alveolars used for velars: 'cat' was not changed to "tat". The nature of the changes. A second question is whether we can find, among those subjects who thought that a child would alter a particular word, cluster, or segment, any consensus as to what the nature of that alteration would be. We have already indicated that there was a variety of treatments of 'spaghetti', including one / s p e D i / and 3 (out of 12) /skeDi/. There was only one instance of the projected 'pisghetti', Some of the respondents were able to transcend normal English phonetics and make highly appropriate subphonemic alterations in their renditions: one subject gave [spageHhi]. Aside from the four register marker words, we found some consonants and clusters for which there was a clear consensus as to how children would treat them, but others for which there was a split or a wide scatter. We can say that the stereotype exists, but we must add that it is not fully specified. The clearest consensus were 18/20 replacements of initial and medial / r / by / w / , 11/14 replacements of initial / d / by / d / , and 80 to 100% replacement of the / r / or / l / in the various stop plus liquid clusters (including / s t r / by / w / . It should be noted that there were a number of unpredicted changes made by our subjects; however, there was only one counter-predicted cluster simplification, namely, one subject rendered a / b r / cluster by / r / , producing 'roke' for the target 'broke'. The other 11 of the 12 changes in / p r / and / b r / clusters were, as expected, replacements of the / r / by / w / . On the other hand, there was a split in the responses to items with final / 0 / : among 13 changes, 4 were / f / , four were [th], and the others were scattered. Split and scattered responses were evident for initial / 8 r / , which presents two problematic segments: of 23 responses, 9 were unchanged, 6 were / t w / , 3 were / t r / , 2 were / 0 w / , one was / f w / , one was / 0 / , and one was the interesting / 0 t / . Study I thus shows a wide agreement among naive subjects about the kinds of sounds that young children might find problematic, and varying consensus on the nature of children's pronunciation of these sounds, ranging from near unanimity on some segments, such as the rendition of / r / by / w / , to rather scattered responses on others. However, we also found several instances in which the subjects used appropriate subphonemic variants of adult phonetic targets; in addition to the rendition of 'spaghetti' already mentioned, there were

Babytalk as a Stereotype and Register

117

velarized 1 and labiovelarized 1 for / l / , the bilabial glide u for / r / , and various hyper-aspirated and palatalized stops. Such responses appear to reflect genuine knowledge of the speech patterns of young children, rather than being based on a stereotype.

Study II This study asked parents to report on their own children's pronunciation. Method. The parents of 22 two-to-five year old children participated in this study. All families were middle-class native speakers of English, living in the Greater Boston area. The children, 11 boys and 11 girls, were all first-born, and were participating in a larger study of parent-child interaction and the acquisition of communicative competence at Boston University. (Other aspects of that study have been reported e.g. in Bellinger and Gleason 1982; Gleason, Perlmann, and Greif 1984; Gleason and Weintraub 1978; Greif and Gleason 1980; Masur and Gleason 1980.) All the children's names have been replaced by pseudonyms in this paper. As part of the study, each child made two vists to our laboratory, once with each parent, and participated in a structured play situation which was videotaped and audiotaped for analysis. At the end of the second session, each parent was interviewed individually about his or her child. This interview, the Parent Awareness Measure, called upon the parent to predict how the child would perform on a number of linguistic, cognitive, and social test items. The linguistic items included questions about the child's pronunciation of certain words. Parents' responses were correlated with the actual performance of the children on these measures. In order to generate a list of the child's own words for comparison with parent report, after the first session for each family we listened to the tape recording of the session and selected out two sets of words. The first set consisted of words on which the child's pronunciation was noticeably immature; the second set consisted of an approximately equal number of words in which the child's pronunciation was adequate but which contained sounds or sound patterns generally considered to be difficult for children to say. (For some of the older children in the group, there were no words which would be considered immaturely produced, so only words of the generally-considered-difficult type were used.) Depending on the number of immature forms which

118

Lise Menti and Jean Berko Gleason

the child used during the first session, the list contained between 5 and 12 items, usually about 8. In the individual parent interviews conducted at the end of the second laboratory session, parents were asked about these words, which were presented orally by an interviewer from our research team who did not know how the child pronounced them. The words were mixed in order of presentation in order to minimize the likelihood of the parent noticing any pattern to the questioning. To avoid suggesting to the parent that her or his own child was in any way slow in phonological development, we used the following standard set of instructions "Here are some words which children X's age often have trouble with. How does X say each of these words?" The words were then read individually and the responses written down by the interviewer. Results and Discussion. Because our sample of children at any one stage of phonological development was small, and because the lists were individualized rather than standardized, quantitative analysis of results is virtually impossible. Examples of parent responses are presented in Table 3 for inspection. Qualitative analysis is, however, very suggestive. Five sources of apparent bias in parents' reportage can be identified: the stereotype explored in study I, sociolinguistic factors, and three types of overgeneralization of the child's own productions. 1) The conventional stereotype. The patterns of parents' reporting suggest that it is indeed biased by stereotypic notions about the way children speak. The parents may expect and report patterns generated by simplification and substitution rules such as those described in Study I. In addition, they may also be influenced by other models of infantile speech in the culture. One such example is the cartoon character mentioned above, who has (in addition to the features found in Study I) the replacement of velars by alveolars in syllable-initial position, and general replacement of fricatives by stops. When the child has just one of these patterns that is common in the stereotype instanced in a word, the parent may report this fact accurately. Our subject John had a general rule changing velars to alveolars; his mother spontaneously reported that she knew he said't' for 'k'. However, when the child made multiple changes in a word, this fact was rarely reported correctly even if all the changes were stereotypic ones. John's mother missed his changes of / r / and / l / to [w] in the words 'green' and 'milk', produced by her son as /dwin/ and /miwt/ respectively.) In Study I, we described the reduction of / b r / to / r / rather than to / b / or / b w / as 'counter-expected'; the parents almost never reported

Babytalk as a Stereotype and Register

119

Table 3. Examples of Parent Reports Child

Age

Taped production

Mother's report

Father's report

Nanette

2:1

/waebit/ /awej/ /bEA/

rabbit only better "trouble with'th'"

rabbit only better

/to/ /toil/ /toz/

/ t o / (show) /tofi/ (coffee) /toz/, close (close)

/6\k/

truck

/pun/

/pun/ (spoon)

/fwi/ (parents str ongly ove

restimate) /so/ /tofi/ /kwoz/ /SAk/ /ÒAk/ /SAk/ /pun/ (parents fre quently u nderestimate) Victor

David

2:5

three

4:2

/sor/ /sto:/ store sky /sai/ sky /saberi/ /stoberi/ (strawberry) /stoberi/ /sirjz/ things /tigz/ /tigz/ (parents misreport lo eus of error in /iiC/ clusters) John 4:2 /dwin/ /dwin/ /dwin/ (green) /s:wu/ /stru/ screw /tae:n/ /taent/ can't /mivvt/ /milt/ /miAlk/ (mother ov îrgenerali zes velar > dentaJ pattern; for be>th parents, accuracy reduced in 'rrlilk', whic i has two substi tutions, 1 > w an t )

counter-expected rules, such as this or deletion of the stop consonant in an / s C / cluster. However, two of our children always deleted stops from / s C / clusters, and two others did so in /str/ clusters (producing 'strawberry' as "sawberry"). Not one of the parents of these four children reported the words affected by this rule correctly. Instead, they reported the words as being correct, or as being affected by one of the stereotypic rules. For example, David said [sor] for 'store'; his father thought he was able to say the word correctly, while his mother reported that he deleted the final / r / . For the CCC cluster /str/, children

120

Lise Menti and Jean Berko Gleason

saying [saiberi] were reported to produce 'trawberry' or 'stawberry', i. e. as using one of the two stereotypic rules that could apply to /str/, liquid deletion or fricative deletion. Another example of the misreporting of nonstereotypic child forms can be found in the data from Martin: on our tape, he said [aza] for 'other', but both his parents reported that he said this word as 'udder'. 2) Sociolinguistic factors. We suggest that a second factor influencing parent reports is the extent to which the child uses socially stigmatized forms, e. g. 'dese, dem, and doze'. This negative sociolinguistic loading of the child's output form appears to make its pronunciation more salient. The evidence for this claim is that parental report accuracy was higher for the child's use of / d / instead of ó than for any other error; by comparison, the rate of accurate report of use of a [w]-like sound for prevocalic / r / was between 25% and 30%. Since this is the only negatively-loaded sound substitution in our data, we cannot be sure of the explanation, however; work in other languages or possibly other dialects of English would be required. Overgeneralizations. The remaining factors are all types of overgeneralization of patterns actually present in the child's speech. 3) Parents under-report the variability of their children's speech in several ways; since children in the midst of phonological development may have very variable productions (Menn 1983), there are many opportunities for this sort of error to occur. Some parents are apparently critical, and report the less mature of the child's current forms, while others are optimistic and report that the child 'used to say' some immature form that he/she still says a large part of the time. Both of these parents behaviors were found in reportage on children who were showing a d / ó variation; some of their parents told us, incorrectly, that the child were still saying / d \ / only, while others told us, equally incorrectly, that the child now always said /óa/. 4) Closely related to this is the fourth factor, failure to keep up with the child's changes through time: parents may report a child as using a relatively immature form while we hear only a more accurate form on our tapes. Very probably, the parent is reporting an earlier stage of development, although it could be that they are reporting a form belonging to the stereotype and which their child actually never used; this hypothesis of course cannot be tested without longitudinal data. 5) The fifth factor, on the other hand, is one for which we have very clear evidence: parent's overgeneralization of children's rules to phonological contexts in which they do not in fact apply. A clear example is John's mother's report that her son said "strew" for 'screw'. As men-

Babytalk as a Stereotype and Register

121

tioned earlier, she had volunteered the statement that her son u s e d ' t ' for 'k', which was in fact true in almost all environments; when she was questioned about 'screw', she sounded out the word carefully and seemed to be making the substitution of / t / for / k / as she said it. The child's form was actually [siwu]; thus, the mother's report is unlikely to represent an earlier stage in John's development. Another misreport, from Katie's parents, on the other hand, probably represents a confound of factors four (reporting an earlier stage) and five (overgeneralizing the phonological environment): they report her as using / w / for / r / in all words, while we heard a fairly clear [r] after initial / t / and the [w]-like sound elsewhere. Another possible explanation for Katie's parents' failure to hear her / r / in consonant clusters is that adults may have beliefs about the likely order of children's phonological development. In Katie's case, the fact that her approximation to [r] was better in / t r / clusters than in singleton word-initial and intervocalic position is counter-intuitive. Direct experimentation asking adults to estimate the age of children based on recordings of their speech could be used to test the hypothesis hazarded here. Conclusions. In these small studies, we have shown that undergraduates who are not parents share a general notion of how young children speak. In particular, they have been in agreement that those words which we have called 'register markers' are likely to be mispronounced by children; at the same time, they did not generally agree, for instance, that young children would say 'pisghetti', even though we believe that everyone will recognize this form as belonging to a stereotypic register. Other features that were common in the stereotype include the substitution of / w / for / r / and / l / (especially in consonant clusters), and / d / for / d / . However, the students did not use / t / for / k / , and there were many sounds for which students showed a wide variety of substituted segments; what they agreed on was only that children would make some sort of alteration of the target segment. We also showed that parents frequently misperceive the phonological errors made by their children, that those misreports are fairly typical in nature, and that they do correspond in part to the kinds of errors that might be predicted if one were operating on the stereotype alone. However, two other types of factors appear to be involved: 1) when the child's from coincided with a socially stigmatized dialect form it was reported more frequently, and 2) parents overgeneralized their child's actual errors over time or over phonological contexts.

122

Lise Menti and Jean Berko Gleason

The relevance of these findings for language development, and clarification of the terminology concerning the language spoken to and by children will occupy the rest of this paper.

Implications for Language Development The bias that we have demonstrated in adults' sensitivity to children's immature pronunciation might serve various functions vis-a-vis the child. For example, it may enable an adult who hears part of such a pattern quickly to identify the speaker as a child at a particular stage of development and to produce appropriately modified speech. Most studies of child-directed speech have concentrated on the syntactic properties of children's speech that may trigger the simplified speech that adults direct to them. Here we have shown that adults are aware of immature phonology: it is therefore reasonable to assume that finetuning on the part of the adult may be in response to phonological as well as to grammatical indices on the part of the child. We have also shown that the adult, while aware of the child's phonological immaturity, may have only a general or stereotypic notion of how the child pronounces words. This misperception may serve a positive, rather than negative, end. By assimilating the child's immature pronunciation to a familiar cultural pattern, the adult may feel reassured that the child is developing normally, whereas an acute awareness of the child's phonology, especially if the adult thought it atypical, might lead to attempts to correct or otherwise interfere with the child in a way that would not be useful. The perception that a child is 'speaking babytalk' may also function like other signals of immaturity in evoking protective and nurturing behavior on the part of the adult. As Charles Ferguson has pointed out in 'Babytalk as a simplified register' (1977) and elsewhere, simplified registers as used by adults serve many purposes: most of them, such as those directed to foreigners and to retarded people, exhibit clarification and simplification processes. Child-directed speech, as a distinct register, also has a large affective component, and it may be that the child's immature phonology is one important factor in engendering adult affection.

Babytalk as a Stereotype and Register

123

Stereotype, Register, and the Uses of the Term 'Babytalk' Stereotype I: 'How Babies Talk', (babytalk 1) Labov's succinct definition of sociolinguistic stereotypes bears repeating here: "Stereotypes are socially marked forms, prominently labeled by society ... A social stereotype is a social fact, part of the general knowledge of adult members of the society; this is true even if the stereotype does not conform to any set of objective facts. Stereotypes are referred to and talked about by members of the speech community; they may have a general label, and a characteristic phrase which serves equally well to identify them (1972, p. 314)." (Labov goes on to mention, as an example of the latter, one which we have already alluded to: "'Deses, dems, and doses'. General characteristic of working class speech in the United States..."). Attempting to characterize the 'way babies talk' virtually implies that a stereotype will be constructed: since children vary considerably in their approaches to 'learning to pronounce', no one notion of childish pronunciation could represent them all. The stereotype does, however, appear to reflect the commonest strategies used by young children, as far as we can tell: for example, relatively few children delete the stop in a stop + liquid cluster; most of them delete the liquid or replace it by a /w/-like sound, and the stereotype of how children talk reflects this majority strategy. The stereotype of child speech that we have sketched out in Study I certainly fits Labov's definition, and it has characteristic or marker words, such as 'pisghetti'. It also has a name: it is called 'babytalk'. However, this term has several other meanings in the literature and in common parlance. Stereotype II: 'How adults talk to babies', (babytalk 2) 'Babytalk' is also the term for the speech that adults report using to babies, and it is in this sense that the word has been used in the earlier linguistic and anthropological literature, most notably by Ferguson (1964), Baby Talk in Six Languages. There is an overlap between reported speech to babies and reported speech of babies, and adults may believe that words in the 'babytalk' they use are based on the 'babytalk' used by babies. As Ferguson has said in a number of instances (opp. cit. passim), the influence is probably bidirectional: some features of adult 'babytalk', such as simplification of consonant clusters, may be in imitation of infant phonology, while others such as the use of special lexemes like 'bunny' for "rabbit', are clearly influenced by the adult. Insofar as adults who report particular forms as 'the way people talk to babies'

124

Lise Menn and Jean Berko Gleason

may never use those forms themselves when talking to babies, this reportage is a stereotype, i.e. a mental construct whose features may never all be realized simultaneously. (And some may never be found at all: do cowboys really say 'Howdy partner'? Do adults ever address a baby with 'kitchy-kitchy-coo'?). The Babytalk Register. (babytalk3) In talking to babies, people bring to the situation their knowledge of stereotypical babytalk, but the actual language they use is a product both of their own individual stylistic choice and of the interactional features that impinge upon them; for instance, a parent may choose to use or not to use a particular babytalk item such as 'nighty-night', but is constrained to speak in short sentences in order to be comprehended by the young child. The resulting speech is also called 'babytalk', even though many of its features are unrecognized by its speakers; much current literature refers to this register as 'Child Directed Speech' (nee 'motherese'). The way children actually speak, (babytalk 4) The fourth use of the term 'babytalk' is common parlance in referring to the syntactically and phonetically modified speech used by very young children; this is the primary definition of'babytalk' given e.g. in Webster's Third New International Dictionary. This use of the term needs no further discussion here; in the developmental psycholinguistic literature this use of the term is now replaced entirely by 'child language'. In this paper, we have shown that there is culturally shared stereotype of young children's speech. Parents appear to be affected by the stereotype when they listen to their own children, so that they sometimes misperceive the children's phonology in the direction of their expectations ; for instance, parents reported that their child said 'trawberry' when the child actually said 'sawberry'. At the same time, parents are aware of some aspects of their children's phonology, and can give an approximate report of their children's pronunciation, even though they may generalize or normalize the sounds that they hear. Their stereotypic notions thus blend with their actual observations; parents reportage of their children's speech is affected both by their own psycholinguistic processes and by the 'babytalk' stereotype current in our society.

Babytalk as a Stereotype and Register

125

Notes 1 The research on which this paper is based was supported in part by NSF grants BNS 76-80278 to K.N.Stevens and the first author, and BNS 75-2109A01 to the second author. 2 Dresel (1977) reports that a trickster-hero speaks a stereotypic babytalk in Hausa folklore. 3 A related study was conducted by Bond (ms.) at about the same time as our Study I was being carried out.

References Anderson, Elaine S. 1984. The acquisition of sociolinguistic knowledge: some evidence from children's verbal role play. Western Journal of Speech Communication 45:125-144. Bellinger, David, and Jean Berko Gleason. 1982. Sex differences in parental directives to young children. Journal of sex roles 8:1123-1139. Bond, Zinny S. (to appear) Baby Talk: Adult prediction of child pronunciations. Channary R. and Shockey, L. (eds) (Festschrift for Ilse Lehiste). Dresel, Linda Hunter. 1977. Palatalization in Hausa child language. Paper read at the annual meeting of the Linguistic Society of America. Ferguson, Charles A. 1964. Baby talk in six languages. American Anthropologist 66:103-114. Ferguson, Charles A. 1977. Baby talk as a simplified register. In Catherine E. Snow and Charles A.Ferguson (eds.), Talking to Children. Cambridge: Cambridge University Press, pp.209-235. Garvey, Cathering, and M.Ben-Debba. 1974. Effects of age, sex and partner on children's diatic speech. Child Development 45:33-42. Gleason, Jean Berko, and Sandra Weintraub. 1978. Input language and the acquisition of communicative competence. In K. Nelson (ed.), Children's Language, Vol. 1. New York: Gardner Press. Gleason, Jean Berko, Rivka Y.Perlmann, and Esther Blank Greif. 1984. What's the magic word: learning language through politeness routines. Discourse Processes 7:495-503. Greif, Esther Blank, and Jean Berko Gleason. 1980. Hi, thanks, and goodbye: more routine information. Language and Society 9:159-166. Labov, William. 1972. Sociolinguistic Patterns. Philadelphia: University of Pennsylvania Press. Masur, Elise, and Jean Berko Gleason. 1980. Parent-child interaction and the acquisition of lexical information during play. Developmental Psychology 16: 404-409. Menn, Lise. 1983. Development of articulatory, phonetic, and phonological capacities. In Brian Butterworth (ed.), Language Production, Vol.2. London: Academic Press, pp. 3-50. Sachs, Jacqueline, and J. Devin. 1976. Young children's use of age-appropriate speech styles in social interaction and role playing. Journal of Child Language 3:81-98.

A Note on Ferguson's Proposed Baby-talk Universals1 Lilith M. Haynes Universität Essen Gesamthochschule Essen and

Robert L. Cooper The Hebrew University of Jerusalem Jerusalem 1. Introduction Ferguson (1964), in a seminal article, hypothesized that baby talk (BT), that register of a language which is regarded as primarily appropriate for talking to young children, is a stable, conventionalized part of all languages. Using six languages as examples, he suggested that BT has defining lexical, phonological, syntactic, and paralinguistic features, as well as rules for its use. Later, Ferguson (1978) examined published accounts of BT in 27 languages2 and concluded that in every human society people modify their normal speech when talking to children whether it be the lexicon, phonology, prosody, or discourse formulae they change - into a coherent and conventionalized register which seems to be an important factor in the socialization of children. Ferguson's survey of the published literature led him to hypothesize 22 characteristics as universal for BT registers. These characteristics are almost fully documented for Standard American English, but, with respect to the other accounts of BT which he used, the original researchers often did not look for or mention many of them. Thus there are many gaps in the existing data. The importance of Ferguson's proposals for characterizing human language, its acquisition, and its use suggest that the actual occurrence of these BT characteristics might be usefully studied in the languages considered by Ferguson, as well as in other languages.

128

Lilith M. Haynes and Robert L. Cooper

2. Procedure A convenient opportunity for surveying many languages was afforded by the linguistic heterogeneity and sophistication of the participants in the 1977 Linguistic Institute of the Linguistic Society of America, which was held at the University of Hawaii. These participants were graduate students and postdoctoral fellows who were enrolled in the Institute's six-week courses. We administered a questionnaire to 34 of these persons during the fourth and fifth weeks of the Institute. Half of these questionnaires were completed by the first author's interviewing respondents for approximately one hour, and half were filled out by respondents working on their own. Approximately equal numbers of men and women constituted these two subgroups. The 34 languages or language varieties are Amharic, Anggor, (Tunisian) Arabic, Assamese, Betawi, Cantonese, Catalan, Dutch, (Australian, British, Canadian, Guyanese, and Hawaiian) English, German, Hebrew, Hokkien, Indonesian, Javanese, Korean, Palauan, (Brazilian) Portuguese, Sindhi, Soddo, (American, Castilian, and Cuban) Spanish, Sundanese, Swahili, Tagalog, Tarascan, Tumbuka, Ukrainian, Ullogooli, and Yiddish. Of these, 22 had not been surveyed by Ferguson, but Arabic, Dutch, English, German, Portuguese, and Spanish are common to both studies.

3. Results Most of the 22 characteristics which Ferguson suggested as BT univers a l were found in at least half of these 34 regionally and genetically diverse languages or language varieties. Only five characteristics were reported in fewer than half of the languages, the smallest incidence being 12 languages (35%). While none of Ferguson's characteristics was reported for every language, there are eight which were reported for at least 27 (79%) of the languages. These are high pitch, exaggerated intonational contours, shorter sentences, repetition, special terms for kin and body parts, reduplication, pronoun shift, and the use of BT by older children. It is of interest that all five of the categories which Ferguson used to classify the modifications which adults make when talking to children (prosody, syntax, lexicon, phonology, and discourse) are represented by at least one of these eight very frequently cited characteristics. Of Ferguson's categories, prosody and syntax are most often represented, accounting for four of the eight most frequently reported

A Note on Ferguson's Proposed Baby-talk Universals

129

Table 1. Reported Presence of Baby-Talk Characteristics Number of languages1

Characteristic

22 languages not treated by Ferguson Yes No Don't No know resp.

All 34 languages Yes No Don't No know resp.

Prosody 1. High pitch 2. Exaggerated contours 3. Slower rate

18 20 15

3 2 3

0 0 2

1 0 2

30 32 24

3 2 6

0 0 2

1 0 2

Syntax 4. Shorter sentences 5. Parataxis 6. Telegraphic style 7. Repetition

21 10 10 18

0 7 6 2

1 4 4 2

0 1 2 0

32 1 17 11 18 10 30 2

1 5 4 2

0 1 2 0

Lexicon 8. Kin terms & body parts 9. Infant games 10. Qualities 11. Compound verbs 12. Hypocorism

16 3 18 0 13 6 6 11 14 3

3 4 3 5 4

0 0 0 0 1

27 4 25 2 20 10 12 13 24 4

3 7 4 8 5

0 0 0 1 1

Phonology 13. Cluster reduction 14. Liquid substitution 15. Reduplication 16. Special sounds

9 8 19 12

8 6 3 3

3 6 0 6

2 2 0 1

16 15 30 18

9 8 4 4

6 8 0 11

3 3 0 1

Discourse 17. Questions 18. Pronoun shift

7 10 15 4

4 2

1 1

13 14 27 4

6 2

1 1

Extended uses 19. Between children 20. Animals 21. Adult intimacy

20 0 9 10 11 6

2 3 5

0 0 0

31 0 19 11 21 6

3 4 7

0 0 0

10

1

15

11

1

Other 22. Variation in degree 1

6

5

7

One respondent per language or language variety

features. It is also of interest that four of the six characteristics suggested by Ru^e-Draviija (1976) as probable universals of BT (high pitch, reduplication, names of body parts and events closely associated with the young child, pronoun shift, diminutives (hypocorism), and onomatopoeia) are among our eight most frequently reported characteristics.

130

Lilith M. Haynes and Robert L. Cooper

In most cases, the number of languages for which the presence of a characteristic was attested was far greater than the number for which the existence of that characteristic was denied, even when the number of attestations was relatively small. Thus, for example, a relatively small number of respondents, 18 (53%), attested the existence of special sounds in the BT of their communities, whereas the presence of this characteristic was denied by only four respondents (12%). There were seven characteristics which were denied for at least nine (26%) of the languages, although for most of these the number of attestations was still considerably higher than the number of denials. For two characteristics, however, the number of denials was greater than the number of attestations: the use of compound verbs made with general purpose auxiliaries (e. g., go bye-bye) and a relatively higher incidence of questions, denied respectively for 13 and 14 of the 34 languages. The frequencies of attestations and denials for each of the 22 characteristics are shown in Table 1. Ferguson's 22 characteristics were found more often in our 34 languages and language varieties than in his 27 languages. Our generally higher frequencies of occurrence support his assumption that the presence of these characteristics would be confirmed once the gaps in his schema were investigated. The percentages of occurrence of each of these 22 characteristics as found by us are displayed in Table 2. A stonger piece of evidence to support Ferguson's assumption that the gaps in his data would be filled by positive reports is seen in the data for five of the languages common to both studies: Arabic, Dutch, German, Portuguese, and Spanish (Table 3).3 If Ferguson had been able to find a report (either positive or negative) about the existence of each of 22 characteristics for each of the five languages, there would have been (22 x 5 = ) 110 entries, but he only found slightly over half (57) that number. Of the 53 blanks, most (43) were filled in by our respondents, and most of these (33) proved to be positive. It is of interest that of the 54 items common to both studies, there were only seven disagreements, all of them for Dutch and provided in our study by a native speaker, the father of a young child who was also present in Hawaii. Inasmuch as our respondent's disagreements run counter to the observations for the other languages, as well as to the data which Ferguson found for Dutch, it is plausible that our respondent's disagreements reflect faulty observation or reporting on his part. On the other hand, since he also reported variation in degree of babyishness for Dutch BT registers, his observations perhaps hold for a "less babyish"

A Note on Ferguson's Proposed Baby-talk Universals

131

Table 2. Percentage of Languages for which Baby-Talk Characteristics Attested: Present Study and Ferguson (1978) Characteristic

Percentage Ferguson (1978)

Present study

N = 27

Languages not All languages in Ferguson N = 22 N = 34

Prosody 1. High pitch 2. Exaggerated contours 3. Slower rate

37.0 33.3 14.8

81.8 90.9 68.2

88.2 94.1 70.6

Syntax 4. Shorter sentences 5. Parataxis 6. Telegraphic style 7. Repetition

18.5 11.1 18.5 22.2

95.5 45.5 45.5 81.8

94.1 50.0 52.9 88.2

Lexicon 8. Kin terms & body parts 9. Infant games 10. Qualities 11. Compound verbs 12. Hypocorism

77.8 51.9 59.3 44.4 74.1

72.7 81.8 59.1 27.3 63.6

79.4 73.5 58.8 35.3 70.6

Phonology 13. Cluster reduction 14. Liquid substitution 15. Reduplication 16. Special sounds

44.4 48.1 77.8 25.9

40.9 36.4 86.4 54.5

47.1 44.1 88.2 52.9

Discourse 17. Questions 18. Pronoun shift

25.9 40.7

31.8 68.2

38.2 79.4

Extended uses 19. Between children 20. Animals 21. Adult intimacy

7.4 22.2 22.2

90.9 40.9 50.0

91.2 55.9 61.8

Other 22. Variation in degree

29.6

27.3

44.1

variety used with his child. Alternatively, the discrepancies may reflect dialect or other variation. There was considerable disagreement among our respondents as to whether there are linguistic differences in BT characteristics associated with social stratification. We do not know to what extent these differ-

Lilith M. Haynes and Robert L. Cooper

132

ences of opinion reflect real differences between speech communities (e.g., in the degree of social stratification and linguistic diversity) or differences in the accuracy of the respondents' observations. There was more agreement, however, on the characteristics associated with variability of use, with a majority claiming that in their communities women use BT more than men; that BT is used more frequently in priTable3. Presence ( + ) or Absence ( —) of Baby-Talk Characteristics in Five Languages as Reported by Ferguson 1978 (a) and Present Study (b) Characteristic

Language Arabic

Dutch

German

(a) (b) (a) (b) (a) (b)

Portuguese

Spanish (a) (b)

(a) (b) Prosody 1. High pitch 2. Exaggerated contours 3. Slower rate

+

+ + +

+ +

+ + + +

+ + + +

+

+ + + + +

+ + + + +

+ +

+ + +

Syntax 4. Shorter sentences 5. Parataxis 6. Telegraphic style 7. Repetition Lexicon 8. Kin terms & body parts 9. Infant games 10. Qualities 11. Compound verbs 12. Hypocorism Phonology 13. Cluster reduction 14. Liquid substitution 15. Reduplication 16. Special sounds Discourse 17. Questions 18. Pronoun shift Extended uses 19. Between children 20. Animals 21. Adult intimacy Other 22. Variation in degree

+ + +

+ +

+ + +

+ +

+ + +

+ +

+ +

—

-

—

—

•

+ + + + +

+

+

+

+ +

+

+

+

+ + + + +

+ + + +

+ + + + + + + +

+ +

+

+

+

+

•

+

+ +

+

+ +

+ + +

+

+

+

—

•

+ —

+

+ + +

—

+

+ -

+ + + + +

+

-

+ + +

+ +

—

+ + + + +

—

+ -

+ + +

+

+

+

+

+ +

+ + +

+

+

+

A Note on Ferguson's Proposed Baby-talk Universals

133

vate than in public; and that frequency of usage is not related to the sex of the child being addressed. While our data were collected to test Ferguson's propositions, some of our respondents' remarks about the use of BT in their speech communities also provide striking, if informal, support for the sociolinguistic assumption that a community's values are reflected, at least in part, by its use of language and by its attitudes towards language use. Thus, for example, in the Palauan speech community, where adults attempt to imbue the child's world with joy, the superlative form of ungil, which is hot, 'the best', is preferred in BT over the uninfected form; in some communities which practice corporal punishment, BT was reported to be used to indicate the authority of the adult or to signal that physical punishment is imminent; our respondents for Yiddish and for Yoruba claimed that their communities place great value on the child's quick acquisition of adult speech norms, and that the use of BT in those communities is therefore less frequent than it might otherwise be. Most communities were described as using BT not for promoting the child's acquisition of language per se but for ensuring its general socialization in the community. Indeed, where BT is seen as interfering with the acquisition of the adult language, by keeping the child babyish in language behavior, it was reported to be discouraged.

4. Summary and Conclusions While we did not obtain attestations of every characteristic for every language in our survey, our data suggest that the characteristics Ferguson listed are indeed typical of BT and that many of them are highly likely to occur. The most common modifications adults make when talking to babies appear to be high pitch, exaggerated contours, shorter sentences, repetition, special terms for kin and body parts, reduplication, and pronoun shift. If any of these characteristics should in fact prove not to be universal, it is nevertheless likely that all BT registers show modification in at least prosody and syntax, and probably in discourse, lexicon, and phonology as well.

134

Lilith M. Haynes and Robert L. Cooper

Notes 1 We are grateful to Charles Ferguson for letting us work with a prepublication draft of his article (1978); to Lily Fillmore and Joshua Fishman for permission to conduct a preliminary survey in their classes as well as for their comments on our study; to Catherine Snow for her advice on an earlier draft of this paper; and to our respondents. We would like to take this opportunity to acknowledge our admiration of and affection for Charles Ferguson, who has shown us the finest qualities of a teacher, colleague, and friend. 2 These languages are (Syrian) Arabic, (Neo-) Aramaic, Bengali, Berber, Cocopa, Comanche, Dutch, (American) English, German, Gilyak, Greek, Hidatsa, Hungarian, Japanese, (Havyaka) Kannada, Kipsigis, Latvian, Luo, Maltese, Marathi, Pomo, (Brazilian) Portuguese, Rumanian, Samoan, Serbo-Croatian, Spanish, and Tzeltal. 3 Our data for Spanish reflect the reports for three varieties - American, Castilian, and Cuban. Although English was common to both studies, it is not shown in Table 3 because there were virtually no gaps in Ferguson's data for English.

References Ferguson, Charles A. 1964. Baby talk in six languages. American Anthropologist 66.6 (Part 2). 103-14. - . 1978. Talking to children: a search for universals. Universals of human language, ed. by J. H.Greenberg, C. A. Ferguson, and E.A.Moravcsik, 203-24, volume 1. Stanford: Stanford University Press. Rü^e-Draviga, Velta. 1976. Gibt es Universalien in der Ammensprache? Akten des 1. Salzburger Kolloquiums über Kindersprache ( = Salzburger Beiträge zu Linguistik2), 3-16. Tübingen: Verlag Günther Narr.

Children's Dispute and Negotiation Strategies: A Naturalistic Approach Muriel Saville-Troike University of Illinois Urbana-Champaign, Illinois

Children's arguments have perhaps received more attention from researchers than has any other isolatable speech event which occurs in childhood settings. Such attention has been justified from three overlapping disciplinary perspectives: cognitive, social, and linguistic. Research from the cognitive perspective has focused primarily on the developmental phenomena which are evidenced in children's disputing behaviors, and on the nature of the underlying competence which may be inferred from them. Piaget (1959), for instance, noted that children under the age of six years use physical responses or primitive arguments without logical justifications, and attributed the change in strategies at that age to increasing cognitive skills. While the study of child language in natural interaction contexts has convincingly demonstrated that younger children use much more complex strategies for argumentation and negotiation than Piaget credited them with (e.g., see Cook-Gumperz 1981), the cognitive perspective generally continues to explore these phenomena from an experimental paradigm. The social research orientation perceives conflict and its resolution as a highly pro-social activity, with children's performance in such events considered evidence for their level of social competence. Children's growing ability to take the point of view of peers in reaching agreement is considered to be related to their overall social-cognitive development (cf. Piaget 1959), including their understanding of social beliefs, values, and role-relationships (Anderson and Messick 1974, Genishi and Di Paolo 1982). Another important concept is that of cul-

136

Muriel Saville-Troike

ture-specific 'framing' of adversative routines (e.g. Mitchell-Kernan and Kernan 1975, Boggs 1978, and Lein and Brenneis 1978). The possible relationship of these social differences to cultural values and to child-rearing practices is elaborated by Boggs (1978), Genishi and Di Paolo (1982), Heath (1983), and Maynard (1985). The linguistic orientation focuses on conflict and resolution which are realized primarily through the medium of language, and emphasizes the range and complexity of verbal strategies to which children have access in directing and manipulating one another (e. g. Garvey 1975, Ervin-Tripp and Mitchell-Kernan 1977, Cook-Gumperz 1981). Most inquiry has been at the level of speech act analysis, but larger structures in interactional discourse have also been taken into account (e.g. Eisenberg 1978, Eisenberg and Garvey 1981, Garvey 1984, and Adger 1984). Findings from research conducted from each of these perspectives have contributed to our understanding of children's disputation as patterned, rule-governed cognitive/social/linguistic activity. As isolatable speech events, arguments have been shown to have predictable content and sequence, and to a great extent have been found to follow predictable developmental stages in their cognitive complexity. Social knowledge and influence is evidenced in accordance with patterns of group membership, and verbal/nonverbal strategies are bound by regular developmental sequences in language acquisition. Without questioning the value or validity of the regularities which have been discovered, it seems important to note that in spite of the great quantity of data which has been collected on this topic, very little of it has been truly naturalistic. Much 'interaction' research, for instance, is based on role-play situations which are artificially contrived, or on interviews with children in which they are asked to project their response to a hypothetical conflict, or on interaction with a puppet through which an adult researcher actually speaks. The primary question which must be asked of the existing knowledge base (also raised by Eisenberg 1978) is the extent to which children have responded to such tasks as they would in natural interaction with peers, or the extent to which their choice of strategies is being influenced by the fact that an adult is controlling the scene. Interaction research which uses dyads of children being observed and videotaped in small rooms from behind one-way mirrors (such as that reported by Garvey and Eisenberg) is a great improvement in this respect, but a number of questions related to naturalness remain. To what extent do contexts affect the nature and content of conflict and disputing? For

Children's Dispute and Negotiation

Strategies

137

example, what do children argue about in real nursery school and kindergarten situations? How do they resolve conflicts when there are no unnatural constraints on using physical force or merely leaving the scene? What roles and strategies are invoked when more than two children are involved in a dispute, or when adults are nearby in the role of caretaker? And while experimental control for linguistic and cultural variables is to be appreciated, how are disputes and negotiations conducted with heterogeneous groups of children who do not all share the same language and culture? Perhaps one can profitably take a purely naturalistic perspective only from the vantage point provided by prior regularized and idealized research formulations, although that is open to dispute. A great deal of variability is clearly to be expected when all experimental controls are removed, and the interpretation of such variable data should be greatly facilitated by having reference to experimental findings. The reverse is also the case, however: the interpretation of experimental data should be enlightened (and ultimately validated) by findings based on an empirical, naturalistic approach. In embarking on this research, I was especially heartened by the decision of Ferguson and Slobin not to limit the selection of exemplary child language studies in their landmark collection (1973) to those with 'clean data', but to include those which present provocative data as well [emphasis theirs]. I take further inspiration from Ferguson's (1979) clear demonstration that important contributions are to be made to the understanding of child language development by considering particular languages or varieties and even individual production, as well as language universals. In the present paper I hope to demonstrate the value of a naturalistic approach in the study of disputes, and to show that the empirical study of natural events can lead to the discovery of significant patterns which can advance our knowledge of this interactional genre. The results to be presented here represent a preliminary analysis of transcripts from 28 nursery school and kindergarten sessions that were videotaped over a period of several months, for a total of 76 hours. For each session a wireless radio microphone was attached to one of 10 focal children with varied native language backgrounds (4 English, 3 Japanese, 2 Chinese, and 1 Korean). The majority of the subjects were thus limited English speakers, and some spoke no English at all at the beginning of this study. The sound track on the video recordings picked up only what was said by or to these focal children, or within their immediate environment. Utterances which were produced in languages other than English are presented here in English translation.1

138

Muriel Savilie-Troike

The university-operated nursery school used in this study enrolled 16 three and four year old children, and had 4 adults in caretaker roles (1 teacher and 3 assistants). The focal subjects in this group were 2 Chinese and 3 English speakers, ages 3.2 to 4.3 at the beginning of the data collection period. The Chinese focal subjects were two brothers from Taiwan. Another Chinese speaker (a girl), three Korean, one Japanese, and one Norwegian speaker were also enrolled in the nursery school; the remaining children were native speakers of English. One adult caretaker was a Greek-English bilingual, and the others were monolingual English speakers. The kindergarten was part of a Midwestern public school system, and enrolled 26 children. Eleven were native English speakers (8 white and 3 black), 4 spoke Japanese, 3 Arabic, 2 Spanish, 2 Swahili, 1 Korean, 1 Chinese, 1 Hindi, and 1 Dutch. The 5 focal subjects in this group were speakers of Japanese (3), Korean, and English, ages 5.2 to 5.6 at the beginning of the study. Three adults were in caretaker roles (1 teacher, 1 student teacher, and one part-time aide); all were monolingual speakers of English.2 One of the first steps in this analysis involved speech act coding (using an adaptation of Dore 1978), which yielded ample justification for the question raised earlier about the validity of experimental research methodology for this topic. Directives indeed take very different form depending on the addresssee, even among the younger children in this sample. For example, the following directives/requests were all uttered by one English speaking boy, Michael (age 3.7), during the course of a single outdoor play period in the nursery school. They are categorized according to the role and status of the individual he was speaking to, and it is clear that his sociolinguistic competence already includes variability for different levels of politeness and formality. To peers: Stop that. Your're gonna get a swat. Let's go fishin'. Come on you guys. Come on. Green slime. Fix my wheelers. Let me see what's that. To a girl he wanted to persuade to play with him: You can pull me in there too. OK, Michelle? You get in the back and I'll drive. OK? You pull me now. OK? Let's go to see a show Michelle. OK?

Children's Dispute and Negotiation

Strategies

139

Michelle, do you wanna come with me to the show? Let's go get a ... this time you wanna go get a ice cream cone? Wanna go to Baskin-Robbins? To adult caretakers: Would you go get Stevie a bike? Would you put on my shoes? Additional evidence for children's sensitivity to the participant structure of disputing events comes from their nonverbal behavior. Before producing or responding to an 'adversative act', children in the study frequently looked around to see if an adult was watching them, and often ducked down out of sight behind partitions in the rooms before making their 'move'. They were very conscious of what adults might consider appropriate disputing behaviors, and this would be likely to affect even hypothetical responses made in the public/adult view. The first step in isolating disputes for analysis was to locate sequences between peers that consisted of an initiating move followed by a protest or countering move of some form. Nonverbal as well as verbal moves were included.3 This first round in a dispute exchange established the theme or topic. The dispute then developed further in ways which will be described, or it was terminated without further interaction. A total of 142 child-child dispute events appeared in the videotapes and transcripts which were examined - 91 in the nursery school sessions and 51 in the kindergarten classes. The naturally occurring rate of disputes for focal subjects in this naturalistic study is much lower than has been found in experimental settings, e. g., the one every 6 minutes reported by Eisenberg (1978). The relatively low rate of conflicts in the situations under investigation can be attributed in part to the greater variety of objects and activities that are accessible to children in these natural contexts vs. experimentally controlled environments (cf. Smith 1972), in part to the greater choice that focal children have in a natural situation to interact with particular other children (or with no one), and in part to the controls that are exerted by teachers and other adult caretakers in the natural scene. Categories of topics or themes which were found in these 142 dispute events were determined empirically by analyzing the individual exchanges, and it is perhaps significant that they do not exactly coincide with categories that have been formulated from other perspectives. I do not wish to argue that they constitute a necessarily universal set, but only to point out that they exhaustively characterize the dis-

140

Muriel Saville-Troike

putes which occurred in my data. These categories are listed below in order of relative frequency of occurrence, along with examples of opening sequences which illustrate each type. (It should be kept in mind when some of the utterances appear not to be idiomatic English that they may have been produced by limited English speakers, or represent literal translations.) 1. Action [A] disputes, or disagreements about doing or not doing something (N = 40; 28.8%). Scene: At hamster cage. Ch. # 1: Poke it. (Making jabbing motion with stick) Ch. # 2 : No! Scene: Clean up time. Ch. # 1: Move it! Move it! (Pointing to block) Ch. # 2 : 1 don't have to. 2. Object [O] disputes, including disagreements about possession, turn-taking, and the relative quantity or size of one's own holdings (N = 37; 26.1%). Scene: In playhouse. Ch. # 1: (Takes toy cup away from Ch. # 2) Ch. # 2: This is my cup. This is my cup. Scene: At board with magnetized letters. Ch. # 1: Give this one to me. Ch. # 2 : 1 have fewer. This is mine. 3. Participant/role [R] disputes, or disagreements about the inclusion or exclusion of participants in an activity, or about the role they will play (N = 33; 23.2%). Scene: At 'jail' (box with slats). Ch. # 1: Hey, we're in jail. (Gets in box with other children). Ch. # 2 : Getoutahere. Scene: In block area. Ch. # 1: Now I'm gonna be the dog. Ch. # 2 : No! 4. Space [S] disputes, including arguments about personal space and place in line or seating arrangement (N = 16; 11.3%). Scene: In line at monkey bars. Ch. # 1: (Breaks into line in front of Ch. # 2). Ch. # 2 : No! My first! Scene: Assembling for circle time. Ch. # 1: (Starts to sit in a chair.) Ch. # 2: Get out of the way. (Said in a low, very gruff voice.)

Children's Dispute and Negotiation

141

Strategies

5. Predication [P] disputes, or challenges to the 'right/wrongness' of another's assertion (N = 2; 8.5%). Scene: Outdoor play. Ch. # 1: There is a spider. There spider. Ch. # 2 : It is a roach. Scene: In block area. Ch. # 1: We're running out of blocks. Ch. # 2 : No, we ain't. 6. Competence [C] disputes, or arguments about the ability of a participant to do something (N = 4; 2.8%). Scene: In block area. Ch. # 1: You can't do anything. Ch. # 2 : Yes, you do. Scene: In playhouse. Ch. # 1 (toCh. # 3 ) : Jane's stupid. Ch. # 2 (Jane): You guys shut up. The distribution of these dispute types is not equal across nursery school and kindergarten sessions, which may be attributable to both developmental and situational factors. Relative distribution of childchild disputes in terms of percentage of occurrence in each context is as follows:

Nursery Kndg

A

O

R

S

P

C

25.3 33.3

35.2 9.8

27.5 15.7

3.3 25.5

5.5 13.7

3.3 2.0

The relatively higher occurrence of space disputes among older children, for instance, may be largely attributed to the fact that they are much more frequently required to form lines and sit in chairs, and the lower occurrence of object disputes to the fact that they are more likely to be involved in assigned tasks rather than have a free choice. Thus context appears to be a strongly significant factor in determining the type of dispute event which occurs. Conversely, there is no evidence of age differences for object disputes within the nursery school when percentages are calculated separately for three and four years olds. In an additional 35 disputes from this sample, the initial exchange was between a child and an adult caretaker. The most common category was action disputes (N = 18; 51.4%) for this pairing, as well as between peers. Most frequently (N = 14; 77.8%) these involved a child rejecting a request for action or a directive from the teacher. For example:

142

Muriel Saville-Troike

Scene: Preparing to come inside. T: Would you carry this in ? (Pointing to a cushion) Ch: It's very big. I don't want to. Scene: Cleanup time. T: I have something for your to clean up. Come on. Ch: No. In some cases this may have reflected a failure on the part of the child to recognize an indirect command as such, as in the following example where the child is a Japanese speaker just learning English. In this exchange he does not appear to realize that he does not really have a choice until the teacher repeats the 'question': Scene: Coloring a worksheet in ESL class. T: Do you want to color the kittens ? Ch: No. T: No? Ch: Yes. No. No color. T: Do you want to color them now ? Ch: OK. (He goes to get crayons) Less frequent in number were disputes between adults and children over objects ( N = 6 ; 16.7%), participant/role (N = 5; 14.3%), and predication (N = 5; 14.3%). The following are examples for each of these categories: Scene: Snack time. Ch: This is my juice. T: This is Jimmy's juice. Scene: Playing 'Squirrels and Trees'. Ch: I don't want a squirrel. I want a tree. T: You be a squirrel. Mary will be a tree. Scene: Board with magnetized letters and shapes. T: That's not a letter. Ch: Yes, it is. Children and teachers never disagreed over the other's competence, and in only one instance, was a child-adult dispute over space: Scene: In the playhouse. Ch: You, you can't stand here. T: What? I was here.

Children's Dispute and Negotiation Strategies

143

The most frequent initiating move for child-child action disputes was nonverbal (62.5%), with one child doing something that another objected to. A number of disputes in this category in the nursery school were also initiated verbally, with one child directing another to do or not do something (30.4%), but that occurred only once in the kindergarten data. The most common opposing move in the first round for both age groups was a simple negative. The younger nursery school children sometimes responded only with physical force (21.7%), but they also gave reasons in their responses (21.7%) more frequently than did the older kindergarten children (5.9%). Only the kindergarten children responded with an explicit statement of wrong-doing, such as "You can't do that," and "You are wrong" (meaning 'you are misbehaving'), or with a demand for an explanation, such as "How come you hit me?" The older children's greater reliance on nonverbal action and simple negation must be interpreted in the context of the heterogeneous class they were in; in this situation it is, seemingly paradoxically, evidence for their higher level of metalinguistic awareness. More complex forms of initiation or response never occurred in action disputes between speakers of different languages in the kindergarten. Younger children, on the other hand, gave directives and reasons in their native language even when the child with whom they were disputing could not understand them (see Saville-Troike 1985 b for a discussion of this phenomenon). All responses to both verbal and nonverbal strategies appeared to be quite appropriate no matter what the linguistic identity of the participants, lending further weight to the claim that children can communicate quite successfully without a common language when they are involved in contextualized face-to-face interaction (Saville-Troike, et al. 1984). Nonverbal initiating moves were also most common for disputes about object (51.4%) and space (87%), while participant/role, predication, and competence disputes were more likely to begin with verbal assertions (e.g., Hey, we're in jail), directives (e.g., Go away), statements of intent (e.g., Now I'm going to be the dog), requests (e.g., Can I come in?), and offers (Want dinner?). Initial response moves in all categories followed the pattern described for action disputes, with simple negation or rejection by far the most frequent response. Beyond the initial round, the most striking feature about the disputes was not their consistent patterning, but their extreme variability. In part this was owing to the different participant structures which occur in natural settings, in contrast to experimental studies. Of the 142

144

Muriel Saville-Troike

disputes which involved an initial exchange between peers, only half (54.9%) were abandoned or resolved without including any other participants outside of the dyad; other children became participants in just over a third of the total (35.2%), and adults in almost as many (28.9%). In the following dispute, for instance, four kindergarten children argue over possession of a book which accompanied the record they were listening to: Scene: At the listening center. # 1 pulls the book a little closer to her. # 2 holds the book back. # 1 I want this. # 2 I want. # 3 Who want? # 1 Auch. (sic) # 2 I make this you do. # 3 Like this. (He takes the book.) # 1 Auch. I want. # 3 Like this. (He puts the book in the middle of the table and grabs # l's hand.) # 1 No. (She pulls the book back in front of herself.) # 3 takes another book and looks at it. # 2 Hey, we got this. (He pulls on the book; # 1 holds fast.) # 1 No. # 2 Yes. # 1 No. # 3 Don't. # 4 taps on the table and points to another book, which is beside # 1. # 1 gives that book to # 4 and points to a book that is beside # 4. # 2 and # 3 take the book they had been fighting over earlier and put it on their side of the table. In the next example of group participation, the kindergarten teacher joined three limited English speaking boys in a dispute that began with teasing, and then devolved into a truth-value question of whether or not fire engines have bathrooms in them. The example is significant for how extensive a dispute can be maintained with such limited linguistic means. It is also interesting as an example of 'foreigner talk' (Ferguson 1975) on the part of the teacher accommodating to the English ability of the children.

Children's Dispute and Negotiation

Strategies

145

Scene: Standing by door where there is a poster of a fireman and fire engine. Taki ( # 2) has just come out of the bathroom. # 1 This is Taki. (He points to the picture on the door.) # 2 No. That's a fireman. # 3 No, Taki go bathroom and he go in the bathroom. # 2 I tell you teacher. T Bathroom in the house ? Yeah. # 1 Bathroom in the fire engine. # 2 Your. Your bathroom. # 3 No. Your. This is a fireman. T Firemen go to bathroom. (She is not really clear about what the children are discussing.) # 2 No. # 3 Yeah. # 2 No. # 3 He do. T In his house. In bathroom, he go to bathroom. # 2 No. No. # 3 They don't go to bathroom. T Well, Mitsu, everyone goes to bathroom. Only 27 (19%) of the disputes ended after the first round. These included nine involving physical action which were immediately settled by an adult. The most prolonged dispute lasted for 25 consecutive rounds, during which the three year old Chinese boy, Didi (D below) argued with his four year old brother, Gege (G), about the relative number of plastic magnetized letters each possessed. The following excerpts (translated from Chinese) illustrate the range of strategies which was brought to bear in this single dispute and its resolution. Scene: At board with magnetized letters and shapes. D I have fewer. This is mine. G Don't take any. They must be the same. D These are not as many as yours. No! Stop! No! No! G D You have letters. You have so many. G I don't have more than you have. D Oh, these are few. Oh, elder-brother has so many. Then I'll have very few. I also want so many. G D I have so few. I only have so few. All I want is English letters.

146

Muriel Saville-Troike

G D G

Do you want Chinese characters ? English. I want English letters. No. Chinese letters. Do you want me to return this "O" to you? D nods head affirmatively. G I return this to you, then do you return this to me? D shakes head negatively. G I return to you two, OK? This to you. Again I return this to you. Thus you have very many. Thus you have more. Didi accepts this arrangement and stops complaining. The resolution of this and similar disputes between the Chinese children was almost always in the younger brother's favor, which is surprising from an American point of view since he was much smaller and weaker. (It is interesting to note that when the four year old did reject his younger brother's requests or actions, he generally switched to his limited English to do so.) The reason for this outcome can be understood in part by listening to their mother's parting words (in Chinese) to the four year old each day: "You take good care of youngerbrother and teach him many things." Not only did the older brother accept this responsibility, he obviously wanted to, and even disputed with an unrelated four year old Chinese girl in the group (E) over who got to sit next to the three year old. Scene: Assembling on the rug for story time. G (to E) I don't want to be with you. E I don't even want to be with you. I want to be with younger-brother. G I am with younger-brother. E I also want to be with younger-brother. G Everybody protects younger-brother, right? E Younger-brother likes me very much. Do you know he likes me ? All three children sit down with the three year old in the middle. Part of the variability in these disputes can thus also be attributed to cultural differences in the groups of children. Further evidence for this can be seen in the extent to which the Chinese children temporized or compromised with one another, and the rarity of these strategies in disputes involving other children. The following are exchanges among the same three Chinese participants:

Children's Dispute and Negotiation

Strategies

147

Scene: In the playhouse. Didi takes a toy telephone away from his brother, Gege. G Mine, mine, mine. (He takes it back.) E Oh, no. Didi tries to get the telephone again, but his brother doesn't let him have it. E Put it away for today. Both brothers abandon the telephone and find other toys to play with. Scene: At the pegboard table. G OK, lend me. (Some rubber bands.) D No. G Lend me. D After using. Scene: Selecting activities during free choice time. G I still want to play with it. E I don't want to play with it. G Then I don't want to play with it either. I want to follow you. I want to play with things you want to play with. The Chinese children were also the most likely to use moral arguments, as in the following object dispute. This strategy was also used by Korean and Japanese children, but never by English speakers. Scene: Playing a Candyland game. Gege takes the spinner from E. Didi is watching. G It's my turn. E I'll give it to you. Don't snatch it away. Snatching is a very bad thing. D Is stealing, stealing, stealing is thief. To some extent children from the same language and cultural background joined forces in disputes against other children, and at the kindergarten level even consciously invoked ethnicity as an argument. In the following space dispute among five year old boys (all in English), for instance, two are Japanese (J1 and J2), one Arabic (A), and one Chinese (C). Scene: Lining up to return from a pullout ESL class. J1 pushes to the front of the line to join J2, who is already standing there. A You can't do that! I was going first. Can't do that. J1 But I got her. I got teacher, (i. e. 'I will get to hold the teacher's hand when we walk down the hall.')

148

Muriel Saville-Troike

A

Then you can't. (He tries to block the door with outstretched arms.) C (to A) Oh, wait. Give me your hand. A I don't want. I was first right here and you was getting the place. J1 But I could. I'm Japanese, right? J2 Right. J1 Just the Japanese people could. A accepts this 'logic', and the two Japanese boys lead the line back to the kindergarten room. In disputes between speakers of different languages, children who were just learning English used essentially the same verbal strategies in their limited second language as did the native English speakers, although the form of their utterances often was not grammatical. Even when help was not sought by the limited English speakers in this study (and perhaps not needed), English speaking children frequently assisted them, 'tattling' to the teacher in their behalf or intervening as their advocates in disputes with others. In the following kindergarten space dispute, for instance, J is Japanese, and El and E2 are English speakers. Situation: Claiming chairs for circle time. J leaves his chair briefly to get a picture book. El claims # l ' s chair. J My chair! My chair! Get up. E2 (to the teacher) He take it. Alex take it. E2 (to El) He get first, (i. e., 'J got the chair first.') Personality factors which appeared across cultural boundaries also contributed to the amount of variability that was found. These primarily involved the extent to which children capitalized on either their strength or their weakness. Children who used physical force to get their way did so in all kinds of disputes, while children who whined also used that tactic consistently. In the following disputes, for instance, the same three year old child Michael ( # 1 ) whose sociolinguistic competence was illustrated earlier, ultimately used force in all types of arguments: Situation: In the playhouse. # 1 Stevie, dinner's ready.

Children's Dispute and Negotiation Strategies

149

Want dinner? I made ... there's topping with the eggs. Topping, OK? Eat them. They're really good. They're really good. (He tries to force feed the (plastic) eggs to # 2.) C'mon. Eat 'em. Eat 'em. # 2 pushes him away. No. # 1 Eat 'em. # 2 No. # 1 Good. # 2 No. T Maybe he doesn't like eggs. Situation: In the playhouse. # 1 You get outa here before I kill ya. I'm gonna cut off your neck. # 2 I'll tell the teacher on you. If you cut off our necks. # 1 I'll kill you too. Situation: Outside in the yard. # 1 Let's go to see a show Michelle. OK? Another boy starts to pull her in a wagon. # 1 I want her to (hits other boy) go to the show with me. T That's wrong, Michael. Children who developed successful specializations in either threats or pleading were those who also mastered appropriate accompanying paralinguistic features, such as a gruff voice or a whine. Child # 2 in the following exchanges lowered the pitch of his voice and used a rasping quality for his 'disputing register'. Although smaller than most of the other kindergarten children, he managed to intimidate them quite successfully. Scene: In block area. # 1 You hit me. # 2 So? Scene: Between tables. # 1 is in # 2's way as he rides a large truck. # 2 Watch it. Move out of the way. # 1 moves out of the way. # 2 Move it, move it, move it. Scene: In coat area. An unseen participant pushes # 2. # 2 Somebody watch it!

150

Muriel Saville-Troike

While as noted at the outset, the findings presented herein must be qualified as preliminary, I believe enough has been shown to demonstrate the scientific value and potential importance of a naturalistic-anthropological approach to the study of children's disputing behavior. The data examined here revealed regularities which permitted a functional classification of disputes into five categories based on their semantic-pragmatic focus : action, object, participant/role, space, predication, and competence. Inasmuch as they were exhibited by children from a variety of language and cultural backgrounds, a tentative claim as to their universality would appear to be justified. It is clear, however, that context (nursery school vs. the more formally structured kindergarten) affected the frequency of particular types, and both individual and cultural differences, as well as degree of sociolinguistic sophistication, affected the utilization of particular negotiating strategies or tactics. These findings reinforce questions about the possible artifactual limitations of experimental studies, and support the need for naturalistic-anthropological research to complement experimental-psychological research, both as a source of hypotheses to be tested and as a means of validating experimental results. Only through adopting a holistic orientatiofi and admitting a variety of research approaches can we hope to reach a full understanding of the parameters of this universal human trait, and thé course of its development among different cultural groups.

Footnotes 1 For assistance in the transcription and translations presented here, I am especially grateful to Charlotte Blomeyer, Rey-Mei Chen, Jo Anne Kleifgen, and O Ook Whan. 2 Additional findings derived from analyses of parts of this database may be found in Kleifgen (1985) and Saville-Troike (1985 a, b). 3 The transcripts used for this analysis did not include a description of all nonverbal exchanges, so while they are included to indicate the range of forms that occur, no conclusions may be drawn from these data about the relative frequency of verbal versus nonverbal disputing behaviors. For a more complete discussion of the role of nonverbal behaviors in children's arguments, see Maynard (1985).

Children's Dispute and Negotiation Strategies

151

References Adger, Carolyn. 1984. Accommodating cultural differences in conversational style: a case study. Paper presented at the Delaware Symposium VI on Language Studies, University of Delaware. Anderson, S. and S. Messick. 1974. Social competency in young children. Developmental Psychology 10.282-93. Boggs, Stephen T. 1978. The development of verbal disputing in part-Hawaiian children. Language in Society 1.325-44. Cook-Gumperz, Jenny. 1981. Persuasive talk - the social organization of children's talk. Ethnography and language in educational settings, ed. by Judith Green and Cynthia Wallat, 25-50. Norwood, NJ: Ablex. Dore, John. 1978. Requestive systems in nursery school conversations: analysis of talk in its social context. Recent advances in the psychology of language: language development and mother-child interaction, ed. by R. Campbell and P. Smith. New York: Plenum Press. Eisenberg, Ann R. 1978. An analysis of the preschooler's use of language in the resolution of an adversative episode. Master's thesis, Johns Hopkins University. Eisenberg Ann R. and Catherine Garvey. 1981. Children's use of verbal strategies in resolving conflicts. Discourse Processes 4.149-70. Ervin-Tripp, Susan and Claudia Mitchell-Kernan. 1977. Child discourse. New York: Academic Press. Ferguson, Charles A. 1975. Towards a characterization of English foreigner talk. Anthropological Linguistics 17.1-14. Ferguson, Charles A. 1979. Phonology as an individual access system: some data from language acquisition. Individual differences in language ability and language behavior, ed. by Charles J. Fillmore, Daniel Kempler, and William S.-Y.Wang, 189-201. Ferguson, Charles A. and Dan Isaac Slobin. 1973. Studies of child language development. New York: Holt, Rinehart and Winston. Garvey, Catherine. 1975. Requests and responses in children's speech. Journal of Child Language 2.41-63. Garvey, Catherine. 1984. Children's talk. Cambridge, MA: Harvard University Press. Genishi, Celia and Marianna Di Paolo. 1982. Learning through argument in a preschool. Communicating in the classroom, ed. by Louise Cherry Wilkinson, 49-68. New York: Academic Press. Heath, Shirley Brice. 1983. Ways with words: language, life, and work in communities and classrooms. Cambridge: Cambridge University Press. Kleifgen, Jo Anne. 1985. Skilled variation in a kindergarten teacher's use of foreigner talk. Input and second language acquisition, ed. by Susan Gass and Carolyn Madsen. Rowley, MA: Newbury House. Lein, Laura and Donald Brenneis. 1978. Children's disputes in three speech communities. Language in Society 7.299-323. Maynard, Douglas W. 1985. How children start arguments. Language in Society 14.1-29.

152

Muriel Saville-Troike

Mitchell-Kernan, Claudia and Keith T.Kernan. 1975. Children's insults: America and Samoa. Sociocultural dimensions of language use, ed. by Mary Sanches and Ben G.Blount, 307-15. New York: Academic Press. Piaget, Jean. 1959. The language and thought of the child. London: Routledge and Kegan Paul. Saville-Troike, Muriel. 1985 a. Cultural input in second language acquisition. Input in second language acquisition: learners' use and integration of language in context, ed. by Susan Gass and Carolyn Madden. Rowley, MA: Newbury House. Saville-Troike, Muriel, 1985 b. Dilingual discourse: communication without a common language. Paper presented at the Second Language Research Forum, University of California at Los Angeles. Saville-Troike, Muriel, Erica McClure, and Mary Fritz. 1984. Communicative tactics in children's second language acquisition. Universals of second language acquisition, ed. by Fred R. Eckman, Lawrence H. Bell, and Diane Nelson, 60-71. Rowley, MA: Newbury House. Smith, W. M. 1972. Behavioral influences of arctic community environments. Behavior, design, and policy aspects of human habitats, ed. by W. M. Smith. Green Bay, WI: Office of Community Outreach. [Cited in Eisenberg 1978.]

Development of Negation in Tamil-Telugu Bilingual Children B. Lakshmi Bai Osmania University Hyderabad, A. P.

Introduction Acquisition of negation by children speaking English, German, French, Russian and Japanese has been studied in both its semantic and syntactic details by a number of scholars (Slobin 1966, Klima and Bellugi 1966, McNeill and McNeill 1968, Bellugi 1967, and Bloom 1970,1978). No such attempt has been made so far in the case of Indian languages, especially the Dravidian languages. Nirmala's work (1982) on four Telugu children contains a few observations on negation but, as the aim of her study was not to focus on this topic, a total picture of stage by stage development of negative expressions is not available in her study. The present research seeks to explore the development of the semantic categories of negation as well as the formal aspects of negative utterances in the speech of my children, Chetan (C) and Deepa (D). C is a boy older than his sister D by four years four months. The study is a longitudinal one covering C's speech from 1 ;0-5;0 and D's from 1 ;0-4;3. The children's father is Telugu and speaks to them essentially in Telugu. I use mainly Tamil with them. The children, thus, have had good exposure to both the languages right from their birth and are fluent bilinguals. Besides Telugu and Tamil, they have also been exposed to English from an early age as it is a language commonly used among friends and relatives both at home and outside. Detailed diaries of the children's speech were maintained since they were one year old. Tape-recordings were also made of their spontaneous speech at regular intervals of 10 to 15 days. Each such recording consisted of an average duration of ten minutes. Apart from sam-

B. Lakshmi Bai

154

pies of spontaneous speech, elicited data on matters relating to different phonological and grammatical points were also obtained from the children. A brief description of negation in Tamil and Telugu adult speech would give us the proper perspective for examining the same in the children's speech.

Negation in Tamil and Telugu Negative sentences in Ta(mil) and Te(lugu) differ from the corresponding positive ones in having either a full fledged negative verb or a negative marker, which can be a suffix or a free form, included in the verb phrase. The major negative verbs in Telugu are waddu [oddu] 'not needed', lee(du) '(it) is not' (verbal predication), and kaadu 'it is not' (nominal predication) indicating 'rejection', 'non-existence', and 'denial' respectively. Tamil has only two negative verbs, viz. veeNDaam to indicate 'rejection' and illai to express both 'non-existence' and 'denial'. A few examples are given below from both the languages:

Positive

Negative

Telugu (1) naaku kaafii kaawaali to me coffee is wanted 'I want coffee' (2) ikkaDa pustakam undi here book exists 'The book is here' (3) idi eluka this rat 'This is a rat'

(la) naaku kaafii oddu to me coffee is not wanted 'I do not want coffee' (2 a) ikkaDa pustakam leedu here book does not exist "The book is not here' (3 a) idi eluka kaadu this rat is not "This is not a rat'

Tamil (4) enakku taNNi to me

water

veeNuum is wanted

'I want some water'

(4 a) enakku taNNi

veeN Daam to me water is not wanted 'I don't want water'

Development of Negation in Tamil-Telugu Bilingual Children

(5) taNNi irukku water is 'There is water' (6) itu maaDu this cow 'This is a cow'

155

(5 a) taNNi illai water is not 'There is no water' (6 a) itu maaDu illai this cow is not 'This is not a cow'

The negative verbs with the exception of Te. kaadu also function as negative markers in their full or shortened forms in conjunction with other main verbs in different negative constructions as can be seen later. In Telugu, the positive finite verbs (defined by Krishnamurti and Gwynn (1985) as main verbs carrying gender-number-person (GNP) agreement with the subject of the sentence) are: (a) Imperative, (b) Past, (c) Future, (d) Durative (present, past, and perfect continuous), and (e) Hortative. Tamil, on the other hand, shows the following distinctions: (a) Imperative, (b) Past, (c) Present, (d) Future, (e) Durative, and (f) Hortative. Thus, unlike Tamil, Telugu does not distinguish the present tense form of the verb from the durative and uses the same form for both. The positive finite verbs in the two languages have the structure: Stem + tense/aspect-suffix-I-(G)NP. For example, Telugu (7) neenu I 'I read (8) neenu I

cadiw-EE-nu read - past suffix - 1st p. sg. it' annam tiN-Taa-nu food eat-future-lstp. sg. suffix 'I will eat my food'

A durative finite verb in Telugu has the structure stem + t/T (nonpast suffix) +un 'to be' (in its finite form), e.g. was-t-unnadi'She/it is coming or was coming'. The Tamil durative verb has the structure: Non-finite form of the main verb + past or present finite form of the verb iru 'to be'. The main verb itself is formed by adding the durative marker -ND to a stem with the past tense suffix. For example: (9) naan paDi-cc (u)-ND-u I read past durative suffix marker 'I was reading'

iru-nd- een be-past- lstsg. marker

156

B. Lakshmi Bai

Imperatives in Tamil and Telugu are inflected for the 2nd person singular and plural, e. g. Te. paaDu + u (—*• paaDu) 'you (sg.) sing', paaDu+aNDi (—>• paaDaNDi) 'you (pi.) sing'. The hortative is a type of imperative which includes the speaker also (see, for details, Krishnamurti and Gwynn 1985). The hortative suffixes are -daa in Telugu and -laa in Tamil. The hortative suffix is followed by 1st person plural suffix -m(u) in Telugu, and -m in Tamil, e. g. Te. poo-daam(u), Ta. poo-laam 'let us go'. Obligative verbs are formed by adding to the infinitive stem of a verb Te. -(w)aali and Ta. -Num, e.g. Te. weLL-aali 'one must go', Ta. kuDukku-Num 'one must give'. Besides these, the two languages also have a periphrastic future construction in which the infinitive stem of the main verb is followed by the present and durative finite forms of the verb poo 'to go' respectively in Tamil and Telugu, e. g. Ta. tara pooreen 'I am going to give', Te. ceppa bootunnaa 'I am going to say it'. The positive verb forms in both the languages are built on a simple stem or a complex stem consisting of the main verb and a vector, e.g. Te. cadiw-EE-nu (simple stem), 'I read it', cadiw-ees-EE-nu (complex stem) 'I have finished reading it'. The negative verbs in Telugu occur in the following paradigms: (a) Past, (b) Future (habitual), (c) Prohibitive or negative imperative, (d) Hortative, and (e) Obligative. The future negative verb in Telugu carries (G)NP contrast; the past and durative do not. The future negative verb has the structure: Stem + a (negative suffix) + (G)NP, e. g. tina-nu 'I will not/do not eat', tin-a-Du 'He will not/does not eat'. In the formation of the Telugu past negative verb, leedu (the 3rd neut. sg. of the existential negative verb lee) is added to the infinitive stem of a verb, e.g. tina leedu 'I/you/he, etc. did not eat'. In the negative imperative or prohibitive, the 2nd person suffixes viz. -u (sg.) and -aNDi (pi. and polite sg.) are added to the negative stem consisting of a verb stem+the negative suffix -ak, e.g. weLLa-ak-u (weLLaku) 'You (sg.) don't go', weLLa-ak-aNDi (weLLakaNDi) 'You (pi.) don't go/please don't go'. Telugu has yet another prohibitive verb in which the infinitive stem of a verb is followed by waddu the negative verb of refusal, e. g. cadawa waddu 'Don't read or let us not read'. Negative permissive is formed in Telugu by adding kuuDadu to an infinitive verb stem, e. g. nuwwu cuuDa kuuDadu 'you must not see'. As distinct from Telugu, Tamil has the following contrasts in negative verbs : (a) Past, (b) Present, (c) Future, (d) Imperative, (e) Hortative, and (f) Permissive. Of these, the future negative expresses GNP contrasts, but the present and past negatives do not. The negative past

Development of Negation in Tamil-Telugu Bilingual Children

157

verb consists of the infinitive stem followed by the negative verb illai 'to be not', which occurs in its reduced form -la in fast speech, e.g. naan vara-la 'I did not come'. The negative present in Tamil is expressed by adding illai to a verbal noun in -du, e.g. avan varadu illai (avan varadillai) 'He is not coming'. Tamil future negative verbs are formed by adding one of the finite forms of the verb maa TTu (which in old Tamil had the meaning 'can, will', etc. Arden 1976) to the infinite stem of a verb, e.g. naan vara maa-TT-een 'I will not come', avan vara maa-TT-aan 'he will not come', etc. Negative imperatives are expressed in Tamil by inflecting the verb with the negative suffix -aa, to which are added the 2nd person sg. suffix -dee or its plural counterpart diingaL, e. g. varaa-dee 'you (sg.) do not come', varaadiingaL 'you (pi. or polite) do not come'. Negative hortative in Tamil is formed by adding veeNDaam (the negative verb of refusal) to an infinitive verb stem, e. g. nambaL vara veeNDaam 'we need not come or let us not come'. Negative permissive in Tamil is formed by adding the negative verb kuuDaadu 'should not, must not', to the infinitive stem of a verb, e. g. vara kuuDadu 'one must/should not come'. The negative equivalent of the periphrastic future construction in Tamil is formed by adding illai 'to be not' to a complex verbal noun made up of the infinitive stem of a verb -I- the verbal noun pooradu, e. g. naa vara pooradu illai 'I am not going to come'. Unlike the positive verbs, negative verbs do not generally take the vectors in these two languages. The use of a vector in a negative verb has special semantic effect.

Semantic Categories of Negation The semantic categories of negation which occur in children's language are identified by Bloom (1978) as (1) Non-existence/disappearance, (2) Rejection, (3) Prohibition, and (4) Denial. Bloom has observed that there is a clear progression in the development of negative categories not only in the speech of English speaking children studied by her and others, but also in the case of a Japanese child studied by McNeill and McNeill (1968). She points out that 'The three semantic categories appeared in the children's sentences in the order non-existence, rejection, and denial and there was a corresponding development in the form of their syntactic representation' (Bloom 1978:191). The semantic category prohibition according to her, was the last one to emerge.

158

B. Lakshmi Bai

Though Tamil and Telugu have distinct verbs illai and leedu respectively to indicate non-existence, they were not the first verbs to emerge in our children's speech. In both the children's speech the negative verb of rejection namely Te. oddu and Ta. veenNDaam emerged before illai and leedu. Between Te. oddu and Ta. veeNDaam, both the children showed an early preference for the former as perhaps, it is phonetically simpler. The word addu (oddu) appeared in C's speech at 1; 0.27 when he was still at the one word stage. For instance, (10) Mother: nii taataa kiTTa poo you grand near go father 'Go to grandfather' (11) Chetan: addu addu no no 'No, I dont' want to go' Likewise addu emerged in D's speech at 1 ;2.20, which she used to reject food offered to her by her mother. Roughly after a fortnight, she started using it in her two-word utterances also. At 1;3.7 when her mother wrongly thought that she wanted some water and offered it to her D said the following: (12) addu nannaa do not want water 'I don't want water' (The word nannaa was D's version of Ta. taNNi 'water'.) The words oddu and veeNDaam can be used to express either rejection by the speaker (i.e. when the speaker does not want something or when he does not want to do something), or rejection of what some one else does (Bloom's 'I don't want... and Don't...'). Both the children first used oddu (or veeNDaam) to express the meaning 'I don't want' and later to convey 'don't do it'. As was shown earlier, the first of these meanings was expressed by C at 1; 0.27 whereas the second was expressed by him at 1; 6.20 as can be seen in the following examples: (13) Mother: iru nii aZudaa puunaya kuuppiDuveen wait you if you cry cat acc. I will call 'Wait, if you cry I will call the cat' (14) Chetan: addu addu 'Don't don't'

Development of Negation in Tamil-Telugu Bilingual Children

159

(15) Mother: enna oddu what don't 'What is it that you don't want?' (16) Chetan: muuna cat 'Cat' The Tamil verb of rejection veeNDaam was attempted by D at 1; 5.20 which she tended to use side by side with Te. addu, and learning later to limit it to Tamil alone. The word veeNDaam occurred in C's speech when he was 1;8.19 as shown below: (17) Mother: ceetan Suus pooDaTTumaa Chetan shoes shall I put on 'Chetan! shall I put shoes on you?' (18) Chetan: veeNaa uwaa no sore 'Don't, (I have) a sore (on my foot)' The second negative verb to appear in the children's speech was Ta. illai in preference to Te. leedu. But while D used this verb to indicate the notion of 'non-existence' first, C expressed 'denial' with it. Compare the first use of illai by D and C in the following sentences at 1; 4 and 1; 8 respectively: (19) Deepa:

appaa ilia father is not 'Father is not there'

(The care-taker takes D to the garage where father's scooter is usually kept. D notices that the scooter is missing and concludes that he is not there.) (20) Chetan:

(21)

tittu picci ilia chetan mad not 'Chetan is not mad' amma guni ilia mother clean is not shaven head 'Mother is not one with shaven-head'

(C said (21) after his father jocularly remarked that mother was guNDu, i. e. one with cleanly shaven head, and C disapproved of it.)

B. Lakshmi Bai

160

The semantic category of 'non-existence' appeared in C's speech only when he was 2; 0.20: (22) Chetan:

babulu viiTTukku aaNTi ilia name of house-to aunt is not there a boy 'Aunt is not there at Babulu's house'

(C said (22) returning from Babulu's house whose mother was not at home at that time.) In D's speech the category of denial appeared when she was 1; 7. But instead of choosing a new verb kaadu appropriate for this purpose in Telugu, she extended the form oddu, which was already known to her, to perform the new function: (23) Chetan:

(24) Deepa:

diipaa appaa aSaDumaa Deepa father is crazy 'Deepa, father is crazy' oddu camattu no good 'No. He is good'

The expected adult equivalent of (24) in Tamil is illai camattu and in Telugu kaadu manciwaaDu. D started using Te. kaadu for denying something when she was 1 ;7.20 as in the following: (25) Deepa:

amma idi eeNTi? Mother this what 'Mother what is this?' (pointing to mother's shadow) (26) Mother: neZal maa shadow dear 'It is a shadow' (27) Deepa: neZal kaaduammaa muunji shadow not mother's face 'It is not a shadow. It is mother's face' We have thus noticed that the order of emergence of the negative categories in D's speech is rejection - non-existence-denial and in that of C rejection-denial-non-existence.

Development of Negation in Tamil-Telugu Bilingual Children

161

Formal Expression of Negation As far as the formal aspects of negation are concerned, both the children used the imperative forms of the verbs in the positive first; the corresponding negative imperatives were the first to emerge and gain in productivity among negative verbs. D's early negative imperatives were Telugu and C's Tamil: Negative

Positive Chetan (l;8-2;0) Ta.akka Ta. vey Ta. taa Ta. pooDu Te. tii Deepa (1;3-1;4) Te. etko

'sit down' (2;1) Ta. caapaada'don't eat it' 'keep it' Ta. toDaada 'don't touch it' 'give' 'put it on' 'take it out' 'lift me'

(1;6) Te. kuutowoddu 'don't sit' Ta. aapu 'eat' Te. poo woddu Te. tii 'take it out' 'don't go' Ta. aandko 'move' Te. koTTa woddu 'don't beat'

Except for the imperatives, the other forms of the negative paradigm were not produced by D until productive use of present, past, and future verb forms was achieved by her in the positive set. That is, between 1; 0 and 1; 8, she had 26 future, 15 present, and 14 past verbs (both in 1st p. sg. and 3rd p. sg.) in the positive. But, except for the stray case of raale 'did not come' at 1; 7, there is no example of a verb form inflected for negation in her speech. It is interesting to note here that during this period all her negative answers typically involved negative markers like oddu, illai, leedu, etc., and no verb inflected for negation as will be obvious from the following conversation: (28) Q. appa inTLo unnaaDaa leeDaa father in the is or not house 'Is father at home or not?' (29) Deepa: leedu 'No'

162

B. Lakshmi Bai

(30) Q. annam tinnaawaa leedaa? food did you eat or not 'Have you eaten or not' (31) Deepa: leedu 'No' (32) Q. paal taagutaawaa? milk will you have 'Will you have some milk?' (33) Deepa: oddu naanaa (Ta. veeNDaam) not not 'No' When she started showing negative inflection for verbs at 1;7.20, there were more instances of future negative forms in her speech than the others, which is in line with her preference for future forms in the positive. Future negation was expressed by her with -a-nu {-a, neg., -nu, 1st sg.) in Telugu and was gradually generalized to include Ta. verbs also. Thus, besides Te. verb forms ceyyanu 'I won't do', peTTanu 'I will not keep it', she also created new verbs with Ta. stems inflected like the Te. ones, e. g. collanu 'I will not say so', toDanu 'I will not touch it'. Within a few weeks, she started using future negative verbs of Tamil also, e.g. tara maaTTeen 'I won't give'. As distinct from D, C's early verbs in the positive set were present and past tense forms and not the future. That is, from 2;0 to 2;1, there were 6 present verbs, 8 past ones and no future verb in his speech. Correspondingly, at 2; 1.23, only 6 negative past tense verbs (all Tamil based) occurred in his speech, and there was a conspicuous absence of all other negative verb forms. Of the six past negative verbs, valla 'did not come' was a solitary instance of an unanalyzed form though used by him correctly to suit the context. It occurred in the following utterance at 2; 1.10: (34) danni edukku valla a girl's name why did not come 'Why did not Dhanni come?' The early negative past forms used by C deviated from the adult norm in the following ways: (a) The negative verb illai was used by him for marking negation instead of its shortened version -la, which gets fused with the verb stem in adult speech. For example:

Development of Negation in Tamil-Telugu Bilingual Children

163

(35) naa ettu ilia I take not 'I did not take it' (The adult equivalent of (35) is naan eDukkala) (b) Unlike the adult model in which a negative past verb is built on the infinitive stem, C's speech showed a tendency to generalize one of the two already existing stems of the positive, namely the imperative and the past to the negative set also. For example: (36) tolenju ilia (use of past positive stem) lost not 'It is not lost' (The adult form is tolaya-la) (37) vccuDu ilia (use of imperative stem) keep it not 'I did not keep it' (The adult form for (37) is vekka-la) Subsequently, however, starting from 2; 10 C's negative past sentences underwent a number of reorganizational changes. Between 2; 10 and 3; 0 his speech showed several examples in which the Ta. negative verb illai was added to an affirmative sentence to render it into a negative one. For example: (38) (2; 10.23) koSu kaDi-ccu- du ilia mosquito bite- past- neut. not suffix sg. "The mosquito did not bite me' (Lit. 'the mosquito bit me not'). The adult form of (38) is koSu kaDikala. The next stage in the reorganization of the past negative was characterized by the realization that past negative sentences in Tamil do not take the finite form of a verb, i. e. the form inflected for GNP. He started, therefore, to attach the negative marker illai to the verb stem without GNP inflection. But still his negative utterance deviated from the adult pattern in that instead of using the infinitive stem, he continued to use the corresponding stem of the positive paradigm. Moreover, unlike the adult pattern, the negative element was not fused with the verb stem in his speech. For example: (39) (3;3.20) naa aDiccu ilia I beat not 'I did not beat'

B. Lakshmi Bai

164 (40)

(41)

naa aDi- cceen I beat-past- 1st p. sg. suffix 'I beat' naa aDi- kka-la I beat-inf.- not suffix 'I did not beat'

(positive correspondent in adult speech)

(adult equivalent of 39)

The next step in the reorganization was to use the negative suffix -la instead of the erstwhile illai, thus bringing his forms closer to the adult pattern. But he continued to use past stem of the positive instead of the infinitive stem of the adult model, e. g. aDiccala 'did not beat', paDiccala 'did not write'. It is interesting to note here that D's past negative sentences of Tamil also showed some of these stages as will be evident from the following: (42) Deepa:koTTi - T T - een leedu spill past 1st sg. not suffix 'I did not spill it' (Lit. - I spilled it not') (43) koTTi -TT- een (corresponding positive in spill past 1st p. sg. adult speech) suffix 'I spilled it' (44) koTT - a la (adult equivalent of 42) spill inf. not suffix 'Did not spill it' (45) Deepa:adu onju -ttu leedu that break past neut. not suffix sg. 'That one did not break' (46) adu onju-ttu (corresponding positive in that one break-past- neut. sg. Deepa's speech) suffix 'That one broke' (47) adu oDay - a-la (adult equivalent of 45) that break inf. not suffix 'That one did not break'

Development of Negation in Tamil-Telugu Bilingual Children

165

It should be pointed out here that the tendency to recaste a negative sentence by adding a negative element to the corresponding positive sentence was not confined to the Ta. negative past alone. Such a tendency was observed in both the children's speech in some of the other negative constructions as will be evident from the following: (48) Deepa (1 ;9) tuunga -laam oddu (Hortative) sleep hort. not suffix 'Let us not sleep' (Lit. 'let us sleep not') tuunga -laam (positive correspondent (49) in adult speech) 'Let us sleep' (50) tuunga veeNDaam (adult equivalent of 48) sleep not needed 'Let us not sleep' een oddu (51) Deepa (1; 9) tuunga poo -rpres. 1st p. not (Periphrastic sleep go suffix sg. future) 'I am not going to sleep' (Lit. 'I am going to sleep not') (52) tuunga poo -rdilia (adult equivasleep go pres. nominal not lent of 51) suffix suffix '(I am) not going to sleep' di leedu (53) Deepa layT occ- in(1:9.22) light come past neut. not suffix sg. 'The light did not come' (Lit. 'the light came not') layT raa leedu (adult equivalent of 53) (54) light come not 'The light did not come' a poo -reen ilia (55) Chetan naa va- rnot (2; 5.25) I come pres. inf. go pres. 1st suffix suffix p. sg. 'I am not going to come' (Lit. I am going to come not) (56) naa va- ra poo -rilla dI come pres. inf. go pres. nominal not suffix suffix suffix 'I am not going to come' (adult equivalent of 55)

B. Lakshmi Bai

166

Though this tendency was of a general application, it must be borne in mind that C and D did not always add a negative element to a positive sentence to arrive at its negative counterpart. Note, for example, that at the same time when D was rendering a past negative sentence in Tamil by appending a negative element to a corresponding affirmative sentence, she did not do so in the Te. sentence with the exception, however, of occindi leedu 'it/she came not' (for adult raa leedu). Similarly neither D nor C subjected future negative sentences in Tamil or Telugu to such an analysis. That is, there were no instances like the following in their speech: (57) poo -veen go fut. 1st p. suffix sg. 'I will not go' (58) poo-taa -nu go fut. 1st p. suffix sg. 'I will not go'

maaTT will not

een 1st p. sg.

leedu not

What the children seem to have done is, wherever there was an asymmetry between the positive forms and their negative counterparts in the types of stems used, they tended either to postpone sorting out the details of such negative constructions by adding directly a negative element to an affirmative sentence, or to regularize the negative paradigm to bring it closer to the positive correspondent. This general strategy accounts for the following characteristics observed in the children's negative constructions: (a) Unlike the adult grammar where a complex verb involving a main verb+vector does not normally constitute a stem of the negative verb form, the children's speech showed abundant use of such a complex verb stem. For example, D's aDi-ccu-Du veeNDaam 'don't beat' for adult aDikka veeNDaam. (b) Infinitive stem of Ta. past negative verb was replaced by the corresponding past tense stem of the affirmative. (c) Suppletive stems tended to be replaced by the regular ones. For example, occindi leedu, occa leedu in D's speech for raa leedu. (d) Though negative verbs like Ta. Te. illai, leedu, kaadu, oddu, veeNDaam were learnt by the children at a very early age in their semantic details, at a later stage of syntactic development they were reanalyzed into a positive verb + a negative element, e.g. D's veeNum oddu and C's veeNum ilia for Ta. veeNDaam 'not needed'.

Development of Negation in Tamil-Telugu Bilingual Children

Discussion In the preceding sections, a description of the development of semantic categories of negation and the formal expression of negation in the speech of two bilingual children was presented. We have noticed that the order of emergence of the negative categories in D's speech is rejection-non-existence-denial and in C's speech rejection-denial-nonexistence. Nirmala (1982) also observed that the verb of rejection, namely oddu appeared before non-existence in the speech of the four Telugu children studied by her. The findings of our research on children speaking Dravidian languages thus, do not support the developmental priority of non-existence over rejection as indicated in the research of McNeill and McNeill (1968), Bloom (1970), and Shapiro and Kapit (1978). These children rather share their preference for rejection with the children studied by de Villiers and de Villiers (1979). As for formal expression of negation, Wode (1977) sets up a general theory of the development of negation, according to which children's negative utterances show the following four stages: Stage I consists of one-word negation. Stage II is two-or multi-word negation in which the negator is placed either before or after the rest of the utterance. Stage III is characterized by an intrasentential negation, though the placement of the negator may not match completely with that of the adult model. In Stage IV, the negative element is correctly placed as in the adult model. A close examination of C and D's negative utterances show clearly that though in Tamil and Telugu negation is expressed essentially by means of verbal inflection unlike English, which involves the use of a negative particle, the strategies adopted by these children are similar to those reported for English and many other languages. Both C and D's speech showed stage I negation of Wode, when their negative utterances consisted solely of addu (oddu) to express rejection. Stage II negation is reflected in such utterances as addu aayaa (D, 1; 2.20) 'no need to use the toilet' or paal addu (D, 1; 2.20) 'I/do not want milk', or C's aato veeNDaam 'don't want to go in the auto'. A continuation of stage II negation is reflected in both C and D's attempts at creating negative utterances by appending a negator to an otherwise affirmative sentence. D's (1 ;9.11) koTTiTTeen leedu, instead of koTTala 'I did not spill it' or adi layTonjupeettu kaadu, instead of anda layT oDeyala, and C's koSu kaDiccudu ilia (2; 10.23) for koSu kaDikala 'the mosquito did not bite me' are all examples of this tendency. What is interesting to note here is the fact that when C and D

167

168

B. Lakshmi Bai

started sorting out negative inflection of the verbs, in all those cases which were complex in their inflectional details, e.g. suppletive stem alternation, periphrastic constructions, or negative paradigms built on a stem different from the corresponding affirmative paradigm, the strategy of an earlier stage was continued for a prolonged period of time. In such negative constructions of the adult model which require the combination of a main verb and a negative verb following it (e. g. Ta. veeNDaam, illai, Te. oddu, leedu), Wode's stage III would reflect only vacuously in the children's speech, since right from the beginning such a negator was regularly placed by them utterance-finally in their multiword utterances. But, in the case of constructions involving inflection for negation, Stage III would mean a total integration of the negative element within the verbal complex rather than being treated as extraneous to the verb form. That is, C's Ta. past negative forms like tolenju ilia, 'it is lost not', and veccuDu ilia 'keep it not', etc. represent Stage II negation while D's kiinjala 'it did not tear' (for adult kiZiyala) comes under Stage III negation, though at the three examples cited above are deviant from the adult standard. To conclude, an examination of the step by step development of negative utterances in the speech of C and D shows that the development of the semantic categories did not follow the same pattern as has been reported for children speaking major IE languages. But, the formal expression of negation involved the same strategies in our children as have been observed in such varied languages as German (Wode 1979), Estonian (Vihman in press), and English.

References Arden, A. H. 1976. A progressive grammar of the Tamil language fifth reprint. Madras: The Christian Literature Society. Bellugi, Ursula. 1967. The acquisition of negation. Doctoral dissertation, Harvard University. Bloom, L.M. 1970. Language development, form and function in emerging grammars. Cambridge, MA: MIT. - . 1978. Development of language content/form: variation and complexity. Language development and language disorders, ed. by Lois Bloom and Margaret Lahey, 165-200. New York: John Wiley and Sons. de Villiers, P. A. and J.G.de Villiers. 1979. Form and function in the development of sentence negation. Stanford University Papers on Research in Child Language Development, 37-64.

Development of Negation in Tamil-Telugu Bilingual Children

169

Klima, Edward, S. and Ursula Bellugi. 1966. Syntactic regularities in the speech of children. Psycholinguistic papers, ed. by John Lyons and R.J. Wales, 183-208. Edinburgh: University Press. Krishnamurti, Bh. and J.P.L.Gwynn. 1985. A grammar of modem Telugu. New Delhi: Oxford India McNeill, David and Nobuko McNeill. 1968. What does a child mean when he says 'no'. Proceedings of the Conference on Language and Language Behaviour ed. by E.M.Zale, 51-68. New York: Appleton Century Crafts. Nirmala, Chervela, 1982. First language (Telugu) development in children: a short descriptive study. Doctoral dissertation, Osmania University, Hyderabad. Shapiro, Theodore and Richard Kapit. 1976. Linguistic negation in autistic and normal children. Journal of Psycholinguistic Research 7, No. 5,337-351. Slobin, Dan 1.1966. Abstracts of Soviet Studies of child language. The genesis of language: a psycholinguistic approach, ed. by F.Smith and G.A.Miller, 129-148, Cambridge, Mass: MIT Press. Vihman, Marylyn M. (In press.) Language differentiation by the bilingual infant. To appear in Journal of child Language. Wode, H. 1979. Four early stages in the development of LI negation, Journal of Child Language 4. 87-102.

Acknowledgement * I express my sincere thanks to Professor Bh. Krishnamurti for reading the draft of this paper and making valuable suggestions.

Italian and Vernacular Writing in Southern Regions of Italy: A Semantic Approach to Children's Language Behaviour Maria Antonietta Pinto University of Rome Rome

This paper will report and comment on the results of a research carried out in three southern regions of Italy: Calabria, Apulia, and Sicily. It will focus on the kind of differences in emotional, cognitive and cultural approach that 5th grade children show through written performances in the Italian and vernacular languages. The reader may be legitimately surprised of hearing of written vernacular, which in fact contrasts with formal school practice. Moreover, in the small towns that have been selected (Lamezia Terme in Calabria, Lucera in Apulia and Gela in Sicily), it can definitely be asserted that one would not find any official sign or document in vernacular, aside from sporadic attempts of folklore poetry. In other words, the contexts under focus are neither those of official bilingualism, as for instance in the Aosta Valley (French-Italian) or in Trentino (German-Italian), and concomitant diglossia with local vernaculars; nor those of simultaneous diglossia and bilingualism Italian-vernacular, as for instance in Venetia, in Friuli and in some areas of Sardinia.1 The varieties spoken in the context of the present research are all Neo-Latin, and nothing has been done so far for a full aknowledgement of them. However, it can't be said that it is a clear-cut diglossic situation, since Italian, at least the regional variety of it, is not confined to purely formal usage and increasingly breaks into oral conversation. This is partly due to socioeconomic differences, because high socioeconomic level people are taught Italian as their mother-tongue and have at least a passive knowledge of vernacular, while low socioeconomic people are taught vernacular as their mothertongue and Italian at school. On the other hand, Italian inevitably

172

Maria Antonietta

Pinto

tends to replace vernacular through technology and mass-media information, and this is a widespread phenomenon across all social classes.2 Such situations may be considered as relatively trivial in the Italian context, if compared to those of many other regions of this country where linguistic contrasts between varieties are sharper both on a structural and on a functional level, and are often strongly underlined by native speakers. Accordingly, it would be even more interesting to see what kind of cognitive and emotional processing is possibly linked to Italian or to vernacular. With reference to the general purposes of the present work, the following points were tested: 1) the possibility that elementary school children would accept writing in vernacular, whereas their whole cultural environment would hardly conceive of such a possibility both for technical and normative reasons; 2) the hypothesis that by mere codeswitching an identical topic would be angled along two different kinds of mental attitudes; 3) the hypothesis that these differences, which are supposed to be in some way code-dependent, were akin to the Bernsteinian dichotomy "elaborated" vs. "restricted code", which seems to be social-class dependent.3 However, the meanings of "elaborated" and "restricted" are taken here in very general terms and do not claim to follow the specificity of the linguistic markers singled out by Bernstein and his team.4 The sense of the above hypothesis is that by asking an Italian version of a given topic, the output would likely be more universalistic, explicit and context-independent than the vernacular version of the same topic. The coding grid that will be presented below, will illustrate the kind of linguistic markers adopted; 4) the hypothesis that a variation in semantic areas would induce a sort of "cognitive switch" that is likely to underlie the Italian-vernacular distinction; 5) the hypothesis that these semantic differences would also be a function of social class differences.

Experimental design In each of the three selected towns, a group of 30 5th grade children was singled out, which summed up to a total of 90 children. Each group was split into two socioeconomic levels through cluster analysis,5 following the results of a questionnaire.

Italian and Vernacular Writing in Southern Regions of Italy

173

Tasks Each child was asked to do 20 written performances about 10 different topics, 5 of which related to local culture and the other 5 to more general themes. Each topic was asked to be developed once in Italian and once in vernacular, after a two or maximum three days interval. The order of the language required by the task was systematically reversed: the Italian version first, then the vernacular one, and vice-versa. Here is the list of the 10 topics. Local culture.

General culture.

Good and bad qualities of my classroom desk-mate. A square of my home town. A Sunday outing to the countryside. A little party at home. A typical day at school.

Injustice in the world. The pollution. During your sleeping hours some people are at work. Starvation in the world: as you celebrate Easter some people are starving elsewhere in the world. The energy crisis.

The coding grid The coding grid mostly consists of semantic categories grouped under two major headings related to two opposite and complementary attitudes. The psychological common denominator of one of these, which will be hereon abridged as S (corresponding to Subjectiveness), is proximity, both in a mental and in spatial sense. This trait is embodied by any utterance where immediate subjectiveness is underlined, often with emotional involvement, and also where visible and tangible objects and events are mentionned as instances of general ideas. In other words, within the S area, outer reality tends to be related to the subject's own experience. The opposite standing is polarized on objectiveness (abridged as O), in that personal involvement is avoided in favour of external and universalistic principles. One's own emotions are admitted only if conveniently ruled by moral statements of a general value, which often sound rethorical. Within the O area, the speaker tends to blot himself out for the sake of otherness and generality. In the specific categories listed below, the reader will be able to recognize a blend of linguistic functions underlying both the S and the O area: in Jakobson's terms, the referential, the phatic, the expressive

174

Maria Antonietta Pinto

and the conative functions. What differs from one to the other area is that the referential function is realized on a particularistic and concrete level (see cat. SI and S5), whereas in the O area the same function is realized on an objective and universalistic level (see cat. 0 1 , 0 2 and 03). In the S area, the expressive, conative, and phatic functions are realized on a colloquial level, often in a jokeful way (see cat. S2, S3, S4), whereas in the O area the phatic function is not represented, and the expressive and conative ones are realized on a patently accurate level (see cat. 04).

S Area. 51) Self-references, singular or plural. Linguistic units (L.U.): personal and possessive pronouns and adjectives. Meaning: the speaker simply refers to external things as belonging to his own immediate sphere. SI units have no introspective value but merely mark a possessive relation to things. 52) Symbolic involvement of the interlocutor in dialogue. L. U.: one or more utterances. Meaning: S2 units appear when the speaker handles a given topic by addressing an imaginary interlocutor. 53) Picturesque descriptions ofsingle aspects of a given person or of whole scenes. L. U.: adjectives, nouns, idioms, one or more utterances. Meaning: S3 units refer to all these visibly connoted expressions as opposed to purely neutral descriptions. They appeal to comic sense through markedly colloquial style. 54) Devaluative anaphoric and exophoric references. L. U.: all kinds of anaphoric and exophoric deictics. Meaning: colloquial style sometimes shifts to stereotyped and generally rough modalities of reference, which lack specificity if compared to the above S3 units. 55) Illustration of general concepts through practical instances. L. U.: one or more utterances. Meaning .the focus here is individuality as opposed to generality.

Italian and Vernacular Writing in Southern Regions of Italy

175

01) Historical, geographical, artistic, cultural references, surnames and Family Names. L. U.: nouns and adjectives. Meaning: the above references are supposed to witness a certain acknowledgement of external facts or entities in their autonomous existence. This category focusses on the otherness and objectiveness of things. Therefore the presence of the signifier perse will be considered as more important than the assessment of the correct understanding of the meaning. 02) General statements. a) factual statements; b) prescriptive, normative statements. L. U.: one or more utterances. Meaning: the 0 2 units embody a generalization process, just as the above S5 units embody a particularization process. In 0 2 b, such a tendency goes as far as to assert what "should be" versus what simply "is" (02 a). 03) Conceptually complex categories. L. U.: nouns, simple or compound, adjectives. Meaning: this category refers to all attempts to give a more acceptable look to linguistic expressions, whether by ostensibly censuring socially disapproved contents, or by exhibiting rethorical precoded formulas. Stylistic purposes are realized here on a formal level.

Intersection between general and particular (I): 0 2 + S5 or vice-versa. L. U.: at least two utterances, explicitely or implicitely connected. Meaning: as a sort of compromise between some of the categories of the two global areas, the I category typically stands up when subjects develop general topics, where they often illustrate a given concept through instances immediately related to this concept. This suggests to code the whole as one and only unit. Aside from semantic categories, three more variables were considered: Information Richness (IR), expressed by the number of predicates, Syntactic Complexity (S. C.) expressed by the total coordinate clauses divided by the total subordinate clauses. In the end, vernacular interferences with Italian (V. It.) and Italian interferences with vernacular (It. V.) were also counted by considering the whole text as the appropriate unit of analysis. A given threshold was assumed for a text to be qualified as affected by interference.

176

Maria Antonietta Pinto

Experimental hypotheses A first cluster of hypotheses concerns the relationships between the semantic categories, on one hand, and: l a ) code variations: Italian-vernacular; b) content variations: local-general culture topics; l c ) socioeconomic status variations: High Socioeconomic level (HSE) and Low Socioeconomic level (LSE) subjects. The hypotheses were that: 1 a) Each S and O semantic area was tendentially linked to one code in particular: vernacular to S categories and Italian to O categories ; therefore, S (V) totals were compared to S (It.) totals and O (It.) to O (V) totals. 1 b) Content variations were also supposed to play a role in selecting S and O categories: general topics (G) were expected to select greater frequencies of the O and of the I categories, whereas local culture topics (L) were expected to induce S categories options. 1 c) Semantic options were also supposed to be a function of socioeconomic differences: HSE subjects were expected to orient themselves towards the O or the I area rather than towards the S area, whereas the LSE subjects trend was expected to be exactly the opposite. As for non-semantic variables, Information Richness was supposed to vary in relation to socioeconomic status, while Syntactic Complexity was supposed to vary in relation to both code variations (Italian versions being expected to give rise to higher syntactic complexity than vernacular versions) and of socioeconomic status variations. Interference-loden texts from Italian into vernacular (It. V) were expected to be more numerous than those in the opposite direction (V.It.). The hypotheses were also that It.V interferences were due to general culture rather local culture topics, and with HSE rather LSE sujects. The basic units for data processing related to the 1 a, 1 b, and 1 c hypotheses were weighted means (xw), namely means of the overall frequencies of S or O or I categories divided by the Information Richness of the corresponding texts. For the remaining hypotheses, such weighting was not deemed necessary and numerical values are just raw means. To both kinds of means the T Student two-tailed test was applied at the two confidence levels of p < .01 and p < .05.

Italian and Vernacular Writing in Southern Regions of Italy

177

Table 1: Results Concerning Hypotheses 1 a), 1 b), 1 c). Calabria

Apulia

Sicily

Hyp. 1 a): Categories (S, 0) and Codes (It., V.) DF=29 xw S (It.) = .56 xwS(V) = .77 t = - 5 . 4 1 +Sign. Ol

xw S (It.) = .73 xwS(V) = .77 t = — 1.41 + NS

xwS(V) = .90 xw S (It.) = .84 t = 12.40 +Sign. Ol

xw 0 (It.) = .48 xw O (V)=.39 t = 4.17 +Sign. 01

xw 0 (It.) = .46 xw O (V) = .36 t=5.15 +Sign. Ol

xw O (It.) = .45 xw 0 (V) = .24 t=5.88 +Sign. Ol

Hyp. 1 b): Categories (S, O, I) and Contents (G, L) DF=29 x w S(L) = .97 x w S(G) = .36 t=9.60 + Sign. 01

x w S(L) = 1.18 x w S(G) = .32 t = 19.36 +Sign. Ol

x w S(L) = 1.50 xw S (G) = 1.19 t = 5.91 +Sign. Ol

xw 0 (L) = .25 x w O(G) = .62 t 10.80 +Sign. Ol

xw O (L) = .18 xw O(G) = .64 t = —13.08 +Sign. Ol

x w O(G) = 1.28 XW0(L) = 1.02 t=6.93 +Sign. Ol

xw I (L) = .000 xw 1(G) = .037 t = 13.76 +Sign. Ol

xw I (L) = .00 xw 1(G) = .62 t = -10.99 +Sign. Ol

xw I (L) = .002 xw 1(G) = .037 t=-10.45+ Sign. Ol

Hyp. 1 c): Categories (S, O, I) and: Social classes (HSE, LSE) DF=28 x w S (LSE) = 1.12 x w S (HSE) = 1.54 t=-2.38-Sign. 05

x w S (LSE) = 1.58 x w S (HSE) = 1.41 t = 2.01 + NS

x w S (LSE) = 1.27 xw S (HSE)=1.39 t = — .93 —NS

xw O (LSE) = .92 xw 0 (HSE) = .82 t=1.09-NS

xw 0 (LSE) = .71 xw O (HSE) = .94 t 3.42 +Sign. Ol

xw O (LSE) =.61 xw 0 (HSE) = .78 t = —1.50 + NS

xw I (LSE) = .08 xw I (HSE) = .10 t = — 1.60 + NS

xw I (LSE) = .062 xw I (HSE) = .063 t = — .07 + NS

xw I (LSE) = .056 xw I (HSE) = .057 t = —.10 + NS

Summary of the results and discussion In order to simplify the reading of the above tables, the signs " + " and " — " have been placed in every cell to mean that the result reported in any single cell fulfils or does not fulfil the corresponding hypothesis,

Maria Antonietta Pinto

178 Table 2: Results Concerning Hypotheses H2, H3, H4. Apulia

Calabria

Sicily

Hyp. 2: Information Richness (IR) and Social Classes (HSE, LSE) DF=28 x IR (LSE) = 334.93 x IR (HSE) = 296.13 t = 1.71 — NS

x IR (LSE) = 378.33 x IR (HSE)=454.13 t = — 1.28 + NS

Hyp. 3 Syntactic Complexity (SC*) and

x IR (LSE) = 344.13 x IR (HSE) = 287.86 t=1.43 —NS Codes (It., V) D F = 2 9 Social^Classes (HSE, LSE)

x SC (It.) = 12.6 x SC (V) = 16.6 t=6.94 +Sign. Ol

x SC (It.) = 15.06 x SC (V) = 18.22 t = — 1.59 + NS

x SC (It.) = 11.64 x SC (V) = 12.48 t = — .82 + NS

x SC (LSE) = 25.17 x SC (HSE) = 33.44 t 2.94-Sign. 0 5

x SC (LSE)=44.84 x SC (HSE) = 21.72 t=5.32 +Sign. Ol

x SC (LSE) = 25.58 x SC (HSE) = 22.68 t = 1.07 —NS

Hyp. 4 Interferences (INT) and

, Direction (It.V) or (V,It.) D F = 2 9 | Contents (G, L) DF=29 I Social Classes (HSE, LSE) D F = 2 8

x INT (It.V) = .23 x INT (V.It.) = .30 t = —.32 —NS

x INT (It.V) = .93 x INT (V.It.) = .00 t = 2.25 +Sign. Ol

x INT (It.V) = 2.03 x INT (V.It.) = .00 t=3.50 +Sign. Ol

x INT (It.V)L= .23 x INT (It.V)G= .00 t = 2.54-Sign. 0 5

x INT (It.V)L=.40 x INT (It.V)G = .53 t = — 1.00+NS

x INT (It.V)L= 1.13 x INT (It.V)G = .90 t=1.42 —NS

x INT (It.V)LSE = .00 x INT (It.V)HSE = .46 t = - 2 . 8 2 +Sign. 0 5

x INT (It.V)LSE = .00 x INT (It.V)HSE = 4.06 x INT (It.V)HSE = 1.86 x INT (It.V)LSE = .00 t = - 2 . 4 3 +Sign. 0 5 t=4.53 +Sign. Ol

* As S.C. is counted by dividing the total coordinate clauses by the total subordinate clauses, the lower the value, the higher the S.C.

aside from its statistical significance. Now, at first glance, it appears that the hypotheses which gave rise to almost uniformly significant results are the first two, one concerned with semantic categories and code variations on one hand, and the other with semantic categories and content variation on the other (1 a and 1 b). Although " 4-" signs are scattered throughout many other cells, the results related to hypotheses 1 a and 1 b offer strikingly different pictures from the other ones, and moreover their general patterns almost overlap. This would

Italian and Vernacular Writing in Southern Regions of Italy

179

mean that what causes significant changes in selecting empathic, concrete, and particularistic modalities or non-personal, generalizing accounts of the world, is the very fact of expressing it in vernacular or in Italian, and also the fact that the subject must deal with general or local themes. Compared to these two major sources of semantic variations, socioeconomic differences appear much less significant. This might seem strange, given the partial overlapping between the Bernsteinian concepts of "elaborated" and "restricted" codes and some of the present research categories. Code-dependency seems also illustrated by the results concerning the S. C. and I. R. hypotheses, although in a less marked way. However, it must be immediately added that Bernstein's studies never aimed at focussing verbal abilities in disadvantaged children competent in two structurally and functionally different codes. Moreover, appropriate reading of the results to hypotheses 1 a and 1 b must be completed by reading the results concerning. It.V interferences. Here, first of all, it is clear that italianization of vernacular is stronger than the opposite process, as it had been supposed, and that the LSE subjects do not produce at all interference-loden It.V texts. In other words, they do know their vernacular better than their HSE pears. Thus their knowledge of the two codes appears to be more distinguished and better defined. As a matter of fact, what has been asked is a vernacular and an Italian version but the real output is what children deem to know about the two codes, whether this self-evaluation is correct or not. Therefore, in terms of real knowledge of the two codes, there is a marked difference between HSE and LSE children. But than it is even more striking that when subjects, in each of the socioeconomic groups think of writing in vernacular, whether this is true or not, they fall into S modalities, whereas when they think of writing in Italian, they fall into O or I modalities. Such an inner semantic switching appears quite interesting with regard to pedagogical implications because it points to a potential flexibility which has been so often observed in accomplished bilinguals. On the other hand, the school seems to ignore such a potentiality because it does not admit of the fact that a pre-existing vernacular competence has linguistic relevance. From this point of view, there is another remarkable phenomenon related to phonological-graphemic correspondences in written vernacular. Many impressive instances could be brought of the graphemic creativity shown by nearly all the 90 subjects in the research throughout the three different vernacular-speaking regions. Since it is absolutely impossible that these children have been taught to write in vernacular, and external opportunities to see vernacular writings are prac-

180

Maria Antonietta Pinto

tically none, the only explanation is that they re-worked sound-writing correspondences learnt at school within and for the official code, which is Italian. Here again, in this curious blending of submission and independence, of respect of external rules and of personal re-creation of them, seems to lie the core of mixed identity. Thus local, parochial identity and national and potentially international identity, constantly alternate not only semantically but also graphemically. This would point to deep cultural identification. Only pedagogical awareness of such implications in terms of identities will be able to transform a hybrid into metalinguistic and meta-cultural unity.

Notes 1 2 3 4 5

Freddi 1983. Mioni 1975. Bernstein 1971,1973,1975. Adlam 1977, Cook-Gumperz 1973. The whole statistical data processing has been carried out by Miss Marta Lepore (Facoltà di Scienze Statistiche, Demografiche ed Attuariali-Università di Roma-"La Sapienza"), whereas the data collection has been carried out by Miss M.G.Gambatesa, Miss L. Pallone and Miss M.Polara, who are gratefully aknowledged. 6 Titone 1971,1981, Cummins 1978, Ben Zeev 1977, Duncan, De Avila 1980, Hakes Evans, Tunmer 1980, Tunmer, Pratt, Herriman 1984.

References Adlam D.S., 1977, Code in Context, London, Routledge and Kegan Paul. Ben Zeev S., 1977, Mechanisms by which childhood bilingualism affects understanding of language and cognitive structures. In P.A.Hornby (Ed.), Bilingualism: Psychological, Social, and Educational Implications, New York, Academic Press. Bernstein B., 1971,1973,1975, Class, Codes and Control, Voll.l, 2,3, London, Routledge and Kegan Paul. Cook-Gumperz J., 1973, Social Control and Socialization, London, Routledge and Kegan Paul. Cummins J., 1978, Bilingualism and the development of metalinguistic awareness, Journal of Cross-Cultural Psychology, 9,131-149. De Avila E. A., Duncan S. E., 1980, Field dependence-independence of traditional and dualistic Chicano communities and Anglo-communities, in Domínguez D. (Ed.), Cross-Cultural Investigations of Cognitive Style, Austin, Tx., South west Educational Development Laboratory.

Italian and Vernacular Writing in Southern Regions of Italy

181

Dittmar N., 1973, Soziolinguistik, Frankfurt, Athenäum Verlag GmbH. Ferguson C. A., 1959, Diglossia, Word, 15,325-340. Fishman J. A., 1972, The Sociology of Language. An Interdisciplinary Social Science Approach to Language in Society, Rowley, Mass., Newbury House. Fishman J. A., 1980, Bilingualism and biculturalism as individual and societal phenomena, Journal of Multilingual and Multicultural Development, 1:3-16. Freddi G, 1983 (a c. di) L'Italia Plurilingue, Bergamo, Minerva Italica. Mioni A. M., 1975, Per una sociolinguistica italiana. Note di un non sociologo in Fishman J. A., 1975, La Sociologia del Linguaggio, Roma, Officina. Pinto M. A., 1980, Svantaggio linguistico, Ambiente Sociale, Educazione. Critica a B. Bernstein. Roma, Bulzoni. Titone R., 1972, Bilinguismo Precoce e Educazione Bilingue, Roma, Armando. Titone R., 1981, Educazione bilingue precoce e sviluppo cognitivo, in Titone R. 1981 (a c. di) Avamposti della Psicolinguistica Applicata, Voi. 2, Roma, Armando. Tunmer W. E., Pratt C., Herriman M. L., 1984, Metalinguistic Awareness in Children, Berlin, Springer-Verlag.

Some Social Traits of the Bilingual Child's Personality Renzo Titone State University of Rome Rome

There is at present sufficient evidence in favor of the claim that, given equal conditions, the intellectual development of the bilingual child is not only fully normal but even superior to that of the monolingual child. Studies supporting this conclusion are numerous and increasingly convincing (see Titone 1972,1979). But one of the principal aims of our research (Chair of Educational Psycholinguistics, University of Rome) consisted in examining the effects of early bilingualism on the growth of the child's personality. At present the data gathered is not sufficient for the forming of hypotheses on this issue. Diebold (1966) suggested, from a dynamic and clinical viewpoint, that the bilingual person possesses more adequate defences against schizophrenia and frustration. If psychological difficulties arise, the reason is that the person is faced by a crisis in social and/ or personal identity; such a crisis could be solved by "code switching" both linguistically and culturally. Our research just started to face the problem of mental normalcy in the bilingual child. But it will be a long time before we can decide on the significance of certain results. Meanwhile, we are trying to measure a few personality traits of our young subjects. First, we have been applying some of the tests of developmental potential (Haussermann 1958). Secondly, we have been conducting observations on some personality dimensions by means of an evaluation scale of the child's personality, including relations with parents, peers, other-perception, and socialization modalities. What is appearing at this point is that our young subjects manifest normal, even superior cognitive growth, and utterly normal development of their self-concept. The social environment of our children is

184

Renzo Tifone

uniform (middle class). Other studies or further phases of our own research will take account of subjects living in different social and cultural contexts, including diglossic settings. One important phase of our investigation has been concerned with the examination of some social dimensions of the personality of elementary school children. We restrict our discussion here to some aspects of the bilingual child's social growth.

Child bilingualism and social growth Is there a relationship between bilingualism and socialization? Present research dealing with this relationship has yielded contrasting results: while Genesee and Hamayan's data (1980) indicates that there is none, Wong-Fillmore's investigation (1976) has found a positive correlation between second language competence and socialization. Our study examined a sample of 182 children attending three private schools and one state school in Rome at the primary level. An experimental sample was extracted out of this group consisting of 27 children who were Italian-English, Italian-French, or Italian-German bilinguals aged 5:6 through 8:6, and attending three bilingual schools. The main objective was to determine whether there was a significant difference between the bilingual experimental group and the monolingual control group and whether such a difference depended on bilingual competence. The problem, however, appeared to be twofaceted: Is bilingualism a factor favoring socialization? Or is socialization a factor favoring bilingualism? It was also imperative that we avoid a few errors in planning our study. That is, one of the drawbacks of previous investigations lay in the heterogeneous composition of the samples both with regard to age and to socio-economic class. In our project all the subjects were within a specific age range and all came from middle class families. Another pitfall may lie in the failure to determine precisely the level and degrees of bilingual competence by including in the same sample Ss endowed only with receptive competence and Ss with productive competence. Our group was tested and selected on the basis of both receptive and productive competence in both languages. Two tests were administered: the Bilingual Measures Test by Fraser, Bellugi, and Brown (Sentence Comprehension Test), and a sociometric test (Bastin 1963). The first (bilingual competence) test was used

185

Some Social Traits of the Bilingual Child's Personality

both receptively and productively. Out of the original 80 items only 40 were used (i. e. only those adapted to the Ss' age), and out of these 20 were used for testing comprehension and production in L, (Italian) and 20 in L2 (French, English, German). The sociometric test implied not only choice of associational preference but also selective self-perception, considering the individual's self as target and source of interrelations, that is each child's perception of his own personality and position in his group. Social dimensions considered here were: (1) popularity vs. isolation; (2) affective expansion; (3) active refusal to participate in the group; (4) self-perception within the group. The sociometric test was based on one suitable criterion, viz. the "playmate" role.

Discussion of results Following is a summary of the main results: I. Socialization: In the table the first line is filled by the bilingual Ss, the second line by the monolingual Ss, and they are coded as S+ (very Sociable), N (normal), and S — (less sociable or isolated). The totals are the result of pre-test selection. Table I

Bil. Mon.

S+

N

S-

9 54 63

6 71 77

5 40 45

20 165 185

TheX2 Value (1.52) was not significant. Therefore, the null hypothesis was accepted: namely, there is no difference between the bilingual and the monolingual children in terms of contact expansion. Both types of children have equal capacity for affective expansion. II. Refusal to participate in group activities: In the table, traits are indicated as Pa-I- (strong refusal), N (normal or average), and as Pa — (weak refusal).

186

Renzo Titone

Table II

Bil. Mon.

Pa +

N

Pa-

12 52 64

4 61 65

6 46 52

22 159 181

TheX2 Value appears to be non-significant. Therefore, the null hypothesis is to be accepted; namely, there is no difference between the two groups with regard to refusal to participate in group activities. III. Popularity vs. isolation: P refers to Popularity, N to average acceptance, and I to Isolation. Table III

Bil. Mon.

P

N

I

12 39 51

6 26 32

4 71 75

22 158

A significant difference was found between bilinguals and monolinguals. The bilingual group was significantly more popular than the monolingual group. IV. Degree of realistic perception: This concerns the self-choices and rejections imagined by the individual as a member of a given group. Thus an individual may be realistic (R+), moderately realistic (N), or unrealistic (R—). Table IV

Bil. Mon.

R+

N

R-

7 36 43

10 54 65

3 62 65

20 152 172

The X2 Value was not significant and consequently the null hypothesis was accepted; namely, there are no differences between the two groups with regard to degree of realism in social self-perception.

Some Social Traits of the Bilingual Child's Personality

187

Analysis of results The data obtained from the comparison of the bilingual and monolingual groups evidences a number of interesting aspects related to social personality, including the following: (1) Socialization or "affective expansion". The choices made by our two groups of children were not significantly different. This suggests that early bilingual competence is no limitation to the development of social contact from early childhood. Bilinguals are just as normal as monolinguals. (2) Tendency to segregation and refusal to participate in group life. The number of refusals made by our children was not significantly different between the two groups. Thus it seems that choices as well as refusals are determined by momentary motivations, by the goals sought by the individual and by interests and affinities, not by the mastery of two languages versus only one. (3) Popularity vs. isolation. Here a significant difference emerged with regard to the subjects' acceptance or rejection within their groups. The bilingual group revealed greater acceptance on the part of their schoolmates. As a matter of fact, the most popular subjects belonged in the bilingual group, whereas the most isolated ones were members of the monolingual group. Thus bilingual children are preferred by their peers. Why? The reason does not seem to be due to the bilinguists' greater school achievement as they are not all overachievers, although none of them was under average. It seems that the bilingual child is considered "different" from the others and he is given a mark of prestige. (4) Self-perception of the individual as a member of a group (one's imagined position and value in a group of peers). Here there was no significant difference between the two groups. Self-perception may vary in degree of realism, but it varied to the same degree in the two groups.

Conclusion The general conclusion is obvious: early bilingualism does not carry along certain disadvantages as far as personality development is concerned; being able to use two languages from early childhood does not imply difficulty in or a different mode of socializing; bilingual children are socially as normal as monolingual children.

188

Renzo Tìtone

On the other hand the bilingual child's privilege of being generally preferred, esteemed, and considered as a potential leader by his monolingual peers (as indicated by the sociometric test), can in the long run influence his growth and achievement in a positive manner (Falcone 1981).

References Bastin, G. 1963 Le tecniche sociometriche, Torino: SEI. Falcone, M. 1981 Bilinguismo precoce e socializzazione, Unpublished dissertation, Dept. of Psychology, University of Rome. Genesee, F. & Hamayan, E. 1980 Individual differences in second language learning, Applied Psycholinguistics, Vol.1, no.l, 95-100. Haeussermann, E. 1958 Developmental potential of preschool children, New York: Grune & Stratton. Titone, R. 1972 Bilinguismo precoce e educazione bilingue, Armando: Roma. Titone, R. 1979 Bilinguismo precoce e sviluppo della personalità, Rassegna Italiana di Linguistica Applicata, Vol. XI, nos. 2-3,1-20. Wong-Fillmore, L. 1976 The second time around: cognitive and social strategies in second language acquisition, Ph. D. dissertation, University of California at Berkeley.

The Hal Construction And The Main Verb in the Sentence Peter Abboud University of Texas Austin, Texas

Despite the fact that the IJal construction is widely used in Arabic, it remains one of the most difficult, controversial, and least understood of its syntactic features. One of the problems that will concern us here is the fact that Arabic grammars have for the most part discussed the Hal in terms of its own internal structure, and mainly, with the exception of a couple of instances to be indicated in the body of the text, without reference to the semantics of the main verb. Such a treatment is deficient in that, among other things, it fails to account for and elucidate many of the properties of the Hal. The purpose of this paper is to demonstrate the importance of studying the Hal in relation to the main verb in the sentence and to show that the semantic classes and features of the main verb are crucial to an understanding of the Hal construction. More specifically, the study will show that these features of the main verb must be taken into consideration in determining the occurrence of a Hal, the type of Hal used, and the interpretation of a Hal and of whether a construction is a Hal or not, and that a consideration of these features will help explain some well-known peculiar usages of Hal. To this end, four arguments are offered in what follows. I. The first argument maintains that there are constraints, semantic and others, on the nature of the H^l, which in turn are determined by the semantic nature and syntactic properties of the main verb. (1) Definitions of the Hal refer to the fact that it occurs fi waqti dalika al-ffli 'simultaneously with that verb' (Ibn Ya°is, II, 55), or that ya-

taqayyadu biwaqti husuli madmunihi [al-fFl] 'it is bound by the time

during which its content (the verb's) occurs' (al-'Astarabadhi, 1,199).

Peter Abboud

192

(In these references, al-fiFl, the verb, refers to the camil of the Hal, i.e., the main verb.) This constraint implies that the event of the main verb must itself be of some duration over a span of time, and hence that verbs that do not have a 'duration' feature or content cannot have Hal complements. On investigation this is found to be indeed the case. Thus, the verb yanduru 'it is rare' and other stative verbs like it, one of whose semantic features is that they do not have a 'duration' aspect, cannot take a Hal, while a verb like intadara 'wait', an activity which assumes duration over time, can. (2) Some of these non-duration verbs are impersonal verbs whose subjects is a noun clause, masdar mu'awwal, or a nominalization of it, i. e., masdar sarih. It so happens that a Hal cannot have a clause for its antecedent or referent, sahib. It follows that a verb that must have a clause for a subject or object cannot have a Hal whose sahib is that subject or that object clause. It must be noted that the verb in the embedded clause may itself take a Ual. Thus, yajuzu 'it may be', or yanduru 'it is rare', does not take a Hal itself, but the verb in its subject noun clause (or its masdar in case it is sarih) may. In the sentence: yajdura bind wanahnu fi hada al-maqami 'an naqiila ... It behooves us, while we are on this subject, to s a y . . . the Hal is governed by the verb naqiila of the subject noun clause, and not by yajduru. (3) The following sentences are universally rejected as aberrant: 'ansada al-sfra mayyitan jalasa waqifan masa jalisan

Being dead, he recited poetry. He sat, standing. He walked sitting.

There is nothing inherently wrong with the Hals in the above sentences, and one can generate many sentences where they are appropriately used. What makes the sentences aberrant is that they refer to states that are not compatible with the semantics of the main verb: you cannot be dead and recite poetry, or sit and stand at one and the same time, etc. In the words of Slbawayhi, they are muhal 'impossible, incongruent'. (4) In the controversy over the occurrence of a masdar as Hal, some grammarians allow only masdars that have been sanctioned by common usage,1 samaF, while others allow for free usage by analogy, qiyas, either unrestrictedly2 or under the constraint that the masdar be navfan Ifamilihi (see c Abbas Hasan, II, 372, note 3) 'a subcategory of

The fial Construction And The Main Verb in the Sentence

193

its (the Hal's) regent'. If this last position is correct, then it would constitute another example of the dependence of the Hal on the features of the main verb, in which case of course one will have to spell out what is meant by a subcategory or type of the verb. II. The second argument stems directly from the semantic classification of the main verb. First, consider the following example: ra'aytihi maridan This sentence is subject to two interpretations: maridan 'sick' may be a Hal or a second object, maf°ulun damn, depending on whether ra'ayti is taken to be a j¥lu hass 'verb of sense', in which case the sentence translates: 'You (f.s.) saw him while he was sick', i.e., as a Hal, or fFlu qalb 'a verb of heart', in which case the sentence reads: 'You (f. s.) saw that he was sick' (Wright, II, 50). The clear implication is that the semantic class of the main verb crucially determines whether a construction is a Hal or not. Second, consider examples of what I will call inceptive-resultive verbs,3 such as rakiba 'he mounted, rode', labisa 'he put on, wore' which are verbs carrying the meaning of both the event inception, usually one of shorter duration, and the resulting event, of much longer duration. The sentence ja'a Zaydun yarkabu farasan has one interpretation: 'Zayd came riding a horse', but ra'aytu Zaydan yarkabu farasan is subject to two interpretation: 'I saw Zayd mounting a horse' and 'I saw Zayd riding a horse'. The reason for this is that either the event inception aspect or the resulting event can occur simultaneously with 'seeing', but with 'coming', which is normally an activity of longer duration, the former, meaning 'mounting', which is of short duration, though possible, is highly unlikely (one does not travel, mounting!). It is worth noting that ra'aytu Zaydan rakiban farasan has only one interpretation: 'I saw Zayd riding a horse', because the Active Participle for the verb rakiba, as for many other verbs, contrary to the commonly-held view of the meaning of the Active Participle (see next section), is really perfective, not imperfective, in aspect. Thus it is equivalent to ra'aytu Zaydan waqad rakiba farasan 'I saw Zayd having mounted/ridden a horse'. III. The third argument has to do with the role of certain classes of verbs in the occurrence or non-occurrence of waw al-Hal 'the circumstantial waw'. The discussion in Arabic grammars regarding the optional or obligatory use of waw is given totally in terms of the structure of the Hal itself and without any reference to the role played by the main verb in the process. Thus, for example, a waw is obligatory, ac-

194

Peter Abboud

cording to most grammarians, with a perfect verb in the Hal, but it is dropped with an imperfect verb, la hajata 'ila al-waw lima bayna al-JFli al-mudarfi wa-smi al-fafili min al-munasabah 'there is no need for the waw 'because of the similarities between the Imperfect and the Active Participle' (Ibn Ya°Is, II, 66). But does the main verb have a role? Consider the following sentences: ja 'a yadhaku tahaddada 'ilayna yadhaku 'istaqbalnahu yadhaku

He came laughing. He talked to us, laughing. We welcomed him, laughing.

No speaker of Arabic today has any problem with the first sentence, but most will be uneasy, to say the least, with the second and third. (Whether these last two sentences were acceptable in Classical Fu$h& is another matter, of course.) If we are to explain the marginality of these two sentences, then we cannot escape the conclusion that the optionality of the waw with an imperfect is determined crucially by the main verb. It thus seems that with at least one semantic class of verbs, verbs of motion like ja'a 'he came', dahaba 'he went' etc., is the omission of the waw permissible. With some classes of verbs it is not. It is to be noted that either type of Hal dahikan or wahwa yadhaku is acceptable in all the sentences above. I call these Hals marked, the markers being the accusative -an in the former and the waw in the latter. As such, the rule seems to be that, with the semantic class 'verbs of motion' and perhaps others, Hals may be marked or unmarked. With other verbs, the Hal must be marked. IV. The fourth argument upholds the contention that an analysis that takes into consideration the semantic classes of verbs will help explain a perplexing usage of the Hal. As is well known, the Hal in most cases is muqarin, 'simultaneous with the event of the main verb'. Now consider the following examples: ja'a 'ilayhiycfuduhu 'ata 'ila cayni ma'in yasrabu 'arsala yiflimuhu kataba yaqulu

He came to visit him (lit. visiting) He came to a well to drink. He sent to inform him. He wrote to say (or saying).

It can be easily seen in these well-known examples that the Hal does not always occur simultaneously with the event of the main verb. Arabic grammars have proposed a second type of Hal, known as Hal muqaddar 'a Hal which indicates future state',4 to account for these examples. This typology, however, leaves some serious questions unanswer-

The Hal Construction And The Main Verb in the Sentence

195

ed. Is this Hal muqaddar possible with all verbs? If not, then which verbs do and which do not take a Hal muqaddar? The foregoing examples and many others suggest that these Hals modify a certain class of verbs whose semantic content refers not only to the event itself but also to the eventual completion of the event,5 which I will call event-completion verbs. Applying the notion of eventcompletion class of verbs obviates the need for a Hal muqaddar. Seen in this framework, the Hal, then, renders the meaning of concomitance and simultaneousness with respect to one or more aspect of the main verb, which are determined by the semantic features of that verb. Thus, (1) With event-completion verbs, a Hal may refer not only to the completion aspect, but also to the event aspect or to both. The semantics of the verb in the Hal construction will determine which is which. Thus, in ja'aycfuduhu 'he came to visit him', the 'visiting' cannot occur simultaneously with the event of 'coming', but rather with the completion aspect of it, 'the getting there'. In ja 'a yatahada 'he came, swinging rhythmically', the 'swinging' does not usually occur at the completion of, but during, the 'coming'; hence the Hal is simultaneous with the event. But in ja'a yabki, the 'weeping' can occur with both aspects of the verb; hence it is capable of two interpretations, though the listener is likely to select the one with the event aspect in mind, i. e., 'he came weeping', rather than the one with the completion aspect, 'he came to weep', as the more probable. (2) It is important to note that with these verbs the Hal with wawis equivalent to the Hal of the event aspect and not of the completion aspect. Thus, yd'a wahwa yatahada has the same meaning as ja'a yatahada 'he came, swinging' and ja'a wahwa yabki has one interpretation, namely simultaneous with the event, 'he came weeping', not the other. The sentence ja'a ya°uduhu has no equivalent with waw: *ja'a wahwa ya cuduhu is not a well-formed sentence, semantically. (3) Finally, it is possible to have in the same sentence a different Hal for each aspect of the verb. Thus, yd'a musrfan 'ila °ayni ma'in yasrab 'he came in a hurry to a well to drink'. This provides strong supportive evidence for the analysis presented in this paper. To summarize, in order to understand the various subtle meanings of the Hal, it has been shown that it is imperative to take into consideration the semantics of the main verb which it modifies. This approach has proven to be productive not only in explaining and interpreting jjals, but also in showing the way of resolving some of the issues related to the Hal construction.

196

Peter Abboud

Footnotes This is the position of Slbawayhi. This is now the official position of the Arabic Language Academy in Cairo. For this class, which he calls 'inceptive', see McCarus. The English sentences 'he came calling' or 'he came acourting' come close to conveying the meaning of this construction. English usage dictates that the above sentences be translated with a to-construction, but the Arabic sentences do not have the meaning of'in order to'. To illustrate the difference, it has been said of Jesus that 'he came eating and drinking', not to eat and drink! 5 Wright, II, 19 suggests: "The imperfect forms a secondary subordinate clause, expressing the state . . . in which the subject of the previous perfect found himself when he completed the act expressed by that perfect."

1 2 3 4

References c

Abbas tfasan. al-Nahw al-Wäfi, 4 vols., Dar al-Ma°ärif biMisr, 4th ed., 1971. al-'Astaräbädhl. Sarh Käfiyat Ibn al-Häjib. Beirut, Där al-Kutub al-°llmlya, 1975. 'Ibn Ya°Is. Sarh al-Mufassal. Cairo: al-Tibäca al-Munlnya, n.d. McCarus, Ernest N. A Semantic Analysis of Arabic Verbs. In Michigan Oriental Studies in Michigan Oriental Studies in honor of George G. Cameron. Department of Near Eastern Studies, The University of Michigan, Ann Arbor, Michigan, 1976. Slbawayhi. al-Kitäb. Ed. Hartwig Derenbourg. Paris 1881 and 1889. Wright, W. A grammar of the Arabic Language? Cambridge, 1967.

The Possessive Predicate in North African Vernacular Arabic Bernard Comrie University of Southern California Los Angeles, California

1. The possessive predicate construction In most varieties of North African vernacular Arabic - as in many other varieties of Arabic - the usual translation equivalent of the English verb have is as in Tunis Arabic sentence (1) below:1 (1) 9and-ii ktaab. at-lS book 'I have a book.' In order to maintain a uniform, though informal, terminology in discussing examples from different varieties of North African Arabic, and to facilitate comparison with English where relevant, I will introduce the following terms: the possessor refers to the individual who is said to own some entity, i. e. in (1) the possessor is expressed in English by the subject pronoun I and in Tunis Arabic by the first person singular object suffix on the preposition 9and; possession refers to the overt expression of the relation of possession, i. e. in English to the verb have and in Tunis Arabic to the preposition 9and; possessum refers to the entity possessed, i. e. in example (1) to the book; these terms apply irrespective of the syntactic or morphological encoding of these relations in an individual language. The apparent syntactic construction illustrated in sentence (1) is as follows: the possessor is expressed as a prepositional object; possession is expressed by means of the preposition 9and, and the possessum is expressed by means of a bare noun phrase. This apparent structure is thus identical to another construction that I will refer to as the loca-

198

Bernard Comrie

tive predicate, whereby an object is located in some location, as in Tunis Arabic sentence (2): (2) li-ktaab 9and-ii. the-book at-lS "The book is at my place.' The preposition 9and can also be used to mean 'at (the place of)'. Although there is generally a difference in word order between the locative and possessive predicate constructions, this is far from absolute, and depends on the topic-comment structure of the sentence rather than on the possessive/locative opposition - many sentences are therefore ambiguous between possessive and locative interpretations. These parallel possessive and locative constructions are found in the dialect of Tunis and in Maltese, both of which I have investigated in some detail with a range of native speakers; they are also found in the dialects of Kairouan (Tunisia), Reghaia (near Algiers), and Tripoli (Libya), which I have investigated in much less detail, with only a single native speaker for each location. I imagine that similar systems would be found in many other locations in North Africa and more generally throughout the Arabic-speaking world, but the investigation of these other varieties of Arabic from this viewpoint remains a task for future research. Despite the apparent similarity between the possessive and locative predicate constructions, I will argue in this contribution that their actual syntactic structures are radically different. The syntax of the locative predicate construction is essentially as suggested by its morphology: the entity located is subject of the clause, the expression of location (9and) is a preposition, and the place where the entity is located is object of this preposition. The syntax of the possessive predicate construction is much more complex, but is never identical to that of the locative construction. In what follows, I will apply various syntactic tests that establish differences between the two constructions and serve to indicate the actual syntactic structure of the possessive predicate construction. To save on undue repetition, I will cite examples from the Tunis dialect alone where a given point applies to all dialects (with only appropriate minor phonological and lexical substitutions - also orthographic substitutions, in the case of Maltese, the only variety with an established orthography).

The Possessive Predicate in North African Vernacular Arabic

199

2. Full noun phrase possessors In general, where a preposition occurs with a full noun phrase rather than with an unstressed pronoun, vernacular Arabic simply has the preposition followed directly by that full noun phrase, i. e. object suffixes on prepositions and full noun phrases are in complementary distribution. This applies in particular to the locative predicate construction, as in example (3): (3) li-ktaab 9and 91ii. the-book at Ali "The book is at Ali's.' This substitution is not, however, possible in the possessive predicate construction. If one wishes to say 'Ali has a book', then it is necessary to retain the object suffix on the preposition 9and and to express the possessor as a bare noun phrase, usually initially in the construction, as in example (4): (4) 91ii 9and-uu ktaab. Ali at-3SM book 'Ali has a book.' In this respect, then, in the possessive predicate construction the complex 'possession -I- possessor' does not behave like a regular prepositional phrase, suggesting that in some sense either the expression of possession is not a preposition or the expression of the possessor is not a prepositional object (or both). There is thus potentially a discrepancy between morphological categories and syntactic categories in this construction.

3. Negation In order to understand the operation of negation with the possessive predicate constructions, it will first be necessary to investigate the operation of negation in other constructions whose structure is less controversial. If a clause has a verbal predicate in these varieties of Arabic, then negation is expressed by means of the combination of the proposed particle maa and the suffix -s, as can be seen by comparing (5) and (6) below:

200

Bernard Comrie

(5) 91ii DRab il-kalb. Ali hit-3SM the-dog 'Ali hit the dog.' (6) 91ii maa DRab-s il-kalb. Ali not hit-3SM-NEG the-dog 'Ali didn't hit the dog.' With a nonverbal predicate, however, as in the locative predicate construction (in the present tense), negation is expressed by inserting a pronoun of the person, number, and gender of the subject, and then adding preposed maa and the suffix -s to this pronoun, as can be seen by comparing (2) above with (7): (7) li-ktaab maa huu-s 9and-ii. the-book not he-NEG at-lS 'The book is not at my place.' (The noun ktaab 'book' is masculine, whence the third person singular masculine pronoun in (7).) With the possessive predicate construction, in all the varieties of vernacular Arabic considered, the negation used is verbal negation, as in (6), and not the negation appropriate for a nonverbal predicate, as in (7). Thus the negative of (1) and (4) are, respectively, (8) and (9): (8) maa 9and-ii-s ktaab. not at-lS-NEG book 'I don't have a book.' (9) 91ii maa 9and-uu-s ktaab. Ali not at-3SM-NEG book 'Ali doesn't have a book.' This suggests that, despite the apparent morphological structure of the expression of possession as a preposition, syntactically it functions as a verb, i.e. in the possessive construction (1) above 9and-iiis syntactically a verb (with one argument, expressed by means of a pronominal suffix), whereas in the locative construction (2) it is a prepositional phrase, in agreement with the morphology.

4. Verb agreement According to the tests discussed above, all varieties of Arabic considered here behave alike. We now turn to tests where there are important differences among these varieties. In both the locative and the posses-

The Possessive Predicate in North African Vernacular Arabic

201

sive predicate constructions, in the present tense there is normally no overt morphological verb. In other tenses, however, the tense is carried by the appropriate form of the verb kaan 'be'. The verb kaan, like verbs in Arabic in general, agrees in person, number, and gender with its subject. One way of testing for the grammatical relations in sentences is therefore to examine subject verb agreement: if a verb agrees with some constituent by means of the appropriate subject agreement affixes, this is prima facie evidence that this constituent is its subject. With the locative predicate construction, the evidence of verb agreement indicates unequivocally that the entity located, and only this constituent, is subject of the construction : the verb must agree with this constituent, and cannot agree with any other constituent. Thus, given that the noun dzaazaa 'chicken' is feminine, and that the default form of a verb (i. e. the form used if the verb does agree with anything) is third person singular masculine, we find that in (10) the verb must be in the third person singular feminine, agreeing with dzaazaa, and cannot be third person singular masculine (default agreement) or first person singular (agreement with the place at which the chicken is located): (10) id-dzaazaa kaanit/*kaan/*kun-t 9and-ii. the-chicken was-3SF/-3SM/-lS at-lS 'The chicken was at my place.' It will be useful also to cite the Maltese equivalent of (10), to show that the Maltese locative predicate construction behaves like that of the other varieties of North African vernacular Arabic in this respect, even though we will see below that Maltese is otherwise somewhat idiosyncratic in its possessive predicate construction: (11) It-tigiega kien-et/*kien/*kon-t ghand-i. Let us now turn to verb agreement with the possessive predicate.

4.1 Default agreement In the possessive predicate construction, one option that is available in all the varieties of Arabic under investigation is default agreement, i. e. a third person singular masculine verb irrespective of the person, number, and gender of either possessor or possessum; in other words, a construction is possible suggesting that neither possessor nor possessum (nor anything else) is subject of the sentence. This is illustrated for Tunis Arabic by (12) and (13):

202

Bernard Comrie

(12) kaan 9and-ii dzaazaa. was-3SM at-lS chicken 'I had a chicken.' (13) kaan 9and-ii wlaad. was-3SM at-lS children 'I had children.' Most of the varieties of Arabic considered here are exactly parallel to the Tunis dialect in this respect. The situation in Maltese is somewhat more complex, although ultimately it is possible to demonstrate clearly that default agreement is also possible in Maltese. For the possessive predicate in the past and future, Maltese does not use the preposition ghand, but rather a form that derives etymologically from the third person singular of the past or future of 'be' plus the dative preposition 1, followed in turn by the appropriate prepositional object pronoun, as in (14)-(15), where in (14) kell-i derives etymologically from kien-l-i, i.e. 'was-3SM-to-lS', and in (15) jkoll-i derives etymologically from j-kun-l-i, i.e. '3SM-be-to-lS: (14) Kell-i tigiega. had-lS chicken 'I had a chicken.' (15) Se jkoll-i tigiega. FUT have-lS chicken 'I will have a chicken.' The parts kell- and jkoll- of the possessive predicate construction are indeed invariable, irrespective of the person, number, and gender of possessor or possessum, but one might argue that this is perhaps for independent reasons, for instance the fused nature of these expressions: the identification of a third person singular masculine verb form in kell- and jkoll- is primarily etymological and of dubious synchronic relevance. There is, however, a more complex tense construction that argues unequivocally for default agreement. In order to express the future-in-the-past or conditional, Maltese uses the past tense of 'be' with the future tense of the main predicate, as in (16): (16) Kon-t se n-oqtol it-tigiega. was-lS FUT lS-kill the-chicken 'I was going to kill the chicken.' As (16) illustrates, the past tense form of 'be' must agree with the subject. In the equivalent form of the possessive predicate, however, we

The Possessive Predicate in North African Vernacular Arabic

203

find the invariable third person masculine verb form kien, i.e. default agreement: (17) Kien se jkoll-i tigiega. was-3SM FUThave-lS chicken 'I was going to have a chicken.' Indeed, as we will see below, in Maltese default agreement is the only possibility in the possessive predicate construction; the other varieties of Arabic investigated all allow at least some other possibilities. 4.2 Possessum agreement In all varieties of Arabic investigated, with the exception of Maltese, it is also possible, under certain circumstances, for the verb 'be' in the possessive predicate construction to agree with the possessum, but the precise details vary from one dialect to another. Since the possessum is almost always third person, and since in the third person in these dialects there is essentially a three-way opposition masculine singular versus feminine singular versus plural (with no clearcut gender distinction in the plural), the question boils down to the following: if the possessum is third person singular feminine, can the form of 'be' stand in the third person singular feminine? if the possessum is third person plural, can the form of 'be' stand in the third person plural? If the possessum is third person singular masculine, then there is no way of telling whether one has possessum agreement or default agreement. In the dialect of Tunis, it is possible to have possessum agreement if the possessum is third person singular feminine, but not if it is third person plural, i.e. one has the pattern as in (18)-(19): (18) kaan-it 9and-ii dzaazaa. was-3SF at-lS chicken 'I had a chicken.' (19) *kaan-uu 9and-ii wlaad. was-3P at-lS children 'I had children.' This difference in judgements between gender and number agreement is, incidentally, absolute and confirmed by a number of native speakers. It is, moreover, impossible to 'split' gender and number agreement, i. e. to have a feminine singular form of 'be' referring to a feminine plural possessum, as in (20); if the possessum is plural, then only default agreement is possible:

204

Bernard Comrie

(20) kaan/*kaan-it/*kaan-uu 9and-ii bnaat. was-3SM/-3SF/-3P at-lS daughters 'I had daughters.' For the dialects of Kairouan, Reghaia, and Tripoli, gender agreement (in the singular) was accepted without hesitation, i. e. sentences parallel to (18) above. For plural agreement (sentences parallel to (19) above), I am less confident of my data, given that I was able to work with only one native speaker of each dialect, but in fact the speaker I worked with from Kairouan accepted sentences like (19), as did the speaker from Tripoli; the speaker from Reghaia did not give me consistent results, thus pointing to the limitations of the elicitation method I employed. It would be particularly interesting to know if the distinction between gender and number agreement, illustrated clearly by the Tunis data, is more widerspread among Arabic dialects. In Maltese, as indicated in section 4.1, only default agreement is possible, i.e. one cannot say (21): (21) * Kien-et se jkoll-i/tkoll-i tigiega. was-3SF FUT be-lS chicken 'I was going to have a chicken.' (The form *tkoll-i'm(21) is simply an attempt to form a feminine singular counterpart to jkoll-i, where the initial,./is, at least etymologically, a third person singular masculine prefix; however, whatever form is chosen, (21) remains ungrammatical.) To the extent that agreement is possible with the possessum, this suggests that the possessum is subject of the possessive predicate construction. We may now turn to the possibility of agreement with the possessor, to test whether perhaps this noun phrase is subject of the construction.

4.3 Possessor agreement Data on possessor agreement are much less clearcut than those on possessum agreement. For the Tunis dialect, investigated in greatest detail from this perspective, possessor agreement seems to be possible, but clearly less preferred than default agreement or possessum agreement (the latter, of course, only where the possessum is feminine singular), but judgements varied from complete acceptability through complete rejection. Incidentally, the person, number, and gender of the posses-

The Possessive Predicate in North African Vernacular Arabic

205

sor seem to have little or no influence on the possibility of possessor agreement, unlike the case with possessum agreement. I therefore assume that sentences like (22)-(24) are grammatical, but note the existence of idiolectal variation and the limitations of the elicitation techniques used: (22) kun-t 9and-ii dzaazaa. was-lS at-lS chicken 'I had a chicken.' (23) faaTmaa kaan-it 9and-haa ktaab. Fatimah was-3SF at-3SF book 'Fatimah had a book.' (24) li-wlaad kaan-uu 9and-hum ktaab. the-children was-3P at-3P book 'The children had a book.' The data for Kairouan, Tripoli, and Reghaia are, of course, less reliable, given that only one native speaker was available for each location (and there is always the possibility of the same kind of idiolectal variation as found for Tunis), but the native speakers with whom I worked unequivocally allowed possessor agreement in the case of Kairouan and Tripoli and just as unequivocally rejected it in the case of Reghaia. In Maltese, as already indicated, possessor agreement is excluded, the only possibility being default agreement: (25) *Kon-t sejkoll-i/nkoll-i ktieb. was-lS FUT have-lS book 'I was going to have a book.' (The form *nkoll-i in (25) is an attempt to create a form parallel to jkoll-i but with a first person singular prefix in place of the (etymological) third person masculine singular prefix of the latter.) To summarize the data on verb agreement, we see that they do not permit any clear conclusion to be drawn on grammatical relations. For Maltese, where only default agreement is possible, they suggest that the possessive predicate construction has no subject. For the other varieties, the construction may apparently either have no subject, or have the possessum as subject (though subject to unexpected idiosyncrasies, at least in the Tunis dialect), or (perhaps as a less preferred variant) have the possessor as subject.

206

Bernard Comrie

5. Possessum as direct object In Maltese, not only does the impossibility of possessum agreement suggest that the possessum is not subject of the construction (in contrast to the locative predicate construction), there is moreover evidence suggesting that the possessum is actually direct object of this construction. In Maltese, most direct objects appear as bare noun phrases, but certain direct object noun phrases either require or allow the preposition ///(which otherwise introduces indirect objects); in particular, personal names as direct objects require lil: compare (26) and (27): (26) Raj-t (*lil) ktieb. saw-lS to book 'I saw a book.' (27) Raj-t *(lil) Pawlu. saw-lS to Paul 'I saw Paul.' Precisely this same distribution of lil is found with the possessum in the possessive predicate construction: (28) Ghand-i(*lil) ktieb. at-lS to book 'Ihave a book.' (29) Ghand-i *(lil) Pawlu. at-lS to Paul 'I have Paul.' For the mainland varieties of vernacular Arabic, I have no comparable data; in general, the native speakers with whom I worked simply felt that such constructions with a human possessum are impossible, so it is conceivable that this particular test is just not applicable in these dialects.

6. Other syntactic tests For Maltese, there are some further syntactic tests which suggest that the possessor - despite its inability to trigger subject verb agreement is nonetheless syntactically a subject; these tests concern primarily constraints on identity of constituents across clause boundaries. The verb seta' 'be able', for instance, requires that its subject be (»referential with the subject of the verb dependent on it.2 Nonetheless, it is pos-

The Possessive Predicate in North African Vernacular Arabic

207

sible for seta' to be followed by a possessive predicate construction where the coreference is with the possessor, as in (30): (30) N-ista' ghand-i kelb. lS-can at-lS dog 'I can have a dog.' I have not been successful in finding parallel examples for the mainland dialects of North African vernacular Arabic: in general, the native speakers with whom I worked simply replaced the possessive predicate construction by a construction using a morphological verb, such as the equivalent of 'get', and sentences like (30) do typically have an ingressive rather than a stative interpretation.

7. Possessor as topic In cases where the possessor seems not to have subject status - for instance, in the mainland dialects where there is no possessor agreement - an alternative analysis that one might attempt is to argue that the possessor is a topic, rather than a subject. In all the varieties of Arabic discussed, topics that are not subjects are in general readily distinguishable in formal terms from subjects. Whereas subjects trigger subject verb agreement, other topics are resumed by an object clitic pronoun, either on a verb or on a preposition, and cannot trigger subject verb agreement, as in the Tunis dialect example (31): (31) id-dzaazaa li-wlaad DaRb-uu-haa. the-chicken the-children hit-3P-3SF 'The chicken, the children hit it.' In (31), it is impossible for the verb to show subject agreement with iddzaazaa, i.e. to appear in the third person singular feminine form DaRb-it. Likewise, one might try to account for the absence of subject verb agreement between possessor and verb in Maltese by arguing that the possessor is a topic, although in Maltese this would face the problem that the possessor does satisfy some syntactic tests for subjecthood, as noted in section 6. If the possessor is analyzed as a topic, then a number of facts do indeed fall into place. In particular, that the possessor normally appears sentence-initially and is resumed by an object pronoun suffix on the preposition 9and would follow automatically from its status as a topic. However, there is also a serious problem with this analysis, namely

208

Bernard Comrie

that it is not necessary for the possessor to be interpreted pragmatically as a topic, i. e. the use of the term 'topic' here would be an otherwise unmotivated extension of the term. The possessor can, for instance, be the focus of a WH-question, in which case it bears the pragmatic role of focus rather than topic, as in (32): (32) skuun9and-uuktaab? who at-3SM book 'Who has a book?' Likewise, it is possible for the possessum to be the position relativized in a relative clause, in which case by definition the possessum is topic within the relative clause, whence the possessor cannot be topic within this clause, as in (33): (33) id-dzaazaa llii li-wlaad 9and-hum the chicken that the-children at-3P 'the chicken that the children have' This suggests that any attempt to analyze the possessor literally as being a topic will be doomed to failure; at best, one could maintain that the possessor is a grammaticalized topic, where this notion would be identical neither to topic nor to subject, but somewhere inbetween the two.

8. Conclusions The data discussed and analyzed in this contribution lead to a number of conclusions. Perhaps the clearest - were it not already clear from Professor Charles Ferguson's work on vernacular Arabic - is the importance of continuing work on the syntax of Arabic dialects, especially as so much of the earlier work has tended to concentrate on phonology, morphology, and lexicon; obviously such work should not depend, as my own contribution has, entirely or perhaps even primarily on elicited material, given the clear limitations that have emerged on the applicability of this method, even with the limited range of data discussed in this contribution. The data presented do not lead to clearcut, simple conclusions concerning the grammatical relations in the possessive predicate construction, especially in so far as these data are often internally contradictory (e.g. for many sentences in some dialects, either the possessor or the possessum or neither may be treated as subject for purposes of verb

The Possessive Predicate in North African Vernacular Arabic

209

agreement). It is, however, clear that the apparent structure of the possessive predicate construction, based on the usual correlations between morphological exponents and syntactic structure, is not an adequate basis for its syntactic characterization. When I embarked upon this project, I hoped to find evidence for a linear development from an earlier stage, where the syntax would be more or less as is implied by the morphology (possessum as subject, possession as a preposition, and possessor as object of the preposition, perhaps topicalized), towards a final stage where the possessor would be subject, possession a verb, and the possessum direct object. Just such a development has been outlined for Hebrew by Ziv (1976). The data at my disposal provide no evidence for such a chronological development, but they do instantiate various systems intermediate, in a logical sense, between the two extremes, especially with verb agreement varying between agreement with the possessum, with the possessor, and with neither. An unexpected, but nonetheless clear, result is the marked division between Maltese and the mainland dialects, even such mainland dialects as those of Kairouan, Tunis, and Tripoli, to which one would expect Maltese to be particularly close on historical grounds. In Maltese, the only kind of verb agreement permitted in the possessive predicate construction is default agreement, whereas all the other varieties permit at least some possessum agreement, many also possessor agreement. Although strictly comparable data perhaps still remain to be uncovered for the mainland dialects, Maltese also seems unique in permitting the possessum to be encoded as a direct object, and in allowing the possessor to satisfy conditions on subject coreference. While the uniqueness of Maltese within North African vernacular Arabic and its offshoots is far from being a novel discovery, in the past this uniqueness has tended to rest purely on phonological and lexical features; more detailed consideration of syntax may well uncover further differences, and perhaps even lead to unraveling of the various influences and crosscurrents that have led to the present-day differentiation of vernacular Arabic in the broadest sense.

Footnotes 1 For data on the Tunis dialect, I am grateful to Hafedh Halila, who also checked the crucial data with several other native speakers; for data on the Kairouan dialect, I am grateful to Khelifa El-Ghali, for data on the Tripoli dialect, to Abdel Monem, and for data on the Reghaia dialect, to Hamoud

210

Bernard Comrie

Salhi. Data on Maltese have been culled from varous sources, but I am particularly indebted to Albert J. Borg and A. Lydia Sciriha. Much of the material on the Tunis dialect was gathered during a Field Methods class at the University of Southern California, and I am grateful to all participants for their comments, especially to Alicia Ocampo. Earlier versions of this paper were presented to an informal seminar at the University of Southern California and to the 16th Conference on African Linguistics, Yale University, March 1985. Portions of the data discussed here have previously been discussed in Comrie (1982) and Comrie (forthcoming) - the latter based on my presentation at the Eastern States Conference on Linguistics, Ohio State University, September 1984 - but the conclusions reached in this contribution supersede those of these earlier works. The following abbreviations are used: IS (first person singular), 3P (third person plural), 3SF (third person singular feminine), 3SM (third person singular masculine), FUT (future), NEG (negative). The notation (*X) means that the sentence is ungrammatical with the inclusion of X, while *(X) means that the sentence is ungrammatical without the inclusion of X. 2 Most grammars of standard written Maltese give the citation form as sata', but my impression is that even among educated speakers of what would otherwise be considered standard Maltese the vocalization seta'is the norm.

References Comrie, B. 1982. 'Syntactic-morphological discrepancies in Maltese sentence structure'. Communication and Cognition 15.281-306. Comrie, B. Forthcoming. 'Agreement as a research tool'. To appear in the proceedings of ES COL 1984. Ziv, Y. 1976. 'On the reanalysis of grammatical terms in Hebrew possessive constructions'. Studies in Modern Hebrew Syntax and Semantics, ed. by P.Cole, 129-152. Amsterdam: North-Holland.

The Verb 'See' in Arabic Dialects Alan S. Kaye California State University, Fullerton Fullerton, California

Comparative Arabic dialectology owes to our y 'see' has cognates in Ethiopic (\/rty), Epiprahpic South Arabian ( y ^ y ~ \/r>), Hebrew \/r>y (= rá^áfh)), Moabite (\/r^y) and possibly Ugaritic ( y ^ y ) • Cf. also Egyptian ?rt 'eye'. Ugaritic uses the verb ph 'to see' with thematic vowel in -a- or -&-. Cf. Gordon (1965: 467). Ugaritic also uses the roots b$r and cyn for 'see'. (y°n) occurs and is presumably to be vocalized yefinu 'he sees' = classical Arabic < ayana (Gordon 1965:455), imperfect yu(ayinu, form III. 3 Historically, there is no question that Maltese is an Arabic dialect (rather closely connected with, say, Tunisian Arabic). However, synchronically, I prefer to regard it as a separate distinct Semitic language rather than a dialect of Arabic. There are two major reasons for doing so: (1) Maltese, if an Arabic dialect, synchronically, would be the only one normally written in Latin script, and (2) it does not have diglossia nor does it share cultural affinity with the "rest" of the so-called Arab world. 4 Cowan (1966:416-17) also notes that in the Spanish Arabic of Granada, as recorded at the end of the fifteenth century by Pedro de Alcalá, the verb

The Verb 'See' in Arabic Dialects

5 6

7

8

9

10

217

'see' is \fr>y. Cowan (1966:417) cites 'I see' as nari, and Corriente (1977: 125) cites rayt 'you see' (after ida 'when') = classical Arabic ra ?ayta with final vowel apocope, glottal stop deletion, and vowel haplology. Spanish Arabic also knows the following forms (Corriente [1977:169]): ra, yard, tiara, yarl, turl, yura, nurd, arayt, ara, arra, ray, war, kit-tara, kin-nara, riya, muraH, murayi, miri, mura, mira >a, amriyya, muray. Cypriot Maronite Arabic is spoken in the city of Kormakiti in Cyprus and has many other unique phonological and morphological features for a modern Arabic dialect. See Kaye (1972) for details. It may now be extinct. Other examples are cited too, e.g., (1984: 52). See Leslau (1969) for the full range of literature available on this topic, especially (1969:181, fn.8). Leslau has pioneered this field of inquiry. For example, Geez (classical Ethiopic) /wald/ (pronounced [wold] in the Amharic pronunciation of Geez) 'child' is not very frequent, but the pronunciation /wund/ is based on an assimilated form of /—»• n in /wundim/ 'brother' (lit., 'the child of the mother') = Arabic walad himm since it is far more frequent. Frequency is, by the way, an area which could profit by a careful reexamination by modern linguistic theories and techniques. The lack of an emphatic r in mra is one piece of evidence why it might be a loanword, although many North African Arabic dialects have alternations in the roots between emphatic and plain r, e.g., far 'mouse', pi. firan. There is a phonemic opposition of / r / vs. / \ / in the Judeo-Arabic dialect of Tafilalt, Morocco (Heath and Bar-Asher [1982:41]). Cf. also Jordanian Arabic mrce. I am unsure how to treat Cohen's (1912:122) citing ret 'I saw' with emphatic rand e vs. his (1912: 251, fn.l) retu 'I saw him' with rbut without emphatic p. This may be a misprint. There are only two verbs in the Jewish Arabic dialect of Algiers which preserves the second radical glottal stop: \fr>y 'see' and \/s>i 'ask'. Cf. Cohen (1912:196, fn.l). For an overall summary and comprehensive bibliography of the Judeo-Arabic dialects of North Africa, Yemen, and Baghdad, cf. Leslau (1945 b). Although the Judeo-Arabic dialect of Baghdad uses y for 'show', which is ?aya (note r > y in this lexeme, but cf. qa 'he read', where one might expect *qaya). In all cases discussed, Leslau (1945 b: 78, English summary) concludes that the Judeo-Arabic dialects are quite different from their Muslim counterparts in terms of phonology, morphology, and vocabulary. However, there has been little investigation of this important arena of study. One is very tempted to see an analogy between Serbian and Croatian here. The Serbs and Croats speak the same language (this is why scholars call it Serbo-Croatian), but the former are Greek Orthodox and use the Cyrillic alphabet while the latter are Roman Catholic and use the Latin script. There is also a parallel between Hindi (in terms of religion and script) and Urdu (which is why this language too is often hyphenated as Hindi-Urdu). ba$ura ~ba$ira means 'see' in Modern Standard Arabic, bassara is its causative, whereas Form IV also means 'see'. FormV means 'look at' and

218

Alan S. Kaye

Form X means 'be able to see'. I believe also the Egyptian Arabic verb ba$$, yibufs li- 'look at someone' or fi 'look at something' is related to y / b§r as is Modern Standard Arabic ba$$a, yabi$$u 'glow, shine'. Egyptian Arabic ba$$a$ 'spy, detective' is the profession form of the root \/b$$, viz., 'someone who is always looking for someone or something'. Some dialects use \/bsr for in 'mirror', viz., mab$ara (Aden, Yemen) (cf. Landberg [1923:1056]) - misyatin Datina. Modern Standard Arabic has both mir?dh and miraya. The former has two plurals: mara^in and maraya. 11 Leslau (1953) collected letters of Yemenite Jews at the Hashed Camp in Aden (in June 1950). All the Yemenite Jews passed through the Camp and then flew to Israel (called "Operation Magic Carpet"). The last letter published (1953:261-2) is in Arabic (with some Hebrew mixed in) but, unfortunately, it does not contain the verb 'see'. A search of the first six published letters in Hebrew with Arabic vocabulary and expressions reveals no occurrence of the verb 'see'. In terms of statistical frequency, the verb 'see' is not so common as one may sometimes think. 12 Landberg (1923:1046-56) discusses the dialectical developments of classical Arabic ra?d 'see', which he thinks developed into V ^ y (1923: 1046, specifically). From the point of view of semantic development, it is not impossible that 'see' > 'wait', especially considering that it collocates with it (viz., 'wait and see'). ra f is documented for Sudanese Colloquial Arabic. Cf. Kaye (1976: 44 and passim). This also occurs in Cypriot Maronite Arabic za< 'he came' = classical Arabic ja ?a. Owens (1981:93), in reviewing Kaye (1976), derives rate 'regard' from ra(a 'graze, tend a flock, observe'. Considering Landberg's work and plausible phonological and semantic development, there is much evidence in support of Landberg's thesis that V ^ y > V ^ y - T h u s e v e n Owen's cited forms in Nigerian Arabic ra: 'regard, look at' and ra:H 'shepherd' (Kaye 1982:37 glosses it better as 'herdsman' because raHlqanam is 'shepherd' [1982: 76]) can best be understood diachronically as 'one who regards/looks at/looks after (a flock) = 'herdsman'. Other developments pointed out in these pages are: Western Sudanese Arabic aretak y as a more archaic level. We agree with Jastrow's conclusion (1983: 206-7): " . . . we now see much more clearly than we did 20 years ago that the contemporary panorama of Arabic dialects is not a predominantly static endproduct of developments that occurred in the first Islamic centuries, but rather of a permanent process which has persisted until today and has gained momentum in the last decades."

References Blanc, Haim. 1964. Communal Dialects in Baghdad. Harvard Middle Eastern Monographs X. Cambridge, Mass.: Harvard University Press. Blau, Joshua. 1983. "Vestiges of tanwin -un and the Case Ending -u as Attested in Yemenite Judeo-Arabic Texts from the Seventeenth Century." Bulletin of the School of Oriental and African Studies, University of London, XLVI: 3: 529-31. Brunot, Louis and Elie Malka. 1939. Textes Judéo-Arabes de Fès. Rabat: Publications de l'institut des Hautes Études Marocaines. - 1940. Glossaire Judéo-Arabe de Fès. Rabat: Publications de l'institut des Hautes Études Marocaines. Cohen, David. 1969. Le parler arabe des Juifs de Tunis. The Hague and Paris: Mouton. - 1975. Le parler arabe des Juifs de Tunis. Vol. II. The Hague and Paris : Mouton. Cohen, Marcel. 1912. Le parler Arabe des Juifs d'Alger. Paris: Librairie ancienne H. Champion, Éditeur. Coniente, Federico. 1977. A Grammatical Sketch of the Spanish Arabie Dialect Bundle. Madrid: Instituto Hispano-Arabe de Cultura. Cowan, William G. 1966. "Two Notes on Arabic Dialectology." Journal of the American Oriental Society. 86.4:416-18.

The Verb 'See' in Arabic Dialects

221

Ferguson, Charles A. 1959. "The Arabic Koine," Language 35:616-30. Goitein, S. D. E. 1934. Jemenica: Sprichwörter und Redensarten aus Zentral-Jemen. Leipzig: Otto Harrassowitz. Gordon, Cyrus H. 1965. Ugaritic Textbook. Analecta Orientalia 38. Rome: Pontificium Institutum Biblicum. Heath, Jeffrey and Moshe Bar-Asher. 1982. "A Judeo-Arabic Dialect of Tafilalt (Southeastern Morocco). Zeitschrift fur arabische Linguistik 9:32-78. Heine, Bernd. 1982. The Nubi Language of Kibera - an Arabic Creole. Berlin: Dietrich Reimer Verlag. Jastrow, Otto. 1978. Die mesopotamisch-arabischen qgltu-Dialekte. I. Phonologie und Morphologie. Wiesbaden. - 1983. Review of F. J.Cadora, Interdialectal Lexical Compatability in Arabic: An Analytical Study of the Lexical Relationships Among the Major Syro-Lebanese Varieties (1979) in Journal of Semitic Studies 28:1.204-07. Kaye, Alan S. 1972. "Review of Maria Tsiapera, A Descriptive Analysis of Cypriot Maronite Arabic, Journal of the American Oriental Society92:4: 536-38. - 1976. Chadian and Sudanese Arabic in the Light of Comparative Arabic Dialectology. The Hague and Paris: Mouton. - 1982. A Dictionary of Nigerian Arabic. Malibu: Undena Publications. Landberg, de, Le Comte. 1923. Glossaire Dalinois, vol.11. Leide. - 1942. Glossaire Dalinois, vol. III, ed. by K. V. Zettersteen, Leide. Leslau, Wolf. 1944. "Texts on Yemenite Folklore." Proceedings of the American Academy for Jewish Research 14: 221-51. - 1945 a. "Hebrew Elements in the Judeo-Arabic Dialect of Fez." Jewish Quarterly Review, New Series 36: 61-78. - 1945b. "Judeo-Arabic Dialects." Yivo Bletter 26: 58-78 (written in Yiddish with English summary). - 1946. "Linguistic Observations on a Native Yemenite Document." Jewish Quarterly Review, New Series, 36:261-79. - 1953. "Contemporary Yemenite letters." The Joshua Starr Memorial Volume: Studies in History and Philology (Jewish Social Studies, Publication 5). New York: pp.249-62. - 1969. "Frequency as Determinant of Linguistic Changes in the Ethiopian Languages." Word 25: 180-89. (erroneously listed as [1968] in Monica S. Devens, "An Annotated Bibliography of the Works of Wolf Leslau," in Stanislav Segert and Andräs I. E. Bodrogligeti, eds. [1983], Ethiopian Studies Dedicated to Wolf Leslau, Wiesbaden: Otto Harrassowitz). Newton, Brian. 1964. "An Arabic-Greek Dialect." Word 20:43-52. Owens, Jonathan. 1981. Review of Kaye (1976), Journal of African Languages and Linguistics 2.91-94. - Forthcoming. Reviews of Heine (1982) and Bernd Heine, The Waata Dialect of Oromo: Grammatical Sketch and Vocabulary in Afrika und Übersee. Panetta, Ester. 1943. L'arabo parlato a Bengasi. Vol.1 Testi con traduzione e note. Rome: La Libreria dello Stato. Talmoudi, Fathi. 1984. "Notes on the Syntax of the Arabic Dialect of Süsa." Zeitschrift fur arabische Linguistik 12: 48-85. Vinnikov, 1.1962. A Dictionary of the Bukharian Arabic Dialect, Pal°est°insk0i Sbom°ik, vol.X.

The Comparative Study of Medieval Arabic Grammatical Theory Jonathan Owens Yarmouk University, Irbid

0. The Theory of grammar developed by the medieval Arabic grammarians is undoubtedly one of the great structural theories in the history of linguistics, and hence appreciating its full significance, its strengths and weaknesses, depends in part at least on comparing it to other structural models of linguistics. Most importantly, one should look initially at modern linguistics, first because it is most accessible to linguists writing today, and secondly because in modern linguistics is found the greatest variety of linguistic models: it provides the richest range of theories on which to base comparisons. This latter point is an important one, because it guards against the tendency to identify medieval Arabic grammatical practice too completely with one modern theory or another. This identification is pernicious in two ways: it obscures a proper understanding of Arabic theory and it deprives us of potential insights which might be drawn from Arabic theory. This is not to say that Arabic theory should be regarded as dramatically different from modern linguistic practice. Indeed, I am struck more by the similarities than by the differences, as will be apparent from the subsequent discussion. What I will argue is that Arabic grammatical theory shares a number of basic methodological and theoretical assumptions with modern western linguistic practice, that different aspects of Arabic theory are closer to the western than are other parts, and that Arabic theory cannot be understood exhaustively by comparing it to only one aspect of modern grammar. In the following I look at four essential aspects of Arabic practice, and interpret them in terms of modern western theory: substitution, dependency, markedness and pragmatics.

224

Jonathan Owens

1. Substitution Carter (1973 a), in probably the first demonstration of the similarity between Arabic theory and modern practice,1 shows that Slbawaih (d. 798), the first and most important Arab linguist, used substitution techniques to define the status of syntactic items: the syntactic status of larger units was defined relative to that of smaller items. The following example is Carter's (1973 a: 154-155) summary of Slbawaih (1:352). 1.

Verb

Object Topic I Pssd

Pssr

! Object i ! I idrib ayy man hit which whoever "Hit whichever one that you consider better." man ra'ayta= al-qawmi ayyu man ra'ayta= alladina

Comment I j Verb ! ! • ra 'ayta afdalu consider you better

The example has a complex sentence with verb and object (cf. (10)), the object in turn consisting of a topic-comment structure (cf. (12 b)), with the topic in turn containing a complex structure of possessed (ayyu) + sentential possessor (man ra'ayta). Slbawaih establishes the syntactic status of the larger sentential units through substitution with single nouns: man ra'ayta "whoever you consider" has the position of a noun like al-qawmi (hit whichever people are better"), and similarly the unit ayyu man ra'ayta is equivalent to the relative pronoun, alladina ( = "hit the ones who (alladina) are better"). The verb itjlrib has a sentential object, ayyu man ra'aytu afdal, which is a complex sentence consisting of topic (cf. e.g. (12b)) ayyu with a sentential possessor complement man ra'aytu. Through substitutions this is reduced to the more manageable, idrib alladina afdal. What is noticeable here is that each substitution of larger for smaller units involves S (sentence) as the larger and N (noun) as the smaller. In general the smaller unit in Arabic theory is only N, a fact which distinguishes Arabic theory from modern constituency analysis (e.g. Wells, 1947; cf. Owens 1984a: 55 for further discussion). I will return to this point in the next section.2

The Comparative Study of Medieval Arabic Grammatical Theory

225

2. Dependency The Arabic grammarians used substitution techniques for precisely the same reason they are used today, to reduce a large number of superficially dissimilar structures to a limited number of grammatical types. Syntactically the grammatical types are related to each other in dependency terms, as I have shown in detail (1984 a). Interestingly, this fundamental aspect of syntactic analysis - fundamental because it provides the basic means by which any two syntactic items relate to each other - foreshadows in detail the four formal conditions which define a dependency relation: only one item is independent; all other items depend on another; no item depends on more than one item, and heads and dependents tend to occur next to each other. These were as much explicit principles of Arabic theory as they are in modern dependency grammar (cf. Mubarrad IV: 80, 156, Sarraj I: 102, II 152, Ibn YacIsh III: 67 for Arabic principles). The dependency analysis of sa°ada zaydun gulama l-rajuli "Zayd helped the man's boy", is essentially the same in Arabic theory and in Tesniere, Robinson, and other modern writers: I (2) sa°ada helped

» zaydun Zayd

1I I gulama l-rajuli (arrow points to dependent) boy -acc def-man gen

Indeed, the Arabic grammarians went further than Robinson's definition, for example in recognizing that sequence relations among items could be ordered in terms of the generalization that heads precede dependents in Arabic (cf. e.g. Sarraj 1:108, Xa$ I: 298, Bat: 148, 149, Ibn Ya°Ish I: 74, and Tesniere 1959: 22 for the modern discovery of this principle). Although the two are not contradictory, a grammar does not need to represent syntactic relations both in constituency and dependency terms,3 and in implicit recognition of this fact Arabic grammar operates almost completely in dependency terms. The one exception pertains to the category S, which represents all items in a sentence as a unit. I have given a structural reason for this: the initial independent item in an Arabic sentence can be a verb, particle or implicit operator, hence there is no unique item to represent the sentence as a whole. I (3 a) tadrubu

^ hindun

I zaydan

I b) lan tadrub-a hindun neg/fut sbjc

"Hind is hitting Zayd." 1 zaydan

"Hind won't hit Zayd."

Jonathan Owens

226

In (3 a) the initial independent item is the verb tatjirubu while in (3 b) it is the particle lan which governs the verb in the subjunctive. In Arabic theory there is no unique independent item in a sentence. In modern dependency grammar by contrast the verb is generally assumed to be head of a sentence (cf. Owens 1984 a: 40,41 for details). Another reason the category S rather than Verb may have come to represent the sentence is the importance that came to be attached first to the notion of sentence as consisting of subject-1-predicate=jumla (e.g. Sarraj I: 70, 76, II: 9, 30) and later to that of independent sentence, kalam (Astarabadi SAT:8, Ibn Hisham ML:490).4 As units jumla and kalam represent exocentric relations: jumla most basically is topic-(-comment or verb + agent (cf. (12)); kalam represents an independent sentence in its entirety. (4) zaydun Topic

(awilun "Zaydistall" Comment=jumla (any subject + predicate) and = kalam (an independent sentence)

The emphasis is on the combination of two units which collectively form an essential whole, one constituent without which meaningful utterances are impossible. We are now in a position to explain the constraint alluded to in the previous section on the necessity of using a single word as the basic member of a substitution pair. Since all relations are dependency ones involving relations of one word to another, the defining member of a pair will always have to be a single word, since only they (and S, which being the largest unit cannot be the basis for substitution by a smaller one) have a structural status.

4. A$l/farc; basic/marked The terms a$l/farc have been interpreted in essentially two ways in the recent literature, either as "deep" and "surface" or something closely equivalent (Ayoub and Bohas 1983: 32, 47, cAmayreh 1985)5, or in terms of inherent rights of grammatical elements (Versteegh 1978). The former is obviously inspired by generative transformationsl grammar while the latter relies more directly on terminology of medieval intellectual thought. I will return to the first interpretation below. Versteegh's summary aptly characterizes the vital status of the 'a$l/ farc dichotomy within Arabic grammatical thought: every element in grammar has a certain basic right (agl) within a given structure, and un-

The Comparative Study of Medieval Arabic Grammatical Theory

227

der certain specified conditions the element can either have certain rights added or taken away, in which case it takes on a farc (lit. "branch") status. Versteegh couches his explanation within the juridical terminology which inspired the original formulation of medieval Arabic terminology (Carter 1972, Haarman, 1974). I will suggest here that the terms a$l/ far0 have a direct analogue in the basic (unmarked)/marked dichotomy of modern linguistics. A complete and detailed exemplification of this point is outside the scope of this paper, for as Versteegh explains the a$l/far° distinction pervades virtually every aspect of Arabic grammatical description. I will thus content myself with exemplifying the point with one relatively clear instance of the compatability of terminology, and remark briefly on other examples below. One of the areas of grammar where the a$l/farc distinction was most developed concerned partially inflected nouns (gayr munsarif). These are nouns which necessarily lack the indefinite -n suffix, and which usually lack distinct accusative/genitive forms, two essential defining qualities of nouns (Anban Asrar: 10ff.; Carter 1981: 76). The Arabic grammarians used markedness to explain why these inflections were lacking (cf. below). The explanation went as follows: if a noun contained two (exceptionally, one) of the following marked characteristics, it became partially inflected (Zajjaj: 2, Xa$ 1:174ff.). (5)

unmarked, a$l

marked, far0

noun noun singular indefinite simple Arabic origin regular

verb adjective plural definite compound non-Arabic origin substituted (macdul)

A noun like ibrahïmu "Ibrahim" lacks the indefinite -n and has only one oblique form ( ibrahïm-a acc/gen) because (a) it is definite (proper nouns are inherently definite) and (b) because it is of foreign origin.6 This very brief account overlooks many difficulties and exceptions among various sub-classes of nouns, exceptions which were dealt with only in relatively ad hoc fashion (e.g. FarisI °Askariyàt 145,146, Anbârî, Asràr: 313, 314), though regardless of the details of the analysis, what is immediately striking about the list is its reasonableness and

228

Jonathan Owens

plausibility. To verify this we can look at the behavior of marked/unmarked categories in the Arabic language in respect of two of Greenberg's (1966: 26-31) characteristics of morphological markedness. Here I will not be so much interested in whether the Arabic grammarians adduced the following arguments to justify their classifications as in showing that in Arabic the categories in (5) do correlate with universal principles of markedness. (1) Neutralization: if a contrast is lost in a given context the unmarked item occurs. Nouns contrast as sg./pl. in Arabic; however, when they occur with numbers over 10 only the singular form is used. (6) arbacu darahima "four dirhams" c ishruna dirhaman "twenty dirhams" (dirham = singular form) (2) Syncretization: the distinctions existing in the unmarked category are lost in the marked. The classic case in Arabic theory is the one exemplified above: the three-fold distinction nominative-genitiveaccusative reduces to a two-way morphological contrast, nominativeaccusative/genitive, and the two-way indefinite form-definite form to a single form, though only when a noun has two marked categories. Examples could be given confirming that the other markedness categories do divide along the lines set out by the Arabic grammarians, though I think the two examples suffice to illustrate the general point. They also can be used to demonstrate that in some cases the reasons the Arabic grammarians gave for setting an item in its markedness category correspond to the motivation in modern linguistics, and sometimes they do not. For example, although singular is considered marked against plural, I know of no linguists who attempted to justify this classification with examples like (6). Rather, it appears to be derived by fiat or logical deduction rather than through inductive reasoning (e. g. Asrar: 48). The marked status of definiteness, on the other hand, is justified implicitly by the observation that the marked member of the pair lacks certain morphological inflections: a noun which is a loan word and definite, for example, lacks two morphological features. There remains, however, one point of difference between the two traditions. This concerns the use made of markedness categories in linguistic explanation. The difference can be diagrammed as follows.

The Comparative Study of Medieval Arabic Grammatical Theory (7)

Basic

Marked

+ feature x

— feature x

zaydun

ibrahimu

proper noun of Arabic origin fully inflected

proper noun of non-Arabic origin lacks full inflection

229

Arabic theory: a noun has certain marked categories ( ~ f ) therefore it lacks inflection Western: a noun lacks certain inflections, therefore it is marked, and the categories it represents are marked In a Greenbergian treatment one uses the skewed distribution of case and definiteness inflection to argue for the marked status of definite/foreign (syncretization occurs in marked categories). Here the nouns themselves are the independent variable, the marked status of the categories definite/foreign derived from their behavior. For the medieval Arabic grammarians the categories themselves are treated as independent variables such that by definition definite/ foreign (and other members on the right-hand side of (5)) are marked. The form of the nouns, on the other hand, is a dependent variable so that a causal explanation follows: ibrahimu lacks distinct accusative/ genitive forms and indefinite -n because it has two marked properties. The line of reasoning may be circular, but it leads to the same results as a western approach in this instance. As a final point I would like to return to the question of what relation the terms a$l/farc have to the notions of deep/surface in generative-transformational grammar. So far as the examples so far go the association as a general explanatory concept will have been seen to be suspicious if only because a$l/ far0 relate items at different levels of grammar (e. g. the morphology example above) whereas deep/surface apply specifically to syntax (Chomsky 1965:71). The point that markedness, not deep/surface, is the more correct analogy is confirmed when one looks to find a general processual term that is used to signal the movement of an item between the a$l/far° categories. Although no one term is used, I believe the most general one is daxala "enter", which is used in the formulation that a marked category (far0) enters in on a more basic (a$l) one. This formulation applies at all levels of grammar, as a few examples will show. In all the

230

Jonathan Owens

examples below the items on the left are the basic category, those on the right the marked. Phonology. Nouns which have a basic root beginning with a -CC sequence "enter" an epenthetic 7- prefix (hamzatu l-wa$l). (8 a) bn-u "son-nom" —• 'ibnu (Mubarrad II: 67, 88, $aymari: 436) Morphology. The prefix 'in- "passive" is said to "enter" onto a basic verbal root. (8 b) kataba -* 'in-kataba "write-be written" (Ibn YacIsh, Shark al-Muluki: 69) The definite article "enters" a noun to make it definite. (8 c) rajulun "a man" —>- al-rajulu "the man" (Lughda: 244, Farisi, cAskariyat: 123, $aymari; 220, etc.) Syntax. Various modality and negation particles "enter" a basic sentence without changing the inflectional form (though the meaning may change). (8d) zaydun darifun "Zayd is nice." —• hal zaydun darifun nom nom Q "Is Zayd nice?" ma zaydun darifun neg "Zayd is not nice." (Farisi, °Askariyat: 81, I(lah: 94, §ayman: 130, Ibn Jinni, Xa$ I: 168) In each pair the term daxala is consistently applied: it describes a situation where a non-basic category enters upon a basic one, giving the entire structure a more marked (far0) status. It might be, and indeed has been, argued that other terms could be applied to the a$l/far° distinction. cAmayreh (1984:108,115) for example, applies the notion of kernel sentence (jumlatu l-asli, p. 96) and transform sentence (jumla tahwiliyya) to pairs of non-Q - Q sentences and non-negative - negative ones, as in (8d). (9) zaydun darifun = kernel —>• ma zaydun darifun (asl) The implication clearly is that the basic (asl) can be understood in the sense of kernel sentence (Chomsky 1957: 45).7 One has to agree that there is a relevant analogy here, though on close inspection, while

The Comparative Study of Medieval Arabic Grammatical Theory

231

it does not completely break down, it does get subsumed under a more general one. Leaving aside technical objections to the analogy,8 the a$l/farc distinction in medieval Arabic theory applies at all levels of analysis, as illustrated in (8). The notion of kernel sentence, on the other hand, applies exclusively within the realm of syntax, and hence is inapplicable to examples like (8 a-8 c). Moreover, even within the syntactic level there are many differences in detail between medieval theory and TG. Indefinite and singular, for example, are basic (a$l) categories, definite and plural marked (far0), yet in most versions of generative grammar both pairs of the categories indefinite/definite, singular/plural are base, not derived categories (e.g. Syntactic Structures: 28, 39, Postal 1970: 62 ff., Stockwell et alia 1973: 74ff.). In sum then, it is difficult to see how the terms deep/surface as understood in TG grammar could be applied consistently to all the examples cited in (8), even if they can be applied to a sub-set of them. On the other hand the dichotomy basic/marked accommodates all the pairs, and hence offers the modern linguist a better tool for the understanding of the Arabic terms.

5. Jurjani and Pragmatics Arabic theory became increasingly prescriptive the further it moved from its original foundation with Sibawaih (cf. Carter 1973 b, 1984). Later theory was not without its theoretical innovations, however, and perhaps the most important of these was Juijanl's (d. 1078) work on word order variation. Juijanl's contribution can be looked at in terms of pragmatics and discourse structure. I will give only two examples from his analysis of word order variation, though I believe it suffices to indicate the general flavor of one of his areas of research. In Arabic VSO, VOS and SVO are all possible word order variants (cf. Yasin 1985 for some discussion). (10a) 'a tabni hadihi l-dar-a V(S)09 Q build this def-house-acc "Are you building this house ?" b) 'a anta tabni hadihi l-dar-a Q you build

"Is it you who are building this house?"

(11 a) qatala l-jundiyyu l-xarijiyya killed def-soldier def-stranger "The soldier killed the stranger"

232

Jonathan Owens

b) qatala l-xarijiyya l-jundiyyu (same meaning, cf. below) acc nom Variations in word order such as these had been duly noted by Jurjanl's predecessors, though almost always they were more interested in the structural implications of word-order permutations than in semantic differences (e.g. Sarraj I: 208ff. II: 131 ff., Zajjajl, Jumal: 10). Jurjanl's great contribution was to relate them to differences of meaning by taking into account discourse and pragmatic factors. The choice between (10 a) and (10 b) is not arbitrary, but rather depends on presuppositional facts (Dala'il: 87-96). (10a) would be appropriate if the questioner is concentrating on whether the action occurred, while (10 b) is relevant if the action is known to have occurred (is presupposed), but the doer of the action is in question. Similarly, (11 a) is appropriate if the background information has indicated that l-jundiyyu is the more important of the two complements in the sentence, ( l i b ) if l-xarijiyya is more important. For each pair the word order is dictated by one pragmatic rule: the important item is placed first. JurjanI provides similar word-order variation pairs based on presuppositional factors for negatives (Dal: 96-99) and indicatives (99-102). Juijani's insight is fundamental, and seems to me foreshadow in particular the multi-systemic analysis of Halliday (1967, 1968). This becomes clear when further aspects of Jurjam's work are taken into consideration. JurjanI wrote an extremely detailed, though very orthodox work on grammar called al-Muqta$id in which he largely recapitulates the standard grammatical analysis of the day. One aspect of this analysis is the division of sentences into two basic types, verbal and nominal. Verbal sentences begin with a verb, nominal with a nominative noun.10 (12a) verbalS qama zaydun got up Zayd "Zayd got up." V Agent b) nominal S zaydun qama nom "Zayd got up." Topic Comment The two sentences, though virtually synonymous (Xa$ 1:184,196) have different analyses. (12 a) is a verbal sentence where zaydun has

The Comparative Study of Medieval Arabic Grammatical Theory

233

the grammatical role of agent, qama of verb while (12 b) is a nominal sentence with zaydun in the grammatical role of topic (,mubtada'), with qama comment (xabar). This is a classification which JuijanI clearly embraced (Muqtagid: 93) and hence the sentences in (10) have two different syntactic analyses, as indicated in (13). (13 a) 'a tabni( ) hadihil-dara Verb (Agent) Object b) 'aanta tabnihadihil-dara Topic Comment Note that the relations between (10/13 a/b) could not be considered transformational in any sense since both represent basic sentence types in Arabic theory. I believe that Jurjani is suggesting that sentences have a multi-systemic analysis which consists of two parts: a syntactic analysis as indicated in (13) and an information structure, which can be indicated as follows. (14) 'a tabni hadihi l-dara 'a anta tabni hadihi l-dara New Old New information comes first, old, presupposed follows. This yields a composite structure as follows, with new information underlined. (15) 'a tabni Verb 'a anta Topic

hadihil-dara Object tabni hadihi l-dara Comment

There are various caveats to be added here, not the least of which is to note that Jurjani himself never worked out a structure like (15) schematically, which means that modern readers must use their own inferences. Further, for Halliday information structure is marked by intonation, and Arabic grammarians unfortunately gave little attention to this aspect of speech. Moreover, a Hallidayean system has to take into account the thematic structure, and it is not clear that this is given separate recognition by JuijanI. Indeed, the differences between (11 a/b) are perhaps better understood as thematic ones: the most important noun comes first, but in Jurjani's description of the conditions under which (11 a/b) would be used, "important" is best identified as "the-

234

Jonathan Owens

matic", not "new", whereas in (lOa/b) I believe "important" is best identified as "new". What is clear, though, is that JuijSnl recognized that a pragmatic level independent of the strictly grammatical had to be posited for an adequate understanding of language.

6. Conclusion I hope to have shown that the comparative study of medieval Arabic grammatical theory requires an eclectic approach. Its different parts can be seen as similar to various ideas in modern linguistics: it used substitution techniques similar to those that inspired IC analysis; its dependency-based account of syntactic relations perhaps more than any other of its aspects correlates in detail with one modern theory, that of dependency grammar; a theory of markedness pervaded all levels of grammar, and at least for Juijani and his successors it had a discrete level of pragmatic analysis. Trying to explain Arabic theory solely in terms of one of these modern analogues fails to do justice to the system of medieval Arabic theory as a whole; looking at it from different modern perspectives gives us greater scope to appreciate its theoretical bases and, who knows, may herald in the day when modern linguistic theory takes some cues from its medieval predecessor.

Notes 1 Diem, 1970/71 also is moving in this direction in places. 2 The smaller unit is always a noun, never a larger item. The larger unit is usually S, though informally the equivalence between N and sub-sentential phrases is noted; cf. Owens 1984b: 64ff. 3 Hudson, 1976 argues for a mixed constituency-dependency approach, though largely abandons this in Hudson, 1984, except for coordinate phrases (on not altogether strong grounds, cf. Owens, 1985). 4 The term jumla is used hardly at all by Sfbawaih; in book IV of Mubarrad (d. 898), the volume with the greatest emphasis on syntax and the one containing the chapter on predication (isnad), I count three uses ofjumla, all in the sense of dependent sentence (pp. 132,347,348). In Sarraj (d. 928), in the first 100 pages alone the number jumps to 9 {jumla = either dependent or independent sentence), which indicates an increasing importance for the term.

The Comparative Study of Medieval Arabic Grammatical Theory

235

5 Ayoub and Bohas use the term "abstract representation" for the basic word order. They claim to be using the term only in a sense utilized by the Arabic grammarians, though I believe the term has too many associations with the abstract representations of generative grammar not to conjure up questions of equivalence. 6 One can contrast a proper noun of Arabic origin, like zaydun "Zayd" which though definite takes the indefinite -n and all three case inflections (zayd-un/zayd-an/zayd-in, nom/acc/gen). 7 A kernel sentence is one derived after application of all obligatory transformations, and before the optional ones. By Chomsky, 1965, it no longer figures in Chomsky's version of TG. 8 For example, transformations in TG relate sentences to sentences (Chomsky 1957: 46 ff.); the a$l/farc distinction can apply to sub-sentential elements. Kernel sentences are derived from a more abstract base, which has no actual representation. In Arabic theory a sentence which is of the a$l category (e.g. (8d), left side) can be an actual surface sentence, as are the farc ones. I might point out that in the respect the sentential asl/farc pairs bear closer similarity to Harris' (1957) transformational theory, where a base form can be an actual sentence, than to Chomsky's. 9 The agent is expressed in the verb, and is assumed to occur after it; cf. Owens 1984: 58 for discussion. 10 Ayoub and Bohas explain some marginal exceptions to this statement; cf. Anbari, Asrdr: 83 for structural justifications of the distinction between the two basic sentence types.

References Medieval Arabic Sources (with identifying abbreviation used in the text in the margin) Asrär. Ibn Al-Anbäii, Abü 1-Barakat (d. 1187). Asrär al-cArabiyya. Ed. by M. al-Bltär. Damascus: al-Majma0 l-cArabi, 1957. SK. Al-'Astaräbädl, Radi I-Din. (d. 1286). Sharh Kitäb al-Käfiya. Beirut: Dar al-Kutub al-cIlmiyya (n. d.). Bat. Al-Batlyüsi, Abü Mohammad, (d. 1127). Kitäb al-ffilalfi I$läh al-Xilal min "Kitäb al-Jumal". Ed. by Sa°Id Sa0CüdI. Baghdad: Wizärat al-Thaqäfa wa l-'Icläm. (n.d.). c Ask. Al-Farisi, Abü cAlI (d. 987). (a) Al-Masä'il ai-cAskariyät fi l-Nahw al-cArabi. Ed. by cAlI 1-Man§üri. Baghdad: Baghdad University, 1982. Idäh. Al-Färist, Abü °AlI. (b) Al-Idäh. (text included in Jurjänl, a). ML. Ibn Hishäm, Abü Mohammad, (d. 1360). Mughnì l-Labìb °an Kutub al-'Acärib. Ed. by Mäzin Mubarak. Beirut: Dar al-Fikr, 1969. Xag. Ibn Jinnl, Abü 1-Fatlj. (d. 1003). Al-Xafä'if. Ed. by Mohammad °Ali 1Najjar. Cairo, 1952-1956. Muqtajid. Al-Juijänl, c Abd al-Qahir (d. 1078). (a) Al-Muqtasid fi Sharh al-'Idäfr. Ed. by K. B.al-Muijän. Baghdad: Ministry of Culture and Information, 1982.

236

Jonathan Owens

Dal. Al-JuijanI, cAbd al-Qahir. (b) Dalà'il al-lcjàz. Ed. by Moljammad Rida. Beirut: Dar al-Macrifa, 1978. Lughda. Lughda, Abu cAlI (d. 911). Kitàb al-Nahw. Ed. by °Abd al-Husayn alFatll. Al-Mawrid3/3: 221-246,1974. Mubarrad. Mubarrad, Ibn Yazld (d. 898). Al-Muqtadab. Ed. by c Abd al-Xallq c U 'ruler'4 while sós occurs in the meaning 'shepherd' in Coptic in all the dialects except Bohairic, that of the Delta. While it does have hieroglyphic antecedents it does not occur until New Egyptian as i^-wbut means 'Beduins northeast of Egypt and their country' not 'shepherd' in general. In considering the history of the Egyptian language, we may note that most writers on this topic recognize five periods which in some instances are initiated by changes in the writing system as well as in the spoken language. A few including Sethe and Grapow distinguish a sixth and earliest period exemplied chiefly in the language of the Pyramid texts. In the present context I will consider six periods as providing a richer basis for discussion. They may be rather briefly and in somewhat oversimplified fashion be characterized as follows.

1. Earliest Egyptian This is chiefly the language of the Pyramid Texts which appears in hieroglyphic form on the walls of the pyramids of five kings of the 5th and 6th dynasties (2494-2181) at the end of the Old Kingdom. Although their first written attestation is later than that of Old Egyptian, there is general agreement that the archaism of their language indicates an earlier linguistic stage.

276

Joseph H. Greenberg

2. Old Egyptian This is the literary language of the Old Kingdom from the 3rd through the 6th dynasties (2688-2181). No consecutives texts are known from the two first dynasties constituting the Proto-Dynastic period and what there is is often difficult to interpret. Old Egyptian occurs written in hieroglyphics on monuments. Alongside of it there is hieratic used for writing on papyrus. Hieratic characters are written in detached form and there is an essential one-to-one correspondence between the individual characters written in each script. Hence hieratic texts are often published in hieroglyphic transscription. There are also no differences in vocabulary or grammar between the two forms.

3. Middle Egyptian After the First Intermediate Period, of uncertain length and during which no central authority existed, national unity was restored under the 11th Dynasty and continued under the 12th (2233-1786). This period is called the Middle Kingdom and the literary language which came into use in this period Middle Egyptian. There followed the Second Intermediate Period during which there was once more no single centralized authority. The period of Hyksos rule in the eastern Delta is part of this period. National unity was once more restored inder the New Kingdom (Dynasties 18,19 and 20,1567-1085). During the earlier part of the New Kingdom, Middle Egyptian with hardly any detectible change continued to be used as the ordinary literary language, except for a few instances to be mentioned later. Middle Egyptian is generally considered to be the "Classical" form of the language and a grammar simply called that of Egyptian will be primarily concerned with this form of the language. Like Old Egyptian it concurrently used hieroglyphic and hieratic systems of writing.

4. New Egyptian The famous religious innovator of solar monotheism Akhenaten, who reigned 1364-1347 was also an innovator in matters of language. No doubt as part of his cult of "naturalness" he introduced what was primarily the form of Egyptian that he himself spoke. It differs considerably in grammar and vocabulary from Middle Egyptian but continues

Were there Egyptian Koines?

277

to be written in the hieroglyphic script and its accompanying hieratic form. However, Middle Egyptian still enjoyed such prestige that it continued in use, however artificially, until the time of the Roman Emperor Domitian in 296.

5. Demotic There followed another period of internal Egyptian disunity, the Third Intermediate Period. This included two Dynasties of Lybian origin and several native dynasties which were partially concurrent and none of which ruled the entire country. National unity was restored under a foreign, but culturally Egyptianized ruler, Piankhi of the 25th, or Kushite, Dynasty (715-656). This was followed by the national restoration which began under the Saite or 26th Dynasty and during which the entire country was under indigenous rule. It was however interrupted by an initial period of Persian conquest (525-404) and, after a second period of native rule, by the Persian reconquest under Artaxerxes III in 341. After a brief interval there followed the conquest by Alexander the Great in 332. Thereafter Greek rule was continued under the Ptolemies until the incorporation of Egypt into the Roman Empire. After a transitional period of "Abnormal Hieratic" (dated by Grapow, 1944, 213 as occurring about 1100), there appears beginning in the Kushite period during the eighth century a highly abbreviated script involving numerous ligatures of previously separate hieratic characters and known as Demotic. It began in business but was officially adopted during the Saite restoration and even used for monumental purposes. Demotic writing follows the same basic principles as hieroglyphic and hieratic writing being partly ideographic and partly phonetic but the forms, abbreviated from hieratic, have been in some instances so drastically altered that the hieratic source is uncertain. Demotic also represents a new stage linguistically. It continued in general use until the third century of our era. It is one of the three languages of the Rosetta stone alongside Middle Egyptian and Greek.

6. Coptic The last stage of the Egyptian literary language involves the use of the Greek alphabet supplemented by a few consonantal signs of Demotic

278

Joseph H. Greenberg

origin to represent sounds not occurring in Greek. Biblical translations into Coptic began in the third century at a time when Demotic was still in active use by pagan Egyptians. However the use of the Greek alphabet to represent Egyptian antedates its employment by Christians. At least a century earlier we find the Greek alphabet used on mummy labels and in magical papyri in which it was believed that the efficacy of the spells depended on their being pronounced exactly. The language of these glosses is called Old Coptic. Coptic is divided into a number of literary dialects. There has been much discussion of the number of distinct dialects to be recognized and their localization but something approaching a general consensus has been reached.5 From south to north in the Nile valley there are Achmimic, Subachmimic, Oxyrhynchus and Saidic. In the Faiyum, that is, the area on and near Birket Qarun (the ancient Lake Moeris), and probably extending to the west bank of the lower Nile at this point, is Faiyumic. Finally there is Bohairic, the dialect of the Delta. By the fifth century Saidic had apparently replaced all the other dialects of the Nile valley but not Faiyumic. It was the vehicle of an original Christian literature, not merely translations from the Greek. Later, Bohairic assumed greater importance and about 1000 it became the official liturgical language of the Coptic Monophysite Church and continues in this function alongside of Arabic up to the present day. How is the relationship among these six successive stages to be envisioned? As the starting point we may take the discussion in the important paper of Sethe (1925). This study, in spite of its title which refers to the relationship of Demotic and Coptic and its main content which considers the relative similarities of Demotic to New Egyptian which preceded and Coptic which followed, concludes with the first overview of the entire historical development of Egyptian. It is accompanied by a diagram which takes the following forms. The spoken language is represented as a single dotted line indicating a continuous and gradual development from the Earliest Egyptian of the Pyramid texts to the last stage, that of Coptic. At intervals, as the gap between the written and spoken form became intolerable, new literary languages are introduced quite abruptly and they continue virtually unchanged until the separation between the literary and spoken language once again becomes unbearably great. There is just one exception; Sethe considers Demotic to be a direct continuation of New Egyptian in a new orthography while deemphasizing the resemblances between Demotic and Coptic which were central in the standard grammar of Demotic by Spiegelberg (1925).

Were there Egyptian Koines?

279

Sethe's theory was dubbed 'cataclysmic' by Gunn (1924, ix) since each literary language was portrayed as having a period of sudden birth from the spoken language at specific intervals. Sethe's theory is somewhat modified in Strieker (1945) who tends to see each form of the literary languages as assimilating some of the features of the spoken language over time. Strieker also takes into account the earlier occurrences of Late Egyptian forms in the literary language before the period of its official use by Akhenaten. What strikes a linguist, however, is not that Sethe's theory is 'cataclysmic' but rather that the spoken language is conceived as a single and countinuous though changing entity for thousands of years. Nowhere in either Sethe's or Strieker's treatment is there any mention of dialect variation. Others, however, have taken this factor into consideration. Of these the first seems to have been Edgerton, and later others as, for example, Edel and Callender. Thus Edgerton (1951,11) states that he considers it quite possible that Middle Egyptian and New Egyptian may represent different parts of Egypt as well as different time periods. In what follows I shall try to present, at least in rough outline, what I consider to have been the course of development of the Egyptian language. Virtually all of the individual points have already been made by other writers who will be referred to at the appropriate points in the discussion. My own conclusions have been arrived at by an independent evaluation of the evidence. To my knowledge this is the first attempt since Strieker to give an overall view beyond the usual enumeration to successive periods of the language. As a preliminary step it will prove useful to discuss some of the main factors and assumptions, both linguistic and nonlinguistic which need to be taken into account in the construction of an overall theory. One of these is the realization that Egypt, when it was politically united, which holds for most of the long period we are considering, was a highly centralized state which utilized labor on a vast scale in royal constructions and in national military expeditions and supervised a complex irrigation system. All of this required a constant flow of emissaries delivering instructions and receiving reports from provincial centers and also required a vast and ubiquitous army of scribes. Not only was writing essential but a central language of administration was virtually indispensible. One would expect the language of the political capital to play the central even if not the exclusive role in the development of successive official languages. A further major factor in Egyptian history that has to be taken into frequent account is the fundamental division of the country into the

280

Joseph H. Greenberg

Nile valley (Upper Egypt) and the Delta (Lower Egypt). The beginning of the Dynastic period is marked by the conquest of Lower Egypt by Upper Egypt under the leadership of Menes. This is based on Manetho's account but has been verified by the discovery of the Narmer tablet and by other evidence. 6 According to Breasted (1908,119), the kings of the 5th Dynasty still felt themselves to be southerners ruling the north. The king of Egypt wore a double crown combining that of Upper and Lower Egypt and the treasury and other departments had southern and northern divisions. Egypt was often called 'the two lands', a designation which lasted until the Coptic period. The persistence of the distinction can, however, be noted in Coptic itself. It is clear that the Coptic dialects can be classified into two groups, one consisting of Bohairic alone, the speech of the Delta, and the other of the remaining dialects. Before proceeding further one linguistic problem requires discussion, what was earlier called the succession problem. Does a particular stage in a literary language continue an earlier stage of the same dialect or does it stem from a different region and dialect? A characteristic and usually decisive phenomenon may occur when there are three successive periods, say A, B and C when C derives from the same area as A but B has a different provenience. What we may get is an apparent reversal in historical change. A linguistic feature x in stage A is replaced by y in stage B but then reappears as jc in stage C. The following is an example from morphology. The oldest attested form of Ethiopian Semitic is Geez, also known as Classical Ethiopic. Various writers on comparative Semitic have either considered the present Semitic languages of Ethiopia all to have been the direct descendents of Geez (Wright 1890,29) so that Geez is in effect Proto-Ethiopic or that Geez is the direct ancestor only of Tigrinya in the northern subgroup of Ethiopian Semitic and spoken in the same general area as Geez while the other contemporary languages descend from a sister dialect of Geez (Brockelmann 1961,1.31). However there is at least one feature of Geez which suggests strongly that it cannot be the direct ancestor of any present day Ethiopian Semitic language, an opinion in which I am in agreement with a leading specialist in this group of languages, W. Leslau. The Semitic verb had a derived, so-called intensive, form which involved the doubling of the second consonant in the dominant triconsonantal pattern. However it also used gemination of the same second consonant in a particular tense-aspect form, the present or continuative which only survives in Ethiopic Semitic and in Akkadian but

Were there Egyptian Koines?

281

whose Proto-Semitic status is further assured by its occurrence elsewhere in Afroasiatic (Greenberg, 1952). From this twofold function of the geminated second consonant there arises an ambiguity in that the present continuative, usually called the imperfective in grammars of Ethiopian languages, is identical in both the simple and intensive forms of the verb. This ambiguity was removed in Geez by an innovation of unknown origin; the a preceding the geminated consonant was replaced by e in the intensive form. However in all present day Ethiopian Semitic languages either the ambiguity persists or is eliminated in some other manner; nowhere outside of Geez is this e found. Had Proto-Ethiopic left written records it would presumably show ambiguity in these two forms. Hence we would have Proto-Ethiopic a, Geez e and once more modern a and there would be an apparent reversal in the course of change. The most powerful evidence of this kind is probably phonological. When there is a complete merger of two originally distinct sounds they cannot be unscrambled to produce the former state in the normal course of change. In Ferguson's article on the koine he notes the merger of dad and (td ? which never have distinct reflexes in any of the koine forms of Arabic but which are distinguished in Classical Arabic. However the dialects of the Arabian peninsula do in general retain this distinction and in the same lexical items as Classical Arabic. This is strong evidence that they are not the direct descendent of the ancestral koine language. Should the peninsular dialects ever become the basis of a new koine, an utterly improbable occurrence on sociolinguistic grounds, we would have an apparent reversal in the course of change from unmerged (Classical) to merged (present day koine languages) to unmerged in the hypothetical new koine. With these considerations in mind we consider the six successive stages of Egyptian in their relationship to each other using both linguistic and non-linguistic evidence. The first stage is generally agreed to be Earliest Egyptian, the language of the Pyramid texts, which exhibits clearly archaic features not found in Old Egyptian. It must therefore precede Old Egyptian which, as we shall see is to be identified with the language of the Old Kingdom and the First Intermediate Period. Earliest Egyptian would then be Proto-Dynastic (1st and 2nd Dynasties) or, as some have thought (e.g. Sethe and Breasted), Predynastic belonging to the period before the unification of Upper and Lower Egypt. However as Erman pointed out long ago (1891), the Pyramid texts already portray the king as wearing the double crown of Upper and Lower Egypt. There is, however, as Erman also indicated, the pos-

282

Joseph H. Greenberg

sibility that the Pyramid texts are not all of the same age but, as far as I know, this question has never been subject to serious investigation. Manetho tells us that Menes, the first king of a united Egypt came from the Thinite nome in Upper Egypt. Its center was Abydos and it was here that the tombs of the kings of the Proto-Dynastic age were discovered. The religious and perhaps the older capital of the southern kingdom was Nekheb, called Hieraconpolis by the Greeks and the center of worship of Horus who was always identified with kingship in Egyptian religion. Whether Abydos, or Nekheb we have, in either case, to do with an Upper Egyptian dialect. In a subsequent section I shall adduce certain further linguistic evidence for this identification. By Old Kingdom times the capital had shifted to the more strategically placed Memphis on the southeastern border of the Delta and from which the Delta could be more easily controlled. Edel (1955,11), the author of the standard grammar of Old Egyptian, commits himself, though somewhat cautiously, to the view that Old Egyptian was based on a dialect of Lower Egypt. He states that the position of Memphis as capital of the Old Kingdom and the existence of the Pyramids and of most burial sites of officials in the necropolises in the vicinity of Memphis shows that a Lower Egyptian dialect played an important role in the formation of Old Egyptian. Similarly Edgerton (1951,11-12) states that Standard Old Egyptian was perhaps based on the local speech of the region of the Pyramids. The degree of linguistic similarity between Old and Middle Egyptian suggests that the latter was the continuation of the former even though the political center of the Middle Kingdom was Thebes in Upper Egypt, a dialect which only much later at the time of Akhenaten attained general literary currency. The foregoing opinion is the one expressed by Callender in his grammar of Middle Egyptian (1975,1) in which he asserts that Middle Egyptian texts are written in a language closely related to that of the Old Kingdom adorned with inherited linguistic material from Old Egyptian. He voices his suspicion that Middle Egyptian was actually the spoken language of the late Old Kingdom and of the First Intermediate Period. It may be noted that Thebes was, as it were, an upstart city whose god Ammon was identified with the more prestigious Re in the compound form Ammon-Re. Re was the god of Heliopolis who played a special role in the rise of the 5th Dynasty of the Old Kingdom. Linguistically New Egyptian may be considered the critical point in the long development of Egyptian in that it marks what is probably

Were there Egyptian

Koines?

283

the most thorough change in its long history (Paull, 1975 citing Korostovtsev's views). It is generally agreed that its use as a spoken language long precedes its first introduction into general, including monumental use by Akhenaten. In the Carnarvon tablet Kamose of the 17th Dynasty, whose reign ended in 1567 at least two hundred years earlier than Akhenaten, talks to his advisers about the campaign against the Hyksos, as Erman observes (1968, 2) in the "purest New Egyptian". Sometimes workmen are represented as speaking New Egyptian early in the New Kingdom but if Kamose speaks to his closest advisers in New Egyptian he could have only spoken Classical Middle Egyptian to the gods! A considerable body of Egyptological opinion believes that in fact New Egyptian was already the ordinary spoken language of the Middle Kingdom e.g. Vergote (1973, II.9) who cites Strieker (1945) approvingly to this effect. New Egyptian would then be a dialect based on that of Thebes in Upper Egypt which was the capital during both the Middle and New Kingdom but would not have attained general literary currency until late in the 18th Dynasty. In fact it never completely displaced Middle Egyptian which continued to be used even into the Roman period. The view presented here is already found in Grapow (1944, 211-212) who called the triumph of New Egyptian the breaking out of the folk language of Thebes in Upper Egypt through the crust of the old dominant high language (Hochsprache) of Memphis, without however completely eliminating the latter. The next major change in the standard literary form of Egyptian takes us once more to Lower Egypt. After the Third Intermediate Period a new kind of writing, Demotic, involves with it a new stage of the Egyptian language. Piankhi, the Kushite (i. e. Nubian) king of the 25th Dynasty established his capital at Memphis and the Saites of the 26th Dynasty had their capital in the Delta. It is generally believed that the Demotic script was invented in Lower Egypt (Lloyd 1983,333). Coptic is to be viewed simply as the linguistic continuation of Demotic in a different form of writing. This is contrary to Sethe's views but others disagree with him. Demotic documents can often be identified as reflecting particular Coptic dialects and Lexa (1934, 162) maintained that such dialect differences could be traced as far back as the middle Ptolemaic period during the third century before the present era.7 The period before this would be one in which the original Lower Egyptian koine had not yet developed sufficient differences to be observed in Demotic. Given the nature of Demotic writing it would not be surprising if they existed but could not be detected.

284

Joseph H. Greenberg

We may summarize the foregoing review of Egyptian linguistic history as follows. Four linguistic periods are posited. I. Earliest Egyptian, whose main records are the Pyramid texts, is based on an Upper Egyptian dialect and is coincident with the first two dynasties (ca. 3100-2686). II. Old and Middle Egyptian with its linguistic basis in Lower Egypt, most probably with the dialect of Memphis playing the major role in its formation (Old Kingdom, First Intermediate Period, 2686-2133). III. New Egyptian, based mainly on the dialect of Thebes in Upper Egypt, dominant as the spoken language already in the Middle Kingdom and probably lasting through the New Kingdom and the Third Intermediate Period until the 25th (Nubian) Dynasty or the Saite Restoration (2133-ca. 750). IV. The language, of Lower Egyptian origin, which underlies written Demotic and continues as Coptic until Egyptian ceased to be a spoken language. Although there is no direct evidence for this, we conjecture from what we know of linguistic history in general that each of these was a koine rather than a local dialect which spread in pure form without any characteristics drawn from other dialects. If the foregoing hypotheses are valid, then I and III would both be Upper Egyptian in origin, while II and IV would be Lower Egyptian. There would then be three instances of collateral rather than direct descent, the replacement if I by II, of II by III and III by IV. We could look for the types of reversals which were discussed earlier in the sequences I-II-III and II-III-IV. These would show themselves in the recurrence in period III (New Egyptian) of some features of I (Earliest Egyptian) and missing in the intervening period of Old and Middle Egyptian. Similarly we might expect resemblances between II and IV not found in III. The first set of recurrences (I and III) would be Upper Egyptian and the second (II and IV) Lower Egyptian. In fact, and without any reference to an overall scheme of the development of Egyptian such as that presented here, Edgerton (1951) notes a number of linguistic peculiarities which occur in what he calls Old Egyptian, but which, it is clear from the context, refer to the language of the Pyramid texts, and recur in New Egyptian without appearing in the intermediate period of Old and Middle Egyptian (II). Edel (1954, 12) both modifies and amplifies Edgerton's examples. Edel's version is cited with approval by Vergote (1973, II. 9). As Edel puts it, these phenomena could be explained by assuming that the colloquial language of Upper Egypt retained some elements of the old language of the Pyramid texts which then resurfaced in New Egyptian. The two most prominent of these features are grammatical, the occur-

Were there Egyptian Koines?

285

rence of a prosthetic ? in certain verbal forms and the distinction of gender and number in the demonstrative functioning as a copula which is replaced in period II by an unchanging masculine singular. Edel (1954, 11) notes the variant forms of the word for 'master, lord' Pyramid texts nbw as well as nb (w is often omitted in spelling) as against the exclusive writing nb in this extremely common word in Old and Middle Egyptian. This contrast seems to be reflected in the Coptic period in which the reflex of nbw appears in Upper Egypt while that of nb is seen in Bohairic, the dialect of the Delta. Parallel differences in other words with regard to final w in noun forms are cited in other instances (ibid. 96). In fact, reconstructions based on a comparison of the non-Bohairic dialects sometimes result in forms incompatible with Bohairic suggesting the retention of earlier dialect features in the latter. The lapse of time is of course enormous and suggests caution. The schema presented here may, with the application of the principles of dialect geography, assist in solving what has by writers concerned with the phonological history of Egyptian been considered a strange phenomenon of "reversal". An earlier h (a back voiceless fricative whose exact phonetic nature is uncertain) while remaining in certain words, is replaced in others by sin the Pyramid texts but then is replaced again by the earlier h in Old and Middle Egyptian. There are two roots, however, which continue to written with s, smm 'to be hot' and ss 'scribe' and with different vocalization 'to write'. That the s in these two roots is merely graphic is shown by the fact that they reemerge as x in Bohairic and Achmimic and as h in the other forms of Coptic. The apparent reversal h>s>h is called remarkable by Erman (1928,46); Albright calls it 'the elusive h problem' while Edel seeks to eliminate it as a problem altogether by calling it purely graphic (1955, 52). In regard to this change it is important to take into account, as has been noted generally by students of the history of the Egyptian sound system, that the partial shift of h to s while h remains in other words is paralleled by the partial shifts of k to 1 and g to d- The t and d were probably palatalized velars which became palatalized alveolars later and then became t and d in most instances during the late Old Kingdom. The conditioning factor for the fronting of A: and g was in all probability vocalic and, as we have seen, vowels are not indicated in hieroglyphic writing. A striking indication of the vocalic factor is the second person singular masculine possessive and verb subject suffix k as

286

Joseph H. Greenberg

against the feminine I which is found in the Earliest Egyptian and continues into later periods. These forms agree exactly with the second person singular masculine-feminine contrast in Semitic ka versus ki with similar forms in Chadic (e. g. Hausa). Given the parallelism of the changes in the three consonants, it is clear that we have here a general sound shift like those stated in Grimm's First and Second laws in Germanic although involving different phonetic features. In such instances we may find, as with Grimm's second sound shift that the various sounds do not change simultaneously and that the entire shift originates in one region and is propagated to others. It is in the region of origin that the shift is carried out most completely. For example, as part of Grimm's second sound shift t> ts, p>pf and k> kx. The phonetic completeness and the order of their origin and spread are stated here with the most complete first. The shift started in the south so that, for example Swiss German has kx where standard German has k. None of these changes spread to the northern, Low German, part of the German linguistic area. Now if the modern standard literary language had earlier been Swiss German and then were replaced by the more northern dialects of High German which underlie the present literary language we would have an apparent reversal from pre-Grimm k to Swiss kx and back to more northern k. Such a reversal would not occur with regard to the change from t to is which had spread to all of High German. If we hypothesize that the Egyptian palatalization affected h later than k and g and that, in accordance with the general thesis presented here, the Earliest Egyptian of the Pyramid texts which shows the h > s shift was based on an Upper Egyptian dialect from which the palatalization spread north, we can explain the restoration of the h, which did not affect the words that had an original s and therefore involves the unscrambling of an earlier merger, by the development of a new standard language based mainly on Memphis in the North to which the change h > s had never spread. Finally, we ask whether the Upper Egyptian Proto-Dynastic language was the earliest form of Egyptian concerning which we have any evidence. That such earlier forms would be spoken in Egypt is stated by Butzer (1976,11) who asserts that there is no reason to doubt that the Predynastic peoples of the Egyptian Nile valley spoke Egyptian. He bases this conclusion both on the linguistic separateness of Egyptian within Afroasiatic and the cultural continuity revealed by the archeological record.

Were there Egyptian Koines?

287

The Egyptian writing system itself at the earliest time that we find it gives evidence of a fairly long previous development. A blank piece of papyrus has been found in a tomb of the Second Dynasty showing that even at that time hieratic must have existed as a writing system alongside of hieroglyphic. James (1979,464) argues that this is an indication that long texts could have been written as early as the initial part of the First Dynasty. The most powerful indications, however, derive from the earliest hieroglyphic writing itself. Certain symbols have phonetic values which are different from the ideas they represent as expressed even in the language of the Pyramid texts. That these phonetic readings actually expressed the sounds of words that existed in Predynastic Egyptian is shown by the fact that several of them have cognates in other branches of Afroasiatic. Among these are d 'hand' (Gardiner 1957, D 46) with which we may compare Akkadian idu, Arabic and Hebrew yad etc., whereas throughout the historic period the word for hand was dr.t which survives with the usual phonetic changes into Coptic. Another is Gardiner F21 'ear of ox' with the phonetic value Idn cognate with Arabic ?dn, Hebrew ?ozen etc. all meaning 'ear', whereas the historic word is an obvious new derivative formation m.sdr literally 'place of lieing down or sleeping'.9 To these and others cited in Vycichl (1934) we may add the symbol for 'foot' (Gardiner D 58) to indicate the sound b, a probable cognate to the root b V 'come' or 'go' found in all other branches of Afroasiatic (Greenberg 1963, p. 54, no. 20 'to come'). In historic times the Egyptian word for 'foot' is rd. Assuming on the basis of the foregoing indications the existence of an earlier form of Egyptian we may ask where it was spoken. Both linguistic and non-linguistic arguments suggest that the earliest form of Egyptian stems from the Delta and that it was here that the system of writing was invented. Because of the 35 feet or so of alluvium deposited in the last 6,000 years or so it is not surprising that there are no surviving Predynastic records from the Delta proper. Because of the relatively advanced nature of the latest Predynastic culture, the Gerzean or Naqada II as it has more recently been called, it is widely assumed that the Delta was more advanced culturally than Upper Egypt in the immediate prehistoric period; some even propose (e.g. Sethe) that before the unification of the country from the south by Menes there was an earlier conquest of Upper Egypt by Lower Egypt. A further piece of linguistic evidence, already cited in Vycichl (1934) points to Lower Egypt as the region in which the writing system was invented. In Egyptian hieroglyphs there is no symbol for I. Where

288

Joseph H. Greenberg

in the Pyramid texts and in all later forms of hieroglyphic we would expect I from Afroasiatic cognates we get rin some words and n in others. It is difficult to see why the inventors of the script would not have a symbol for I if it existed in their speech. The writing system was evidently designed by speakers of a dialect in which / had become r in some instances and n in the others. The fact that /survived in some dialects is shown by the fact that a sign was invented for it in the Demotic of the Ptolemaic period by adding a short diacritic stroke to the rand by its emergence in Coptic which used the Greek alphabet which had a sign for /. Among the words which had I in Demotic and Coptic are some which have Afroasiatic cognates with I, e.g. the word for 'tongue' Sahidic las related to forms in all other branches of Afroasiatic except Cushitic. Vycychl already pointed out that the very name of the old religious capital of Upper Egypt Nekheb, spelled nxb in hieroglyphic Egyptian actually had an initial / as can be seen from the name of the present village which is on the same site, Al-kab in which the initial consonant has been reinterpreted as part of the Arabic definite article. Hence the dialect without / in which the writing system originated would be in Lower Egypt. There is a further indication. The word spelled bnr 'outside' first occurs in the New Egyptian of the 19th Dynasty which, as was seen earlier is based on the Upper Egyptian dialect of Thebes. That the sequence nr was an attempt to spell I by using the two most similar consonants to / is shown by the occurrence of bal 'outside' in Sahidic Coptic with corresponding forms in other Coptic dialects. In the light of the foregoing evidence I consider it reasonable to posit a Predynastic form of Egyptian spoken in and near the Delta which was replaced in the Protodynastic period by a koine based on the speech of Upper Egypt. The difficulties and uncertainties of the problems with which this paper has been concerned are obvious. It is presented simply as a set of working hypotheses in need of further corroboration or even major correction in the light of future investigation.

Footnotes 1 The Greek grammarians also use the word koine in these two senses. For references see Liddell and Scott (1925-1946) s.v. koinds. 2 All dates given in this paper are in accordance with those in the Cambridge Ancient History (1970-1982).

Were there Egyptian Koines?

289

3 For a detailed account of Coptic after the Arab conquest see Wiet (1927). 4 Hieroglyphics are transcribed in the usual manner, with the consonants only, because the vowels are not indicated, and with a dot to denote the boundary between morphemes in the word. 5 For a discussion of the various proposals concerning the classification of Coptic dialects see Vergote (1973 1.88-89, II. 8-11) which includes maps. 6 Egyptologists are in disagreement concerning the identity of the figure on the Narmer palette. For a discussion see Emery (1961,14). 7 Johnson (1972) which might add to this discussion is an unpublished dissertation of the University of Chicago which was not available to me. 8 In order to avoid the problem of representing hieroglyphics in text I have referred to them in accordance with the sign list in the appendix of Gardiner (1957) and with his accompanying description. 9 The consonant transcribed as ? is probably a semivowel and corresponds etymologically to both y and the glottal stop of Semitic.

References cited Albright, William F. 1946. Review of "Phonétique Historique de l'égyptien" by J. Vergote, Journal of the American Oriental Society 66:316-320. Breasted, James Henry. 1908. A History of the ancient Egyptians. New York: Scribner. Brockelmann, Carl. 1961. Grundriß der vergleichenden Grammatik der semitischen Sprachen. Hildesheim: Olms. Butzer, Karl W. 1976. Early Hydraulic Civilization in Egypt. Chicago: Chicago University Press. Callender, John B. 1975. Middle Egyptian. Malibu: Undena. Cambridge Ancient History. 1970-1982. 3 vols. 2nd and 3rd Editions. Cambridge: Cambridge University Press. Edel, Elmar. 1955. Altägyptische Grammatik. Rome: Pontificium Institutum Biblicum. Edgerton, William F. 1951. Early Egyptian dialect interrelationships. Bulletin of the American School of Oriental Research, Nr. 122,9-12. Emery, Walter B. 1961. Archaic Egypt. Edinburgh: Penguin Books. Erman, Adolf. 1891. Zur Erklärung der Pyramidentexte, Zeitschrift für ägyptische Sprache und Altertumskunde, 29: 39-44. - . 1928. Ägyptische Grammatik. 4th revised edition. Berlin: Von Reuther and Reichard. 1968. Neuägyptische Grammatik. 2nd edition revised by W.Erichsen. Hildesheim: Olms. Erman, Adolf and Hermann Grapow. 1931-1959. Wörterbuch der ägyptischen Sprache. Berlin: Akademie Verlag. Ferguson, Charles A. 1959 (a). The Arabic koine, Language 35:616-630. - . 1959 (b). Diglossia, Word 15:325-340. Gardiner, Alan. 1957. Egyptian Grammar. 3rd. revised edition. London: Oxford University Press.

290

Joseph H. Gremberg

Grapow, Hermann. 1938. Vom Hieroglyphischen-demotischen zum Koptischen. Sitzungsberichte der Preußischen Akademie der Wissenschaften, Berlin. Phil. Hist. Klasse 322-349. - . 1944. Ägyptisch. Vom Lebensverlauf einer afrikanischen Sprache, in Hans H.Schaeder ed. Der Orient in deutscher Forschung, 205-216. Leipzig: Otto Harrassowitz. Greenberg, Joseph H. 1952. The Afroasiatic Present, Journal of the American Oriental Society 72:1 -9. - . 1963. The Languages of Africa. The Hague: Mouton. Gunn, Butiscombe. 1924. Studies in Egyptian Syntax. Paris: Geuthner. James, Thomas G. H. 1979. Egypt, History of, in Macropedia 6:463-465. Chicago: Encyclopedia Britannica. Johnson, Janet H. 1972. Demotic verbs and dialects. Ph. D. Diss. University of Chicago. Lexa, Frantisek. 1934. Les dialectes dans la langue demotique, Archiv orientälni6:161-172. Liddel, Henry G. and Robert Scott. 1925-1946. A Greek-English Dictionary Oxford: Oxford University Press. Lloyd, Alan B. 1983. The Late Period in Trigger et. al. 1983,279-348. Moulton, James H. 1966. A Grammar of New Testament Greek. Edinburgh: T. and T. Clark. Pauli, C. H. S. 1975. Review of M. Korostovtsev, Grammaire du neoegyptien, Journal of Egyptian Archeology 61:285-286. Sethe, Kurt. 1925. Das Verhältnis zwischen Demotisch und Koptisch und seine Lehren für die Geschichte der ägyptischen Sprache. Zeitschrift der deutschen morgenländischen Gesellschaft 19 :290-316. Spiegelberg, Wilhelm. 1925. Demotische Grammatik. Heidelberg: Winter. Stricker, Bruno Hugo. 1945. De indeeling der Egyptische taalgeschiedenis. Leiden: Brill. Trigger, Bruce et. al. 1983. Ancient Egypt, a social history. Cambridge: Cambridge University Press. Vergote, J. 1973. Grammaire Copte, 2 vols. Louvain: Peeters. Vycichl, Werner. 1934. Hausa und Ägyptisch. Mitteilungen des Seminars fiir Orientalische Sprachen, Berlin. 37, III: 36-116. Wiet, G. 1927. art. Qibt, Encyclopedia of Islam, ed. Houtsma, M.T. et. al. Leyden: Brill.

Lexical Retention in Ethio-Semitic: Checking up on a Myth M. Lionel Bender Southern Illinois University Carbonéale, Illinois

1.

2.

Ethiopia has an abundance of myths concerning language, most of them probably false. One of these is that Giiz is the ancestor of contemporary Ethio-Semitic languages. Hetzron 1972 has shown that this is not credible. In fact, Giiz is a sister-language to Tigre and Tigrinya in North Ethio-Semitic, while South Ethio-Semitic is the second major branch. A second is that Giiz is still spoken in remote northern Eritrea (which means that Tigre is taken to be latter-day Giiz) or that on the contrary, Tigrinya is the "closest language to Giiz". Hetzron shows that the historical question has not been answered as to whether either Tigrinya or Tigre is closer to Giiz, though he considers Tigrinya more likely. As a descriptive question, "no methodical comparative work of this kind has been carried out." I propose to try such work on the basic lexicon of these languages plus representatives of all other branches of the Ethio-Semitic family. My method is as follows: a basic 100-item list and Leslau 1963 are the starting points. The languages are: Giiz, Tigre, Tigrinya (all North), Amharic, Harari, Silti (all Transversal South), Gafat, Soddo and Muxir (latter two "North Gurage"), Mesqan, Cheha, Innemor (all West Gurage). These represent all major branches of Hetzron's family tree (1972:119). The 100-item list was revised as follows: "feather" and all grammatical items ("I, this, thou, we, what?, who?, ye") were omitted, and several items were divided to represent alternative roots of interest ("cut" and "split", two roots for "fly", two for "give", night" and "become evening", "skin" n. and "skin" vb., two roots

292

3.

M. Lionel Bender

for "sun", "tongue" and "lick", "tooth" and "molar", "tree" and "wood"). This yields 101 items in all. Where the Harari dictionary did not include a basic item, other sources were used, and in all cases other sources were consulted as a check on the dictionary, though few discrepancies were found. The list is given in Table 1, together with Giiz roots and indication of cognates in other languages. Appleyard 1977 was a very useful supplement to Leslau 1963 here. The results are four-fold.

3.1 An attempt was m a d e to see to what extent Giiz roots are found in

other Semitic outside of Ethio-Semitic ("Other Semitic" column of Table 1). 74 such were identified, with six more doubtful cases, leaving 21 negatives. This means that Giiz shares about 77% of its lexicon with other Semitic (Table 1). Of the 21 negative cases, seven seem to be fair isoglosses for Ethio-Semitic as a whole (found in most or all major branches): burn ndd, cloud cfammsna, egg ?ank'ok'Ro, fish (asa, many bzx, person sabi?, warm mwk'. Table 1. Comparative Ethio-Semitic Lexicon: Summary No.

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17a 17b 18 19 20

Gloss

all ashes bark belly big bird bite black blood bone breast burn claw cloud cold come cut split die dog drink

Giiz Root kullu fiamad vb. k'rf ksrs abiy anf 'afiadu lela sabP znm k'yfi

Found in:

X

X

X

X

-

X

X

X

-

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

?

X

X

X

X

X

X X

X

X

X

X

X

X

X

X

-

X

-

X

X

X

X X

X

X

X

X

X

X X

X

X

X X

X

X

X

X

X

-

X

X

X

-

-

X

X

X

-

-

X

X

X

X X

X

X

X

X

X

X

X

-

X

X

X

X X X

X

?

294

M. Lionel Bender

No.

65 66 67 68 69 70 71 72 a 72b 73 74 75 76 77 78 79 80a 80b 81 82 83 86 87 a 87 b 88 a 88 b 89 a 89b 90 91 92 94 96 98 99

Gloss

road root sand say see seed sit skin vb. skin sleep small smoke snake stand star stone sunt sun2 swim tail thin three lick tongue tooth molar tree wood two warm water wet white woman work

Giiz Root finot sirw fios'a bhl r>y zr> nbr maHs t'bfi -nhy®® n^us tan kaysi k'wm k w ak 3 b ^bin s'ahay >er -wny zanab -k't'n sis Iiis lisan sin dirs J om i is'3 kiPe mwk' may rt'b ns'fi Wisit -fiby

Other

Sem Te Ty Am Hr Sd SI Mx Ms Ch In G f X

X

?

X

a occurs as auxiliary b "break off" h "hail"

X

X

X

X

x

X

X

X

X

X

X

X

X

-

X

X

X

-

X

s

X

X

X

X

X

X X

X

a

a

X

X

X

X

X

X

X

a X

X

X

X

X

X

X

X

-

X

X

X

X -

-

X

X

X

X

X X

X

X

-

X

-

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X X

X

X

X

X

X

X

X

X

X X

X

?

X

-

X

X

X X X

X

X

X

?

X

X

X

X

X

X

-

X

X

X

-

X

-

X

X

X

X X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

X

-

X

X

X

X

X

X

X

X

X

-

X

X

X

-

-

X

-

X

X

X

X

X

X

X

X

X X

X X

X

X X

X

x negative ? doubtful - data lacking otherwise (blank) indicates positive

Abbreviations:

X

s

Notes

Key:

Found in:

-

X

X

-

-

s found in Soqotri only s "show" v "vanquish"

Symbols:

* = "fight" ** from Cushitic # = "belly" # # ="moon" ® possible loan ®® = "recover"

-

295

Lexical Retention in Ethio-Semitic

3.2 Tigre and Tigrinya are nearly the same with respect to their sharing roots with Giiz (Table 2: Tigrinya 80%, Tigre 77%). The difference is surely not statistically significant. By Black 1973: F2, a difference of 9 points is required to achieve a confidence level of 20% that the difference is significant. Thus, the myth can be laid to rest: Tigre and Tigrinya are indistinguishable as far as synchronic lexical resemblances to Giiz is concerned. Table 2. Results

Giiz vs. Other Sem. Tigre Tigrinya Amharic Harari Soddo Silti Muxir* Mesqan Cheha Innemor Gafat

pos.

doubt.

neg.

missing

ratio

%

71 77 81 73 76 71 59 61 59 58 58 55

7 0 0 0 0 0 0 0 1 1 0 1

23 23 20 28 25 29 38 38 40 41 42 37

0 1 0 0 0 1 4 2 1 1 1 8

74.5/101 77/100 81/101 73/101 76/101 71/100 59/97 61/99 59.5/100 58.5/100 58/100 55.5/101

74 77 80 72 75 71 61 62 60 59 58 60

* Since a disproportionately large number of items (21) were not available for Muxir, those which correspond to negatives in the closely related languages Mesqan-Cheha-Innemor are counted as negative for Muxir also (15 items) and those found to be positive in Mesqan-Cheha-Innemor are counted as positive (4 items).

Incidentally, the same answer is indicated for phonology. Hetzron and Bender 1976: 31 shows that they agree and disagree in the same way in their consonant inventories, except that Tigre has an extra trilled r phoneme and lacks labiovelars k", k'w, g*. For vowels, one can refer to Ullendorff 1955: 179,188, which states that Tigrinya has "diphthongized" the vowel e as compared to Giiz and that Tigre has not. Again Tigrinya seems to have a slight edge, but it is of doubtful significance. In grammar, Hetzron favors Tigrinya, as noted above, and Ullendorff (1955:227) considers the question a "terminological quibble", though indicating a tendency of Tigre to preserve archaisms. A careful large-scale investigation would be welcome. 3.3 It is interesting to note that a loss of20-23% of common lexicon between Tigre-Tigrinya and Giiz is close to what one would expect

296

M. Lionel Bender

under the assumptions that our records of Giiz represent a sisterlanguage last spoken about 1000 years ago. At that time GiizTigre-Tigrinya was probably a dialect cluster, and it will thus introduce little distortion to assume that this cluster was directly ancestral to Tigre and Tigrinya of today. Since the glottochronological rate of lexical loss is about 20% per millennium, this shows a rate of change like that of "normal" change in Indoeuropean (mainly) as measured in the original work on glottochronology by Swadesh. However, an important note of caution is needed here. The present work departs from the requirements of glottochronology in two ways: use of a non-standard list and "searching for cognates" by allowing semantic shifts. It is not clear what overall effect the changes in list items will have, but it should be small since the changes involve largely using other items of the original 200-item list while keeping the same balance of semantic categories (body parts, verbs, etc., but purposely omitting grammatical forms). The effect of "searching for cognates" is to increase percentages (cf. my figures of Bender 1968: 2 for Giiz-Tigre 70%, Giiz-Tigrinya 71%, Giiz-Amharic 67%), but this is offset somewhat by using two roots for some items, since these tend to be in complementary distribution (i. e., if a language has one root it will generally lack the other). The 1968 figures of 70-71% would lead to a time-depth of about 1500 years for Tigrinya-Tigre of today as against Giiz. It should be safe to say then that the range is 1050 to 1500 years, based on judicious interpolation in the glottochronology graph given by Gleason 1955: 88. 3.4 The results for all of Ethio-Semitic are:

Tigre 77 Tigrinya 80

Amharic 72 Harari 75 Soddo 71

Siiti Muxir Mesqan Cheha Innemor Gafat

61 62 60 59 58 60

Ave.

Ave.

Ave.

60

78.5

72.5

Thus, Transversal South Ethio-Semitic is probably significantly below Tigre-Tigrinya in its lexical relationship to Giiz, and Outer South is certainly significantly behind Transversal. It is

297

Lexical Retention in Ethio-Semitic

striking that all three groups act as units with no more than 4% differences in their group-internal results. Projecting guesses at time-depths is not easy, since one is dealing with languages which are not contemporary (present-day Ethio-Semitic against 1000-year-old Giiz), and one cannot assume that Ethio-Semitic languages other than Tigre and Tigrinya were close enough to attested Giiz to form a dialect cluster in the past. If we did assume that Transversal Ethio-Semitic were in such a position, an average of about 72.5% separation would date to about 1400 years, whereas if we assumed that Giiz were contemporary with Transversal Ethio-Semitic today, it would indicate about 700 years. Since the North Ethio-Semitic separation from Giiz is at least 1050 years, one must assume that Transversal Ethio-Semitic is closer to the higher figure of 1400, say 1200 years. In the case of Outer Ethio-Semitic, the average of 60% would lead to guesses of about 1150 to 2300 years on the two assumptions (Giiz a contemporary, Giiz an ancestor). A good guess might be 1800 years. Since the percentages are all on the high side because of "cognate searching", the glottochronological results must be on the low side. Referring to the figures of Bender 1971:173 I find the North Ethio-Semitic average 69.5%, Transversal Ethio-Semitic 56.5%, Outer Ethio-Semitic average 50%. I now consider these figures to have been based on too conservative a cognation procedure and lower than the new figures of this paper by an average of nearly 12%, so that an average of 6% reduction in my present percentages would seem a fair compromise. Thus we arrive at the following, which shows the results based on the new compromise percentages, high and low time-depths based on the assumptions of Giiz being ancestral or contemporary, and a compromise time-depth closer to the higher figures.

ComproAncestral mise Assumption Percentages Giiz vs. North E-S Transversal E-S Outer E-S

72 66 54

1350 1900 2900

Years: Contemporary Assumption

Compromise

675 850 1450

1100 1500 2300

298

M. Lionel Bender

The new results put the separation of the North Ethio-Semitic languages from Giiz at about the year 800 A.D., separation of Transversal at about 400 A. D., and separation of Outer at about 400 B.C. 3.5 Historical Speculations. These dates fit in with the presumed prehistory of Ethiopia as follows. Accepting Hetzron's primary division of Ethio-Semitic into North and South, based on morphological isoglosses, one is faced with the problem of how to reconcile the much closer lexical relationship between Transversal and North than Outer and North. Hetzron's account (1972:122-125) of putative migrations based on his isoglosses has a common migration southward from Axum followed by a "vanguard" south group soon moving out and settling furthest south, with a "transversal" group later heading southeast and maintaining a constant contact with the northern civilization. It is this differential in continued contact with the north which accounts for greater lexical (and grammatical) similarity between North and Transversal than with Outer South. Note, however, that the linguistic data does not determine directionality. The process could have been the reverse, with Outer South Ethio-Semitic staying closest to home and North representing an early migration, as proposed by Hudson 1977. In either case, the data agree with both Hetzron's and Hudson's judgments of greater time-depth than could be accounted for by the classical account of South Arabian migrations to Axum, with the presentday differentiation resulting in only the thousand years since the breakup of the Axum Empire (e.g., Ullendorff 1955:225-231).

References Appleyard, David, 1977: "A Comparative Approach to the Amharic Lexicon", Afroasiatic Linguistics 5.2:43-109. Bender, M. Lionel, 1968: "Remarks on Glottochronology of Northern Ethiopian Semitic Languages", Journal of Ethiopian Studies 6.1:1-11. -. 1971: "The Languages of Ethiopia: A New Lexicostatistic Classification and Some Problems of Diffusion", Anthropological Linguistics 13.5. Black, Paul, 1973: "Multidimensional Scaling Applied to Linguistic Relationships", Bell Laboratories, Murray Hill. Gleason, Heniy A., 1955: Workbook in Descriptive Linguistics, Holt, Rinehart, and Winston.

Lexical Retention in Ethio-Semitic

299

Hetzron, Robert, 1972: Ethiopian Semitic: Studies in Classification (Journal of Semitic Studies Monograph No. 2), Manchester University Press. -. and M.Lionel Bender, 1976: "The Ethio-Semitic Languages", pp.23-33 in M. L. Bender, J. D. Bowen, R. L.Cooper, C. A. Ferguson (Eds.): Language in Ethiopia, Oxford University Press. Hudson, Grover, 1977: "Language Classification and the Semitic Prehistory of Ethiopia", Folia Orientalia 17:119-166. Leslau, Wolf, 1963: Etymological Dictionary of Harari, (University of California Publications, Near Eastern Series 1), Berkeley and Los Angeles. Ullendorff, Edward, 1955: The Semitic Languages of Ethiopia: A Comparative Phonology, Taylor's (Foreign) Press, London.

The High Central Vowel in Amharic: New Approaches to an Old Problem Dick Hayward University of London London

The Ethiopic syllabary consists of a matrix in which each row presents a distinct consonant and each column presents a distinct vowel. The symbol appearing in a given cell of the matrix is a composite consisting of a consonant-depicting element and a vowel-depicting element built into a single graph1. Symbols of the sixth column or 'order', however, are unique in having a dual function in writing, for on some occasions they represent a consonant plus what is usually pronounced as a high central vowel (approximately IPA [i]), while on others they represent a simple consonant with no following vowel. That the appearances of the sixth order vowel are subject to a high degree of predictability has not passed unnoticed by linguists, and a full article has been devoted to this topic (see Hetzron 19642). In general it has been suggested that the sixth order vowel (hereafter /) is usually to be regarded as epenthetically introduced. While I am wholly in sympathy with this interpretation I do not think that certain issues connected with this subject have been addressed with sufficient rigour. The conditions of f-epenthesis have not been stated with precision, nor has the motivation for the process been defined clearly. Considerable advances have been made of late in our appreciation of syllable structure and of ways to integrate such essentially phonotactic information into the formalisation of rules expressing phonological processes. It is my purpose in this paper to examine ¿-epenthesis within the context of the syllable phonology of Amharic. Applying a basically autosegmental approach I shall consider the issues outlined above. I shall also explore a further matter relating to ¿epenthesis, namely, why it is that given the facts of Amharic phonology the epenthetic process should involve i rather than some other vowel. Fortu-

302

Dick Hayward

nately for this study Archangeli's recent exposition of underspecification theory makes it possible to pose such a question and to find some satisfying answers. I It can be claimed that in very nearly every case the occurrence of i-epenthesis correlates with syllable structure. More precisely, the process occurs whenever the structure of a word does not permit an exact division into syllables; /-epenthesis is Amharic's stock anwer to unsyllabifiable words. In order to establish what is meant by syllabifiability it will be necessary to propose an analysis of the syllable structure of the language, and section II will be given over to this exercise. In the discussion of syllable phonology, however, I shall assume certain hypotheses about the nature of phonological representations which have been made available by recent work in autosegmental phonology. These assumptions are outlined in the present section. Kahn (1976) has to be seen as the seminal work which reinstated the syllable in phonological theory, but there is a difference which has appeared in most subsequent work. This involves the representation of certain aspects of phonological substance on a level which could be said to mediate between the syllable and any of its terminal elements. The possibility of representing different components of the phonological composition of an item on multiple planes or tiers is, of course, the fundamental insight of autosegmental phonology. Although not all linguists working within this model agree as to the exact details concerning this mediating tier, there is general agreement that durational features have their expression here, rather than in the segments themselves3 (see Clements and Keyser 1983: 10ff.). We may simply note how that the representations in (1) resolve the central problem of the analysis of phonologically long segments, namely, how it is that an unbroken single articulation can exhibit the durational properties (and very often the distributional properties) of two-segment sequences. (1)

timing tier segmenttier

V

I

V

[ ]

V [ ]

short vowel

long vowel

V

V

V

C

I I V [ ]

C

[ ] [ ] vowel sequence

long consonant

C

C

I I

[ ] [ ] consonant sequence

The High Central Vowel in Arnharic

303

Allowing for an adjustment to be dealt with in a moment such representations are assumed to be correct. Another difference that has marked work in syllable phonology subsequent to Kahn (1976) concerns the recognition of hierarchical constituent structure.internal to the syllable. Although there is considerable variation in the degree to which this internal structure is elaborated 4 , there seems to be a common, albeit not universal, feeling about the need to recognise a 'rhyme' (R) constituent comprising the nuclear vowel (or syllabic consonant) and everything following it. Everything preceding the rhyme belongs to an 'onset' (0) constituent 5 ; see (2)a. Yet more recent work goes somewhat beyond this in proposing that syllables are 'headed' structures. The nuclear 'syllabic' itself is said to constitute that head and all additional elements of the rhyme constituent are seen as projections of the head (pace the familiar notions of X Syntax (Jackendoff 1977)). Within this conceptualisation the onset can be viewed as essentially an adjunct, rather than a dependent element (see Levins 1983:14 ff.). There is something intuitively satisfying about this. However, although heads and their dependants are distinguished structurally, it is not at all obvious that any of the terminology introduced for the purpose of labelling constituents is necessary, and Archangeli (1984) dispenses with them, referring to rhymes simply as 'syllable heads'. In the diagrammatic representation in (2)c. the head is distinguished by an upright line. Elements dependent on the head are linked to the upright line by left-to-right dipping lines, and the onset is similarly linked by a right-to-left dipping line. (2)

Various syllable trees for the Amharic monosyllable li) 'child' a.

( = a)

c.

O

C 1

In the present paper I shall make use of the mode of representation of (2)c. In representing the timing tier in terms of elements of two types, ie. Cs and Vs (as in (1) and (2)), a basic question arises. The question concerns the 'content' of C and V slots. It has never been suggested in the

304

Dick Hayward

literature that the linearly ordered C and V elements are themselves a string of partial distinctive feature matrices in which C and V are a convenient shorthand for [ + cons] and [ — cons] respectively. According to Clements and Kevser (1983: 32): 'Unless otherwise stipulated in the grammar or lexicon of a given language, V-eiements of syllable structure are freely allowed to dominate [ — consonantal] segments, and C-elements are freely allowed to dominate both [ + consonantal] segments and [-I-high, — consonantal] segments. Other associations are possible only when admitted by language specific rules. For example, some languages allow post-vocalic consonants to be dominated by V if they are tautosyllabic with the preceding v o w e l . . . As a further example, some languages allow tautosyllabic VC sequences to dominate single consonant or vowel segments . . . ' This makes it clear that C and V dominate matrices where such features as [ ± cons] would be specified, but that they themselves are featureless. It is evident that they are simply cover symbols. Moreover, the passage quoted makes it clear that the definitions of C and V are not constant, and that under language-specific conditions a good deal of flexibility is tolerated. The main purpose in having both Vs and Cs is to distinguish elements nuclear to the syllable from elements marginal to it. Such a differentiation accomplished on the timing tier makes the feature [ ± syll] a redundancy on the segmental tier. But it can be argued that recognition that a particular element on the timing tier is a syllable head accomplishes essentially the same distinction, so that the differentiation of Cs and Vs is itself redundant 6 . In recent work by Levins (1983) and Archangeli (1984) all Cs and Vs are replaced by strings or 'skeleta' of X slots, that arc explicitly recognized as empty and identical to each other. Levins (1983: 12ff.) and Archangeli (1984: 164) advance convincing, yet distinct, arguments for this change, but it is not necessary to repeat them here; some additional support for this position will emerge in the discussion of syllable weight in Amharic (see below). Although in discussion in the text itself it will often be convenient to make use of ' C and ' V in an informal way, in formal representations of syllable structure X slot skeleta will be adhered to.

II Examination of Amharic monosyllables suggests that there are the following syllable types: VC, VCC, CVC, CVCC and, very rarely, CV. Depending on our analysis of the so-called 'labiovelars' (see below) a fur-

305

The High Central Vowel in Amharic

ther range of types may or may not need to be added. An examination of polysyllables suggests only the one further possibility of a V syllable type in words such as alii 'he said'; but this presupposes adequate criteria for syllabification. Amharic has a 7-term vowel system, viz: (3)

i e

i a a

u o

The distribution of these vowels tends toward a pattern. The central vowels i, a and a commonly occur in syllables closed by two consonants, eg: (4)

t'irs 'tooth', k'irb 'near', silk 'telephone', libs 'clothes', malk 'appearance', madf'a rtillery', ayn 'eye', amd 'pillar', mann 'who?'

Consideration of the final VCC sequences of polysyllables shows many more, eg: (5)

t'ik'imt 'name of a month', siddist 'six', tim(h)irt 'learning', bilamd 'if I get used to', bisadd 'if I send', indik'alt' 'so that it melts', milac'c' 'razor', tamars 'you(f) have learned', bittalf if she passes', binnawk' 'if we know'

The front and back 'peripheral' vowels i, e, o, u are comparatively uncommon in CVCC or VCC syllable types, though they are common enough in syllables closed in a single consonant, eg: (6)

monosyllables: fit 'face', kis 'pocket', t'im 'beard', set 'woman', k'es 'priest', k'ob 'type of cap', t'or 'spear', (i)rob 'Wednesday', boy 'ditch', sum 'chief, (i)ruz 'rice', t'ut 'breast' polysyllables: maggabit 'name of a month', sarawit 'army', ganzaben 'my money', samicces 'I having heard you(f)', gimbot 'name of a month', agarun 'his country (accusative)', sammus 'they heard you(f)'

Amharic has a strict word structure constraint which permits syllables closed by two consonants to occur only in word-final position. Interestingly, CVC and VC syllables in which the nuclear vowels are i, e, o, u also tend to be restricted to word-final position. CVC and VC syllables occurring in other positions generally contain central vowels. It is difficult to escape the impression that there is some sort of'equivalence' between (C)VCC syllables in which V is i, a or a and (C)VC syllables in which V is i, e, o or u.

306

Dick Hayward

A sort of explanation for this twofold patterning of vowels is obvious enough to anyone knowing something about the origin of i, e, o and u, for these are usually explained as the reflexes of long vowels or of short vowels followed by glides7. Therefore, a word such as bet 'house', deriving from earlier *bayt, in which a sequence of two nonsyllabics followed a short vowel, could be said to be structurally similar to a word like sarg 'wedding'. The fact that syllables with i, e,ooxu tend not to take more than one following consonant would be seen as the reflex of an earlier constraint limiting syllables to two Cs finally. Whatever the adequacy of this as an explanation for the origin of the present-day 'equivalence' between syllable types, it does not tell us what that equivalence is synchronically. For modern Amharic one would certainly not want to say that i, e, o, u had to be derived from underlying sequences of a or i plus a glide. At the most obvious level of objection such an approach would oblige us to explain away surface contrasts such as the following. (7)

vowel samt[oj'hz having heard' betocc[u] 'the houses' ayn[e]'my eye' t'igab[i]nna 'eat your fill and . . . '

vowel 4- glide samtfaw]'they having heard' sdmmacc[iw]'she heard him' nfej]'cornel' [ij(i)]nna 'see and

Yet the equivalence has persisted as a tendency in syllable phonotactics, and invites an explanation. A generally satisfying account is, I believe, to interpret the vowels i, e, o and u as long vowels. (This implies, of course, that the historical development of these vowels would often have involved compensatory lengthening.) If vowel length is what is concerned, the equivalence with which we are concerned is simply another instance of something commonly encountered in discussions of metre, accentual placement, etc. I refer to the traditional notion 'heavy syllable', the exponents of which are frequently represented as the disjoint pair CVC and CV. (The maximal syllables of Amharic are, of course, longer in that a further C is involved, ie. CVCC and CVC. These are what have sometimes been called 'super-heavy syllables'. However, nothing is lost for the present discussion by subtracting the common factor of the final C.) To recognize that the phenomenon of equivalence between syllable types is not unique to Amharic does not, of course, explain it. A comparison of the phonological structure of the two syllable types makes it clear that the basis of the equivalence is accounted for in terms of their common two-slot rhyme structure, viz:

307

The High Central Vowel in Amharic

X

X

X

X

[ ] [ ] [ ] C

V

C

X

X

[ ]

[ ]

C

V

X X

[ ] [ ] C

V

The essence of this has sometimes been expressed in recent literature in words to the effect that heavy syllables have branching rhymes, while light syllables do not. It should be noted that it is not simply a matter of counting slots, for if all skeletal slots were of equal value, one would expect differences of syllable weight to correlate with the presence/absence of a slot to the left of the syllable head in the same way that it correlates with the presence/absence of a slot to its right. This is clearly not the case here, nor does it ever seem to be so. It is worth making the point here that compared with the transparency of the explanation of heavy syllable equivalence afforded in (8) an account utilizing a mediating tier distinguishing Cs and Vs is far less perspicuous. In order to express the fact that ' . . . long vowels and diphthongs are functioning equivalently to VC sequences . . . ' Clements and Keyser (1983: 33) suggest that long vowels (and diphthongs) be represented as in (9).

in spite of the fact that this would be at variance with the insightful representation of long vowels noted in (1) above, which requires one element on the segmental tier to be dominated by two V elements on the CV tier8. One finds it difficult to attach much meaning to the terms C and V. It may be necessary to deal with certain objections at this point. Anyone familiar with Amharic (especially in its written form) will probably wish to point out that there are many words containing noncentral vowels either followed by two consonants in word-final position, or followed by one consonant in a nonfinal position. Many of the apparent exceptions fall into two classes: (i) morphologically complex forms of 'hollow' verbs9, such as

308

Dick Hayward

(10) hon+s 'you(f) became', rot+k/h 'you(m) ran', som + k/h 'you(m) placed sb in authority', hed+s 'you(f) went' etc. (11) words in which the syllable onset contains a 'labiovelar' or some other consonant with a labialized offglide, such as (11)

gwfiijnc'

'cheek', Fwfujlf

'lock', Tcw[u]ntai '100 kilos',

w[u]rc'

'frost', w[o]rrada 'he descended', gwfojddala 'it decreased', mw[o]kkara 'he tried', etc.

Concerning 'exceptions like those of (10) it is argued by Taddese Beyene (forthcoming) and Hayward (forthcoming) that all instances of noncentral vowels appearing in 'hollow' verbs derive from sequences of underlying central vowels, so that they have to be considered as a rather special case. The 'exceptions' in (11) are easily disposed of, since the fact that the onset contains a segment with lip-rounding in all cases enables one to claim that the vowels in all these forms result from a process of assimilation to the rounding of the preceding segment (see end of section IV). There are a number of genuine exceptions, eg: (12) k'ent 'elegance', mist 'wife', sost 'three', gint' 'scorpion', bore'

'paunch', tembar 'postage stamp', (also there is the plural formative -occ) but I believe that it would be rash to abandon such generalisation as the preceding analysis affords on account of such a small number of counterexamples. It has been observed that with a short vowel a rhyme may contain up to two nonsyllabics. There appear to be only two systematic constraints governing such clusters. The first disallows any cluster the second member of which is a glide. The second disallows any cluster in which the second member is a sonorant when the first is an obstruent10. These constraints may be most simply expressed in terms of negative syllable structure conditions such as (13)a. and (13)b. (13) X

X ] [ — cons]

X

X

[ — son] [ + son]

X

X

[ + cons] [ + son]

Excepting h all nonsyllabics may be geminated, and many geminates occur as syllable-final clusters. Among these we find geminate y in

The High Central Vowel in Amharic

309

words such as biláyy 'if I separate (tr)' and bitayy 'if I am seen'. It should be observed that as a consequence of the way in which geminates are represented (see (1)) these do not violate (13)a. A few of the permitted clusters are the following (see also (4) and (5))(14) birz 'honey-water', dink' 'wonderful', t'imb 'carrion', dims' 'voice', maz/'anvil', birát'b 'if it gets wet', bisagd 'if I bow in worship', bik'át'f'ií I cheat', binat'k' 'if I snatch', binat'b 'if it drips', bac'd'if I reap', biwak't' 'if I crush' It will be noted that (13)b. permits clusters of consonantal sonorants, eg: (15) (h)ilm 'dream' aln 'we said', indittamr 'so that she appears attractive', bamn 'if I believe', indik'amliso that he removes lice', etc. For some speakers, however, the pronunciation of such items would require the insertion of an epenthetic vowel. For such speakers (13)b. would need to be expanded to (13)c. The analysis of the so-called 'labiovelars' (kw, gw, k'w, hwn) poses a number of problems in Amharic, especially if one attempts to correlate pronounced with written forms (cf. the remarks in Ullendorff 1951:76, 82). For the present purpose the main question concerns the number of segments these sounds represent, ie. whether they are to be analysed as one segment or two. Inspection makes two things clear. Firstly, the labiovelars are phonetic complexes consisting of an obstruent with a labialized offglide. The degree to which labialization marks the obstruent varies, but its effect on a following non-low central vowel can be so marked that, apart from duration factors, i and a come close to the pronunciations of u and o respectively. Before the low vowel a the offglide is maximally distinguished and here one might feel that it is a distinct segment. Secondly, we see that the labiovelars appear only in syllable onsets12. This distributional restriction makes the labiovelars idiosyncratic with respect to all other single segments, but there is ample reason for analysing them in this way. In section III we shall see that Aepenthesis always splits pre-syllable head sequences of two X slots. This never happens in the case of labiovelars, arguing that a single X slot analysis is appropriate. Strong support for this view comes also from the morphology. The various 'prosodic templates' for verb forms never treat a labiovelar as two consonants; thus, gw-d-l'decrease' behaves entirely like its fellow triliteral verb g-i/-/'kill' - never like a quadriliteral such as g-l-b-t'

310

Dick Hayward

'overturn' 13 . The revealing analysis of complex segments made available in autosegmental phonology (see (16)a.) captures the fact that although two distinct phonetic articulations are concerned (each of which requires a separate feature matrix), from a functional point of view there is a single segment. (16)b. shows the representation appropriate for the monosyllable gwad 'comrade' under this analysis. (16)

a.

X [

]

b.

a

[ ]

^ X

X w

g

X a

d

The labiovelars are not an isolated series. Except for the scarcely nativized p, the entire natural class of [-cor] consonants can appear with labialized offglides, though the range of occurrence and number of examples of certain types is very restricted. Thus alongside (17)

tagwaza 'he travelled', gw[o]dda 'he injured', gw[u]rsa 'a mouthfull', irikwan 'even', lcw[o]nndna 'he condemned', Tcw[u]rat 'pride', k'want'a 'dried meat', k'w[o]ssala 'he got wounded', k\v[u]rs 'breakfast'

we have (18)

mwamwa

'it dissolved', 7nw[o]kkara 'he tried', mw[u]ggit

'law

suit', tanfwak'k'ak'a 'he dragged himself tafw[o]nnana 'it got cut (of a nose)', ijfw[u]Mit 'viper', Invambwa 'pipe', Tm[o]zzana 'he became idle', bw[u]din 'team' ¡Pwagme (~ T^wagme) 'the "13th month" of the Ethiopian calendar' Wider possibilities are hinted at by the otherwise unique words t'[w]at 'morning' and t'[w]af'candle'. And these possibilities are realised in a rather full way within certain morphologically complex word structures. When the 2p or 3p subject agreement suffixes in the perfect or imperfect (-u suffixes), or the 3ms subject agreement of the converb (a -o suffix) is followed by any element beginning with a, 'desyllabicization' of the rounded vowel suffix occurs in normal speech, eg: (19)

t+rad+u

+ at + alla + acchu —• tirad[w]atallaccihu

you(p) will

help her k'azk'z + o + all —• k'azk'izfwjall

it has got cold

a + labbas + u + accaw —• alabbas[w]accaw they clothed them samt + o + at+all

—• samt[w]atall

he has heard her

The High Central Vowel in Amharic

311

It is clear that this process enlarges the series of consonants having labialized offglides (hereafter abbreviated to C L O ) , since many of the [ + cor] consonants are involved. Observe too what happens in similar verb forms the stem-final consonants of which are velars, since this would pose a peculiar difficulty f o r any autonomous phonemic account of the labiovelars ( ± rounded labials). (20)

ndkk + u + accaw —• nakk[wjaccaw they touched them as + darrag+u awk'+o

+ at—* asdarrdgfw]at they made her d o it

+ acchu + all —* awk'[w]accih[w]all

amm + o + at + all—• amm[wjalall

he has known you(p)

she feels ill

A commitment to biuniqueness would force one to treat the C L O s resulting from desyllabicization (ie. those of (20)) differently from similar phonetic entities in words such as gwador k'wank'wa 'language'. In terms of syllable structure all the examples of C L O s shown in (17)-(20) are identically analysed, ie. as in (16). A second problem which concerns all C L O s and, of course, w itself (the distinctive feature properties of which all C L O s share) is that following them one cannot find the full range of seven vowels functioning contrastively. The problem is acute in the case of the labiovelars and rounded labials because their [-rd] congeners often acquire labialization before [ + rd] vowels. This results in a situation where there is considerable surface variation. The simplest hypothesis will be to posit a systematic defective distribution for underlying labiovelars, rounded labials and w with respect to the vowels e, i, o and u. This has a surface counterpart in terms of a negative syllable structure condition which holds true for the entire class of surface C L O s , viz:

X

([ D

X

l + rd]

X

[ ]

In fact, (21) is not restrictive enough, nor does it show certain symmetrical phenomena involving palatal glides; but treatment of these things would require too lengthy a digression 14 . It is sufficient for the present purposes to have established that all surface C L O s are exclusively syllable onsets.

312

Dick Hayward

III Syllabification is the division of the elements of the skeletal tier into syllables in conformity with certain universal principles and in accordance with the syllable types of the language concerned. It seems intuitively correct to assume that this starts to take place as soon as a rhyme comes to dominate a particular slot on the skeletal tier. The way in which this is achieved need not concern us here, and for the present we shall simply assume an association between syllable heads and certain X slots. The mechanism for syllabification operates as follows. 1. Co-syllabify all Xs to the left of the syllable head subject to the language-specific constraints governing onsets. 2. Co-syllabify all Xs to the right of the syllable head subject to the language-specific constraints governing rhyme structure. Both 1 and 2 are also subject to language-specific word structure constraints. The application of 1 before 2 is in line with what Clements and Keyser (1983:37) refer to as the 'Onset First Principle'. It ensures a unique syllabification for one or more intervocalic segments in cases where the syllable structure of the language would permit more than one possible analysis. Thus, Amharic words such as t'afac"sweet' and inkwan 'even' are syllabified as t'a.fac' and in.kwan, rather than as *t'af.ac'or *ink.wan. It will be obvious that effective syllabification requires elements on the segmental tier to be associated with X slots. Without this there could be no way whereby Xs could be differentiated for acceptability or unacceptability as onset or as rhyme components. This association is simply assumed for this paper. While it is a fact that all surface forms of words can be exhaustively syllabified, it is not necessarily the case that this condition obtains at more remote levels of phonological representation. On the contrary, there are good reasons to believe that in many cases the possibility of surface syllabification depends upon the operation of i-epenthesis. Consider, for example, the alternations seen in the jussive stem of a verb such as k'-m-s 'taste'. (22) a. fcimas'taste!' b. yik'mas 'let him taste!' The distribution of these two alternants k'imas ~ k'mas is governed entirely by considerations of syllable structure. The alternant k'mas could not occur in (22)a., since, as a word, it would be unsyllabifiable. However, the occurrence of k'mas in (22)b. is perfectly explicable,

313

The High Central Vowel in Amharic

since the preceding 3ms subject agreement yi- renders the word syllabifiable, viz: yik'.mas. But the 3ms subject agreement itself does not always appear with the alternant yi-; in the following forms based on verbs with imperfect stems we see that the 3ms subject agreement has the form y-. (23) a. ayk'amsim 'he will not taste', b. yawk'all 'he knows' In (23)a. the presence of the preceding negative prefix a-, and in (23)b. the fact that the verb a-w-k' 'know' begins with a vowel, guarantee syllabification, viz: ay.k'am.sim, yaw.k'all. There are many formatives like the 3ms subject agreement that have alternants containing /, and there are hundreds of verbs with jussives like k'-m-sthat have alternants with i, so that a grammar would be very inadequate indeed if it failed to show that the presence/absence of these i vowels correlated with the surface syllabifiability of words. It has been claimed that the jussive stem k'mixs cannot appear in its assumed underlying form (ie. as k'mtis) because it cannot be fully syllabified. A generalised epenthesis rule such as (24) would certainly accomplish the desired result, (24) 0

/ ' / # [ + c o n s ] - [+cons]

but it would fail to show that what was going on had anything to do with syllable structure; as it stands, the rule appears to be motivated simply by word structure. This brings us to the notion of extrasyllabicity. Given the assumptions and procedures of the preceding section, it would follow that a string of segments such as k'mas could be partly syllabified. While in (25)a. the partial string 'ma's'does satisfy syllable structure, the initial segment cannot be incorporated in any way. If such 'extrasyllabic' elements are marked in some way so as to show that they have failed to syllabify, any rule which operated to create new syllable structure (a rule such as /-epenthesis, for example) would be able to make explicit reference to that mark, so relating that rule to nonsyllabifiability. In the literature extrasyllabic elements are marked with a prime, as in (25)b. (25)

a.

b.

CT

a

X

X

X

X

X

X

X

X

k'

m

ä

s

k'

m

ä

s

314

Dick Hayward

Moreover, in order to accomodate extrasyllabic elements, what Aepenthesis does is to create the minimal essence of an additional syllable, ie. a head. A characterization of f-epenthesis that would point these things up might appear as (26)

(26)

I 0 —• X / X' — X

As we shall see in section IV (26) contains some redundant information, and the extraction of this will make for a deeper insight into Amharic phonology. Presently, however, we need to note that f-epenthesis does not take place only at the beginning of words. The word b[i]tt[i]sab[i]r 'if she breaks' could be said to contain only one nonepenthetic vowel. If we assume the partially syllabified form of (27), it will be seen that there are four extrasyllabic slots. (The final r cannot be syllabified with the preceding syllable on account of the negative syllable structure condition (13)b.) (27)

ct X

X

b

X t

X

X

X

X

s

à

b

r

(In (27) and similar subsequent diagrams representation of the internal structure of syllabified strings is omitted ; only the extent of syllables is demarcated.) Applying (26) from left to right and allowing syllabification (abbreviated to 'sbn') to apply after each new syllable head is created accounts for only two of the high central vowels, viz: (28) CT CT CT CT

A

I

X X X X X XX

I V

I I I I

b

s à b r

t

A

A

26

I b i

V

I I I I

t

s à b r

CT CT CT CT CT

I A

XXXX'XXXXX'^ 26

I I V

b i

t

A

A

X X X X X X X X -H. X X X X X X X X

I I I II

i s à b r

A

sbn

I I V b i

/I

A

XXXXXXXX

sbn I I \ j b i t

|

Il

I

i s à b

I I I I t

X

I

r

s à b r

315

The High Central Vowel in Amharic

To account for the final [/] we need another rule, for the sequence X X' does not fit the structural description for (26). (29)

| 0 —• X / X — X' i

Applying (29) renders the word completely syllabifiable, but it should be observed that on account of the Onset First Principle, syllabification involves a resyllabification of stem b with the new syllable, viz: (30) a

a

a

A M

a

X X X X X X X X X X A 29

| | V t b i

a

a

a

A /I /I A XXXXXXXXXX

I I I I I I sbn | | \ / i s a b i r b i t

| | | | | I i s a b i r

So far all the instances of f-epenthesis that have been considered could be said to be justified insofar as they permit us to maintain a commitment to morpheme invariance. There are, however, many thousands of occurrences of the high central vowel in monomorphemic words where there are no alternations. Such occurrences belong to two situations: (i) situations where the location of / just where it would be predicted by rule; (ii) situations where / occurs in other locations, and so could be said to have an independent or lexical status. Examples of the first situation - and this I believe to be by far the most frequent one - are such as the following. (31) zfijnab 'rain', k'[i]t'al 'leaf, t[i]ll[i]k' 'big', t[i]m(h)[i]rt 'learning', sakfijm 'load', etc. Examples of the second situation are such as the following. (32) [ijnc'at 'wood' (*n[i]c'at), [i]ne 'I' (*ne), [i]bab 'snake' (*bab), s[i]d[i]b 'insult' (*s[i]db), ak[i]st 'aunt' (*aks[i]t), etc. The key question is whether it is justifiable to posit an underlying form for words such as those exemplified in (31) without high central vowels, ie. as (33) X X X X 1 z

I n

I a

X X X X I b

'rain'

I t

V I

I k' 'big'

316

Dick Hayward

etc., and supply [/] by means of epenthesis rules. It seems to me that this is perfectly justifiable, f-epenthesis rules are exceptionless, so that the alternative to generating surface forms of the type represented in (31) by means of these rules would still require us to have such rules for alternating forms, but (on account of exceptions such as those of (32)) to ignore the fact that thousands of nonalternating morphemes show surface patterns of distribution for [/] which happen to coincide exactly with those in forms generated by such rules. The fact that there is a phonetic identity between the products of epenthesis and the realizations of lexical i is, of course, irrelevant. There are three other predictable occurrences of f-epenthesis, but I shall treat these only briefly. Firstly, there are numerous occurrences of [f] before geminate consonants. The type of representation for geminates shown in (1) would predict that geminates could not split in the same way that a cluster could. The rule in (35) ensures that underlying forms such as (34)

a. X X X X W n

b. X X X X X

I I a t 'mother'

II \/ n g g

I r 'speech'

syllabify correctly without violating the 'integrity' of geminates15, as in (36). (35)

| 0 —• X / — X' X I '

(36)

V [

]

a.

b. X I

i

X X X X \/ n

a

11 t

o /

X n

\

a

a

/ \

/

\

i

r

xxxxxxx i

g

i

g

Some interesting cases arise in the case of Type A triliteral verbs containing labiovelars. The morphological patterns of Type A require infinitives and 3ms jussives to have the forms má-CCáC and y-CCaC respectively. Thus, we have mágdal and y[i]gdal from g-d-l 'kill'. The surface forms of these for gw-d-l 'decrease' are mágw[ü]dál and y[ijgw[ú]dal. A moment's reflection will show that the occurrence of an epenthetic [ü] (see end of IV for the explanation of this) would not be predicted by (26) since gw has been allocated to a single X slot, so that the full syllabification ought to have occurred (as in (37)a.).

317

The High Central Vowel in Amharic

(37)

a. * a

/\

a

/\

b.

A

CT

CT

A

CT

/\

X X X ^ X X X

XX

x x x x x

m ä g w d ä l

m ä g w ü d ä

I

Epenthesis occurs in this case because gw (like all consonants with labialized offglides) can occur only in an onset. The correct form of (37)b. results from (38). (38)

| / 0 —• X / X' — X M

][ ]

In Amharic word-initial r always has an optional pronunciation with a preceding [/]16, viz: (39)

| 0 —• X / # - X

For example: (40) [i]riz ~ riz 'beard', [i]ret ~ ret 'aloe', [i]rob ~ rob 'Wednesday', [i]ruz ~ ruz 'rice', [i]rist ~ rist 'inheritance', [i]rajj[i]m ~ ra)j[i]m 'tall, long', [ijrab ~ rab 'hunger' It does not seem possible to relate this case of f-epenthesis to syllable structure. Some words show interesting alternations which depend upon whether this optional process applies or not, for in cases where a word-initial r would be extrasyllabic the rule of (26) has to apply if (38) doesn't, eg: (41) [i]rgib ~ r[i]gib 'pidgeon', [i]rguz ~ tfijguz 'pregnant', [i]rda ~ r[i]da 'help!'

IV Amharic, as we have seen, has no less than five distinct rules of epenthesis, four of which are motivated by considerations ultimately related to syllable structure. Yet in every case the the choice of the epenthetic vowel is the same. At a result of recent and illuminating work on

318

Dick Hayward

'underspecification theory' by Archangeli (1984), an explanation for this choice becomes apparent, for in the Amharic vowel system the maximally underspecified vowel is i. To appreciate this, it will be necessary to present a brief resumé of underspecification theory17. Within this theory phonemes in underlying representations are given a minimum of distinctive feature (hereafter abbreviated to DF) specifications. Not only is the number of DFs kept to a minimum, but the theory operates with the premiss that there should be no DF requiring specification in every phoneme. For all underlying representations, moreover, only one value (either ' + ' or ' — ') may be selected for any DF. The choice of DFs to be employed and the choice of which value it is that is to be specified (when this is necessary) for any given DF are not matters that can be determined without reference to the phonology of the language concerned. For example, in Qafar (a Cushitic language of the Horn of Africa) having a five-vowel system, if we operate with just the three DFs [hi, lo, rd] (instead of [rd] we might have used [bk]) quite a number of 'minimally specified matrices' are possible; three of them are shown in (42). (42)

a. [hi] [rd] [lo]

i e a o u -

+ + +

b.

i e a o u +

+ +

c.

i e a o u -

+

Deciding which of these (if any) is the correct one would involve sensitive consideration of the phonology of Qafar. One thing that emerges in (42) is that in each case one of the vowels is completely without specifications. Underspecification theory makes the interesting prediction that (among other things) this maximally underspecified vowel is the one we might expect to find as the epenthetic vowel of a language having epenthesis. The minimally specified matrices resulting from the application of the principles outlined above usually attain full specifications in surface forms by the operation of redundancy rules. But this is not the only way whereby unspecified values may be completed, for phonological rules may also apply to minimally specified matrices. Phonological rules differ from redundancy rules in that they can change the underlying values of DFs. Redundancy rules may only insert feature values, and a provision is made in the form of the Distinctness Condition to ensure that values are inserted only if no value is already specified.

The High Central Vowel in Amharic

319

(43) "The Distinctness Condition' (Archangeli 1984:46) 'The input to a redundancy rule is not rendered distinct from the output by application of the redundancy rule.' Since minimally specified matrices of different phonemes do not contain different specifications for the same DF, they do not confer distinctness, according to one definition of that term, viz: 'Two units Ui and U2 are distinct if and only if there is at least one feature F such that Ui is specified [aF] and U2 is specified [/7F] where a is plus and ¡3 is minus . . . ' (Chomsky and Halle 1968: 336). Distinctness will be achieved by application of redundancy rules. Phonological rules, however, by changing feature specifications may prevent a latent distinctness from being realized. There are three sorts of redundancy rule: 'default rules', 'learned rules' and 'complement rules'. It is claimed that default rules are universally available. The following are examples of default rules. (44) a. b. c. d.

[ [ [ [

,+lo] ] —»- [-hi] / [ ] —[-lo]/[ ,+hi] ] —>- [ard] / [ , abk, -lo] ] —• [-rd]

While the actual specifications in underlying representations have to be learned, the mechanism for forming complement rules can be regarded as universally available. I believe underspecification theory to contain a great deal of relevance for a satisfactory treatment of Amharic phonology18, but for the moment I wish to focus only upon its relevance in accounting for ¿epenthesis. It does not seem possible to establish a minimally specified matrix for the 7-term vowel system of Amharic without the use of four DFs.

320

Dick Hayward

Operating with the customary DFs used for vowels, ie. [hi], [lo], [rd] and [bk], plus knowledge of the fact that the epenthetic vowel is i (cf. Archangeli 1984: 64ff.), the simplest and most plausible 19 matrix would be that of (46)a., which, after the application of default and complement rules (no learned rules are required) will end up as (46)b. a.

i

e

[hi] [bk] - [rd] [lo]

i

a a o

-

u

+

+

+

b.

i

e

i a a o u

[hi] + - + — — — + + + + + + [rd] + + [lo] + —

—

Before considering what redundancy rules might be required, I should like to dissent from the view that (46) shows the most appropriate matrix for Amharic. My objection concerns the use of the feature [bk], Amharic is a language with true central vowels, and although it is a common assumption (cf. Schane 1973:12) that nonfront vowels are necessarily [ + bk], it could well be that this assumption is wrong 20 . It is not simply phonetic accuracy that is at stake; there are phonological processes of the language which are obscured by insisting on this feature to the neglect of a better one. Stem-finally in verb forms the nonhigh coronal consonants (t, d, t', s, z, n, I) regularly palatalize (to c,j, c', s, z, n, y respectively) before the vowel i in suffixes such as the 2fs imperfect and jussive subject agreement (-i) and the agentive (-i) and instrument (-ia) nominalizing formatives. A rule attempting to express the naturalness of this process would make crucial use of the feature [hi]. However, precisely the same palatalizations take place before the Is converb subject agreement, in which the vowel concerned is e. It is clear that no amount of juggling with the features [hi] and [bk] can make this eminently natural process appear so in a rule. Arguments similar to these were advanced by Fromkin (1968:161) in support of a proposed feature 'palatal'. In the absence of any convincing arguments to the contrary, I propose to replace [bk] by [pal](atal), and so revise (46) to (47). (47) a.

i e i a a o u

[hi] [pal] + -I[rd] [lo]

-

+ +

+

b.

i

e

i

a a o u

[hi] + - + [pal] + + [rd] [lo]

+ + + + - -

321

The High Central Vowel in Amharic

It is observed that this substitution creates no difficulties for the matter under consideration. It is relevant to note, however, that feature minimization, which is a maxim of underspecification theory 'forces the use of the same features for consonants as are necessary for vowels' (Archangeli 1984: 54). Here too no problems emerge, for as far as A m haric is concerned there is no compelling need for the feature [bk] for consonants. The redundancy rules required to get from (47)a. to (47)b. are as follows 21 . (In (48) default rules are marked ' D R ' and complement rules are marked ' C R ' . ) (48)

i. ii. iii. iv. V. vi. vii. viii. ix.

[ [ [ [ [ [ [ [ [

]-[-hi]/[ , + lo] D R ] —• [-rd, -pal] / [ , + lo]DR ] —• [-lo, -pal] / [— ] —»- [-lo, -rd] / [

, + rd] D R

— , -1-pal] D R

] — e "S Bj u s

c u e o a

(2 at/i >

I

u

E

&

1 9 m Tf

c o

"S a

•o u g «2 « s

vo (N £

2

I>-OO t- aw J

•o C .ts V S i 1w t» w a U-) •e o 00 "2 II

> 2 J! 05 to

(2

*

* * OS ,

'So ed 3 ^ * • s? «ed -O g fi ? co •o 2 u a r-\ n< a c

ed IX 3 60

c 60 'S s e M(A h fi ì '3 c o ci ^ •J3 4) o -a 2 -2 2 . 2 2 fi o Ic I >M o3 r5t *o > ** COo 43 o a c ed

•ts t-i 5 .S ? S o t „ » e c 2 6 I ® fi -o I OI ^ e $ 3o a J S «i J2 >> u> eo§ e ? u" 3 I l i fad> et] eo 2 « ® 8o «>s s« ed a> a 01 > g a g " ^ 3 S 1fi fl5 1 5 cd ? E - 5a 8 E 60

fi

. .

t-

h O h 60 O O- >Q ~

fi CO 4J 'l- M) ^ o ed U fi ^ ti fi 2 cd K ed a - o i> u « s-s-g CO — u fi £ ^ ^ •«2 ' -1 -S o u C ed 5 5 7 5 § 3. I fi 'S § ÌC- o§ K2 S4J >>

The Concept of Politeness and its Formulas in the Romanian Language

P

8 a

-S

.

M-s^-c

u T3 3

y

6 0

as.a««

®

® -2 l i , x> r

S S!

a

J i ^ >>

o

c

la

a

2 J ™ a 03 ©

8

4> 2 i "S o

H b

a

t-l o

|

.a c .SJ o tu §

|I t oS. £a s— s x8>

u J3

I

§ ts r

j

"

01

S

S

i

« " I a 3 5 tJ 2 5 S g O > ¡5 O l t> & 0 J 3

-

a

O

CO

-o 0

I

I

i) h

£

W 1> •C

>, 2 a 1 3

§

a 5 1

>> g—;. ¡2 ^

e 2

,

2

S

u lo 11 o

ed -o

N W I I u « l „ O 3 B « ¡3 u _ w „1 •S x> Ji s j g » I s ^ l ^ cd -t-» a l S ' E . - S ' s . i I > 1 rabi "J o a S . ss .5s t¿s» « a a is

i

cq.

2 u •e u

^

..

-o

3 6 9 'P £ S "s o >> " r " -js a> c '2; ro 3 r

£ 0 ~ ts fe JS

£ .5

H

3 3 j-to o * i 'g " - a « 2 •§ a ^ 5 -15

G o

t«

i 'S A a o f c S S t) c 8 § " a o ' g S 2 « h O C v 5 0

rT T -

a• S f i S f

00

».i- o x>

e a ..«S

1 ^ E v i S 3 o -o 5 U

o

1

S JS £ £ wr S . - 1 3 - c ' 5 , 3 i . S 2 ' 3 . S (2 a

•O o

ss| o

s

«> > c . S -a a K 5 c r " .8 O en i

7 6

•a

48

Tatiana

ed

g

V

a ? . t? >> ^ ^ ° ^ 53 © g g-g? ® °

4.S a 5 >> ed

€

ed P

8 c ; "a ft. e t® -e E-

«a> e S

.2 ^ §

J

2

S 3 « 9

^ —

o

C o 2 u > V Ol

2

„

O

C •a •5 E S

60 u x> C

2 I D co cn 0> « ~ ^ -a

60" C ~ •n « ^ u

E a _ « a1 Cd © •• j"

-t 3

Kj u

60

E

60 60

u E o U

3

«

l 2; ì e a © ¡3 | o

E u « ji © © e « 9> X •>>•% fi .

« S u il a « g ' S - i « e

»

u

g

U« w

&

e . ai 13

Slama-Cazacu

The Concept of Politeness and its Formulas in the Romanian Language (and diversified) for wishes. I also considered some invocations as PFs, when expressed as a positive wish in favour of the partner (an intention of solidarity with him). That is why, also the partner's words or actions can be the object of the emitter's solicitude: (44) Nu fi-ofigura poleita '(I hope) you're not golden mouthed' [when the partner expresses anxiety at a future bad event, the emitter wishes his partner not to be a Cassandra, a prophet of bad events]; or: (45) Sa fie la el acolo 'May it stay with him/her' [when the partner speaks about somebody's illness - a tumor, etc. - and shows on his/her body the sick place, the interlocutor wishes that the 'bad' thing 'remain' on the evoked person's body]. Expressions of thanks are also numerous and differentiated.12 An interesting fact is the reduced number of compliments (as against English). The attitude of paying a compliment is indirectly expressed, by questions, which show the emitter's interest in the interlocutor's condition or situation, which has to be followed by a PF: Sa o porfi sanatos see (39). The compliment was perhaps prohibited also because of the superstition (for the 'evil eye': after a compliment, the PF [26] had to be used). To prevent 'evil', there exist Fs of negative wishes (some of Turkish origin, as: [46] Sa fie pe capul dufmanului 'May it be on the enemy's head'). Good faith plays an important role in relations of solidarity and therefore the partner has to be convinced that the emitter is sincere: hence asseverations, vows and oaths invoking the dearest members of family, the most precious parts of the body (eyes, head).13

7.0. Obligatoriness of PFs has two aspects: concerning formal coercion (obligation to use certain forms: a fixed or relatively invariable structure), and concerning the obligation of using specific Fs in certain situations. 7.1. Obligatoriness of form. The fixed forms are not aleatory: they are set in patterns that are repeated in all Fs (in RL the optional variants are less frequent than in English - see Wolfson 1982a: 127 -, and they generally respect the basic form).14 Even variants are set within the general pattern: this peculiarity is connected with the force of usage of the PFs,

49

50

Tatiana Slama-Cazacu

still existing in RL. With the withdrawal of the coercion of using certain Fs in special situations (e. g. 'thanking for a meal' in present day French society, while in Romanian it is still in use), the fixed form disappears too and substitutes may be used (as the necessity of 'saying something' - e.g. in France, after a dinner - is felt however). That is why, in such cases, the speakers no longer feel the character of a ' F for the expressions which they use (cf. Manes and Wolfson 1981:126,127 ; and my first observations with young people in Ban, Italy). In these communities, one embroiders on a canvas with vaguely outlined contours, while in RL very precise patterns are followed or at most there are variations of colour or of details inside the pattern. Certainly, some Fs are no longer used (see [47]), or are in use only with older people or in certain milieux; some others begin to be used by the younger generations, argotic expressions are created or Fs from foreign languages are adopted (as: [48] Ciau, several years ago). There are Fs which disappear or are modified because mentality changes (invoking divinity mostly, or based on superstitions): it is no longer fashionable in RL (as in Italy, it is not in keeping with the "Galateo", the "Code des manières élégantes") to say Noroc (Bless you) when somebody sneezes.

7.2. Obligatoriness of using PFs. The use of a PF whatever its form is felt as obligatory, as a ritual of P which regulates human interaction. The variations in the concrete form of this obligatoriness are however to be taken into consideration, as a general phenomenon.15 Historical and social evolution, cultural modifications, changes in mentality, bring about such variations, as well as the loss, creation or borrowing of Fs. In RL, in the late 18th - early 19th centuries, the great influence of the Phanariot princes (Greeks from Constantinople, who bought from the Porte the right to be 'Prince' Domnitor of Wallachia and Moldavia) imposed certain oriental PFs, mostly at court. The population continued to use the traditional Fs, but little by little some of the foreign Fs were adopted also by them. Another influence, that of French étiquette, began to be felt, in the mid 19th century. The use of PFs is also much influenced, in the same community, by the age of interlocutors, their relations, the register (formal/informal situations) etc. On the synchronous level, variations appear according to the social communities, to geographical areas (dialects, patois etc.). The same individual uses of course Fs according to interlocutors, to situations, etc. The 'shame' for

The Concept of Politeness and its Formulas in the Romanian Language not using a PF disappears sometimes with people who are not acquainted with their occasional partners (on a train sometimes; cf. also Wolfson 1981 a: 127,129). I have collected the use of PFs also in the process of work, where, for communicational economy, some Fs are not obligatory, but some of them, in special situations, must be used: (49), Noroc bun 'Good luck' is the usual greeting among miners. Special research of social psychology, connected with more precise and minute observation of linguistic and psycholinguistic facts16 would lead to interesting conclusions in this respect.

7.3. Obligatoriness is expressed (in 7.1. and in 7.2. as well) by the frequency of occurrences (usually, directly proportional with obligatoriness). There are however some subjective factors that prevent frequency from being a very exact indicator of obligatoriness (e.g. some PFs may become obligatory because recommended in school, though not having great frequency in general, in the beginning at least. The prestige of some persons who use special PFs may modify, later on, their frequency in a collectivity.

7.4. Both obligatoriness ofform and that of usage in specific situations can be analyzed from a psychological point of view (as far as we know, it has not yet been done in the bibliography of PFs). Such an analysis can provide explanations for certain aspects of PFs and their functioning. The use of PFs can be included among automatized behaviours. That is why they are used very promptly (and sometimes they are even uttered mechanically by persons who do not afford them in their initial meaning: e.g. Bozhe upasi 'God forbid' said by atheists in modern Russian). Another psychological aspect is the learning of PFs (based on memory, which is helped by the relatively fixed patterns, by their brevity, and by some prosodic elements. In the Romanian community, the PFs are as a rule taught by adults to younger generations at a very early age (they are part of the buna creftere 'good breeding'), at home and in preschool institutions. They are learned in association with the situations where they must be used and included in a whole system of be-

51

52

Tatiana Slama-Cazacu

havioural rules necessary for social interaction. In our corpus concerning the dialogue of children (Slama-Cazacu 1961/1977: 58, 113, etc.) there are many PFs, used even under 3 years of age:17 (49) - S. (3 years, 2 months) Poftim [the PF for: 'Here, take some']. - M. (3; 2) Mulfumesc 'Thank you'; (50) - Z. (4; 2) Hai, te rogfrumos. Ilenuca 'Come on, I beg you' (51) - A. (4; 6) [the play of "buying food"] Va rog [etc.], 'Please give me a cabbage'; (52) - F. (6; 3) [the play of "calling up"] Te rog [etc.] 'Please, come to me to-night'. The awareness of the coercion of PFs appears quite early: (53) - M. (2; 8) Poftim [while she "feeds" her doll]. Spune me[r]si, la revedere 'Say thank you, good bye'; (54) - A. (7; 3) ["answering the phone" at the question Ce maifaci - see (23)] Bine'Fine'. O. (6; 11) [correcting him]. Bine, multumescTm fine, thank you'. Forgetting is naturally correlated with the process of learning such automatized expressions. A person may continue to use some PFs, without being aware of using them; a F may be apparently 'forgotten', but an effort at recalling might reactivate that F (a subject told me, while asking him about PFs - which he did not remember, but 'found' them again - : "Look. - [there are] expressions which I used to know in my childhood and I have forgotten since!"). The use of PFs is connected with emotional attitudes - generally positive (we are referring here to the attitude towards the use of PFs). Many persons (belonging mostly to the older generations) show discontent when a partner does not use specific PFs and characterizes an interlocutor according to the use of PFs or the failure to use them. Generally, the obligatoriness of PFs is not resented as a negative coercion. Provided specific PFs are known (in their traditional form or a new one), they are used with pleasure. At any rate, all the subjects answered in my own research in RL - as well as in that performed in Italy - with interest, even with satisfaction (cf. also Manes and Wolfson 1981) and with a desire of cooperation.18 The generally positive attitude towards the coercion of these Fs is also generated by the fact that they offer instruments for social interaction, which function adequately, hence efficiently, in special situations.

8.0. Conclusions 8.1.

The repertory of PFs in RL is rich and includes many expressions with powerful stylistic features, with some specific peculiarities of structure and function (or subfunctions). They have also aspects common to PFs

The Concept of Politeness and its Formulas in the Romanian Language 53 existing in other languages and mostly in the ancient Balkan area (subject to the Turkish influence) and in the oriental Mediterranean area. Even some PFs of Latin origin got new connotations and forms - with many picturesque peculiarities, ritualized in chains of an exchange of Fs, etc. - adequate to the local historical contexts of life.

8.2.

The PFs only reveal their characteristics when analyzed as a unitary ensemble (including, 'thanks', 'greetings', 'wishes' etc.) having a common basic function: that of creating or maintaining social interaction, mirroring a sense of solidarity within the community but also an attitude of omenie, of respect for a general class of 'friends' or non-enemies. 8.3.

The repertory is still in use as far as most of the Fs are concerned and preserves all its categories (finalities and relation with various situations). 8.4.

The investigation in contexts of real life allows one however to notice a great dynamics of these PFs. Being subject to social determinism and to psychological laws, many of the Fs disappear and are continuously replaced by others while some remain in the passive storage of older generations or continue to be used as automatisms in daily communication even by the younger generation, often deprived of their original contents. 8.5.

The social finality of these PFs is that of pragmatic necessities of interaction, as well as of the human sense of solidarity for a common life under difficult historical conditions. Originally, these Fs were magic verbal invocations, marking the intercommunitary solidarity by asking divine forces to prevent the evil

54

Tatiana Slama-Cazacu

fate for the interlocutor (or to send the evil fate to his/her enemy): "May the day be good for you" - Buna ziua - , "May you live many years" - La mulfi ani, etc. with most speakers, the evolution of mentalities mainly since the beginning of the 20th century has voided the majority of these Fs of the magic intents and meanings. They are still in use however, in interactional strategies - where only a few would recognize fossils of magic invocations in the most usual Fs: as the universal Buna dimineafa 'Good morning', or Sanatate '(I wish you the best of) health'.

Footnotes 1 First presented in Romanian, in extenso (also including the repertory and comparative considerations, with examples in various other languages, in Slama-Cazacu 1983; cf. also in press. 2 Besides the direct observations I personally made (in various language communities, collecting bibliography, or using my questionnaire in preliminary investigations for Italian, Russian, German, and other languages), a large research project is currently underway. It is done under my direction (using the same questionnaire and my classification) in Italy, by four teams covering various geographical areas (Abruzzi-L'Aquila, Bari, Bologna, Torino), each with 40 subjects of different age, social stratum, and economic category. 3 One of the rare exceptions is the study of Tannen and Oztek 1981; it is also a valuable study as to the languages (Greek and Turkish) from which they made the corpus of PFs. The article has however some methodological shortcomings: e.g. data were only partly collected directly in the respective countries (Turkish subjects were mostly students at Californian universities), the questionnaire for Greek was sent to be completed by persons living in Athens, classification of Fs is not consistent with well-grounded criteria (why 'arrivals' are included into 'gains'?) etc. 4 A minor Romanian author of the 19th century (M. F. Livescu) underlined this feature of P (called in Romanian buna cuviinfa). 5 The difficulty in studying the PFs in Italian nowadays (see note 2) also consists in defining the concept of P and even its naming (because cortesia has special connotations; that is why I proposed a term which is completely obsolete, but has no such connotations: politezza). 6 Without living in the context where the PFs are used and where they evolve and change their frequency, sometimes very rapidly, one cannot make classifications and propose interpretations that correspond to reality. An example: Gheia sou (mentioned in Tannen and Oztek 1981 as having a low frequency, with the meaning 'Your health') is reported by people going to Greece or living there as having a high frequency, with the sense of 'Hi', in various milieux (instead of Kalimera 'Good morning' [or 'day'] or Kalispe-

The Concept of Politeness and its Formulas in the Romanian Language

7

8 9

10 11 12

13

14 15 16

17

18

55

ra 'Good evening'). Or: the questionnaire now used in our investigation in Italy reveals a lower frequency of Buon appetito 'Bon appétit' in some regions and with younger generations. In RL spoken by younger people (teen-agers, students), such PFs as Bunâ ziua, Bunâ seara 'Good evening' etc. are, as a rale, replaced by Salut 'Hi', 'Hello', or Bunâ 'Good (morning/ afternoon)'. Almost all PFs have an exclamatory intonation (see 5.9). For economy of space we do not put the exclamation mark. Ad litteram (literal) translations are usually given (rarely, the corresponding F in English). Translation is generally given only at the first occurrence of a F. When a part of the F in RL cannot be given in good English, it is placed between []; when a part of the F is added in the English translation, it is placed between (). In RL the personal pronoun is not normally used before the verb as a subject (e.g.: Merg '(I) go') ; nor are other pronouns obligatory with verbs : Mulfumesc "Thank (you'). A very interesting linguistic fact is the word order of Noapte bunâ 'Good night', which differs from the rest of the series, having the adjective after the noun: Bunâ ziua, Bunâ dimineafa 'Good morning', Bunâ seara 'Good evening', Noapte bunâ is only used when departing: that is why, my explanation for this word order is that the F only remained in the form of an answer (see [14]) or when leaving somebody. It is the usual PF in RL, corresponding to: 'How are you?'; it requires an answer. A subject (from a rural milieu) told me: "If [he] does not answer, you tell (yourself): This man did not hear, or he is angry with me: I told [gave] him Bunâ ziua and he did not answer." It is very interesting that in South-Danube dialects of Romanian (Macedoromanian) PFs for thanks and for requests almost do not exist (or are used only for very special, important situations): the offer of help was felt as an obligation, in those closed communities. In the past, the beard was a very precious detail in the highest categories of boyars [ = noblemen] it was cut down by the prince when the boyar was found guilty, therefore an asseveration or oath was: (47) Pe barba mea 'On my beard'. In Italian, we have found greater variety: in rural lif due to the existence of dialects, in urban milieux mostly due to an evolution towards loss of a great number of ancient traditional PFs. Studied by us in RL, and now verified, with a large team, in various situations and regions, in Italy (see supra note 2). Their absence (as in Mikula 1977) prevents arriving to in-depth studies in this field (the importance of the German PF Entschuldigen Sie..., which is used by the subjects, is not noticed by the above-mentioned author, who only analyzed the utterance following this F: p. 7,22). PFs are taught very early in many languages: sometimes they are among the first words learned by a child: Bye-bye etc. In Romanian, a young child is told: (Intinde mina fi) zi: "Pa", or "Mersi", or "La revedere" '(Give [the lady, etc. your] hand) and say: "Pa" [ = the baby-talk word in RL for "Byebye"] (cf. also Manes and Wolfson 1981:127, Eisenberg 1982). Young Italian subjects, who had not the awareness of many such PFs, became interested in the topic and later joined the teams devoted to their study (see note 2).

56

Tatiana Slama-Cazacu

References Akisina, A., and Formanovskaja, N. 1978. Russkij reöevoj etiket (Posobije dlja studentov-inostranicen), Moskva: Izdat. Russky jazyk, 2nd ed. Atlasul Lingvistic Roman (ALR). 1966. Serie nouä, vol.V, Bucure§ti: Ed. Acad., s.v. Salutul(maps 1312-1315). Brown, Penelope and Levinson, Stephen. 1978. Universals of language usage: Politeness phenomena. In Goody 1978 (ed.): 56-289. Brown, R., and Ford, M. 1964. Address in American English. Language in culture and society, ed. by D. Hymes, 234-44. New York. Cipariu, Timotei. 1855. Portarea de buna cuventia intre oameni. Blasiu: Tipografia Seminariului. Corsaro, William. 1981. Communicative processes in studies of social organization: Sociological approaches to discourse analysis. Text 1, 5-64. Coulmas, Florian. 1981. Introduction: Conversational routine. In Coulmas 1981 (ed.): 1-18. - 1981. "Poison to your soul. "Thanks and apologies contrastively viewed. In Coulmas. 1981 (ed.): 69-92. - 1984 (ed.). Conversational routine. Explorations in standardized communication situations and prepattemed speech. The Hague - Paris-New York: Mouton. Courie, B. 1975. Polite plurals and predicate agreement. Language 51, 2, 406-18. Edmondson, W. 1977. Gambits in foreign language teaching. Kongreßdokumentation der 7. Arbeitstagung der Fremdsprachendidaktiker, ed. by H. Christ and H.Piepho, 45-8, Limburg: Frankonius. - 1981. On saying you are sorry. In Coulmas 1981 (ed.): 273-88. Eisenberg, A. 1982. Understanding components of a situation: spontaneous use of politeness routines by Mexican two-year olds. Papers and reports on child language development 46-54. Department of linguistics: Standford University, vol.21. Ferguson, C. 1976. The structure and use of politeness formulas, Language in society 5.137-51. Goffman, E. 1971 [1955]. On face-work: an analysis of ritual elements in social interaction. Communication in face-to-face interaction, ed. by J. Laver and S. Hutcheson, 319-46. Harmondsworth: Penguin books[1972]. Golescu, Iordache. 1975. Poväfuiripentru buna cuviinfä. Bucure$ti: Ed. Eminescu. Goody, Esther. 1978. Questions and politeness: Strategies in social interaction. Cambridge: Cambridge University Press. Gumperz, J., and Hymes, D. 1972 (eds.). Directions in sociolinguistics. New York: Holt, Rinehart and Winston. Hofmann, J.B. 1926. Lateinische Umgangssprache. Heidelberg: Winter. House, Juliane. 1980. Opening and closing phrases in German and English dialogues (Projekt Arbeitspapier; mimeo, 26 p.). - and Kasper, Gabriele. 1979. Politeness markers in English and German. In Coulmas 1981 (ed.): 157-86.

The Concept of Politeness and its Formulas in the Romanian Language

57

Hymes, D. 1972. Models of interaction of language and social life. In Gumperz and Hymes 1972: 35-71. Iordan, 1.1975. Stilistica limbii române. Bucureçti: Ed. çtiinjificâ. Jakobson, R. 1960. Concluding statement: Linguistics and poetics. In Style in language, ed. by T. A.Sebeck, 350-77. Cambridge: The Technology Press of M.I.T. Keller, E. 1981. Gambits. Conversational strategy signals. In Coulmas 1981: 93-114. Kendon, A., and Färber, A. 1973. A description of some human greetings. Comparative ecology and behaviour of primates, ed. by R. Michael and J. Crooks, 591-668, apud Goody 1978: 318. Komnenich, Pauline. 1977. Decision-making aspects of greeting behaviour among Serbians and Montenegrians. International journal ofpsycholinguist s 4, no .1 [7]. 31-50. Lakoff, Robin. 1973. Rules of politeness. The logic of politeness. Papers from the Ninth Regional Meeting of the Chicago Linguistic Society ed. by C. Corum et al., 291-305. Chicago: Chicago Linguistic Society. Apud House and Kasper 1981. Leontiev, A., Sorokin, JU., and Tarasov, E. 1977 (eds.). Nacional'no-kulturnaja specifika reöevogo povedenija. Moskva: Nauka. Manes, Joan, and Wolfson Nessa. 1981. The compliment formula. In Coulman 1981 (ed.): 115-32. Matisoff, James. 1973. Psycho-ostensive expressions in Yiddish. Maledicta, apud Tannen and Öztek 1981: 526. Mikula, Gerold. 1977. Bitteformulierung und Hilfeleistungsverhalten, Berichte aus dem Institut für Psychologie der Universität Graz. Pietreanu, M. 1977. Salutul verbal in limba romänä [Abstract of Ph. D. dissertation], Bucureçti : Tipografia Universitafii. Ralea, M. 1949. Explication de l'homme. Paris: P.U.F. Sacks, H., Schegloff, E., and Jefferson, G. 1974. A simple systematics for the organization of turn-taking in conversation. Language 50, no. 4 696-735. Schegloff, E. 1972. Sequencing in conversational openings. In Gumperz and Hymes 1972 (eds.): 346-80. Searle, J.R. 1971 (ed.) The philosophy of language. London: Oxford University Press. Slama-Cazacu, T. 1957/1984. Verhältnis von Sprechen und Denken bei Vorschulkindern. Berlin: Akademie Verlag (Rom. ed. 1957). - 1959/1961. Langage et contexte. The Hague: Mouton (Rom. ed. 1959). - 1961/1977. Dialogue in children. The Hague-Paris-New York: Mouton (Rom. ed. 1961). - 1964. Comunicarea in procesul muncii ('Communication in the process of work'). Bucureçti : Ed. $tiin{ificä. - 1968/1973. Introduction to psycholinguistics. The Hague-Paris: Mouton (Rom. ed. 1968). - 1970. The power and limits of social context of language behaviour, Cahiers de linguistique théorique et appliquée VII, 31 -41. - 1971. Die dynamisch-kontextuelle Methode in Soziolinguistik. Soziologie der Sprache, ed. by R. Kjölseth and F. Sack 73-86. Opladen: Westdeutscher Verlag.

58

Tatiana Slama-Cazacu

- 1972-1973. Sur le concept "socio-psycholinguistique", Bulletin de psychologie, 26. no.304/5-9,246-251. - 1973 (éd.). Cercetàri asupra comunicàrii ('Researches on communication). Bucureçti: Ed. Academiei. - 1976 a. The role of social context in language acquisition. Language and man, ed. by W.McCormack and S. Wurm, 127-48. The Hague: Mouton. - 1976 b. Nonverbal components in message structure. In McCormack and Wurm (eds.) 1976:217-28. - 1981. Interpersonal relations and dialogue structuring: on "dialogic syntax". International journal of psycholinguistics 8.4 [24], 57-103 [in extenso: Studii §i cercetan lingvistice, 1982.3.211-24 and 4.301-21], - 1983. Formule de politefe în limba româna: structura çi funcfii ('Politeness formulas in Romanian: Structure and functions'), Studii çi cercetari lingvistice 34. no.3.237-63 and no.4.287-310. - in press. Les formules de politesse en roumain: Considérations comparatives. In Festschrift W.Bahner, Berlin: Akademie-Verlag. Tannen, Deborah, and Ôztek, Piyale Cômert. 1981. Health to our mouth. Formulaic expressions in Turkish and Greek. In Coulmas 1981 (ed.): 37-54 (from Proceedings of the 3rd Annual Meetings of the Berkeley Linguistic Society, 1977). Tohyama, Yasuko. 1983. A semiotic analysis of meeting and parting rituals in Japanese and English. Proceedings of the XIHth International Congress of Linguists(1982). 1207-13. Tokyo: Gakushuin University. Verschueren, Jef. 1981. The semantics of forgotten routines. In Coulmas 1981 (ed.): 133-54. Wolfson, Nessa. 1981a. Compliments in cross-cultural perspective, Tesol Quarterly 15, no. 2.117-24. - 1981 b. The descriptive analysis of rules of speaking. Paper at the 6th Aila World Congress, Lund 1981 (mimeo, 13 p.). Youssouf, I., Grimshaw, A., and Bird, C. 1976. Greetings in the desert, American ethnologists, no.4. 794-824, apudCorsaro 1981: 63. Zanne, 1.1890-1912. Proverbele Românilor, Bucureçti. 10 vols.

Language and Society: The Case of Japanese Politeness J.V.Neustupny Monash University Clayton, Victoria*

The problem I shall discuss in this paper is somewhat narrower than the problem of Language and Society in general. The general problem contains issues both of social structure constraining language and language constraining society. Although there is much we still do not understand about the former issue, the points to be attended to are less controversial and I shall not discuss them here. However, the latter issue remains highly debatable. Does language constrain our non-linguistic behaviour or does it not? The question has been with us for a long time and, for the time being, is likely to stay. In its strong version the theory of linguistic relativity, which answers the question affirmatively, implies that the shape of language fully determines our perception of social facts and our social behaviour. A weak version of the theory merely claims that the shape of language can maintain, reinforce or alter certain social facts because they are, for instance, frequently referred to. It is possible to deny that the etymology of the Japanese word kanai "my wife" (etymologically "inside the house") fully determines our perception of matrimony (the strong version), but still maintain that when the word is used, the assertion is made that one's wife is someone who belongs "inside the house" and should not take equal part in the outside social life (the weak version). If this is the case, language does indeed influence society. One cannot easily avoid using the word kanai (there is no other stylistically neutral word), and when one does use it, the theory of linguistic relativity claims, one asserts a particular social ideology. Notice, however, that no assumption is made in the weak version about the directionality of the relationship. Linguistic facts may be secondary to

60

J. V. Neustupny

social facts, but still exert influence on social situations. In this paper only the weak version of the relativity theory will be considered. Sociolinguistics of the 1960s and 1970s was highly sceptical with regard to the strong version, or clearly denied its validity (Fishman 1960, Hymes 1966, Grimshaw 1971, Fishman 1980; an excellent summary by a psycholinguist is Steinberg 1979). On the other hand various types of the weak version of the relativity theory have survived into the 1980s and can hardly be ignored. The study of discourse which developed in the more recent years of sociolinguistics throws new light on the issue. It reminds us that the question we should ask is not how language possibly constrains behaviour in abstracto, but how much and in what ways it exerts influence in actual discourse. To contribute to this inquiry is the aim of my paper.

1. Japanese honorifics and Japanese society The relationship between honorifics and society is a question often raised in connection with Japanese honorifics, one of the most highly developed honorific systems known (cf. Martin 1964, Hinds 1976, Harada 1975, Neustupny 1978 a, 1978 b, Ide 1982). To what extent does the existence of the system reinforce traditional social relations? And if it does, what is the mechanism of such determination? The role of honorifics is to communicate distance between personnel in discourse (cf. Neustupny 1978:198, 219), but do they also contribute to the regulation of such distance? Whenever an honorific form is used, does such usage affect social relationships?

2. Study of the honoriflcs/society relationship in discourse The most efficient way how to answer all these questions is the study of honorifics as they are used in natural discourse. Since the ultimate location of all language is in discourse, this method of investigation is not merely the most efficient one but cannot be avoided. Observe, for instance the following three pieces of discourse uttered in particular speech situations: (1) O-cha nomu? "Will you have a cup of tea?" (A middle aged mother to a 14 year old son) (2) Taberu? Taberu yo! "Will you have your dinner now?" "Yes, of course!" (A wife to her husband 10 days after their wedding)

Language and Society: The Case of Japanese Politeness

(3)

61

Ikaremasu ka? "Will you go there?" (A young city clerk to a middle aged citizen who should go to the police to report a theft). From the point of view of the language system the noun o-cha "tea" in (1) contains the honorific prefix o- (cf. Shibata 1972). The 'Mikado' style of speaking about Japan would normally translate the expression as "your honorable tea", but in the given sentence the honorific meaning is totally absent. The speaker has no choice. "Tea" is for her o-cha and the use of the expression is completely automatic. In particular, it does not involve any consideration of the level of politeness. I shall say that the honorific meaning in this discourse is COVERT. Apart from the case of male speakers of some sociolects of Japanese (for whom the form cha "tea" exists), it can only be retrieved (overtized) under special circumstances, as when one discusses the etymology of the word. In this case the etymology is revived. It is still a part of the knowledge of the language, not merely a knowledge restricted to academics.1 In the case of the second example the situation is totally different. The wife can also use meshiagaru, taberareru, tabemasu and possibly other forms with exactly the same referential meaning (to eat, to have dinner) but with different honorific shades. Ten days after the marriage the honorific usage is OVERT and means that a certain social relation of equality and intimacy has been established. Discourse (2) contributes significantly to the reinforcement of the relationship. Other factors which contribute to the same effect include other communicative behaviour (such as facial expression, etc.) and substantive behaviour (such as who does what for whom). However, after a year of marriage the selection of one of the available forms (in our case taberu) will be completely automatic and within the particular situation will carry no honorific meaning whatsoever. It will be the only neutral expression available for the occasion. Within a sentence which sounds identical with (2) the honorific meaning will be covert. However, we can assume that even after a few years it can be overtized under special conditions, for instance in the presence of other personnel (the husband's parents, etc.). In this case the wife may feel that the form taberu would too strongly assert equality and intimacy between herself and her husband, and may select a different honorific form. Most honorifics used within a family, among friends, and among colleagues seem to be normally covert in the sense used here and do not, therefore, contribute to the creation or maintenance of social relationships, unless overtized. Of course, the study of the conditions of

62

J. V. Neustupny

overtization of meanings in general is a most important task of further research. With regard to sentence (3) I shall only say that the particular honorific form ikaremasu "you (weak superior) go" was overt for the citizen involved who had a strong negative attitude towards it and would probably have taken it for socially significant in any discourse in which it would be used in reference to his own action. Obviously, the problem of the relationship between language and society can be reformulated in terms of its mode of existence in discourse. The relationship only exists if it is overt. To know when a relationship is overt and when it is covert is an issue of primary importance for the study of language and society.

3. The study of overtness of honorific meanings What kind of evidence can be found about the overtness or covertness of an honorific meaning? The simplest case obtains when participants are aware of the social significance of individual segments of speech as it takes place. The following sentence occurred in a conversation: (4) . . . ko san wa ima oshiete rassharu no? "Axe you now teaching?" The 'are -ing' segment is expressed through an honorific verb '(i)rassharu'. In a follow-up interview with the speaker she said that she did hesitate at the moment of producing the sentence whether to use the honorific but decided that this was 'proper' for the relationship between herself and the addressee and for the situation. The interview confirmed thus that for the speaker the usage was overt. For the speaker the character of the relationship was reaffirmed, the situation was defined. For the hearer, however, there was no awareness of the usage, and nothing seemed to indicate that the usage would be overt for her. What are the other tests of overtness? Firstly there are non-verbal reactions, such as body movements, facial expression, violent reactions, etc. This method includes attitudes assumed at the time of interaction which are sometimes verbalized later. Participants conclude, for instance, without being able to quote any support for their claim, that other participants are 'shinsetsu' (kind), 'yosoyososhii' (indifferent), etc. Such attitudes indicate that a message about how the other participant conceives of the social relationship was received. Of course, unless the interaction was extremely short, it may be difficult to estab-

Language and Society: The Case of Japanese Politeness

63

lish which of the several possibly overt forms was actually overt, significant for the speaker. A further kind of evidence is furnished in the case an unaware corrective adjustment takes place. Observe sentence (5): (5) Bobbu . . . , uchi no shujin nanka, mattaku sono hantai de . . . "Bob, someone like my husband, is a completely opposite case..." In this case the speaker obviously corrected the too intimate reference to her own husband with a first name (Bobbu) by replacing it by the phrase "someone like my husband". This can be taken as an attempt to regulate the social relationship between herself and her interlocutor. The follow-up interview did not reveal that the speaker would have been aware of the corrective adjustment at the time it occurred. However, she did admit, on reflection, that the added phrase did probably have the meaning. Finally, negative evidence of an honorific form being overt can be supposed to exist when the form obviously fulfils a non-honorific function. For instance, in (6) (see(7h)) the use of the polite form iki masu (I'll go) fulfils a hypersyntactic function: it signals that the speaker is returning to a previous sub-topic of conversation.2 Of course, this kind of evidence is not conclusive in itself. While it is possible that one function excludes others, it can also be that the honorific function is added to whatever other functions the honorific form may carry. Other than honorific uses of honorific forms include mainly their use as hypersyntactic markers, or their use to communicate who are the various referents referred to in the sentence. The latter usage is connected with the fact that in Japanese subject, object, etc. are often not represented in the surface sentences and can only be identified indirectly, for instance thanks to the honorifics. Another case of possibly covert use of honorifics obtains when a speaker uses certain honorifics automatically to virtually all addressees. In a case like that, one can safely assume that for the speaker the honorifics are not destined to assert, reinforce, or introduce any particular social relationship. Of course, the usage may be (and in Japan frequently is) overt for participants other than the speaker. In many instances we do not possess at present clear cues to decide whether an application of an honorific form - or for that matter any other form - is overt or covert. Analogy from other similar cases may remain the only cue.

64

J. V. Neustupny

4. How many honoriflcs are overt: addressee honorifics The extent in which Japanese honorifics contribute to the creation, maintenance or alteration of social relation in Japanese society is thus in direct proportion to the extent in which honorific forms used in speech carry overt honorific meanings. To suggest, as has often been the case in the past, that each honorific form in each of its applications affects social relations is obviously incorrect. Only overt forms possess the capacity to create social reality. Forms which are rarely, if ever, overt cannot be claimed to exert influence on social relationships. Before we can obtain a satisfactory picture of the problem of the relationship in general, much particular work will have to be done. In order to assess some problems which may appear on the occasion of such analysis, a conversation of the duration of 22 minutes was recorded and analysed. The conversation took place between two Japanese women in their late 40s, educated in the Tokyo area. Both speakers were from elite middle class families, with fathers and grandfathers who graduated from the University of Tokyo. Because of their common background they became close friends after having permanently settled in Australia where (at the home of Jl) the conversation was recorded. J1 was approximately two years younger than J2 and this fact may be responsible for her higher rate of honorific usage. In the conversation they spoke freely about recent events in their families and about their common friends. In the following analysis I base my discussion on so-called addressee honorifics ('desu/masu forms'). Addressee honorifics are forms which principally communicate the degree of distance between the speaker and the addressee. The basic level of addressee honorifics used in the conversation was the 'plain style'.3 This is the most neutral style used within families and among good friends. For instance, while speaking about the opening of an exhibition the subjects diverge to a discussion of a common acquaintance: (7) a Irassharu? "Will you go?" b Soo na no. Kayoobi ni nanka, Monashu no sensei ga hiraku n'datte. "That's right. A teacher from Monash, is that right, will open it on Tuesday or when." c U n , . . . sensei ne. "Yes, D r . . . " d Nee, nanka, Nihongo dekiru tte "Tell me, what, they say that he can speak Japanese."

Language and Society: The Case of Japanese Politeness

65

e Nihongo mo joozu yo. Hooritsu no sensei da kedo. "His Japanese is good. But he is teaching law." f Nihon de hooritsu o yatta no? "Did he study law in Japan?" g Soo ja nai to omou wa. Ano, Nihon ni mo ryuugaku nasatta koto mo aru to omou kedo, hooritsu wa kotchi no daigaku de. "I don't think so. Well, I think he also studied in Japan, but he studied law at a university here." h Taitei ikimasu. "Most likely I'll go." Except for the last predicate (which is in the polite desu/masu style, cf. (6) above) all predicates in this section of the conversation are plain. None of these plain forms was reported as significant, the speakers did not mention any special emotive or attitudinal behaviour, the plain forms were not corrected, and there was no ground to believe that they would have been used for some other purpose - in other words, there was no evidence that the use would be overt, significant. No doubt, the exchange of the plain forms between the two speakers may occasionally be overtized (as when they speak in the presence of another participant, the relationship with whom still is to be decided), but in the conversation studied here, the use was automatic, and no special meaning was attached to it. The same phenomenon is known from established TU usage in European language: once firmly automized the use of the TU pronoun does not have a politeness meaning, except when overtized. However, apart from (7h) the conversation also contains a number of predicates which use 'polite' forms. Does this mean that the speakers reinterpret, in the middle of the conversation, their relationship and assert such a reinterpretation through the use of honorifics? Certainly, there must be occasions when this is appropriate: one can for instance 'put someone in his place' or express apology by using particular honorific forms. All speakers of Japanese know that this frequently happens in daily conversation. The polite forms in our conversation can be accounted for in various ways. Firstly, there are forms for which no plain alternative exists in the variety of the speaker. For instance, the form desho(o) "probably is; isn't it" is formally a polite form, but its plain counterpart daro(o) carries an additional 'hard' feature and would not be used by either of the two speakers except in anger, for instance when scolding her own child. A polite form for which the speaker has no plain alternative (or almost no plain alternative) is the expression n'desu kedo(mo) "it is, but" for J1 who used it 5 times (though once she used the plain counterpart n'da kedo) and n'desu tte "it is, they say". I shall hypothesize

66

J. V. Neustupny

that all this usage was automatic and covert for Jl. A possibility exists that the n'desu kedo(mo) or n'desu tte expressions are overt for J2, signalizing a certain degree of distancing of Jl from herself, but the possibility is remote. J2 did not respond positively to a suggestion that this could be the case, and in view of the long established relationship between the two, it would be most improbable. Secondly, in the following sentence it seems that a polite form is used in order to soften the written style construction de wa atta "true, it was, but" rather than communicating the distance between the participants. The sentence is (8) Da kara, kyoogaku de wa arimashita kedo ne "Hence, it was coeducation, but, you know" The conversation contained only one example of this construction. In this case, too, the honorific meaning seems to be covert. Thirdly, as already mentioned above, the use of the polite form ikimasu in (7h) is most likely a hypersyntactic marker of closing a subtopic and does not seem to be interpreted as a carrier of a politeness meaning. The politeness meaning which is available for this form in the system of the Japanese language is almost certainly covert here. There is one sentence which probably has an overt politeness form, but the overtness has resisted attempts at being clearly identified. (9) Ato de kaite kudasaimasu ka? "Will you write it down for me later?" In this sentence Jl uses a polite predicate kudasaimasu "do for me" instead of the possible plain form kudasaru. Kudasaru would be sufficient in view of the relationship of the two interlocutors. There is a possibility here that Jl wanted to be 'polite' in view of the sentence transmitting a request. This could have been interpreted by J2 in the intended way. However, no clear evidence has been found to support this interpretation. Finally, exchange (10) is an example of a discourse in which several factors intersect. (10) J2 Watashi no, totte gomen nasai. "I'm sorry for taking . . . , with my..." Jl O-satoo supuun wa o-satoo tsubo ni arimasu. "Sugar spoon is in the sugar pot." J2 apologizes for using her wet coffee spoon to take sugar. Jl does not accept her apology (she implies that there is no reason for one) but turns her attention to the sugar spoon (which J2 has not noticed) as a more convenient way of putting sugar in her coffee. Since the mistake resulted in spilling a few drops of coffee on the

Language and Society: The Case of Japanese Politeness

67

table, this was an embarrassing event for both subjects. The follow-up interview established that they both remembered the circumstances of the situation well. Neither of them was aware of the use of the polite form (arimasu), but J1 was aware of her attempt to be polite (soft) at the moment, and J2 thought that J1 handled the situation graciously. I assume, that the polite honorific meaning was encoded and decoded overtly in this case by both participants. Of course, the process may have been more complicated here. Should J1 have used the plain form aru (instead of the polite arimasu), the norm of the language would have required that a sentence final particle such as wa, wa yo, kedo, etc. be attached. However, this choice would have been difficult, because the semantic content of the final particle might have introduced an unwanted meaning, such as a strong assertion. It was therefore safer to use the polite form, with which a final particle is not required. Moreover, I assume that the polite form was also used in its function to express subordination in response to anger or apology, and that it was decoded as such. We can conclude that the use of addressee honorifics in our conversation was mostly covert and was overtized on one or two occasions only. It is difficult to imagine that it would have contributed in a significant way to the reinterpretation of the social distance between the two subjects. The relationship was fully established before the interview took place. Except for sentences in which honorifics appeared in other than the politeness meaning and for the one or two sentences with overt honorifics, the participants did not select honorific forms separately on each occasion. They seem to have applied what I call (Neustupny 1979) a Situational Set Rule. This rule is applied only once, normally at the opening point of the situation and selects a particular set of features for the whole situation. In the case of repeated situations such as exemplified in our conversation, there is a reason to believe that the set rule is only applied at the beginning of the first of such situations and is not operated on later occasions. Whenever features of the situation are the same, the set rule applies.4 When a rule is not operated it is covert, automatic, and is not 'significant' for the participants. Note that addressee honorifics are not necessarily as strongly covert in all situations. Okuda-Kawasaki (1977) described a situation in which a group of female university students was communicating with a male student who was only casually known to them. She commented on a number of points of the mixing of plain and polite forms in her data, and it is not surprising that many of these seem to have been

68

J. V. Neustuprty

overt. In her situation the distance between the participants was in the process of being defined and honorifics were used to communicate what was the perception of the situation in the case of various participants. For the conversation analyzed in this paper the usage of 'referent honorifics' was also investigated. The occurrence of honorifics which referred to the 'superior' position of the 2nd person was limited. J2 referred to J1 only once and the verb appeared in its honorifically neutral form (oboete iru). On the other hand J1 referred to J2 six times and out of these six cases used an honorific form three times. It is interesting to note that the three cases were all usages of the 'irregular' (suppletive) honorific forms (kudasai, irassharu, nasaru) while verbs which remained in the neutral form were those which required a morphologically produced 'regular' honorific forms (wakaru, utsuru, ireru).51 suggest that J1 possessed a situational set rule (applicable to a large number of participants and settings) which specified that the irregular and only the irregular honorific forms will be used in such situations. Should the rule be correct, it would imply that for J1 all referent-honorific usage was automatic and hence covert. No doubt, a longer text might have revealed cases when the set rule was temporarily switched off and a referent honorific was used overtly - in the same way as we could note in the case of the addressee honorifics. Unlike 2nd person referent honorifics, 3rd person referent honorifics were used abundantly in the text (remember that the subjects mostly gossipped about their common acquantainces). As one would expect, the usage was much more overt than in the previous two cases.

5. Conclusions The above discussion clearly indicates that the question of overtness is basic for the consideration of the influence of honorifics on Japanese society. Where the honorific meaning is almost 100 percent covert (as in o-cha "tea", go-han "rice", etc.) the possibility of any social significance of such honorifics is remote. I wonder whether there are any honorifics which would be 100 percent overt for an average middle class, Tokyo educated Japanese. Perhaps the very high honorifics such as asobasu "to do" are of this type. However, the more interesting issue is that of the degree of overtness. Only those honorifics which are overtly used are related to processes developing within Japanese society. What are the full conditions

Language and Society: The Case of Japanese Politeness

69

of overtization and covertization? To what extent do they apply to individual honorific categories and forms? What is the extent of individual and other variation? Interviews with native users of Japanese indicate that resistance to honorifics is surprisingly small. One reason for this may be the type of honorific meanings that are carried by honorific forms. It is not only the superior vs. inferior relationship that matters. However, another reason is perhaps that in daily use honorifics are more often covert than one would expect. They are there but, at the same time, they are not there. When covert, honorifics represent nothing more than a formal device without any social significance. The same is true for any other suggested influence of Japanese linguistic features on Japanese society. One often speaks of the lack of the plural, the character of Japanese tenses, the deletion of the subject... This all is empty talk unless one can show in discourse that the features are overt and do intervene with social facts. Of course the same argument can be applied to the same problem in other languages. The concept of covertness and overtness may not solve the whole problem of'language and society' but it has a potential to contribute to its solution in a significant way.

Notes * The author is grateful to Anthony Backhouse, Michael Clyne and Björn Jernudd for valuable comments on the first draft of this paper. 1 Some readers may be tempted to say that the meaning of o- in o-cha is not covert, but from a synchronical point of view simply is not there. However, two circumstances should be borne in mind: (1) The case of o-cha represents an extreme point on a scale of increasing covertness (decreasing overtness) - from this perspective it makes sense to speak of a (close to) zero overtness rather than of absence of the honorific meaning. (2) The possibility of clearly identifying, in the future, conditions under which the prefix becomes overt in the speech of some male speakers remains real - our present capability to elicit overtizing environments should not be overestimated. 2 Concerning the use of honorifics for hypersyntactic discourse function see Neustupny 1978a: 228,1978c: 140, Ikuta 1983. 3 Traditionally, three levels of addressee honorifics are distinguished in Japanese: plain (used in ingroup), polite (used in outgroup), and deferential (used in situations of great distance). The issue is of course quite complicated but since details are not immediately relevant for my argument in this paper, I shall not discuss them here. 4 In the 1960s and 1970s linguists sometimes assumed that the generation of

70

J. V. Neustupny

each sentence implied the application of all rules that 'generate' the sentence. Should this be true, no speaker could speak. The principle of the (Situational) Set Rules explains why we can use language after all. When I say "fill it up" at my local petrol station I apply only one rule, the situational set rule for that situation. I do not select lexical items, grammatical forms etc. for the sentence on the basis of individual rules. The principle applies widely in language and communication. 5 Acutally, the honorific form for ireru was avoided here (cf. Neustupny 1983). "Die sentence was O-satoo nashi? "No sugar?" for the usual O-satoo ireru? or (honorifically) O-satoo o-ire suru? "Will you have sugar?". In any case, a full verb has not been used.

References Fishman, J. A. 1960. A systematization of the Whorfian hypothesis. Behavioral Science 8:323-339. Fishman, J. A. 1980. The Whorfian hypothesis: varieties of valuation, confirmation and disconfirmation. International Journal of the Sociology of Language 26:25-40. Grimshaw, A.D. 1971. Sociolinguistics. Advances in the Sociology of Language, ed. by J. A. Fishman, Vol. 1,92-151. The Hague: Mouton. Harada, S. 1975. Honorifics. Syntax and Semantics 5. Japanese Generative Grammar, ed. by M.Shibatani, 499-561. New York: Academic Press. Hinds, J. 1976. Aspects of Japanese Discourse Structure. Tokyo: Kaitakusha. Hymes, D. H. 1966. Two types of linguistic relativity. Sociolinguistics, ed. by W.Bright, 114-158. The Hague: Mouton. Ide, S. 1982. Japanese sociolinguistics: politeness and women's language. Lingua 57:357-85. Ikuta, S. 1983. Speech level shift and conversational strategy in Japanee discourse. Language Sciences 5.1:37-53. Martin, S. E. 1964. Speech levels in Japan and Korea. Language in Culture and Society, ed. by D. Hymes, 407-15. New York: Harper and Row. Minami, F. 1974. Gendai Nihongo no Koozoo [The Structure of Modem Japanese], Tokyo: Taishuukan. Neustupny, J.V. 1978 a. Post-Structural Approaches to Language. Tokyo: University of Tokyo Press. Neustupny, J.V. 1978b. Japanese honorifics: an outline for the teacher. Japanese Language Teaching Newsletter 2:1-10. Neustupny, J.V. 1978 c. The variability of Japanese honorifics. Proceedings of the Symposium on Japanese Sociolinguistics (The Summer Institute of the Linguistic Society of America, The University of Hawaii, July 28-29,1977), 125-150. San Antonio: Trinity University. Neustupny, J.V. 1979. Gengo koodoo no modern [Models of language behaviour]. Gengo to koodoo, ed. by F.Minami, 33-66. Tokyo: Taishuukan. Neustupny, J.V. 1983. Keigo kaihi no sutoratejii ni tsuite [On strategies of honorific avoidance]. Nihongogaku 2:62-7.

Language and Society: The Case of Japanese Politeness

71

Okuda (Kawasaki), A. 1977. Desu-Masu in Conversation. Unpublished manuscript. Shibata, T. 1972. The honorific prefix "o-" in contemporary Japanese. Linguistic Communications 7:29- 77. Steinberg, D.D. 1979. Language, thought and culture: a mentalistic integration. Working Papers in Linguistics, University of Hawaii 11.3: 87-110.

Questions and Non-Answers J.B.Pride Victoria University of Wellington Wellington

This paper attends mainly to some common ways of not answering questions. It is in three parts. In Part One, the sometimes inhibiting effect of what may be intended as quite innocuous information questions is discussed (from largely cross-cultural points of view) with reference to the prepositional content of the questions themselves. In Part Two, consideration is given to the varied illocutionary and perlocutionary force of information questions generally, in the contexts of ongoing discourse. In Part Three, the paper proceeds in more detail to discuss nine strategies (sorted into the three groups: 'lies', 'refusals', and 'evasions') for withholding information which has been asked for. The strategies are: lies; unqualified refusals, objections, and 'remedied' refusals; unclear responses, 'not answering the question', implicit responses, delayed responses, and forestalling the question. Emphasis is placed upon misunderstandings (particularly in cross-cultural contexts) which tend to arise from the use or non-use of the strategies, and from the means of expression employed. On the express train to Lublin, a young man stopped at the seat of an obviously prosperous merchant. 'Can you tell me the time?' he said. The merchant looked at him and replied: 'Go to hell!' 'What? Why, what's the matter with you! I ask you a civil question in a properly civil way, and you give me such an outrageous rude answer! What's the idea?' The merchant looked at him, sighed wearily, and said, 'Veiy well. Sit down and I'll tell you. You ask me a question. I have to give you an answer, no? You start a conversation with me about the weather, politics, business. One thing leads to another. It turns out you're a Jew - I'm a Jew. I live in Lublin - you're a stranger. Out of hospitality, I ask you to my home for dinner. You meet my daughter. She's a beautiful girl - you're a handsome young man. So you go out together a few times - and you fall in love. Finally you come to ask for my daughter's hand in marriage. So why go to all that trouble? Let me tell you right now, young man, I won't let my daughter marry anyone who doesn't even own a watch!!'

74

J. B. Pride

1. Prepositional Content In a footnote in Schegloff (1968) the story serves to illustrate how 'conversational oaks may out of acorns grow'. For our purposes it illustrates equally well how even the most innocuous of enquiries may prove to be highly sensitive to situational context, immediate or (as in the story) otherwise. It also reminds us how, in 'real life', it may be quite a difficult business to open a conversation with a stranger; all the more so when that person is someone from a different language and cultural background. What topics can be used to break the ice? What topics constitute social taboos? Personal questions such as, for example, 'Are you married? (How come you aren't?)', 'Do you have children? (And if not, why not - or if only one, why not more?)', 'How much do you make a year?', and so on, which would be offensive in English, are often asked in Tagalog, and by Filipinos in English upon their initial contact with a foreigner, as a sign of interest. Questions like 'What is the average rainfall in the Philippines?', or 'What is the population of Manila?' on the other hand, 'do not interest the average Filipino, and they embarrass him because he does not know the answers' (Goulet 1973). The problem of what to talk about in such cases directly arises from the character of cross-cultural differences generally. 'In Tagalog, conversation openers tend to be related to home and family life. There is a tendency to trace one's roots, one's family and regional background ... It seems an individual is important not so much for himself but for his family ties ... In contrast, the American ice-breaker 'What du you do?' reflects American achievement orientation - doing is more important than being, and it is not crucial to learn about another's family background.'1 Even conventions of phatic communion (usually thought of as speech used to make social contact, without necessarily conveying a particular meaning) differ greatly from one society to another. There was the recently-arrived Hindi-speaking student at a northern English university whose bewilderment was all too apparent: 'You know Mr. Pride I said to this man "Where are you going?"; and do you know what he said to me? "Mind your own bloody business!" he said.' But the harmful effects of getting it wrong can be much more severe than this, not least when related to the choice of propositional content. Using real data, Basso (1979) draws a graphic picture of the kinds of traumatic experience frequently undergone by Western Apache Indians when subjected to Anglo-American greeting routines. Towards the end of a long ritual joking episode in which 'the Whiteman' is being very

Questions and Non-Answers

75

effectively parodied we get this piece of'Whiteman' language: 'I don't know. Maybe you get sick. Maybe you don't eat again long time?'. Basso comments: 'Most Western Apaches hold firmly to the understanding that talking about trouble and adversity can increase the chances of its occurrence ... Anglo-Americans are observed to speak about such matters often and in what Apaches regard as a supremely casual fashion, a practice that gives rise to the bizarre impression that Whitemen are eager to experience hardship and disaster.' Often enough of course the problem may be one of distinguishing phatic from other types of communication in the first place. Take for example the job interview. In the following excerpt from the data presented by McGregor and Williams (1982) an Indonesian job applicant has interpreted his Australian interviewer's questions as merely phatic, when their real intention had been to seek answers of commitment to the firm: A: Right, O. K. So that's really why you want the job. It's for travelling. B: Yes, I love travelling. Cross-cultural problems with the propositional content of questions can arise no less in respect of their sheer scope, their degree of specificity, the difference for example between 'Tell me something about the festival' and 'What did they play at the concert last night?'. Thus: 'Connected with this whole phenomenon of Aboriginal responses to questioning, is the fact that they resist another person's efforts to get specific answers. To demand specificity is often considered rude . . . For example, if you try to get specific information about the next high t i d e . . . one will normally get very vague information. And if you persist, you will be met by quite definite passive resistance or a bored expression' 2 (Harris 1978:152). In the same part of the world, among South-East Queensland Aborigines local events are not commonly described for the benefit of those not present at them. That is to say, while it is possible to witness events one does not generally report on them or respond to questions about them; one is more likely to refer to a communal camp verdict on an event than to give details about the event itself (Eades 1982 a). The underlying issue in this last example is that of rights to knowledge, all knowledge in Aboriginal society being considered to be an inalienable part of relationships between people, having no value of its own outside of those relationships 3 . "This is where a lot of White researchers make a big mistake. They go to a community and ask "Can

76

J. B. Pride

anyone here speak the language?", or "Does anyone know the story of what happened on such and such an occasion," and people say "I don't know" or "couldn't say". The White person often misinterprets this to mean - "there is no one here with any knowledge on that topic". Whereas what the Aborigines mean is "we can't give you that sort of information - we don't know you - we don't know what you'll do with the information" . . . etc.' (Eades 1982 b). But if life is difficult for the researcher it is usually much more difficult for the teacher: "The knowledge which is most prized is not open. It is vested, not in teachers, but in persons recognised as wielding power and social control in the community. To possess this knowledge is to hold it in trust. To release or to discover it inadvisedly is to occasion shame. The knowledge will not be released to children nor to those who ask questions. Many of the sociolinguistic rules of speech use in Aboriginal communities are concerned with the exercise of due discretion with regard to closed knowledge and its holders. The school is likely to assume a totally different attitude to knowledge from that of the Aboriginal community, and Aboriginal pupils may, in the school context, appear lacking in a spirit of inquiry and lacking in motivation to master decontextualized knowledge' (Malcolm 1982). Even in the classroom, to get an answer from one's pupil may be a rare privilege, not a right, and may depend entirely upon what exactly is being asked and how carefully one goes about asking it!

2. Illocutionary and Perlocutionary Force One encounters many designations for types of question in the literature on speech acts and discourse: there are so-called 'real', 'examination', 'rhetorical', 'starter', 'triggering', 'confirmation' questions, and many others. All however are requests in one way or another, and one might well agree with Labov and Fanshel (1977) that all requests go in any case through what amounts to an initial (perhaps quite subliminal) filtering process in the listener's mind which asks in effect: 'Is this a request for action or merely for information? Am I being asked to do something, or not?'. Very often of course there is no problem, as in those settings where information-cum-action requests are fully understood as such, as likely as not quite routinely responded to in the same manner. One such setting is the shop counter, where highly elliptical questions often save time and patience. Merritt (1976) gives the nice example 'Do you have

Questions and Non-Answers

77

coffee?' - 'Milk and sugar?', and points out that it may be far from easy to decide what exactly has been left out, or if in fact there is a full form at all. His own fertile imagination provides the following possibilities, but there may be others: 'Yes we do. Would you like some?' 'Yes please' - 'Would you like . . . ? ' ; or'... if you do I'd like some' 'Yes we do. How would you like yours, w i t h . . . ? ' ; or 'Yes we do. I assume you'd like a cup. How would you like yours, with . . . ? ' ; or 'Yes we do' - 'All right, can I have a cup please?' - 'O. K. You want . . . ?' (326). What is clear in all cases is that the information being provided by the assistant's 'Milk and sugar?' is multiple, answering to various levels of presupposition implicit in the customer's request - this being itself a request for both information and action. Misunderstanding can however go deep, even in the most familiar exchanges, perhaps never surfacing in the minds of the protagonists themselves. Take the following piece of dialogue, reported and discussed at length by Tannen (1979), between an American wife and her Greek-American husband: Wife: John's having a party. Wanna go ? Husband: O.K. (later) Wife: Are you sure you want to go to the party? Husband: O. K. Let's not go. I'm tired anyway. The wife's initial 'Wanna go?', intended as a straight information question, was clearly heard as an action request, and was responded to accordingly. However, the husband's bare 'O. K.' must have been heard in turn as showing lack of enthusiasm, so a second information request was put to him. But this, too, was heard as an action request, this time signalling a change of mind, and was again responded to accordingly. We might say that the information sought by the American wife is in fact being withheld by her Greek-American husband on account of its - to him - supposed irrelevance4. On questioning a carefully selected body of informants afterwards about this single piece of dialogue, Tannen found that Americans with no Greek ancestry consistently preferred literal or direct interpretations of the wife's questions; that Greeks preferred such indirect interpretations as 'My wife wants to go to this party, since she asked - I'll go to make her happy', etc.; and that Greek-Americans stood midway, or vacillated. Tannen goes on to refer to the more extreme case of the Gonja people in Ghana who (according to the researches of Goody 1978) do not ask information questions in teaching and learning situations. In Goody's own words: 'Gonjans

78

J. B. Pride

are trained early on to attend above all to the command function of questioning - the information question simply hasn't got a chance!' (Goody 1978:40). But of course the information question format alone can easily serve to bind together the whole of a normal discourse. Many everyday interchanges take the form of numerous overlaps of the three elements questions (Q), responses to questions (R), and evaluations of those responses (E), in such a way that a great deal of what is actually said takes the form of unbroken successions of plain questions. Here is an example from Mishler (1975): 1. A: What do you want? Is this the same? 2. B: Yeah 3. A: What do you want to know? Do you have a question? 4. B: Is that the same? 5. A: Is that the same as this? 6. B: Yeah is that the same as that 7. A: Do you think it is? 8. B: No 9. A: O.K. You're right

(Q) (R) (E/Q) (R/Q) (E/R/Q) (E/R/Q) (E/R/Q) (E/R) (E)

An annotation of the six evaluative moves claimed here by Mishler might go something like this: in 3, A tacitly accepts B's reply to one of his questions by not commenting, but repeats and elaborates his own unanswered question; in 5, A's reply to B's question comments on its lack of comprehensibility by rephrasing it; in 6, B accepts A's rephrasing with 'yeah' and by repeating the question; in 7, A now queries the appropriateness of B's repeated question by turning the question round; in 8, B accepts A's judgement in doing so by giving A's question a straight reply; and in 9, A approves of B's 'no'. 'Withholding information which has been requested' is by no means therefore a simple question of 'not answering questions'. 'Refusing to cooperate' might be a more pragmatically Gricean way of putting it.

3. Nine Strategies Those we shall look at may be placed into three main groups, based on the intentions of the responder.

Questions and Non-Answers

79

(a) 'Lies':

intent to hide reluctance to answer by misrepresenting facts (or reasons for not answering) - strategy 1 (b) 'Refusals': intent to reveal reluctance to answer - strategies 2,3,4 (c) 'Evasions': intent to hide reluctance to answer by taking advantage of ambiguities in the situation - strategies 5,6,7, 8,9

1. Lies Lying is not the same thing as not answering. The question is acknowledged and responded to in an entirely cooperative manner - except in respect of the truth. Lies may give false information, or false reasons for not supplying information (in the latter case it may be explicitly and directly, as with 'I don't know'; explicitly and indirectly, as with 'How do you expect me to know?'; or more implicitly, by pretending one did not hear the question, did not hear it as a question, etc.). Cultural, social, and individual differences abound when it comes to motivations for lying. One of the most widespread (and often quite normal and expected) is that of saying what one thinks the other person wants to hear.

2. Unqualified refusals Straight refusals are probably not very common, certainly not to questions of a potentially incriminating nature. Changing the subject is a different matter, and need not be the transparent unsophisticated device it may seem to be; it probably constitutes a very real part of the acquisition of communicative competence as one grows older. Alternatively, refusals can take the form of silence . . . The uses and significances of silence vary greatly, however, hence have to be understood in much broader contexts before turning to our present one. Sheer tolerance for silence for example may range from virtually nil in one society (or part of one society) to easy acceptance in another - where mutual silence may for example be one way of marking shared familiarity and intimacy ('Some Danes appear to "nourish" a silence as one might appreciate a cosy fire': Reisman 1974). The Paliyans of south India 'communicate very little at all times and become

80

J. B. Pride

almost silent by the age of forty. Verbal, communicative persons are regarded as abnormal and often as offensive' (Hymes 1972). In contrast, Dr. Johnson once said of Oliver Goldsmith: 'Goldsmith had no settled notions on any subject; so he always talked at random. It seemed to be his intention to blurt out whatever was in his mind, and see what would become of it'. Silence as a response to information questions will surely draw much of its significance from cultural and other group norms of many different kinds. In 'Western' societies silence can certainly be 'heard' as a refusal to answer. It can alternatively (or additionally) indicate incomprehension, astonishment, embarrassment, feelings of inferiority, confusion, unawareness that a question has been asked at all (or that one is oneself the intended addressee), a temporary - or not so temporary - delay in answering, and so on; or it may just accompany a non-verbal response ('his looks spoke volumes'). The sociolinguistic literature touching on silence is quite extensive; see for example Basso (1970), Darnell (1979), Dumont (1972), Eades (1982 a), Harris (1980), Gumperz, Jupp and Roberts (1979), John (1972), Labov (1969), Malcolm (1980-82), Metge and Kinloch (1978), Philips (1972, 1976). Philips (1976) is particularly revealing on the roles of silence and delay in responses to information questions (see strategy no. 8 below) in a culture in which tolerance for silence is high.

3. Objections These do not necessarily convey or accompany some kind of refusal; after all, one might wish to object to a question before going on to answer it - having lodged one's protest, and having perhaps, gained extra clarification, time, etc. in the process. Objections may be raised on the basis that one or more of the preconditions of the Labov/Fanshel (1977) Rule 1 have not been met, or that the question is 'loaded', 'complex', ambiguous, or not understood.5 Note that a very slightly modified version of the Labov/Fanshel rule could read: 'If A requests B to answer a question X at a time T . . . ' etc. Accordingly, objections directed to the first pre-condition of the rule might take such forms as: 'I've already told you', 'What's that got to do with it?', etc. To the second: 'How should I know?', 'You expect me to answer that in five minutes?', etc. To the third: 'I don't have to tell you that', 'Why should I tell you?', etc. And to the fourth: 'Mind your own business!', 'What gives you the right to ask?', etc. More like-

Questions and Non-Answers

81

ly, however, objections will be a great deal more implicit, and may refer simultaneously to more than one of the preconditions: 'Why don't you ask Fred?'. Or be wrongly understood to be such, as in the case of the Black American pupil who answered his teacher's question with 'I don't know', spoken with the rising intonation associated in the White teacher's mind with objections-cum-refusals, nevertheless meaning 'I need some encouragement' (Gumperz and Tannen 1979). 'Loaded' questions ('When did you stop beating your wife?') and 'complex' questions ('Are you one of those unthinking Liberals?') could well be objected to with 'What do you mean?'. Ambiguous or otherwise unclear questions might attract either this or the ubiquitous general-purpose 'Why?' especially if the ambiguity is of the 'action or information only?' type. Thus, 'Why?' in answer to 'When do you plan to come home?' could mean any of (or more than) the following: 'Why are you asking? Do you want me (or not want me) to come home, and if so why? Why do you think I'm in a position to come home? What makes you think I can tell you? Why should I come home? Why should I tell you? What gives you the right to ask me to come home? Or to ask me such a question?'. On top of all this, there is the matter of situational appropriateness: 'Why?' could mean 'Why are you asking me that now, or here, or in the presence of these people . . . etc.?'. No wonder 'Why?' is so often countered with 'What do you mean 'Why?'!'.

4. Remedied refusals Goffman (1972) discusses three types of 'remedied refusals': accounts or explanations ('I don't know', etc.), apologies (tending to occur with accounts), and requests ('Do you mind if I don't answer that?', etc.) A fourth could be added, namely offers of alternatives ('I'm afraid I can't answer that one, but I can tell you about the time when...'. And fifth, in effect a special kind of alternative: 'I don't know, but this is what X thinks or has said or written on the matter' - the latter amounting perhaps to what the questioner as examiner had wanted or expected to elicit in the first place . . . Some questions themselves offer (or are phrased as offering) specific possibilities for remedied refusal ('I don't suppose you could tell m e . . . ' , etc.), or rather - more exactly - enquire about the 'felicity conditions' for saying or doing things (see Brown and Levinson 1978:137).

82

J. B. Pride

5. Unclear responses A brief comment may suffice on just one aspect of this limitless topic, relating to misunderstandings which can arise cross-culturally from the mistaken assumption that one's question has been answered in a deliberately vague or roundabout manner. A fundamental problem is that normal modes of directness and indirectness vary considerably from one language and society to another. The Japanese language, for example, is 'one of indirect social forms, indicative of politeness, and status identifiers . . . utilising circumlocutions so that the intended object of the particular speech is reached by a circular rather than a linear route' (Lyman 1978: 285). Accordingly, in the conversational English of second-generation American-born Japanese Nisei, 'circumlocutions and indirect speech are regular features' (287). The Nisei 'do not immediately approach the main point of a conversation', so that it is the duty of the listener to ascertain the important point from the context and his knowledge of the speaker . . . Nisei are troubled by direct questions; they may refuse to answer, change the subject, or subtly redirect the conversation back to its concentric form . . . ' (288). In a similar vein, Gumperz and Tannen (1979) point out: 'In making an argument, Indian speakers take great care to formulate the background for what they are going to say, and use increased stress to mark this background information, then shift to low pitch and amplitude on their own contribution. The strategy is something like raising one's voice to get attention, then stating one's message in a low voice' (321). Gumperz, Jupp, and Roberts (1979) show how, in job interviews in Britain, the Asian interviewee may well be wrongly perceived as evading the question that has been asked, for just these reasons.

6. 'Not answering the question' 'How old are you?' - 'Don't worry, they'll let me in to Jimmy's.' Weiser (1975) uses this as an example of 'selection by reply', in which the addressee has interpreted the intention behind the question and answered it (instead of, perhaps, asking 'Why?' . . . ) . In doing so he may very likely not be seeking to evade anything at all. Alternatively, however, he may be aiming to hide his proper understanding of the intention (even the literal meaning) of the asker of the question, giving instead a direct answer to what he knows not to be that real intention. There are

Questions and Non-Answers

83

those cases too where the literal meaning of the question is answered, although this may be well understood not to have been in line with the intention of the questioner: 'Have you talked to Harry about my suggestion as you promised you would?'. In a sense, 'selection by reply' when used as an evasion or avoidance strategy takes advantage of (or pretends to take advantage of) ambiguities in the question. This may include pretending that the question is hardly a real question at all, rather a 'rhetorical' question, a mere greeting formula ('How are you?'), and so on; that the scope of the question is narrower than in fact it is; or that all that is being asked for is some comment associated with the general drift or content of the question ('Are you going to do something about the lighting?' - 'Ah yes, lighting is of course important, I've always thought so . . . ' ) . The latter ('triggering') function of questions may of course be their real purpose anyway ('When on earth are they going to do something about the lighting?'), in some contexts highly conventionalised, as in many interviews (on the radio, with the press, for jobs, etc.). There are, one suspects, many cross-cultural bunkers liberally strewn throughout this whole area. Apparently deliberate cases of 'not answering the question' may be traceable to differences in well-established cultural norms for the signalling of intentions behind questions. Thus, in Gumperz, Jupp and Roberts (1979) the job interview question 'What exactly do you do in your present job?' was not duly understood by the Indian applicant to mean 'Tell me about your present duties and how these relate to this job your are applying for'. To the questioner, Mr. Sandhu's reply 'does not sound immediately relevant or sufficiently explicit' (28). Much the same situation prevailed in the case of Polynesian interviewees in Wellington, New Zealand, who (in marked contrast to native New Zealand applicants) consistently failed to pick up slight but conventional cues from their interviewers to elaborate on their answers in their own way - so as (crucially) to show interest, intelligence, fluency, etc. (Harris 1983). In some ways however the most revealing examples of cultural relativity in the use of triggering questions (or strategies: they may not always be expressed as questions) are those in which the questioner from another culture has perceived and fallen in with the norms of the culture of the person addressed, succeeding thereby in penetrating what might otherwise seem to be an impenetrable wall of taciturnity. Thus, of the Aboriginal people of South-East Queensland, Eades writes: 'In the most common triggering strategy the person desiring substantial information makes a relevant statement, presenting something he al-

84

J. B. Pride

ready knows about the topic. This is followed often by silence and then by the knowledgeable person talking on that topic, if he desires' (Eades 1982a: 76). Similarly, from the same part of the world, reflecting the experience of the researcher when talking informally with her young informants: 'The experimental task required children to discriminate between local and non-local birds on the basis of coloured photographs . . . Not surprisingly, the more traditionally oriented children showed greater skill in the task, and also volunteered a great deal of extra information on bird life not called for by the task itself' (Kearins 1982:15). We are reminded of the care taken by Labov (1969) and other linguists in eliciting natural responses from adolescent Black Americans, responses which may well be more substantial, logical, incisive, altogether to the point, than the lengthy verbalisations of socalled 'advantaged' groups from the same society - who have their own ways of 'not answering the question'.

7. Implicit responses Where the last strategy ('not answering the question') could be said to take advantage of ambiguities (or pretended ambiguities) in the question, the present strategy introduces a note of ambiguity, even vagueness, into the response itself. The two strategies are alike however in utilising, for their own purposes, perfectly normal, productive, and economical devices for the exchange of information: in the one, the addressee short-cuts his way to his questioner's implicit intentions; in the other, the questioner perceives the implicit meaning in his addressee's answer. 'How old are you?' - 'Don't worry, they'll let me in to Jimmy's', and 'Are you stopping in today?' - 'Well, the forecast's fine', and that is to say alike in their reliance on shared knowledge and understandings. In fact both could serve to illustrate both strategies together. It is for this reason that both devices are so amenable for use as strategies of evasion: when 'not answering the question' one pretends to have misunderstood the intentions of the question itself; when giving an 'implicit response', one pretends to have misunderstood the ability of the questioner to infer one's own intentions.

8. Delayed responses Many societies tolerate delay in responses to information questions (that is to say in the spoken language, and allowing 'responses' to

Questions and Non-Answers

85

cover provisionally acceptable replies pending acquisition of the desired information) no more easily than silence itself. The ambiguities which can arise are of the order 'Is he going to answer or not? Does he want to answer? Has he understood the question?' and so on. Delayed responses, or what appear to the questioner to be delayed responses, can be exact and explicit ('Well now let me think about this . . . ' ) hesitant and verge on the incoherent ('I would say we are um, let me put it this way . . . ' ) , take the form of 'side-sequences' ('It depends what you mean b y . . . ' ) , adopt modes of indirectness before 'getting to the point' (see also strategy no. 5), come back to questions after much intervening discourse other than side-sequences (on this, see especially Philips 1976), or simply take the form of silence (with perhaps absence of even a non-verbal response). Motivations for delay must be legion. One which is probably more common than might be supposed (at any rate in school settings) has been referred to as 'shadowed replying, in which one delays one's reply until one can give it in the 'shadow' of some other speaker' (Malcolm 1979).

9. Forestalling the question There are various ways of heading off a question which one can see coming: talk about the topic (or otherwise behave) in such a way that the prospective questioner can perceive the unsatisfactory nature of one's likely response; steer the conversation in a direction which makes the question difficult to fit in at all; take advantage of any advance warnings already given in question form ('What are you doing tomorrow?'); ask the question oneself, so as to learn more about the other's opinions, intentions, etc.; and so on. Some of these tactics are of course akin to 'changing the subject' (under strategy no. 2), in this case however before the question is asked, not after. The ambiguity lodged in the questioner's mind, if he perceives any at all, will amount to 'Does he know what question is coming? Is he warning me off? Is this a refusal?' etc. - more likely a sense of ill-defined frustration. Strategies of these sorts are rarely employed singly: they may be sequential; simultaneous, perhaps multiple, perhaps indistinguishable ('How should I know?' as any or all of the following: transparent lie about own knowledge, refusal, objection to the Labov/Fanshel second precondition, remedied refusal, unclear in any case - inviting such as 'What do you mean how should you know?', implicit, a good 'delaying tactic', and no doubt more besides, given the right circumstances); one

86

J. B. Pride

used to achieve another or to avoid another (forestall as a way of refusing or not refusing); mutually contradictory in the one response; demanding on both participants differently ('Are you stopping in today?' - 'Well, the weather's fine' can be quite ambiguous in both parts, allowing strategies 6 and 7 at the least); and tied up in other ways no doubt. Indistinct, indirect, implicit, are the orders of the day. Levels of awareness are probably not high at all (what actual prospects are there for participant analysis of all such considerations in code-switching?). At what levels do we find universals? Where rights and obligations to give and take the knowledge asked for are concerned, contexts of situation, linguistic and non-linguistic means of expression, not far down: all or much is likely to seem relative on inspection of cases. The strategies themselves could be universal in one employment and one form or another. Universal or relative, popular metalinguistic expressions ('missing the point', 'dodging the issue') may be worth studying as 'talk about talk'; on the periphery of which, might the terms 'strategies' and 'tactics' offer here any useful understood contrast? Why not, for example, refer to the three groupings 'lies', 'refusals' and 'evasions', as strategies, the rest (admitting many more?) as tactics?

References Basso, K,H. 1970. 'To give up on words': silence in Western Apache culture. Southwestern Journal of Anthropology. Basso, K. H. 1979. Portraits of 'the Whiteman': linguistic play and cultural symbols among the Western Apache. Cambridge University Press. Brown, R. and Gilman, A. 1960. The pronouns of power and solidarity. Style in language, ed. by T. A. Sebeok, 253-76 M. I.T. Press. Brown, P. and Levinson, S. 1978. Universals in language usage: politeness phenomena. Questions and politeness: studies in social interaction, ed. by E. N. Goody, 56-289. Cambridge University Press. Darnell, R. 1979. Cree interactional etiquette. Working Papers in Sociolinguistics55. Austin, Texas: Southwest Educational Development Laboratory. Dumont, R.V. 1972. Learning English and how to be silent: studies in Sioux and Cherokee classrooms. Functions of language in the classroom, ed. by C.B.Cazden, V.John and D.H.Hymes. New York: Teachers College Press. Eades, D. 1982 a. You gotta know how to talk . . . : information seeking in South-East Queensland Aboriginal society. Australian Journal of Linguistics, 2,1. 61-82. Eades, D. 1982b. English as an Aboriginal language. Aboriginal languages and the question of a national language policy, ed. by Jeanie Bell. Aboriginal languages Association, August 1982.

Questions and Non-Answers

87

Edwards, J., Rosberg, M. and Pryme Hoy, L. Conversation in a West Indian Taxi: an ethnolinguistic analysis. Language in Society 4.295-322. Geertz, C. 1960. The religion of Java. Free Press. Goffman, E. 1972. Relations in public. Penguin Press. Goody, E. N. 1978. Towards a theory of questions. Questions and politeness: studies in social interaction, ed. by E. N. Goody, 17-43. Cambridge University Press. Goulet, R. M. 1973. A comparison of conversational openers in English and Tagalog. Parangal Kay Cecilio Lopez. Special Monograph Issue of Philippine Journal of Linguistics 4. Gumperz, J.J. and Tannen, D. 1979. Individual and social differences in language use. Individual differences in language ability and language behaviour, ed. by C.J.Fillmore, D.Kempler and W.S.-Y.Wang. New York: Academic Press. Gumperz, J., Jupp, T. and Roberts, C. 1979. Crosstalk. Southall, Middlesex: National Centre for Industrial Language Training, Havelock Centre. Harris, S. 1978. Culture and learning: tradition and education in Northeast Amhem Land. Darwin: Northern Territory Dept. of Education. Harris, J.L. 1983. Linguistic problems in the interview situation. M.A. thesis, Victoria University of Wellington. Hymes, D. H. 1972. Models of the interaction of language and social life. Directions in sociolinguistics, ed. by J. J. Gumperz and D. H. Hymes. Holt, Rinehart and Winston. John, V. 1972. Styles of learning - styles of teaching: reflections on the education of Navago children. Functions of language in the classroom, ed. by C.B.Cazden, V.John and D.H. Hymes. New York: Teachers College Press. Kearins, J. 1982. Cognitive strengths of Australian Aboriginal children. Wikaru, 11.2-22.

Labov, W. 1969. The logic of nonstandard English. Georgetown Monographs on Language and Linguistics 22. Labov, W. and Fanshel, D. 1977. Therapeutic discourse: psychotherapy as conversation. New York: Academic Press. Lyman, S. M. 1972. Generation and character: the case of the Japanese Americans. East across the Pacific, ed. by H. Conroy and T. S. Miyakawa. Santa Barbara, Cal.: American Bibliographical Center Clio Press. Malcolm, 1.1979. The West Australian Aboriginal child and classroom interaction - a sociolinguistic approach. Journal of Pragmatics 3. Malcolm, I. 1980-82. Speech use in Aboriginal communities: a preliminary survey. Anthropological Forum, 5,1. 54-104. Malcolm, I. 1982. Communication dysfunction in Aboriginal classrooms. Aboriginal education, ed. by J.Sherwood, 153-172. Creative Research, North Perth, Western Australia. McGregor, A.L. and Williams, R.T. 1982. Cross-cultural communication in the employment interview. Paper presented to the Seventh Annual Congress of the Applied Linguistics Association of Australia, 21-28 November 1982. Merritt, M. 1976. On questions following questions (in service encounters). Language in Society 5. 315-358. Metge, J. and Kinloch, P. 1978. Talking past each other. Victoria University of Wellington Press.

88

J. B. Pride

Mishler, E.G. 1975. Studies in dialogue and discourse: types of discourse initiated by and sustained through questioning. Journal of Psycholinguistic Research 4,2. Philips, S.W. 1972. Participant structures and communicative competence. Functions of language in the classroom, ed. by C.Cazden, V.John and D.H.Hymes, 370-394. New York: Teachers College Press. Philips, S.U. 1976. Some sources of cultural variability in the regulation of talk. Language in Society 5. 81-96. Reisman, K. 1974. Contrapuntal conversations in an Antiguan village. Explorations in the ethnography of speaking, ed. by R.Bauman and J. Sherzer. Cambridge University Press. ScheglofT, E.A. 1968. Sequencing in conversational openings. American Anthropologist 70.1075-95. Tannen, D. 1979. Ethnicity as conversational style. Working Papers in Sociolinguistics55. Austin, Texas: Southwest Educational Development Laboratory. Tanner, N . 1967. Speech and society among the Indonesian elite: a case study of a multilingual community. Anthropological Linguistics 9,3.15-39. Weiser, A. 1975. How not to answer a question: purposive devices in conversational strategy. Chicago Linguistic Society Regional Meeting Papers, 11.

Riddling and Lying: Participation and Performance John R.Rickford Stanford University Stanford, California

Introduction Ferguson's work on proverbs and riddles (Ferguson and Preston 1946 a, 1946b, Ferguson and Echols 1952, Ferguson 1954) is one of the least-known aspects of his scholarly activity as a linguist, but as he has noted (p.c.), it 'shows one aspect of my interest in language and society.' It certainly qualifies as the earliest published manifestation of this interest. According to Dil's (1971) bibliography, Ferguson completed his dissertation on Bengali and had his first paper published ('A Chart of the Bengali Verb') in 1945; the papers on Bengali proverbs and riddles were published in the very next year. The practice of following or combining a study of a people's language with a study of their folklore or verbal art is less common now than it was in the first half of this century, but in adhering to it, Ferguson was following the anthropoplogical linguistic tradition of Herder, Boas and Sapir. As Bauman (1977) has noted, these legendary predecessors of Ferguson's represented an 'integrated vision of language, culture and verbal art,' one which faded as linguistics asserted its 'disciplinary autonomy'. I share Bauman's view, however (ibid: 17), that a unified perspective is beginning to re-emerge through work in the ethnography of speaking by linguists, anthropologists and folklorists. The scope of this area has been outlined by Hymes (1964:101): In one sense this area fills the gap between what is usually described in grammars, and what is usually described in ethnographies. . . . In another sense, this is a question of what a child internalizes about speaking, beyond rules of grammar and a dictionary, while becoming a full-fledged member of

90

John R. Rickford

its speech community. Or, it is a question of what a foreigner must learn about a group's verbal behavior in order to participate appropriately and effectively in its activities. The ethnography of speaking is concerned with the situations and uses, the patterns and functions of speaking as an activity in its own right.

The correctness of Bauman's assessment is shown by Ferguson's work itself, which includes, in its most recent phases, attention to topics like 'The collect as a form of discourse' (1976) and 'Sports announcer talk' (1983) which fall squarely within the ethnography of speaking. In this paper, I will attempt to follow Ferguson's earliest and most recent phases by studying the verbal artistry of a people whose grammatical usage I have previously described (1974,1985), but doing so within the framework of an ethnography of speaking. My focus will be on the nature of riddling and lying on one of the Sea Islands off the coast of South Carolina. Collections of Sea Island folklore like Parsons (1923) include a rich stock of riddles and lies, but they are abstracted from the setting in which they were performed. There is also an extensive literature on the structure, definition and significance of riddles and tall tales (see, for instance, Potter 1950, Taylor 1951, Georges and Dundes 1963, Scott 1969, Maranda 1971, Thomas 1977, Pepicello and Green 1984), but, with rare exceptions (Abrahams 1983, Bauman, in press, Evans 1976) this literature tells us little about riddling and lying as verbal activities. Following the terms of the ethnography of speaking, we wish to specify what one needs to know to participate appropriately in riddling and lying; not just what riddles or lies are like, but how instances of each genre are performed and responded to in socio-cultural context. Riddles and lies are only two of the genres which occur in a speech event referred to in the Sea Island area as 'telling riddle', 'telling lie' or, most generally, 'telling story.'1 Other genres which occur in this speech event include: folk-tales (ranging from Brer Rabbit and other animal stories to 'Preacher' stories; see Greene and Sharpe 1984); 'smutty' or stag stories; and short rhymed verses or 'toasts.'2 Although some of the individual genres still survive - even flourish - in parts of the Sea Island area, 'telling story' as a whole was, by native accounts, more frequent in the 'old days,' that is, four or five decades ago. At that time, there were more people in the area (1,000 or more on the island on which I worked, compared with less than 100 today), and they would often get together after farming or fishing all day, and 'tell story' for entertainment. Another popular setting for this speech event was during the packed boat-trip to Savannah, Georgia to sell produce, or in Savannah itself, where residents from different is-

Riddling and Lying: Participation and Performance

91

lands, would compete with each other in telling stories as well as selling produce.3 Today, because of ecological and economic factors,4 the population of many Sea Islands has dwindled considerably, and the incidence with which one encounters the traditional linguistic patterns and folklore which made this area such a rich resource for students of AfroAmerican history and culture (Jackson et al. 1972) has diminished. This is not to suggest that intriguing African and Afro-American traditions do not continue on the Sea Islands (see Jones-Jackson 1983), nor that distinctive patterns of verbal artistry do not remain. But among the middle-aged and younger Sea Islanders, television, pop-music and other speech events have replaced 'telling story' as central means of entertainment. To get an idea of what riddling, lying, and other 'telling story' genres were like in their hey-day, one has to depend on the older residents of the area.

The data The data for this study consists of a tape-recording made in 1972 of three older residents on one of the South Carolina Sea Islands participating in a 'telling story' session: Mr. Walters, 71, his wife Mrs. Walters, 69, and their close friend, Mr. Jones, 69. (These names are all pseudonyms.) At the time, they were the only representatives of the over-sixty age group on their side of the island, and constituted a closely-knit, natural interaction group. The data is artificial insofar as it was not a spontaneous 'telling story' session on which I happened to stumble, with tape-recorder conveniently in hand. As part of a project to use the resources of the community to create reading materials for the island's two-room school, I went around recording life-histories, narratives and folklore from the residents of the island. This particular session was arranged by me, and it was clear that the participants had not, in recent times, actively taken part in 'telling story.' Mr. Walters and Mr. Jones kept asserting that they knew 'a whole bunch o' story' but that they had 'gone from them now.' As Mr. Jones explained, his voice trailing off wistfully: 'Since we get old now, we can't remember t h e m . . . we can't remember them There is something else missing from this session, apart from lack of practice. It comes out in Mr. Walter's comment as he tries to remember more riddles:

92

John R. Rickford

I know a whole BUNCH o' riddle, but, you k n o w . . . if I had somebody to START it, who could, ah, dig up my remembrance, and, ah, TUNE me up, WARM me up, and what n o t . . .

Tuning up would, ideally, require more participants, ready with their own stock of stories, riddles, and lies. The content (and the form!) of their 'stories' would remind other performers of related 'stories', which they could then tell in turn; the competitive element would be stronger; and the positive (and negative) evaluations of the audience made more powerfully. In this particular session, there are really only two performers. Mrs. Walters tries throughout to contribute, but aside from introducing one brief riddle which is easily solved, she serves more as audience and evaluator than performer. I suspect that this had less to do with her competence in the relevant genres than with a restriction on performance in 'telling story' to adult males; almost all the traditional stories which I collected on this island were from older men, and Heath (1983: 183), describing the southeastern Afro-American community of Trackton, notes that it was boys who excelled at lies there ('the performance of a "true story" in which they "talk junk."') I myself attempted to play the role of audience and respondent too, as well as my native experience in a similar Caribbean culture would allow, but in this role I was hardly as effective as Mrs. Walters. Limited numbers of participants and lack of practice do not make for the most memorable story-telling session ever. But what resulted from the session that evening, with only two performers and one 'audience,' was quite successful. A total of nine folktales or stories, ten riddles, and four 'lies' was generated, and, for the rest of my stay on the island, the participants frequently talked about the fun they had had during this session and asked to hear the tape-recording of it. Even more significant as evidence of the validity of the data was my almost total inability to determine the direction of the session. I would sometimes use lulls in the session to ask for specific material, but in almost every case, my requests were either ignored, or acknowledged without being acted upon, quite unlike the case with individual interviews. For instance, after Mr. Walters had told a highly successful preacher story, there was a pause of about eleven seconds which I tried to fill with Labov's (1972) 'danger of death' question. Note what happened: 5 JRR.:

Well, Mr. W., you ever, you ever, am, you ever been in a situation where you thought you was close to death? Like, you know, hunting, or out at sea, or anything like that?

Riddling and Lying: Participation and Performance

93

Mr. Walters: Ooh, Yeeaah! JRR: (Warming up) Where you thought you was a/most gone?! (2 second pause.) Mr. Walters: Ah, ah, I gon tell y'all another lil piece about a preacher again. Now, let me see, let me see how that go. . . . Ah! (Breaks into preacher-story) Mr. Walter's reaction here made me aware that my question was inappropriate in several ways. In the first place, I was not a performer, and only performers are allowed to change the current genre (although they can respond to suggestions on this point if they see fit). In the second place, even performers don't shift genres as abruptly as I was trying to do. If A tells a lie, B must respond with another lie. Successive turn-taking in that genre will depend on who knows more instances of that genre and can most swiftly fill in the pauses between successive performances, but a switch to another genre must be naturally engineered. We will see what this involves when we discuss the switch from riddles to lies in this session. Thirdly, my illegal interruption was threatening Mr. Walter's right to perform a story which he was merely perfecting in his head before bringing it unto the floor. Finally, the genre into which I was attempting to channel the session was not even one which normally occurs in 'telling story.' This happened at another point when I asked for proverbs, and was similarly ignored; proverbs do not come up in these sessions unless they occur in a story. I consider these failures on my part, and the fact that I was made to discover them, a victory for the participants, and a vindication of the integrity of the session as an exemplification of 'telling story.' In discussing the data, we have already touched on several general characteristics of 'telling story,' but before going on to examine what is specific to riddling and lying, one other characteristic of the larger speech event in which they occur should be emphasized. The goal of story-telling, like other local speech events, is not just entertainment, nor enlightenment, nor education (though it is all of these), but the earning of positive evaluation and approval.6 This approval derives, in the first instance, from the specific audience before which one performs, and, by extension, from the larger community in which one's reputation is established. Abrahams' (1971:28) characterization of folklore as 'all conventional expressive devices available for performance and the achievement of performer status within a socially bounded group' is relevant here, as is Bauman's (1975: 293) formulation of performance as 'a mode of spoken verbal communication . . . [involving]

94

John R. Rickford

the assumption of responsibility to an audience for a display of communicative competence.' Whether riddling or lying, performers try to match or outdo each other's displays of competence, and to win more positive evaluation from the audience. As we discuss the individual genres, we will pay particular attention to the evaluation process, and the strategies which performers use to win status.

Riddling Riddles are brief, stylized forms which contain a puzzle for respondents to solve. Regardless of whether the surface structure is that of a question or not, every riddle has the underlying form of a question (Maranda 1971:195), or the illocutionary force of a request. Defining riddles in such a way as to distinguish them uniquely from other genres is a challenging exercise. For our purposes, it is enough to note that riddles usually contain at least one proposition or deep structure sentence (Scott 1969:137) which expresses the puzzle and simultaneously provides the clue to its solution in a metaphorical comparison between one object or event and another. Here are a few examples from the recorded session, all of which occur in Taylor (1951): 1. Ten thousand window and one door. (A cast-net.) 2. What come up to the door and don't never come in? (A road.) 3. What round as a dollar and busy as a bee? (A watch.) In riddling, the roles of the participants shift continuously according to whether they act as riddler/performer or respondent.7 The role of riddler is granted by unspoken consent to the first participant who can pose a complete, formally coherent riddle to the assembled group. Exactness of form is important in both riddles and their answers, and participants often spend the moments between the solution of one riddle and the start of another trying to reconstruct all the elements of a riddle which they remember only in part. The first participant to formulate a complete riddle holds the floor, and he or she performs as riddler until the correct answer is given by respondents or revealed by the riddler. Participants who try to hold the floor as they stumble to formulate their riddles may be displaced by others who blurt out complete riddles: Mrs. Walters: What go - what go round the - the - ah Mr. Walters: (Rapidly interrupting) A plate of money you can't count, and cake you can't cut!

Riddling and Lying: Participation and Performance

95

Once a formally complete riddle is on the floor, the business of respondents is to try to answer it. How participants respond right after the riddle is posed provides the first evaluation of the riddle's difficulty (and worth). A respondent might say right away, 'I know that,' and give the correct answer. Riddles that are so readily answered earn little credit or approval, and the strategy of the riddler is to produce riddles which are not immediately known to the respondents, and allow him to hold the floor for a longer time as respondents try to figure it out. Only three of the ten riddles produced in this session were answered in the first turn of respondents. In one case, the person who responded first did not have the exact answer at her fingertips, and had to keep justifying her right to the floor: Mr. Jones:

What holler all night, and never, don't never stop holler? Mrs. Walters: I know that! I know that! Ah - wait - wait - Oh! Treehollow. More typically, respondents will indicate that they do not know the answer. They might then repeat the riddle to themselves, over and over, or ask the riddler to say it again, or try to figure out the answer silently, or admit that they used to know the answer but can't remember it. All of these responses please the performer, but the most positive evaluation in this phase of the riddling exercise is when respondents indicate that they are totally stumped, by making statements like: 'You got me now, with THAT,' or 'You got me,' or 'Got me there' (all involving 'got'). Note that responses of this type are not admissions of defeat or requests for the riddler to provide the answer. They are acknowledgements of the difficulty of the riddle, in response to which the riddler typically offers additional clues. These may involve references to related items in the immediate surroundings, or references to the shared experiences of participants. For instance, after Mr. Jones produced riddle 2 above, it was followed by silence, and he added, 'You come through it just now. You come through it just now.' Another frequent clue is, 'Most everybody got one. Most everybody got one,' indicating, incidentally, that riddles are usually about everyday objects or events, disguised in metaphor to look unfamiliar. Even if respondents indicate that the riddle is difficult to solve, they are still expected to attempt an answer: Mr. Jones:

(In response to a riddle from Mr. Walters): Now, I might say right, and I might say wrong. Mrs. Walters: Well, say it anyhow!'

96

John R. Rickford

In response to wrong answers, the riddler might say, 'Nooo,' following this by repeating the riddle or providing other clues, indicating in either case that the riddle remains unsolved. There is a standing rule that no new riddles will be introduced until the one on the floor is solved. This rule is invoked when Mr. Walters tries to introduce a new riddle (actually, it is the set-up for a lie, the genre to which Mr. Walters wants to switch) while Mr. Jones still has an unsolved riddle on the floor. Mrs. Walters has just suggested one possible answer to Mr. Jones' riddle. Mr. Jones has just answered, 'Nooo,' and Mr. Walters breaks in: Mr. Walters: What's the largest watermelon you ever seen, Mr. Jones?' Mr. Jones answers: I don't know. I don't know. But y'all ain't get THAT bout the ten thousand window and one door.' And Mr. Walters has to admit defeat on Mr. Jones' riddle before he can launch into his own performance. Respondents indicate defeat on a riddle, and request the riddler to furnish the correct answer by saying, 'I crave.'8 Once one respondent says this, the riddler checks with other participants to ensure that they all 'crave,' which they may do by verbally repeating the formula, or nodding their heads in the cultural signal for 'yes.' The riddler then provides the correct answer. The reaction of respondents after the answer is given may be simply, 'Oh, I see,' or 'Yes,' indicating that they understand the metaphorical puzzle involved in the riddle, but do not find it earth-shaking. Alternatively, however, they may reveal deeper admiration for the artistry involved, by means of prolonged laughter (compare Potter 1950: 938 on this point), or statements like, 'Aah! Well, tha's allright. Tha's allright!' There are thus two points at which evaluation of a riddle's excellence are made: right after the riddle is presented, where respondents indicate whether it is easy or hard for them to answer, and right after the correct answer is provided by the riddler if no one has been able to guess it. However, perhaps because riddles themselves are so short and so numerous in any one session, a performer needs to produce several good riddles in a session to earn credit as a good riddler. From this perspective, it is clear that Mr. Jones is a more successful riddler in this session than Mr. Walters. Even though Mr. Walters produces two unsolved riddles, Mr. Jones outmatches him, producing seven in all; three of them remain completely unsolved, and two are solved only after extensive cluing on his part.

Riddling and Lying: Participation and Performance

97

Mr. Jones's strategy throughout the riddling phase of the session is to keep the floor and maintain the role of riddler, which he does easily with a ready stock of riddles. When his riddles are solved, he sometimes uses the turn in which he is supposed to acknowledge the correctness of the answer to introduce another riddle, accomplishing two moves in one turn: Mr. Jones:

Yes. Yes m'am. Well, what turn over all day and never stop turn over? Turn over right now. Never stop turn over.

By using this strategy, he dominates the riddling section of the session, producing the most riddles, occupying the floor as he offers participants clue after clue, and even, at one point, demanding the positive evaluation he deserves for providing the correct answer to an unsolved riddle: Mr. Jones: A net. Mrs. Walters: Oooh! (laughter) Mr. Walters: What you say 'bout a net ? Mr. Jones: A net. Cast-net. JRR: Yeah - ten thousand window and one door. (Pause of 3 sees. Mr. Walters makes no further comment.) Mr. Jones: (To Mr. Walters and Mrs. Walters) What y'all think bout that? Tha's allright? Mrs. Walters: Yeah,that's ALLright! Mr. Walters: Yeah, we'll hold that until we learn better. In this example, we can see an important part of Mr. Walters's strategy. Although it was he who introduced the first riddle, Mr. Jones soon demonstrated that HE was the master-riddler, and it was only near the end of the riddling section that Mr. Walters could remember and introduce another riddle. Unable to match Mr. Jones'as a successful riddler, Mr. Walters equalizes the situation by playing the role of respondent with less enthusiasm than he might. When he knows the answer to a riddle Mr. Jones has posed, he gives it right away, but when he does not know it, he doesn't keep guessing, as a good respondent should, but remains silent, as if trying to remember a riddle of his own. When pushed to respond, he says simply, 'I crave,' and when the correct answer is provided, his positive evaluation is minimal. The principal means which he uses to redress the situation, however, is to change the genre from riddling to another genre in which his competence is stronger. He does this in a natural way, by asking a

98

John R. Riclrford

question which has the superficial form of a riddle, but is not: 'What's the largest watermelon you ever seen?' There is no neat 'riddle-answer' to this question, since it is really the set-up for a lie. But in acknowledging the incorrectness of the answers he gets, the way is cleared for Mr. Walters to launch into the lie itself. The strategy is a valid one, smoothly executed on the second try (the first try was inappropriately placed in the middle of Mr. Jones' unsolved riddle, as noted above). Once Mr. Walters performs this first lie, Mr. Jones starts lying too, and we hear no more of riddles.

Lying Lies, also known as tall tales, little jokes, or lying stories here and elsewhere (Thomas 1977, Tanna 1984:57), are like folktales in length. Like folktales, they also involve a detailed recounting of experience, but unlike them, the central character is not some conventionalized other (Preacher, Brer Rabbit, and so on), but the narrator himself or herself. The focus of the lie is an experience which the narrator has had with a person, animal, plant or thing whose attributes are larger than those normally encountered in real life; so much so that we know at once that the experience is not real, but fictitious. Lies usually begin with a sentence indicating that they are about something that happened to the narrator at some time in the past. They may also specify, at this point, where the experience took place, but this detail may be saved for the ending or the post-lie discussion. Although the body of the lie contains episodes or complicating actions similar to those of folktales and other narratives (Greene and Sharpe 1984), the action is peripheral rather than central, and description is lavished on what we may refer to as 'exaggeration features' - each of which amplifies the larger than life quality of the animal or object which the lie is about. In addition, lies should contain enough detail and vividness to make them seem plausible as real-life experiences, even though everyone knows that they are not. The most successful lies appear to be those with the richest exaggeration features. The last line is usually a punch-line with a final, climactic exaggeration feature. In order to clarify these remarks and pave the way for further discussion, the four lies which were performed during this session are presented below (titles were added by me):9 (A) 'The largest watermelon I ever seen' (by Mr. Walters) I see a man raise a watermelon once. This watermelon was on

Riddling and Lying: Participation and Performance

99

a hill, was on the side of a hill, you know, and they had - just like you gone up on the mountain and had them road, road on the side of the mountain - it was on the hill, like that. And that watermelon grow so large unti they - he roll it up on the side there in a low place, and been using it for a BRIDGE. (Mr. Jones laughs; Mrs. Walters says, 'Abridge! Craz(y)!') People used to come by from everywhere to look at that watermelon. It was growin so large, until it grow so high until cars couldn't even much go OVER the bridge! (Mr. Jones: 'Goodness!') And they had to - every two or three days they had to put another wedge on the side of um to keep um from rolling down, you know. (Mr. Jones: 'Umm-hmm'!) And they been about five thousand people been there been looking at that melon that Friday afternoon. And one of them wedge slip out, and that melon start down the hill. And when it strike itself in front of them rock, it break in half, and the water come out of um drowned over five hundred head of people! (B) 'Mosquitoes on the railroad track' (by Mr. Walters) You know, I was, I was in, ah, in Florida. And ah, I'll tell you the place directly - in Augess, Florida. (Mrs. Walters, in tone of incredulity: 'What!') Augess, a little place I been in they call Augess, Florida. A little old country town. And, ah, we was laying, working on the railroad laying steel. Another time that day, railroad been going on just right by this big oak tree. And just like, we didn't going no place off - we just get our lunch, and, nothing to sit on, but some people been sitting up on the border of the rail, some been sitting on the cross-guards. That was, one of them cross-guards was MY seat. And was so much mosquito up there until you had to mind when you open your mouth to put your food in it, that the mosquitoes don't go in along with the food! (Mr. Jones: 'Mm hm.') And then, some of them mosquitoes was so large up there, would just come up and hit you in your back, would knock you over on you face. (Mr. Jones: 'Mm hm'!) And I was sitting down, right 'gainst the border of the railroad line. And, ah, one of them mosquitoes 'light on the cross-guard in the back there, and crawl and come up to the steel and sting me in my back! (Mr. Jones: 'In my back'!) And when I jump up and look round, that scoundrel bill-where, where, where that mosquito bill went through that steel and sting me, he been there pulling back, trying to get his bill out. And that hammer I had there for driving

100

John R. Riclrford

them, ah, spike down with, I took my spike, my hammer, and I brad that son-of-a-bitch on the other side, on the side I was sitting on. And he couldn't get out. Couldn't get it loose. And now, I was the last man on the end. I didn't tell nobody nutten 'bout it, you know. And later on that day, when the freight car come, freight train come, when the train get right there, with the load of steel beeen going on further up the road for the stop for we to unload it, it wreck that whole flat-car load of steel. Nobody didn't know what was the trouble. They been there THREE WEEKS trying to clear up that track right there where all that steel had that thing all wind up there where they had that big wreck. And after I see all that been going on, I took my cap and I leave, before they find out that I was the man who brad that mosquito-bill on that railroad. Cause, see, they would put me to trouble for that, you know. And I gone, and never did come back. But I know, the last report I hear they was STILL working there for three weeks, been trying to get that old mosquito frame out the way, where they could lay that railroad track in back through there again. (Mrs. Walters: 'Oh-oh'! Mr. Jones: 'Ye-es, sir! Yes, sir!') And that was a SMALL mosquito! (C): 'The Big [Alli]gator I killed' (by Mr. Jones) Well, I been in the woods one time (Mr. Walters starts laughing in anticipation), and I kill - a gator. I kill a big old gator, seven foot and a half. And that gator was so big, after I kill the gator, I couldn't DRAG it. I couldn't drag it. That gator was so large, I went back home, and I get my cow, and my piece chain, fix it to the gator, pull it. And after I - ah - pull that gator home, that gator was, I'd say about, I don't know, am, about ten or twelve feet. That gator was 'bout TWELVE FOOT long. (Mrs. Walters: 'Oh?!') And might be longer than that! (JRR: 'Really'?!). That was the biggest gator I ever did see in my life. I couldn't get it to hold. I just gone home, went home, and get my cow and cut it out. When I got home I cut it. When I got home, I cut it. Cow took it home, and skin it. But I cut it, and I ain't get but fifteen cents for it! (D):'The time I went fishing' (by Mr. Jones) I been in the woods one time, and cap, I caught some FISH! I caught some lil ones and some Big ones, in the net. (JRR: 'Mm hm.') And I cast, cast, cast, cast, c a s t . . . The first one got out. Ah, and I cast again - caught the other one. Allright, Tchoops (sucking teeth

Riddling and Lying: Participation and Performance

101

to indicate disgust, annoyance). The biggest fish get out and left all the little one in. (Mr. Walters laughs softly). And I say 'Don't know what I gon do with this small one.' I throw um 'way. And then, with my last cast, I come and bring um home and cook it for the dog. In the transcripts of these lies, we have included comments made by the audience during the telling of the lie. We shall take up shortly the nature and significance of comments made after the lie is completed. First we need to discuss briefly the role of the performer, and that of the audience, in lying. Apart from the basic responsibility of the performer to provide a good lie (see below), he or she is required to do two things in the telling of it: maintain fluency (don't stumble, stutter, or be inconsistent about details), and keep a straight face (don't laugh!). Both are important for credibility. As in riddling, the audience shows its evaluation of the lie at two (different) points: during the telling of the lie itself, and after it is finished. The evaluation may take the form of laughter, or overt complimentary statement ('That's allright!') or perpetuation of a mock seriousness which extends, and supports the reality of the performer's lie. Comments like, 'My goodness!' or 'Good lord!' - uttered in feigned shock or amazement - are in this last category. Also in this category are requests for further details about the experience, which give the performer a chance to expand his lie. For instance, after the initial loud laughter following (A) had died down, Mr. Jones asked Mr. Walters if this had happened on a particular island, and Mr. Walters responded: 'No, ah, that, that been a place I was they call Junior City!'. The audience responds to this with fresh waves of laughter, and Mr. Walters adds an even finer detail: "That's, that's seventy-five miles on the other side of Spider City.' The audience breaks into laughter again at these specific names for fictitious places. The extent and quality of audience reaction at these two points indicate their evaluation of the lie and its performance. On the basis of audience reaction, it is clear that Mr. Walters is a better liar than Mr. Jones, just as Mr. Jones had been the better riddler. The audience reaction during and after Mr. Walters's lies is characterized by long, hearty laughter, and the ready participation of the audience in the mock seriousness referred to above. Mr. Walters' second lie (B) is marked by loud hooting laughter from Mrs. Walters, and repeated comments of 'Oh boy! Oh boy!' from Mr. Jones. Mr. Jones's first story (C) does not provoke much audience reaction throughout, but afterwards merits

102

John R.Rickford

mock amazement from Mrs. Walters: 'Fifteen cents?!' and further elaboration from Mr. Walters: "That wasn't enough to pay the postage of it!.' Mr. Jones builds on these comments, adding, with a sad shake of the head: 'No sir. Not at all. All that - work too hard, and only fifteen cents!', and the audience breaks into peals of laughter. However, Mr. Jones's second story (D) does not get any audience reaction at all when it is finished. There is a nine second pause after the last line, indicating that the audience did not even realize that the lie had been finished. After this pause, Mr. Jones recycles the last part: 'Last cast, carry um home and cook it for the dog,' and breaks into a giggle. Everyone realizes now that the lie is over, and there is brief laughter (more 'polite' than hearty). Mrs. Walters says 'Aw, Lord,' but there is no further comment. Mr. Walters mutters softly to himself for a while, and then changes the genre to 'smutty stories' (his term), satisfied that he has won this round. Judged by audience reaction, he has scored two big successes, and Mr. Jones only one. Mr. Walters' lies are superior to Mr. Jones' in the close detail which they are spelled out, and in the richness of the 'exaggeration features' which they contain. (B) for instance, contains at least eight: 1. Mosquitoes were so plentiful that you had to mind that when you opened your mouth for food, they didn't go in 2. the mosquitoes were so big that if they hit you in your back, thay would knock you over on your face. 3. This particular mosquito was so strong that it was able to ram its bill though the steel rail and sting the narrator. 4. It took a hammer and spike to 'brad' it to the track. 5. The mosquito frame was so bulky that it caused a wreck of the approaching freight-train. 6. The wreck was so bad it took three weeks to clear. 7. The situation was so serious that the narrator had to quit his job to protect himself. 8. With all this, the mosquito in question was only a SMALL one. Mr. Jones's second story, however (D), contains only two 'exaggeration features': 1. He caught a BIG one, but it got out. 2. He caught some little ones, so small that he took them home and cooked for his dog. Even these points are not skillfully deployed. The performer simply asserts the BIGNESS of the fish that got away: he does not dramatize it by comparison to other recognizably big things, or describe some fantastic consequences of its huge size. Even the action of taking the small fish home to cook them for his dog is not that unusual in this community. The audience fails to respond positively at the end of Mr. Jones's lie because it remains too close to real life, too close to the ordinary chit-chat that they exchange about the events of the week. They may chit-chat more often than they engage in lying or the bigger speech-

Riddling and Lying: Participation and Performance

103

event of which it is a part, but they have not forgotten the rules and standards which characterize this genre. Even in an engineered session like this one, they adhere to traditional conventions.

Conclusion We have considered the nature of participation and performance in riddling and lying from the perspective of an ethnography of speaking. Although the data for this study were drawn from a Sea Island community off the coast of South Carolina, readers familiar with similar genres in other communities will undoubtedly recognize some elements. Some may wonder why linguists should not leave the analysis of genres and verbal activities like these to folklorists and anthropologists. The answer is that material of this type can lead us to fruitful revisions of our concepts of competence and performance (see Hymes 1971), can challenge and enrich our attempts to describe the structure of units beyond the sentence, can illuminate our interest in language use and the relation between language and social life (see Roberts and Forman 1972, Ferguson 1977), and can lead us to the answers to questions we have barely begun to ask, such as, 'What is language good for, anyway?' and 'Why does anyone say anything?' (see Hymes, ibid., Labov 1972: 48). Space does not permit us to develop these larger implications, but it is these that should entice us to attend to riddling and lying in addition to phonology and morphology, as Ferguson first did four decades ago. We have as much to bring to and gain from the study of these verbal activities as the folklorists and anthropologists and literary scholars. If we each make our contributions, the harvest should be bountiful for all.

Footnotes This is a revised version of a paper originally written for a class taught by Dell Hymes at the University of Pennsylvania several years ago, and it has benefitted from his comments. I am also grateful to Charles A. Ferguson for copies of relevant papers by himself and others (although he did not know what I was writing this paper for) and to Roger Abrahams and Richard Bauman for feedback and references. I also wish to thank my wife Angela for helpful comments and encouragement.

104

John R. Rickford

1 The nasal in 'telling' is usually dental or alveolar rather than velar, but in order to avoid phonetic notation and maintain readibility, we will not generally attempt to indicate phonological features of Sea Island speech in this paper. 2 For example, from Mr. Jones: 'Little Jack Horner/Sitting in the corner/Eating he Christmas pie/He juk in he thumb/And he pull out a plumb/Oh, look what a good boy we am !/Haw haw haw, haw haw haw !/Oh, look what a good boy we am!/. 3 The popularity of Savannah as a meeting-place for 'story-tellers' in the old days is documented in Parsons 1923: xv. 4 These include: the pollution of oyster-beds whose farming had employed hundreds, emigration to urban centers in search of employment, the effects of World War II, and the displacement of locals by hoteliers and real-estate dealers seeking to turn the sea-islands into the tourist paradise of the South. 5 I should note that in other fieldwork within this community and elsewhere, I have found that questions involving danger, death, superstitions, ghosts and the like often do get the interviewee interested and speaking excitedly along the lines predicted by Labov (1972). For my own views on the issues raised by Wolfson (1976) about the use of strategies of this type in fieldwork, see Rickford (to appear). 6 Mr. Walters and Mr. Jones were both deacons in the local Baptist church, with well-deserved reputations as pray-ers and singers; but even in these sacred speech events, the performance element was still present. For instance, after being commended for a particularly moving prayer which he delivered at a prayermeeting, Mr. Walters said, beaming: "I can kick up a little bit of dust sometimes - if I got somebody to start the fire for me, and kind of warm me up. It'll make you rub your eye, anyhow." 7 According to Tanna (1984: 48), one element in the appeal of riddles in Jamaica, which similarly occur in story-telling sessions, is the fact that 'the role of questioner changes rapidly and that everyone can participate in answering.' Note, incidentally, that the Jamaican riddles have a formulaic opening, 'Riddle me this, riddle me that, guess me this riddle and perhaps not' (ibid.) which is not attested in our data, but may once have been current in this area and may still be on some islands. 8 Compared with 'I give (up)' - the equivalent formula in parts of the USA and Caribbean, this is a more powerful admission of defeat for respondents and success for the performer. 9 Richard Bauman has drawn my attention to the fact that the watermelon story and the mosquito story occur on p. 540 and p. 509 respectively of Baughman 1966.

References Abrahams, Roger D. 1971. Personal power and social restraint in the definition of folklore. Journal of American Folklore 84:18-30. Abrahams, Roger D. 1983. "A riddling on St. Vincent." Western Folklore XLII : 272-95.

Riddling and Lying: Participation and Performance

105

Bauman, Richard. 1975. Verbal art as performance. American Anthropologist 77:290-311. Baughman, Ernest W. 1966. Type and motif-index of the folktales of England and North America. The Hague: Mouton. Bauman, Richard. 1977. Linguistics, anthropology and verbal art: toward a unified perspective, with a special discussion of children's folklore. Linguistics and Anthropology ( = Georgetown Univ. Round Table 1977), ed. Muriel Saville-Troike, 13-36. Georgetown: Georgetown Univ. Press. Bauman, Richard. In press. "Narrative variation and the shifting contexts of traditional storytelling." Story, performance and event. Cambridge: Cambridge Univ. Press. Dil, Anwar S. 1971. Bibliography of Charles A. Ferguson's works. Language structure and language use: essays by Charles A. Ferguson, selected and introduced by Anwar S. Dil, 313-27. Stanford: Stanford Univ. Press. Evans, David. 1976. Riddling and the structure of context. Journal of American Folklore 89:166-88. Ferguson, Charles A. 1954. Review of Modern Lebanese proverbs, collected at Ras al-Matn, Lebanon, by A.Frayha. Journal of American Folklore 67: 223-6. Ferguson, Charles A. 1976. The collect as a form of discourse. Language in religious practice, ed. W.J.Samarin. Rowley, Ma: Newbury House. Ferguson, Charles A. 1977. Linguistics as anthropology. Linguistics and Anthropology ( = Georgetown Univ. Round Table 1977), ed. Muriel SavilleTroike, 1-12. Georgetown: Georgetown Univ. Press. Ferguson, Charles A. 1983. Sports announcer talk: syntactic aspects of register variation. Language in Society 12 (2): 153-72. Ferguson, Charles A., and J. M. Echols. Critical bibliography of spoken Arabic proverb literature. Journal of American Folklore 65:67-84. Ferguson, Charles A., and W.D.Preston. 1946.107 Bengali proverbs. Journal of American Folklore 59:365-86. Ferguson, Charles A., and W. D. Preston. 1947. Seven Bengali riddles. Journal of African and Oriental Studies 66:299-303. Georges, Robert A., and Alan Dundes. 1963. Toward a structural definition of the riddle. Journal of American Folklorel6\ 111-8. Greene, Ervin L., and Priscilla Baptista Sharpe. 1984. Structure of the Sea Island folktale. Paper presented at the Fifth Biennial Meeting of the Society for Caribbean Linguistics, Univ. of the West Indies, Mona, Jamaica. Heath, Shirley Brice. 1983. Ways with words: language, life, and work in communities and classrooms. Cambridge: Cambridge Univ. Press. Hymes, Dell. 1964. The ethnography of speaking. Readings in the sociology of language, ed. Joshua A. Fishman, 99-138. The Hague: Mouton. Hymes, Dell. 1971. The contribution of folklore to sociolinguistic analysis. Journal of American Folklore 84:42-50. Jackson, Juanita, Sabra Slaughter, and J. Herman Blake. 1974. The Sea Islands as a cultural resource. The Black Scholar 5:32-9. Jones-Jackson, Patricia. 1983. Alive: African traditions on the Sea Islands. Negro History Bulletin 46 (3): 95-6,106. Labov, William. 1972. The design of a sociolinguistic research project. Paper presented at the Sociolinguistics Workshop held in Mysore, India, May-June 1972.

106

John R. Rickford

Maranda, Elli Kongas. 1971. The logic of riddles. Structural analysis of oral tradition, ed. Pierre Maranda and Elli Kdngas Maranda, 189-232. Philadelphia: Univ. of Pennsylvania Press. Parsons, Elsie Clews. 1923. Folk-Lore of the Sea-Islands, South Carolina, Memoirs of the American Folk-lore Society, No. 16. Cambridge, Mass.: The Cosmos Press. Pepicello, W. J., and Thomas A. Green. 1984. The language of riddles: new perspectives. Columbus: Ohio State University Press. Potter, Charles Francis. 1950. Riddles. The standard dictionary of folklore, mythology and legend, ed. Maria Leach. New York: Funk and Wagnall. Rickford, John R. 1974. The insights of the mesolect. Pidgins and Creoles: current trends and prospects, ed. David De Camp and Ian F. Hancock, 92-117. Georgetown: Georgetown Univ. Press. Rickford, John R. 1985. Ethnicity as a sociolinguistic boundary. American Speech 60 (2): 99-125. Rickford, John R. To appear. Linguistic variation and the social order. New York: Academic Press. Roberts, John M., and Michael L.Forman. 1972. Riddles: expressive models of interrogation. Directions in sociolinguistics, ed. Dell Hymes and John Gumpeiz, 180-209. New York: Holt, Rinehart and Winston. Scott, Charles T. 1969. On defining the riddle: the problem of a structural unit. Genre 2:129-42. Tanna, Laura. 1984. Jamaican folk tales and oral histories. Kinston, Ja.: Institute of Jamaica Publications, Ltd. Taylor, Archer. 1951. English Riddles from Oral Tradition. Berkeley and Los Angeles: Univ. of California Press. Thomas, Gerald. 1977. The tall tale and Philippe d/Alcripe: an analysis of the tall tale genre with particular reference to Philippe d/Alcripe's "La Nouvelle Fabrique des Excellents Traits de Verite" together with an annotated bibliography of the work. St. John's, Newfoundland: Department of Folklore, Memorial University. Wolfson, Nessa. 1976. Speech events and natural speech: some implications for sociolinguistic methodology. Language in Society 5:189-209.

Stereotype Foreigner Talk in German Literature Peter Miihlhàusler Linacre College University of Oxford Oxford

1. Introduction By coining the term 'foreigner talk' in 1968, (published 1971), Charles A. Ferguson made linguists, and in particular pidginists and creolists, focus on a phenomenon which they had previously only dealt with in an unsystematic fashion, if at all. Foreigner talk, (henceforth FT), it appears, forms an important part of the contribution of speakers of the superordinate culture to the development of pidgins, particularly in their early phases. Whilst FT explanations of pidginization are valuable in understanding the universal and language-specific forces in linguistic development, the importance of the concept is much wider. As has been recently pointed out by Duranti (1981) for Samoan, most grammatical descriptions of this language reflect a FT register (p 165): "the variety elicited through interviews does not closely correspond to either informal or formal varieties of most spontaneous interaction in traditional settings". Very much the same state of affairs may be found in the descriptions of numerous other 'exotic' languages, making them a much less reliable basis for typological and comparative work than is widely believed. The aim of this paper is not to discuss such methodological issues, however, but to present a body of data of German FT and brief comments on its theoretical implications.

2. Methods Ferguson (1981:12-3) discusses the various research methods used in the study of FT, emphasizing the importance of direct observation and experimental methods. He adds (p 13) that "it is also possible to inves-

108

Peter Miìhlhàusler

tigate FT by the analysis of written material, especially literary use of FT to represent the speech of a foreigner". In the case of historical research this is the only method available. The data discussed in this paper illustrate its use and its limitations. In the case of FT German, as with FT Dutch, it appears that the time depth one can reconstruct is not terribly great, though the fact that no literary examples have been found, from before 1800 does not mean that FT was not used at this time. Three types of literary FT German can be distinguished: i) attempts to record the speech of foreigners ii) literary constructions iii) translations The latter category is particularly important in the case of German. I will not consider it here as I have not completed my data collection in this area. As regards the other two categories it is frequently difficult to keep them apart and multiple origins will have to be assumed to underlie many text samples. Any study of early FT will also be concerned with the very mundane problem of obtaining the data. The FT register is much more common in low than in high literature and, even when it is used in high literature (as in literary portraits of gypsies, foreigners and inhabitants of various Utopias and dystopias), it rarely attracts the attention of literary critics. The data considered here are part of my very large collection of materials. This collection, at present, is far from complete or even representative as it fails to cover a number of areas and periods. It includes extracts from satirical magazines, folk humour, exotic novels and plays. Occasionally, a particular author or journal would seem to justify a separate study.1 Thus, a closer investigation of the language and metalinguistic views found in Karl May's work seems highly desirable.2 My present study is an example of such a 'local' investigation, i. e. it is concerned with FT as found in a single journal, Die Jugend. An important consideration in FT studies is that the data should be relevant to current issues in linguistic theory. Thus, it would seem desirable to promote a comparison between FT, interlanguage grammars and bioprogram hypotheses. In my own research I have tried to distinguish between aspects of FT which reflect: i) postulated bioprogram features ii) language-derived simplification processes iii) cultural stereotypes These categories could be extended by adding Ferguson's (1977:29) iv) clarification processes v) expressive and identity processes

Stereotype Foreigner Talk in German Literature

109

thus incorporating the important, and possibly constitutive, role of pragmatic and sociolinguistic factors in FT production. Such a classification, whilst still a strictly pretheoretical one, could be a significant step in distinguishing between the natural and cultural bases of FT. Moreover, it would provide us with more general measures for the reliability of written sources and for the reconstruction of early pidgins and other simplified registers. A final note on literary FT is that it usually portrays a variety spoken by foreigners whilst the term 'foreigner talk' in its original sense refers to the kind of language used to address foreigners. However, it is widely held that the literary speech of foreigners is sufficiently close to the spoken speech used towards foreigners to ignore the differences. For the purpose of this paper, I will also have to ignore the fact that the choice of literary medium can introduce new material not encountered in spoken FT,3 e. g. eye-dialect, an example from my data being the use of the spelling ferertes instead of German verehrtes 'honoured' to portray Hungarian speakers.

3. T h e d a t a Having discussed the longitudinal aspects of the development of a literary German FT elsewhere (Miihlhausler 1984), I shall concentrate here on the FT found in a single satirical magazine Die Jugend (Munich), particularly texts which appeared between 1900 and 1914. My choice of data is motivated by the desire to exclude as many variables as possible and to concentrate on the reasons for variation within a relatively homogeneous body of data. Thus, i) in all instances are we dealing with original literary fabrications rather than recoded FT or translations ii) all examples were produced within a brief span of time. Very similar data were found for the preceding and following 5 years, i. e. they reflect a relatively stable tradition iii) the same editorial policies appear to have been applied throughout the period chosen. The FT data under consideration fell into a number of different literary genres, including jokes, cartoon captions, satirical poetry, fabricated letters, fabricated interviews, among others. Of these, fabricated letters by fictitious foreign writers are the most numerous and will therefore be given prime attention. The texts are represented as having been produced by speakers of a number of languages, notably Italian, Slavic

110

Peter Mühlhäusler

Languages, Hungarian, English and French. Very few examples were found of speakers of the non-European races. I shall present examples for each of these groups. Brief remarks on their salient linguistic properties will be given at the end of each example. The translations are my own. Example 1. Foreigner German by Italian (Die Jugend, Vol 43, p 868,1904): Empfang bei Signore Domenico Katzeimacher Signora Katzeimacher: Du, Domenico, sein sie draußen der Eccellenza, die ministro von das Hunterickt! Wollen sie bissei sprechen mit Dir! Domenico: Sollen sie nur spettare un poco, warten ein wenig! Aben ick jeß niente tempo, keiner Seit! Signora Katzeimacher (kommt nach einer geraumen Weile wieder): Lassen sie son serr bitten der Eccellenza, daß sie werden emfang! Domenico: Per me! Meinetalb! (Der österreichische Unterrichtsminister Ritter v. Harteil tritt mit tiefen Bücklingen ein). Härtel: Buon giorno! Mille complimenti! Riverisco! Umillissimo servo! Unterhänigster Diener, verehrungswürdiger Signore! Nun werde ich mir mit der Errichtung der italienischen Rechtsfakultät in Innsbruck wohl endlich Ihre hochgeschätzte Zufriedenheit und Wohlgeneigtheit errungen haben? Domenico: Sein sie ick gar nix contento, gar nix sufried! Per bacco! Aben sie dock versprocken der governo, das Regierung, su verlegen die corso juridico von das maledetto Hinnsbruck! Reception by Signore Domenico Wop Signorina Wop: Hey, Domenico, be she outside the Excellency, the ministro of heducation! Want speak a bit with you! Domenico: Must she just spettare un poco, wait a little bit. Have I now niente tempo, no time. Signorina Wop: (returns after a good while): Lets you already a lot ask the excellency, that she be receive. (The Austrian Minister of Education, von Härtel, enters bowing and scraping). Hattel: Bon giorno. Mille complimenti. Riverisco, Umilissimo servo, humble servant, honourable signor. I hope to have finally, by setting up the Italian Faculty of Law at Innsbruck, deserved your esteemed satisfaction and favourable inclination. Domenico: Be she I not at all contento, not at all satisfied. Per bacco! Have you however promised to the governo, the government to shift the corso juridico from the maledetto Hinnsbruck. Apart from a number of features meant to illustrate interference from Italian, a number of FT constructions are also present, including: (a) use of infinitive instead of inflected verb forms, (b) phonological simplification (e. g. shibbilants), (c) the use of the negator nix

Stereotype Foreigner Talk in German Literature

111

Example 2. A fictitious speech by the speaker of the Italian House of Representatives (Die Jugend Vol 54, p 996,1915) Libbi Gazzelmitmakkeri! Ewiva Italia! Uns gehti slekt! Mi sammi fridliskste nazione vun di tutto mondo - ewiva Italia! Aber, was kenna mi mache, wann eigen Vabindete, fratre austriaco - maladetto - hamtikisch uns anpakke? He? Dreissik Jahn lange uns imme hamtikisch anpakke durk angeblike Bindnis!? Dreissik Jahn lang uns imme die Haend halte, damit mi ihm kenna nit Gorgola zudrukki!? Ewiva Italia!... Dear fellow wops! Ewiva Italia! We are suffering! We are the most peaceful nazione of di tutto mondo - ewiva Italia! But, what can do if own ally, fratre austriaco-maladetto - treacherously holds us, yes? Thirty years has treacherously constrained us through socalled treaty!? Thirty years held our hands, so that we can not throttle his throat! Ewiva Italia!... Example 3. Foreigner German by a Czech writing to a Pole on the occasion of an incident in the Austrian Parliament (Die Jugend vol 43, p318, 1904): Hab ich mich ghert, liwe Brude meinige bolnische, das hastu gmaust in Parlament oestreichische Masse Briefboegerl, Bleischtifzkrl und Schreifzeigerl, was hade vafluchte Schwein deitschi aufgelgtowäd fir Verbreiterung van Kultur unte Abgeordnete. Ale! Frei ich mich ganse herzinnige, wann siechte bolnische Nazion as amol ein, dass ise ganse Kultur deutschi nix wert als dasse uns kunnte gschtuhlen wem! Awe hab ich mich blus vawundrowäd, dass Brude bolnische Dich haste lassn verwischn! Mi werine ihnen stehlns nich blus Briefboegerl und Blaistiftzkrl sundern ganse Barlamentzki und boehmische Landle unte Sitzbanerl weg, ohne dass merkens, womit begrisse ich Dich heite, libe Brude Dein Navratil Pschsbzkbonesszl Schneidemasta, Praha Have I heard myself, dear brother mine Polish, that you have stolen in Parliament Austrian lot paper, pens and writing things what had damned pig German put there for spread of culture among representatives. Ale! Delight I myself very much, when realized Polish nation that entire culture German is not worth stealing. Have wondered why brother Polish let himself get caught. We shall steal them not only paper and pens but entire Parliament and Bohemian country under their very seats without them noticing, whereby I greet you for today, dear brother, your Navratil Pschsbzkbonesszl tailor, Praha Salient properties of this text include: (a) the use of reflexive verbs where German does not require them (b) adjectival postmodifiers instead of German adj ^Vorder (c) subjectless sentences

112

Peter Mühlhäusler

(d) variable deletion of post-vocalic (r) (e) variable word-order tending towards VSO Example 4. A complaint by a Polish speaker (Die Jugend Vol 49, p94,1910): Wir arrmen Pollen! Habben wirr gestimmt firr Steierrefform, habben wirr gezzogen Karre aus Dreck und was ist jetzt? Ha, Dank vom Hause Preissen! Werr uns hat gegebben Stimme, wird gemisshandelt in Kattowitz und verrsetzt! Und dabbei habben wirr uns immerr gedduckt und warren immerr bescheiden. Habben wirr jemmals verrlangt, das Deitschland soll Nammen andern in Pollenland? Habben wirr jemmals verrlangt, dass Thronredde soll verrlesen werrden in polnischer Sprache? Habben wirr jemmals verlangt, dass militaerische Kommandos in der Armee sollen gegebben werrden in polnischer Sprache? We poor Poles. Have we voted for tax reform, have we pulled cart out of dirt and what now? Ha, thankyou from Prussian Monarchy. Who has given us vote is maltreated in Kattowitz and dropped? And really have we always been obsequious and were always modest. Have we ever demanded that Germany should change name into Poland? Have we ever demanded that royal speech should be read in Polish? Have we ever demanded that military order in the Army were to be given in Polish language? Apart from the many spelling devices underlining the foreigness of this text, it does not deviate greatly from standard German. Deviations are restricted to (a) some changes in word order, particularly the position of auxiliaries (b) variable absence of articles and other determiners. Example 5. Hungarian-German FT (Die Jugend, Vol 28, p 574,1904) The insistance of the authorities of Kolozsvär university that letters should not be addressed to Klausenburg Universitaet evokes the following satirical letter: Nix daitsch! Wort' nur, Schwöb verdommter, wonn Du Dir nächtesmoi nit konnst Nomen merken von ungarischer Stodt - hat, so schick' ich Haiduk! Der soll donn mit papriziertem Stecken Nomen Dir auf Buckel schraiben! Teremtete! Wirst Du schon lernen, ob es gibt ein Klausenburg im gonzen Ungarn, wonn Du dickes Pfefferrehrl kriegst zu schmecken! Hot Schwöb Frechheit gonz gemaine, zu behaupten, daß er Kolozsvar gegründet! Baratome, gestotten mir spekulative Frage: Wonn auch Schwöb gegründet Stadt, hot er darum ain Recht, der Stodt zu geben daitschen Nomen? Schwöb hot überhaupt kain Recht! Schworzer daitscher Hund soll froh und donkbor sain, doß ihm hot Magyar erlaubt, Stodt zu gründen! Mit ollerhond Hochochtung Grof Hunyady Janos No German Just wait, Swabian damned, if you cannot remember, the name of Hungarian town next time, hat, will send I Haiduk! He will write the name on your back

Stereotype Foreigner Talk in German Literature

113

in red pepper! Teremtete! Will you already learn whether there is Klausenburg in whole of Hungary when you taste red pepper stick! Has Swabian insolence most vicious to maintain, that he founded Kolozsvar! Baratom, excuse my asking speculative question: Although Swabian founded town, has he for that reason right the town to give German name? Swabian has no right at all! Black German dog must be happy and grateful that Magyar gave him permission, town to found! With all sorts of respect, Count Hunyady Janos Again, only the pidgin features observed in the previous text are present in this letter, and again there are a number of other changes, such as deviant word order and phonology signalling Hungarian substratum influence. Again, ignoring the numerous instances of eye-dialect, deviations from standard German grammar are not many. Salient features include: (a) deviant word order including Adj ./Vorder (variably) (b) stylistic infelicities Example 6. Monologue by an Englishman who has caused a Russian to trip over his feet (Die Jugend Vol 43:1043,1904): Oho, Mister Iwan, you have mir trampled auf der Bain! Wat will you denn aigenlick! Wat glauben you, dass ick me uerde lassen that gefall? Oh no! Oh no! Ik uerde you zaigen, uer ist master hier of the place. Ick uerde box You in the ground and Bod! Damned biest! Ick verlangen, dass you ganse Stiefel directly mir ausliefern! Ganse Stiefel! Uas hat mir getreten auf the corn! Ferstanding? You wil not? you will me bios geben returning the nails, der Naegel, uas mir have getreten? o yes! Very good, I uerde nehmen der Naegel... You wil me not geben der Naegel? You wil me bloss lassen sehn the place uo sein the nails gewesen, uo mir have trampled auf the corn? Very good! O yes! Oh, wat have you for ain schoenes Stiefel! Oh! You have gekauft der schoenes Stiefel from me? Oh! Mister Iwan! Very fine! All right! You can have noch vieler Stiefel! from me. Goodbye! Gluecklicher Reisen! Oho, Mister Iwan, you have trod on my leg. What do you want? What believe you, that I will tolerate? Oh no! I will you show, who is master here of the place. I will box you in the ground and soil. Dammed Biest! I demand that you entire boot directly me give! Entire boot! What has trod me upon the corn! Understanding? You will not? you will me just give returning the nails, the nail, what me have trod? O yes! Very good, I will take the nails . . . You wil me not give the nails? You wil me just let see the place where have been the nails, where me have trampled upon the corn? Very good! O yes! Oh, what have you for a lovely boot! Oh! You have bought the lovely boot from me! Oh! Mister Iwan! Very fine! All right! You can have still many boots from me. Goodbye. Happy journey!

114

Peter Mühlhäusler

This text differs from the others in a number of ways, in particular in that we have a very considerable amount of code switching and crosslingual identification (syncretisms) made possible by the existence of numerous cognates in English and German. Specific devices noted in this text include: (a) the tendency to pronounce German [v] as [w] (b) use of infinitives instead of verbal inflections (c) highly variable word order, usually different from that which would have been appropriate in German (d) calqued idioms Of particular interest are texts where speakers from a number of different backgrounds all produce versions of German FT. One such example is the following. Example 7. An extract from the speeches held at the International Symposium on Divorce: (Die Jugend Vol 42: 89,1903): Lord Laurence Pips aus London: Oh, gentlemen, maine Haeren, ick sain very sorry, saehr betruebet, Ueil my wife, main Gemahlin haben the brandy, die Snaps maehr als mick geliebet! Grof Kegloedy Istvan aus Budapest: Hat, erlauben, kerem aläsom, hob ich gehairothet Prinzessin; wor serr schoene Fajer! Barätom, muss ich lossen schaiden! Worum? Kommt mir der Soch' viel zu thajer! Don Ramiro y Rosalva de los Castellos aus Lissabon: Ick aben, per los Dios, das schwerste de los tuttos Fallos, Sein mir meines Donna gebrennen durck mit einer Kneckt von dos Stallos! Oh, Gentlemen, my gentlemen, I be very sorry, very sad, because my wife, my spouse, have the brandy, the snaps, more loved than me. Hat, with permission, kerem alasom, have I married princess, was very nice celebration. Baratom, must I let divorce! Why? Is for me matter much too expensive! I have, por los dios, the most unbearable de los tuttos cases, Be my Donna run away with one lad from stables.

Note that in this last example, we have a macaronic mixture of various Romance languages rather than Portuguese. Apart from the more general features such as changes in word order there are a number of areas in which language-related differences can be noted. These include (i) German haben to have is represented as haben in FT used by an English speaker, as hob in that of the Hungarian speaker and aben in that of the Portuguese.

Stereotype Foreigner Talk in German Literature

115

(ii) German ich appears as ick (for English and Portuguese speaker) and as ich for the Hungarian speaker. (iii) English and Portuguese speakers change German [sC] clusters to [sC] A final text illustrating different types of FT is the following: Example 8. A verbal duel between Miss Duncan and Madame Madeleine (Die Jugend Vol 43:219,1904): Miss Duncan: Und ich uissen ueberhaupten nicht, uie Sie koennen sprechen mit mich von dances ! Haben Sie gewest in Griechenland, zu tanzen in the theatre of Dyonisius! Haben Sie gesungt mit little Griechenknaben? Madame Madeleine: Oh, ma chère, dass ick nickt lacken! mon dieu; Ihrer kleines grieckiges Knabs! Wenn sreit meiner Kinder, si schoenere Ton gieben von sick, als Ihrer ungesundes Knabs mit ihre verflixte Quack Quack! Miss Duncan: Uas verstehen Sie of my dear boys mit gockelhelles Stimm? And why, uarumueshalben Sie tanzen nicht mit nacktes Fuss? Mme Madeleine: Pourquoi? Weil ick 'aben das Ansickt, dass Beet'oven nickt 'aben gesreibt seiner Musiken fuer nacktes Fuess! Sondern fuer 'erz in Bauch! Miss Duncan: Ueil Sie versteh nothing at all! Und halten Sie veilleicht so schoener Ansprach an der Leut? Sie müssen extra bringen with you einer Mr. Schrenck-Rotzing. And why? Uarumueshalben? Weil Sie nicht koennen sprechen der Deutsch so perfectionally, wie ich. Oh, Sie haben keiner Ahnung, uie ich es werden bringen so ueit! So ensetzlich ueit, dass kommen werden the people not, um zu seh meine little dances, sondern for zu hoeren meiner speach. Ich nich uerden nur reformiren the dance, ich uerde reformiren the deutsche Sprach too! Madeleine: Fuer so Vorstellung artistique sein zwoelf francs mehr, als assez bien! Ick aber, moi, kann demander à meine Besucher vingt, was sein so viel als swansickü Alors - sie müssen sehen ein in Ihre Kopp - ick sein der groessere Tanzmusikantin! Et - après tout wir kannen fraggen der Publikum um seiner opinion. Miss Duncan: Very well. Uer is dancing mehr gut der Musik, lieber Publikum? Stimme aus dem Publikum: Alle zwo seids guat! Was aus 'm Ausland kimmt, is allweil guat. Do feit si nix! Miss Duncan: And I know not at all, how you can speak with me of dances! Have you been in Greece to dance in the theatre of Dyonisius? Have you singed with little choir boys? Mme Madeleine: Oh, ma chère, that I not laugh. Mon dieu; Your little Greeks boys! When cry my brats, they nicer sound make than your unhealthy boys with their ruddy Quack Quack!

116

Peter Miìhlhàusler

Miss Duncan: What do you understand of my dear boys with crystal clear voice? and why, for what reason you not dance with naked foot? Madeleine: Pourquoi? Because I 'ave the opinion, that Beet'oven not 'ave written his music for naked feet! But for heart in stomach! Miss Duncan : Because you understand nothing at all ! And hold you perhaps nice speech for people? You must especially bring with you a mister Schrenck-Rotzing. And why? For what reason how come? Because you not can speak the German so perfectionally, as I. Oh, you have no notion, how far I will go ! So terribly far, that will come people not to see my little dances but for to hear my speech. I not will reform the dance, I will reform the German language too. Madeleine: For performance artistique be twelve francs more than assez bien! But I, moi, can demander a my visitors vingt, what be as much as twenty. Alors - you must realise in your head - I be the greater dancing musician! Et - après tout we can ask public of his opinion. Miss Duncan : Very well. Who is dancing more good the music, dear public? Voice from the floor: You are both good. Whoever comes from abroad must be good. No worries.

Of the constructions of interest here, some are found with both English and French speakers, e.g. the use of infinitives instead of inflected forms, the use of clarification devices such as repetition, near-synonyms and changes in word order. Language specific changes include the incorporation of French and English lexical material and differences in pronunciation.

4. Analysis I shall now try to organize the FT observed in the above and similar texts in terms of the parameters given in the introduction to this paper. Given its exploratory nature and the absence of representative sampling the results will be very much a tentative first step rather than a definite answer. However, even within these limits, a number of interesting trends can be observed. Before discussing these I would like to give a brief list of the properties identified in FT registers and their status in the texts under investigation: Key: + categorically present - categorically absent 0 variably present

Stereotype Foreigner Talk in German Literature 1. Bioprogram features focus movement use of definite article TMA Relative and Subject Cop. Absence of Cop. Adjectives as verbs question form question words 2. Language-derived simplification processes avoidance of passive invariant word order loss of gender distinction loss of agreement reduction of synonymy avoidance of marked sounds

117 (cf Bickerton 1981) — — — — — 0 — — — — 0 — 0

3. Cultural stereotypes nix 'general negator' other special FT lexicon eye-dialect malapropisms morphological mixing superfluous reflexives

0 0 + + + +

4. Clarification processes repetition paraphrasing translating in text attention-getting devices (tags etc)

+ + + +

5. Identification processes diminutives/hypocoristics status marker pronoun nationality markers

0 0 +

6. Other macaronic mixture confusion of inflections cross-lingual identification of forms

+ + +

Other categories which have not been examined but could be of use in future analysis include the changes in type-token relations between FT German and ordinary German and the role of redundancy at different size levels. The extent to which the various pretheoretical categories used here would be of use in a theoretical account of FT cannot be determined at present. One suspects, however, that several of the categories used here would have to be split and/or refined and that others may turn out to be of limited use.

118

Peter Miihlhàusler

5. Discussion There are a number of interesting trends inherent in the data. First, it appears that literary FT is not influenced greatly by grammatical considerations. Indications are that neither bioprogram grammar nor language-specific grammatical simplifications have played a major role. On the other hand, indexical and pragmatic factors are of great importance and, in several texts, are the only parameters of FT. It appears that the texts rely heavily on the following devices: i) the violation of norms of German to signal that the speaker is an outsider ii) the use of lexical borrowing and mixing to establish the speaker's nationality or group membership iii) the use of non-fluency markers to emphasize the deficiencies in the communicative competence of foreigners. There is a second tendency, also evident in texts other than those examined in the present paper, namely that of using features differentially to signal distance from or solidarity with foreign language speakers. Thus, the more 'barbaric' a foreigner is perceived to be, the more deviations from German are introduced. Moreover, certain constructions such as the loss of inflections or the absence of subject nouns and pronouns, tend to be reserved for less 'civilized' foreigners (e. g. negroes, Eastern Europeans) whereas cross-lingual identification is found mainly in the FT of people German speakers can identify with (e. g. English speakers). These findings suggest that there may be a considerable discrepancy between stereotyped literary versions of FT on the one hand and actually used FT on the other. Pragmatic and lexical intuitions of writers appear to be much more developed than grammatical ones. One is led to conclude that the reliability of literary registers of FT and pidgins may be somewhat less than researchers had hoped for, a conclusion also arrived at by investigators studying the literary FT registers of other languages. They would seem to be best studied as a cultural phenomenon, reflecting attitudes towards outsiders and stereotypes about second-language learning. These stereotypes have no doubt entered into the formation of pidgins and other makeshift systems of cross-cultural communication, but in a much more indirect fashion than is sometimes suggested. The study of stereotyped FT registers remains relatively undeveloped. I hope to have shown that the suggestions made by Ferguson can be used profitably to extend our knowledge of this area of linguistic behaviour.

Stereotype Foreigner Talk in German Literature

119

Footnotes 1 Such studies could also be of a comparative nature. I am thinking of a comparison between early and later editions and translations of Robinson Crusoe. The data provided for the Dutch edition by Werkgoep (1978) would be an excellent basis for such a study. 2 Such a suggestion was first made by Clyne (1975). 3 We are concerned here with the wider principle that a change of medium promotes a change in form and function, i.e. that written language cannot be regarded as derived directly from spoken forms.

6. References Bickerton, Derek. 1981. Roots of Language, Ann Arbor, Karoma. Clyne, Michael G. 1975. 'German and English Working Pidgins', Linguistic Communications 13:1 -20. Duranti, Alessandro. 1981. The Samoan Fono, Canberra, Pacific Linguistics B80.

Ferguson, Charles A. 1971. 'Absence of Copula and the Notion of Simplicity' in Dell Hymes (ed) Pidginization and Creolization of Languages, London: Cambridge U.P.: 141-150 Ferguson, Charles A. 1977. 'Simplified Registers, Broken Language and Gastarbeiterdeutsch' in C.Molongy, H.Zobl & W.Stoelting (eds) German in Contact with other languages, Kronberg, Scriptor, 25-39. Ferguson, Charles A. 1981. 'Foreigner Talk as the Name of a Simplified Register' in International Journal of the Sociology of Language 28:9-19. Miihlhâusler, Peter. 1984. 'Tracing the Roots of Pidgin German', Language and Communication 4,1: 27-57. Werkgroep. Werkgroep Taal Buitenlandse Werknemers, Nederlands Tegen Buitenlanders, Amsterdam, Publications of the Institute for General Linguistics 18

Role of Hypercorrection in French Linguistic Change Nicole Gueunier Université François Rabelais Tours

My best thanks to Melissa Metcalf, Elisabeth Savaton and Michèle Wahl for having helped me with the translation. In the present study on linguistic change, I shall follow - while trying to eschew an overmechanistic sociology of language - the intellectual tradition of linguists whom Labov (1972: 274) calls: "the 'social'group [who] would pay close attention to socialfactors in explaining change" and for whom "data from the speech community will continue to form an essential part in the analysis of linguistic change". (id.: 182) One can roughly distinguish in that trend of thought widely illustrated in France by the works of A. Meillet, F. Brunot, M. Cohen, a few major explanatory principles of linguistic change, linked to the social position of the subgroup in which it originates. If change is imputed - rightly or wrongly - to a subgroup at the bottom of the social hierarchy, it will be interpreted in terms of'vulgarism'; if it comes from a prestige subgroup, then it will be called 'innovation'; lastly, in the case, studied by Labov (1972), when it takes form in a subgroup moving upwards and hence in a position of linguistic insecurity, such as the lower middle class in New-York, it will be termed 'hypercorrection'. In agreement with Labov, I give this term not only its original definition:

122

Nicole Gueunier

"irregular misapplication of an imperfectly learned rule, as in the hypercorrect case marking o/whom did you say is calling?"1 (id.: 126) but all the forms of the tendency to overuse the forms known as correct2. In order to clarify the statement, here are a few points of French pronunciation, which respectively exemplify these tendencies. For each of them will be given an example of successful change - e.g. which actually gave birth to a new form, taking the place of the former one, and an example of unsuccessful change, in which the old form may maintain itself together with the new one or prevent it from developing. 1) vulgarism a) successful change: delabialization of / œ / in /Ë/ "This simplification of the articulation is of popular origin and (... ) seems to have first appeared during the XVIIth century". (Straka 1981:182) b) unsuccessful change: palatalization of / k / in / k / and of / g / in / g / when followed by a front vowel. This palatalization, which seems to have originated in the XHIth century, and for which we can find graphic evidences such as guerre written guière or quel written quieu, did not develop: "Entirely popular, this pronunciation was never accepted in correct language". (id.: 180) 2) innovation a) successful change: substitution of velar or dorsal / k / for the apicoalveolar / r / . As early as 1913, Martinon (quoted by Straka 1981:182) wrote: "There is no point in trying to get back the vibrating r, which used to be pronounced with the tip of the tongue. This r has alost completely disappearedfrom common usage, at least in towns". So much so that the apical / r / is considered to day as a ruralism. b) unsuccessful change: the liaison of nasal vowels before a vocalic initial, especially in the case of the indefinite un. In 1836, the "Société Grammaticale" had decided to pronounce un like une in that context.

Role of Hypercorrection in French Linguistic Change

123

Ex.: [ynami], in reaction against the appearance of [denami], considered to be vulgar. But in 1913, Martinon noticed that [ynami] had completely disappeared. 3) hypercorrection a) successful change .the loss of / a / ~/a/ opposition. Straka notices that in the popular parisian pronunciation, / a / tended to be closer to / o / . The confusion between / a / and / a / originated in the desire to react against this pronunciation, considered as unrefined (Straka, id.: 219). b) unsuccessful change: sigmatism of intervocalic / r / . As early as Palsgrave's time, the tendency of intervocalic / r / to be changed into / z / appeared. During the XVIth century, all grammarians agreed to reject this pronunciation: hypercorrections, in the restrictive sense of the term, could even be found. So / r / was restablished in words in which / z / was normal, as shown in this text, quoted by Marcello-Nizia 1979:86: "Madame, je vous rayme tan May ne le dite pa pourtant les musailles ont des rozeilles". Given that language observers, grammarians, chroniclers or linguists, succeed but rarely in placing themselves above ideological preconceived notions, I shall investigate, within the limits of a few points in French, the respective part played by each of these three explanatory principles which account for linguistic change. It is very likely that, in linguistics as in other fields, those who are the most reticent towards change will have a tendency to attribute it to the subgroup which they like least. On the contrary, those who, like Dell Hymes 1961, quoted by Labov 1972, don't reject the idea that change can be an improvement will insist more on the part played by the subgroup which those who hold the contrary opinion underestimate. Thus, some sort of sociologization of the principle of the least effort leads to attribute to people the tendencies to "laziness" or to "articulatory laxity" that cause the loss or the transformation of some oppositions. According to the same principle, changes due to hypercorrection would be attributed to the attempts of'sociological mutants' to rise to the level of prestige groups. In this case, one can see in "language change a kind of massive testimony to original sin" (Labov 1972:323). Thus the age-old debate started by the ancient grammarians between analogists and anomalists is the subject of ideological interpretations that lead, for instance, to ex-

124

Nicole Gueunier

plaining analogical regularizations as "simplifications" or "impoverishments" whereas to the differentiative functions of the anomaly, one opposes the risks of neological inflation or of borrowings supposedly perverting language. This pessimism can be seen, too, in the premedieval folklinguistics for which each postbabelic language is but a degradation of the adamic: thus Isidore de Seville (Vllth century) considers that "lingua prisca" was altered, going through the different phases "latina", "romana" and "mixta" (Muller 1981). Modern purism, whatever the causes specific to each linguistic community, continues with this tradition, whose function Ferguson analyses as follows: "The functions of expressing beliefs in deterioration and the need for action seems to be to provide an accepted channel of reaction to social stratification. It is sometimes a nice way of pointing out that other people do not have the benefits of one's own education". The hypothesis I would like to present, drawing on some of the data gathered for a previous paper on "the crisis in French" (Gueunier 1985) is the following: linguistic change in French has most often been understood in terms of vulgarism, imputed to the influence of popular tendencies, against which the detainers of the legitimate norm, policymakers, exemplars, teachers or linguists (see Carrington 1981) should unite in order to become its maintainers. Though it has often been pointed out when it occurred, the part of ascending classes' hypercorrection has been underestimated and I would like to emphasize a few points which show that its importance seems more decisive than is usually claimed. In France, an age-old tradition attributes to popular vulgarism the points of linguistic change which are said to be "natural" and to which intellectual upper class currents put up a scholarly resistance3. As early as Old French, the latter led to a distinction between popular and scholarly language4, which showed itself during the XVIth century, under the influence of classical humanism, and during the XXth with the development of technico-scientific terminologies, great consumers of greco-latin. Since the XVIth century grammarians, while condemning excesses of relatinization, have been reticent towards evolutions which they deemed popular. P. Fouche, treating of change in phonetism, underlines this: "The Renaissance grammarians, though they sometimes unwittingly adopted one or other popular pronunciation, neverfailed to criticize what

Role of Hypercorrection in French Linguistic Change

125

one of them (H. Estienne) calls 'the corruptions and depravities' of 'humble folk'. They constantly use the words 'plebs', 'vulgus', 'rabble'". (1952: 67) Without pretending to outline the whole history of this tradition (see François 1959), it can be said that it has continued up to the present day and informs the ideology (if not the theory) of linguistic change which I have pointed out and which is illustrated, among many others, by Bauche's classic book on early XXth century popular language: "popular language first modifies itself; then it gradually makes its way into the language of the social upper classes, with which it mingles to form the new language. Popular language goes from the pantry to the drawing-room, from the workshop in the factory to the office of the industrial manager, from the shop-girl in a big store to the rich customer". (1920:27) Yet Bauche wants to keep some kind of neutrality: "It would be as wrong to grieve over this phenomenon as to rejoice at it". (id.: 27) But the names that he applies to it leads one to doubt of this disposition: "inundation", "dangers", "destruction", "muds and garbage", "weakening", "decline" etc . . . As for Martinon, another classic, he minds his words even less : "Unfortunately, the popular language has drastically invaded the middle classes. It is quite likely that French has never been spoken so badly as today". (id.: VII) At present, under the influence of linguistic objectivism and of a clearer awareness of the links between language, social community and democratic development, such radical positions are less naively adopted and inbetween the extreme tendencies of the reactionnary purists (e.g. Tanguy Kenech'du 1984, Daguet 1984) and of the irresponsible laxities, a reasonably dirigistic trend has taken place. Thus Dupré and Batany 1972, Sauvageot 1978, Hanse 1983 are both aware of the fact that linguistic change is normal and are favourable to the diversity of its origins and expressions. A pleasant instance of this attitude is given by Cellard 1972: "As for us, let us remember with gratitude the numerous anonymous speakers of 1500years of history whose laziness and ignorance have

126

Nicole Gueunier

made French, in spite of academic barriers, a loveable and well-loved language. They are none other than the people". (ed. 1979:49) What about hypercorrection? Though I maintain that it has been systematically underestimated, it must be pointed out that it has not been completely ignored in the last centuries. But its perception has mainly confined itself to two fields, relatively restricted, compared to its true dimensions, the first: judgements concerning French in contact, the second: what we could call 'inflation' in its grammatical, lexical and discursive aspects.

Hypercorrection As for all languages, the history of French is marked by stages when contact with one or other of the neighboring or classical languages was more intense, whatever the circumstances (conquests, colonizations, trade or, as regards culture, revival of classical humanism). Three of these moments were outstanding in our linguistic history: scholarly relatinization, from the XlVth to the XVIth century, italianism in the XVIth century and anglicization, from the 'anglomania' of the XVIIIth century to the 'franglais" of the XXth century. In the three cases, however different the forms might be, the prestigious character of the new loanwords gave rise to hypercorrection behaviour, mainly in the middle classes, who would overuse the innovations imported by subgroups of dominant classes. The scholarly caricatures of the former by the latter give evidence of this assertion. Thus, when the grammarian Etienne Pasquier (1581) scoffs at the ridiculousness of the 'latiniseurs', two of his characters, one of whom was fictional, the other a living person, represent subgroups that can be called, mutatis mutandis, 'middle class'. The first one is Rabelais' 'escholier limousin' ('the scholar from Limousin'), the second, a provincial figure too, is a womannovelist from Picardy, whose novel Les Angoysses douloureuses ['painful pangs', 1538], though well known among language historians is not deemed a masterpiece of French literature (Chaurand 1977:69-73). Renaissance italianisms spread mainly under the influence of the Court and learned circles. Therefore, they have left most traces in the vocabulary of war, arts and science, which constitutes a good example of linguistic innovation, coming from the upper classes. Yet the caricatures which were made of them lead us to think that a certain infatua-

Role of Hypercorrection in French Linguistic Change

127

tion for the Italian fashion had reached the middle classes before undergoing a considerable recession (see Pasquier 1578)7. As for anglicisms and 'franglais' of the XXth century, they belong only in part to the heritage of the postclassic 'anglomania'. The latter, concentrated in prestigious activities such as political life or sport conveyed the upper and middle classes' snobbery, whereas the spreading of present day anglicisms is due to more complex and various causes. Thus, according to Chaurand (1977:156), one must take into account "a natural or technical curiosity that took a dominant place" because of "the world economic situation, industrial expansion and the supremacy acquired by the United States". Regarding the integration of borrowings, the 'innovatory' tendency of those presently responsible for linguistic planification appears in: 1) phonological, graphic and morphologlocal gallicizations: scanner

>- scanneur

listing

• listage

2) calques, merely coming from translations: architecturalfirm clip

• agence d'architecture • agrafe

3) creations more freely inventive, in comparison with the original word: brain-storming design drive-in cinema

• remue-méninges >• stylique • ciné-parc (see Fantapie and Brulé 1984).

Yet anglicism as hypercorrection remains real in everyday life. Without pretending to be exhaustive, I shall give some examples which seem to me to prove the part played in their expansion by the linguistic insecurity of the middle classes. The average Frenchman, anxious, consciously or not, to turn his secondary education to good account, finds a 'profit de distinction', according to Bourdieu's terminology, when he can display signs of his knowledge of English, which is the most widespead second language in French secondary schools. Hence, for fear of making a mistake, he shows a strong tendency to give an English pronunciation to any foreign or foreign-looking name. E.g. Daguet (1984: 88) quotes the following utterances: [insbKœk] for Innsbruck, Alexandre [njuski] for Nevsky, [kli] for Klee, [bitov] for Beethoven. The Frenchmen Eric Satie

128

Nicole Gueunier

and Jack Lang [the present Minister of Culture] don't meet with a better fate: [eKajksataj] and [djaklag]. The fact that these pronunciations were used in radio or television programs by official announcers is a further argument in favour of the interpretation of these forms as hypercorrections. In their morphosyntax, advertising language, signs and trade names of products can be seen to have a tendency to substitute the anglo-saxon order enlargement/nucleus for the regular French order nucleus/enlargement Poitou-œufs, Tours-pneus, Touraine-Véhicules-Industriels, Jules Bar or even Sport-Boutique, Moderne Hotel, Mondial Tapis. Without wishing to betray the secrets of the linguistic conscience of all the proprietors of the former "Hotel Moderne" which have become "Moderne Hotel" (or even "Modern Hotel"), one can reasonably suppose that the advantage expected from such a transformation comes from a certain idea of aesthetico-linguistic prestige. The proof is that when shopkeepers are questioned about the reasons which incite them to call their shop "pressing" rather than "laverie" or "hair dresser" rather than "salon de coiffure", they reply unanimously that "it sounds better to the customers". Finally, it can be seen with P. Lerat (1984: 73) that in the linguistic field itself, the anglicisms of "research papers, works of vulgarisation and pedagogy" indicate the tension that exists on the terminologie market: "sandhi rules (vs "les règles de sandhi") has an aura of M. I. T. which is not only a prestigefactor, but the mark of the ascendency of a school".

An inflated French The theatrical tradition of Comedy has always mocked the pedants who, from a sociolinguistic point of view, represent cultural parvenus whose linguistic insecurity is shown by various forms of overstatement. This is the point in common between Moliere's Trissotin, the semi-incultured teachers of the postclassic novels and vaudeville policemen, who have just learned how to replace "alors" by "subsequemment" when writing their reports. In modern and contemporary French, language chroniclers have rightly pinpointed the real sources from which this type of inflation springs, the most spectacular manifestations of which are primarily lexical and syntaxic. They are the "linguicides" denounced by Grandjouan's lampoon (1971): teachers, administrative staff, translators, journalists and advertising agents. Although it is not

Role of Hypercorrection in French Linguistic Change

129

possible to mention all the facts pertinently observed by Grandjouan, I shall quote some of his conclusions: School: "School wages war on living French in the name of a French which has never existed". 1971:172 N. B. On this point, it must however be noted that the existence of various pedagogical movements has produced developments which at least in part contradict this judgment. Administration "A language that is sacred is a good thing to have: he who speaks and writes it can say what he wants without the vulgar understanding. It is a good way to communicate with the other initiates (...). Those who perfectly possess the administrative language know all that one can draw from "il nous appartient" or "prendre les mesures appropriées". id.: 186. Translators "Everything that is pompous, everything that is insipid and long, vague and doubtful, all that will most certainly attract and engross them". id.: 226. Journalists "Grandiloquence and studied elegance, allied to a schoolboy's fear of repetition, make refined pseudo-synonyms flourish: "instance"for "tribunal", "forum" or "tribune" for "journal", "dans cette enceinte" for "ailleurs". id.: 235. On the syntactic level, one can note the overuse of the narrative imperfect and "adjectivitis": "le phénomène prostitutionnel" for "la prostitution". Advertisements "Les réclamiers" are our "Précieux". id.: 264. This last phenomenon can readily be perceived on observing such syntaxic examples as: a) the desarticulation of sentences designated by Cellard (1979: 119) under the name of "style coup de points" :

130

Nicole Gueunier

"Chez Hag, nous avons des mélanges différents. Que nous décaféinons différemment. Pour avoir un arôme différent". [with Hag, we have different mixtures. That we decafinate differently. To obtain a different aroma.] b) the excessive use of the inversion: "Elle est belle, intelligente. Mais la ronge le dégoût de vivre (...). L'attire un moment l'innocence". [She is beautiful, intelligent. But gnaws at her the disgust for living, attracts her for one instant, innocence.] Quoted by Sauvageot (1978:93.) Hence we have located some of the places and instances in which the tendency to hypercorrection appears. To conclude, I would like to draw attention to a field where it seems to me that hypercorrection is still more manifest, that of pronunciation. Not that morphosyntax is not deserving of attention. Radiophonie or televisual conjugation would provide many an example. For hypercorrections which show up as mistakes, the use, recorded by Cellard, of résolveront for résoudront: "résolveront is a hypercorrection. The writer (and the corrector) know that the conjugation of the verb is tricky. They deduce confusedly that the most complicatedform is the right one". 1979:1973. In the same way, I noted in February 1985, in the news on Radio, the following form: "il serait étonnant que le Parti Communiste* [aplodisa]". Is it * applaudissa vs applaudit or * applaudissât vsapplaudît? Whatever the exact form intended, one can suppose that the combined search for the subjunctive required by the completive sentence and the past tense consequent on the conditional serait produced this parasite form, preferred to the simple [aplodi]. In view of the polyvalence of the conjugaison on this point, [aplodi] could lead one to think that the speaker ignored the imperfect subjunctive. But let us leave morphology. Indeed the field of pronunciation, less contrôlable by the speaking person, seems to me, because of this very fact, richer in instruction as concerns our subject. I shall therefore take two sorts of examples, some arising from the obvious behaviour of hypercorrection, others which deserve discussion.

Role of Hypercorrection in French Linguistic Change The first obvious case concerns the liaison in the spoken press. Traditionally, it was considered that popular language tended to reduce liaisons: "Liaisons are made less and less in French. In popular French they have almost completely disappeared" writes Bauche (1920: 56), who adds: "There are suppressions of liaisons which are essentially popular and which indicate an appartenance to the common people (...) So son arrivés or on arrivé for sont arrivés". An opinion shared by linguists: Guiraud (1965: 106), Delattre (1966: 58) and all the works quoted by Encrevé (1983). Functioning as a stereotype, in the sense given by Labov (1972), the liaison will therefore normally be the object of especial care on the part of those requiring and of those providing normative models. Among the latter, the announcers, among whom chroniclers spot: 1) an increase of mistaken liaisons: * quatre mille- z - ouvriers vs quatre mille ouvriers * sans même-z- hésiter vs sans même hésiter Grandjouan 1971:242. 2) an increase of excessive liaisons : ? il part-t-en courant vs ilpart en courant ? le président-t-américain vs le président américain Cellard 1979:152. 3) a corresponding increase of popular forms with zero liaison: * ils rentr'à Paris vs ils rentrent-t-à Paris How can one interpret the contradictory coexistence of 1) 2) on the one hand and of 3) on the other? The explanation given by Grandjouan is the following: "This modest chapter of liaisons perfectly illustrates the deep-rooted malaisefrom which the press suffers: it is so busy with its "talons rouges" that it no longer sees in what crap it steps". ibid. N. B. "talons rouges" is the name given to French XVIIIth century refined aristocrats and courtiers who wore shoes with red heels.

131

132

Nicole Gueunier

Explanation confirmed by Cellard: "The fashion is today, on radio and television, to liaise without rhyme or reason, for fear no doubt of talking as the people to the people himself (... ) irritating manifestation of the wish to "faire bien" (to make a good impression) and not to "bien faire" (do well). ibid. Such an interpretation may lead to formulating new hypotheses in regard to some facts of linguistic change traditionally interpreted as vulgarisms. Thus mistaken liaisons of pronominal morphemes: a) b) c) d)

* les gens qui-z-y vont * les gens qui leur-z-ont dit *fous lui-z-y sur la gueule * donne moi-z-en Frei, 1929.

A traditional explanation should, it seems to me, be rejected, that of the introduction of an euphonic phoneme destined to avoid the hiatus. If this is the case, how can one account for the hiatus [kâ?5di] (quand on dit), [s5?aKive] (sont arrivés) so frequent in the spoken press? In the same way, one must probably question a syntaxical interpretation of a), often rendered by attempts at a graphic representation of popular speech: les gens qu'ils y vont for [kizivô] The underlying hypothese to such graphies is a morphological reduction of the paradigm of simple relatives qui que dont to the simple elidible form que: * les gens qu(e) ils y vont vs quiyvont * les gens qu(e) parle vs dontje parle Personally, I should propose to distinguish cases a) b) on the one hand and c) d) on the other. The last two examples effectively illustrate the allomorphs of the second pronouns, with an initial vowel: [i] [â]

[zi] [zâ]

(y, z'y) (en, z'en)

rather those of the first ones: ? [mwaz]

? [lqiz]

In these contexts, the forms [zi] and [zâ] seem simply analogous with those that one finds after imperatives with final consonant whether the

Role of Hypercorrection in French Linguistic Change

133

consonant has a morphological function, as in fous-y [fuzi], or a euphonic one as in vas-y [vazi], (on this point, see Blanche-Benveniste 1983:34-35). On the contrary, in the cases of [kiz] and of [lœKz], a likely hypothesis is that the subjects, aware that most NP carry the marks of the number, have a tendency, orally, to generalise the rule to invariable pronouns, especially in so far as, by doing so, they are brought to make use of the liaison, source of "profit de distinction". Now this type of rule generalization is, according to Decamp (1972:87) very much a definitive feature of hypercorrection. To support the same hypothesis, I shall quote a recent phonetic change, since it dates from the 1970s. It is what Encrevé (1983) calls the liaison without linking (liaison sans enchaînement), of the type feavez'ÊKev] instead of foavezëKev] (I had a dream) [393Y3ez 5esâsjel] instead of [333y3£zesâsjel] (I considered essential) It is defined both by the presence of a liaison consonant (as in the linked liaison) but also by the absence of linking the following syllable, which has a vocalic initial. Studied in broadcast or television speeches by six contemporary Heads of State and nineteen politicians, it is not without some relationship to "spelling pronunciation" and Encrevé presents it as a change moving towards a prestige form. Is it a hypercorrection or a cultivated innovation? According to Encrevé it does not only characterize upper class lects but also: "most speakers from categories of professionals in public address, journalists on radio and television, intellectuals (...), preachers, barristers etc ... It can also be observed among almost all non-professionals of public speaking who effect optional liaisons". p. 61. "Today it appears in great increase among speakers generally, most of whom have been to schoolfor a considerable time and [can be explained, among otherfactors by] the tendency tofollow the school norm as regards optional liaisons, a tendency made more acute by the fear of the hiatus and the awareness of the social valorisation of these liaisons". p. 67. Another change, for which hypercorrection seems to me to provide the explanation, is the increase in the utterance of twin consonants in words such as [gKammeK] vs [gKameK], [ilb3ik] vs [ilo3ik] etc . . . Al-

134

Nicole Gueunier

though the point has not been the subject of a survey as far-reaching as the liaison (see Straka 1981: 244, 245), the evolution of the latest pronouncing dictionaries (Warnant 1968, Lerond 1980, not to mention older ones), is significant in this respect, but only a detailed sociolinguistic inquiry could confirm the hypercorrective hypothesis. The latter should also be tested to treat of the use of doubling / in pronoun le : je l'ai vu feellevy] qui I'a cru [killakKy] while traditional explanations are functional from a linguistic point of view (in the second example the doubling does away with the ambiguity between qu'il a cru and qui I'a cru) and in favour of the vulgarism, from a sociolinguistic point of view. Among doubtful cases, that is those in which one may hesitate between hypercorrection, vulgarism or scholarly innovation in the interpretation of facts of linguistic change, one can quote that of spelling pronunciations of mute letters, final or not, in words such as [dotceK] or [dSptcBK], [ftel] or [feptel], [mceK] or [mceKs], [o] or [os] etc . . . In this field, tradition itself is hesitant. Dupuy (1836, quoted by Straka) states that "people who speak well" will never say donter, dontable e t c . . . An opinion confirmed by the Academy (1878) but contradicted by Littré. Even today, [doptoeK] is rejected by Hanse (1983) but accepted, not without reserve, by Lerond (1980). While Warnant (1968) recommends [le] for legs and [mceK] for moeurs, Lerond declares: "[leg]; archaic: [le]" and [mceKs]; archaic: [mceK] "and Hanse also accepts them, in current speech, it is true. To explain this type of evolution - for which detailed surveys have yet to be made - I should again be inclined to prefer the hypothesis of hypercorrection to the two others in so far as it is doubly integrated in the more general hypothesis of prestige expected to be gained from graphic pronunciations by the middle classes and in which this type of pronunciation, echoing the French adaptations on English models, in which final consonants are performed, also corresponds to a demand for prestige. Other cases in which the interpretation is doubtful require a strictly paradigmatic study. Thus, the performance of vocalic length oppositions is difficult to account for, in so far as it is necessary first to neutralize the regiolectal factor responsible, even in cultivated lects, for numerous long variants. The same is true for oppositions in vocalic pitch, although we have been able to show, as regards the opposition / e / ~ / e / in different socio-cultural milieux and in different French urban areas, how its normative pronunciation is function of hypercorrection (s. Gueunier et al. 1978).

Role of Hypercorrection in French Linguistic Change

135

In the consonantic paradigm, it would certainly be interesting to measure the respective roles of hypercorrection and of innovation in the recent increase of the nasal velar / q / borrowed from English. Regional surveys by Walter (1982, 1983) could be completed from this point of view 1) by a survey on the parallel expansion, though less frequent, of the phoneme /&$/ also borrowed from English, or in proper names (Jack, James, Jenny, Jerry, Jim, John), or in common nouns, whether it appears in the initial position (jazz, jean, jerk, job, jogging) or in median (management, teen-ager, etc.). 2) by sociolinguistic investigations making possible a systematic correlation of usage according to sociological categories and to speech situations. To conclude this survey of the perspectives that the explanation of change due to hypercorrection opens to French sociolinguistic studies, I should like to underline that they appear interesting not only from an empiric but also perhaps from a theoretical point of view. If these hypotheses are verified in wider fields and domains than those mentioned here, they will make it possible to question the belief, still held by a majority, according to which change is to be attributed to the extension of popular forms of speech. We shall have to be careful, moreover, not to fall into the opposite extreme of overestimating the role of hypercorrection in the middle classes! On this point, one should take into account the warning given by Gumperz, who invites to beware of "the assumption that to each category there is a specificform of speech, or idiom, distinct from all others". 1966:28. It is indeed important not to reify these lects and to see the dialectic relationships which control the dynamic, to avoid superimposing on reality a preconceived theory, all the more so that between the "elite" and the "envious mass" whom Fischer (1958) makes the principal actors of linguistic change, there are many cross-channels. Interesting therefore on the sociological plane, this hypothesis is also from a strictly linguistic point of view, if it can help to find out the general lines of development, for example, if it confirms the generalization of the number markers in forms that are still invariable in NP or the borrowing, not only lexical but also syntaxic, from English. Lastly, on the epistemological plane, I would point out the fact that this hypothesis enables us to question some folklinguistics notions at-

136

Nicole Gueunier

tached to previous attempts to explain change, such as "naturalness" and "graduality" of popular change, as well as those of "simplification" and "elaboration".

Notes 1 Here is a corresponding example in French : the barbarism * vous médites (vs vous médisez) comes from the overgeneralization of the rule which prescribes for faire and dire specific forms of the second plural person: faites, dites. The hypercorrect * medites is due to the reaction to the analogical tendency which leads to producing *faisez, *disez. 2 Thus the statistical increase, in idiolects or sociolects, of the percentage of correct realizations for unstressed French pronouns, as found in formal contexts, such as fôœ] ~ [3] vs current French [f] before unvoiced consonant [ty] — [tij] vs current French [t] before vowel [il] vs current French [i] before consonant etc... 3 The opposite arguments, who emphazise scholars' and exemplars' role in linguistic change can also be found (François 1959, Steinmeyer 1981), but they are in the minority. 4 Schane 1973 holds this division responsible for the high degree of abstraction of French phonology. 5 This gradual conception is today refuted by the current theories of linguistic change (see Weinreich and al. 1968). 6 E.Pasquier 1586 charges these characters' language as "escorche-latin". 7 The following specimen of italianized French (quoted by Brunot) shows how important this background movement is: "l'ay bonnes iambes (de quoy Dieu soit ringratié), mais i'ay batu la strode..." 8 "Hypercorrection is always a function of rule generalization, though the converse does not hold" {p. 87).

Bibliography Bauche Henri, 1920. Le langage populaire. Grammaire, syntaxe et dictionnaire du français tel qu'on le parle dans le peuple de Paris avec tous les termes d'argot usuel, Payot. Blanche-Benveniste Claire, 1983. "L'importance du 'français parlé' pour la description du 'français tout court', Recherches sur le français parlé, n° 5, Groupe Aixois de Recherches en Syntaxe, Université de Provence, Marseille, éd. Jeanne Lafitte. Carrington L.D., 1981. "Le conflit linguistique à l'école dans les Caraïbes", Bulletin de Psychologie, juin-août.

Role of Hypercorrection in French Linguistic Change

137

Cellard Jacques, 1979. La vie du langage, Chroniques 1971-1975, Le Monde, Le Robert, coll. "L'ordre des mots". Chaurand Jacques, 1977. Introduction à l'histoire du vocabulaire français, Bordas, coll. "Linguistique". Daguet Dominique, 1984. Langue française à l'épreuve, Troyes, Librairie Bleue. Decamp David, 1972. "Hypercorrection and Rule Generalisation", Language in Society, n° 1. Dupré P. et Batany Jean, 1972. Encyclopédie du bon français dans l'usage contemporain, Editions de Trévise. Encrevé Pierre, 1983. "La liaison sans enchaînement", Actes de la Recherche en Sciences Sociales, n° 46. Fantapie Alain et Brulé Marcel, 1984. Dictionnaire des néologismes officiels, Franterm, diff. Nathan. Ferguson Charles, 1979. "National Attitudes toward Language Planning" in Alatis J. E. and Tucker G. R., Language in public life, Georgetown University Round-Table on languages and linguistics, Washington, Georgetown University Press, 1979. Fischer John L., 1958. "Social Influence in the Choice of a Linguistic Variant", in Dell Hymes, Language in culture and society, 1964, N.Y., Harper and Row. Fouché Pierre, Phonétique historique du français, Klincksieck. Francois Alexis, Histoire de la langue française cultivée, Genève, Alexandre Jullien. Frei Henri, La Grammaire des fautes, Geuthner. Grandjouan J.O., 1971. Les linguicides, Didier. Gueunier Nicole, Genouvrier Emile, Khomsi Abdelhamid, 1978. Les Français devant la norme. Contribution à une étude de la norme du français parlé, Champion. Gueunier Nicole (in press), "La crise du français" in La Crise des langues, Québec, CLF, Paris, Le Robert, coll. "L'ordre des mots". Guiraud Pierre, 1965. Le français populaire, P.U.F., coll. "Que sais-je?" Gumpenz John, 1966. "On the Ethnology of Linguistic Change" in Bright William, Sociolinguistics, Mouton. Hanse Joseph, 1983. Nouveau dictionnaire des difficultés du français moderne, Brucelles, Duculot, Paris, Gembloux. Labov William, 1972. Sociolinguistic Patterns, University of Pennsylvania Press, Philadelphia. Lerond Alain, 1980. Dictionnaire de la prononciation, Larousse. Marchello-Nizia Christiane, 1979. Histoire de la langue française aux XlVe et XVe siècles, Bordas. Martinon Philippe, 1927. Comment on parle en français, Larousse. Lerat Pierre, 1984. "Anglicisme et emprunt terminologique", Le Français dans le monde, n° 183. Muller Bodo, 1981. "Remarques sur la prospective de la langue française" in La prospective de la langue française, Colloque de Sassenage, Paris, C.I.L.F. Pasquier Etienne, 1586. Lettres, in Rickard Peter, 1968, La langue française au XVIe siècle, Cambridge University Press. Sauvageot Aurélien, 1978. Français d'hier ou français de demain, Nathan.

138

Nicole Gueunier

Schane Sandford A., 1973. "Sur le degré d'abstraction de la phonologie du français", Langages, n° 32. Steinmeyer Georg, 1981. Historische Aspekte des Français Avancé, Droz. Straka Georges, 1981. "Sur la formation de la prononciation française d'aujourd'hui", Travaux de linguistique et de littérature, Université de Strasbourg, Klincksieck diff. Tanguy Kenech'du, 1984. Avatars du français - De Rivarol aux néo-linguistes, Tequi. Walter Henriette, 1982. Enquête phonologique et variété régionale du français, P.U.F., coll. "Le Linguiste". Walter Henriette, 1983. "La nasale vélaire, un phonème du français?", Langue Française, n° 60. Weinreich Uriel, Labov William, Herzog L. M., 1968. "Empirical Foundations for Theory of Language Change" in Lehmann W. and Malkiel Y., Directions for Historical Linguistics, University of Texas Press.

Naming in the Kashmiri Pancjit Community: Sociolinguistics and Anthroponymy Braj B. Kachru University of Illinois Urbana, Illinois

In earlier research on onomastics the concentration has been on two types of studies: One type focuses on the data on naming processes with reference to a well-defined social, regional, or religious group. The other type is essentially comparative, and has implications for defining and illustrating concepts such as "linguistic area" or "sociolinguistic area". There is a profusion of studies of the first type, but rather few on the second type. An example of a comparative study with typological and sociolinguistic interest is Emeneau 1978. In this preliminary investigation in anthroponomy of a small community of India, generally known as Kashmiri pandits and locally known as bati (Skt. bhartri), I have attempted to combine these two approaches. The pandit community forms an extremely small minority in India, not exceeding over one hundred thousand people. Of this total, a significant number live in Kashmir valley, mainly in Srinagar, and the rest have settled in the major cities of Allahabad, Agra, Delhi, Lucknow and Jaipur in northern India. During the last seven hundred years, this minority community has passed through a series of social, political, and cultural vicissitudes.1 Traces of these changes are naturally present in their language and in their naming system. The language of Kashmiri pandits is called kasur (or Kashmiri). This language has traditionally been treated as a member of the linguistic sub-group termed "the Dardic family", which is within the Indo-Aryan branch of languages. The Dardic family has been identified on the basis of several phonological, syntactic, and lexical characteristics of three language groups: the Kafiri-group, the KhowSr-group, and the Dard-group.2

140

Braj B. Kachru

However, the classification of Kashmiri continues to be controversial: There still is a need to draw evidence from those areas of language and culture which might throw more light on the controversy concerning the affiliation of Kashmiri language, and the types of convergence it has gone through. A study of names should provide evidence to show that Kashmiri forms a part of South Asian sociocultural and linguistic area. In a stimulating paper "Toward an onomastics of South Asia" Emeneau (1978) has rightly suggested that, in the case of South Asia, onomastics provides interesting ethnographic, cultural, and historical evidence. In his paper Emeneau examines " . . . features of the naming system and their spread in the South Asian linguistic area" (128). There is evidence that the patterning of the naming system, the choices available within the system, and the functional use of such choices provide insightful data for characterizing South Asia as a "sociolinguistic area". I shall not go into any detail in explaining here the concept "sociolinguistic area" since this forms the topic of a study by D'souza (1985).3 What I propose to do is to discuss selected processes of naming among the pandits of Kashmir within a sociolinguistic context. These observations may hopefully, provide more data from an interesting linguistic and cultural area of India for our understanding of India as a "sociolinguistic area". The first issue one faces in a study of onomastics is to determine the proper approach. The body of literature on onomastics is aweinspiring, with a long tradition of research from various disciplines: among others, anthropology, (e.g., Evans-Pritchard 1948), etymology (e.g., Bradsley 1915; see also for a bibliography Smith 1950,277-282), linguistics, (e.g., Algeo 1973; Uhlenback 1969) and philosophy (e.g., Donnellan 1970). There are also several studies which provide useful theoretical discussions from various viewpoints (e.g., Gardiner 1940, Pulgram 1954, and Sorenson 1963). In addition, there are numerous atheoretical studies. However, the sifting through of this abundant body of material is not always rewarding. Zgusta (1974: 819) rightly warns us that "the study of names is plagued with imaginative and fantastic, but unfounded and unscholarly, publications; therefore, caution should be exercised in the choice of sources." In earlier literature very little attention has been paid to the linguistic aspects of onomastics and its contribution to linguistic and sociolinguistic theory and description, since linguists have considered onomastics "etymologically explanatory rather than systematically descriptive" (Allen 1968; quoted in Uhlenbeck 1969:321). There are very

Naming in the Kashmiri Pandit Community

141

few attempts by linguists, like that of Uhlenbeck (1969: 321-335) who make an effort to show "that the synchronic study of proper names may be an interesting and rewarding subject for the linguist as well as for the ethnographer and that at least in some linguistic communities personal names may form a well-defined subsystem within the whole of the linguistic structure" (321-322). In his detailed study Utley (1963:145-176) draws attention to the linguistic component of onomastics and the neglect of this field by linguists. He observes that " . . . the responsibility for correction lies not only on the historian, logician, and etymologist, but also on the modern linguist, structural or transformational, who has been slow to plow on onomastic pastures" (145). The Indian languages are no exception to this general neglect by linguists.4 The result is that the linguistic aspects of onomastics in these languages have not attracted serious attention of linguists, and no attempt has been made to provide typological statements which contribute to our understanding of India as a linguistic area or a sociolinguistic area (Emeneau 1978 is an exception). There are, however, several language-specific studies which provide useful data and descriptions (e.g., Dutta 1981; Jhungare 1975; Sjoberg 1968; and Temple 1883). There are not many observations in earlier literature on Kashmiri or Kashmiri papdits, except in Koul (1924:18-23). This study, therefore, is one of the first attempts toward understanding anthroponymy of Kashmiri papdits in relation to sociolinguistics. Naming and gotra It is claimed that traditionally Kashmiri papdit last names indicated gotra (exogamous groups); these goira-marking names were initially restricted to three names which Koul (1924: 20) terms the "three principal divisions", i. e., bhat, pandit and rázdán. According to him, from these three "are derived the distinctive appellations of koul, sopori, papdit, and rainá. From these three families, [bhat, papdit and razdan] as each took to a particular occupation, or by adoption or intermarriage with other gotras, other gotras came into existence" (20). These gotras bear the names of risis ("sages"), and to some extent, as Koul says, they form a hierarchy. In this hierarchy the kol (Kaul or Koul) are considered very high (Koul 1924: 20). It is claimed that by intermarriage and intermixture "with other Brahmins the number of gotras multiplied to 199" (Koul 1924: 20). As time passed, the number of last names substantially increased. I will discuss the reasons for this later in this paper.

142

Braj B. Kachru

Naming-Ceremony and Nameability Criteria The ritualistic aspects of naming among the papdits are not significantly different from the brahmap groups of other parts of India. In a traditional papdit family the naming-ceremony (Hindi, namkaran) takes place on the eleventh day after a child's birth. It is called kahnethir (kah "eleven" and nethir "wedding"). At this ceremony a child is expected to receive his or her given name. The names of the new-born may be chosen by various methods. In certain families the family priest pandith (Skt. pandit) or gor (Skt. guru) might suggest the first letter of the name. The suggestion of the kaligor (Skt. kulguru "family priest") is based on clues from the horoscope of the new-born child. Suggestions may be made by relatives from the mother's side or the father's side. Naming after parents, or grandparents, as is common in the Western world, is not practiced among the Kashmiri Papdits. The earlier practices of namkaran ('naming-ceremony') are, however, slowly changing as we will see in the following section. The two productive processes for naming are Sanskritic and Persian. Muslim first names do not show the process of Sanskritization. However, a number of Sanskrit last names (Kashmiri zath) have been preserved by some Muslims after their conversion to Islam, for example, bath (Skt. bhatta), pandith (Skt. pandit).5 On the other hand, Hindu given names and last names show the influence of both Sanskritization and of Persianization. The result is, as Emeneau (1978:117) also attests in another context, "Hindu names of hybrid form" such as jawahar ("jewel"). In Kashmir, this hybridization of names still continues.

Dynamics of Kashmiri Papdit Names The dynamics of the papdit names show mainly three types of changes. Generation-distance is the main factor in the changing pattern of the given names. As in other language areas, the changing social, political, and linguistic influences on the land and the people of Kashmir have left their impact on naming patterns. Consider, for example, given masculine names such as aftab 'the sun', and toti 'parrot' (as in totikakh), or feminine names such as goni 'virtue', (as in gonivati('the possessor of virtues'). These given names are now only of antiquarian interest and form part of the diachronic onomastics of the papdit com-

Naming in the Kashmiri Pan/fit Community

143

munity. These names suggest a distance of more than one generation. Names such as makhni (Hindi-Urdu, makkhan 'butter') and muti (Hindi-Urdu, moti 'jewel') are much less frequent in the present generation.

Components of a Pagdit Name The naming system of the pandits follows the pattern of other IndoAryan languages. A pandit name has three components: The first name or given name, the middle name, and the last name (zath) or 'family' name. The first and last items are obligatory and the middle item is optional.6 (For example, madissdan (Skt. madhusudan ("the killer of the demon Madhu"), roginath (Skt. raghunath "master of Raghu dynasty"); omkar(Skt omkar). The given name has the structure of a noun phrase in which the family name occupies the position of head, unlike names in the Dravidian languages in general. In Telugu, as Sjoberg (1968: 314) shows the Family names appear to stand in an adjectival relationship to given names. That is, as an adjective in the Telugu language precedes the noun it qualifies, so too the family name precedes the given name, which is clearly a noun.

The types of compounds and modifiers which form a given name again fall into a typical Indo-Aryan pattern as does the constituent structure of names. The lexical sets which occupy the modifier and head positions may be characterized in terms of semantic sets. These sets provide attitudinal, locational, and other clues. I shall discuss some of these sets below.

Given Name and the kram The given names are chosen out of a variety of lexical sets, the most common being theophoric names, or names which refer to various gods of the Hindu scriptures and mythology. As mentioned earlier, a new-born child is not necessarily given a name at its birth. A name may be given to the child at his kahnèthir "naming ceremony" or at a convenient later time. In the meantime, a child may be called niki (fem. nic) 'baby' until it is given a name. In certain cases niki (nic) may continue to be used as an alternate name, both for reference and address.

144

Braj B. Kachru

The term kram (a class name) is traditionally used to make a distinction between the nick-names and traditional last names or zath (e. g., last names such as Koul, Bhatt). The kram slowly gets institutionalized as the last name. The last name, therefore, need not be an indicator of gotra, but might develop out of the sobriquets (nick-names) acquired by the family for various reasons. Sobriquets are acquired in numerous ways: occupation, color, physical characteristics and so on. Toward Structuring kram The kram or 'nicknames' acquired by each family may be listed in terms of lexical sets of sobriquets. Sobriquets were originally aptly called in English ekenames (an 'also-name'; see Smith 1950: 74-92). These sets share semantic characteristics and provide semantic classes for the zath of Kashmiris. The following /cram-denoting lexical sets are illustrative. Attitude-marking: The use of attitude-marking terms is a good example of making a generalization from an individual to the whole family. Such terms must have originally started as nick-names, and, then, slowly acquired the status of kram. There are, therefore, cases among Kashmiri Pandits where two brothers might use two distinct last names, one using the kram and the other the original last name. For example, one brother may use kol (Koul), and the other brother and his children use tut (Anglicized as Tutu). Kol is goira-marking ancestral last name, and tut is the acquired Aram-marker. The attitude-marking terms have an underlying meaning which conveys an attitude and focuses on characteristics of certain types. Such terms can be subdivided into a number of classes on the basis of the shared characteristics of the members of a class of last names. Consider, for example, the following sub-classes. Animab:

brar 'cats'; gagar, 'mice'; hapath, 'bears'; hast', 'elephants'; hSgal, 'stage'; khar, 'donkeys'; panz, 'monkeys'. Birds: bulbul, 'nightengale'; kav, 'crows'; kokar, 'roosters'; kotar, 'pigeons' Edibles: vugri, 'watery rice' Fruits: badam, 'almonds' Spices: ganhar, 'poppy seeds' Utensils: