Language Death: Factual and Theoretical Explorations with Special Reference to East Africa 9783110870602, 9783110134049

Table of contents :
Part I
Social contexts of language death
Theory of language death
Codeswitching as a mechanism of deep borrowing, language shift, and language death
Language decay and contact-induced change: Similarities and differences
Part II
East African case studies
The fate of ethnic languages in Tanzania
Language shift in Tanzania
Reduction in Kore reconsidered
Dahalo: An endangered language
Language death and the origin of strata: Two case studies of Swahili dialects
Chifundi and Vumba: Partial shift, no death
Lexical retention in language shift: Yaaku/Mukogodo-Maasai and Elmolo/Elmolo-Samburu
Dialect death: The case of Terik
Language shift among the Suba of Kenya
175 years of language shift in Gweno
Part III
A survey of language death in Africa
Names of scholars
Subject index
Index of languages (and variants)

Citation preview

Language Death in East Africa

Contributions to the Sociology of Language

64

Editor

Joshua A. Fishman

Mouton de Gruyter Berlin · New York

Language Death Factual and Theoretical Explorations with Special Reference to East Africa

Edited by Matthias Brenzinger

Mouton de Gruyter Berlin · New York

1992

Mouton de Gruyter (formerly Mouton, The Hague) is a Division of Walter de Gruyter & Co., Berlin. ® Printed on acid-free paper which falls within the guidelines of the ANSI to ensure permanence and durability.

Library of Congress Cataloging in Publication Data

Language death : factual and theoretical explorations with special reference to East Africa / edited by Matthias Brenzinger. p. cm. — (Contributions to the sociology of language ; 64) "Articles ... based on contributions initially prepared for an international symposium on 'Language death in East Africa,' held at the Werner-Reimers Foundation in Bad Homburg (Germany) from January 8 — 12, 1990" — CIP preface. Includes bibliographical references and index, ISBN 3-11-013404-7 (acid-free paper) 1. Language obsolescence—Africa, East — Congresses, 2. Code switching (Linguistics) — Africa, East — Congresses. 3. Sociolinguistics—Africa, East—Congresses. I. Brenzinger, Matthias, 1957— , II, Series. P40.5.L332A3535 1992 306.4'4'09676—dc20 92-25843 CIP

Die Deutsche Bibliothek — Cataloging in Publication Data Language death : factual and theoretical explorations with special reference to East Africa / ed. by Matthias Brenzinger. — Berlin ; New York : Mouton de Gruyter, 1992 (Contributions to the sociology of language ; 64) ISBN 3-11-013404-7 NE: Brenzinger, Matthias [Hrsg.]; G T

© Copyright 1992 by Walter de Gruyter & Co., D-1000 Berlin 30. All rights reserved, including those of translation into foreign languages. No part of this book may be reproduced or transmitted in any form or by any means, electronic or mechanical, including photocopy, recording, or any information storage and retrieval system, without permission in writing from the publisher. Typesetting: Danny Lee Lewis, Berlin. — Printing: Gerike GmbH, Berlin. — Binding: Lüderitz & Bauer, Berlin. Printed in Germany.

Preface

The articles compiled in the present volume are based on contributions initially prepared for an international symposium on "Language death in East Africa", held at the Werner-Reimers Foundation in Bad Homburg (Germany) from January 8-12, 1990. The round table meeting, organised by the Institut für Afrikanistik of the University of Cologne, was attended by fourteen invited participants, who contributed papers on the social motivation for language contraction in East Africa, and the subsequent structural changes in the grammars of these languages. The main aim of the symposium was to provide case studies in order to help to understand the processes of language shift better and to stimulate further research towards a theory of language death. Part I of this volume deals with a wider range of languages and pertains to the theory of language death. Part II constitutes a collection of case studies from East African, and finally Part III provides an overview of African languages relevant to our topic. Different approaches were chosen by the authors in describing and analysing the various aspects of specific language shift situations and cases of language death. Not only because of this diversity but also because of the original data and the intimate knowledge of the presented cases which the authors have, we hope that the reader will benefit from this work. We would like to thank the Volkswagen-Foundation for sponsoring the symposium, and the Werner-Reimers Stiftung for providing an excellent atmosphere in their "Tagungsstätte". Furthermore, we are grateful to Carla Butz for drawing the maps in the volume. Matthias Brenzinger

Köln, November 1991

Contents

Part I

1

Matthias Brenzinger and Gerrit J. Dimmendaal Social contexts of language death

3

Hans-Jürgen Sasse Theory of language death

7

Carol Myers-Scotton Codeswitching as a mechanism of deep borrowing, language shift, and language death

31

Hans-Jürgen Sasse Language decay and contact-induced change: Similarities and differences

59

Part II

81

Matthias Brenzinger East African case studies

83

Herman Batibo The fate of ethnic languages in Tanzania

85

Karsten légère Language shift in Tanzania

99

Gerrit Dimmendaal Reduction in Kore reconsidered

117

Mauro Tosco Dahalo: An endangered language

137

Wilhelm J. G. Möhlig Language death and the origin of strata: Two case studies of Swahili dialects

157

Derek Nurse and Martin Walsh Chifundi and Vumba: Partial shift, no death

181

vili

Contents

Matthias Brenzinger Lexical retention in language shift: Yaaku/Mukogodo-Maasai and Elmolo/Elmolo-Samburu

213

Bernd Heine Dialect death: The case of Terik

255

Franz Rottland and Duncan Okoth Okombo Language shift among the Suba of Kenya

273

Christoph Winter 175 years of language shift in Gweno

285

Part III

299

Gabriele Sommer A survey of language death in Africa

301

Names of scholars Subject index Index of languages (and variants)

419 425 431

Part I

Social contexts of language death Matthias Brenzinger and Gerrit J. Dimmendaal

Setting aside the rare case of language extinction through genocide (in Africa only reported within the Khoisan language family so far), all instances of language death are the result of language shift. Investigating processes leading to language death, therefore, means studying language shift situations. Here it should be kept in mind, of course, that language shift does not necessarily result in language obsolescence. Very often, ongoing processes of language shift are halted or even reversed, thereby resulting in revitalization, a case in point being "Suba" on the western shores of Lake Victoria, as described by Rottland and Okombo in this volume. No consensus seems to exist as to when a particular language is to be regarded as "dead"; in the present context, a language is considered to be extinct when there is no longer a speech community using the language. This applies to East African languages discussed in the present volume such as Kore discussed by Dimmendaal, Yaaku and Elmolo discussed by Brenzinger. With regard to the phenomenon of language death two levels seem to be involved: (1) the environment, consisting of political, historical, economic, and linguistic realities; (2) the speech community, with its patterns of language use, attitudes, and strategies. Concerning the first level, factors such as status, demography, institutional support, and cultural (dis)similarity are relevant (cf. Giles 1977, AppelMuysken 1987). These influence the second level, that of the speech community in which a particular language is left to die. In terms of causal relations, then, changes within the speech community very often have to be understood as reactions towards environmental changes. Most case studies in the present volume deal with minority languages since they are the ones threatened by extinction in language shift situations. Such languages can only survive if its speakers are willing to maintain an active interest in retaining them, i.e. in showing strong language loyalty, as

4

Matthias Brenzinger and Gerrit J. Dimmendaal

with the Dahalo, described by Tosco in the current volume. The minority language has to be valued highly by the members of a speech community in order for it to survive a generally hostile environment. Apart from internal deficits experienced by minorities, such as the limited communication yield caused by the restricted distribution of the language, there are often external pressures on minorities, due to stigmatization, exclusion from education and political participation, and economic deprivation. Whereas in general one may arrive at diverging conclusions as to whether language is the most important component in constituting ethnicity, the situation is different with minorities. In larger groups, language may be of secondary importance for self-identification; with groups speaking minority languages, however, language is always an eminent component, since minority languages are doomed to vanish, if strong language loyalty is no longer present. Patterns of language choice reflect language attitudes. Therefore, in cases of language shift one has to investigate underlying changes in attitudes towards the languages involved, namely the abandoned language and the target language (cf. Thomason-Kaufman 1988). As observed by Heine in his contribution on the Terik in Western Kenya, there is often a discrepancy between self-assessment in terms of speech behaviour and one's actual language use in this respect. Being "a good Terik speaker" is determined more by environmental factors than by purely linguistic facts. Stable bilingualism is also a widespread feature of African societies. Secondary languages are widely spoken in such domains as market places, schools, or at work, especially by members of minorities, since they are generally forced to use a language other than their primary language in communicating outside the group. Examples of language shift at the local, regional, and national level and their effects on notions of ethnicity can be found in the presentations by Batibo and Legère for Tanzania. Such patterns of language shift put a minority language into jeopardy, because the use of the language tends to be restricted to such domains as home, traditional ceremonies, peer group meetings, etc. The actual process of abandoning a language may be observed in a decrease in a) number of speakers, b) functional domains, c) competence. Next to these aspects of the social context of language death, there are structural-linguistic consequences of language death. Sasse's contribution, "Theory of language death", focusses on these latter issues. The author elaborates on Nancy Dorian's pioneering work in this respect, and investigates the nature of the reduction, as well as problems of interpretation regarding internal versus external causes of reduction processes.

Social contexts of language death

5

Myers Scottoti addresses linguistic and social-psychological aspects of selection by bilinguals of material from an embedded language in a matrix language in conversation; it is argued that the same mechanism operates in contact-induced change resulting in language death, through a process of grammar shift in codeswitching situations. The rather complex issue of interference and substratum in closely related Bantu languages or dialects of the same language at the southern Kenyan coast is dealt with by Nurse-Walsh, and Möhlig from different theoretical angles.

References Appel, René-Pieter Muysken 1987 Language contact and bilingualism. Giles, H. (ed.)

London: Edward Arnold.

1977 Language, ethnicity and intergroup relations. New York: Academic Press. Thomason, Sarah Grey-Terrence Kaufman 1988 Language contact, creolization, and genetic linguistics. Berkeley: University of California Press.

Theory of language death1 Hans-Jürgen Sasse

1. The state of the art The most macabre of the numerous anthropomorphic metaphors linguists provide for their subject matter is that of language death. The extinction of a language is in fact a distressing matter, since the cultural tradition connected to it and the socio-cultural or even ethnic independence of the group that speaks it very often perish together with it. Yet it is a very common phenomenon. In the last five hundred years about half the known languages of the world have disappeared; hundreds of languages are in danger of becoming extinct by the end of the century. In Africa alone, nearly 200 languages are endangered (Sommer, this volume). Of course, as we know from history, languages have always disappeared: Gothic, Etruscan, Iberian, Sumerian, Hittite, Egyptian, etc., but the world-wide colonization of many small ethnic units by a minute number of large ethnic units, the formation of big national states, the development of transport technology, the spread of supraregional communication media, etc. have led to an enormous increase in the extinction of smaller languages which can hardly be stopped. It seems strange that such a frequent and well-known phenomenon has not been studied earlier in greater depth; nevertheless it is a fact that the investigation of language death is a new and developing field which has emerged as something like an independent subdiscipline of linguistics towards the end of the seventies. This comparatively embryonic stage of the field should be kept in mind throughout the following discussion. It may be useful to start with some preliminary remarks to clarify what our subject-matter is.2 The title of this paper is misleading insofar as it suggests the presentation of at least a rudimentary theoretical or methodological framework within which the data discussed at this conference could be handled. If anybody expects something of this kind, he will be disappointed. As yet there is no theory of language death.

8

Hans-Jürgen Sasse

Extensive material on the process of language extinction, covering all kinds of relevant information is available for only a limited number of European minority languages. Such information includes the following: socioeconomic factors, their underlying historical causes, linguistic and sociolinguistic details in different phases of linguistic decline and contraction, and speech behavior of different layers of imperfect speakers immediately proceeding language extinction. The two best-documented cases are the Albanian periphery dialect of Greece, the so-called Arvanitika language, and the East Sutherland variety of Scottish Gaelic. On Arvanitika there will be a four volume documentation by this author, covering approximately 25 years of continuous research, the first volume of which just appeared (Sasse 1991). It deals with all of the possible facets of Arvanitika history, from its first attestation until the final phase of its extinction. Moreover, there is a very detailed sociolinguistic monograph on Arvanitika by Lukas Tsitsipis (1981), which is especially devoted to the obsolescent phase of this language and examines some of the crucial methodological and theoretical issues to which its investigation gives rise. Numerous further studies of article length deal with various linguistic and sociolinguistic aspects of this language. The linguistic history of Arvanitika is amply documented; the historical and sociolinguistic circumstances as well as the social and economic conditions under which the Arvanitika-speaking community lived can be followed through several centuries. The same is true of Gaelic, which has been attested in written form for many centuries and whose external history in different parts of Great Britain is well known. The final phase of one of its Scottish dialects, East Sutherland Gaelic, has been described in utmost detail in countless studies, some of them of monograph length, by the world's leading authority on language death, Nancy Dorian. No other European minority communities which speak an obsolescent language have been studied in comparable detail. There are a number of interesting smaller case studies among which are those on Breton by Wolfgang Dressier and his students and on Hungarian in Austria by Susan Gal, which provide valuable material on certain aspects, but do not cover the whole array of phenomena whose investigation is necessary in order to obtain a clear picture of the entire process. Longitudinal studies (covering the development within a single self-contained speech community, e.g. a village, over a sufficient stretch of time) have rarely been carried out, a notable exception being Breu (1991) for Italo-Albanian. In addition to the work on European cases of language death, there are a number of scattered studies from other parts of the world, and these may serve as a corrective to the rather uniform picture that emerges from the

Theory of language death

9

European minority studies. Two of the most outstanding scholars who have contributed to our knowledge of "exotic" obsolescence situations are Hill on Nahuatl (Mexico), and Schmidt on Dyirbal (Australia). But there are huge geographical areas for which no comprehensive studies are available, one of these being the region under discussion in this volume. Furthermore, all this work taken together does not suffice as an empirical basis for a theory of language death. Theoretical or model-establishing approaches have therefore been scarce, and only few attempts have so far been made to give a broad overview of current research in this field (cf. Dressler - Wodak-Leodolter 1977, Rindler-Schjerve 1989, and particularly Dorian 1989). Nevertheless, the studies available thus far offer a sufficient amount of research to serve as a point of departure for asking - and hopefully answering - a number of fundamental methodological questions: 1. What are the relevant problems and research goals in this field? 2. How can we pick out, clarify and separate from each other the principal levels of research and their interaction? 3. Is it possible, given the extremely small number of comprehensive case studies, to base some generalizations on the cases known so far? 4. Are these generalizations sufficiently general in order to serve as a rudimentary model of language death, within which it is possible to unify terminology and establish criteria for further research?

2. Levels of research A brief look at the literature on language death reveals that the problem is often tackled from quite different angles: some authors deal with the socioeconomic factors which have given rise to a certain situation but do not describe the actual linguistic events; others investigate structural phenomena of dying languages without paying attention to the sociolinguistic status of the speech form under consideration. In order to understand the entire process, however, a holistic approach is necessary which takes the interplay and the possible causal connections of the phenomena investigated into account. I will therefore begin by separating the different levels of research and their aims and goals and then try to show the interaction of the empirical facts they are concerned with. Three types of phenomena relevant to the study of language death must be clearly distinguished. 3 First of all, there is the entire range of extra-

10

Hans-Jürgen Sasse

linguistic factors: cultural, sociological, ethnohistorical, economic, etc., processes, which create, in a certain speech community, a situation of pressure which forces the community to give up its language. I will call this the External Setting (ES). The linguist studying obsolescence is not so much concerned with ES phenomena per se, the detailed investigation of which is the task of historians, sociologists, and specialists in other neighboring disciplines. Nevertheless, ES phenomena must be carefully taken into account because they constitute the trigger for the entire process. I will come back to this point directly. Let me first introduce the second set of phenomena, which I will class under the general term Speech Behavior (SB). By this I mean the regular use of variables, which, in a given speech community, are bound with social parameters, e.g. the use of different languages in multilingual settings, the use of different styles of one language (Fishman's famous Who speaks what language to whom and when), domains of languages and styles, attitudes towards variants of languages, and so on. For the investigation of these phenomena an integrated sociolinguistic macro-approach is necessary which would have to combine an extended and modernized version of empiricist sociolinguistics of the Labovian type with interpretive models such as those of Gumperz (ethnomethodology) or Giles (social psychology). 4 Since the political and social conditions are primary, ES phenomena have a strong impact on SB. Attitudes towards languages and styles develop on the basis of political, social and economic pressure, and this pressure in turn develops on the basis of the historical situation in which a speech community finds itself. It is therefore possible, even very likely, that differences in ES induce differences in SB. This is the most delicate point for a model based on European findings because of the relatively uniform conditions of the wellstudied minority situations. The few better-known "exotic" cases (Native American, Dyirbal), do not differ very much in this respect. African material will serve as an important corrective here. 5 The third set of data which is being studied in the investigation of language death is the purely structural, substantial-linguistic set of phenomena, e.g. changes in the phonology, morphology, syntax and lexicon of the language threatened by extinction. I will simply call these Structural Consequences (SC) phenomena. It is necessary to emphasize once more that the study of language death involves all three areas of research, i.e. a combined historical, sociolinguistic and structural-linguistic approach. There have always been attempts to reduce the investigation of language contact and language death to the structural domain and to aim at system-linguistic explanations for this kind of phenomena. This approach is unrealistic and counterintuitive. The idea that a language

Theory of language death

11

can "kill itself' by becoming so impoverished that its function as an adequate means of communication is called in question so that it must be abandoned for structural reasons is not compatible with the empirical facts. Structural impoverishment and so-called "bastardization" may help accelerate the process of language death in the final stage (we will come to this point below), but it will always be the consequence rather than the reason for linguistic obsolescence. Reasons are found exclusively in the ES area. As we shall see below, endangered languages remain functionally intact and are therefore not structurally identifiable ("deviant"), until they reach the terminal stage of extinction. On the other hand, studies confined to ES and SB phenomena (unfortunately the majority of language death research is of this kind, especially for Africa) are defective, given that the main interest of research in language death is merely in the interaction of external and internal phenomena. The restriction to non-structural phenomena neglects just that kind of information in which both linguists and historians 6 have the most vital interest. This all leads to the conclusion that an explanatory level of research can be reached only when the whole array of sublevels is equally well served. That is, a complete study on language death will have to encompass a historical analysis of the External Setting, a sociolinguistic analysis of the community's Speech Behavior, and Structural Descriptions of different speaker categories in different stages, preferably obtained in longitudinal studies covering a sufficient period of development. (In case longitudinal studies are impossible because of the rapid progress of decay, they can partly be compensated by diatopic and/or diastratic comparison.)

3. The Gaelic-Arvanitika model of language death As outlined in section 1, it is possible to conceive a model of language death which takes into account the interrelation of all the relevant phenomena, but this model will be based empirically on a very small number of comparatively uniform case studies. Since the bulk of evidence comes from East Sutherland Gaelic and Arvanitika, I will call it the Gaelic-Arvanitika-Model (GAM). 7 GAM is compatible with several smaller studies on European language death situations as well as with the results of other research (Uto-Aztecan, Dyirbal, etc.), and has been slightly generalized in order to include these results. It shows the interactions and causal relations of ES, SB and SC in a very neat way, but it cannot be overemphasized that the situations on which it is based are very similar both in their socio-economic and in their structural

12

Hans-Jürgen Sasse

linguistic phenomenology. Indeed, the question of whether differences in ES phenomena result in differences in SB, which in turn result in different structural changes, remains open. At the level of generality I am mostly dealing with, the lack of a sufficient number of detailed back-up studies is not crucial, but it "remains a weakness in the baseline from which any of us theorize" (Dorian, p.c.). The challenge and, as I believe, the usefulness of the approach presented here lies in the possibility to test the applicability of GAM to cases which look different at first sight, and to use it as a heuristic guideline for the comparison of case studies. Of course "every case is special" (Dorian 1989: 7, quoting an experienced field researcher). Even the case studies which form the empirical input of GAM present numerous specific idiosyncrasies which GAM, as proposed here, cannot predict or otherwise explain. Nevertheless, I believe that at least the general design of the model can be applied everywhere. This particularly holds for the postulate that any case of language death involves external causes, changes in the sociolinguistic patterning of variants, and language change (at every grammatical level). The results may be different from GAM, but the general characteristics remain the same. Moreover, it seems reasonable to assume that different ES conditions lead to different SB/SC results. Africa would be an ideal testing ground for this hypothesis. As we know, there are cases where ES factors quite different from those underlying GAM have been described (e.g. the Yaaku and Elmolo cases, cf. Heine 1982, Brenzinger this volume, and Dimmendaal 1989). A careful examination of the sociolinguistic and grammatical data obtained in these case studies is necessary to understand the wider connections. Let this suffice as a general introduction; I shall now try to construct GAM. At the present state of research it seems clear that the interrelation of the three sets of phenomena discussed in the preceding section is that of an implicational chain: ES phenomena induce a certain kind of SB, which in turn results in certain SCs in the dying language. It would seem, moreover, that this implicational chain is paralleled by the historical development: the extralinguistic factors appear first; a change in speech behavior then obtains due to or as a reaction to the extralinguistic factors. Finally structural changes emerge as a consequence of the change in speech behavior. While the first appearance of the factors in each case is probably phase-displaced, they continue to be operative throughout the entire process. The simple diagram in Figure 1 illustrates this. Every case of language death is embedded in a bilingual situation which involves two languages, one which is dying out and one which continues. Following a terminology frequently used in recent studies, the language given

Theory of language death

13

t —• ES I SB Ì

SC I Figure 1

up is called the Abandoned Language (henceforth A), and the language which the former speech community of A continues to use is called the Target Language (henceforth T). In all well-known cases on which GAM is based, the story begins when a substantial portion of a bilingual speech community shows a simultaneous or nearly simultaneous shift in their primary (P) language from the A language to the Τ language and a consequent shift in their secondary (S) language from the Τ language to the A language. 8 I propose to call this situation primary language shift. How is primary language shift initiated? It is triggered by the decision of a speech community to cease to transmit their language to their descendants. The result is an interruption in language transmission (LT). By LT we mean the purposive, directed passing-on of a language from one generation to the next. The pragmatic correlate of language transmission is the language transmission strategy (LTS), which, as recent cross-cultural studies of language acquisition have revealed, seems to be partly intuitive and partly communityspecific (traditional), cf. Ingram (1989: 127). In any case, it always seems to involve a specific way in which mothers (or other language-transmitting adults) talk to their children - this is called "motherese" by child language specialists (cf. Kaye 1980) - as well as repetitions, exercise games, corrections and other types of metacommunication, especially discussions about word meaning, and - last but not least - a strong tendency to assist and encourage children in their own efforts to improve their linguistic skill. LTSs play an extremely important role in language acquisition, whose impact on language death will become apparent later on in the the discussion. It is now appropriate to say a few words about the reasons for interrupting language transmission. Although the motives for such a decision may vary from case to case, especially in the historical details (restrictive language policy in one case, economical reasons in the other), studies of language death situations available thus far indicate that there is always one common element,

14

Hans-Jürgen Sasse

viz. the presence of socio-economic and/or socio-psychological pressure phenomena which move the members of an economically weaker or minority speech community to give up its language. This happens - not always but very often - via the development of a negative language attitude which results in collective doubts about the usefulness of language loyalty. The attitude towards A is often not entirely negative; it may be schizophrenic in that the retention of the language is valued positively for one reason, and negatively for another. For example, according to my experience with Arvanitika and Aromunian in Greece there are cases in which people still retain a positive attitude towards their language as far as its role as a signal of group identity is concerned, but a negative one otherwise: it is claimed that the language must be given up "because it is ugly and useless". It can be assumed that this constellation chiefly characterizes situations in which there is an aggressive language policy on the part of the dominant language community (i.e. the monolingual Τ community), which consciously and deliberately sets criteria for the negative valuation of the recessive or minority language. It may be of help to interrupt here for a moment to draw a more precise picture of the implicational relations of External Setting, Speech Behavior and Structural Consequences up to this point. We have a multilingual community, which - for whatever historical reasons - displays an uneven distribution of languages. For further political and/or economic reasons the unevenness becomes the source of strong social pressure, which may create a negative attitude towards the language of the recessive group and leads to the decision to abandon this language. Uneven distribution of languages in a bi- or multilingual speech community always results in complementary distribution of domains, which consequently leads to lexical loss or failure of lexical development in domains where the dominant language is favored. Due to the restrictions of domains, collective bilingualism increases, because the speakers are forced to learn the dominant language in order to use it in domains where the recessive language cannot be used (for a number of examples cf. Hill 1973, Dressier - Wodak-Leodolter 1977, Tsitsipis 1981). This may increase interference and simplification (e.g. loss of complex morphophonemic systems, etc.), but the A language nevertheless still remains a functionally intact language. However, once the decision to abandon A falls and language transmission comes to be interrupted, the situation changes radically. The formerly primary language A becomes secondary and begins to show serious symptoms of imperfection. Due to the lack of LTSs the only source of A data for the infant is what he occasionally hears in his environment. However, simple exposure to a language is obviously not enough in order to develop normal language proficiency (Tsitsipis 1981: 342 ff., quoting

Theory of language death

15

personal communication from Dorian). Further, in a situation in which nearly all domains have been conquered by the Τ language, A is simply no longer used in a number of important speech styles (e.g. narrative, formal, etc.). From a sociolinguistic perspective, this restricted use of A has a feedback effect on the speaker's sociopsychological evaluation and contributes to the development of a negative attitude towards A (cf. Dressler 1982: 324 f.). It is at this point that we enter a new phase - and perhaps the crucial one - of the process of language death. This period is characterized by a phenomenon called language decay. Language decay is defined as the serious linguistic disintegration which is typical for the speech of so-called semi-speakers, i.e. that speaker generation which results from the interruption of language transmission.9 When a considerable number of infants in a bilingual speech community is regularly subjected to LTSs only in one of the two languages (and perhaps even discouraged to acquire the other), the trivial consequence is an imperfect acquisition of the language whose transmission is suppressed. Semi-speakers are therefore characterized by an imperfect knowledge of A. Their morphology is extremely defective, they lose important grammatical categories such as tense, aspect or mood, even if these categories are present in T. Their speech often shows a pidgin-like simplification of syntax and a strong insecurity in the mapping of forms and functions. They are hardly able to master the phonological distinctions of A and show extreme variation in their pronunciation. I will not examine details of language decay here because this will be the subject of my second paper (Sasse, "Language Decay and Contact-Induced Change: Similarities and Differences", this volume). I will only briefly touch a number of issues which may help us understand the nature of language decay, its development and the theoretical consequences which arise. One crucial problem which has been underestimated in the literature on language death is the proper differentiation between phenomena of normal language contact and those of distorted speech in the phase of language decay. The problem has been touched on a number of times in very recent literature, notably by Campbell and Muntzel (1989: 195) and Rindler-Schjerve (1989), however, without having ever been discussed in extenso. Two terms used in Creolistic studies have been brought into the debate by Trudgill (1977), "simplification" and "reduction". Simplification is loss of external complexity, while reduction is loss of essentials and results in defectivity. In terms of the nineteenth century distinction between form and substance, one could say that simplification mainly involves readjustments in substance, while reduction involves considerable loss of both form and substance. The fundamental problem which resides in the distinction between simplification

16

Hans-Jürgen Sasse

and reduction and its application to the investigation of language decay is that of its limited operationalizability. First of all, we do not exactly know what an essential part of human language is. If "essentiality" is the only criterion that distinguishes reduction from simplification, 10 the distinction is largely arbitrary. Second, it is difficult to find objective criteria for the definition of "simplicity"; what may be called simple in one perspective may seem complex in another. Third, since loss is involved in both cases, it can often not be decided whether a certain phenomenon must be attributed to a reductional process; this can only be stated for processes which result in defectivity. Nevertheless, my claim is that we can and must distinguish language decay from normal language contact phenomena (including some instances of "simplification") by identifying the reductional character of decaying varieties of obsolescent languages. All studies of speech forms of obsolescent languages in the terminal phase of language death (admittedly few but uniform in this respect) show deviations from the norm of intact versions of the A language which cannot be attributed to language contact: loss of grammatical means for expressing entire category systems (such as the tense/aspect/mood distinctions mentioned above, unsystematic decay of person marking on verbs, etc.), even if the Τ language possesses the categories in question. Language contact phenomena ("borrowing" in the broadest sense) involve the transfer of substantial material, of patterns and of category distinctions, they can always be explained as the imitation, in one language, of some linguistic trait of another. In the case of decay, however, we are not dealing with transfer in any sense, but with downright loss leading to a heavy expression deficit. This is something quite different from the phenomena observed in normal language contact situations, and must therefore carefully be distinguished from the latter. The process of morphosyntactic borrowing connected with any situation of intensive language contact may involve "negative borrowing" 11 in the sense that a category can be lost in the replica language because of its absence in the model language. However, this type of "loss" is always compensated by functionally equivalent means of expression which imitate the morphosyntactic pattern of the model language, while in the case of language decay a true reduction takes place in that there is no compensation at all. That is, in contrast to the normal language contact phenomena, which do not affect the functionality of the system, reduction is pathological in the sense that it results in functional defectivity. What remains of the A language in the phase of decay is not a language in the sense properly understood (a structured code), but an amorphous mass of words and word forms, stereotype sentences and phrases, formulaic expressions, idioms and proverbs, which are learned in "chunks", whose forms are im-

Theory of language death

17

perfectly known and whose functions are poorly understood. When used in actual conversation, these linguistic fossils are put together in some random linear order without fixed syntactic rules. The notion of "loss" must therefore be handled with care. For example, the disappearance of the optative in early 20th-century Arvanitika (Tsitsipis 1981: 313 ff.; Sasse 1985: 70) is a normal case of negative structural borrowing due to intensive language contact: the optative was abandoned because the model language, Modern Greek, did not possess it and used the conjunctive instead, which was present in Arvanitika at all events, so the Arvanitika conjunctive simply extended its meaning to cover optative meaning of its Modern Greek model, and dropped the luxusious optative category. The loss of the aspect distinction in the Modern semi-speaker version of Arvanitika (cf. Trudgill 1977) is a clear case of reduction: since the Modern Greek aspectual system is more elaborate than the Arvanitika one, there was a tendency in last-generation full-speaker Arvanitika to extend the system according to the Greek model (Sasse 1985: 74, a case of positive structural borrowing). The underdistinction of aspect totally counteracts this recent development, and since the Modern Greek-Arvanitika bilingual masters the Greek aspect distinction perfectly, he will clearly feel an expression deficit in his Arvanitika speech 12 . The locus of language decay is the semi-speaker. Due to the lack of linguistic skill and the absence of the corrective mechanism normally connected with LTSs, he never acquires the relative proficiency of a full speaker of the language. Semi-speakers often remember an amazing amount of vocabulary, but may get totally lost with morphology and syntax. Having worked extensively with semi-speakers from different speech communities, I have observed at first hand that their situation causes more serious psychological problems than expected. In spite of their being normal full speakers of T, they suffer from the awareness of their linguistic deficiency in A, especially as long as A is still represented in their environment by a sufficient number of full speakers. This creates a kind of collective language-pathological situation which can be overcome by the acceleration of language death. Many semi-speakers avoid speaking a language in which they cannot easily express themselves and which they conceive of as a bastardized, pidginized non-language (the typical attitude of a semi-speaker: "X is not a language"). 13 In order to complete the model, let us finally fix the point where a language is definitely dead. Since this is not an empirical question but a matter of definition, one has to choose among certain alternatives. Was Manx a dead language when Hurlstone Jackson worked with its last speaker, Ned Maddrell, or did it die when Maddrell died? Was Hebrew a dead language before its revitalization in the form of Ivrith, or did it never die? Perhaps the answer

18

Hans-Jürgen Sasse

can only be given from case to case. For the present purpose my proposal is to define the final point of language death as the cessation of regular communication in the language (hence the bold type line, symbolizing an obituary notice, in Figure 2). Arguments in support of this definition will be presented in the next section. A dead language may leave residues of various kinds. It may continue as a ritual language, as a secret language, as a professional jargon, etc. It may leave a codified version, which in turn can be used for ritual or other purposes. It may finally leave a substratum influence (especially lexically) in the dialect of Τ which the former speech community of A continues to speak. A summary of the GAM is found in Figure 2.

Terms and Definitions A = Abandoned Language (Language which is dying out). Τ = Target Language (Dominant language which is continued). Primary Language = L with higher degree of lexical, grammatical and pragmatic competence. Secondary Language = L with lower degree of lexical, grammatical and pragmatic competence. Primary Language Shift = Shift from A as Primary to Τ as Primary and from Τ as Secondary to A as Secondary. Language Replacement (= Complete Shift) = Total replacement of A by Τ (possibly TA, i.e. an Α-influenced variety (dialect) of T). Language Transmission = Purposive, directed passing-on of a language from one generation to the next. Language Transmission Strategies (LTS) = The whole array of techniques, used by adults to assist their children in first language acquisition, e.g. "motherese", repetitions, exercise games, corrections, metacommunication, etc. Language Decay = Pathological language disintegration. Semi-Speaker = Member of the post-Language-Transmission break generation with imperfect knowledge of A. Terminal Speaker (Sometimes confused with imperfect speaker) = Last generation speaker. Simplification = Removal of linguistic complexities. Reduction - Removal of significant/essential/functionally necessary parts of the language.

Theory of language death ES

SB

Figure 2

SC

19

20

Hans-Jürgen Sasse

4. Revitalization In this short section we will be concerned with the question of revitalization in different stages of language death. In the preceding section it was proposed to view a language as definitively dead when its use for regular communication has ceased. I think this is the point up to which "natural" revitalization processes are possible. Revitalization could thus be used as justification for this definition. Before talking about revitalization, let us summarize the different phases of the language death model conceived in section 3. GAM, as outlined above, can be divided into three phases: the primary/secondary language switch, the decay phase, and the terminal phase of death of A and its total replacement by T. The three phases are summarized in Figure 3.

I.

Primary Language

II. Language

Shift

Decay

Emergence of Semi-Speakers Reduction of Style Repertoire —> Reduction of Grammatical System Pragmatic Incompetence —» Structural Incompetence III. a. Language

Death

b. Language

Extinction of Communication in A —> Extinction of Creativity in A

Figure 3

Replacement

Full monolingual Proficiency in Τ (possibly substratum phenomena; emergence of a Τ dialect on A substratum)

Theory of language death

21

A language is usually called "healthy" before it enters stage I, but it becomes "threatened" immediately after this point. Once a new language becomes dominant in a certain speech community the old one is potentially endangered unless there exists a very strong motivation to retain it. Such motives can easily develop on the basis of an altered political situation, such as the emergence of separatist movements (cf. Basque), support from an ethnically or linguistically related community from outside (cf. Italo-Albanian as supported by Albania), removal of social pressure on the part of the dominant community, migration and gain of new prestige in the new homeland, etc. This may push an endangered speech community during phase I to revitalize A by a renewed reversal of A-T/P-S relationship; i.e. the A language is made primary again by a reinforced interest in its transmission. At stage II, where a language is "in the process of dying", revitalization becomes more difficult. As long as there is a sufficient number of full speakers left, these same speakers may begin to teach their children or grandchildren the language as soon as they observe their renewed interest in it. I have sporadically come across such cases in Arvanitika communities in Greece, where certain young adherents of leftist movements recently started to learn their parents' or grandparents' language (the interest normally decreases when they realize that Arvanitika is quite deviant from Tirana Albanian). When phase II is at an advanced stage, revitalization seems possible only by "creolization" (either by mixture with some related standard language or by mixture with a non-related language, perhaps T, on the basis of semi-speaker material; in extreme cases this is possible by a generation skip (oldest generation still speaks language; middle generation oscillates between semi-speaker and zero; youngest generation acquires language anew). Whether the creation of "regular mixture" (in the sense of Ma'a - module A from language Li and module Β from language L2) is possible in the course of this process cannot be examined here. Revitalization by means of creolization is close on the fringe of natural revitalization. It is questionable whether one should take such cases - if they ever occur - to be continuations of the same language. It would seem more useful to treat them as language renewal ("language birth"), i.e. the creation of an entirely new language. This would enable us to define discontinuation of linguistic tradition in a straightforward way: any total interruption of language transmission results in language death; any revitalization after total interruption of language transmission results in the creation of a new language. From phase III on, only artificial revitalization on the basis of thesauruslike, codified material is possible. The most conspicuous example of such

22

Hans-Jürgen Sasse

an event is Ivrith, Modern Hebrew, which was created on the basis of the codified holy texts after more than 2000 years of interruption of regular language transmission.

5. Discrepancies and open questions: evidence for alternative models? Campbell and Muntzel (1989: 182ff.) distinguish between four different types of language death: 1.

SUDDEN DEATH: The case where a language abruptly disappears because all of its speakers suddenly die or are killed (e.g. Tasmanian). 2. RADICAL DEATH: Rapid language loss usually due to severe political repression, often with genocide, to the extent that speakers stop speaking the language as a form of self-defense (e.g. Lenca and Cacaopera in El Salvador). 3. GRADUAL DEATH: Language loss due to gradual shift to the dominant language in language contact situations. 4. BOTTOM-TO-TOP DEATH: The so-called "latinate pattern" where, according to Hill (1983), "the language is lost first in contexts of family intimacy and hangs on only in elevated ritual contexts" (e.g. Coptic or Southeastern Tzeltal).

GAM was designed as a model of gradual death because the instances on which it is based are characteristic cases of gradual death. Moreover, gradual death seems to be the prototypical case of which the others are merely variants. Cases of sudden death are of no linguistic interest, since no changes in SB and no SCs are observable. Cases of radical death may be conceived as a gradual variant of gradual death. "Overnight" abandonment of a language characterizes an entire community only in extreme cases of menacing pressure exerted on the whole group or prohibition; in contrast this occurs sporadically in all situations of "gradual" death with which I am familiar (i.e. some people shift more rapidly than others, depending on the strength of the pressure they feel). In other words, the normal situation is a mixture or a continuum between radical and gradual death, where both types of shifters, the rapid and the gradual, are present in a single obsolescent speech community. Rapid shift is characterized by the absence of an intermediate A/P-A/S switch and consequently by the lack of a semi-speaker generation in the sense defined above, but this produces a different type of imperfect speaker whom

Theory of language death

23

Menn (1989: 345) aptly has termed a "rusty speaker", a person whose interaction opportunities have been limited for a long time and who has to invest a great deal of energy in retrieving words and putting sentences together (cf. Sasse for more on this type of speaker, this volume). In studies of gradual death usually both types of imperfect speakers are described. 14 It is questionable whether bottom-to-top death constitutes a separate type at all. In many cases of gradual death a residue is left which may be maintained for special purposes ranging from ritual language, secret language and professional jargon to special terminology which remains as "substratum vocabulary" in the dialect of Τ that the former speech community of A still uses. For instance, in certain parts of the formerly Arvanitika-speaking community, Albanian weaver terminology, plant names and a number of pastoral items have been retained in the dialect of Modern Greek. 15 I have also observed the continuation of a ritual language in a situation of gradual death among the Kemant, an obsolescent Agaw community in Central Ethiopia. Kemant is a dying language mostly spoken by elders over 40, but younger adherents of the Kemant religion (a mixture of Christian, Jewish, and traditional Cushitic elements) still know a number of prayers composed in an archaic variety of Kemant, even if they are semi-speakers in the modern variety. Thus "bottom-to-top" phenomena belong to the general field of residual maintenance and do not contradict the conditions of gradual death: the process of obsolescence in normal communication may well run according to the usual pattern independent of what material is continued for a specialized purpose. However, cases have been reported where after the extinction of normal communication the resulting special language forms a hybrid whose morphosyntactic matrix comes from the Τ language, but whose vocabulary (at least in part) is a residue of A. These cases deserve special attention because they raise the question of how and at which point during the extinction process such mixtures can arise. A case in point is Krekonika, a secret language used by masons of the Peloponnese in an area where formerly Arvanitika was spoken. The morphosyntactic basis of this language is Modern Greek, but most of the secret vocabulary is Arvanitika (Konstantinopulos 1983). The Africanist is reminded of Ma'a, the most widely discussed "mixed language", which is composed of Cushitic vocabulary and a Bantu morphology. A few more of such languages have been reported in the literature, but no plausible theory of their origin is as yet in sight, especially due to the fact that we know very little about the actual history of the hybrid languages involved. 16 The above considerations lead us to the conclusion that the empirical findings of gradual death situations can in fact be fairly well generalized into a comprehensive model of "prototypical" language death. What remains is

24

Hans-Jürgen Sasse

to test its applicability to a broad variety of cases, especially those where deviation from the pattern described in section 3 are reported. At first sight, most of the instances dealt with in the literature on language death seem to fit rather well, including those treated in this volume. Nevertheless, a number of fundamental questions remain open. In the cases of Suba (Rottland), Gweno (Winter), Kore (Dimmendaal), Yaaku (Brenzinger) and Elmolo (Brenzinger), the ES phenomenology seems to be quite standard: all involve weaker or stronger pressure situations that urge a minority group to develop a more or less negative or "schizophrenic" attitude towards their ethnic identity and decide 17 to integrate into a larger ethnic group. In all cases this results in spectacular changes of SB. The problem is that we do not know very much about structural consequences. It is therefore difficult to draw conclusions with respect to the entire process. There are certain discrepancies which simply cannot be resolved as long as we don't have detailed descriptions of the three separate domains which we tried to disentangle in section 2. For instance, it is said that Elmolo adults seem to have insisted on speaking the language to their children, but that the children refused to speak Elmolo and replied in Samburu (Heine 1982: 177). Does this make a reconsideration of the LTS hypothesis necessary? On the other hand, Heine's (1980) Elmolo, elicited from the last Elmolo speaker, is clearly a semi-speaker product (cf. Sasse, this volume). This should not be the case if imperfect language learning results from the suppression of LTS's. 1 8 The reverse problem is posed by Yaaku: in spite of the explicit decision to give up the language and to interrupt language transmission, a general lack of semi-speakers is observed. Given the dearth of information we cannot decide whether this is significant 19 or whether a relatively short and not very spectacular decay phase has simply escaped the field workers. This returns us to the opening considerations of this paper. We are still not sufficiently equipped for a better understanding of the interaction of the three domains of social structure including its political and economical aspects (ES), sociolinguistic patterns (SB), and structural consequences (SC). Despite its general applicability, GAM is dependent on certain specific conditions which are not present everywhere. I cannot help but conclude this paper with the stereotype dictum that much more work must be done in all areas of research. Africanists will have to exploit the opportunity of carrying out longitudinal studies in endangered but still "healthy" speech communities such as the Dahalo (Tosco, this volume). Furthermore, they will have to be careful not to draw premature conclusions on the basis of limited material, as the entire field of linguistic obsolescence is still too poorly understood to allow for broad generalizations.

Theory of language death

25

Notes 1. I am indebted to Nancy Dorian, Wolfgang Dressler, Mechthild Reh, Rosita Rindler-Schjerve and the participants of this conference for valuable comments on an earlier draft of this paper. 2. It goes without saying that we will not be concerned here with cases of language extinction due to the sudden extinction of speech communities. Where a language disappears because all of its speakers die or are killed is of no linguistic interest. Nor do we mean dead languages as opposed to living languages in the same sense used in schools (Latin and Greek vs. French and Spanish). In our view Greek is not a dead language because it survives in the form of Modern Greek. 3. A clear distinction is not always made even in most recent and otherwise highly interesting studies. A good example is Lanoue (1991), cf. esp. p. 91. 4. I am indebted to Rosita Rindler-Schjerve for drawing my attention to the necessity of an integrated approach. 5. In the model devised in section 3, I have therefore kept the ES part as general as possible. In order to investigate correlations between ES and SB, an extended model will have to establish a number of ES and SB parameters. For instance, it will have to specify the various conditions under which an uneven distribution of languages in a multi-lingual setting comes about, e.g. migration, conquest, intrusion, gradual gain of importance of one language over another, conscious integration, and so forth. In the SB domain, it will have to specify the groupdetermining features of the two languages, e.g. how language is tied to ethnicity, profession, etc. Finally, the relation of the two languages is of importance, e.g. whether L| is a dialect of L2 or a totally different language, whether L] is a written language and L2 is not or vice versa, etc. 6. A considerable number of structural phenomena connected with language death are being studied particularly for the general linguistic and/or historical issues they give rise to: 1. substratum identification: reconstruction of historical events from linguistic residues that obsolete languages have left in superseding languages; 2. typological and universalistic implications of language contact and decay (e.g. its relation to pidginization and creolization processes); 3. patholinguistic implications of imperfect speech in the decay phase; and many others. 7. The label "Gaelic" here is merely an abbreviation for "East Sutherland Gaelic" (ESG). This has to be stressed for two reasons. First of all, ESG is virtually the only Gaelic dialect whose fading has been investigated in sufficient detail; there isn't even one other Scottish Gaelic dialect that has been studied in a way that parallels Nancy Dorian's work on ESG. That means that her findings aren't even safely generalizable for Gaelic, never mind for other Celtic languages or for non-Celtic languages. Secondly, the label "Gaelic" is not thought to suggest that the language as a whole is currently a fit subject for language death studies, rather than the ESG variety in particular.

26

Hans-Jürgen Sasse

8. The use of the terms "primary language" and "secondary language" roughly follows Weinreich (1967: 74 ff.). There, the distinction is based on two criteria, firstly the degree of lexical and grammatical competence, secondly the relative stylistic proficiency, i.e. the degree of pragmatic competence. According to Weinreich the difference is a matter of the order of language acquisition. 9. Tsitsipis (1981) uses the term "terminal speakers". Dorian's "semispeakers", introduced in Dorian (1977), is preferable, because terminal, i.e. last generation speakers must not necessarily be imperfect speakers. Semi-speakers, however, are defined in terms of their imperfect language competence. 10. Regarding certain problems with the notion of simplification cf. also Andersen (1989). 11. For the notion of negative borrowing cf. Sasse (1985). 12. Nancy Dorian (p.c.) has reminded me of another whole category of non-contactinduced change that does "not involve loss of grammatical categories and heavy expression deficits": reduction of syntactic options (e.g. merger of two syntactic constructions for expressing the passive into a single one, or reduction of three negative imperative constructions in one, both in East Sutherland Gaelic). It is disputable, however, whether such "microcosmic style reductions" (Dorian) connected with particular syntactic structures really make up a category of their own. I am inclined to believe they don't because on closer inspection they often turn out to be either style reductions connected with register loss or indirect contact-induced changes (i. e. some sort of "negative borrowing") due to the fact that the subtle conceptual distinctions implicit in these options simply do not match with distinctions found in the contact language. More research in this area is certainly necessary, and I am grateful to Nancy Dorian for having brought this fact to my attention. 13. There are some striking parallels between semi-speaker speech and speechpathological situations, which are worth being investigated. As far as I know, except for Menn (1989) no serious attempt has been made so far to compare language decay with aphasie conditions. 14. In the Arvanitika case, for instance, rapid shift, reconstructible by the absence of a decay phase and the presence of rusty speakers, is observable mainly in families of inferior social status. This can easily be explained as a strategy of suppressing language as a diagnostic feature of social inferiority (cf. Sasse 1985: 43). 15. Such elements usually do not have a long life because Α-influenced Τ dialects are themselves subject to dialect death being absorbed by the standard variety of T. 16. Although this is outside the scope of our present discussion because it concerns "language birth" rather then language death, a few words about this type of "hybridization" are in order here. Seven cases of linguistic hybrids roughly composed after the pattern "morphology from L], vocabulary from L2" have been found so far: in addition to Ma'a there are three Romani dialects (English, Armenian, and

Theory of language death

27

Spanish), further Michif (Cree and French), Media Lengua (Quechua and Spanish), and Mednyj-Aleut (Aleut and Russian). The most recent summary discussion is found in Thomason - Kaufman (1988). Several models have been proposed to deal with this phenomenon. One is that of "gradual shift" by disruptive borrowing, i.e. a successive replacement of basic elements, especially a replacement of the morphosyntactic frame. This would imply that A borrows (nearly) all of the morphology of Τ but at first continues using the vocabulary of A. This is then gradually replaced by elements of Τ until the "normal" state of Τ is reached. The model is extremely problematic because borrowing of morphology to the extent that entire systems are replaced has never been attested in an observable case and is only assumed for conjecture here. An exception is dialect contact (i.e. contact between very closely related morphological systems), where much of the morphological material is already identical or very similar, so that interdialectal analogies can operate. (Gradual shift between Bantu languages is usually of this kind.) This is not the case in the instances under discussion, the languages involved usually being extremely dissimilar. A more promising explanation (for some of the cases at least) would be to assume relexification of Τ after the shift has already taken place: assume a speech community where A and Τ are spoken side by side. The younger generation is quasi monolingual in T, but the vocabulary of A is still "present". Under these conditions, A vocabulary can be freely inserted into the morphosyntactic frame of T, just as modern German jargons use English verb roots in handle-n 'to handle', click-en 'to c l i c k \ f l i p p - e n 'to flip', rock-en 'to rock shift-en 'to shift', drift-en 'to drift', and many more. A hybrid, initiated in this way, can be continued for whatever purpose even after the complete replacement of the original version of A, e.g. as a professional jargon as in the Krekonika case, for the creation of a ritual or secret language for which the disguising character of the forgotten A vocabulary is exploited, or even as a last minute revitalization strategy in a period of increasing group identity (as perhaps in most of the seven cases cited above). 17. A warning is necessary not to take the term 'decision' too literally. Decisions explicitly made by the entire speech community certainly do occur, but these are the exception rather than the rule. Several such cases have been reported for East Africa, the most spectacular one being the Yaaku who decided to give up their language in favor of Maasai in a public meeting sometime in the 1930s (cf. Heine 1982: 30). In most of the cases however, "decision" is to be read as "tacit collective agreement". In the extreme cases of menacing force such a collective agreement may even be collective coincidence due to the fact that all members of the community react to the repression by means of the same survival strategy. 18. There would be a way out of the dilemma if we knew more about SB patterns in the Elmolo community of the 1920s and 1930s when the shift occurred. Experience with Arvanitika has shown that even in cases where the decay phase is initiated by suppression of LTSs for the overwhelming part of the community, there may exist a small group of conservative language users who insist on speak-

28

Hans-Jürgen Sasse

ing A, but whose children deliberately refuse to learn it once they internalize the "modern" attitude against it. 19. Perhaps because the socio-economic gap between those elders who still speak the hunter-and-gatherer language and the "modernized" pastoralists who gave it up is so striking that the latter refuse to have anything to do with it?

References Andersen, Roger W. 1989

"The 'up' and 'down' staircase in secondary language development", in: Nancy C. Dorian (ed.), 385-394. Β reu, Walter 1991

"Zur aktuellen Situation in den nördlichen italoalbanischen Kolonien", in: Walter Breu - Rolf Ködderitzsch - Hans-Jürgen Sasse (eds.), Aspekte der Albanologie. (Akten des Kongresses "Stand und Aufgaben der Albanologie heute", Köln 1988.) Wiesbaden: Harrassowitz, 1-16. Campbell, Lyle - Martha C. Muntzel 1989

"The structural consequences of language death", in: Nancy C. Dorian (ed.), 181-196. Dimmendaal, Gerrit J. 1989 "On language death in eastern Africa", in: Nancy C. Dorian (ed.), 13-31. Dorian, Nancy C. 1977

"The problem of the semi-speaker in language death", in: Wolfgang U.

Dressier - Ruth Wodak-Leodolter, 23-32. Language death. The life cycle of a Scottish Gaelic dialect. Philadelphia: University of Pennsylvania Press. Dorian, Nancy C. (ed.) 1989 Investigating obsolescence. Cambridge: Cambridge University Press. Dressier, Wolfgang U. 1981

1972

"On the phonology of language death". Papers from the Eighth Regional Meeting of the Chicago Linguistic Society. Chicago: Chicago Linguistic Society, 448-457. 1981 "Language shift and language death - a protean challenge for the linguist". Folia Linguistica 15: 5-27. 1982 "Acceleration, retardation, and reversal in language decay?". In: Robert L. Cooper (ed.), Language spread. Studies in diffusion and social change. Bloomington: Indiana University Press, 321-336. Dressier, Wolfgang - Ruth Wodak-Leodolter (eds.) 1977 Language death (= International Journal of the Sociology of Language 12). The Hague: Mouton.

Theory of language death

29

Fishman, Joshua A. 1965 "Who speaks what language to whom and when?" La Linguistique 2: 67-88. Gal, Susan 1976 Language change and its social determinants in a bilingual community. [Unpublished Ph.D. dissertation, University of California, Berkeley.] 1979 Language shift: Social determinants of linguistic change in bilingual Austria. New York: Academic Press. Heine, Bernd 1980 The Non-Bantu languages of Kenya. (Language and Dialect Atlas of Kenya, Vol. II.) Berlin: Dietrich Reimer. 1982 "Traditional fishing in the Rift Valley of Kenya: A linguistic survey". Sprache und Geschichte in Afrika 4: 7—40. Hill, Jane H. 1973 "Subordinate clause density and language function", in: Corum, C. - T.C. Smith-Stark - A. Weiser (eds.), You take the high node and I'll take the low node: Papers from the Comparative Syntax Festival. Chicago: Chicago Linguistic Society, 33-52. 1983 "Language death in Uto-Aztekan", International Journal of American Linguistics 49: 258-76. Ingram, David 1989 First language acquisition. Method, description and explanation. Cambridge: Cambridge University Press. Kaye, K. 1980 "Why don't we talk 'baby talk' to babies", Journal of Child Language 7: 489-507. Konstantinopulos, Christos G. 1983 Oi paradosiakoi chtistes tes Peloponnesou. Athens: Melissa. Lanoue, Guy 1991 "Language loss, language gain: Cultural camouflage and social change among the Sekani of Northern British Columbia". Language in Society 20: 87-115. Menn, Lise 1989 "Some people who don't talk right: Universal and particular in child language, aphasia, and language obsolescence", in: Nancy C. Dorian (ed.), 335-345. Rindler-Schjerve, Rosita 1989 "Sprachverschiebung und Sprachtod: Funktionelle und strukturelle Aspekte", in: Heinrich Beck (ed.), Germanistische Rest- und Trümmersprachen. Berlin-New York: Walter de Gruyter, 1-14. Sasse, Hans-Jürgen 1985 "Sprachkontakt und Sprachwandel: Die Gräzisierung der albanischen Mundarten Griechenlands", Papiere zur Linguistik 32: 37-95.

30

Hans-Jürgen Sasse

1991 Arvanitika. Die albanischen Sprachreste in Griechenland. Teil 1. Wiesbaden: Harrassowitz. Schmidt, Annette 1985 Young people's Dyirbal: An example of language death from Australia. Cambridge: Cambridge University Press. Thomason, Sarah Grey - Terrence Kaufman 1988 Language contact, creolization, and genetic linguistics. Berkeley: University of California Press. Trudgill, Peter 1977 "Creolization in reverse: Reduction and simplification in the Albanian dialects of Greece". Transactions of the Philological Society 1976-7, 3 2 50. Tsitsipis, Lukas D. 1981 Language Change and Language Death in Albanian Speech Communities in Greece: A Sociolinguistic Study. [Unpublished Ph.D.Diss., University of Wisconsin, Madison.] Weinreich, Uriel 1967 Languages in Contact. The Hague-Paris: Mouton.

Codeswitching as a mechanism of deep borrowing, language shift, and language death Carol

Myers-Scotton

Introduction The problem which occasions this paper 1 is the following: in some situations of heavy contact between two languages, one language takes on not only cultural lexical forms from the other language, but also core 2 lexical forms (i.e. words for which the borrowing language already has its own equivalents) and even system morphemes 3 (i.e. function words and inflections) and syntactic patterns. Or heavy contact, and the requisite sociolinguistic climate, may lead speakers to shift from their LI to an L2 as their primary means of communication, carrying with them features of the LI. In East Africa, we have an excellent example of the result of such contact: Ma'a (Mbugu), spoken in northeastern Tanzania. Thomason (1983) offers a thorough discussion of Ma'a as a Cushitic language which has borrowed from a neighboring Bantu language or languages numerous features of its morphosyntax (also see Thomason - Kaufman 1988). Brenzinger (1987), however, sees Ma'a as a result of language shift from a former Cushitic language; that is, he characterizes Ma'a as a Bantu language with a substrate of some Cushitic features, notably much lexical material. Whichever scenario is accepted, there is no question that the resulting language is a composite, leading earlier writers to cite Ma'a as a prime example of a linguistic hybrid which is not a pidgin or creole (Whiteley 1960, Goodman 1971). But other, if less dramatic, examples exist elsewhere in the world. Some are discussed in Thomason and Kaufman (1988), e.g. Greek as spoken in scattered communities in Asian Minor (Dawkins 1916, although the extent of Turkish grammatical material is less dramatic than the Bantu material in the Ma'a case); Romani spoken by English Gypsies, consisting of a Romani lexicon with English grammar; and Aleut as spoken on Mednyj in the Aleutian Islands, which has had its extensive finite verb morphology replaced by

32

Carol Myers-Scotton

that of Russian (Menovscikov 1969). Lesser known cases seem to be the Javanese heavily laden with Indonesian lexicon spoken by the Peranakan Chinese of East Java (Wolff 1983) and Media Lengua, an Amerindian contact language spoken in Ecuador, consisting largely of a Quechua base and a Spanish lexicon (Muysken 1981). Further, back in East Africa, there are a number of languages which are still maintained while sustaining inroads from another language, e.g. Dahalo, a Cushitic language, spoken along the northern Kenya coast, but with serious incursions from Swahili (Tosco this volume); and Okiek in Kenya's Narok District, which shows, at the least, examples of deep lexical borrowing from Maa (Maasai) (Kratz 1986). Thomason (1983) and Thomason and Kaufman (1988) deserve our appreciation for drawing linguists' attention to the facts that cases of pervasive grammatical borrowing exist and that language shifts are often not unitary. Their work is so important because many linguists, under the influence of Sapir ( 1921 ) 4 and others, have assumed that the core systems of language (especially the morphology) are relatively impervious to interference. For example, many linguists have dismissed contact studies as of no more than descriptive interest, assuming that most relevant features are cultural lexical forms and that therefore such studies have nothing to offer to the quest for explanatory theories about the internal nature of language. But, given certain types of language contact situations, languages can borrow a good deal more than lexical loans for new objects and concepts. Also, a consideration of the process in contact communities showing language shift and language death may suggest that such changes are less en masse than current models concentrating on end products imply. Codeswitching, as a synchronic phenomenon, is also bare-faced evidence that languages are far from single or sealed units, but does suggest some possible modularity or sub-systems. That is, all parts of a language may not march to the same drummer. This likelihood, as well as differences in psycho-sociolinguistic profiles between groups in a community, may result in stages in structural change and discontinuities in community repertoires. This is counter to a general assumption that most change is gradual and is most likely to result in losses for the sociolinguistically minor language. Rule gains and compromises, instead of losses, are distinct possibilities. Denison (1977: 21), for example, raises the issue of whether rule substitution can "if it goes far enough, ultimately amount to 'language death' ". But it seems unlikely that changes in their structure make languages disappear. They disappear because of changes in language use patterns in the community; their speakers, not the languages, disappear. Mohan and Zador (1986: 317), in invoking the notion of punctuated equilibrium from evolu-

Codeswitching as a mechanism of deep borrowing

33

tionary biology to characterize language death, suggest that languages "do not die because of the nature of their linguistic structure" but because of a blow from the outside, a shift in language loyalty in parts of the community. They write (1986: 313) of what we may call a "dying" language as, in fact, already "dead at its source, but with a now finite community of native speakers continuing, like the earlier light of a dead star, to travel its original course and give the illusory picture of vitality". Surely considerations of what can be borrowed and the processes languages undergo in language shift/death are relevant to general theoretical issues as universals, markedness and the organization of language in the brain, as well as discussions of such specific matters as positive or negative evidence of constituent structure (i.e. what must go with what). The issue, however, remains, what is the mechanism for the borrowing of non-cultural lexical forms and parts of the core systems of phonology, morphology, and syntax, and for the structural composite found in some results of language shift? Many, of course, have offered "heavy cultural contact" as behind such results; but a moment's reflection reveals that "contact" (and accompanying bilingualism) may be necessary backdrops, but they are not means or mechanisms. This paper argues that one important mechanism for "deep borrowing" and shift phenomena is codeswitching. It is also suggested that, for the same reasons, codeswitching is involved in language death. Rather than looking upon language death as necessarily a gradual loss of a language's grammatical apparatus, I will suggest that at least some instances of language death may involve the pervasive addition or substitution of the grammar of another language in the codeswitching situation.

Types of borrowing compared with codeswitching Let us begin by defining some abbreviations which will be used in this discussion. Borrowing (hereafter B) is the incorporation into one language of material from another language. Codeswitching (CS) is the selection by bilinguals (multilinguals) of material from an embedded language (EL) in utterances from a matrix language (ML) in the same conversation. A major difference between Β and CS is that under B, EL forms become part of what constitutes the lexical competence of an ML speaker, while EL forms in CS undergo no such incorporation, but are accessed from the EL only to serve the socio-pragmatic needs of the current exchange.

34

Carol Myers-Scotton

The ML has the more structurally influential role as the language of the "frame" of CS utterances, as will become clearer below. CS can involve units at all levels from a single morpheme up to several or more sentences. Note that this is a different position from that taken by some researchers on CS who would restrict "true" CS to stretches minimally as long as phrases and clauses. As will become apparent, what makes CS an important mechanism in the process of introducing material from one language in another is that CS does apply at the morpheme level, possibly resulting in constituents consisting of system morphemes from one language and content morphemes from another language. Lexical Β involves two types of forms: cultural loans, which stand for objects or concepts new to the Β language's culture; and core loans, which stand for objects or concepts already encoded by the Β language. Of course, (and in line with the non-unitary approach of this paper), it is an oversimplification to speak of many Β forms as loans into the language as a whole; some Β forms only become part of the competence of certain groups of speakers, not all; and some Β forms never achieve general currency, but are limited to certain styles. Also, the degree of phonological (and even morphological) integration of a Β form may differ from speaker to speaker, depending on the speaker's sociolinguistic profile (including degree of bilingualism). Deep borrowing (DB) will be used specifically for Β involving not only core lexical forms, but also system morphemes, and syntax. DB also includes phonological features, but they will not be considered here. The process of CS facilitates DB between the ML and the EL, it will be argued. If CS took place in a sociolinguistic vacuum, the borrowing could be in either direction; but in a living community the borrowing is asymmetrical: the flow is mainly from the sociolinguistically dominant language to the other one. Cultural and core lexical Β forms enter the borrowing language (the ML) by different processes. There need be only a slender bond between CS and the process of borrowing cultural Β forms; because such forms fill a lexical vacuum, they can enter the language without an additional impetus. Although core Β forms also may enter a language without widespread CS among its speakers, CS is an obvious mechanism to introduce such loan words. This is because CS displays potential Β forms in juxtaposition with material from the Β language. The weight of this juxtaposition, over time and many interactions, results in B. That is, a core Β form typically starts out as a CS form, meaning there is a continuum between these Β forms and CS forms. While some cultural Β forms also enter the language through CS, these Β forms often

Codeswitching as a mechanism of deep borrowing

35

enter the ML abruptly, in the same manner as the object/concept for which they stand. CS is differentiated from Β in two major ways. First, ML speakers may be monolingual and still use Β forms, but those who use CS forms must show some degree of bilingualism. Second, Β forms have acquired status as part of the grammar of the ML and therefore their relative frequency for encoding the concepts they stand for in a large data corpus is more similar to that for native forms than it is to CS forms. Also, this frequency is predicted to show up across a number of speakers. However, because they have no status in the ML, CS forms have a frequency prediction quite different from that of native ML forms; they may occur only once or twice and possibly only in the speech of a single individual. Saying that CS is involved in the incorporation of EL core lexemes and morphosyntactic material into an ML is not new, of course. Where this paper claims to make its contribution is in going beyond this general observation to demonstrating the actual mechanisms which seem involved in this transmission of material from one language to another. In order to make these claims with possibly diachronic consequences, I must first sketch the model of synchronic CS outlined in Myers-Scotton (1992).

A frame-based process model of CS This model follows the premise that CS proceeds in two processing steps (whether simultaneous or sequential) and that the more important is the step which builds the frame for CS utterances. The model applies to intrasentential CS and views such utterances as consisting of three types of constituents: (1) Those constituents of most interest in this paper are constituents composed of morphemes from both the Matrix Language (ML) and the Embedded Language (EL); these are called ML+EL constituents. (2) The remaining two types of constituents are ML or EL islands; these are composed entirely of morphemes from one language only (ML or EL) and must be well formed according to the grammar of that language. The first step in building the frame involves applying two ML hypotheses (discussed below) to ML+EL constituents and specifying which constituents may be islands. Filling in content morphemes is the second step. Prior to building the frame, the two (or more) languages involved in CS must be differentiated; that is, the ML must be identified. The ML is identified on psycholinguistic and sociolinguistic grounds independent of its role in

36

Carol Myers-Scotton

CS. The ML tends to be the speaker's more dominant language in terms of proficiency and it is the language for which community norms make it the more unmarked choice for the interaction in question. 5 Two hypotheses regarding the M L are offered to explain frame building in M L + E L constituents in CS. The Morpheme Order Hypothesis predicts that the morpheme order of the M L prevails in such constituents. The System Morpheme Hypothesis predicts that all productive system morphemes in such constituents must come from the ML. 6

I. Support for the matrix language hypotheses Consider the following examples from Swahili/English CS in Nairobi, Kenya. They come from naturally-occurring conversations among second language speakers of Swahili and English who do not share the same first language; Swahili is their more dominant language in terms of proficiency and it is also the more unmarked choice for casual, inter-ethnic conversations. Swahili is the M L in these CS utterances. Examples (1) through (3) show Swahili morpheme order: (1)

[Mungu anaweza y ote muamini ataweza kukubadilisha mambo mengi NEW-mapya katika maisha yako. mambo things

mengi many

NEW /nu/ new

-mapya cl.6-new

katika in

na utakuwa na] maisha life

yako your

'[God is able to do all (if) you believe he will change you and you will have] many new things-new in your life.' (Myers-Scotton 1989: 4) (2)

ImeTURN BLACK sana. i-me-turn /tasn/ BLACK /blaek/ it-PERF-turn black 'It had turned very black.'

sana very (Myers-Scotton 1989: 2)

(3)

H-ao

wa-na-taka

TIMING

PROPER

timing proper 'Those (people) want proper timing.' (Myers-Scotton 1989: 2)

DEM-cl.2

cl.2.-NON-PAST-want

Examples (4) and (5) show that all productive system morphemes in the M L + E L constituents come from the ML, Swahili:

Codeswitching as a mechanism of deep borrowing

(4)

37

[... Ukikaa huko Baringo,] u-na-CHANGE, [mazee. Unafikiri kama watu wa huko jo! Ukija huku watu wanashangaa. Unaanza] ku-BEHAVE [kama watu wa huko] wa-na-vyo-BEHAVE. u-

-na-

CHANGE (/ceinj7

2nd sing

PRES

change

ku-BEHAVE / b i h e i v /

wa-

-na-

-vyo-

INFIN-behave

3rd pi

PRES

MANNER

BEHAVE behave

'[If you live at Baringo,] you change, [my friend. You think as people from there! If you come here (Nairobi) people will be amazed.] You [will begin to] behave [as people from there] behave.' (Myers-Scotton 1989: 5)

(5)

[Haukuona] a3rd sing

a-ki-ni-BUY-i-a

-kiPROG

-ni1st sing obj

[beer siku

hiyo]?

BUY /bai/- -ibuy BENEFACTIVE

-a INDICATIVE

'[Didn't you know] he was buying [beer] for me [that day]?' (Myers-Scotton unpublished data)

II. M L a n d E L islands As noted above, in addition to ML+EL constituents, intrasentential CS also includes both M L and EL islands, constituents entirely in one language. Which constituents may be islands seems to vary from language pair to pair. But the principle that such islands exist in CS utterances is universal, and their specification is part of the frame activated whenever speakers engage in CS. Examples of islands follow. (6) illustrates that a predicate may be an EL island in English in Swahili/English CS; (7) illustrates what seems to be a universally possible island, a set phrase {PEOPLE ON THE MOVE):

(6)

I-na cl.9-with

shida problems

ny-ingi cl.lO-many

yaani that is

GOOD EXAMPLES

i-na

m-ambo

good examples

cl.9-with

cl.6-matter

ku-SET iNFlN-set m-engi

cl.6-many

'There are a lot of problems, that is, to set good examples has (requires) many things.'

38 (7)

Carol Myers-Scotton

We we you (emphatic)

u-na-ju-a 2ndS-NON-PAST-know-INDIC

hi-i

FASHION

ni

C1.9-DEM

fashion

is cl.9-of C1.2-DEM cl.2-person Sasa, wa-tu kama wewe

PEOPLE ON THE MOVE.

people on the move. amba-o wa-me-baki

y-a

bwana mister

now

wa-le

cl.2-person nyuma

wa-tu

as

REL-cl.2 cl.2-PERF-remain behind ha- mu- wez- i ku-APPRECIATE

hi-yo

NEG-2ndPL-able-NEG

DEM-C1.9

INFLN-appreciate

you

'You know, mister, this fashion is of those people, people on the move. Now, like-people like you who remain behind, you cannot appreciate this.' While islands are important components in the process model, they do not figure crucially in this paper. Rather, it is the Morpheme Order and the System Morpheme Hypotheses which will be suggested as the explanatory mechanisms for contact-induced change.

Types of language interference and where they come from Recognition of the different results for the language in a borrowing vs. a shift situation is one of the most important insights in the Thomason and Kaufman (1988) model for interference phenomena. Under B, the ML is maintained, but sustains incursions from the EL; under language shift, the ML is dropped and its speakers shift to the EL (their new ML). Basically, Thomason and Kaufman claim that under Β (we will call it the "maintenance condition"), contact phenomena first (and primarly?) consist of lexical loans; however, under the "shift condition", ML speakers shift to the lexicon of the EL, but may bring with them elements from ML phonology and morphosyntax. They do allow, however, that given certain types of cultural pressure, even under the "maintenance condition", an ML may borrow from the phonology and morphosyntax of the EL. For this reason, it seems, they argue that Ma'a, with its Bantu morphosyntax, can still be considered the result of a "maintenance condition". This will be discussed further below. Thomason and Kaufman cite case after case to illustrate their model. They and others also offer ample discussion, albeit often only in the most general sense, of the types of socio-psychological factors which obtain when either

Codeswitching as a mechanism of deep borrowing

39

type of linguistic interference occurs. To put it briefly, whenever the "outside" language (the EL) plays dominant role(s) in the community - notably in terms of socio-economic and political factors or in terms of numbers of speakers the ML only maintains itself by sustaining numerous borrowings from the EL, or a shift condition (to the EL) develops. What has not been well discussed, however, is how these contact-induced changes are accomplished from the structural point of view. While surely no researchers believe Β forms simply materialize in a language, or that in the shift condition speakers go to bed speaking their LI and wake up speaking an L2 (as a "new" ML, as well), researchers have been silent about the linguistic processes involved. The purpose of this paper is to propose scenarios offering mechanisms both for Deep Borrowing (core lexical Β forms and morphosyntax) under the maintenance conditions and for the incorporation of LI material (especially from the morphosyntax, but also from the lexicon) into the L2 under the shift condition. At the heart of the scenarios is CS.

The interference process: background considerations Prior to discussions of the proposed outcomes, it is important to raise two premises. One premise has to do with the social functions of CS and the other refers to the central role of the ML in structuring ML+EL constituents in CS utterances. Since the proposed scenarios require CS to occur, let us consider the motivations for CS. In brief, CS has this overall social purpose: to signal identity with the socio-psychological attributes associated with the language to which a speaker switches, or identity with the two or more languages in the case of a pattern of overall switching (i.e. what will be discussed below as "CS as the unmarked choice"). In Scotton (1988b) four social motivations for CS are presented as possibilities under a markedness model of code selection in general. Only one of these, CS as the unmarked choice, is relevant to this paper. The others will be mentioned briefly. First, in a community showing widespread bilingualism, one function of CS is to present "sequential unmarked choices". That is, given the norms of the community and the interaction in specific, one language is more unmarked or expected than the other as the medium of the interaction. However, if one or more social features of the interaction change (the topic, a new participant joins in, etc.), then the unmarked medium also may change. In such circumstances speakers may begin an interaction in one

40

Carol Myers-Scotton

language, the unmarked choice, but then switch to the other language (now the unmarked choice) with a change in situational features. Second, if there is no change in situational features but a speaker switches codes, such a switch is unexpected, given societal norms; it is a "marked choice". When the speaker makes a marked choice, he/she is negotiating a change in the social distance holding in the interaction, in the rights and obligations which exist and are created between participants. An example of "CS as a marked choice" follows in (8): (8)

Setting: a bus in Nairobi. The conductor has taken money for the passenger's fare, promising to return with change, but the passenger is about to alight and the conductor has not yet given him back his change. Finally, the passenger asks for his change. The conversation has been entirely in Swahili, the unmarked choice for such a service encounter in Nairobi. But when he still does not get his change, he switches to English. The conductor also switches. A s a marked choice, for this encounter, the message in using English is one of education and likely authority. Passenger (English): / am nearing my destination. Conductor (English): Do you think I could run away with your change? (Scotton - Ury 1977: 17)

"CS as an exploratory choice" is a third possibility. In an unfamiliar interaction or when the applicable norms conflict, a speaker may be unsure of the "best" linguistic choice. In such a case, he/she may use CS as an exploratory device, trying out first one language to see what the interlocutor's response is, and then trying out another. Within a few turns, a choice is established as the medium (accepted along with its socio-psychological attributes and therefore accepted as the basis for the rights and obligations which hold between participants for the interaction). All of these three motivations for CS may occur in any community where CS is a conversational strategy. In contrast, the type of CS of interest in this paper, CS as an unmarked choice, occurs only in certain types of bilingual (or multilingual) communities and generally only for ingroup (usually casual) interactions (Scotton 1986, 1988a). Such CS only occurs between peers who positively evaluate the identities indexed by the two or more codes for their ingroup interactions. That is, it is not enough that the speakers are fluent in both languages; they must hold certain positive attitudes toward them. Typically, the message of CS as an unmarked choice is simply that it is unmarked for the speakers involved to have not a single identity, but dual identities, those symbolized by the codeswitched languages. For example, in Nairobi,

Codeswitching as a mechanism of deep borrowing

41

such CS between Swahili and English is common among educated speakers for their informal inter-ethnic interactions; Swahili symbolizes a relatively de-tribalized, pan-East African urban identity, while English symbolizes education, authority, and a world perspective. Speakers using such CS associate themselves with both of these sets of attributes. The reason this type of CS does not occur in all communities is that the languages available for CS may symbolize values (ethnic or otherwise) seen as conflicting by the community, and speakers do not wish to associate themselves with both sets of values. CS as an unmarked choice also differs from the other three types discussed above in the social significance of each individual switch. Under CS as an unmarked choice each switch is not intended as a negotiation of interpersonal relations; rather, the overall pattern of switching indexes the negotiation, which is one of coordinate identities. Examples (1) through (7) all exemplify this type of switching. It is CS as the unmarked choice which interests us in this paper because of its to-and-fro action, since it is the overall pattern which is socially symbolic, rather than a single switch. This type of CS "displays" both languages in the same utterance, facilitating DB, it is claimed. Further, in those communities where it occurs such CS is very frequent since it is occasioned by a "state of mind", a perception of self and one's fellows. Also as background material, let us consider further the role of the ML in switching, returning to the frame model of CS sketched above. Evidence from much data gathering in Nairobi shows Swahili alone or Swahili/English CS as an unmarked choice for inter-ethnic ingroup conversation, especially of a casual or non-public nature. This is the finding for a data corpus consisting of more than 20 hours of naturally-occurring conversation between Kenyans who speak both languages as a second language but for whom Swahili is the more dominant of the two. These conversations (recorded in 1983) as well as some loosely-structured interviews (recorded in 1988) are the source of Swahili/English CS examples in this paper. When CS occurs in these conversations, Swahili is almost always the ML. This is in accord with the psycholinguistic and sociolinguistic criteria cited for identifying the ML vs. the EL (relatively higher speaker proficiency in the ML and the ML as the more unmarked choice for the interaction). However, a crucial point for this paper is that the designation of ML may change over time for a community (with an accompanying change in relative speaker proficiency and in relative socio-psychological unmarkedness of the varieties involved). Such a community-wide change would typically be preceded by changes for specific interactions only. Further, in communities where the two (or more) languages involved in CS have much currency,

42

Carol Myers-Scotton

whether for similar or different reasons, it is possible which language is the ML in CS is somewhat variable. In most cases, this variability may represent a change in progress in the designation of ML for CS in that community. It is also possible that for some speakers their ML changes from interaction to interaction, depending on the situational features of the interaction. 7 For example, consider the following examples from the Nairobi data. The ML changes for the same speaker in consecutive sentences. One may suppose Swahili is the ML because of the speaker's proficiency in it and because of Swahili's position as the national language and its currency in Nairobi in particular; English becomes the ML because the speaker is accustomed to using it when talking about job-related matters. In (9a) English is the ML; English morpheme order prevails in the ML+EL constituent. In (9b) Swahili is the ML; Swahili morpheme order is followed in the NP of the prepositional phrase: (9)a. [If a customer has] any ΑΡΑΤΕ service]}

MALALIMIKO-[NI

WAJIBU WANGU KUMSAIDIA ILI

any

MALALIMIKO = S t a n d a r d S w a h i l i

malalimiko

yo

any

problems

problems

any

yote

'[If a customer has] any complaints-[it is my responsibility to help him/her so that he/she gets service].' (9)b. [Nimefanyia kazi yangu hapa Nairobi katika]

OFFICES

tatù /ofisaz tatù/.

[THIS ONE IS MY THIRD PLACE],

offices offices

tatù = Standard Swahili three

ofisini offices

tatù three

'[I have done my work here in Nairobi] in three offices. [This one is my third place].' (Myers-Scotton 1989: 5) The great majority of sentences in this data corpus show Swahili as the ML. However, there are enough examples of English as the ML to demonstrate that its designation is not irreversible. See (10) for another example of English as ML: (10)

[HATUENDI college]

Saturday

ASUBUHI.

Saturday asubuhi = Standard Swahili

asubuhi mornings

za of

Jumamosi Saturday

'We don't go to college on Saturday mornings.' (Myers-Scotton unpubblished data)

Codeswitching as a mechanism of deep borrowing

43

Following the CS model introduced above, hatuendi college 'we don't go to college' must be analyzed as a Swahili EL island (Verb + Locative), with college as a Β form in Swahili. Saturday asubuhi is an ML+EL constituent, with English as the ML. Its relative frequency (vs. Swahili jumamosi) does not indicate Saturday is a Β form; but even if Saturday were considered a Β form in Swahili (many English time expressions are Β forms), the morpheme order and construction of this NP comes from English. In English the NP is a compound noun; in Swahili a genitive construction with an inflected -a "of association" is required. There is limited, but compelling, evidence from other CS studies that over time speakers may well change from one language to the other as the ML. One would expect this is what happens in an immigrant community as it becomes more assimilated into the L2 culture; that is, the L2 becomes the ML. For example, Nishimura (1986) implies that English is as often the ML as is Japanese in CS examples drawn from Japanese-American immigrants in North America. The next step for these speakers would be a switch to English as the only ML.

CS-based scenarios of contact-induced change Let us turn now to possible outcomes, either under a maintenance condition (DB occurs but the LI is maintained) or a shift condition (LI features appear when there is a shift to the L2). Outcome I: content morphemes as core lexical Β forms This outcome is defined by the borrowing of core content morphemes from the EL, but not system morphemes. Some CS, particularly CS as an unmarked choice, is a prerequisite for wide scale borrowing of such content morphemes, it is argued. It is acknowledged that simply a bilingual condition and regular speaker use of the languages in question encourages such borrowing. But these are not sufficient conditions to explain extensive borrowing of this type. If they were the main triggering factors, there would be no reason why numerous system morphemes would not be borrowed alongside content morphemes. This is not the case. Rather, our hypothesis about CS as the crucial mechanism in such borrowing gives the correct outcome. We argue that the reason core content morphemes are borrowed, but not system morphemes, is that only content morphemes from the EL occur freely in ML+EL con-

44

Carol Myers-Scotton

stituents, making them highly salient because of their co-occurrence with ML morphemes. But EL system morphemes cannot occur (productively) in such constituents, so they lack this salience. They occur only in EL Islands where they are "buried" in other EL morphemes. CS as the unmarked choice contains more ML+EL constituents than other types of CS (because it shows much switching back and forth between languages). Unfortunately data have not been collected in such a way to test the hypothesis there is a connection between much borrowing of core Β forms and CS as an unmarked choice. It is the case that a large scale study comparing French Canadian communities in Quebec and Ontario Provinces, Poplack, Sankoff, and Miller (1988) found more borrowing in the Ontario communities where there also was more CS. These, compared to the Quebec communities, are where the psycho-sociological factors should promote CS as the unmarked choice. Unfortunately, the authors do not differentiate core Β forms from cultural loans, nor do they look at types of CS. The process of core borrowing probably begins with nouns and verbs (because they have the most "psychological salience") and then proceeds to discourse markers and adverbs (because they have "positional salience"). Even though cultural nouns or verbs as Β forms may always outnumber core Β forms, many cases of core Β forms are detailed in the literature, unfortunately without an accompanying consideration of CS. For example, on such Β forms from Luganda and Swahili in Ateso, see Scotton - Okeju (1973). Or see Pugh (1989) on the borrowing of Russian verb stems into Balto-Finnic dialects. Possibly in a later stage in this outcome, discourse markers are borrowed, lexemes hovering on the border between content and system morphemes. As noted, position is important for these morphemes. For example, they often occur in initial position in a conversational turn, or at least as heads of constituents (e.g. as prepositions in PPs). Their position make them almost as attractive as nouns and verbs to become Β forms; possibly they are even more attractive, since, as free forms, they do not have to be extracted form inflections, as is often the case with nouns and verbs. Quantitative evidence that a pattern of alternating between the use of two languages for ingroup conversations results in such discourse markers becoming Β forms is found in a study of the incidence of the English lexeme so as used by French Canadian speakers in Ontario, Canada (Mougeon Beniak 1987). (The study looks at so as a discourse marker or consecutive conjunction.) Considering the association between so incidence and a number of independent variables, including sex, socio-economic status, locality of residence, and an index of French maintenance, Mougeon and Beniak con-

Codeswitching as a mechanism of deep borrowing

45

elude that French maintenance is the best predictor of the use of so. But they go on to point out that the relationship between level of French maintenance and frequency of use of so is not a linear one. Rather, it is the mid level users of French, who show roughly balanced alternation between French and English, who are the greatest users of so. Mougeon and Beniak suggest that so is lacking in the French of others because their linguistic behavior is more compartmentalized. They write (1987: 40): . . . the low maintainers by definition speak little French in the private domain while the high maintainers speak mostly or only French in the said domain. In other words, it would appear that bilingualism has to be of the 'unpatterned' type .. .in order for ( this) core borrowing to become prevalent, [emphasis added]

Interestingly, although Mougeon and Beniak's data did not specifically detail how the heavy users of both French and English use them (simply speaking English and French at different times during the day, or speaking both languages in the same interaction when situational factors change (CS as sequential unmarked choices), or producing an overall pattern of CS (CS as the unmarked choice), they conclude by suggesting a clear connection between core lexical borrowing and CS. They write (1987: 43): . . . that sentence connectors and other kinds of discourse organizers are often reported in lists of core lexical borrowings may not be a coincidence since these items all occur at prime switching points. If this view of things is correct, then core lexical borrowing (or at least some of its manifestations) would simply be a by-product of code switching.

My own research on Β in Africa reinforces these findings and interpretations. Study of Swahili Β forms in Lwidakho, a Luyia variety spoken in Western Kenya, and also of English Β forms in Shona, as spoken in Harare, Zimbabwe, produces similar findings. That is, Lwidakho speakers who use both Lwidakho and Swahili in their everyday lives show a number of core Β forms from Swahili. (Most of the speakers studied use Lwidakho on its own at home, but regularly use both Lwidakho and Swahili in their daily contacts outside the home, sometimes as CS to fellow Lwidakho speakers, although Swahili in its own is spoken to outsiders.) Common Β forms include Swahili conjunctions mpaka 'until' and lakini 'but'; also the discourse marker sasa 'now' is a Β form. See (11): (11) A male farmer, age 38, with primary education only is being interviewed in Lwidakho: Question: Maduma kenaka khulwenulu kakhola karina mumilimi yumu? 'How is the maize, or rather what do you think about it now?'

46

Carol Myers-Scotton

Answer: SASA, maduma yaka kavele kali kunu Iwenolo kakaholakhu vulahi, LAKINI kavule kenyanga khurulitsa vusimpu nunundi Iwenndu ¡DAD1K1HII kavele nikonekhanga Iwa ikori. 'Now, the maize which had flowered is doing pretty well, but that flowering now, this minute, might get spoiled.' Note: IDAKIKI HII = Standard Swahili dakika hii minute this In interviews (conducted in Shona) in Harare, one finds many examples of discourse markers as Β forms as well. Most prominent are because and but. A quantitative study (Bernstein - Myers-Scotton forthcoming) shows that in 130 interviews, there are 367 possibilities for encoding the concept of 'because'. For these, the Shona lexeme nokuti occurs 341 times while because occurs 26 times or 7 per cent. Out of 315 possible encodings for 'but', the Shona lexeme asi occurs 289 times while but occurs 26 times, or 8 per cent. Thus, this test of relative frequency indicates that because and but have moved far enough along the continuum between CS forms and Β forms to qualify as incipient Β forms. There is no need to claim they must totally replace their Shona equivalents to qualify as established Β forms; in fact a long period (or permanent state) of duplication may be the norm for such types of Β forms. For example, see (12) in which the same speaker first uses but and then the Shona asi : (12) The interviewed is a 47 year old building inspector with primary school education but many additional short courses. In response to the interviewer, he says: Na-ΤΕΝ O'CLOCK tofanira kunwa tii BUT dzitnwe nguv hasvivioiti huti tinwe tii pamusana pebasa asi tine bvumo yekinwa katatu. 'At then o'clock we have tea, but sometimes it will never happen for us to drink tea because of work. But we have the right to drink tea three times.' (Myers-Scotton unpublished data) Further analysis of the Lwidakho, Shona, and Swahili data sets would show that many core Β forms are common. These include most prominently the discourse markers already discussed, as well as verb stems (e.g. Swahili/English CS a-li-DECIDE 'he decided'). Almost all PPs referring to times in the Shona data are in English, but with the preposition in Shona as a prefix (e.g Ndinofanira kumuka n a - H A L F PAST FOUR OR FIVE O'CLOCK Ί have to get up at half past four or five o'clock').

Codeswitching as a mechanism of deep borrowing

47

Note, however, that in Outcome I the Β forms are content morphemes, not system morphemes. As long as the speaker's native language (or "national language" in the case of Swahili) remains the ML in CS interactions, system morphemes from the EL can only occur in EL islands. Even if EL system morphemes are relatively frequent in islands, they are not very "available" for borrowing because system morphemes are either bound forms or "parasitic" parts of constituents (e.g. determiners).

Outcome II: Relexification with EL content morphemes Taking Outcome I's scenario a step further, we arrive at a near-relexification of the non-system morphemes of the ML, a much rarer outcome but still attested in a handful of cases. The Javanese spoken by Peranakan Chinese in East Java (Wolff 1983) seems to be an example. The Peranakan are people of Chinese descent in Indonesia who have given up almost all vestiges of their background; most are descendants of people who left China at least 100 years ago. Wolff notes that Peranakan Javanese differs only in small ways from Pribumi (native) Javanese in its phonology and morphosyntax. But he goes on to say (1983: 593-594): "However, when we look at the lexicon, we are in for a surprise, for we find coexisting side by side with Javanese items a complete array of items of Indonesian provenience." (Wolff indicates he prefers to call these lexical items of "Indonesian provenience" because they are often Javanese caiques and do not occur in any Indonesian dialect.) About example (13) Wolff states (1983: 594), "The markers and affixes are of Javanese origin; the phonemics, morphophonemics, and syntax are Javanese; but the contentives are purely Indonesian." 9

(13) INI these

SPATU shoes

ta? by-me

PAKI?e JA PA let-them-be used- only- question particle

'I'll just wear these slippers, ok?' Note: Capital letters indicate Indonesian

Wolff even goes on to say that some markers (i.e. some system morphemes) can be in Indonesian and gives an example (1983: 594) in which the Javanese marker send appears with its Indonesian analogue yang as well as a composite of other Javanese system morphemes and Indonesian content morphemes:

48

Carol Myers-Scotton

(14) Kembange the-flowers

KAN isn't-it

seng the-one-who-man

LAK1 himself

'Isn't the groom himself who brings the

flowers?'

dewe YANG mBAWA? the-one-who-it-brings

Muysken (1981) provides another example of relexification of content morphemes in the case of Media Lengua, a contact language spoken in several communities in the Ecuadorian Highlands. According to Muysken (1981: 52), the language is Quechua in its grammatical structure and almost 90 per cent Spanish in its lexicon. (15) Quechua Med.L. Sp

yalli-da dimas-ta

tamia-pi-ga, llubi-pi-ga,

mana no

too-much

rain-.S't/5-to,

not

ri-sha-chu i-sha-chu g o - / FUT-NEG

si lleuve demás, no voy a ir 'If it rains too much, I won't go' (Muysken 1981: 54)

Note that the type of relexification which has taken place in both Peranakan Javanese and Media Lengua would be predicted as the type of Β which is a permissable outcome of CS as a mechanism of Β when the M L is maintained. The M L Principle's System Morpheme Hypothesis requires all the productive system morphemes in M L + E L constituents to come from the ML; other (content) morphemes may come from either language. And it is precisely the system morphemes which are being maintained in the M L while the content morphemes are relexified. Other cases of similar relexifications while an M L is being maintained as such are not hard to find. For example, Haugen in discussing the shift from Norwegian to English in America describes the state of Norwegian (presumably while it is still the ML) in the very terms the frame-based CS model would predict (Haugen 1969: 71) quoted in Romaine (1989: 250): In becoming bilingual within the American cultural environment they [speakers of Norwegian] were forced to modify their Norwegian if they wished to continue using it. At practically every point they maintained the basic phonetic and grammatical structures of their native dialects, but they filled in the lexical content of these structures from the vocabulary of English, [emphasis added]

Also, Tosco (this volume) sums up his discussion of Dahalo as an endangered, but maintained, language in some of the exact words which the framebased model of CS would use. That is, he writes, "Taking into consideration both assimilated and unassimilated material, the net result is often some kind

Codeswitching as a mechanism of deep borrowing

49

of Mischsprache, in which Swahili is framed into Dahalo."[emphasis added] (The only problem might be the use of Mischsprache, depending how it is intended; according to the frame model, there is certainly no "mixing" in any non-systematic way.)

Outcome III: a possible turning point: the ML changes in CS Outcome III is highly speculative, but it accounts in a principled way for the melange of languages found in Ma'a, as well as for other dramatic cases of pervasive incursions of an L2 into the morphology or syntax. Also, it explains the many cases of more limited, but still extensive, borrowing of system morphemes. Under Outcome III, speakers of an LI take a step just short of shifting to an L2 which has become sociolinguistically dominant. Instead, they acknowledge the L2's dominance by shifting to the L2 as the ML when they are engaged in CS. Recall the role the ML plays in ML+EL constituents in CS and it is obvious why this "flipping of the ML" in CS is so important. Both morpheme order and productive system morphemes in these constituents come from the ML. If the LI loses its preeminence as the ML in CS, with the L2 taking its place, the result is "outside goes to inside" change. Denison (1977: 21) offers us a vivid metaphor for such replacement, recalling "the Russian fable of the wolf which ate the sleigh-horse and thereupon found itself in harness as a horse-substitute". Let us speculate how such replacement would affect the Ma'a scenario. Of course, the first step would be bilingualism in one or more local Bantu languages, soon followed by CS in conversations with their Cushitic brethren who were in the process of shifting to a Bantu language as their LI. The next step would be a change in the ML of this CS - from Cushitic to Bantu. This would happen under the pressure of sheer numbers of Bantu speakers (Ma'a is a very small [Cushitic] island in a very large sea of Bantu), not to mention possible accompanying socio-political pressures. This shift in ML would mean that the system morphemes as well as morpheme order in ML+EL constituents in CS would come from the Bantu language, not from the Cushitic one. Once this replacement is made, it is easy to see that the ancestors of today's Ma'a could carry over the practice of using non-Cushitic system morphemes and syntax to non-CS renditions of Ma'a. This scenario fits present day results since in Ma'a almost all of the system morphemes and morpheme order come from Bantu. As Thomason and Kaufman (1988: 224) remark,

50

Carol Myers-Scotton In morphology, Ma'a retains only one systematic inflectional feature of Cushitic original, namely, suffixed prononiminal possessors . . . Otherwise, Ma'a preserves just a few fossilized Cushitic inflectional features, while the entire productive inflectional apparatus is borrowed from Bantu - the functional categories, their ordering, and the affixes themselves, [emphasis added]

It does seem as if the social situation has to be unusual for speakers to shift f r o m their o w n L I to L 2 as the M L in C S interactions. In the case of M a ' a , there is a Cushitic enclave w h i c h distinguishes itself by language f r o m not only its Bantu neighbors, but even its ethnic neighbors (those w h o have shifted to Bantu languages, largely Pare). It is not hard to see h o w , in such a situation, the M a ' a could shift their M L in C S . It is k n o w n that the M a ' a have regular interactions, not to mention ethnic ceremonies, with their brethren w h o have shifted f r o m Cushitic (now Pare speakers). In order to m a k e themselves understood by the Pare, it is reasonable to a s s u m e remaining M a ' a speakers must m a k e Pare, not M a ' a , the dominant language of such encounters. But they could symbolize their ethnic identitiy by not using Pare exclusively, but rather insisting on a C S pattern (CS as their u n m a r k e d choice for such interactions). For such SC, Pare would be the M L . Therefore, C S as an u n m a r k e d choice is suggested as the m a j o r m e c h a n i s m by which m o r p h o s y n t a x enters a recipient language. T h e same type of scenario m a y apply to the case of M e d n y j Aleut, for one. This language has incorporated Russian inflections for finite verbs (see e x a m p l e [16]). Aleuts w h o are bilingual in Russian (wives of Russians?) m i g h t use a C S pattern in conversing with Russians. Further, it is likely that they must m a k e Russian the M L of such CS, j u s t in order to facilitate c o m munication and because of R u s s i a n ' s socio-political dominance. T h o m a s o n and K a u f m a n (1988: 237) hint at this idea in their statement that "Aleuts bilingual in Russian might . . . have f o u n d it easier to c o m m u n i c a t e with Russians in Aleut if they used Russian endings on the verbs." The C S model presented here provides a m o r e principled statement of the s a m e idea; that is, the Aleuts use Russian inflections because universal constraints on M L + E L constituents in C S require t h e m w h e n Russian is the M L . (16) sg.

1: 2: 3:

Bering Aleut unuci-ku-q unuci-ku-xt unuci-ku-x

M e d n y j Aleut Russian Gloss unuci-ju ja sizu Ί sit' unuci-is ty sidis 'you sit' unuci-it on sidit 'he sits' ( T h o m a s o n - K a u f m a n 1988: 234)

W h i l e such e x a m p l e s of a language with m u c h of the lexicon f r o m one source and important/substantial system m o r p h e m e s f r o m another m a y well

Codeswitching as a mechanism of deep borrowing

51

be rare, a constant theme in all writings about them, a moment's reflection indicates they may only be unusual because they represent incomplete language shifts. After all, the types of social situations which prevailed in the case of Ma'a and Mednyj Aleut are not what is rare; only the linguistic outcome is rare. Unusually such situations result in language shift. In fact, while shift of the ML in CS is probably not a prerequisite to language shift, it is a likely mechanism to initiate shift. Whether what we have in present-day Ma'a is the result of deep borrowing OR shift remains at issue. There is good motivation to argue that presentday Ma'a is a Bantu language which retains Cushitic features as part of its substrate; such a conclusion would be compatible with, but not identical, to Brenzinger's (1987) conclusions. The major motivation for this conclusion is that this brings the Ma'a case into line with other language-contact situations; that is, why create an "exception" when the data can be explained within a shift scenario? Thomason (1983) and Thomason and Kaufman (1988), however, argue that the Ma'a case is unusual, showing especially deep borrowing. Their argument for borrowing, not shift, rests largely on their claim the basic vocabulary is mainly Cushitic. But adherents of the shift hypothesis could counter-argue that (a) only parts of the Ma'a lexicon are Cushitic and (b) those parts are basic vocabulary items, just the sort which would be maintained to symbolize ethnic identity. Outcome IV: language shift It is acknowledged that language shift may occur with little or few of the effects outlined under Outcomes I, II, and III. That is, there is no reason to insist that an overall pattern of CS as the unmarked choice, followed by a shift in the ML in such CS, must result in language shift nor must it be a prerequisite to language shift. Yet, CS, with a shift in ML, is an obvious mechanism which not only promotes language shift, but also explains the process. Other discussions of language shift have been silent on the process. This paper does not pretend to detail the process of a language shift facilitated by CS; but it deals only with the principle mechanism. Much more empirical evidence is needed to suggest details. For example, one possibility is that the shift of ML in CS (affecting both system morphemes and morpheme order in CS) need not be transmitted in a unitary fashion to production of the "old" ML in a non-CS mode. For example, it is possible that the "old" ML may incorporate first only the syntax of the "new" ML, or first only some groups of the system morphemes of the "new" ML. Based on the claims about the imperviousness of inflectional morphology to change and

52

Carol Myers-Scotton

the known cases of syntactic convergence, and if syntax and morphology are separate modules, then it is likely the syntax may be the module more open to change. The example of Mexicano (Nahuatl) spoken in Central Mexico suggests this. Hill and Hill (1986: 233) state flatly that Malinche Mexicano syntax has converged with that of local Spanish and that system morphemes which are function words also come form Spanish. They write: It is possible to speak an intelligible (although not an elegant) Mexicano by simply inserting items of either Spanish or Mexicano etymology, appropriately inflected in Mexican, into Spanish sentence frames held together by Spanish function words.

O u t c o m e V: l a n g u a g e death Finally, any time a language reaches one of the previous four outcomes, a further possible scenario is for all of its speakers to shift their language; that is, language death overtakes the "old" L I . Many, including some of those presenting papers at this symposium, speak of language death as if it necessarily represents morphosyntactic loss or decay. It follows from the argument developed in this paper that the dying language need not have lost features so much as it may have substituted those of the "invading" language. Further, the argument here also accommodates such discontinuous data as that documented by Dorian (1981) for East Sutherland Gaelic; that is, when questioned by the investigator, some speakers show a full competence in the language, but others show comparative loss and reduction. A scenario for language death, with CS as a mechanism in the community, would include the following parts. First, those fluent speakers, who produce the language in its near-full form, would not engage extensively in CS as an unmarked choice. The reason is that their fluent production of the dying language implies loyalty to the identity associated with that language, not to the "invading" language. Recall that speakers must positively identify with the attributes of both languages in order to engage in CS as an unmarked choice. (This does not mean, of course, that the fluent speakers cannot engage in other types of CS; but that is not at issue.) Further, even if fluent speakers do engage in extensive CS, the argument developed here predicts that the dying language must remain their ML in such CS. Thus, the mechanism for extensive incursions into the morphosyntax from the invading language is not available for these fluent speakers.

Codeswitching as a mechanism of deep borrowing

53

Second, other speakers in the community (e.g. possibly Dorian's "semispeakers"), however, have quite different speech patterns, it is hypothesized. That such discontinuities exist in the community is documented by many writers on language death. These speakers use the dying language very little except in a CS pattern, and when this happens, it is the invading language which is the ML, it is hypothesized. That is, they do not see their identities as solely tied to the dying language and, in fact, see them better symbolized by the invading language. Note that for some communities, there are no fluent speakers, but rather the only use of the dying language is in CS. Thus, it is not language decay which kills a language, but rather (a) among the semi-speakers, replacement at all levels, with the final blow represented by morphosyntactic replacement; and (b) among the fluent speakers, their own death.

Conclusion In this paper, several possible outcomes for language contact situations have been suggested, some applying only to maintenance conditions, but others applying also to language shift conditions. Outcome I includes extensive incorporation of core content morphemes as Β forms into a maintained language, which is also the ML in CS as the unmarked choice. It is emphasized that some borrowing of core content morphemes, and even of some system morphemes, is possible without CS. But when there is extensive borrowing of content morphemes, some explanation beyond citing bilingualism is needed. CS as the unmarked choice offers an explanation. Such switching includes ML+EL constituents which consist of frames of morphosyntactic material from the ML. The frames can be filled in with content morphemes from either the ML or the EL. As long as the maintained language is the ML, CS offers no means for EL morphosyntax to enter the ML, but it does provide a mechanism for EL content morphemes to migrate. This prediction matches the results under Outcome I. Outcome II takes incorporation a step further to relexification from the EL, but again largely allowing only content morphemes. Outcome III rests on a more radical CS mechanism; for socio-psychological reasons, speakers shift the ML in their CS. The result is that they will now use the morphosyntax from the new ML as a frame for ML+EL constituents in CS. This shift sets the stage for borrowing the new ML's morphosyntax into their original ML (generally their mother tongue). Outcome IV takes such interference a step further: speakers shift their main language to the new ML of CS. Finally, Outcome V is language death; the

54

Carol Myers-Scotton

old M L is no longer used on its o w n by larger and larger numbers of speakers in the community; eventually, it is dropped entirely, even from CS. This "CS hypothesis for contact-induced language change" suggests a number of specific hypotheses for field testing in African contact situations, using naturally-occurring data. First and most generally, the hypothesis that C S as an unmarked choice is involved in extensive borrowing of core content morphemes is open to testing. Second, that a switch in the M L in CS interactions is what triggers borrowing in the morphosyntax of the old M L also can be tested. Finally, situations in which language death is occurring can be studied to determine if, in fact, semi-speakers use the dying language mainly only in C S and then do so with the invading language as the ML.

Notes 1. This paper is informed by, and includes examples drawn from, fieldwork in East Africa and Zimbabwe conducted in 1983 under a Fulbright Senior Research Fellowship and in 1988 under a summer grant from the Social Science Research Council/American Council of Learned Societies. Support is grateful acknowledged. Comments on the manuscript from Edouard Beniak, Nancy Dorian, Frances Karttunen, Yuriko Kite, Raymond Mougeon, and John Singler are also appreciated, as well as discussion by fellow participants in this symposium; I alone am responsible for the remaining shortcomings. 2. I began using "core loan" in Scotton and Okeju (1973) for: all Β forms for which the recipient language has its own lexemes. Mougeon and Beniak (1987: 337) adopt this usage and aptly amplify on it by noting these loans are "gratuitous" in the sense that "the recipient language always has viable equivalents". Note that this use of "core" necessarily has a much wider scope than the very limited "core" vocabulary on "basic" 100-word lists and the like. 3. The term "system morpheme" comes from Bolinger (1968) and it used as a cover term for both inflectional affixes and for function words. Exactly what is to be included under system morphemes is at issue; however, Myers-Scotton (1992) offers a much fuller discussion of the content vs. system morpheme distinction. 4. In his discussion of morphological borrowings, Sapir (1921: 215-220) went so far as to say "we have no really convincing examples of profound morphological influence by diffusion", and he characterized language as "probably the most selfcontained, the most massively resistant of all social phenomena". He is quoted by Emeneau (1964: 643), who goes on to discuss massive Dravidian morphological influences on the Indie languages. 5. It is acknowledged that the criteria used to define the ML are not water-tight; yet, community members readily recognize which language of a pair (or more)

Codeswitching as a mechanism of deep borrowing

55

in question meets the criteria. But see Myers-Scotton (1992) for a more precise definition. 6. Myers-Scotton and Azuma (1990) includes numerous examples from typological diverse data sets supporting the ML hypotheses about the frame of ML+EL constituents. They also discuss the fact that the System Morpheme Hypothesis applies only to productive system morphemes. That is, fossilized system morphemes from the EL may occur, but not in any productive sense. Also, they discuss the possibility of "bare forms", typically EL nouns or verbs which are not accompanied by the system morphemes of either the ML or the EL. 7. For example, persons bilingual in Language A and B, who because of their sociolinguistic profiles have memberships in communities of both Language A and B, report their ML for CS varies according to the group with which they are speaking. With (native) speakers from Group A (who also know Language B), Language A is the ML; the opposite case applies with bilingual Language Β speakers. For example, Yuriko Kite, a Japanese-English bilingual living in the United States and married to an American, reports her ML, when engaged in CS as an unmarked choice, alternates according to whether the interlocutor is Japanese- or English-dominant. 8. In this example, service is a Β form. 9. Wolff (1983: 597) is careful to point out that because Javanese and Indonesian are so closely related, it is difficult to make an air-tight case that all the system morphemes in Peranakan Javanese come from Javanese rather than Indonesian.

References Agard, Frederick - Gerald Kelley (eds.) 1983

Essay in Honor of Charles T. Hockett. Ithaca: Cornell University Linguistics Department.

Bernstein, Janice - Carol Myers-Scotton forthcoming

"English loans in Shona: Consequences for linguistic systems", to appear, International Journal of the Sociology of Language.

Bolinger, Dwight 1968

Aspects of language. New York: Harcourt, Brace & World.

Brenzinger, Matthias 1987

Die sprachliche und kulturelle Stellung der Mbugu (Ma'a). [Unpublished M.A. thesis, University of Cologne.]

Dawkins, R.M. 1916

Modern Greek in Asia Minor: A study of the dialects of Siili, Cappadocia, and Pharasa with grammars, texts, translations, and glossary. Cambridge: Cambridge University Press.

56

Carol Myers-Scotton

Deaton, Karen - Manuela Noske - Michael Ziolkowski (eds.) 1990 Papers from the 26th annual meeting, Chicago Linguistic Society. Chicago: University of Chicago Department of Linguistics. Denison, Norman 1977 "Language death or language suicide?", International Journal of the Sociology of Language 12: 13-22. Denning, Keith et al (eds.) 1987 Variation in language: NWAV-SX. Stanford: Stanford University Department of Linguistics. Dorian, Nancy 1981 Language death. Philadelphia: University of Pennsylvania Press. Emeneau, Murray B. 1964 "India as a linguistic area", in: Dell Hymes (ed.), 642-650. Joshua Fishman - Andrée Tabouret-Keller - Michael Clyne - Bh. Krishnamurti Mohamed Abdulaziz (eds.) 1986 The Fergusonian Impact, Vol. 2 Sociolinguistics and the sociology of language. Berlin: Mouton de Gruyter. Goodman, Morris 1971 "The strange case of Mbugu", in: Dell Hymes (ed.) 243-254. Haugen, Einar 1969 The Norwegian language in America: A study of bilingual behavior. (2 vols.) Bloomington: Indiana University Press. Heller, Monica (ed.) 1988 Codeswitching: Anthropological and sociolinguistic perspectives. Berlin: Mouton de Gruyter. Highfield, Arnold - Albert Valdman (eds.) 1981 Historicity and variation in Creole studies. Ann Arbor: Karoma. Hill, Jane H. - Kenneth C. Hill 1986 Speaking Mexicano. Tucson: University of Arizona Press. Hymes, Dell (ed.) 1964 Language in culture and society. New York: Harper & Row. (Reprinted from Language 32 [1956].) 1971 Pidginization and creolization of languages. Cambridge: Cambridge University Press. Kratz, Corinne Α. 1986 "Ethnic interaction, economic diversification and language use: a report on research with Kaplelach and Kipchornwonek Okiek", SUGIA 7: 189-226. Lowenberg, Peter H. (ed.) 1988 Language spread and language policy: issues, implications and case studies (GURT '87) Washington: Georgetown University Press.

Codeswitching as a mechanism of deep borrowing

57

Menovscikov, G.A 1969

"O nekotoryx socioal'nyx aspektax evolucii jazyka", in: Voprosy Social'noj Lingvistiki. Lengingrad: Nauka, 386-406. Mohan, Peggy - Paul Zador 1986 "Discontinuity in a life cycle", Language 62: 291-320. Mougeon, Raymond - Edouard Beniak 1987 "The extralinguistic correlates of core lexical borrowing", in: Keith Denning et al. (eds.) 1987: 337-347. Muysken, Pieter 1981 "Halfway between Quechua and Spanish: The case of relexification", in: Arnold Highfield - Albert Valdman (eds.), 52-78. Myers-Scotton, Carol 1989 A look at processing: Codeswitching and borrowing. [Paper presented at 25th Annual African Linguistics Conference, University of Illinois-Urbana.] 1992 "Constructing the frame in intrasentential codeswitching", Multilingua 11: 101-127. Myers-Scotton, Carol - Shoji Azuma 1990 "A frame-based process model of codeswitching", in: Deaton, Karen -Manuela Noske - Michael Ziolkowski (eds.) Nishimura, Miwa 1986 "Intrasentential code-switching: The case of language assignment", in: Jyotna Vaid 123-143. Poplack, Shana - David Sankoff - Christopher Miller 1988 "The social correlates and linguistic processes of lexical borrowing and assimilation", Linguistics 26: 47-104. Pugh, Stefan N. 1989 Russian verbal borrowings in Balto-Finnic: Stem-types and systemic convergence. [MS. submitted.] Romaine, Suzanne 1989 Bilingualism. Oxford: Basil Blackwell. Sapir, Edward 1921 Language. New York: Harcourt, Brace, Jovanovich. Scotton, Carol Myers 1986 "Diglossia and code-switching", in: Joshua Fishman et al. (eds.) Vol. 2, 403-415. 1988a "Code-switching and types of multilingual communities", in: Peter H. Lowenberg (ed.), 61-82. 1988b "Code-switching as indexical of social relationships", in: Monica Heller (ed.), 151-186. Scotton, Carol Myers - John Okeju 1973 "Neighbors and lexical borrowings", Language 49: 871-879.

58

Carol Myers-Scotton

Scotton, Carol Myers - William Ury 1977 "Bilingual strategies: the social functions of code-switching", International Journal of the Sociology of Language 13: 5-20.[Also, Linguistics 193 (1977): 5-20.] Thomason, Sarah Grey 1983 "Genetic relationships and the case of Ma'a (Mbugu)", Studies in African Linguistics 2: 195-231. Thomason, Sarah Grey - Terrence Kaufman 1988 Language contact, creolization, and genetic linguistics. Berkeley and Los Angeles: University of California Press. Vaid, Jyotna (ed.) 1986 Language processing in bilinguals: psycholinguistic and neurolinguistic perpectives, Hillsdale NJ: Erlbaum. Whiteley, W. H. 1960 "Linguistic hybrids", African Studies 19: 95-97. Wolff, John 1983 "The Indonesian spoken by Peranakan Chinese of East Java: a case of language mixture", in: Frederick Agard - Gerald Kelley (eds.), 590-601.

Language decay and contact-induced change: Similarities and differences1 Hans-Jürgen

Sasse

1. The distinction between language contact and language decay In the present paper I shall try to establish a number of criteria for a distinction which I think is extremely important, but which is not always easy to make -and has therefore been the cause of considerable confusion, namely the distinction between the phenomenology of transfer processes in normal language contact situations and the pathological situation of language decay in the final phase of language death. Let me first try to explain in a few words why I consider this distinction necessary. First of all, if it could be demonstrated that normal language contact phenomena can be differentiated from language decay, we would have a structural correlate of language death. We could say that a language is dead once it arrives at the phase of decay. As is pointed out in Sasse (this volume), decay status has consequences both for the psychological condition of the speakers involved in the decay process, who tend to hide or disguise their imperfect language competence and avoid speaking the obsolescent language (a condition which accelerates the final extinction of the language), and for the possibilities of revitalization insofar as a decaying language can be revitalized only by way of drastic creolization. Second, it would be desirable for historical linguistic studies to have a methodological guide on hand, a heuristic yardstick, so to speak, which could help us distinguish between cases of continuous language transmission and cases of interrupted language transmission. If it were possible to identify language decay as a phenomenon sui generis, closely connected with interrupted language transmission, one could perhaps deal with certain problems of genetic relationship in a more straightforward way. Thirdly - and this is a more practical question but nevertheless of extreme importance - it is necessary to evaluate the data we obtain when doing field work in language death situations. If we condict a

60

Hans-Jürgen Sasse

linguistic investigation in a speech community where only semi-speakers are left, we must be aware of the possibility that most, if not all, of the elicited material will be distorted speech that has undergone certain processes of reduction, with the result that the original grammatical system of the former full speakers of the language under investigation will be accessible only by way of reconstruction. The view that borrowing and decay must be kept apart is not unanimously taken by all specialists on language contact and death. On the contrary, it is often said that the types of change observable in a moribund language do not differ principally from those occurring in other kinds of contact situations. Even Dorian (1981: 151) has stressed that the same kinds of changes occur in healthy languages, though the amount and the rate of change may be atypical in the case of language death. A similar view is expressed by Romaine (1989: 71 ff.). Some authors, while tacitly assuming that there is a difference, draw attention to the fact that in some cases it is hard to distinguish the two kinds of phenomena (e.g. Campbell and Muntzel 1989: 195). While this is basically correct (cf. also Sasse, this volume), it can nevertheless be shown that in principle borrowing and interference on the one hand, and irreversible loss and reduction in the system of an obsolescent language on the other, are quite different things. Before linguistic evidence is presented to substantiate my claim, it is necessary to distinguish the two types of individuals responsible for the two types of phenomena under discussion.

2. Bilingual speaker and semi-speaker 2.1. The bilingual speaker The locus of language contact, interference and borrowing is the bilingual individual. It is commonplace in the literature on bilingualism to distinguish between coordinate and compound bilingualism and to discuss at length the psychological and psycholinguistic evidence against and in favor of this distinction, resulting in the conclusion that the reality is too complex to be captured by such a simple dichotomy. Exclusively coordinate bilingualism is a difficult matter because the two languages can never be kept totally apart. On the other hand, given the complex nature of meaning in human language, the differences in the associational networks, etc., it is hard to imagine a bilingual speaker of the compound type with a totally fused semantic system.

Language decay and contact-induced change

61

I shall not go into the complex question of bilingualism here; for further details the reader is referred to such classical works as Weinreich (1967) or the more recent general treatments such as Romaine (1989), and Hamers and Blanc (1989). Suffice it to say here that in cases of long-term collective bilingualism there is usually a drift towards a certain kind of compound bilingualism which amounts to the formula "one form - two substance systems". This is brought about by the fact that, in a totally bilingual community, both languages are transmitted by the same persons over generations in an identical sociocultural setting. The result is that interference phenomena cease to be spontaneous, individual and ad hoc and begin to be transmitted from one generation to the next. The ideal goal over the long term in such a situation is a total isomorphism of the two languages.

2.2. T h e semi-speaker Now to the locus of language decay, the semi-speaker. As a producer of the kind of decayed language we are examining here, he must be carefully set off from other types of imperfect speakers found in language death situations. In the dying speech communities I am familiar with, there were always two types of imperfect speakers. The first type had a comparably good proficiency in the grammatical system of the language and a perfect passive knowledge, but suffered from severe memory gaps, especially in vocabulary, but also in more complicated areas of the grammatical system. My longitudinal studies in the Arvanitika speech community in Greece over a period of 25 years have revealed that these persons mostly develop from former fluent speakers who were on their way to becoming full speakers, but never reached that degree of competence due to the lack of regular communication in the language 2 . These individuals cannot be reckoned among the semi-speakers proper; they are simply "forgetters". Lise Menn coined the catchy term "rusty speakers" which I will use in the following. Rusty speakers may be found in situations of rapid shift, but also in communities characterized by gradual death. In the latter case such persons are probably born and raised in a period where the interruption of language transmission has not been fully operative - several families have begun to quit the language, others still transmit it. The other type of semi-speaker is a person whose command of the language is from the outset imperfect to a pathological degree. This is what we call the semi-speaker proper, the producer of the pathological speech forms we are examining here. The term "pathological" is chosen deliberately in order to express the similarity of this type of distorted speech to certain types of aphasie speech such as agrammatism (cf. Menn 1989). My claim is that

62

Hans-Jürgen Sasse

these semi-speakers result from the interruption of language transmission. The validity of this hypothesis was tested with positive results in an Arvantika village of Boeotia in 1978. All residences of this village clearly identifiable as semi-speakers claimed that their parents did not talk to them in Arvanitika and that they learned the language just by listening to and occasionally talking to elder fluent speakers. On the other hand, fluent speakers explicity stated that they were raised in an environment where Arvanitika was still regularly spoken and where adults with a positive attitude towards further language transmission existed. If one is allowed to generalize the striking results of the Arvanitika survey, there are strong indications that the linguistic input during language acquisition plays a crucial role and that the prototypical semi-speaker is a person who has not learned the language by way of a normal language acquisition process. Given this anomalous language learning process, it is quite clear that there will be differences in proficiency and language production due to several additional factors such as linguistic talent in general, degree of exposure, and the presence in the family of elders with an exceptionally good proficiency and a positive attitude 3 . Semi-speakers therefore form a continuum, a fact which has been referred to several times in the literature. The most detailed classification (though still extremely simplifying because it is not easily possible to quantify a continuum) is given in Campbell and Muntzel (1989). They distinguish between 'nearly fully competent', 'imperfect but reasonable fluent', 'weak' and 'rememberers' (1989: 181). Unfortunately, this classification captures only the degree of competence without specifying where the differences lie. In particular, it does not take account of the difference between "rusty speakers" and "semi-speakers proper". It is reasonable to assume that speakers found at the upper end of the continuum, the "nearly fully competent speakers", are typically "rusty speakers", while "imperfect/reasonable fluent" and "weak" speakers are typically semi-speakers proper. "Rememberers" may represent the terminal stage of both.

2.3. "Later loss" hypothesis vs. " i n c o m p l e t e acquisition" hypothesis The difference between rusty speakers and semi-speakers has been treated in the literature as evidence for competing hypotheses regarding language acquisition in language death situations. The "later loss" hypothesis produces different degrees of rusty speakers, whose competence vanishes because of the lack of regular linguistic experience: "complex linguistic structures that are acquired late in childhood will be lost since this is the very time when children in many communities stop fully using the obsolescent language"

Language decay and contact-induced change

63

(Gal 1989: 314 after Voegelin - Voegelin 1977). The "incomplete acquisition" hypothesis, on the other hand, which produces different degrees of semi-speakers, argues that the persons classified as imperfect speakers never learn these structures. Experience has shown, however, that both types of imperfect speakers often exist within the same speech community. This comes as no surprise since the "later loss" hypothesis and the "incomplete acquisiton hypothesis" seem to be mere variants of each other. The notion of incomplete acquisition has three aspects; 1) Those children who do not learn the language at all by way of language transmission strategies (e.g. children of progressive families) are able to learn the language only by chance. 2) Those children who grow up in a subpart of the speech community which still uses the language may be first socialized primarily in the obsolescent language, but shift to the target language in a later stage of language acquisition, with the result that the more elaborate linguistic skills in the obsolescent language are not learnt (i.e. these people stop somewhere in the middle of the natural language acquisition process). 3) Those children who are socialized completely in the obsolescent language may forget it later due to the lack of communication partners; i.e. the regular continuation of language acquisition by intersubjective communication during adulthood does not occur. In other words, the continuum among semi-speakers and between semi-speakers and rusty speakers may be viewed as a continuum between random acquisition as the one extreme, and varying degrees of "regular" acquisition. 4 In the following we will be mainly concerned with the linguistic output of random acquisition. As argued in Sasse (this volume), random acquisition is the consequence of the interruption of regular language transmission strategies. An infant growing up in a bilingual setting where one language is recessive will be confronted with the collective stigmatization of this language from a very early phase of language acquisition. Stigmatization is, as it were, an integral part of the bilingual language acquisition process. The child will not be taught the abandoned language by way of regular language transmission strategies: he will be discouraged to use it ("don't talk L j , L2 is better"); he will soon realize that he is more often addressed in the target language than in the abandoned language; he will realize that elders (e.g. his grandparents) are criticized for using language transmission strategies such as fairy-tales or nursery rhymes, and so on. 5 The result is that such children do not learn the language by way of a normal language acquisition process which gradually accumulates bits of language knowledge and successively arranges them in a system. Rather, what he learns of the language is learned in chunks. No complete linguistic system is established; linguistic competence remains confined to a closed list of short sentences of everyday use, formulaic

64

Hans-Jürgen Sasse

expressions, phrases, words and different forms whose functions are, for the most part, opaque. This means that linguistic creativity is restricted to putting these elements together in some way; the ability to create new utterances on the basis of principled grammatical patterns and rules is irreversibly lost. The prototypical semispeaker at the lower end of the continuum is unable to utter a narrative text or to carry on a longer conversation; for such purposes he will always switch to the target language. Passive competence is often considerably higher (Dorian 1981, passim), especially as far as a knowledge of the vocabulary is concerned, but obsolescent categories and complex syntactic structures are rarely fully understood. Nevertheless, the language is occasionally used. In the final phase where the obsolescent speech community consists nearly exclusively of semi-speakers and regular communication in the language stops, residue knowledge is used as a "phatic symbol of identification" (Tsitsipis). As a matter of fact, the remnants of the language are employed to signal in-group identity in songs, rhymes, jokes, toasts, obscene words and phrases, or occasionally as a secret language. On occasion a kind of pseudocommunication is possible with the help of stereotype phrases and formulaic expressions from a fixed stock learnt by hearth

3. Linguistic evidence for the distinction between borrowing and decay We will now consider the linguistic evidence for the distinction between borrowing and decay. In a normal contact situation, when a language is exposed to influence from another language, it tends to borrow not only vocabulary but also structure. We are not interested in the mechanism of borrowing here; what we are concerned with is the results. Due to the tendency towards compound bilingualism in a long-term contact situation, there is some kind of ideal goal (which may never be arrived at in effect) to develop a one-to-one relationship between the morphosyntactic systems of the two languages. In particular, there is a tendency towards an identical category system, and a tendency towards structurally similar means to mark these categories. For example, if prior to the contact situation one language distinguishes between a general present and a progressive and the other does not, either the one that made the distinction will tend to abandon it, or the one that did not make the distinction will develop it. Which of the two languages influences which is a

Language decay and contact-induced change

65

matter of prestige and power; it is also possible that both languages influence each other. Developing new categories and abandoning old ones are necessary parts of the same structural borrowing process. If necessary, we can make a distinction between positive borrowing and negative borrowing. Negative borrowing has nothing to do with reductional loss of essentials; it just follows the practice of the model language in dropping linguistic elements that the model language does not have. This is not confined to categories, for it can happen in other areas of the language system, too. If the replica language is of the flexional type and the model language isolating, the replica language will tend to lose its morphology. This does not mean, however, that the language becomes defective: loss of morphology will be compensated by imitating the means of the model language; i.e. the lost morphological categories will be replaced by categories marked by means of function words. If the replica language does not possess a sufficient repertory of function words, it will develop new ones by means of rapid grammaticalization or borrowing. These are indeed all quite well-known facts; hence I need only give a small number of striking examples.

3.1. E x a m p l e s of b o r r o w i n g A case study of extreme contact-induced change is presented in Thomason and Kaufman (1988: 215-222), based on Dawkins' 1916 study of Modern Greek dialects spoken in Asia Minor. Turkish interference, especially in the Cappadocian varieties of Asia Minor Greek, was so great that, in Dawkins' view, "the body has remained Greek, but the soul has become Turkish". The entire syntax was remodeled according to the Turkish pattern. Grammatical gender was nearly abandoned. The definiteness system was reorganized: the article is usually dropped, except for cases where it is used to indicate differential object marking, a category imported from Turkish. Periphrastic verb forms were introduced to match Turkish categories. The most striking feature of Cappadocian Greek was the general shift from flexional to agglutinative morphology. The dialect of Ulaghatsh, for example, adopted Turkish verb categories by imitation of the Turkish pattern using enclitic forms of the verb 'to be' as inflectional endings. Further, it developed a totally agglutinating noun morphology by generalizing some easily analyzable endings of one of the neuter paradigms, the following examples are drawn from my own 1968 field work with one of the last speakers of Ulaghatch in Athens, cf. also Kesisoglou (1951). They represent a slightly more progressed form of the dialect than Dawkins 1916. The paradigm of neuter nouns ending in -i was a particularly well-suited point of departure for agglutinative reinterpretation

66

Hans-Jürgen Sasse

of the case system. In the nominative and accusative the ending -i dropped by regular sound change; thus standard spit-i 'house' becomes spit. The genitive is regularly spityu, developed from earlier spit-i-u, with the regular genitive ending of the o-declension attached to the stem spiti- (the original nominative was spiti-on). Since the nominative/accusative now had a zero suffix, it looked exactly like a Turkish nominative/accusative form, e.g., ev 'house'. The genitive could now easily be interpreted as an agglutinating suffix and equated with the Turkish model:

NOM/ACC GEN

Turkish ev ev-in

Ulaghatch spit spit-yu

Once identified as an agglutinating suffix, -yu was detached from the neuter paradigm and employed as a general genitive ending everywhere. The genitive of neka 'woman', originally nekas, became neka-yu, the genitive of átropos 'man', originally atropu, become atropos-yu, etc. But analogy didn't stop here. Since the Turkish genitive suffix -in (with its vowel harmony variants -in/iin/un) is employed in the plural, too, -yu followed this pattern (which is scandalous for an Indoeuropean language). The plural nominative/ accusative of neka 'woman' is nekes. The genitive plural must originally have been something like nekon (cf. Standard Modern Greek yinekon). It now becomes nekes-yu. The story still goes on. In Turkish, the genitive suffix -in regularly follows the general plural suffix -lerAar. When Ulaghatch speakers found it necessary to have an equivalent of -ler/lar, the neuter paradigm again offered itself: the plural nominative/accusative of spit 'house' is spit-ya (regularly from spit-i-a), which again looks like an agglutinative mechanism zero vs. x. The ending -ya was thus reinterpreted as a general plural ending. The plural of átropos, formerly something like atropi, is now atropos-ya, etc. We have now arrived at an exact equivalent of the Turkish declension pattern:

SG NOM/ACC GEN PL NOM/ACC GEN

Turkish adam adam-in adam-lar adam-lar-in

Ulaghatch átropos atropos-yu atropos-ya atropos-ya-yu

Standard MG anthropos anthropu anthropi anthropon

In the speech of my informant only a few archaic declensional forms had been preserved (much fewer than in Dawkins' material), which may be an indication that regularization of the declension was a rapidly ongoing process in Ulaghatch prior to the expulsion of Greek speakers from Asia Minor.

Language decay and contact-induced change

67

Something similar to the events described for the development of noun inflection in Ulaghatch must have happened in the history of Armenian, though the details are less clear and the exact model is unknown. Armenian, probably in contact with Turkic and Caucasian languages, lost its gender and developed an agglutinating noun and verb morphology very similar to Turkic languages. Particularly striking is the similarity of development between Armenian and Ulaghatch noun morphology. Compare the following (indefinite) paradigms of Old Armenian (between 5th and 7th century A.D.), Modern Armenian, and Turkish: Modern West Old Armenian Armenian

Turkish

jori jorwoy jorwoy jori jorwoy jorwov jorik' joris jorwoç jorwoç joris jorwoç jorwovk'

jori jori-i jori-i

ester ester-in ester-e ester-de ester-den

gorci gorcwoy gorcwoy gorcwoj gorcwoje gorceaw gorcik' gore is gorceaç gorceaç gorcis gorceaç gorceavk'

gorcik' gorcik'-i gorcik'-i

'mule' SG NOM/ACC GEN DAT LOC ABL INSTR PL NOM ACC GEN DAT LOC ABL INSTR

-

jori-e jori-ov jori-ner jori-ner jori-neru jori-neru -

jori-ner-e jori-ner-ov

-

ester-ler ester-1er ester-ler-in ester-ler-e ester-ler-de ester-ler-den -

'tool' SG NOM/ACC GEN DAT LOC ABL INSTR PL NOM ACC GEN DAT LOC ABL INSTR

-

gorcik'-e gorcik'-ov gorcik'-ner gorcik'-ner gorcik'-ner-u gorcik '-ner-u -

gorcik'-ner-e gorcik'-ner-ov

alet alet-in alet-e alet-te alet-ten -

alet-ler alet-ler alet-ler-in alet-ler-e alet-ler-de alet-ler-den -

To add just a few more examples: Salar is a Turkic language which came into contact with Chinese and lost a considerable part of its morphology due

68

Hans-Jürgen Sasse

to imitation of the isolating pattern of Chinese (cf. Tenisev 1963). Arabic dialects of Southeastern Turkey tend to abandon the complex "broken plural" system characteristic of Arabic and generalize the feminine plural ending -at after the Turkish model of -ler/lar (Sasse 1971: 241). Moreover, these dialects have developed a clause-final copula as all other languages of the area have. These languages have also developed an entirely new system of verbal categories (aspect, tense, and mood) on the model of Turkish (Sasse 1971). Amharic developed a complex system of main and subordinate clause verb forms, converbs, etc., on the model of Agaw (Central Cushitic) languages. All these changes are contact-induced radical rearrangements of the typological make-up of the languages involved, which were achieved entirely by using native material without borrowing grammatical morphemes from the model language. Entire grammatical systems are totally altered; old categories that do not match are abandoned and new categories are built up. The whole process is characterized by impressive innovative creativity. If categories or distinctions are lost in the course of this process, it would be inappropriate to call this "simplification" or "reduction"; the abandonment of these distinctions is simply due to the fact that they are absent in the model language. We don't have to do with any kind of contraction here; none of the languages referred to above is threatened by language death.

3.2. Borrowing and decay: the Arvanitika case When Arvanitika, the Albanian dialect of Greece, came into close contact with Modern Greek by an increase of bilingualism in the middle of the 19th century, exactly the same thing happened. The verb lost two categories, the admirative and the optative, which were not present in Greek, and began to develop an imperfective/perfective distinction in certain subordinate clauses by a very clever exploitation of the subjunctive particle on the Greek model (formerly aspect distinction worked only in main clauses). The five cases of the declensional paradigm were reduced to three on the Greek model, and so on. All these changes were contact-induced and do not represent instances of irreparable loss, as they simply follow the Greek pattern. They involve both positive and negative borrowing, and those changes which we would consider negative, the loss of categories, did not affect the normal functional capacities of the language. They simply conformed to the tendency to render the category inventory similar to or identical with that of Greek. If Greek, which is a perfectly normal language, can do without such categories as the admirative and the optative, why shouldn't Arvanitika, as well?

Language decay and contact-induced change

69

At the beginning of the 20th century something happened which one certainly would have a right to call a substantial loss. There was a general breakdown of the native derivational system, which ceased to be productive within a very short period of time. This loss, however, did not affect the functional capacities of the language since it was compensated by the invention of a tricky and complex system of loan word integration, by which the entire Greek lexicon became open to Arvanitika (cf. Sasse 1991). Although the language became in no way abnormal, the total replacement of wordformation mechanisms by a loan word integration system is alarming as a possible sign of incipient language death, because the condition of having no derivational system and being forced to integrate Greek vocabulary for all innovative purposes heavily relies on the contact situation. In other words, after this step, the symbiosis became vitally necessary for Arvanitika. Denison (1979:34), who reports a similar case for the German dialect of Sauris in Northern Italy, regards this development as "system-retaining interference". It seems to be a typical concomitant of very intense contact situations with very uneven dominance proportions. In any case, the language still functions as a normal means of communication and is fully functional. In Arvanitika, the integrated loan words were fully assimilated and considered part of the language. Moreover, it was considered part of the linguistic competence of the full speaker to know which Greek items are to be used in Arvanitika and which Arvanitika lexemes should not be replaced by Greek ones. The terminal phase of language death was reached in the Arvanitika speaking communities not long after that, due to the interruption of language transmission. When Arvanitika speaking parents ceased transmitting the language to their offspring, a constantly increasing group of semi-speakers emerged. The language of these semi-speakers is totally different from the Arvanitika spoken up to this point. It begins to show pathological phenomena. The entire category system breaks down; important distinctions get lost. This reduction cannot be any longer attributed to negative borrowing as Greek possesses all the categories that Arvanitika is losing now. Many speakers use only a few verb forms, which are all employed for the same purpose. If past tense forms are remembered at all, they are always in the perfective aspect. This means that aspect distinction, which is of extreme importance in Greek and was even in the process of being elaborated in Arvanitika to conform to the Greek model, is now being abandoned. Verb morphology breaks down; the conjugational system, once operating with personal suffixes, is now replaced by whatever verb form the speaker remembers plus isolated pronouns serving the indication of personal reference. Articles are dropped or used

70

Hans-Jürgen Sasse

incorrectly, subjunctive and future particles are confused, gender agreement is not observed, and so on. In the following I will try to summarize and systematize the most important characteristics of semi-speaker speech using examples from Arvanitika. Further abundant material can be obtained from Tsitsipis (1981), Dorian's various publications on East Sutherland Gaelic, Dressier on Breton (1981 and elsewhere), Schmidt (1985) on Dyirbal, Schlieben-Lange (1976) on Occitan, and Campbell - Muntzel (1989) and the literature cited therein. 1) Loss of subordinative mechanisms Arvanitika semi-speakers do not use gerund forms of the verb in spite of the fact that there is an exactly parallel form in Greek. The most frequent type of subordinate clause is the short relative clause. Adverbial clauses are avoided except for those introduced by 'when' or ' i f . At the phrase level, modifiers are rare; genitives and adjectives are not frequently used. 2) Loss of systematic integration The elaborate integration system is lost. Greek words are not assimilated any longer to the morphological system of Arvanitika, but are cited, as it were, in their proper Greek form: ea ts soma tileorasi 'let us watch television' (Greek tileorasi 'television', should be integrated as tileoras; the full-speaker version of the sentence would be ea ta fohma tileoras)·, do vip eòe mihaÁi me to öiciyoro sot? 'will Michael, too, come today together with the lawyer?' (to öiciyoro, the proper Greek accusative form with the article, is used here instead of the traditional integrated Arvanitika form öiciyuarra definite accusative 'the lawyer'). A further aspect of these Greek citations in the Arvanitika frame is that fully inflected Greek words even occur in cases where good full speakers would use an Arvanitika lexeme: sehpsarya 'bring fish!' (the word for 'fish', pifk, pi. pife is never replaced in full speaker Arvanitika by the Greek psari, pi. psarya; a full speaker would say seh pife). This phenomenon is akin to code-switching but differs from code-switching proper in that it only pertains to single words at those places where former full speaker would have assimilated loan words.

3) Breakdown of grammatical categories The entire system of tense/aspect/mood categories is becoming mixed up. Semi-speakers often do not differentiate between the future particle do and

Language decay and contact-induced change

71

the subjunctive particle ta. They either confuse them or use a blend of both, approximately de or da. The tense system is usually reduced to the present and the aorist (perfective past). Imperfective past forms are hardly remembered by advanced semi-speakers. The perfect is used only in very prototypical cases (the "experiential" cases such as kam vatur Ί have been [somewhere]', etc.). The imperative is usually best preserved. As far as nominal categories are concerned, plural is often not formed, the singular form being used instead. Definite and indefinite forms are becoming confused. Case forms are mixed up, the accusative being used for the nominative, and vice versa. 4) Agrammatism Closely connected with the breakdown of grammatical categories is the total disintegration of the morphological system. Analogy goes in all directions, suppletive forms are not remembered, irregularities are not observed or are applied to forms where they do not belong. Verb forms are confused (3rd person for 1st person, imperative for present 3rd person, etc.). Nominal plural forms become drastically regularized, on the one hand, but on the other a bewildering tendency towards extension of irregularities is observed. The funniest phenomenon is phantasy morphology (something which sounds like an Arvanitika morpheme being attached to a form) or blends of any sort: Mata 'you (pi.) cry' (for regular kÁani, -ta being the suffix of 2nd pi. aorist), stapirara 'houses' (for regular ftopi, to which the mass noun plural ending -ara is attached and a meaningless -r- inserted), fÁerifl Ί was sleeping', a funny attempt to make the imperfectly remembered 1st person imperfect form of 'sleep' look more like a 1st person form by attaching a suffix -ip, taken from one of the subclasses of the present of the n- conjugation; etc. Syntactic rules are not observed either. The connector article system breaks down, agreement rules cease operating, prepositions occur freely with whatever case form comes to mind: ka martuar vajza mixaÁit 'he married Michael's daughter' (normally ka martuara vajzan e mixaÁit, nominative vajza instead of accusative vajzan(a), and the connector e of the genitive phrase is dropped), e vuri nda dzepi 'he put it in his pocket' (normally e vu nda dzep, with the regularized 3rd person vuri 'put' instead of irregular vu, and the definite nominative dzepi 'the pocket' after the preposition nda which demands the indefinite accusative dzep). Since the verb forms are hopelessly confused, personal reference is often unclear and is therefore disambiguated by independent personal pronouns: u jep ti 'I'll give it to you' (for regular ta jap; jep is 3rd person, the regular proclitic accusative form ta 'thee' is replaced by the independent nominative form and placed after the verb like a regu-

72

Hans-Jürgen Sasse

lar noun object; the confusion of person marking on the verb is somehow compensated by the use of the independent 1st person pronoun u). It is noteworthy that this extreme agrammatism occurs only in spontaneous utterances, when the speaker is forced to make creative use of the language. There are many formulaic "chunks" which are perfectly grammatical. For instance, a test in a Boeotian village in 1978 led to the strange result that nearly all speakers who had something like u jep ti for 'I'll give it to you' were able to produce the perfectly correct negative form s to jap Ί won't give it to you', perhaps because the latter was learned as an unanalyzed idiom.7

5) Word retrieval problems Semi-speakers usually have problems of word-finding. As soon as they are forced to transform thoughts into words for which no ready chunks are at hand, they will shift to Greek (perhaps with a comment "how do they call it, damn").

6) Extreme phonological variation and distortion A few examples may suffice here: Standard /Jtepi/ ~ /stupì/ ~ /stepi/ ~ /stipi/ 'house' Standard /tja ban/ ~ /tsimbin/ ~ /tsembin/ 'how are you' Standard /rricaXa ~ /ricil/ ~ /riceX/ ~ /riciX/ ~ /ricins/ 'tick' Standard /maereps/ ~ /majereps/ ~ /maireps/ ~ /majireps/ 'cook'

7) Phonological hypercorrection Sometimes hypercorrection occurs in order to make some sounds "sound more like Arvanitika". Some young people with a positive attitude, for example, pronounce traditional /h/ before /e/ and /i/ as x. The reason is that Greek doesn't have h, so that the allophony of /hJ was readjusted according to the Greek pattern which has a complementary distribution of χ before back vowels and ç before front vowels. Thus, a word like /hers/ 'time' became /çere/. Semi-speakers, attempting to imitate the "something different" which

Language decay and contact-induced change

73

they remember from the speech of elder full speakers, try to avoid the sound ç, but, being unable to pronounce h, they substitute x. 3.3. Decay in Kemant I will now compare some material from Kemant, an obsolescent Agaw language of Central Ethiopia, in order to show that the very same picture obtains in the semi-speaker variant of an African language. The material was elicited 20 years ago, and not from a perspective of the investigation of obsolescence. Nevertheless, it gives rise to some interesting issues with respect to certain very typical decay phenomena. There are at least 4 regions where Kemant is still spoken, three of which I visited on my Ethiopian field trip in 1971. The sociolinguistic situation differs from place to place. My main informants lived in and around Gondar, the provincial capital, where the dominant language (and the target language for Kemant) is Amharic. They were all in their sixties or seventies and usually had a good command of the language, although some of them had ceased to practice it and showed signs of "rustiness". I have little evidence for semi-speakers in that area, although I remember (impressionistically) that younger people of the village of Wälläqa (5 km north of G.) with whom I tried to communicate in Kemant, had a very limited knowledge of the language. My impression is that language decay is fairly radical in that region; shift from Kemant to Amharic is rapid and leaves little room for an extended decay phase, though there may exist some conservative strugglers whom I simply did not encounter during my superficial acquaintance with the speech community. I assume that the rapidity of shift in Gondar is to be attributed to the fact that the town, which has been a socio-political and economical center for centuries and the capital of the Abyssinian empire between the 1630s and 1850s has exerted a particularly strong influence during the recent increased Amharization and urbanization process connected with the reign of the last two Ethiopian emperors Menelik III and Haila Selassie I, who transformed Ethiopia into a more modern national state. The difference of language maintenance in the other two communities I visited is largely dependent on the degree of preservation of paganism. The Ethiopian Orthodox church ran two campaigns of enforced Amharization against the Kemant, one at the end of the past century, and a second one after the Italian occupation in the early 1940s. Unbaptized Kemant were threatened with loss of use rights to their land (Gamst 1969). The acculturation process progressed at a much faster rate in the eastern Kemant communities, with the effect that the politico-religious center of Kemantland shifted from the traditional area of Karkar to the more

74

Hans-Jürgen Sasse

conservative western area of Chilga. This was reflected in the language maintenance situation observable in 1971. In Karkar, language transmission was largely disrupted, resulting in a considerable number of semi-speakers, while the language was better retained in Chilga, where it was supported by a reinforced religious zealotism which was made possible by a decree which allowed the remnant group of adherents to the Kemant religion to follow their old beliefs and practices without hindrance (Gamst 1969: 122). Semi-speaker material was collected sporadically in both areas. The most striking deviation from "good" Kemant was the general decline of verbal inflection. A spectacular typological trait of Agaw languages is the presence of special morphophonemically complex verb paradigms for each type of subordinate clause with postposed conjunctions, as it were, fused to the personal endings. Moreover, each paradigm appears in 4 variants, perfective, imperfective, each affirmative and negative. This amounts to an enormous mass of verb forms (for Bilin, Palmer 1957 counts up to 10,000), which were still quite well-mastered by elder speakers with a good knowledge of Kemant. Semi-speakers, however, were unable to remember most of the subordinate forms and used the main clause paradigms instead. This is the more striking because the Amharic verbal system is a less complex imitation of the Agaw system which came about by intensive language contact (substratum influence) in the 13th century and thereafter. In other words, equivalents to Amharic categories existed, but they were abandoned due to the lack of normal linguistic skill on the part of the semi-speakers. The loss of the typical subordinate verb forms is a particularly clear example of a process which has been observed in other obsolescent speech communities (Arvanitika, Breton, Gaelic), viz. the general loss of strategies for constructing complex sentences, first observed for Nahuatl by Hill (1973). Loss of systematic integration with the code-switching-like effect as described for Arvanitika occurs in Kemant, too. One of the many striking examples is ansolawm fäf 'take the blanket away!', ansolawsn is the Amharic definite accusative of ansola, an Amharic noun meaning 'blanket', which is normally integrated into Kemant as a masculine noun of the α-declension, whose regular accusative is ansolâs. The same sample sentence was consequently rendered by full speakers as ansolôs f ä f . As in Arvanitika, a general breakdown of the category system was observed. In spite of the fact that Amharic provides good equivalents of the 8 Kemant cases, speakers still confuse the case forms. The "absolutive" and the "subject case" are mixed up, indefinite and definite accusative forms are not distinguished. Local case forms are seldom used. There is a general decline of the verb system in addition to the abandonment of subordinate forms:

Language decay and contact-induced change

75

aspect forms are confused and there is a tendency to substitute all negative paradigms by the negative relative forms, perhaps due to the fact that these are more transparent than the main clause forms. Exactly as in Arvanitika, imperatives are remembered best and sometimes used to substitute indicative forms. Agrammatism is common. Gender agreement rules are disregarded, plurals are formed incorrectly, and wrong suffixes are attached to verb forms. Hybrid forms such as d.9w-yä-n9kw-dn 'you (pi.) are speaking' for dsw-yä-nakw (with the plural -n- attached twice) are frequent. Phonological variation was observed especially with respect to the "unusual" sounds q, γ and χ which do not occur in Amharic. γ tends to become zero, while the other two tend to be equated with Amharic h.

4. Conclusions Now I shall briefly summarize the empirical evidence in favor of a distinction between language decay and heavy contact-induced change. The most important similarities between the two is that both involve loss of linguistic material. Theoretically, contact-induced loss can easily be distinguished from loss due to decay in that the former is motivated by the absence of the respective categories in the contact language, while decay involves loss of categories not motivated in this way. There is of course a practical problem which cannot be resolved easily, namely the fact that decay may occasionally pertain to types of reductions which could, in principle, also be attributed to contact: where the obsolescent language possessed a particular category prior to the decay phase and the target language did not, its loss during the decay phase can be explained in either way. Nevertheless, I think this is a minor problem since the bulk of typical decay phenomena, especially agrammatism, syntactic reduction, and extreme variability, is so different from what happens in normal contact-induced change, that it can clearly be set off from the latter. It is also clear that the difference between language decay and heavily contact-influenced but functionally intact language reflects the difference between two types of language users - the semi-speaker and the "normal" bilingual individual - , and that this difference is to be explained in terms of the different processes of language acquisition involved. A language acquired in the usual way, transmitted according to norms established in the speech community, tested in everyday conversation a whole life long, be it

76

Hans-Jürgen Sasse

as "mixed" as one can imagine (but languages even in extreme contact situations are normally mixed only in a certain sense, as I have tried to show), will have to exhibit a certain degree of stability and will have to conform to certain minimal requirements of a functionally intact language. That is, it will have relatively stable mechanisms of constructing utterances, it will have a paragraph structure, a clause structure, etc., and the appropriate morphosyntactic apparatus necessary to serve them. A language acquired only by chance, however, the competence of which is limited to stereotype-fixed expressions and a small stock of ready-made clauses and isolated word forms will necessarily not provide an adequate means for fluent communication. A category system is virtually non-existent, and the minimal functional requirements of discourse (relative stability, techniques for combining slots with filler categories, etc.) can no longer be fulfilled. This is the pathological situation in which the prototypical semi-speaker finds himself. In conclusion I would like to point out some practical consequences of the distinction advocated here. Differentiating loss of categories in a normal process of assimilation to a model structure from the pathological situation of language decay is essential for the evaluation of data elicited from last generation speakers in a language death situation, and it therefore constitutes a fundamental question of documentation in such a situation. Dimmendaal (this volume) gives us a nice example of the defective knowledge of Kore, where this problem crucially arises. How reliable is the speech of the last Kore speakers and how much does it reveal of the original structure? I was confronted with this problem several years ago when I received Bernd Heine's Elmolo material and attempted to apply it for comparative purposes. I was impressed by the fact that the verbal paradigms were so deviant from those of the neighboring and related Arbore. Moreover, they are inconsistent in themselves: some verbs have only one form for all persons, others distinguish two or three (not always the same!), still others show the whole array of forms normally found in Cushitic languages. Even worse, the same verb reduces all person forms to the 3rd masculine singular in one tense, and retains them in another! (cf. Heine 1980: 89 and 193). In general, Elmolo and Arbore seem to be so closely related that they can be considered dialects of each other. But comparison of Hay ward's (1984) description of Arbore with Heine's (1980) Elmolo data renders Elmolo a broken-down form of Arbore. Disregarding the fact that Hay ward's material is much more exhaustive and more carefully elicited, the impression remains that Elmolo is much "poorer" than Arbore. For example, the pronominal system of Elmolo is an abridged form ofthat of Arbore (cf. Hay ward 1984: 215 and Heine 1980:187). Of six functionally differentiated series only two remained, whose functions

Language decay and contact-induced change

77

have already begun to be blurred, too. Gender distinction in the third person was abandoned, and the form itself is being replaced by the "primitive" circumlocution 'this person'. There is virtually no trace of the case system of Arbore (Hayward 1984: 131 ff.), which distinguished at least four cases with complex distribution. Subject and object are not distinguished, the predicative form of the noun seems to have been abandoned, and the complex genitive construction, which was both head-marking and dependent-marking (Hayward 1984:150 ff.), was replaced by simple juxtaposition of possessed and possessor with or without a linking particle (Heine 1980:184). Further, while Hay ward (1984: 300-325) reports a relatively elaborate system of forming complex sentences, involving infinitives, gerunds, a considerable number of constituent clause types and coordinative clause combining strategies involving the subjunctive, Heine states that Elmolo has "only few morphological devices to mark embedded constructions. In many cases, embedded clauses which follow their governing constituents are introduced without any formal linkage" (Heine 1980: 200). Drastic functional reduction, agrammatism, reduction of subordinative mechanisms are all well-known symptoms of language decay. We found them with semi-speakers of Arvanitika, Gaelic, Breton, Dyirbal, Nahuatl, Kemant, and many more obsolescent languages. The conclusion is that the last Elmolo speakers were semi-speakers and Heine's Elmolo material does not represent "original Elmolo" but a pathologically distorted version thereof. For comparison with other Cushitic languages it must therefore be handled with great care. I am convinced that much of the material elicited in obsolescent speech communities in Africa will be of this kind, and I think the pathological nature of this kind of material should be taken into account especially when drawing historical and comparative conclusions.

Notes 1. I wish to thank Nancy Dorian, Wolfgang Dressier, and Lukas Tsitsipis for valuable comments on an earlier version of this paper. 2. These statements may be somewhat oversimplified on the basis of my personal experience with Arvanitika. Dorian (p.c.) arrives at a four-term distinction for East Sutherland Gaelic. She recognizes (according to different levels of proficiency): fluent speakers, formerly-fluent speakers, and semi-speakers who have two different kinds of acquisitional history but similar outcomes in terms of proficiency: 1. normal childhood acquisition followed by cessation of further acquisition at school entry age, with attrition thereafter; and 2. abnormal acquisition, never nor-

78

Hans-Jürgen Sasse

mal for any development stage, but sometimes with some degree of improvement later in life. The linguistic characteristics of a formerly-fluent speaker are discussed in Dorian (1982). As a matter of fact, we are mostly concerned in the following with the linguistic output of abnormal acquisition. Cf. also note 4. 3. I know of several cases where semi-speakers who were clearly at the near-tofull-speaker end of the continuum reported to have had in their family persons sticking to the language with unusal vigor. In one case there was an elderly aunt, monolingual in Arvanitika, said to be very conservative, who forced children to address her in Arvanitika. In another case, the grandmother was a narrative talent and told many fairy-tales to the children, which explains the relative linguistic skill of that particular informant. 4. The existence of later loss side by side with incomplete acquisition may in fact result in two intersecting continua of language proficiency, in which Dorian's four types of speakers (cf. note 2) represent four prototypical instances: a continuum of later loss cutting across a continuum of incomplete acquisition. In order to get a clearer picture, future studies of imperfect speakers in language death situations should carefully examine the acquisition histories of the informants. 5. "I used to tell him fairy-tales in Arvanitika, but his mother didn't like that". (Grandmother, about 75, in an interview in a Boeotian village) 6. For example, an Arvantika semi-speaker A may say to his companion: ea ta ρ ima njapotir vera 'Come, let's have a glass of wine!'. Β may answer: ta ham tapim ta YÁendisajma 'Let's eat and drink and have f u n ! ' . C joins in: takatsejm 'Let's dance!'. D completes: tabanemi tapa 'Let's get tight!'. And so on. These are all perfectly well constructed Arvanitika forms, but the speakers would probably not be able to construct a paradigm of 'dance'. Pseudocommunication can be retained over a considerable stretch of time, without any information being conveyed. It just serves the purpose of group identification, speakers simply have fun in documenting their Arvanitism to each other. 7. There may be differences in the degree of distortion observable in the morphology of semi-speakers in different obsolescent communities. Dorian's results from East Sutherland Gaelic do not seem to be quite as messy as my Arvanitika ones. She also points to the fact that although individual variation may be extremely high, speakers tend to be systematic within their own usage over many years. That is, "the chaos is there across the group, but not so much on an individual basis" (Dorian, p.c.).

References Campbell, Lyle - Martha C. Muntzel 1989

"The structural consequences of language death", in: Nancy Dorian (ed.) 181-196.

Language decay and contact-induced change

79

Dawkins, R.M. 1916 Modem Greek in Asia Minor: A study of the dialects of Siili, Cappadocia and Phárasa with grammars, texts, translations, and glossary. Cambridge: Cambridge University Press. Denison, Norman 1979 "Zur Triglossie in der Jahre", in: P.S. Ureland (ed), Standardsprache und Dialekte in mehrsprachigen Gebieten Europas. Tübingen: Niemeyer, 2737. Dorian, Nancy C. 1981 Language death. The life cycle of a Scottish Gaelic dialect. Philadelphia: University of Pennsylvania Press. 1982 "Language loss and maintenance in language contact situations", in: Richard D. Lambert - Barbara F. Freed (eds.), The loss of language skills. Rowley, Mass.: Newbury House, 44-59. Dorian, Nancy (ed.) 1989 Investigating obsolescence. Cambridge: Cambridge University Press. Dressier, Wolfgang U. 1981 "Language shift and language death - a protean challenge for the linguist". Folia Linguistica 15: 5-27. Gal, Susan 1989 "Lexical innovation and loss: The use and value of restricted Hungarian", in: Nancy Dorian (ed.), 313-331. Gamst, Frederick C. 1969 The Qemant: A Pagan-Hebraic peasantry of Ethopia. New York: Holt, Rinehart and Winston. Hamers, Josiane F. - Michel H.A. Blanc 1989 Bilinguality and bilingualism. Cambridge: Cambridge University Press. Hayward, Dick 1984 The Arbore language: A first investigation. (Cushitic Language Studies 2). Hamburg: Helmut Buske. Heine, Bemd 1980 The Non-Bantu languages of Kenya. (Language and dialect atlas of Kenya 2). Berlin: Dietrich Reimer. Hill, Jane H. 1973 "Subordinate clause density and language function", in: C. Coram T.C. Smith-Stark - A. Weiser (eds.), You take the high node and I'll take the low node: Papers from the Comparative Syntax Festival. Chicago: Chicago Linguistic Society, 33-52.

80

Hans-Jürgen Sasse

Kesisoglou, I.I. 1951 To glossiko idioma tou Oulagats. Athens: Collection de l'Institut Français d'Athènes. Menn, Lise 1989 "Some people who don't talk right: Universal and particular in child language, aphasia, and language obsolescence", in: Nancy Dorian (ed.), 335-345. Palmer, Frank 1957 "The verb in Bilin. Bulletin of the School of Oriental and African Studies 19: 131-59. Romaine, Suzanne 1989 Bilingualism. Oxford: Basil Blackwell. Sasse, Hans-Jürgen 1971 Der arabische Dialekt der Mhallamiye in der Provinz Mardin (SüdostTürkei). [Unpublished Doctoral Dissertation, Munich.] 1991 Arvanitika. Die albanischen Sprachreste in Griechenland. Teil 1: Die Mundart Nordost-Attikoböotiens. Wiesbaden: Harrassowitz. Schlieben-Lange, Brigitte 1976 "A propos de la mort des langues", in: Atti del XIV Congresso Internazionale di Linguistica e Filologia Romanza, Napoli 1974, Amsterdam: Benjamins, 381-388. Schmidt, Annette 1985 Young people's Dyirbal: An example of language death from Australia. Cambridge: Cambridge University Press. Tenisev, E.R. 1963 Salarskij jazyk. Moskva: Izdatel'stvo vostoinoj literatury. Thomason, Sarah Grey - Terence Kaufman 1988 Language contact, creolization, and genetic linguistics. Berkeley: University of California Press. Tsitsipis, Lukas D. 1981 Language change and language death in Albanian speech communities in Greece: A sociolinguistic study. [Unpublished Ph.D.Diss., University of Wisconsin, Madison.] Voegelin, C.F. - F.M. Voegelin 1977 "Is Tübatulabal de-acquisition relevant to theories of language acquisition?", International Journal of American Linguistics 43: 333-336. Weinreich, Uriel 1967 Languages in contact. The Hague - Paris: Mouton.

Part II

East African case studies Matthias

Brenzinger

Introduction In Part II we present case studies from East Africa. The articles written by Batibo and Legère analyse the situation of minority languages in Tanzania, both pointing out the role which the national language, Swahili, plays in this country. The structural consequences of language shift and language death are dealt with by Dimmendaal (Kore), Tosco (Dahalo), Möhlig (Chifundi/Chwaka); and Nurse-Walsh (Chifundi/Chivumba). Specific case studies analysed from a sociolinguistic perspective are presented by Brenzinger (Yaaku/Elmolo), Heine (Terik) and Rottland-Okombo (Suba). Winter takes a historical perspective to describe the language shift of Gweno. The locations in which the languages under consideration are, or were spoken, are shown on Map 1 (on next page).

84

Matthias Brenzinger

Map 1. East African languages under consideration

The fate of ethnic languages in Tanzania Herman Batibo

1. Introduction Although Darwin's theory was aimed at living species, languages too, spoken over limited space under specific conditions, can be said to be in a state of struggle, and therefore only the fittest will survive. The small or weaker languages will have to give way to the larger or more powerful ones. This analogy, although not wholly valid, 1 could be applied to Tanzania where 126 (according to Molnos 1969) languages are spoken over areas which are often overlapping. This conflict of interest and allegiance, mostly determined by socio-economic and cultural factors, has been the cause of linguistic expansion, reduction or even death of many of these languages. In principle, all language speakers would like to preserve not only their linguistic but also their cultural and ethnic identity. In this paper, I intend to discuss the fate of the more than 120 ethnic languages 2 and dialects which had made Tanzania the only country in Africa to have all the four language families described by Greenberg (1963). I have decided to embark on a more general discussion of the threatened languages rather than treat one or few individual languages. The obvious reason is that, as rightly pointed out by some Tanzanian colleagues in Cologne (Sommer, this volume), all ethnic languages in Tanzania "are dying out in all parts of (the) country". A more general treatment of the problem is therefore more appropiate. I will first discuss some of the causes of language shift in Tanzania and then look into the stages or phases which the Tanzanian ethnic languages seem to be going through on their way to extinction. In the absence of objective data and statistics (as no comprehensive information or statistics on language and ethnicity in Tanzania have been compiled since 1957), many of the statements will be speculative or tentative.

86

Herman Β atibo

2. Major causes of language shift in Tanzania Sommer (this volume) suggests a number of possible causes of language shift such as economic change, size of speech community, urbanization and relative degree of language prestige. It is the last factor which seems to have been the major determinant of language shift in Tanzania. The hierarchy of language prestige is represented in: 1. 2. 3. 4.

Language Language Language Language

of National Prestige of Regional Prestige of Local Prestige without Special Prestige

The above diagram indicates the levels of prestige from a language without any special prestige (4) to a language of national prestige (1). The dominance pattern should not be seen as successive but as directional. The only Tanzanian language that has enjoyed national prestige is Swahili, 3 which was declared the sole national language of Tanzania immediately after independence. However, its prestige increased rapidly in 1967 when its role and importance were reinforced by a number of national events which occurred that year. These events included the famous Arusha Declaration, the creation of BAKITA (the National Swahili Council), the introduction of Swahili language as Primary School Medium, the enhancement of Swahili as official language, 4 the creation of a task force to prepare the use of Swahili in secondary schools, the use of Swahili in all literacy compaigns etc. Such enhancement had the following effects: (i)

(ii) (iii)

(iv)

(v)

The earlier somehow, negative up-country attitude towards Swahili (thought to be a propagator of coastal culture) was reversed. It was now seen more as a functional language rather than a cultural language (Batibo 1987). Swahili began to be used more extensively as a modern inter-ethnic means of communication. Thus it went deep into the grassroots. The domains, where Swahili was used, expanded, naturally at the expense of the ethnic languages. In many cases, the ethnic languages were restricted to only family and village level communication. Many Tanzanians became bilingual in Swahili. Their competence and use of Swahili varied with education, urbanization, mobility, religion, age, sex, literacy, ethnic size and proximity to the coast. While during the colonial times, English was the language of social promotion, Swahili assumed that role. In addition to that, it became

The fate of ethnic languages in Tanzania

87

associated with modernization (education, mass media, business, trade, administration, law, politics, social services etc.), nationalism and membership to a wider world. At regional level, certain languages were considered more prestigious than others. Such status was associated with history, cultural or military superiority, possession of wealth such as cattle, possession of technology or art such as iron-making or pottery, or demographic predominance. The languages which have been considered prestigious due to, at least, one of these factors include Maasai, Sukuma, Gogo, Chagga, Haya, Nyakyusa, Hehe, Ngoni, Makonde, Nyamwezi etc. Languages of local prestige were those languages which had no regional status, but because of their speakers' dynamism, had tended to dominate their immediate neighbours. Typical examples of such languages in Tanzania are the Matumbi and Luo whose history has been that of socio-political dynamism. One clear indicator of regional or local prestige in language is the degree to which its users still want their children to learn it, especially in urban or modern settings. At the University of Dar es Salaam, for example, Zinza or Pogolo children learn Swahili without parental concern, but Haya or Hangaza children would necessarily ensure that their children learn their ethnic languages first. Urbanization is a factor which has tended to favour Swahili as a lingua franca. With the rapid expansion of urbanism (from 5% in 1967 to 19% in 1988), 5 more people are apt to lose their ethnic languages in favour of Swahili. Economic change has been a valid factor of language shift in many traditional hunting and gatherers societies where such groups were forced to depend on agriculturalists and pastoralists when ecological conditions were no longer favourable to their traditional ways of life. There are, however, other factors which have been responsible, to varying degrees of importance, for language shift in Tanzania. Demographic pressure and eventual ethnic movement has been an important factor in language domination and eventual "suffocation". The Maasai and Sukuma expansions due to search for new grounds to graze their cattle and cultivate (in the case of the latter) have caused a number of deaths or near deaths to languages in a number of ethnic groups. The Maasai, believing to have indisputable right to graze their cattle in any area that they may choose, have roamed in all directions from the north of the country. Where they met smaller groups, they subjected them to their language and/or way

88

Herman Batibo

of life. The Tanzanian languages which seem to have suffered much from the hands of the Maasai include Akie, Sonjo, Kwasi and Asax. Cases of the strong assimilating power of the Maasai have been cited elsewhere, e.g. Kore (Dimmendaal, this volume) and Yaaku (Brenzinger, this volume). The Sukuma expansion as a pastoralist and agriculturalist group started in the early 1930s, when many Sukuma people, as a result of demographic pressure and search for more fertile land and grazing grounds, crossed the Busisi Gulf to go and settle in Geita and Biharamulo districts which, up to that time, were inhabited by Zinza and Subi respectively. Although many squabbles were reported between the invaders and the invaded, the situation had to be accepted as a "fait accompli". At present, both districts have, by far, more Sukuma speakers than the original groups. The effects of this situation is that not only are most Zinza and Subi people bilingual in Sukuma, but also, for some, it is their dominant or only language. Some Zinza speakers reported to me that Zinza children, in some parts, spoke only Sukuma. The Zinza language has tended to persist, somehow, on the Kome Island. With the current impact of Swahili, the linguistic situation in this area has become more complex. As the Sukuma expand south and southwest, their demographic and cultural impact are associated with linguistic influence. The other ethnic languages which are currently suffering (although slightly) from such impact are the Sumbwa, some dialects of Nyamwezi, Nilyamba and Shashi. The Sukuma are reported to have moved as far south as Mbeya and even Zambia, in their search for extensive land for cultivation and grazing. The other large ethnic languages in Tanzania which seem to have had sociolinguistic impact on their neighbours are the Nyamwezi (on Kimbu, Konongo and Tongwe), Makonde (on Mwera, Machinga) Shambala (Asu, Bondei), Luo (Suba, Sweta, Sizaki, Kurya, Surwa), Chagga (Ongamo, Arusha), Gogo (Kwadza), Jita (Kerewe), Haya (Nyambo), Nyakyusa (Ndali, Safwa, Niha), Vumba (Segeju, Digo, Chifunzi). 6 However, demographic importance, although often linked to regional or local prestige and wider communication, is not, in itself, a guarantee of language maintenance and/or expansion. For example, Zaramo (Mwaruka 1965; Legère, this volume) and Bondei (Mochiwa 1979) are languages in point. Both of them are "spoken" (i.e. have ethnic membership) by over 200,000 people, but are fast giving way to the Swahili language. Moreover, dialectal homogeneity is important for linguistic preservation. Where dialectal forms are unintelligible to each other, as in the case of Chagga, the speakers have resolved to a lingua franca. This means more substantial use of a "go-between" language, at the expense of the ethnic language. Chagga seems to be the only such case in Tanzania.

The fate of ethnic languages in Tanzania

89

Finally, the acceptability of the minority group is an important factor. It is not automatic that given two languages, under identical conditions, they will both give up their languages in favour of another language; Tosco (this volume) notes that while the Dahalo have given up their language in favour of Swahili, the Boni have not. 7 There must, therefore, be some internal motivation in the group. In Tanzania, much of the "Suicide death" (Denison 1977) was a result of factors such as economic needs, natural catastrophies, acknowledgement of neighbour's prestige or a mere strategy of integration. The earliest such incidents involved break away groups which went to settle among other groups. Itandala (1983) reports a number of linguistic groups as having assimilated themselves among the Sukuma from other language communities. These groups, now known as clans and identified only in the complex Sukuma greeting system, 8 have kept their ethnonyms such as Kamba (from Kenyan Kamba), Binza (from Toro/Nyoro), Lega and Shi (apparently from Zaire), Silanga (from Kerewe Island) etc. Equally many hunting groups, in the past, gave up their languages in favour of the languages of the people with whom they were in symbiotic relationship. The only reported cases we know are those of the Bahi (from Hadza) who are reported to have assimilated themselves to the Nyamwezi/Sukuma 9 and a "Dorobo" hunting group in the Ngulu mountains which is reported to have given up its original language in favour of Ngulu. 10 In current years, a fast assimilation process seems to be going on along the coastal area where speakers of ethnic languages, like Zaramo, Rufiji, Dendereko, Digo, Doe and Kwere, would happily call themselves "Waswahili" 11 in their desire to identify themselves with the coastal culture. This cultural allegiance must have an impact on language shift.

3. Phases of language shift in Tanzania Having discussed the major causes of language shift in Tanzania, I will now look at the phases through which the Tanzanian languages seem to be going. This description is difficult, not only because there is no total homogeneity in language behaviour and competence in any one ethnic community but also because the whole question of language shift can be described both diachronically and synchronically. Moreover, as Dressier (1972) rightly pointed out, language shift is ambiguous since it can refer to linguistic as well as sociolinguistic phenomena. I will, however, side with Dorian (1977), that the two phenomena tend to go hand in hand. I will make a synchronic description of the state of the Tanzanian languages by considering the five phases

90

Herman Batibo

which they are expected to follow on their way to extinction. Such phases should be taken as points on a continuum rather than well defined stages. Another problem is that some languages may be mentioned as belonging to one phase when actually they belong to another. This is because of unreliable and sometimes contradicting information on the state of the ethnic languages, especially since language and ethnicity often do not coincide (Dimmendaal 1989; Batibo 1988). (a) Phase I: ethnic language monolingualism Although all the ethnic languages in Tanzania are assumed to have developed as single languages serving specific ethnic groups, there is presently no true monolingual community in the country. There are only two categories of ethnic communities which would, on the whole, be said to be relatively monolingual. The two categories are (i) the large up-country languages like Sukuma, Haya, Gogo, Ha, Hehe, Nyakyusa, Bena, Nilyamba, Fipa, etc., (ii) the small but geographically remote languages like Hangaza, Kuria, Isanzu and Tatog. Even in this case, one has to recognize variations in the degree of monolingualism/bilingualism due to age, sex, education, mobility, religion, work, urban influence, and degrees of influence from neighbouring languages or dialects. On the Tanzanian map, the languages, in this phase, are still the least threatened by language death. Only a few members may lose their linguistic competence through re-settlement, intermarriage or urban influence. (b) Phase II: bilingualism with ethnic language predominance This is the phase in which the majority of Tanzanian ethnic groups are presumed to be. They use their ethnic languages in all family, village and intraethnic activities, but use another language as a language of wider or specific communication. Although the second language is often Swahili, any of the other regional or local languages may have this role. These communities have had to learn the second language (L2) alongside their ethnic language (LI). They are therefore bilingual. However, their use of L2 has been limited to specific functions such as trade or wider communication. This is a clear case of diglossia. Normally, each variety has a well defined domain. Cases of codes witching, interferences and linguistic erosion12 of LI forms are minimal at this stage. The Tanzanian linguistic groups which are in this phase include many of the small and medium-sized up-country linguistic groups such as Shambala,

The fate of ethnic languages in Tanzania

91

Yao, Pare, Makua, Ngindo, Langi, Kagulu, Pogolo, Pangwa, Safwa, Nyiha, Kwaya, Kimbu, Sangu etc. Although the ethnic languages in this category are closer to extinction than the ones in phase I, they are still enjoying full competence by their users in most stylistic and situational contexts. The speakers are still heavily attached to them. (c) Phase III: bilingualism with L 2 predominating In a number of cases, the range of domains in which L2 has been put to use has expanded, usually at the expense of LI, the extensive use of L2 has not only made it the dominant language but also the form in which the speakers had more confidence. The situation has become both bilingual and diglossie (Ferguson 1959). A lot of code-switching, code-mixing and extensive borrowing 13 from L2 would be expected when members of the community speak LI. The linguistic groups in this phase comprise many of the languages in which the predominance of Swahili or another regional language as a lingua franca is heavily felt. Such languages include Dendereko, Matumbi, Digo, Kami, Kutu, Rufiji, Zinza, Sumbwa, Bondei, Ngulu, Jiji etc. The extensive use of Swahili or regional language has tended to restrict the use of the ethnic languages to family and cultural performances (Polomé 1980). Moreover, some Swahili dialects seem to be heavily influenced by Standard and Urban Zanzibar varieties. The most affected dialects are Kimtangata, 14 south Pemba 15 and Kihadimu (Abubakar 1970). One has to distinguish, however, between Kikae (the earlier speech in Makunduchi area which has almost died), and the dialectal variety of Hadimu and Makunduchi (Batibo 1989). (d) Phase IV: restricted use/competence in L I In a few speech communities, the functions of LI have been so much reduced that the people use LI forms only in specific situations like initiation ceremonies, secret rituals, or folkloric performances. These ethnic communities have therefore lost their original competence in LI or have not taken the trouble to learn it properly. Only a few old men, and especially women, may still be able to use the linguistic forms as originally used. The members of the communities, of course, may not be consciously aware of this fact. In fact, their competence in LI may have suffered not only substantial reduction in its stylistic expression but also significant simplification in its phonological

92

Herman Batibo

system (Dressler 1972) and heavy reduction in its morphology. Moreover, all irregular forms may have been regularized and simplified. The syntactic rules may have been simplified and made more general. The lexicon may have been heavily reduced and may suffer many intrusions from L2. This stage has been known elsewhere as pidginization (Dimmendaal 1989). The linguistic groups which seem to have reached this stage 16 are: (i)

Suba, Sweta and Surwa, under the influence of Luo (Rottland - Okombo 1986). (ii) Ongamo, under the influence of Chagga (Heine - Vossen, 1975). (iii) Kwadza, under the influence of Gogo (Ehret 1980). (iv) Okuma/Arusha, under the influence of Chagga (Winter 1980). (v) Segeju, under the influence of Digo (Nurse - Walsh, this volume). (vi) Kahe, under the influence of Chagga/Pare. 17 (vii) Akie, under the influence of Maasai. 18 (viii) Zaramo, Kwere, Doe and Rufiji, under the influence of Swahili. 19 (ix) Alagwa, under the influence of Langi (Maarten Mous, p.c.). (χ) Mbungwe, under the influence of Maasai. 20 It is rather amazing that there is a clear correspondence between the pidginization process affecting the languages in this group and the cultural erosion or ethnic affiliation. Many coastal people, for example, are ambivalent about their linguistic, cultural and ethnic affiliation and allegiance. Depending on given circumstances they would be known by their narrower ethnic identity (e.g. Mkwere) or their wider cultural identity (e.g. Mswahili). 21

(e) Phase V: LI as substratum In very few Tanzanian ethnic communities, the predominance of L2 has been so high that LI has been completely replaced. LI has therefore died. However, these ethnic communities may have kept their ethnonyms and some of their traditions. Often some of the linguistic characteristics of LI may have remained as underlays or residual features underlaying L2. Such substratum phenomena may have involved prosodie, phonetic, phonological, semantic or lexical features. In some cases LI may have disappeared without leaving any linguistic traces. 22 Unfortunately, due to contradicting data, it is not certain which of the Tanzanian languages have really reached this phase. The problem is that it is difficult to decide between phase IV and V since it is not easy to ascertain whether all the members of a given language group have disappeared

The fate of ethnic languages in Tanzania

93

completely. The languages which have been described as dead (Sommer, this volume) are: (i)

Yeke, a language reported by M. Mann and D. Dalby in their Thesaurus (1987) as extinct. No further information is given. (ii) Hamba, according to R. Moser (p.c.), is a language which was absorbed by Mwera, Ndonde, Ndendeule and Ngindo. (iii) Asax, according to Winter (1976) and Ehret (1980), became extinct when the last speaker died in 1976. However, according D. Nurse (cited by Sommer, this volume), some old speakers still exist. (iv) Bahi, a language related to Hadza, is reported to have been absorbed by Nyamwezi/Sukuma. (v) Kikae, originally spoken in Makunduchi, Zanzibar. Pure Kikae hardly exists any more. It has however left its traces in the Makunduchi/Hadimu variety, especially in the prosody, phonology and morphology as an underlay. Other cases of language extinction took place many years ago when groups moved and integrated into other groups like the case mentioned earlier of the many groups which settled into Sukumaland and were linguistically absorbed. 23

4. Conclusion As it can be seen from the foregoing, although the regional and local languages have played an important role in the elimination of other languages, it is the rise of Swahili, as a national language, which has constituted the major threat. This is because Swahili has all the necessary charms like prestige, modernity, lingua franca, social promotion and wider acceptance. As one Swahili scholar remarked recently in Dar es Salaam, "Swahili is like a Nile perch, it is a delicacy, but it is swallowing all the smaller fish on its way." 2 4 The politicians, however, do not see this as a problem. According to the then President of Tanzania, Mwalimu Julius Nyerere, in his speech to a Symposium of the Association of Swahili Poets and Writers (UKUTA), in 1984 he stated that one of the effects of the growth and expansion of Swahili was the eventual disappearance of the ethnic languages. According to him, that was "a natural and unregrettable development, since any good move has its side effects". 25 Nyerere's statement was a true reflection of the country's linguistic policy which, although has not been explicit on the fate of the ethnic languages, has tended to treat them as marginal, or even as nonexistent. In

94

Herman Β atibo

the proposed language policy document, prepared by the Ministry of Labour, Culture and Social Welfare, 26 the ethnic languages (Lugha za Jamii) are being recognized in so far as they are sources of cultural expression, historical information, linguistic forms for Swahili language enrichment, and a basis for folkloric performances. Evidently, the decision against any inclusion of the ethnic languages in education, literacy campaigns, and wider village life, has tended to minimize their role in modern Tanzania. However, there are some tendencies, in present-day Tanzania, which might slow down (although may not stop completely) the trend of affairs. First, there is a growing regional consciousness among people of both Zanzibar and Mainland Tanzania. Many regional groups and associations are being formed, nearly always based on ethnic identities (although often camouflaged under geographical or administrative designations). This consciousness although aimed at integrated efforts to develop their home areas, can also serve as a basis for ethnic and linguistic preservation. Second, the two moves by the government, namely to decentralize administration to the regions and to liberalize the national economic life, have created some form of autonomy and flexibility in the regions and districts. These developments have helped to make people more local centred, and therefore, indirectly, bring support to ethnic and even linguistic maintenance. 27 Third, as it has already been experienced elsewhere, especially in Spain and France, the outcry for national unity and identity often overrides regional differences when "the iron is still hot". But once things cool down, people realize that they were being nationalistic at the expense of their own local identity. Ethnic consciousness is therefore bound to be stronger in future. Even if the current trend is slowed down, the already dying patients will not be saved. This is exactly one of the reasons why the linguistic atlas project for Tanzania was conceived some years ago. What has been decided now is that we should not waste time making beautiful maps on the distribution of the languages or trying to trace isoglosses marking the boundaries between languages or dialects, 28 rather we should concentrate on thorough data collection, efficient data storage system and then massive description of these languages not only for academic ends, but also (and even more so, for Tanzania) for historical, cultural and linguistic enrichment.

Notes 1. The snag about this analogy is that while species, in this context, could be said to be mutually exclusive, languages are not; only one language may be redundant.

The fate of ethnic languages in Tanzania

95

2. In this study, the term "ethnic language" will denote any linguistic entity which is associated with a well-defined ethnic group, with its specific culture, identity and location. Swahili, in this case, will not be regarded as an ethnic language. 3. English has been left out of this discussion because of its special status and function. 4. In fact from that time, Swahili took over many of the official roles traditionally reserved for English. 5. According to the 1967 and 1988 (provisional results) Tanzania Population Census. 6. Most of the information was gathered through personal communication with university students from these areas. The Vumba case, however, was described by Nurse and Walsh (this volume). 7. Evidently, the situation, as described by Tosco (this volume), is much more complex. 8. In fact, this complex greeting system (Lwimbo) may have been initiated as a way of preserving ethnic identity. In this system the younger person has to mention the "clan" (ethnic) name of the older person. The latter has also to mention the name of one of the grandparents of the younger person. 9. As reported by a university student from the area. 10. As reported by E. Gross (personal communication). 11. The issue of who is "Mswahili" has triggered a lot of debate. This is discussed in Batibo (1988) and Sengo (1987). 12. Other authors have described it as linguistic contraction (cf. Dimmendaal 1989). 13. According to Carol Scotton (personal communication), this occurs when two languages are competing for roles. 14. As reported to me by Hamisi Akida (personal communication). 15. As reported to me by a Pemba student. 16. This again calls for verification in view of contradictory information. 17. As reported to me by a university student. 18. As reported by D. Nurse (cited by Sommer, this volume). 19. As reported by Polomé (1980) and Mwaruka (1965). 20. According to Légère's description (this volume), Mbungwe population in 1967 was only 54% of that of 1948. 21. This ambivalence creates problems in determining language speakers' populations. That is why census figures of 1948, 1957 and 1967 used by Légère (this volume) should be handled with care as the respondents may have confused between linguistic competence, cultural allegiance and ethnic identity. 22. However, cultural or ethnic traces may still remain as in the case of Sukuma above. 23. In fact, there may be more cases of languages which have died or are dying among closely related languages. This is more difficult to detect than when a language is dominated by a language of a different family. 24. This is an interesting analogy. The Nile perch is a large fish which now dominates Lake Victoria waters. It is nutritious, delicious and, above all, found in abundance.

96

Herman Batibo

However, it has eliminated most of the small fish in the lake. Fisheries authorities have mixed feelings about its continued presence in Lake Victoria. Nobody would expect people to have mixed feelings about the continued presence of Swahili in Tanzania! 25. To be fair to Mwalimu Nyerere, here he was faced with a problem of trying to reassure the Swahili language promoters about the uncontestable place of Swahili as National language. But the consequences of his statement were far reaching. 26. This document is now awaiting the approval of the Cabinet and the Party. 27. One of the measures being taken by some ethnic groups is to revive their earlier activities in the ethnic languages. For example, the Sukuma are planning to revive their former newspaper LUMULI which had disappeared in the 1960s. The Haya, on the other hand, have launched again their former fortnightly, RUMULI, which is read very widely in Kagera Region. 28. Evidently, such an operation would be appreciable for historical interest, but would be unrealistic for any practical use, since there is no chance of organizing literacy, education, mass media or radio programmes based on the ethnic languages.

References Abubakar, A.M. 1970

"Hali na maendeleo ya Kiswahili Zanzibar", in: Taarifa ya Semina ya Kiswahili iliyofanyika Chuo Kikuu, Dar es Salaam, 5-9 Juni 1970.

Batibo, Herman M. 1987

"The challenge of linguists in language development: The case of Swahili in Tanzania", in: Proceedings of the 2nd LASU conference (Harare, Zimbabwe, Sept. 1987), 42-61.

1988

"Some crucial sociolinguistic issues in the growth and development of Kiswahili", Journal of Linguistics and Language in Education (University of Dar es Salaam) 3: 44-57.

1989

"Les parlers de Zanzibar, sont-ils des dialectes swahili? Les premiers résultats d'une enquête lexicostatistique", in: Marie-Françoise Rombi (ed.), 42-61.

Denison, Norman 1977

"Language death or language suicide?", International ciology of Language 12: 13-22.

Journal of the So-

Dimmendaal, Gerrit J. 1989

"On language death in eastern Africa", in: Nancy C. Dorian (ed.) 12-31.

Dorian, Nancy C. 1977

"The problem of the semi-speaker in language death", International nal of the Sociology of Language 12: 23-32.

Jour-

The fate of ethnic languages in Tanzania

97

Dorian, Nancy C. (ed) 1989 Investigating obsolescence. Studies in language contraction and death. Cambridge: Cambridge University Press. Dressier, Wolfgang U. 1972 "On the phonology of language death", in: Papers of the Chicago Linguistic Society 8. Chicago: Chicago Linguistic Society, 448-457. Ehret, Christopher 1980 The historical reconstruction of Southern Cushitic phonology and vocabulary. Berlin: Dietrich Reimer. Ferguson, Charles A. 1959 "Diglossia", Word 15: 325-340. Greenberg, Joseph H. The languages of Africa. Bloomington: Indiana University. Guarisma, Gladys - Suzanne Platiel (eds.) 1980 Dialectologie et comparatisme en Afrique Noire. Paris: SELAF. Heine, Bernd - Rainer Vossen 1975 "Zur Stellung der Ongamo-Sprache (Kilimandscharo)", Afrika und Übersee 59 (2): 81-105. Itandala, Buluba 1983 A history of the Babinza of Sukumaland. [Unpublished Ph.D. thesis, Mulhouse, Canada.] Mann, Michael - David Dalby (eds.) 1987 A thesaurus of African languages. London: Hans Zell and K.G. Saur. Mochiwa, Zacharia S.M. 1979 The impact of Kiswahili language on ethnic languages: A case study from Handeni District. [Unpublished M.A. thesis, Dar es Salaam.] Molnos, Angela 1969 Language problems in Africa. A bibliographical summary ( 1946-67) of the present situation, with special reference to Kenya, Tanzania and Uganda. Nairobi: East African Research Information Centre. Mwaruka, Ramadhani 1965 Masimulizi juu ya Uzaramo. London. Polomé, Edgar C. 1980 "The languages of Tanzania", in : Edgar C. Polomé - C.P. Hill (eds.), 3-25. Polomé, Edgar C. - C.P. Hill (eds.) 1980 Language in Tanzania. Oxford: Oxford University Press. Rombi, Marie-Françoise (ed.) 1989 Le Swahili et ses limites. Paris: ADPF. Rottland, Franz - Duncan Okoth Okombo 1986 "The Suba of Kenya: A case of growing ethnicity with receding language competence", Afrikanistische Arbeitspapiere 7: 115-126.

98

Herman Batibo

Sengo, Tigiti Shaaban Yusuf 1987 Some issues concerning Kiswahili scholarship. [A paper presented at a Linguistics Forum, at the University of Dar es Saalam, February 1987.] Winter, Jürgen Christoph 1976 "Language shift among the Aasax, a hunter-gathered tribe in Tanzania", Sprache und Geschichte in Afrika 1: 175-204. 1980 "Internal classifications of Kilimanjaro Bantu compared: Towards an East African dialectometry", in: Guarisma, Gladys - Suzanne Platiel (eds.) 101-132.

Language shift in Tanzania* Karsten Legère

It is well known that the spread of Swahili as a supra-ethnic means of communication in East Africa in general, and Tanzania in particular, took place on the basis of a considerable ethnic and linguistic heterogeneity. Suffice it here to say that the ethnic situation in Tanzania is marked by the existence of approximately 120 nominal ethnic communities. Accordingly, the linguistic situation reflects, as in the majority of other states in Sub-Saharan Africa, to a large extent a dialect continuum. As a rule, languages spoken by members of neighbouring ethnic groups share a high degree of mutual intelligibility (provided that they originate from the same language family) and the communication between their speakers is merely affected. Language plays an important role in any attempt to identify ethnicity. Nevertheless, its influence may not be overestimated. On the strength of socio-economic processes in Tanzania, a number of changes have taken place which affect the knowledge and performance in languages of ethnic groups, although the consciousness of belonging to a particular group has been retained and is overt very well. Thus, it is not strange to observe cases where an active or even passive command of the local language1 (which relates directly to the ethnic background of the interviewee) is no longer found. Other aspects such as e.g. place of birth or parents' origin, traditions and customs etc. are kept alive when identifying oneself with a certain ethnic group, while language knowledge is assigned a less important position. The gap between ethnic origin and knowledge or active command of the local language portrays a large variety of changes at the grassroots level and results in various stages of language shift. On the strength of interviews conducted in various parts of Tanzania some years ago several observations were made which have been verified more recently.2 The summary is presented below: (i) There are areas inhabited by persons who speak Swahili as their mother tongue. These are places where people live who claim to have originated from various ethnic groups or nationalities which identify themselves by

100

Karsten Légère

autonyms like "Shirazi", "Pemba" or even "Arab". Swahili dominates in traditionally Swahili speaking settlements along the coast and the islands, up-country historically in some places along the caravan routes (e.g. Ujiji) and under heterogeneous conditions (e.g. in towns, on sisal plantations, in mining areas etc.) (ii) Swahili is spoken as the first language in the hinterland of the coast where it has spread significantly. This language is learnt simultaneously in childhood with the local language or even before mastering the latter. In addition, it has a stronghold in towns as well as adjacent areas sharing increasingly the same prestige as in the hinterland, local languages having subsequently been given up. (iii) Local languages are used alongside Swahili in some towns and rural areas which are situated in the territory of a dominant ethnic group. In the past, these languages played an important role as a means of communication, whereas now this function has gradually been usurped by Swahili which keeps on replacing the local languages (see e.g. Mbeya, Mwanza or Bukoba). (iv) Local languages prevail in daily communication in rural areas where a particular ethnic group dominates. Taking into account that even in 1988 out of a total Tanzanian population now numbering 23,174,336, 3 where about 80 percent of the people still live in the countryside, it seems a well established fact that the role of local languages continues to be important. This assumption, however, does not take into account extralinguistic factors like school education, involvement in processes beyond the frontiers of traditional ethnic groups, social mobility, administrative measures etc. Accordingly, any approach to local languages has to face the grassroots situation. Thus, in particular, languages of small ethnic groups portray a gradual reduction of their role in the territory of the society under consideration. This coincides well with the observations made in other parts of the world. 4 Consequently, their speakers are, under certain conditions, exposed to the assimilating influence exerted by neighbouring ethnic groups (and their languages) which are either more numerous or influential in the socio-economic field. In Tanzania, processes of language shift have been taking place for centuries. Thus, in the past a number of speakers of local languages gave up their own language in favour of that of their neighbours or of a language with a more distinguished social prestige. Such a tendency was reported by Busse ( 1960: 11 ) who observed that the Nyiha language in Southwest Tanzania (Mbeya Region) faces a strong impact both of the Nyakyusa and Swahili languages. He went on to state that thanks to the efforts of missionaries who succeeded in producing a small number of booklets, the decline of the Nyiha

Language shift in Tanzania

101

language did indeed slow down, though at the same time stressing that the pressure on that language was still rather great at that time. A similar development was observed by Besha (1974) in Lushoto District (Tanga Region) pertaining to the shift from Maa, Mbughu, and Shambala. 5 Occasionally this linguistic assimilation coincides with a change in the ethnic identity. For many years in the past and in particular now at present, changes from local languages to Swahili have gained momentum. Due to the strong socioeconomic status of Swahili speakers in earlier years as well as later resulting from some administrative attempts to strengthen the use of Swahili in official domains, the Swahili language has always enjoyed a high prestige. At the same time, the linguistic assimilation by a neighbouring ethnic group and its language is losing ground or has now come to a standstill. However, Swahili has in fact become the major factor and symbol of linguistic integration which runs contrary to the consolidation of languages spoken by larger ethnic groups. This process, however, does not result in the formation of a large community which identifies itself as the "Waswahili", but, on the one hand, the original ethnic identity is preserved or, on the other, is replaced by the (sometimes vague) feeling and consciousness of being a Tanzanian. So far, the latter tendency to refrain from any ethnic affiliation has not made much progress. For the overwhelming part of the Tanzanian population, thinking in terms of ethnic categories is nothing strange, as even interviews with secondary school students who grew up partly in urban areas reveal. Unfortunately, for present years, reliable data relating to the ethnic situation in Tanzania are not available. In fact, the 1967 population census was the last census where enquiries about the ethnic composition of the Tanzanian population were made. A comparison between the 1967 figures and those of earlier years (esp. 1948) reveals various tendencies which, to a certain extent, are still valid for describing the present situation. These data are quite instructive. Referring, on the first hand, to small ethnic groups (as an arbitrary upper limit groups up to 20,000 people were included) a picture may be drawn (Table l). 6 This summary can be supplemented by figures for the Ankole, Baluhya, Ganda, Kikuyu, Soga, Taita and several other groups the most of whom live in neighbouring countries like Uganda or Kenya. Referring to the types of language contact presented above, some inferences as to the fate and future of languages spoken by small groups may be made. As a matter of fact, a number of those groups live up-country sometimes in remote corners of Tanzanian territory. Their members have not been much exposed to socio-economic processes taking place in the country as a whole. On the contrary, many of them live under conditions which are

102

Karsten Légère

Table 1. Smaller ethnic groups in Tanzania Ethnonym - figures 1948/1967 (territory; growth 1948/1967 in percent/selective) Alagwa/Wasi/Chasi - 12,513/15,746 (Kondoa/Dodoma; 126%) Bende - 8,836/8,269 (Mpanda/Rukwa; 94%) Bungu/Wungu - 8,291/1957: 14,926 (Chunya/Mbeya; 1967 classified as Kimbu?) Burunge - 9,716/9,286 (Kondoa/Dodoma; 96%) Doe - 7,781/13,869 (Bagamoyo/Coast) (Dorobo - 890/2,038 = Masai?) Gorowa - 17,705/17,397 (Babati/Arusha; 98%) (Holoholo - 4,410/10,931 - Kigoma; from Zaire) Ikizu - 9,303/10,647 (Musoma; 114%) Ikoma - 9,711/8,568 (minus Issenye and Nata, Musoma) Isanzu - 12,243/20,355 (Singida) Issenye - /9,235 (Musoma) ?Jiji - /14,957 (apparently toponym Bo. 0 in intervocalic position, e.g. Sw. mali > Bo. mai 'property', gari > gai 'car', hela > hea 'money', etc. Recently, Swahili words like mbolea 'fertilizer', elimu 'education', wilaya 'district' or kipuri (alongside kipui 'spare part') were heard in the conversation conducted in Bondei. This fact transpires even from the Bondei toponym Magia which is similar to Sw. Magila. In addition, the natural development in the morphological field has suffered a sever blow. Various derivational devices no longer function. Thus, recently coined Swahili nouns or verbs are borrowed into Bondei notwithstanding the fact that the latter language avails of the same derivational properties as Swahili. Thus, there is no problem in coining easily appropriate equivalents in Bondei for those terms and expressions which are now used as loans from Swahili, e.g.: Sw.-simama 'stand up' Bo. -gooka 'stand up'

> msimamo 'position' > *mgooko 'position'

Sw. -zaa Bo. -vyaa

'give birth' 'give birth'

> -zalisha > -vyaisa

Sw. -isha Bo. -siya

'come to an end' > mwisho 'end' > *msiyo

'production' 'work as midwife' 'end' 'end'

A last phenomenon to be mentioned is the role of code switching in Bondei speech. This does not take place in discussing traditional topics e.g. in narrating a fairy tale or local stories from the past, customs, songs etc.

Language shift in Tanzania

111

As soon as recent issues are discussed, one observes many cases where the Bondei language gives way to Swahili expressions or even clauses. Below some examples of recorded speech (Sw. in italics) illustrate this observation: (1)

mzee Old

uyu this

nee emph

mwenyekiti chairman

wa of

shina cell

'This old man was indeed the party cell chairman.' (2)

lugha language

ya pesa of money

iyo this

'this language of business' (3)

wakurugenzi wa maendeleo wahusikao na ndima iyo directors of development they-concern-rel with work this 'development directors in charge of this job'

(4)

kuhusu mpango wa kuwega wana shule kama iyo Inf-concern project of Inf-put children school like this 'what concerns the project to send children in a school like this'

As regards the grammatical system of Bondei, this has remained intact so far. The tape recordings evidence that apart from code switching in cases where a Bondei expression does not exist or was not at hand, or was simply forgotten, the borrowings were treated in the same way as the elements from Bondei itself. Swahili expressions or loan words are made to fit the grammatical rules of Bondei, e.g. shiika da maendeleo 'Development Agency' (connective da for class 5/6 words), matatizo yo yose 'no difficulties whatsoever' iyo yose corresponds to Swahili yo yote), kiwandai 'in the factory', i.e. kiwanda + Loe -i), etc. Today Bondei is facing a turning point in its history. It is legitimate to claim that the younger generation is much more inclined to speak the target language Swahili than Bondei. A large and steadily growing number has not been properly taught to speak and to esteem the language of the ethnic group they belong to. Obviously, parents do not care much about this fact. Hence, one may not expect this younger generation to show any language loyalty in the future, namely to stick to the Bondei language as one symbol of their ethnic identity and, contrary to that what they themselves have experienced, to transmit to their children a language they have not mastered very well. More plausible is a prognosis which takes into account the massive language shift at present and infers that supported by the mass media, administration, increasing social mobility etc. Swahili will strengthen its position among the Bondei even faster. As long as no concerted efforts by the Bondei are

112

Karsten Légère

undertaken this development will finally result in the death of their language. It is, of course, too early to formulate any hypothesis on how long this will take, but, nevertheless, linguists should be aware of this fate which awaits this and other Tanzanian languages existing under conditions similar to Bondei. In any case, linguistic contributions in studying Bondei or other local languages facing an uncertain perspective might be highly appreciated by c o m i n g generations.

Notes * The financial support by the University of Dar es Salaam, Faculty of Arts and Social Sciences as well as the Institute of Kiswahili Research to conduct field research within the framework of the project "The Impact of Kiswahili on Local Languages" is gratefully acknowledged. 1. It is clear that in view of the number of people speaking a given language as well as its distribution in a more or less large part of an area, sometimes being even dominant in a region of Tanzania/Mainland, raises questions on this term. It seem, however, motivated in contrast to Swahili which among the Tanzanian languages is the only one which has spread over the whole country. 2. See Légère (1982). 3. Quoted from Uhuru [Freedom], 15/3/1989, No. 8039, p. 1. 4. For a summary of this phenomenon and relevant factors affecting language shift, language death etc. see Dressier (1988). 5. Compare Chamungwana ( 1971: 11). 6. The figures for Tanganyika/Tanzania are according to Moffett (1958) and Tanzania (1971); for Zambia see Ohannessian - Kashoki (1978), and for Kenya see Kenya (1981). 7. About the Zaramo language in his times Mwaruka (1965: 115) wrote: " . . . mpaka sasa sehemu kubwa ya Zaramo ya Mwambao lugha itumiwayo ni Kiswahili, na wengi wao hawaifahamu vizuri lugha yao ya asili ya Kizaramo." [ . . . until now the language used in a large part of coastal Zaramo country is Swahili, and many of them [i.e. Zaramo people - K. L.] do not well master their mother tongue Zaramo.] 8. See "Sehemu Ya Tatú. Haki na Wajibu Muhimu" [Third part. Rights and important duties], in: Tanzania (1985). 9. Uhuru, 2/11/1976, 5. 10. A summary has been compiled in Légère (1982); illustrative studies were also made by Mochiwa (1979) and Rubanza (1979).

Language shift in Tanzania

113

References Abubakar, A. M. 1970 "Hali na maendeleo ya Kiswahili Zanzibar" [Situation and development of Swahili in Zanzibar], in: Taarifa ya Semina ya Kiswahili iliyofanywa tarehe 5-9 juni, 1970 Chuo Kikuu, Dar es Salaam [Proceedings of the Swahili seminar held at the University of Dar es Salaam June 5-9, 1970], Dar es Salaam, 29-33. Batibo, Herman M. 1985 Le kesukuma (langue bantu de tanzania): Phonologie. Morphologie. (Cahier No. 17.) Paris: Editions Recherche sur les Civilisations. Besha, Ruth M. 1974 A socio-linguistic description of Kimaa and its points of contacts with Kishambala. [Unpublished M. A. thesis, Dar es Salaam University.] 1989 A study of tense and aspect in Shambala. (Language and dialect studies in East Africa 10.) Berlin: Dietrich Reimer. Bryan, Margaret 1959 The Bantu languages of Africa. London: Oxford University Press for International Africa Institute. Busse, J. 1960 Die Sprache der Nyiha in Ostafrika. (Veröffentlichungen des Instituts für Orientforschung 41.) Berlin: Akademie-Verlag. Byarushengo, Ernest R. 1975 An examination of the segmental phonology of Haya. [Unpublished M. A. thesis, Dar es Salaam University.] Chamungwana, W. H. Shedu 1971 Primary school entrants' knowledge and use of Swahili. (Dar es Salaam, mimeogr.) Dimmendaal, Gerrit J. 1989 "On language death in Eastern Africa", in: Nancy C. Dorian (ed.), Investigating Obsolescence. Cambridge: Cambridge University Press, 13-31. Dressier, Wolfgang 1988 "Spracherhaltung - Sprachverfall - Sprachtod", in: U. Ammon - N. Dittmar - K. J. Mattheier (eds.), Sociolinguistics/Soziolinguistik (Vol. 2). Berlin - New York: Walter de Gruyter, 1551-1563. Ingrams, William H. 1924 "The dialects of the Zanzibar Sultanate", Bulletin of the School of Oriental Studies iii (3): 533-555. Kenya, Republic of 1981 Kenya Population Census 1979. Nairobi: Central Bureau of Statistics. Khamis, Said Ahmed Mohamed 1984 Urban versus rural Swahili (A study of Pemba Varieties). [Unpublished Ph. D. dissertation, Karl Marx University, Leipzig.]

114

Karsten Légère

Khamisi, Abdu M. 1974 "Swahili as a national language", in: G. Ruhumbika (ed.), Towards Ujamaa. Kampala - Nairobi - Dar es Salaam: East African Literature Bureau, 288-308. 1983 "Kiswahili ikiwa ni lugha ya kimataifa" [Swahili as an international language], in: Makala za Semina ya Kimataifa ya Waandishi wa Kiswahili [Papers of the International Swahili Writers' Seminar], Vol. I Lugha ya Kiswahili [Swahili language]. Dar es Salaam: Institute of Kiswahili Research, 1-18. Kihore, Yared M. 1976 "Tanzania's language policy and Kiswahili's historical background", Kiswahili 46 (2): 47-67. Légère, Karsten 1975 "Zum Verhältnis zwischen dem Swahili und anderen tansanischen Sprachen", Zeitschrift für Phonetik, Sprachwissenschaft und Kommunikationsforschung 28 (3/4): 342-348. 1980 "Die sprachliche Situation in ländlichen Gebieten der Region Mbeya (SW Tansania)", Linguistische Studien, Reihe A, 64: 177-214. 1982 Die sprachliche Situation in Tansania (Festland) und ihre Veränderung von der Mitte des 19. Jahrhunderts bis in die Gegenwart. [Unpublished inaugural dissertation, Karl Marx University, Leipzig.] 1983 "Lokale Sprachen in Tansania und sich verändernde kommunikative Bedürfnisse ihrer Träger (am Beispiel des Bondei)", Zeitschrift für Phonetik, Sprachwissenschaft und Kommunikationsforschung 36 (1): 79-86. 1986 "Kimgao and Southern Kiswahili in mainland Tanzania", Linguistische Studien, Reihe A, 148: 146-159. 1988 "Bantu and Southern Cushitic. The impact of Kiswahili on Iraqw", Zeitschrift für Phonetik, Sprachwissenschaft und Kommunikationsforschung 41 (5): 640-647. Maganga, Clement 1990 A study on the morphophonology of Standard Kiswahili, Kipemba, Kitumbatu and Kimakunduchi. [Unpublished Ph. D. dissertation, University of Dar es Salaam.] Mekacha, Rugatili D. Κ. 1984 Research proposal: phonological processes affecting Ki-nata vowels Dar es Salaam (mimeographed). Mkude, Daniel 1979 Conflicting attitudes towards local languages. Dar es Salaam (mimeograph). Mochiwa, Zacharia S. M. 1979 The impact of Kiswahili language on ethnic languages: A case study from Handeni District. Dar es Salaam: University [Unpublished M. A. thesis, Dar es Salaam University.]

Language shift in Tanzania

115

Moffett J. P. 1958 Handbook of Tanganyika. (Second edition.) Dar es Salaam: Government Printer. Möhlig, Wilhelm J. G. 1974 "Die Bedeutung der Minoritätenfrage für die Sprachsituation in Tansania und Kenya", in: M. O. Hinz (ed.), Ethnische Minoritäten im Prozeß nationaler Integration in Afrika. Hamburg: Helmut Buske, 115-124. Mwaruka, Ramadhani 1965 Masimulizi juu ya Uzaramo [Stories about the Zaramo country]. London: Macmillan. Ndilime, M. M. 1977 Bibliografia ya kazi za fasihi simulizi (methali, vitendawili na hadithi) [Bibliography of works in oral literature (Proverbs, riddles and folkstories)]. Dar es Salaam: Institute of Kiswahili Research (mimeographed). Nurse, Derek 1979 Description of sample Bantu languages of Tanzania. (African languages/ langues africaines 5, 1.) London. Ohannessian, Sirarpi - Mubanga E. Kashoki (eds.) 1978 Language in Zambia. London: Oxford University Press for International Africa Institute. Polomé, Edgar C. - C. P. Hill 1980 Language in Tanzania. London: Oxford University Press for International Africa Institute. Rubanza, Yusuf T. 1979 The relationship between Kiswahili and other African languages: The case of Kihaya. [Unpublished M. A. thesis, Dar es Salaam University.] Struck, Bernhard 1921 "Die Einheitssprache Deutsch-Ostafrikas", Koloniale Rundschau 4: 164196. Tanzania 1971 1967 Population Census, vol. 3. Dar es Salaam: Government Printer. 1985 She ria ya Marekebisho ya taño katika Katiba ya Νchi [Fifth Amendment of the National Constitution], (Acts Supplement no. 1 to the Gazette, no. 11, 66). Dar es Salaam: Government Printer. Whiteley, Wilfred Howell - A. E. Gutkind 1958 A linguistic bibliography of East Africa. (East African Linguistic Studies, no. 2.) Kampala: East African Institute of Social Research.

Reduction in Kore reconsidered* Gerrit J.

Dimmendaal

1. Introduction Despite the fact that a number of languages in eastern Africa are on the verge of extinction, little study has been made of the linguistic processes involved in their gradual, or sometimes sudden, contraction. One of the few actual studies at hand of reduced language competence is Heine and Vossen's (1980) description of the speech of a semi-speaker of Kore (Kenya). The present contribution is an attempt to interpret data in the vocabulary list which Heine and Vossen (hereafter H&V) have left unanalyzed; also, alternatives are proposed to some of H&V's interpretations. In addition, Kore is compared with related and unrelated languages, in order to assess the implications of the Kore case for a general theory of language reduction. As argued below, the Kore data suggests that reduction is to be taken as additional to, rather than equated with, processes of pidginization. In this way it is more comparable to changes occurring in healthy languages, though with a compressed timespan.

2. The Kore and their former language 2.1. The external setting of the Kore case The ethnic Kore, who currently number a few hundred, live on Lamu Island off the Kenyan coast and on the opposite mainland around Mokowe (Kenya). Their political and social history is discussed in great detail in Romero Curtin (1985); a summary of main events in the history of the Kore is presented below. According to oral traditions, the Kore have Maasai ancestry. Their forefathers, who problably belonged to the L-Aikipiak Maasai, were defeated by the Purko Maasai some time in the 1870s. (See Vossen 1982: 69-82 for

118

Gerrit J. Dimmendaal

an extensive account of Maasai history.) After this event, the ancestors of the Kore fled from their home area northwest of Mount Kenya. They were then taken prisoners by Somali people near the current Samburu-Rendille boundary in northern Kenya, after which they ended up as clients or slaves in Somali households. They are said to have been set free by British imperial forces around the end of the 19th century. The Kore people seem to have regained cattle in the years following mainly through raiding. Loss of cattle brought them from Kenya's mainland to Lamu some forty years ago, where they are now farmers and fishermen. These days, the Kore speak Somali as their first language, while Swahili is used mainly as a lingua franca. As a socio-political and historical unit, however, they are distinct from the surrounding Somali and Swahili speaking people. The Kore do not seem to marry ethnic Somali (whom they regard as bandits), even though Somali is their first language. Intermarriage with descendants of Lamu's ex-slave community (who speak Swahili) does take place however. In other words, the Kore have maintained ethnic identity although their society seems to be of a fragmented nature these days - without maintaining their former language. Data on the now abandoned language of the Kore was obtained by Bernd Heine during several visits to the area in 1976 and 1977 on the basis of interviews with - what seem to have been - the only two surviving semispeakers still remembering parts of the former Kore language. They were a sixty year old farmer and former headman, and a seventy year old mentally disturbed woman. Although both Bernd Heine and Rainer Vossen made several attempts to retrieve information on the Kore language in various parts of East Africa, no traces were actually found. In the study of speech behaviour, ideally one would like to investigate proficiency and social stratification on the basis of interviews with a continuum of speakers of different regions in situations where reduced linguistic competence occurs as a result of language contraction. Speech behaviour among semi-speakers of Australian languages such as Dyirbal (Schmidt 1985) or Warlpiri (Bavin 1989), or East Sutherland Gaelic (Dorian 1978), for example, has revealed that the degree of morphological simplification correlates with age. But given the linguistic situation for Kore, it was neither possible to simulate time depth by comparison of contemporary speakers of all reaches of the proficiency continuum, nor could a longitudinal study be carried out. As noted by Carol Myers-Scotton (personal communication), the nature of the material available on Kore also does not allow testing her code switching model (discussed elsewhere in this volume). Idiolectal grammars based on the speech of one informant have been a target of repeated criticism. In the

Reduction in Kore reconsidered

119

case of Kore this choice was again not made voluntarily, as there was only one semi-speaker left who could be interviewed at the time of Heine's investigation. Nevertheless, an interesting pattern emerges from the description provided from H&V, comparable to reduced structures observed for other dying languages. H&V take the position that the reduction observable in the male speaker which Heine interviewed are the same as those characteristic for pidginization. Such processes involve a development from inflectional and agglutinating structures towards analytic and isolating structures, and a suppression of marked categories in favour of corresponding unmarked ones. Below, it will be argued that there are also some striking differences between reduction due to language shift and reduction due to pidginization, which suggests that they are better treated as distinct processes. From the relatively little information that Heine was able to obtain from the male semi-speaker (whose primary language is Somali, as stated above), it became clear that the former language of the Kore belongs to the Maa cluster of Eastern Nilotic, one of the three primary branches of Nilotic. The Maa cluster consists of a number of closely related dialects spoken in Kenya and northern Tanzania: Camus, Samburu (together sometimes referred to as North Maa and spoken in Kenya), and Maasai (or South Maa, spoken in Kenya and Tanzania). As argued by H&V, the former language of the Kore should be classified as an additional variety of Maa. Its last semi-speaker, as interviewed by Heine, presumably grew up in a bilingual community setting in which Somali as the dominant language (or target language in the sense of Sasse; see elsewhere in this volume) was fully acquired; limited access to the Kore language resulted in imperfect mother tongue learning of this language for the interviewed male speaker (and presumably his contempories). The exodus from the Maa area, and the subsequent captivity in the Somali area, took place when the father of the male semi-speaker was still a child. While the father of the interviewed male semi-speaker acquired the Kore dialect of Maa as his primary language, his son probably learned Kore in chunks (rather than unlearning it), and Somali became the language of primary socialization. This ultimately resulted in a total replacement of Kore by Somali. Kore thus became a quickly contracting language, disappearing essentially in one generation, a case of radical death in terms of Campbell and Muntzel's typology (Campbell - Muntzel 1989: 183).

2.2. Morphological reduction in Kore A general characteristic of reduction in contracting languages is morphological simplification. As noted by H&V, these structural consequences are also

120

Gerrit J. Dimmendaal

observable in the Kore data, in particular for the two major categories that are morphologically rich and complex, the noun and the verb. In its most complex form, the noun in Camus, Maasai and Samburu consists of a stem preceded by a gender-marking proclitic and followed by a number suffix. Gender markers are omitted when the noun is preceded by a demonstrative, or when the noun is used in a general (indefinite) or adverbial sense; otherwise, the marker is obligatory. There are three nominal gender markers: masculine, feminine, and one location-marking noun. Their exact form depends on the syntactic role performed by the noun, as well as on the category of number. A noun can be inflected for number in the singular (by means of a singulative marker), or in the plural. For a few nouns there is an additional collective marking form. There are also nouns (many of which are derived form verbs) which have a number suffix both in the singular and in the plural. In Maa and most other Nilotic languages, there is a large set of number markers. Nouns referring to entities naturally occurring in pairs or large numbers tend to have a basic, morphologically unmarked, form in the plural and a singulative suffix in the singular. Some examples from Maasai illustrating gender and number (f = feminine; m = masculine):

(f) (f) (m) (m) (m)

singular

plural

en-tólú εη-kerá-í ol-alá-í ol-ó-ito οΐ-οιη-ónì

in-tolú-ó m-kérà il-álá il-ó-ìk ιΐ-οιη-ók

'axe' 'child' 'tooth' 'bone' 'bull'

Eastern Nilotic languages belonging to the Non-Bari branch (to which Maa belongs) have gender markers generally alternating for singular and plural. Because of these portmanteau morphemes, number inflection in these languages is lexically and syntactically redundant. Despite this redundancy, the highly complex system of number suffixation has been retained, and there are no signs of it breaking down or disappearing (Dimmendaal 1987). The noun in Kore as used by the semi-speaker has become virtually invariable. The category of gender has lost its grammatical function but has been retained as a petrified element attached to the noun, as a comparison with the Maasai examples above illustrates: (2)

singular ndólú ilalá lóik

and plural 'axe' 'tooth' 'bone'

Reduction in Kore reconsidered

121

The loss of the initial vowel in certain allomorphs of the gender marker in Kore is also attested in the Camus and Samburu dialects of Maa. For example, the word for 'axe' in Samburu is n-tólú. (See also Vossen 1988 for additional data.) The vowel loss can therefore be interpreted as a natural historical development. Because morphophonemic alternation for the gender marker has disappeared in Kore, number suffixation would no longer have been redundant. Nevertheless, both number and gender are almost entirely absent as formal categories of the Kore noun (while in Camus or Samburu number inflection is still fully productive). The nominal invariability of the Kore noun is therefore probably a consequence of language reduction. Presumably the generalization and fossilization of the form with (rather than without) the gender marker in Kore is related to its wider syntactic distribution. Only a small set of (presumably) high-frequency words still alternates in Kore: (3)

singular

plural

qgérá-í Ιαϊη-όηί

qgérá laitj-ók

'child' 'bull'

Jn addition to this set, which includes the nouns for 'baboon', 'boy', 'camel', 'sandal', 'sheep' and 'woman', there is an irregular plural form, and one form where tone distinguishes the singular from the plural: (4)

singular

plural

qgiôéq laitórókí

qgishù laitorokí

'cow' 'bee'

For the other nouns, either the original singular or the plural form has been retained. Thus, in the case of 'axe' the singular was retained, while in the case of 'tooth', or 'bone', the plural form was still remembered by the semispeaker, as a comparison between examples (1) and (2) above illustrates. Such forms can refer to singular and plural entities, whereby the numeral 'one', nabô, may be used to specify a singular form. The local markedness of certain singular forms (expressed in, for example, Maasai by inflecting the singular rather than the plural for number) thus manifests itself in Kore in that the plural form was retained. There is one nominal prefix in Kore, ma-, which has been borrowed from Swahili. In Kenya Up-country Swahili, the same prefix is added to a nominal stem if it is to be emphasized that a collective meaning is intended; otherwise the singular form of the noun is used, although the singular form is itself neutral as to number in this pidginized version of Swahili.

122

Gerrit J. Dimmendaal

Retained complex nominal forms in Kore include a few descriptive genitive constructions, as well as noun plus adjective forms. These examples show that there was still some idiom remembered by the semi-speaker. Examples: (5)

Kore a.

b.

Samburu

qguöuk

ló

Qgàfì

η-káúk

è

qk-àjì

mouth

of

house

mouth

of

house

lowarú

qíró

l-owworú

girò

?

striped

?

striped

'door'

'leopard'

(Forms that are cognate with l-owworú in Samburu refer to 'wild beast' or 'carnivorous animal' in other Eastern Nilotic languages.) The verb root in Maa can be modified by affixes expressing valency change, or inflectional markers for aspect, negation, number, person, tense, and voice. The Kore verb is invariable, but, as with nouns, certain morphological traits have been retained, albeit devoid of any meaning synchronically. Compare the potentially complex verb in Maasai with the fossilized verbal formula in Kore set out in Table 1. Table 1. Verb comparison Maasai

Kore

negation tense person marker lexical root derivational suffix(es) aspect voice

? (absent) k- plus third person marker lexical root (absent) (absent) (absent)

The prefix k- in Maa is a subordinating particle co-occurring with first and third person forms in questions and answers, as in the following Maasai example: (6)

k-ε-ίάίρά

suB-3-finish

e-siààì

work

'has he finished the work?' In terms of discourse types, question-answer interactions presumably have a high frequency, as a result of which its formal marker seems to have been generalized as the basic form in Kore. (In addition, there is a second type

Reduction in Kore reconsidered

123

of lexicalized verb form, as shown below.) In Maa, first person singular and plural forms are marked by means of different prefixes; the second person prefix is identical for singular and plural, but the latter is distinguished from the former by means of reduplicating the verb root or through a separate suffix depending on tense/aspect distinctions. The third person is not expressed by means of zero marking, as in many languages, but by means of prefixes; its unmarked status manifests itself through its generalization as the only reference form for conjugated verbs in Kore. The form generalized by the non-full speaker of Kore is the third person singular imperfective tense form (active voice) ε- (and the allomorph e-). Although the majority of verb forms in Kore occurs in combination with this (petrified) third person prefix, the latter no longer functions as a person marker (see also 2.3 below). The generalized form in Kore is the one with the widest distribution in discourse in, for example, Maasai as far as tense/aspect distinctions are concerned; this generalized form is used to indicate present tense, habitual action, future time or no time in particular. Its generalization in Kore may be compared to the extended use of the progressive in the Gaelic of non-fluent speakers as described by Dorian (1978). A Kore example: (7)

aná-lbáyán this-man

k-é-dà have

qgólòq sun/hunger

'this man is hungry' The majority of verbs in H&V's data set is used in combination with the prefix k- and the lexicalized third person marker ε-fe-. The set includes verbs like 'be somewhere', 'do', 'know', 'milk', 'see', 'sweep', and 'want'. For the verb 'go' the perfect form without the discourse marker k- seems to have been lexicalized (H&V 1980: 277); other verbs presumably treated this way are 'die', 'go to', 'pay', 'sing', 'take'. Next to these conjugated forms, H&V give verbs in the infinitive, for example 'breath', 'build', 'give', 'run away'. Furthermore, H&V's wordlist contains forms with a formative t plus a vowel that is identical to the following root vowel. (8)

té-yérà tu-dùmù t-àmà tó-(w)òshò

'cook' 'cut' 'drink, eat' 'pierce, beat'

The same formative occurs as t plus vowel (whose exact form is determined by the following root vowel) in the other Maa dialects, but only with one type of verb: class 1 verbs (see, for example, Tucker - Mpaayei 1955: 56-59).

124

Gerrit J. Dimmendaal

They differ from the other morphological class in Maa, class 2 verbs, in that the latter have an initial high front vowel (historically a prefix), while class 1 verbs do not have such a petrified vowel. This distinction plays a role in Maa verb conjugation and in nominal derivation. Compare the following Maasai examples with the class 1 root -rány and the class 2 root -irrág. The forms in (9a) express general past tense (or narrative mood), those in (9b) are imperatives, and those in (9c) are nominal derivations: (9)

class 1

class 2

a.

é-tá-rány-á '(s)he sang'

é-írrág-á '(s)he lay down'

b.

tà-ràny-à 'sing!'

-írrág-á 'lie down!'

c.

e-rány-átá 'act of singing'

εη-kírrág-átá 'act of lying down'

It should also be noted that the same formative with t is found in combination with some adjectival and predicative forms in Maa, but without the corresponding person markers preceding the marker tV-, as in the form for '(be) bad' in Maasai: to-rrânô. Kore has retained the two morphological verb classes, i.e. it has verb roots with an initial high front vowel (class 2 verbs) and verb roots without such a vowel (class 1 verbs), but there is no indication that this formal distinction plays a role in the grammar, as it does in other Maa dialects. The Kore examples in (8) above lack the person markers of the corresponding past tense in Maasai (as in [9a]), but since absence of a functioning cross-reference system is also found with petrified indicative present forms as observed earlier, the forms in (8) can be interpreted as being the result of a general process of loss of person/number marking. The forms of the Kore verbs most likely are therefore the result of paradigm levelling, whereby past tenses, imperatives and predicative adjective forms merged and past tenses were generalized as the primary reference form for the verbs in (8). This again is a more widespread phenomenon in contracting languages. However, none of the expected corresponding class 2 forms are attested in Kore, i.e. there is no separate past tense form (or imperative form) as for class 2 verbs in Maasai in (9b). Instead, all class 2 verbs (as far as can be concluded from H&V's data) are used in combination with the third person prefix ε-fe- plus the discourse marker k- (i.e. as present indicatives). It is also remarkable that the statistically less numerous class 2 verbs have not been (over)generalized as class 1 verbs. One explanation for their stability - even when language contraction occurs - would

Reduction in Kore reconsidered

125

be their early acquisition in the development of language with children. As argued in Dimmendaal (1991), the distinction between the two morphological classes in Maa and other Nilotic languages is so pervasive, playing a role in determining morphological allomorphs throughout the grammar, that children probably become aware of this formal dichotomy at an early stage. Verbal inflections as such are not to be treated as an unordered set of morphosyntactic properties. Cross-linguistic studies have shown that verbal inflections differ with respect to their relevance to the verb, i.e. the extent to which their meanings directly affect the lexical content of the verb stem (Sapir 1921; Bybee 1985). Although aspect is highly relevant to the verb (its expression through verbal affixes in many languages being evidence of this), more so in fact than cross-reference marking, it is not attested in the reduced variety of Kore. This most likely is due to semantic reduction, a general property of reduced language use and pidginization.

2.3. Reduction at the phrase and clause level Languages like Maasai and other Eastern Nilotic languages have a fairly strict head-modifier structure. The head noun in Maasai is optionally followed by modifiers such as adjectives or numerals. The latter take agreement markers reflecting gender and number features of the head noun. Gender and number are also reflected in demonstratives in Maa, which precede the noun. Some Maasai examples: (10) ε-môny to-rrónS i-múnyi to-rr5k ! endâ=môny ι kundá=múnyi

'a bad rhino' 'bad rhino's' 'that rhino' 'those rhino's'

In Kore on the other hand, adjectives, numerals, and demonstratives are invariable (although the plural demonstratives kuló (masculine) and kúná (feminine) are used to indicate spatial distance (H&V 1980: 276)). Note also that in the Kore example the demonstrative precedes the invariable noun, while in Maasai it would have replaced the gender proclitic: (11) Kore εηά-ljàm ená-Qgídéq

Maasai εΐέ-càni ι €ná=khéq

'this tree' 'this cow'

126

Gerrit J. Dimmendaal

The majority of nouns in Eastern Nilotic languages has feminine gender. For example, in Maasai only twenty-two out of a list of sixty-seven basic nouns have masculine gender (Dimmendaal 1983: 210). In Maasai, the unmarked character of the feminine is also shown in the use of the feminine demonstrative when the gender of a noun is unknown, as in questions of the type "What is this?". The feminine demonstrative has been generalized as a deictic marker of proximity in Kore, a fact which must be linked to the numerical dominance, and, presumably, higher frequency, of feminine nouns in the language family to which it belongs, a language-specific manifestation of markedness it would seem. The word order in Kore is SVO. Although a few VSO-type clauses are also attested (H&V 1980: 274), it is not clear what determines the variation. One example with a nominal subject preceding the verb in Kore: (12) aná-lbáyán this-man

k-é-ôà have

ηgólòη sun/hunger

'this man is hungry' The other Maa dialects and all other languages of the primary branch of Eastern Nilotic to which Maa belongs are verb-initial. The languages of the Bari group, which constitute the other primary branch of Eastern Nilotic, are SVO languages. Elsewhere in Nilotic there are various verb-initial languages, and several of the closely related Surma languages also have a verb-initial syntax. Given, furthermore, the highly idiosyncratic fact that post-verbal subjects (rather than objects, or both) in Eastern and Southern Nilotic are marked for case by way of tonal inflection, the most likely hypothesis is that verb-initial word order in Nilotic is old. The shift to SVO in Kore must be a recent innovation, given the verb-initial order in the other members of the Maa cluster. In Maasai, the order in predicative phrases is predicate (adjective, noun) followed by the subject in the nominative, whereas in Kore the order is subject - copula - predicate, i.e. with the subject in initial position: (13) (Maasai)

kùmòk many

ilkéèk trees

'the trees are many' (14) (Kore)

oljani

ε

gómùk

'the trees are many'

Reduction in Kore reconsidered

127

As observed by H&V (1980: 274), in clauses with pronominal subjects as agents, a syntactic subject is often omitted, despite the fact that there is no cross-reference marking on verbs. This phenomenon has an exact parallel (as an exponent of language obsolescence) in Australian languages such as Kamilaraay for which Austin (1986: 221) notes that "pronouns, both free and bound pronominals, are not known productively and only appear in a few fixed phrases which are no longer morphologically analysed by the speakers". In Maasai, Samburu or Camus, syntactic subject and object positions can be left unspecified phonetically, but because of reference markers on verbs there is no ambiguity. When the subject position is left unspecified in Kore, the primary interpretation seems to be that a first person is the agent of the action, process or state expressed. There is thus a great deal of indeterminacy and consequently a heavy burden placed on the context for the semantic interpretation of syntactic roles in dying Kore. (15) k-é-ôà 'SUB-3-be

ΐ]%όΙόη hunger/sun

Ί am hungry' The absolute forms of pronouns nevertheless were still remembered by the male speaker (except for the second person plural). The subject pronoun precedes or follows the verb but there is no nominative case marking for post-verbal subjects: (16) (ηαηύ) I/me

k-é-nyá SUB-3-eat

ηάάβά food

Ί want to eat' (17) cwá take

ηαηύ I

ndólú axe

Ί took the axe' (18) (ηαηύ) I/me

k-é-dól SUB-3-see

n'myè him

Ί saw/see him' In accordance with a well-established universal, verb-initial Nilotic and Surma languages have SVO order as an alternative (whereas the inverse is not necessarily true). The shift to SVO in Kore could therefore be interpreted as a natural drift, rather than being a consequence of contraction. Absence of nominative case marking on subjects preceding the verb in Kore is what one would expect, as verb-initial Nilotic and Surma languages also lack nominative case with pre-verbal subjects. However, although nominal (and pronominal) sub-

128

Gerrit J. Dimmendaal

jects may still follow the verb in Kore, they lack a crucial feature shared by all other verb-initial Nilotic and Surma languages: nominative case marking; instead, subjects occur in the absolute case (H&V 1980: 274). The loss of nominative case marking for post-verbal subjects in Kore is therefore most probably a further instance of language reduction, whereby the unmarked form has been retained. The absolute case form is unmarked morphologically and syntactically in these languages, because of its wide distribution, and because other case forms (nominative, genitive, locative) can be derived from it. If verb-initial word order is old in Nilotic, the languages of the Bari group in Eastern Nilotic must have innovated SVO order. Probably as a consequence of this shift (which itself needs to be explained), verbal affixes marking syntactic subjects and objects disappeared from the Bari group in most constructions. The break-down (or petrification) of cross-reference on the verb in Kore could therefore be interpreted as a consequence of a shift to SVO along the same lines as for the Bari group. However, the verb in the Bari group is not invariable. The languages of this group have retained affixes marking tense and aspect as well as derivational suffixes marking modification or verbal valency; since most of these affixes are not affected by a shift in the basic position of syntactic subjects, there is no reason why they should have disappeared in Kore. Again, the reduction in Kore must be a consequence of contraction. The loss of nominative case would be parallel to the systematic reduction of ergative allomorphy and subsequent loss of ergative case marking in Dyirbal as used by non-fluent speakers (cf. Schmidt 1985). There is no reason to assume that in the case of Kore the loss of nominative case marking and the shift of word order is due to interference from Somali or Swahili. Somali is a verb-final (SOV) language. But Kore syntax does not look like relexified Somali. Basic word order in Swahili is SVO, but a finite verb with person and number marking affixes and other morphemes attached to verbal stems constitutes a complete utterance, i.e. syntactic subjects do not need to be expressed. Also, contacts with Swahili are probably fairly recent, i.e. from the period when the Kore moved to Lamu. The loss of nominative case marking in Kore is comparable to the loss of ergativity in contracting Australian languages in this respect. The reduction in verbal morphology is also a trait attested in other dying languages. The question remains whether the SVO order - in clauses where syntactic subjects are expressed - is the result of natural drift or a consequence of language contraction. This issue is taken up again in the final section below.

Reduction in Kore reconsidered

129

2.4. Phonological variability Phonological levelling has been shown to be a widespread feature of dying languages (see, for example, Austin 1986; Campbell - Muntzel 1989). H&V mention similar characteristics for Kore segmental structure which, as suggested by the authors, are also typical of African pidgin languages: the opposition between voiceless plosives and voiced implosives has been lost; differences in vowel length and tongue root position for vowels are no longer phonologically distinctive. They give various examples especially of the first type of alternation in Kore (with numerous examples of t/ö alternations). Some examples: (19) ntitò -ipir

or: or:

ntidò -ibir

'girl, daughter' 'jump'

This process is accompanied by a general loss of morphophonemic rules operating over morpheme boundaries, i.e. allomorphic collapse, as noted above. H&V further suggest (1980: 280): "It would seem that tone has lost its distinctive function in Kore to some extent but the data available do not allow for any generalizations on this issue." African pidgins tend to lose tone as a feature when derived from a tonal source language, as is the case with Fanagolo, based on Zulu (Heine 1973). There is also a tendency towards disyllabic words in such pidgins. However, neither of these are found in Kore. Tonal melodies on words have been retained, and no tendency towards disyllabic patterning can be observed. Given the variability in other phonological traits, the tonal stability is even more striking. Apart from the loss of nominative case marking, which is basically a syntactic phenomenon, tone does not seem to have been affected by the structural decay. Although correspondences with Maasai are not entirely regular, each syllable in the variety of Kore known by the semi-speaker still has its distinctive high or low tone and there is no slant towards a fixed accent or stress pattern. There is therefore no evidence for a reduced tone system, nor is there any indication of interference from Swahili or Somali in this respect, both of which have accentual systems (combined with stress in the case of Swahili). The following nouns from Kore, which, when compared with their cognate forms in Maasai (in the absolute case), may help to illustrate the conservative nature of tone in the former language: (20) Kore

Maasai

mbàà ndólú

εηι-bàè en-tólú

'arrow' 'axe'

Kore

Maasai

olósòwàn Ιίύηάηά

ol-ósòwùàn il-túqáná(k)

'buffalo' 'person'

130

Gerrit J. Dimmendaal

(The Kore forms are used both for the singular and the plural.) The study of acquisition of (genetically related) tone languages like Luo by Blount (1970) has shown that children master tone before segmental distinctions. In view of the early acquisition of tone (and intonation) by children, the stability of tone relative to other phonological features in Kore is not without significance. The process of imperfect mother tongue learning must have provided a strong enough basis for a full acquisition of tone at the word level.

3. Some conclusions The lack of interest in a more or less systematic account of language reduction in (eastern) Africa may be rooted at least partly in the assumption that such performance is unsystematic. Only few authors make incidental reference to the speaking competence of semi-speakers or former speakers. Maguire (1927: 251, 259), for example, observes that semi-speakers of the Cushitic language Aasáx in northern Tanzania no longer remembered irregular plural forms, or the numerals four, five, six, seven and ten. The decreasing use of a language in various domains and, ultimately, its extinction, is accompanied by a systematic reduction in structural complexity, as already observed by Bloomfield in the 1920s with regard to semi-speakers of Menomini. Clearly, pragmatic incompetence leads to structural incompetence. The conclusions to be drawn from an analysis of the Kore data on the nature of language reduction are in line with such observations on, for example, Gaelic by Dorian (e.g. 1978, 1981), or Dyirbal by Schmidt (1985). A number of conclusions can be drawn for the Kore case on the basis of the available material, suggesting that the decay is partly comparable to, but also distinct from, pidginization. The language-specific results in Kore can be interpreted as manifestations of a universal tendency shared with pidginization from the marked to the unmarked. This is true for the generalization of feminine demonstratives, the obsolescence of number inflection (with a lexicalization of either singular or plural nominal forms), and the petrification of specific verb forms. A further feature shared with pidgins is the reduced vocabulary and, correspondingly, semantic extension or widening of lexical items. At the syntactic level, dying languages and pidgins also share a trend towards analytic structures. There is, however, a methodological issue involved in the interpretation of word order change in dying languages and pidgins. Schmidt (1985) attributes the shift to a fixed SVO order in the north-

Reduction in Kore reconsidered

131

ern Australian language Dyirbal to interference from English. Austin (1986: 227) notes that the syntax of the eastern Australian language Kamilaraay has been completely replaced by an English type structure including SVO word order. Given the fact that the dominant language English has SVO word order, the conclusion that the changes in Dyirbal or Kamilaraay are due to interference from the former seems straightforward. But the question imposes itself whether the order in Dyirbal or Kamilaraay is due, not to interference but, instead, to a shift towards the unmarked word order resorted to when morphological coding mechanisms are lacking. Given the SVO word order in English, there does not seem to be any unambiguous way of deciding between these two options. Somali has a verb-final syntax, but Kore does not allow such a word order. There is very little interference in general from either Somali or Swahili in Kore phonology, morphology or syntax (not more it would seem than in contact situations between healthy languages). Given the short period of contact between, for example, Kore and Somali, this should not be surprising. It would seem that differences in duration and intensity of contact between languages in general are major factors determining interference or convergence in language contraction or pidginization (or creolization) processes. The substratum issue is still one of the central problems in creole studies. It may well be that no general rules can be given for such contact situations, or, as Singler (1988: 48) puts it: "given the correlation between degree of homogeneity of substratal input and extent of substratal influence, it follows directly that the impact of substrate languages will vary, perhaps widely, from creole to creole." Similar observations have been made by Thomason and Kaufman (1988: 191) with regard to pidgins. Thus, while many pidgins turn out to have SVO word order, Chinese Pidgin Russian has SOV word order. Next to common features, there are differences in functional domains between pidgins and dying languages, involving a distinct network of social and psychological factors. Pidgins typically emerge in multilingual settings, where speakers usually do not give up their mother tongue. Dying languages, on the other hand, tend to be situated in bilingual settings, or, more correctly, situations of 'encroaching diglossia' (Dimmendaal 1989: 16). Pidgins and contracted languages share strong limitations in structure and function. For example, both processes result in a strong reduction of the morphological component in the medium of communication. But the conspicious shrinkage in morphological alternation may lead to the formal retention of affixes and even clitics devoid of any function or meaning in dying languages; this does not seem to occur in pidgins. In other words, areas of morphological complexity still remain e.g. the retention of the morphological verb classes,

132

Gerrit J. Dimmendaal

and compounds (which as such are rare in the Maa group). In many respects changes in dying languages are comparable to restructuring in natural languages, but the timespan involved is far more compressed in the former. In sharp contrast with pidginized forms of African languages such as Fanagolo, tone in Kore has not been replaced by a pitch accent system, nor has there been any tendency towards disyllabic words, contrary to cases like Kenya Pidgin Swahili where such a syllabic reduction did take place. In other words, the prosodie structure has not been affected. It would be interesting to know whether similarly stable patterns can be observed in tonal languages of Central America like Otomi, which is now being superseded by Spanish. Such differences would seem to follow from the relatively early stage at which tone is acquired by children even in the case of incomplete language acquisition.

Notes * I would like to express special thanks to Bernd Heine for making his fieldnotes on Kore available, as well as to the participants of the symposium for their comments and suggestions. Special thanks also go to the Netherlands America Committee for Educational Exchange for the Fulbright Grant, which enabled me to finish the present paper during my stay as a visiting scholar in the Department of Linguistics, University of California at Berkeley. 1. This kind of (mutual) stigmatization seems to be a more common phenomenon obtaining in situations where language replacement has taken place (Dimmendaal 1989). This fact also would seem to contradict claims in, for example, Giles, Bourhis and Taylor (1977: 326) that "ethnic groups members identify more closely with someone who shares their language than with someone who shares their cultural background". 2. For Maasai, the grammar by Tucker an Mpaayei (1955) is still the main reference work. In addition, data can be found in Heine - Claudi (1986). More recently, Vossen (1988) has published a comparative Maa dialectology with data on Maasai, Camus and Samburu. For Kore, Heine and Vossen's (1980) description remains the only available source. 3. Maa has been an assimilating language otherwise. It has absorbed such linguistic units as Aasáx, Yaaku, Elmolo, Akiek and Omotik. 4. Nouns referring to paired entities (e.g. 'ears'), or large numbers ('mosquito's') tend to be unmarked in the plural in many languages (see, for example, Tiersma 1982). 5. Due to the disappearence of the initial vowel in gender prefixes in Kore, consonant clusters emerged which were not allowed otherwise; in those cases, the gender

Reduction in Kore reconsidered

6.

7. 8. 9.

133

prefixes disappeared completely. Compare Maasai ol-menetja-ni 'corpse, ghost' with Kore meneijaní 'corpse'. The principle of marking the first element in an utterance probably has a functional explanation in that grammatical relations which are marked on the verb are established this way at the outset (cf. Nichols 1986: 81; Dimmendaal 1991). Imperatives are probably basic to some verbs (Menn 1989: 342). See also Andersen (1988) for similar conclusions regarding verb-initial word order in Proto-Nilotic. Whether there was any interference in the target language (Somali) is not known.

References Andersen, Torben 1988

"Ergativity in Pari, a Nilotic OVS language", Lingua 75: 289-324.

Austin, P. 1986

"Structural change in language obsolescence: Some eastern Australian examples", Australian Journal of Linguistics 6: 201-230.

Bavin, Edith L. 1989

"Some lexical and morphological changes in Warlpiri", in: Nancy Dorian (ed.), 267-286.

Blount, Ben G. 1970

"The pre-linguistic system of Luo children", Anthropological 12: 326-342.

Linguistics

Bybee, Joan L. 1985

Morphology: A study of the relation between meaning and form. (Typological Studies in Language 9.) Amsterdam: John Benjamins.

Campbell, Lyle - Martha C. Muntzel 1989

"The structural consequences of language death", in: Nancy C. Dorian (ed.), 181-196.

Dimmendaal, Gerrit J. 1983

The Turkana language. (Publications in African Languages and Linguistics 2.) Dordrecht: Foris.

1987

"Drift and selective mechanisms in morphological changes: The Eastern Nilotic case", in: Anna Giacalone Ramat et al. (eds.) Proceedings of the 7th International Conference on Historical Linguistics. Amsterdam: John Benjamins, 194-210.

1989

"On language death in eastern Africa", in: Nancy C. Dorian (ed.), 13-31.

1991

"The geometry of verb paradigms in Teso-Turkana", in Frans Plank (ed.), Paradigms: The economy of inflection. Berlin: Mouton De Gruyter, 275306.

134

Gerrit J. Dimmendaal

Dorian, Nancy C. 1978 "The fate of morphological complexity in language death: Evidence from East Sutherland Gaelic", Language 54: 590-609. 1981 Language death: The life cycle of a Scottish Gaelic dialect. Philadelphia: University of Pennsylvania Press. Dorian, Nancy C. (ed.) 1989 Investigating obsolescence. Studies in language death and language contraction. Cambridge: Cambridge University Press. Giles, Howard - R.Y. Bourhis - T.M. Taylor 1977 "Toward a theory of language in ethnic group relations", in: Howard Giles (ed.), Language, ethnicity and intergroup relations. (European Monographs in Social Psychology 13.) London: Academic Press, 307-348. Heine, Bernd 1973 Pidgin-Sprachen im Bantu-Bereich. (Kölner Beiträge zur Afrikanistik 3.) Berlin: Dietrich Reimer. Heine, Bemd - Rainer Vossen 1980 "The Kore of Lamu: A contribution to Maa dialectology", Afrika und Übersee 62: 272-288. Heine, Bernd - Ulrike Claudi 1986 On the rise of grammatical categories: Some examples from Maa. (Kölner Beiträge zur Afrikanistik 13.) Berlin: Dietrich Reimer. Maguire, R.A.J. 1927 "Il-Torobo", Journal of the African Society 27: 127-141. Menn, Lisa 1989 "Some people who don't talk right: Universal and particular in child language, aphasia, and language obsolescence", in: Nancy C. Dorian (ed.), 335-345. Nichols, Johanna 1986 "Head-marking and dependent-marking grammar", Language 62: 56-119. Romero Curtin, Patricia 1985 "Generations of strangers: the Kore of Lamu", International Journal of African Historical Studies 18: 455-472. Sapir, Edward 1921 Language. New York: Harcourt Brace. Schmidt, Annette 1985 "The fate of ergativity in dying Dyirbal", Language 61: 378-396. Singler, John Victor 1988 "The homogeneity of the substrate as a factor in pidgin/creole genesis", Language 64: 27-51. Thomason, Sarah Grey - Terence Kaufman 1988 Language contact, creolization and genetic linguistics. Berkeley: University of California Press.

Reduction in Kore reconsidered

135

Tiersma, Peter Meyes 1982 "Local and general markedness", Language 58: 32-49. Tucker, Archibald N. - John ole Mpaayei 1955 A Maasai grammar with vocabulary. London: Longmans, Green and Co. Vossen, Rainer 1982 The Eastern Nilotes: Linguistic and historical reconstructions. (Kölner Beiträge zur Afrikanistik 9.) Berlin: Dietrich Reimer. 1988 Towards a comparative study of the Maa dialects of Kenya and Tanzania. (Nilo-Saharan 2.) Hamburg: Helmut Buske.

Dahalo: An endangered language Mauro Tosco

0. Introduction1 Dahalo is a language spoken in the Lamu district of Coast Province, Kenya, by a few hundred people of all ages (see below, 1., for the ethnic denomination and the number of speakers); the speakers are former hunter-gatherers who partly turned to a sedentary existence in recent years. From a genealogical point of view, Dahalo has been classified as a Cushitic language, but about its inclusion in the Southern group (as expounded especially by Ehret 1980) doubts have been expressed (e.g., by Zaborski 1986 and 1987), and the present writer finds a tentative classification within the Eastern Cushitic group (possibly in a subgroup together with Yaaku) more in line with the available, still limited, data (Tosco 1992). Dahalo has a very mixed lexicon, in which a great deal of Somali, Oromo, Swahili and other Bantu language loans are found, together with ancestral words of probable Khoisan origin, as can be argued by the presence of two clicks. In the following, we shall concentrate on the present sociolinguistic status of the language vis-à-vis its major neighbour, Swahili, and on the ongoing process of assimilation of both the Dahalo language and people.

1. Basic information on the sociolinguistic status of Dahalo Statistics dealing with the number of Dahalo-speakers are totally lacking; the Kenya Population Census gives, quite naturally, the ethnic affiliation without regard to the actual language(s) spoken by the communities. In the case of the Dahalo, the ethnic affiliation is concealed under the cover-term of "SanyeBoni", applied for the Boni of the Lamu District, the Waata of the Tana River and Kilifi Districts and the far less numerous Dahalo.

138

Mauro Tosco

Map 1.

Dahalo: An endangered language

139

The same confusion is found in Tucker - Bryan - Woodburn (1977: 319), where Dahalo is considered "one of the so-called 'SANYE' dialects, spoken in coastal pockets north of Mombasa in Kenya", therefore putting the Dahalo together with the Waata, actually speaking an Oromo dialect (on which see Heine 1981). The townspeople's attitude which is mirrored in these works blurs the differences among the different groups of bush-dwellers under a denomination which, being sociological, reflects the traditional opposition in terms of occupation and social status between the inhabitants of the bush and the "others", either agriculturalists, cattle-keepers or townspeople. On the other hand, the groups of bush-dwellers often lack an ethnic selfdenomination, even if they can understand the derogatory meaning of the current appellations. As a consequence of this, the same term can be used for what are ethnically/linguistically different peoples, and, on the contrary, different appellations can be applied to the same people in different areas. The question has been discussed by Heine (1977) with regard to the Boni: Heine refused the denomination "Aweer(a)", claiming it to be a generic term (meaning simply 'hunter'), and adopted the term "Boni", which is today generally used in linguistics, even if "Boni" is in its turn derogatory (being the Somali denomination of various low-caste groups; in Somalia therefore, "Aweera" - a foreign term, and consequently neutral - is preferred). We shall return below to the Boni, their language and its radically different - in comparison with what we find among the Dahalo - sociolinguistic situation. Among the Dahalo themselves, while the term "Dahalo" (d'aháálo; Singulative M: d'ahááloto) is understood, it is never used, and the Dahalo pretend that it means 'slave' in their own language (but my informants were not able to elaborate on this point, such as to give plural forms, etc.); they further say that "Dahalo" is a term used by the Swahili, to which they in return refer to the Swahili as kúúdza (Singulative M: kúúdzeti), which would likewise mean 'slaves'. The Dahalo we met always referred to themselves as dááko (Singulative M: dáákoto). Probably this is just the name of a section of the people, as the informants were also very firm in saying that, while all the Daako are sedentary, not all the Dahalo are Daako; they furthermore demonstrated that they understand the precise sociological value of the term "Sanye", applying it with preference to the little groups of hunters which live only in the bush and do not practice cultivation. This sharp sociological distinction between the two groups is not accompanied, we were assured, by any practical language difference, and we have been told on many occasions that dialect variability

140

Mauro Tosco

is minimal between the D groups, involving above all the lexicon (more Southern groups having been influenced by Pokomo and Elwana). The people we interviewed occasionally used for themselves the denomination of guho gwitstso 'Little People'; on the other side, the denomination of guho gárimaani ('People of the outside') reported by Elderkin (1974: 2) was unknown. The language was called simply ?áfo gúhooni 'the language of the people'. Apart from the Daako, we recorded the following names of Dahalo sections (the Swahili prefix wa- is occasionally used): manta d'ig i ïïma Tíílanitu lúnku

2. Sociolinguistic aspects of language shift Concerning the actual number of Dahalo speakers, it is calculated in "a few hundreds" by Ehret (1980: 12), about 500 (followed by a question mark) by Sasse (1981: 199), while Zaborski "could estimate about 280 of them, though the upper limit may be about 400" (1987: 223-234). The same estimate of "less than 400" is made by Art Rilling of the Kenya Working Group of the Summer Institute of Linguistics, who adds that "This is admittedly a guess, and the figure is probably generous" (Rilling 1986: 5). We think that the figure of 400 cannot greatly exceed the truth: in the peripheral (for the Dahalo people) area of Mokowe we met in one occasion about 50 of them, and we were told that many more lived in the same area. The 1979 Kenya Population Census gives a number of 4,170 "Boni/Sanye", of which 2,212 are living in the Lamu District. If we compare these figures with those of the 1969 Census (3,972 of which 1,276 are living in the Lamu District, with an increase of 73.3%), we can get the impression of a vital community, but when we consider that the Boni are estimated to be about 3,000, the vast majority of them living in the Lamu District (only the very little Wayoore section and the Kilii section living on both sides of the KenyanSomali border; Heine 1982: 13), very little place is left for the Dahalo, and also the figure for the Boni should probably also be lowered. To our knowledge, the Dahalo have never been the object of any sociolinguistic research, with the exception of the survey carried out for the Kenya

Dahalo: An endangered language

141

Working Group of SIL by Art Rilling and Eric Graham in September 1985. The aim of the research, which covered both the Dahalo and the Boni, was to ascertain the languages spoken by these communities, their degree of literacy and the eventual need for written material in their mother tongues. 24 Dahalo were asked to complete a questionnaire made up of 27 questions: the first eight concerning general information (age, sex, origin of the subject), the remaining being sociolinguistic questions, and the last exploring the attitude toward the eventual development of written material in Dahalo. The results are presented in the Boni/Dahalo Report (Rilling 1986), on which we shall largely draw in the following. 2 The results are not surprising. All the 24 Dahalo interviewees (a slight majority of them being from the area of Mikunumbi and males) claimed to be bilingual in Swahili, although "29% said that there were very few monolingual Dahalo speakers, and that these were all elders" (Rilling 1986: 14); it is significant that 21% of the interviewees acquired Dahalo as a second language (i.e., they were brought up in Swahili) and the same rate admitted not to use Dahalo as "the language of the home". 29% claimed to know Swahili better than Dahalo. Our informants repeatedly told us that, while in the olden days there were Dahalo who did not master Swahili, today everybody can at least understand it. As a matter of fact, all the Dahalo we met were bilingual, while nobody had a good command of any other language. In particular, English - Kenya's official language - was totally unknown. Little can be said about age as a factor in language proficiency; the eldest Dahalo-speaker I interviewed in the area of Mkunumbi spoke with the same high amount of Swahili loans as the younger ones, could not remember the Dahalo numerals from 6 onwards, nor many words referring to hunting and other traditional practices (sentences [1] and [2] below were uttered by him). In mixed couples (which are very common, given the small size of Dahalo groups; see below) Swahili is now used as the medium, and the children are therefore brought up in that language; young people we met in Mkunumbi had a lower proficiency in the Dahalo language than their parents. For many of them, exposure to the language was limited to listening to stories narrated by the elders: they had not learnt Dahalo by way of a normal acquisition process, and had become - or were going to become - semi-speakers (in the sense outlined by Sasse, this volume). The answers of the interviewees to the questions about the degree of bilingualism and of literacy, and about the need for written material are very interesting. The SIL researchers (Rilling 1986: 15) comment that "the Dahalo-speaking community seemed to be more proficient [than the Boni;

142

Mauro Tosco

MT] in Swahili, but less interested in mother tongue literacy": 37% admit that Swahili is the most important language, while only 17% pretend to use Dahalo "most often" and "1/4 of the Dahalo-speaking interviewees prefer to use Swahili rather than KiDahalo". None of them had learnt to read, and even the hypothesis of written material in Dahalo seems to have been largely misunderstood: "They seemed personally disinterested though not opposed to the development of mother tongue materials"; the totality of the interviewees answered positively to question 27 ("There is a need for literature in KiDahalo"), but were mostly unable to give reasons for that need; one can note, nevertheless, that eight interviewees answered "To help preserve the language, that it not be lost" (Rilling 1986: 15). On the other hand, Dahalo do not seem conscious of any threat to their language on the part of Swahili. Reactions to question 25 ("KiDahalo will continue in use?") are for 79% positive. Needless to say, the SIL report does not plead for the idea of developing written material in Dahalo, proposing "to re-evaluate the sociolinguistic situation among the Dahalo community after five or ten years", while a deeper interest in the development of material in Boni is recommended (Rilling 1986: 16). The different "score" of the Boni and the Dahalo raises many questions: the Boni show all the characteristics of a vital community: there are monolinguals, foreign languages (Swahili and, to a lesser extent, Somali) are known, but to a lesser degree and with lower proficiency than found among the Dahalo. The answers to the questions concerning the actual use of the mother tongue in different occasions show that Boni is practically the only language used by the Boni-speakers in many social occasions (work, home, meetings, festivals), while Dahalo retains the majority of uses only at home (79% of total) and especially at festivals, dances, etc. (96% of total), but counting, praying and, presumably, working ("in the forest") are carried on in Swahili.

3. Motivations for the language shift The reason for this very different resistance of the two languages cannot just be the difference in numerical strength of the two communities (about 3,000 the Boni, less than 400 the Dahalo). In our opinion, a major role in the decline of the Dahalo language has been played by the proximity to the Dahalo area of such important Swahili centres as Lamu and, in the past, Witu. On the contrary, the Boni have been divided between Swahili and Somali influence,

Dahalo: An endangered language

143

and we can assume that a multiplicity of influences can result in a partial weakening of their strength and, consequently, in a better resistance of the "dominated" language. 3 Stiles (1988) has recently pointed to the mechanism of casting out as at least a partial answer to the problem of the origins of "outcast peoples" such as the Boni; intermarriage and casting out have in our opinion played a decisive role in the formation of the Dahalo people, just as today seem to work for language shift in favour of Swahili. The Dahalo, being a lowcaste people, intermarry with any other people, especially the Boni, but also the Giriama, the Pokomo, etc.; intermarriage is considered by the Dahalo themselves a decisive factor in the lexical differences between the various Dahalo groups. Economic shift to a semi-agricultural way of life is of course playing a decisive role in the process of replacement of Dahalo by Swahili. This shift has been favoured by the increase in land suitable for agricultural use; the government settlement scheme at Lake Mukunguya, or Lake Kenyatta, has attracted in the area many Bantu farmers from the overpopulated Highlands, thus complicating the linguistic and ethnic situation of the area, and exalting the role of Swahili as an inter-community medium; naturally, the precarious role of the little peoples of the interior of the coast has been affected negatively by these developments. Agricultural and townlife terms in Dahalo are with no exception Swahili, often unassimilated loans. The ban on hunting practices in Kenya is likewise contributing to disrupting traditional ways of life, and, from a linguistic point of view, knowledge of the hunting-gathering vocabulary is today low among the young Dahalo we met. And, as we know from other cases of language death in Africa, the linguistic and cultural collapse of the minority peoples has increased and is gathering momentum from the economic upset of the last decades. But that the process of language shift is not at all recent is shown by the fact that even older speakers, living in the bush in the area of Mkunumbi and economically "unassimilated", make an extensive use of Swahili loans and expressions, which cover, in addition to the more technical domains mentioned above, even general "cultural" vocabulary as 'to marry' (hoovaad-, from Northern Swahili -o(w)a), 'to pray' (?onbole?ad-, from Swahili -omba), 'to bury ( n drigad-, from Northern Swahili -öika), etc.

144

Mauro Tosco

4. Cultural aspects of language shift The cultural identity of the Dahalo seems presently realized "negatively" through the only partial assimilation to the agricultural economy predominant today in the Dahalo area. Most Dahalo in the Mokowe area live as hired labour in the shambas, carrying out such duties as fruit-picking, water-fetching, etc., but refusing to sedentarize and cultivate on their own. A "parasitic" attitude is thus developing, radically transforming the traditional symbiosis between hunter-gatherers and sedentary people, made of reciprocal duties and interests (cf. Galaty 1986 for the Torrobo-Maasai relationship). The psychological effects of this marginality are often passivity and a sense of dejection, in their turn leading to drunkenness, which local - not always groundless - rumours frequently associate with the Dahalo (cf. Guenther 1986 for a description of similar disruptive effects of sedentarization coupled with economic marginality among the Bushmen of Botswana). On the other side, it could perhaps be argued that social marginality is slowing down the process of cultural and linguistic assimilation, which would be enhanced by an unconditioned acceptance of the new culture represented by agriculture and townlife.

5. Linguistic aspects of language shift We already mentioned the heavy borrowing of Swahili lexical items in Dahalo. It must be noted that until today all Swahili material has been to a considerable extent successfully assimilated in the grammar. Phonologically and morphologically unassimilated borrowing frequently occurs only in specific areas of the lexicon associated with "modern" cultural domains traditionally alien to Dahalo culture, such as agriculture and housing. In the following we shall provide examples from different parts of the grammar, mainly on the basis of Tosco (1991).

5.1. Phonological aspects of the impact of Swahili All the (very few) descriptions of Dahalo phonology illustrate the exceptional richness of consonants: Ehret (1980) arrives at a chart of 51 consonants,

Dahalo: An endangered language

145

Elderkin (1974 and 1976) works on the basis of a 57-consonants inventory, while Tucker - Bryan - Woodburn (1977) record "only" 39 consonants. Now, similar arrays of phonemes are always suspicious, and indeed we think that true phonemes and allophones have been merged in all previous accounts, together with, probably, idiolectically different realizations. This is probably true of the presence of both a bilabial /ß/ and a labiovelar /v/ fricative, proposed by Ehret (1980), and which would make of Dahalo a unicum among the languages of the world (cf. Maddieson 1984: 46); another case in point is the presence of the dental fricative /δ/, which, when coupled with an alveolar implosive /d'/ and with both dental and alveolar plain stops would make the Dahalo system at least unusual (cf. Maddieson 1984: 32 for relevant figures and the constraints on the presence of different stops in the same places of articulation). On our side, we regard /ß/ and /5/ as intervocalic allophones of Ibi and /d/, respectively. 4 The simultaneous presence of dental and alveolar stops is indeed a feature of Dahalo, but this typological oddity is well explained when one departs from the idea that alveolarity is original and dentality the acquired feature, as proposed by Ehret (1980) in his reconstruction of Proto-South-Cushitic. In order to account for Dahalo reality (which knows overwhelmingly more dentals than alveolare), Ehret has then to propose a rule changing alveolars to dentals in Dahalo. Nurse (1985) supports the same idea; he explains dentality - said to be occurring "in a minority of languages worldwide" (Nurse 1985: 243), which seems to us unproved - as an areal feature, arisen firstly in Dahalo and probably in Boni and then passed to Bantu languages such as Elwana, Pokomo and Northern Swahili. This view is consistent with Nurse's hypothesis of a long-term influence of Dahalo over the Kenyan coast (and of an even more ancient influence of Southern Cushites over most of Kenya, as proposed in Nurse 1986). To discuss the soundness of this reconstruction would lead us too far away from what is our present aim; for what concerns Dahalo, we prefer to consider dentality as original (Cushitic languages as a whole have dental rather than alveolar stops), and alveolarity as borrowed from Bantu (in Dahalo as in the South Cushitic languages): only dentals occur in suffixes and are still predominant in the lexicon; alveolars, on the other hand, are found only in Swahili words or in items of doubtful origin, but which in any case do not look Cushitic and are therefore to be regarded, at least tentatively, as loans. 5

146

Mauro Tosco

In our view, the phonological system of Dahalo can only be understood as the reflex of a process of heavy and continuous borrowing, especially from Swahili. Now, the phonological system of Dahalo is indeed rich and asymmetrical, and these characteristics are a function of the mixed nature of its lexicon, in which multiple borrowings have not (yet) been equally absorbed. In other words, it is simply unrealistic to look at the "sounds" of Dahalo as constituting a system: they do not, because different systems are always present in any utterance of a Dahalo speaker. As it is probably the case in any bi- and multilingual community, the boundaries between the systems, initially strongly perceived by the speakers, gradually fade away, as one of the systems conquers new communicative fields; while in the first phase the loans are adapted to the phonological system of the target-language, successively, as more and more loans creep in, they are simply stored, unanalyzed. In other words, looking at Dahalo phonology one has to distinguish between Lehnwörter and Fremdwörter, and their respective phonological systems. Therefore, the size of the consonantal inventory of Dahalo is a function of the history of the Dahalo people: like all hunter-gatherers, the Dahalo have probably always been subject to multiple, long and deep influences, resulting in heavy borrowing, and, probably, recurrent language shifts; in recent years we have "witnessed" the shift of the Yaaku from their Cushitic language to Maa and, some centuries ago, the shift of the Boni to Somali (Ali 1985 has proposed that the Boni are just Dahalo who switched to a Somali dialect more precisely, the Karre dialect - some centuries ago). As a consequence of all this, many of the consonantal phonemes of Dahalo have a very low rate of occurrence, or are subject to positional restrictions. The following phonemes have been found in initial position only: /dl/ /j'/ /k w 7 /ts'/

(three items) (three items, two Swahili loans) (four items) (three items)

The following phonemes have not been found in initial position: /d/ Ibi

(the only voiced bilabial stop in initial position being the implosive lb'/) Other very rare phonemes without positional restrictions are:

Dahalo: An endangered language

/c'/ Ici /i w / /p'/ /s/ /tl'/ lyl Fld/

147

(found in two items in initial position - probably as an allophone of Ici - and in one item medially) (found only in a few items in initial position and medially in one item - probably a reduplicated stem: cikocik- 'to saw') (found in one item only: iwah- 'to pitch') (found only in initial position - except in one instance in a reduplicated stem and stem-finally in k'arap'- 'to cut' and t'up'- 'to leak') (found in ten items, all of them loans, mainly from Swahili, but also from Somali) (found in six items in initial position and stem-finally in vati'- 'to return' and huntl- 'to chew') (found only in yááyo 'mother (alive)') (two items, loans from Northern Swahili; freely alternating with / n dr/ and ñ t / (as it does in Amu and Baraawa dialects of Swahili).

The labialized velars (probably all of loan origin) and several prenasalized phonemes are for the most part of limited occurrence; in particular, /"g w / occurs only in háng warang wára 'centipede' ; /"j/ in kipunju 'place where the maize is seasoned', /™dz/ in three items; /"/ / (prenasalized dental click) in five. Other phonemes occur only in unassimilated Swahili loans (and are therefore not to be considered as parts of the phonological system of Dahalo), such as /z/, which is constantly replaced by làl - phonetically 161 intervocalically (see above) - in assimilated loans (as in the Swahili word kaskazi 'drought, hot season', realized both as kásikazi and kásikadi). The phonological shape of words has been so much influenced by Swahili that today all Dahalo words end by one (short) vowel. This constraint is surely the effect of heavy Swahili loaning, as no other Cushitic language has anything similar.

5.2. Morphological aspects of the impact of Swahili In the domain of morphology, a first effect of loaning has been the fading away of the deeply rooted Cushitic gender distinction (masculine/feminine). Today, Dahalo nouns do not have any affix which could be interpreted, at least synchronically, as gender-marking. Given the existence of separate masculine (3M) and feminine (3F) forms of the determinants and of the 3 singular verbal forms, an obvious possibility is that nouns are subcategorized according to gender. Now, only semantically feminine subject nouns (i.e. nouns whose referent has female sex) govern feminine forms of the determinants and, as

148

Mauro Tosco

subjects, of 3S verbal forms; this seems to have already been recognized by Elderkin (1976: 292): "Gender in Dahalo is natural". All evidence seems to indicate that the category of gender is gradually dying out in Dahalo. That we are facing here the fading away of a single category, and not of nominal morphology as a whole - and that therefore bilingualism is probably the cause - is demonstrated by the fact that number marking is vital in Dahalo, even more than in those neighboring Cushitic languages (Oromo, Southern Somali dialects, Boni) in which overt expression of the plural on the noun is often avoided if number can be recovered semantically or is already marked (e.g., on the verb). In number marking, the impact of loans has shifted the balance between the different types of Dahalo plurals, which can take the form of: a. b. c. d. e. f.

partial reduplication; partial reduplication plus suffixation suffixation change of final vowel change of tonal pattern (without change in segments) irregular formation

While "original" Dahalo nouns are pluralized by any of the above types, loanwords never form their plural by types b., e. or f., and only very occasionally by type d. Indeed, Swahili loanwords are consistently pluralized through reduplication (type a.) and, less frequently, suffixation (type c.). The high percentage of loans in Dahalo vocabulary has as its major consequence that plural types a. and c. are the most frequent (accounting for, respectively, about 34% and 44% of all plurals in our corpus). Moreover, many nouns which form their plural through the change of the tonal pattern (type e.) have another alternative plural in types a. or c. One may say that plurals of types d. (change of final vowel) and e. (change of tonal pattern) are no longer productive in Dahalo. Still another possibility in number marking is to recur to Swahili plural prefixes. This frequently occurs with phonologically unassimilated loans, and especially with adjectives, which never change their final vowel to -i, as most Dahalo adjectives do. Note the following short sentence: (1)

?addókwa mambo ?ákkale day-this-M things all 'everything is new today'

mapya new-P 6

where mapya is in agreement with mambo 'things, affairs'.

Dahalo: An endangered language

149

As we know it often occurs in bilingual contexts, it may well be that the presence of an unassimilated noun as mambo (a Swahili plural) enhances the adjective in its Swahili plural form; normally, mpia would rather be left unchanged. This means that, when loans enter in a language as phrases, their "strength", i.e. their possibility to be accepted and to resist assimilation, is increased. With nouns my informants were normally able to provide a true plural, or even more than one - as it often happens in Cushitic - , and even for unassimilated singulars, but occasionally the Swahili plurals were used (and they were often given first, therefore demonstrating that they were more "normal" than the use of the Dahalo pluralizing devices); an example is zumba 'rooms', as plural of cúúnba, alongside the proper Dahalo plural cúúnbabbi. Bilingualism can result in the substitution of native Dahalo words with Swahili equivalents, especially if the word is a member of a set in which Swahili is predominant, as the numerals: Dahalo numerals are preserved until '5'; from '6' onwards Swahili numerals are used; but in the following sentence the Dahalo numeral dáwatte '5' has been substituted by Swahili tanu; naturally, ?amma is Swahili, too: (2)

Mfi saîàla ?amma tanu ratstsi men 4 or 5 go-PrfPs-3P 'four or five men went into the bush'

gudde bush

5.3. Syntactic aspects of the impact of Swahili Dahalo has up until today retained the Cushitic SOV basic word-order, but SVO is frequent: while in many cases it can be said that this permutation serves a topical function, sometimes it is just a reflex of Swahili SVO wordorder: (3)

?á+na+te váha O+PERF+AFF see-PrfPs-lS Ί have seen the girl'

dééla girl

Swahili ni is perhaps at the origin of Dahalo ?ini, which is best interpreted as a focus marker (the initial syllable RH, which serves the purpose of bearing the high tone, is deleted when the particle is suffixed); while it would be possible to analyze Tini as an invariable copula (which would be more in line with the value of Swahili ni), we must note that a true copula seems to occur in Dahalo -sú, which is sometimes found in nominal sentences, otherwise without overt marking.

150

Mauro Tosco

The following sentences illustrate the use of Tini: (4)

nát'etstsa ñ't[o] Tini woman-F that-F FOC 'this woman is Dahalo'

d'aháálotstsi D.-F

(5)

nat'á+t[o] Tini te lúbo woman+that-F FOC ÄFF beat-Pr-lS Ί am going to beat that woman'

(6)

?ááta fini dáákotstsi you-S FOC D.-F 'are you (SF) Daako?'

(7)

háájo ?úku Tini man than-M FOC 'who is that man?'

(8)

dééla ?ít[o\ ?íni girl that-F FOC 'who is that girl?'

jíkotu who-M jíkotstsi who-F

Swahili ni has better retained its copular value in Dahalo negative sentences (in which ni is changed to si in Swahili), where it is suffixed to the negative selector b'a (in past sentences) and the negative+irrealis selectors b 'á+ka (in non-past sentences), as a quicker and invariable alternative to the use of the Dahalo verb Peek- 'to be': (9)

mini ?úku ?íni house this-M FOC 'that house is not old'

b'á+ka+ni NEG+IRR+FOC

(10) b'á+ka+ni+va+?a NEG+IRR+FOC+PAST+O 'they were not men'

háfi men

(11) b'á+ka+ni+va+?a NEG+IRR+FOC+PAST+O Ί was not a man'

háájo man

b'ára+ani before+GEN

As can be seen in the last two sentences, this construction is ambiguous as far as the subject is concerned; cf. the similar explicit sentences with the Dahalo verb ?eek-: (12) b'á+ka ?ééketo vine NEG+IRR be-Pr-2S good 'you (SM) are not good'

Dahalo: An endangered language

(13) b'á+ku+va ?eeka NEG+you-SM+PAST be-Inf 'you (SM) were not a man' (14) b'á+va ?eeka NEG+PAST be-inf 'he was not a man'

háájo man

(15) pápi b'á+ni+va we NEG+us+Past 'we were not men'

?eeka be-Inf

151

háájo man

háíi men

A Dahalo conversation is normally interwoven with (Northern) Swahili expressions, such as b'asi 'well!, that's all!', ndro 'well!', héla 'come on!', etc.; other Swahili elements is present in positional terms (adverbs, prepositions): saa 'time, hour', saa sita 'noon', sasa 'now', sikuzote 'always', zamani 'formerly', ?ande 'outside' (cf. Baraawa 71 de), mpaka 'until', and in WH-words: kwani 'why? because', nbona 'why?' (Swahili mbona). MyersScotton (this volume) points to the attractiveness for borrowing of these "discourse markers, lexemes hovering near the border between content and system morphemes. These morphemes have 'positional salience' in that they may occur in initial position in a conversational turn, or at least as heads of constituents (e.g., as prepositions in PPs). Also, of course, they are often free forms" (Myers-Scotton, this volume). Actually, many sentences of a Dahalo text often begin with b'asi, sasa, n dro, etc. In our opinion, the importance of the extensive use of these discourse markers as a hint of deep borrowing and bilingualism has often been neglected. In the case of a bilingual society, it can become fashionable to use them in the LI from the very beginning of the contact, when the material from L2, although extensively borrowed, has not yet been assimilated in LI and the two languages are kept distinct.7 It does not seem possible, therefore, to regard the borrowing of the "discourse markers" as secondary in respect to the borrowing of content words as nouns and verbs. Moreover, these words, being free forms, can be accepted in the language without being assimilated phonologically, but they can have at the same time deep repercussions in the syntax, if, for example, a language with postpositions borrows prepositional terms, as it is the case in Dahalo. In this way, these positional words can act as a sort of "advanced column" toward the relexification of the language. Naturally, Swahili loans enter in Dahalo both as isolated words and as idioms, such as nafáási patad- 'to get a chance, an opportunity' (from Swahili nafasi and -pata)', under the influence of Swahili, a Dahalo item can shift and

152

Mauro Tosco

extend its meaning in order to cover the range of meanings of a corresponding Swahili word; it is the case of the verb pah- 'to hit, beat', used with an object in order to translate many Swahili expressions built with the verb -piga; we have even recorded ripoti pah- 'to report'. Taking into consideration both assimilated and unassimilated foreign material, the net result is often some kind of Mischsprache, in which Swahili is framed into Dahalo; but it may be surprising to know that Swahili loans account for about 15% only of our Dahalo glossary (made up of nearly 1,000 words. 8 This Swahili-borrowed section of the glossary is made up for the most part of nouns (63%), then follow the verbs (23%) and the "discourse markers" (14%). This would seem to rule out the possibility for Dahalo to become a true Mischsprache. Rather, if the sociolinguistic conditions which we tried to sketch persist in the next years, Dahalo will soon be on the verge of extinction, the economic and social conditions which so far have permitted this little language and people to exist having disappeared.

Notes 1. This work originates from a two-month period of field work among the Dahalo in the Lamu District (Coast Province, Kenya), in January-March 1988, which was made possible by a scholarship granted by the Istituto Italo-Africano. The main aim of our work was to collect material on the morpho-syntax of this language. The morphology has been dealt with by Elderkin (especially 1974) and a certain amount of lexicon has been presented in various sources, most notably Ehret (1980) in his reconstruction of Proto-South-Cushitic. Our principal informant was Dawa Hamadi, 30 years old, niece of one of Zaborski's informants. Although she had been living in Lamu town for many years, she was born in the area of Mkunumbi and was fluent in Dahalo. Like most Dahalo, she was bilingual in (Amu) Swahili and spoke no other language. The research was carried out in Lamu town and in the surrounding areas of Mokowe and Mkunumbi. We want to express our thanks to: the Istituto Italo-Africano for the grant which made this research possible; Mr. Athman Lalli, the Curator of the Lamu Museum, who helped us in any possible way during our stay in Lamu, and, naturally, our informants. 2. We thank Mr. Athman Lalli for having provided us with this precious source of data, and Mr. Ed Loving (Kenya Working Group, Language Programs Coordinator) for having granted us the permission to quote it. 3. This would apply only to the last centuries, if Ali's (1985) hypothesis about the origins of the Boni proves correct. See section 5.1.

Dahalo: An endangered language

153

4. In our informant's speech [β] never occurs initially; e.g., Elderkin's (1974: 81) ßuddaö- is for us buddad- 'to ask'. Concerning the use of the subscript [ J , see note (5) below. One can note that Tucker - Bryan - Woodburn (1977: 320) noted the existence of a voiced velar fricative /γ/, unknown to other authors (and which we never heard). It can perhaps be seen as a "negative" proof of the allomorphic character of the other two fricatives /ß/ and /δ/, too. 5. Coherently with this approach, we mark alveolarity (with subscript [ J ) and leave dentality unmarked. 6. The following tags are used in the glosses to the Dahalo sentences: 0: particle-complex marker AFF: affirmative selector F: feminine FOC: focus marker GEN: genitival postposition Inf: infinitive IRR: unrealised action selector M: masculine NEG: negative selector Ρ: plural PAST: past selector PERF: perfect selector Pr: present paradigm PrfPs: perfective past paradigm S: singular + a morpheme boundary 7. One can here mention Maltese, in which a conversation is frequently interwoven with English expressions such as "all right!", "that's all!", even if the English borrowed in Maltese has not been assimilated yet, contrary to the Siculo-Italian material, borrowed at an earlier date and for a longer time. 8. This is a rough - and probably underestimated - figure, obtained by ruling out many probable early loans from Swahili (or some older forms of it), fully assimilated and no more recognizable today as such, not even to the Dahalo speakers. Loans from Lower Pokomo, Giriama, as well as forms for which a parallel can be found in ChiMiini (Baraawa Swahili) are likewise excluded. To draw up an etymological dictionary of Dahalo would require the joint work of both Cushitists and Bantuists (not to speak of Khoisanists), and it is well beyond our possibilities. Ehret - Elderkin - Nurse (1989) is a first step in this direction, although flawed by the "Southern Cushitic" bias of the authors, which makes them recur to Proto-South-Cushitic roots in cases in which borrowing from an Eastern Cushitic language (as Somali and Oromo) is more probable. Still in other cases, the authors are inclined to see in a Dahalo word not a Bantu loan, but, conversely, the source of words found in various Bantu languages of Kenya; this view is consistent with Nurse (1985 and 1986).

154

Mauro Tosco

References Ali, Mohammed Nuuh (: Maxamed Nuux Cali) 1985 History of the Horn of Africa 1000 B.C.-1500 A.D.: Aspects of Social and Economic Change between the Rift Valley and the Indian Ocean. [Unpublished Ph.D. thesis, University of California, Los Angeles.] Ehret, Christopher 1980 The historical reconstruction of Southern Cushitic phonology and vocabulary. Berlin: Dietrich Reimer. Ehret, Christopher - Edward Derek Elderkin - Derek Nurse 1989 "Dahalo lexis and its sources", Afrikanistische Arbeitspapiere 18: 5^-9. Elderkin, Edward Derek 1974 The Phonology of the Syllable and the Morphology of the Word in Dahalo. [Unpublished M.A. thesis, University of Nairobi.] 1976 "Southern Cushitic", in: M. Lionel Bender (ed.), The Non-Semitic Languages of Ethiopia. East Lansing: Michigan State University, 278-297. Galaty, John G. 1986 "East African hunters and pastoralists in a regional perspective: An 'ethnoanthropological' approach", in: Franz Rottland - Rainer Vossen (eds.), 105-131. Guenther, Mathias 1986 "Acculturation and Assimilation of the Bushmen of Botswana and Namibia", in: Rainer Vossen - Klaus Keuthmann (eds.), Contemporary Studies on Khoisan 1. Hamburg: Helmut Buske, 347-374. Heine, Bernd 1977 "Bemerkungen zur Boni-Sprache (Kenia)", Afrika und Übersee 60: 242-295. 1981 The Waata dialect of O romo. (Language and Dialect Atlas of Kenya 4). Berlin: Dietrich Reimer. 1982 Boni dialects. (Language and Dialect Atlas of Kenya 10). Berlin: Dietrich Reimer. Maddieson, Ian 1984 Patterns of sounds. Cambridge: Cambridge University Press. Nurse, Derek 1985 "Dentality, areal features, and phonological change in Northeastern Bantu", Studies in African Linguistics 16(3): 243-279. 1986 "Reconstruction of Dahalo history through evidence from loanwords", in: Franz Rottland - Rainer Vossen (eds.), 2: 276-305. Rottland, Franz - Rainer Vossen (eds.) 1986 Proceedings of the International Symposium on African Hunter-Gatherers. Sankt Augustin, 3-5/1/1985, in: Sprache und Geschichte in Afrika 7(1,2.).

Dahalo: An endangered language

155

Sasse, Hans-Jürgen 1981 "Die kuschitischen Sprachen", in: Bernd Heine - Thilo C. Schadeberg Ekkehard Wolff (eds.), Die Sprachen Afrikas. Hamburg: Helmut Buske, 187-215. Rilling, Art 1986 The Boni/Dahalo report. (Sociolinguistic Surveys in Selected Kenyan Languages.) Nairobi: Summer Institute of Linguistics, Kenya Working Group. Stiles, Daniel 1988 "Historical interrelationships of the Boni with pastoral peoples of Somalia and Kenya", Kenya Past and Present 20: 38-45. Tosco, Mauro 1991 Λ grammatical sketch of Dahalo, including texts and a glossary. (Kuschitische Sprachstudien 8). Hamburg: Helmut Buske. 1992 "The classification of Dahalo: Another perspective", in: Giorgio Banti (ed.), Proceedings of the Second International Symposium on Cushitic and Omotic Languages, Turin, 16-18 November, 1989. Napoli: Istituto Universitario Orientale (in press). Tucker, Archibald Ν. - Margaret A. Bryan - James Woodburn 1977 "The East African click languages: A phonetic comparison", in: Wilhelm J.G. Möhlig - Franz Rottland - Bernd Heine (eds.), Zur Sprachgeschichte und Ethnohistorie in Afrika. Berlin: Dietrich Reimer, 300-323. Zaborski, Andrzej 1986 The morphology of nominal plural in the Cushitic languages. (Beiträge zur Afrikanistik 28.) Wien: Veröffentlichungen der Institute für Afrikanistik und Ägyptologie der Universität Wien. 1987 "Remarks on recent developments in Cushitic", in: Giuliano Bernini Vermondo Brugnatelli (eds.), Atti della 4a Giornata di Studi Camitosemitici e Indoeuropei. Milano: Unicopli, 219-227.

Language death and the origin of strata: Two case studies of Swahili dialects Wilhelm J. G. Möhlig

1. Introduction On the basis of two case studies concerning the Swahili dialects Chwaka and Chirazi (Chifundi), both spoken in South East Kenya, it will be discussed in this paper to what extent language death has contributed to language stratification and how language strata may be taken as evidence for events of language death not otherwise attested by historical sources. The term "stratum" as such implies a historical perspective. In this sense, stratum is generally defined as a complex of surviving features typical of a former language which has been replaced (Bynon 1977: 252). Thus, already the definition of the term has a direct link with language death as a historical event which may have created it. The criteria for a linguistic stratum may be phonal, lexical or grammatical. For the Africanist, who normally has no historical documents of former stages of language development at hand, the empirical base for stratificational analyses consists of data which have been collected among modern languages still being spoken. The language material for the present case studies was collected within the continuing research program of the Language and Dialect Atlas of Kenya being carried out by Bernd Heine and myself since 1970 (Heine - Möhlig 1980). The procedure of stratificational analysis adopted in this paper is based on several methods which have been applied by others and by myself in different contexts and with different aims. The first method is that of dialectometry, i.e. a method which measures the dialectal distances between distinct, but similar idioms of specific areas on linguistic grounds (for more details see Guarisma - Möhlig 1986). The indices of dialectal distance between two idioms often differ from each other depending on the nature of the linguistic features which have been taken as

158

Wilhelm J.G. Möhlig

the dialectometrical gauge. These divergencies turned out to be a reliable diagnostic instrument to discover otherwise hidden linguistic strata (Möhlig 1984-1985), for they reflect historical events which have touched the linguistic sub-structures, such as the sound inventory, the various semantic sections of the lexicon or the grammatical sub-systems, in different ways. Another method from which elements are used here is the historical comparative method. In order to be able not only to reconstruct a common proto language for the languages compared, but also the processes which have led to the linguistic diversity between them, that is the actual sound shifts, the comparative method has been paired with the dialectological method, in particular with its principle not to leave any linguistic no-man's land between the idioms compared (Möhlig, in: Heine - Möhlig 1980: 38-53). Finally, a qualitative method of comparison has been applied. It takes into account the semantic factors rendered by each item, thus constituting the more quantitative results of the other two methods. If, for instance, two languages compared have a certain percentage of the lexicon in common, it is historically relevant, whether this percentage refers to the lexicon in general, to the basic or to the modern cultural vocabulary. Another aspect of this method is the comparison of semantic fields, like terms of agriculture, iron work and house building, or of functional fields, like terms referring to the sphere of women, men, children or elderly people. The following two case studies concern idioms which belong to the Swahili group of dialects spoken in Kenya. As long as no dialectologically collected material of the Swahili dialects spoken in Tanzania is available, we tentatively classify the Kenyan Swahili dialects into three subgroups (Möhlig 1984-1985: 265 ff.): (1) Northern Kenyan Swahili, (2) Central Kenyan Swahili and (3) Southern Kenyan Swahili. In order to avoid the constant repetition of these cumbersome labels, we use in this paper the group names "NK Swahili", "Mvita" (for Central Kenyan Swahili) and SK Swahili.

Two case studies of Swahili dialects

Map 1. Location of languages discussed

159

160

W i l h e l m J.G. M ö h l i g

2. First case study: Chwaka 2.1. The language and its speakers Chwaka is spoken in a small village of about 500 inhabitants in the centre of the peninsula of Shimoni. In spite of its central situation, the area is comparatively remote from the main roads and therefore difficult to reach. At first glance, the Chwaka people give a comparatively backward impression. One can describe them as almost xenophobes. They consider themselves to belong to the Swahili culture. This is confirmed by their way of life which is Muslim in the typically coastal form. The Chwaka idiom is distinct from all surrounding dialects and languages and clearly a linguistic entity of its own. Although its speakers consider it to be a dialect of Swahili like Vumba or Chirazi (Chifundi), one can argue whether it doesn't in fact belong more to the Mijikenda group.

2.2. The dialectometrical context Dialectometrical analyses exhibit the following relations of Chwaka (L. = lexical indices of nearness, P. = phonological indices of nearness): Mvita: Chirazi: Vumba: Pate:

L.79, L.80, L.82, L.71,

P. 64 P.77 P.75 P.52

Tikuu: Digo: Duruma:

L.69, L.77, L.72,

P.54 P.94 P.94

For the purpose of evaluation we compare the indices of Digo with: Chirazi: Vumba: Duruma: Mvita:

L.70, L.72, L.87, L.74,

P.73 P.67 P. 100 P.60

The chart shows three pecularities: First, the phonological indices between Chwaka and the neighbouring Swahili dialects are significantly lower than the lexical indices between the two idioms. Secondly, the highest phonological indices of Chwaka are shared with the Mijikenda dialects. Thirdly, the highest lexical indices link Chwaka with its geographically closest Swahili neighbours Vumba and Chirazi. This dialectometrical pattern strongly indicates a genetic affiliation of Chwaka to Mijikenda rather than to the Swahili group of dialects, as the people themselves believe. On the basis of our general experience with the historical interpretation of dialectometrical indices we can take it almost as

Two case studies of Swahili dialects

161

a rule that, in the case of diverging phonological and lexical indices, higher phonological values indicate genetic links, whereas higher lexical values exhibit relations based on contact. We shall try to substantiate this assumption by a historical qualitative analysis on the basis of phonetic, lexical and morphological comparative data.

2.3. Historical phonological analysis 2.3.1. The modern sound system

Ρ "ph b'

t °th d' °ts "dz θ - "s δ - "ζ "l/r - "η "nd "nz

- Í -

οβ

7 .• °v

"m 0 mb

• 0

mv

-

-

-

-

"k "kh

j'

-

V

•

°y

-

•

"ny -

• "te sh -

V "ng

In anticipation of the results of the following historical analysis, the sounds which can genealogically be traced back to a Proto North Eastern Coastal Bantu are marked with the symbol There are five distinctive vowel phonemes and a word accent on the penultimate syllable. Whether tone plays a distinctive role is not certain; our limited data on Chwaka suggest that it does not.

2.3.2. Reconstructed sound shifts This present sound system is the result of several sound shifts which, according to our historical dialectological method, can be arranged in the following chronological sequence: Phase I: Spirantization of plosives before tense vowels and merger of high vowels: 7 V > 5 V. i.e. *pl/*bl > fi/vi, *mpl/*mbl > **mfi/mvi, *tl/*dl > si/zi, *ntl/*ndl > **nsi/nzi, *kl/*gl > si/zi, *nkl/ngl > **nsi/nzi - *pU/*bu, *tU/*du, *kU/*gU > fu/vu, *mpU, *ntU, *nkU > **mfu-*mpU, *ntU, *nkU > **mfu, *mbU, *ndU, *ngU > mvu (I for tense i and U for tense u).

162

Wilhelm J.G. Möhlig

Phase II: Loss of prenasality with voiceless fricatives and affricates and, as a consequence, merger of the originally prenasalized fricatives with their homorganic non prenasalized counterparts, i.e. **mf/f>f, **ns/s > s, *nts/ts > ts. Phase III: Strengthening of the oral component of the voiceless prenasals by aspiration, i.e. *mp/*nt/*nk > **mph/**nth/**nkh. Phase I Va: Lenition of voiced anterior stops: *b > b> glide/zero, *d > d> l/r. Phase IVb: Lenition of voiceless anterior stops *p > φ > β, *t > hr > hR > (The last step probably as a dissimilatory process to strengthen the phonological contrast between Irl ex *d and IhRJ ex *t). Phase V: Lenition of velar plosives in front of the vowels i and e, i.e. *ki/e > tshi/e, *gi/e > ji/e. Phase VI: Loss of prenasalisation with voiceless aspirated stops, i.e. **mph/ **nth/**nkh > ph/th/kh. 2.3.3. Discussion of the sound shifts The character and sequence of these sound shifts correspond entirely to those which can be reconstructed for the southern languages and dialects of Mijikenda, in particular for Digo and Duruma. In addition, the proto soundsystem reconstructable for Chwaka is identical with that of all Mijikenda languages. In other words, the inferences which have earlier been drawn from the high phonological dialectometrical indices, are fully confirmed by the historical dialectological analysis of the phonological system. In this perspective, it appears that Chwaka is indeed a member of the Mijikenda language group. The remaining sounds of the present sound system of Chwaka which do not originate from Proto Mijikenda can be identified as borrowings from two sources: (1) θ, δ, gh and h derive from Arabic. (2) All the other sounds originate from Sk Swahili; the Arabic loans were most probably mediated by these dialects.

2.4. Historical lexical analysis 2.4.1. The empirical base The lexical analysis is mainly based on our 600-word list, which we have been using with slight modifications throughout our dialectological studies in East

Two case studies of Swahili dialects

163

Africa. The linguistic material of this list has been dialectologically compared and evaluated for all the idioms spoken in the area between Mombasa and the Tanzanian border. The procedure resulted in four groups of words: (1) items shared by Chwaka with Chirazi and Vumba, but not with Mvita; (2) items shared by Chwaka with Mvita and the Northern Swahili dialects, but not with Chirazi and Vumba; (3) items which occur neither in the Northern nor in the Sk Swahili dialects. (4) identifiable loans from Unguja (Standard Swahili).

2.4.2. Items shared with Chirazi and Vumba Chwaka shares with Chirazi and Vumba to the exclusion of the NK Swahili dialects and Mvita about 6% of the 600-word list. The majority of the items belong to the basic vocabulary. Within this conceptual grouping, these words are characterized by a reduced or low frequency of use in communication, and it is important to note that almost all these items also occur in Mijikenda, particularly in Digo and Duruma, for example: Chwaka

Vumba

Mwita

Digo

funda kisukusuku chiganja phazi

fiinda kisukusuku kiganja phazi

tavu funda kivi kisukusuku kitanga fumba ngoyo fundo la mguu

Duruma

NK Swahili

tsavu chikoko gandya ngunyu

itavu kivi kichanga ito la guu

'cheek' 'elbow' 'palm' 'ankle'

2.4.3. Items shared with Mvita and N K Swahili The group of lexical items shared with Mvita and the NK Swahili dialects to the exclusion of Chirazi and Vumba is considerably smaller than the first group. It concerns less than 1% of the 600-wordlist. A few examples are: Chwaka

Vumba

Digo

Mvita

NK Swahili

chiwamba

ukuta

ukuta

ukuta

kiwamba(z)a

'wall'

-tsepuza

-chißuza

tseßula

-tepuza

-tepuza

'to sprout'

-chia khasi

-piga khafi

-piga kh.

-tia kh.

-chia kh. (Tikuu)

(Amu)

'to paddle'

2.4.4. Exclusive items About 12% of the lexical items within the range of the 600-wordlist are not shared with any of the Swahili dialects, but three quarters of these exclusive

164

Wilhelm J.G. Möhlig

items occur in Mijikenda, again with a clear distributional focus on Southern Mijikenda. Semantically, more or less all these words belong to the basic vocabulary like parts of the body, basic activities, natural phenomena or to the basic cultural vocabulary such as domestic activities and domestic animals, for example: Chwaka

Vumba

Mvita

Digo

cheru mumiro -saala -ima lungunzu -sana -vug a ng 'onzi mnyau dia

kidevu khoo -kaa -simama alfajiri -fua -pika khondoo phaka mbwa

kidevu khoo -kaa -simama alfajiri -fua -pika khondoo phaka mbwa

cheru mumiro -saala -ima ligunzu -sanya, -fula -bia, -jita ng 'onzi mnyau diya

'chin' 'throat' 'to remain 'to stand' 'dawn' 'to forge' 'to cook' 'sheep' 'cat' 'dog'

2.4.5. Loans f r o m Standard Swahili About 10% of the lexicon within the range of the 600-wordlist can be identified, on phonological or structural grounds, as loans from Unguja or Standard Swahili. The items in question refer to social concepts and activities, religion, commerce, judiciary, qualities and concepts of modern culture, for instance: Chwaka

St. Swahili

kabira harusi luga sheria kaburi rehema

kabila harusi lugha sheria kaburi rehema

'tribe' 'marriage' 'language' 'law' 'grave' 'mercy'

2.4.6. Discussion of the results of this analysis As the historical result of the qualitative lexical analysis, we may resume that Chwaka underwent strong lexical influences from its Swahili speaking neighbours Vumba and Chirazi. In comparison with its closest relatives Digo and Duruma, this led to a stronger linguistic alienation from the traditional Mijikenda stock. But important parts of the original Mijikenda lexicon have

Two case studies of Swahili dialects

165

still been preserved. The theoretical alternative that the sequence of historical events was the other way round, i.e. a Swahili stock being strongly influenced by Mijikenda, can be clearly excluded on the evidence rendered by the semantic quality of the word material in question: The "Mijikenda portion" of the lexicon is by far the more traditional or archaic one, whereas the "Swahili portion" reflects historically younger concepts. If we compare the lexical relations of Chwaka with Mvita and the NK Swahili dialects, we find the higher indices with the geographically remoter dialects in the North. At first glance, this could lead to the assumption that, in former times, the geographical distribution of the dialects was different, for instance Chwaka being spoken in the vicinity of Tikuu or Siu. However, another hypothesis appears to be more plausible, namely that the lexical items shared by Chwaka and the NK Swahili dialects at the exclusion of Mvita were formerly used also in Mvita, but became obsolete there under the relatively recent influence of Unguja or Standard Swahili. In fact, very old speakers of Mvita still remember having used these obsolete items, and the older poetry in Mvita also attests their former usage. (Abdulaziz 1979, Miehe 1979).

2.5. Historical analysis of the pronouns The analysis is based on: (1) the substitutives (independent pronouns), (2) the possessive pronouns, (3) the demonstratives and (4) the interrogatives.

2.5.1. Substitutives and Possessives 2.5.1.1. Substitutives

Chwaka

Vumba

Chirazi

Mvita

Digo

mimi

mimi

mie

mimi

mimi

l.sg.

wewe

wewe

wee

wewe

uwe

2.sg.

yeye

yeye

yee

yeye

iye

3.sg.

sisi

swiswi

swiswi

sisi

swiswi/sisi

l.pl.

ng 'ming 'mi

ng 'ming 'mi

ng 'ming 'mi

ninyi

ng 'ming 'mi

2.pl

hao

wao

wao

wao

ao

3.pl.

166

Wilhelm J.G. Möhlig

2.5.1.2. Possessives Chwaka

Vumba

Chirazi

Mvita

Digo

-angu -ako -ake -ehu -enu -ao

-angu -o -we -eru -enu -ao

-angu -o -we -eru -enu -ao

-angu -ako -ake -etu -enu -ao

-angu -o -we -ehu -enu -ao

l.sg. 2.sg. 3.sg. l.pl. 2.pl. 3.pl.

2.5.1.3. Comparative commentary On the surface, the singular forms of the substitutive and the possessive appear to be identical with Sk Swahili, the plurals, however, with Southern Mijikenda. On a comparative basis, one would have expected that the 2nd and 3rd persons singular of the possessives were identical with the forms in Sk Swahili and Southern Mijikenda, i.e. -o and -e. Instead -ako and -ake are found. They may both be identified as loans from Standard Swahili.

2.5.2. The Demonstratives 2.5.2.1. Near Demonstrative Chwaka

Vumba

Mvita

Digo

hiyu hano huu hii

huyu hawa huu hii

huyu hawa huu hii

hiyu/yuno hinya/hano huno hino

2.5.2.2. Remote Demonstrative Chwaka

Vumba

Mvita

Digo

yuya hara hura hira

huyule hawale hule hile

yule wale ule ile

yuya hara hura hira

cl.l cl.2 cl.3 cl.4

cl.l cl.2 cl.3 cl.4

Two case studies of Swahili dialects

167

2.5.2.3. Referential Demonstrative Chwaka

Vumba

Mvita

Digo

hiye hao huo hiyo

huye hao huo hiyo

huyo hawo huo hiyo

hiye hao huo hiyo

cl.l cl.2. cl.3 cl.4

2.5.2.4. Comparative commentary Where the demonstratives of Mijikenda and Swahili are structurally different, Chwaka shares the Mijikenda forms. 2.5.3. Interrogatives The interrogatives of Swahili and Mijikenda in general do not differ much. Only the interrogative 'who' is represented in the area of comparison by three alternating forms: ani (= Southern Kenyan Swahili), nani (= Mvita and Standard Swahili), and yußi (= Southern Mijikenda). Chwaka shares the form ani with SK Swahili. 2.6. Stratificational evaluation of the historical analyses With reference to the later period of language history, Chwaka shows four strata. (1) Mijikenda * phonological structure * basic vocabulary (2)

S.K. Swahili

basic vocabulary

(3)

Arch. Mvita

addentai t basic cultural vocabulary

(4)

Standard Swahili * *

modern vocabulary pronoun paradigms

2.6.1. Stratum 1 The oldest of these strata can be genetically allotted to Southern Mijikenda. It manifests itself mainly in the phonological structure: 26 of the 36 consonants of Chwaka can be traced back in a straight genealogical line to the proto

168

W i l h e l m J.G. M ö h l i g

language shared with Southern Mijikenda. In addition, the major part of the basic vocabulary also belongs to the Mijikenda stock.

2.6.2. Stratum 2 The relative chronology of the reconstructable sound shifts does not show when the second stratum, which is of Sk Swahili origin, was founded. Since all Swahili dialects share the historically youngest sound shift with Mijikenda, namely the loss of prenasalization with voiceless aspirated stops, the SK Swahili stratum in Chwaka could have been founded either before or after this sound shift. Yet, a significant part of the basic vocabulary in Chwaka stems from SK Swahili. This speaks in favour of longlasting and/or intensive contacts.

2.6.3. Stratum 3 The third stratum in Chwaka can be identified as archaic kiMvita dating from a period before Mvita came under strong influence of Zanzibarian Swahili. This stratum is defined by the addentai t in the sound system and by significant influences in the domain of the basic cultural vocabulary.

2.6.4. Stratum 4 A fourth stratum originating from Standard Swahili is actually in the process of being formed. It is easily identifiable by its many loan words in all domains of modern life and by its influence on the paradigms of the substitutives and possessives.

3. Second case study: Chirazi 3.1. The language and its speakers Chirazi (Chifundi) is spoken in some small villages along the coastal line of the Funzi Bay, on Funzi Island and on the eastern part of Wasini Island. Also the villages on the mainland are far from the main road (see map). The Chirazi speakers mostly live on fishing and agriculture. They are Muslims of the Kenyan coastal type.

Two case studies of Swahili dialects

169

Together with Vumba, Chirazi forms the core of the Southern Swahili dialects of Kenya. In a genealogical perspective, it is correctly allotted to the Swahili group of dialects. Nevertheless, its stratificational structure reflects an event of its linguistic history which may be relevant for the discussion on the concept of language death in general. Methodologically we follow the same procedure as in the first case study.

3.2. The dialectometrical context Dialectometrical analyses exhibit the following relations of Chirazi:

Indices of Chirazi:

Indices of Mvita (for comparison):

with: Mvita: Vumba: Pate: Tikuu: Digo: Duruma: N.Mjk. Pokomo:

L.87 L.93 L.81 L.77 L.70 L.66 L.65-67 L65-67

P.83 P. 82 P. 67 P.68 P.73 P.73 P.64-69 P.62-67

L.90 L.87 L.81 L.74

P.80 P.83 P.77 P.60

L.71

P.«60

The chart reveals the following relational elements which will have to be scrutinized by qualitative analyses: (1) Chirazi shows remarkably high phonological indices with Mijikenda, in particular with Digo and Duruma. They are 5 to 6 points higher than those with the Northern Swahili dialects Pate and Tikuu. (2) In comparison with the other Swahili dialects, only the lexical indices of Chirazi with Mvita and Vumba fall within the range of 81 to 90, which is characteristic for all the Swahili dialects without Chirazi, whereas the lexical indices with Pate and Tikuu clearly lie outside this range. (3) If one takes for the phonological indices a general Swahili range of 80 to 100, the relationship between Tikuu and Chirazi remains outside, but the indices with the Southern Mijikenda dialects are inside this frame.

170

Wilhelm J.G. Möhlig

3.3. Historical phonological analysis 3.3.1. The modern sound system ρ b'

-

- t - th

7 °m °m¿> -

- θ - ö

- nd «mv -

- t - *th -

-

°s *z °l/r °n °nd °nz

y

-

V

- «y

-

gh

- >y

-

V

-

- h -

- °nj

The sounds which can be traced back genealogically to Proto Eastern Bantu are marked with As in Chwaka, there are five distinctive vowel phonemes and a word accent on the penultimate syllable.

3.3.2. Reconstructed sound shifts The phases of sound shift which can be reconstructed for Chirazi are the same as for Chwaka and Mijikenda up to phase V. In this, Chirazi differs from the sound shift pattern of Northern Kenyan Swahili, Mvita and Unguja. After phase V, Chirazi follows the Swahili pattern of sound shifts, i.e.: Phase I: Spirantization of plosives before tense vowels and merger of high vowels: 7 V > 5 V, i.e. *pl/*bl > fi/vi, *mpl/*mbl > **mfi/mvi, *tl/*dl > si/zi, *ntl/*ndl > **nsi/nzi, *ki/*gl > si/zi, *nkl/ngl > **nsi/nzi — *pU/*bu, *tU/*du, *kU/*gU > fu/vu, *mpU, *ntU, *nkU > **mfu - *mpU, *ntU, *nkU > **mfu, *mbU, *ndU, *ngU > mvu (/ for tense i and U for tense u). Phase II: Loss of prenasality with voiceless fricatives and affricates and, as a consequence, merger of the originally prenasalized fricatives with their homorganic non prenasalized counterparts, i.e. **mf/f > f , **ns/s > s, *nts/ts > ts. Phase III: Strengthening of the oral component of the voiceless prenasals by aspiration, i.e. *mp/*nt/*nk > **mph/**nth/**nkh. Phase IVa: Lenition of voiced anterior stops: *b > b> glide/zero, *d > d> l/r.

Two case studies of Swahili dialects

171

Phase IVb: Lenition of voiceless anterior stops *p > φ > β, *t > hr > hR > (The last step probably as a dissimilatory process to strengthen the phonological contrast between Irl ex *d and /hR/ ex */)· Phase V: Lenition of velar plosives in front of the vowels i and e, i.e. *ki/e > tshi/e, *gi/e > ji/e. Phase Via: Introduction of implosive voiced stops. Phase VIb: Replacement of **dz by j'. Phase Vila: Dentalization and fortition of alveolar affricates, i.e. **ts > t, **nts > nt, **ndz > nd. Phase Vllb: Restitution of a class of voiceless anterior stops, i.e. ρ and t. Phase VIII: The loss of prenasalization with voiceless aspirated stops.

3.3.3. Discussion of the sound shifts The historical qualitative analysis of Chirazi shows three types of sounds: (1) Sounds which can be traced back in a straight genealogical line to Proto Mijikenda. With 25 consonants out of 37 they form the major part of the sound system. (2) Sounds which derive from Proto Swahili, i.e. ρ, t, t, th, b\ d\ j' and nd. (3) Sounds which have been borrowed from Arabic, i.e. Θ, ö, gh, and h, most probably via other Swahili dialects.

3.4. Historical lexical analysis Within the lexical range of the 600-word list, Chirazi shows a comparatively high amount, namely about 6% of items which it shares only with Sk Swahili or Southern Mijikenda, but not with the NK Swahili dialects or with Mvita. As to their semantic properties, these items belong either to the basic vocabulary or to the basic cultural vocabulary, for example: Chirazi

Vumba

funda funda chiganja kiganja -uya -uya -futuka -futuka -lya -lya manga manga

Mvita

Digo

tavu funda kitanga fumba -rudi -ula -fura -futuka -la -rya muhogo manga

Duruma

NK-Swahili

tsavu gandya -ula -futuka -rya manga

shavu kitanga -toka -fura -la muhogo

'cheek' 'palm of hand' 'to return' 'to swell' 'to eat' 'cassava'

In contrast to this, the group of words shared exclusively with one of the NK Swahili dialects or with Mvita is rather small. Only four items, i.e. less

172

W i l h e l m J.G. M ö h l i g

than 1% of the 600-wordlist, could be identified as belonging to this group, and some of the items differ only in one phoneme. These are: Chirazi

Vumba

Mvita

Digo

Duruma

NK-Swahili

chanwa bubwi -winda tisiya

kinywa bubu -winja tisa

kanwa bubwi -winda tisia

domu bwibwi -winza chenda

mulomo bwibwi -inza chenda

kanwa bubwi/bwibwi (v jinda tisia

'mouth' 'dumb' 'to hunt' 'nine'

There are a lot more words which can be identified as Mvita borrowings or as forms influenced by Mvita, but these are also shared with Vumba, Chwaka or South Mijikenda, thus they are not significant for the definition of a special relationship between Chirazi and the North. Furthermore, there appears to be a strong overlay of Standard Swahili forms which is found not only in Chirazi but also in Vumba, Mvita and the other Swahili dialects. This strong impact of a third language upon all the Kenyan Swahili dialects has at least had the dialectometrical effect that the lexical indices within the latter group have risen considerably. In other words, by this still ongoing process, the dialectal distances between the Kenyan Swahili dialects have become less than they were before the impact of Standard Swahili.

3.5. Historical analysis of the p r o n o u n s 3.5.1. Substitutives and Possessives 3.5.1.1. Substitutives Chirazi

Vumba

Mvita

NK-Swahili

Digo

mie wee •yee swiswi ng 'ming 'mi wao

mimi wewe yeye swiswi ng 'ming 'mi wao

mimi wewe yeye sisi ninyi wao

mimi/imi wewe/uwe yeye/iye sisi/isi ninyi/inyi wao/iwo

mimi uwe iye swiswi/sisi ng 'ming 'mi ao

l.sg. 2.sg. 3.sg. l.pl. 2.pl. 3-pl-

The singular forms of the subtitutive are shared with Mvita and Vumba, whereas the 1st person plural swiswi and the 2nd person plural ng'ming'mi are the same in Mijikenda and in Vumba.

Two case studies of Swahili dialects

173

3.5.1.2. Possessives Chirazi Vumba Mvita NK-Swahili Digo -angu -o -we -eru -enu -ao

-angu -o -we -eru -enu -ao

-angu -ako -ake -etu -enu -ao

-angu -ako -ake -etu/-echu -enu -ao

-angu -o -we -ehu -enu -ao

l.sg. 2.sg. 3.sg. l.pl. 2.pl. 3.pl.

The 2nd and the 3rd persons singular of the possessive pronouns -o resp. -e are identical in Chwaka and Mijikenda, however the other forms occur everywhere in the Swahili dialects. 3.5.2. Demonstratives 3.5.2.1. Near Demonstrative Chirazi Vumba Mvita NK-Swahili Digo huyu hawa huu hii

huyu hawa huu hii

3.5.2.2. Remote

huyu hawa huu hii

huyu hawa huu hii

hiyu/yuno hinya/hano huno hino

Demonstrative

Chirazi Vumba Mvita NK-Swahili

Digo

huyulee hawale hule hile

yuya hara hura hira

huyule hawale hule hile

yule wale ule ile

ulee walee ulee ilee

cl.l cl.2 cl.3 cl.4

3.5.2.3. Referential Demonstrative Chirazi Vumba Mvita NK-Swahili Digo huyo hao huo hiyo

huye hao huo hiyo

huyo hoyo hawo hao huo hoo hiyo hio

hiye ao huo hiyo

cl.l cl.2 el.3 cl.4

cl.l cl.2 cl.3 cl.4

174

Wilhelm J.G. Möhlig

The demonstratives of Chirazi are very heterogeneous. The near demonstrative class 1 has two alternative forms: yuyu and huyu. The first is shared with Mijikenda, the second with Swahili in general. For the remote demonstrative class 1 we even find three alternatives: hule, yule and huyule. The first form was only observed in Chirazi, the second also occurs in Mvita and in Standard Swahili, and the third form is shared with Vumba. The referential demonstrative class 1 again is the same as in Mvita and in Standard Swahili: huyo.

3.5.3. Interrogatives As already mentioned in the case of Chwaka, the interrogatives show hardly any differences between the Swahili and Mijikenda dialects except for the surface forms which are different according to the different sound systems. Only the interrogative 'who' has several forms in the languages compared. The form in Chirazi is ani, and also occurs in Vumba, Chwaka and in Mijikenda.

3.6. Stratificational evaluation Chirazi, like Chwaka, shows four strata in the later periods of its history. (1)

Mijikenda

* *

(2)

S.K. Swahili

* *

(3)

Arch. Mvita

* *

(4)

Standard Swahili

* *

phonological base no lexical traces general basic vocabulary occurrence of implosives addentale ρ and t basic cultural vocabulary modern vocabulary pronominal paradigms

3.6.1. Stratum 1 The oldest is of Southern Mijikenda origin, mainly attested by phonological data. 25 of the 37 consonants of the Chirazi sound inventory can be genetically traced back to Mijikenda. In addition, the chronological sequence of the reconstructable sound shifts has a clear break in the genealogical line after phase V, from whereon the further phonological development follows that of the Swahili dialects. It is important to note that within the range of the 600-word list, there is no significant number of word forms in Chirazi which

Two case studies of Swahili dialects

175

is exclusively shared with Mijikenda. Thus, the first stratum is only attested by phonological data.

3.6.2. Stratum 2 The second stratum can be defined as Southern Kenyan Swahili. This is attested by the fact that Chirazi shares a considerable part of the words referring to the more conservative semantic fields with the other Southern Kenyan Swahili dialects at the exclusion of Mvita and the Northern Kenyan Swahili dialects. As far as the sound system is concerned, the sound shifts under phase VI, namely the introduction of voiced implosives and the substitution of **dz by j', may be allotted to this stratum, because it is shared with Archaic Vumba. The same sound shift also occurs in modern Mvita, but there it is due to comparatively recent Unguja influence. The traditional representative of *j of the Northern and Central Swahili dialects is y.

3.6.3. Stratum 3 The third stratum in Chirazi appears in fact to be of Mvita origin. It is attested by the three addentai consonants in the sound inventory and by the restitution of the phonemes ρ and t. As gauged on the basis of Common Bantu, almost 50% of the reflexes with an inital *t in Northern or Central Kenyan Swahili have a t correspondent in Chirazi, whereas the other 50% have r according to the Mijikenda genealogical line. As to the reflexes of *p, the number of word having β in Chirazi seems to be higher than 50% in comparison with those having ρ as their initial. But nevertheless, the number of words which do have ρ is still very important. Semantically it refers to basic cultural concepts as well as to the modern aspects of life.

3.6.4. Stratum 4 The fourth stratum which is still in the process of being formed, originates from Standard Swahili. It concerns, as in the case of Chwaka, all the spheres of modern social and economic life as well as items of modern culture. The alternating forms of substitutives and demonstratives which have been observed in Chirazi most probably are also due to the modern influence of Standard Swahili.

176

Wilhelm J.G. Möhlig

4. Interpretation of the results as to language death 4.1. The Chwaka case The structure of the stratification in Chwaka (see the chart above in section 2.6.) suggests a historical development in four stages. The underlying hypothetically reconstructed events might have been as follows:

4.1.1. Period no. 1 The borderline between strata (1) Mijikenda and (2) SK Swahili and the actual habitat of the Chwaka suggest that speakers of a Southern Mijikenda dialect immigrated into an empty area in Shimoni which was surrounded by speakers of a SK Swahili dialect with a high social prestige. The sociolinguistic situation created a certain readiness on the side of the immigrants to being linguistically and culturally influenced by these neighbours.

4.1.2. Period no. 2 The second retraceable period in the language history of Chwaka is connected with the origin of stratum (2). We may call this stage the Archaic Vumba Period according to the main representative of SK Swahili dominating that period. As a consequence of period no. 1, there followed massive cultural and linguistic influences from the side of the SK Swahili speaking neighbours. Under this influence, the Chwaka speakers began to change their group identity from Mijikenda towards Swahili.

4.1.3. Period no. 3 Period no. 3 is related with the origin of stratum (3). Later in the historical development of Chwaka, there appears to have been a strong cultural and linguistical influence of Mombasa Swahili, i.e. Mvita. A comparison with the other languages and dialects of the area south to Mombasa shows that the impact of Mvita everywhere appears to have been the same. This period marks the political, economical and cultural dominance of Mombasa during the 18th and 19th centuries.

Two case studies of Swahili dialects

177

4.1.4. Period no. 4 The appearance of the most recent stratum (4) of the chart make us suggest that Standard Swahili through mass media, school, government activities, judiciary and external contacts is actually influencing the language to a high degree. We may call this period the Standardization Period. Its traces are attestable in all Swahili dialects. As a general result of these detailed historical inferences, we come to the conclusion that at no stages of the language history of Chwaka did a real language death occur. The qualitative analysis of the linguistic nature of the strata shows that the Chwaka speakers always kept their identity as belonging to one linguistic community distinct from their Swahili speaking neighbours. This is attested by the fact that the non-Swahili stratum (1) is defined by numerous elements covering the whole inventory and structure of the language without a focus on any specific field. The second and the third strata indicate a change of the group affiliation under the political and cultural dominance of either Vumba or Mombasa. This is attested by a massive impact on the sound inventory and in particular on the vocabulary. Even parts of the grammar are affected. However, on the basis of the present material, it cannot be specified whether the change of group affiliation occurred already during the Vumba period or later during the Mombasa period.

4.2. The Chirazi case The stratificational structure of Chirazi (see chart above under section 3.6.) suggests, unlike that of Chwaka, a historical development in only three stages. The hypothetically reconstructed events might have been as follows:

4.2.1. Period no. 1 The fixing of the first reconstructable period refers to the borderline between strata (1) and (2). Speakers of a Southern Mijikenda dialect, Archaic Digo or Duruma, immigrated into the area, where SK Swahili, Archaic Vumba, was spoken. They gave up their original language, but retained most elements of their inherited sound system, thus creating a new dialect of SK Swahili.

4.2.2. Period no. 2 The beginning of this period is related to the origin of stratum (3). The new dialect of SK Swahili came under the strong linguistical and cultural influence of Mombasa, thus sharing the fate of Chwaka (compare above period no. 3 of Chwaka under section 2.6.3.).

178

Wilhelm J.G. Möhlig

4.2.3. Period no. 3 The last period can be related with the origin of stratum (4). At present a new stratum is being formed under the influence of Standard Swahili through mass media, school, government activities, juiciary and external contacts. This process is identical with that described as period no. 4 in the history of Chwaka. The borderline between the first and the second strata marks a historical event which can be subsumed under the phenomenon of language death insofar as an original language was given up whilst its speakers integrated themselves into the language community which already existed in the area of immigration. The language which was given up can be clearly identified as belonging to Southern Mijikenda, because it left substantial linguistic traces, particularly in the sound system, behind. In theory, there are two alternative reasons for these traces: either the group of immigrating Mijikenda was politically and culturally dominant although small in numbers or, quite the contrary, they were numerous, politically unimportant, but ready to integrate themselves quickly into the local population. In the latter case, one could, for instance, think Mijikenda speaking women who in large numbers married men speaking SK Swahili, whatever the social motivation might have been.

5. Final evaluation of the Chwaka and the Chirazi cases If we compare the Chwaka case with the Chirazi case, we see that both show the same number and sequence of strata. Nevertheless two different conclusions as to the question of language death have to be drawn. In the Chwaka case, the borderline between stratum (1) and stratum (2) has to be interpreted as an exposure of Archaic Chwaka to massive linguistic influence of Southern Kenyan Swahili. In the Chirazi case, we have to interpret the similar borderline between stratum (1) and stratum (2) as marking the death of an original Mijikenda idiom, eventually of a sociolect, and the beginning of a new dialect of a different language group. This shows that it is not the existence of strata as such which is indicative of the nature of the historical events leading to their existence, but rather the linguistic criteria defining the strata. In the Chwaka case, the Mijikenda stratum is defined by linguistic elements from all structural domains, be it the sound system, the lexicon or the grammar. In the Shirazi case, however, the Mijikenda stratum is only defined by phonological data. There is evidence from other parts of

Two case studies of Swahili dialects

179

the Bantu area outside East Africa that strata attesting the event of language death may, in addition to the phonological data, also be defined by lexical data. But by their meanings these always appear to point to specific social groups. In Chirazi such lexical items were not discovered. Thus, Chirazi appears to be an example where language death is indicated by a stratum of exclusively phonological criteria retraceable in unilinear sequences of sound shifts to proto languages different from the genealogical group affiliation of the present language. Our dialectological research in East Africa has revealed many examples of this kind. Another one is for instance the neighbouring Vumba, where the Southern Kenyan Swahili shoot has evidently been grafted upon a Northern Pokomo Stem (Möhlig forthcoming). Contrary to the general opinion that strata are complexes of surviving features typical of a former language which has been replaced (Bynon 1977), our case studies suggest that language replacement or - as we call it - language death is only a rare reason for the origin of linguistic strata. Instead, areal influences amounting to a complete alienation from the original linguistic stock appear to be more frequent events leading to new strata.

References Abdulaziz, Mohamed H. 1979 Muyaka. 19th century Swahili popular poetry. Nairobi: Kenya Literature Bureau. Bynon, Theodora 1977 Historical linguistics. London: Cambridge University Press. Guarisma, Gladys - Wilhelm J.G. Möhlig (eds.) 1986 La méthode dialectométrique appliqée aux langues africaines. Berlin: Dietrich Reimer. Heine, Bemd - Wilhelm J.G. Möhlig 1980 Language and dialect atlas of Kenya. Vol.1. Berlin: Dietrich Reimer. Miehe, Gudrun 1979 Die Sprache der älteren Swahili-Dichtung Phonologie und Morphologie. Berlin: Dietrich Reimer. Möhlig, Wilhelm J.G. 1984-1985 "The Swahili dialects of Kenya in relation to Mijikenda and to the Bantu idioms of the Tana Valley", Sprache und Geschichte in Afrika 6: 253-308. Möhlig, Wilhelm J.G. in press Dialect atlas of the coastal Bantu languages of Kenya. Berlin: Dietrich Reimer.

Chifundi and Vumba: Partial shift, no death* Derek Nurse and Martin Walsh

1. Introduction The targets of this study are Chifundi and Vumba, two small and dwindling Swahili dialect communities, spoken on the coast of southern Kenya, and of southern Kenya/northern Tanzania, respectively. The purpose is to: (a) investigate certain innovations, mainly phonological, in Chifundi and Vumba, innovations which give them a distinctly non-Swahili appearance (b) look at models that might explain the diachronic appearance of these innovations. We suggest that these innovations result from a historical shift from Digo, a neighboring language community. (c) link the foregoing to the topic of this volume - language death - by proposing a concrete set of prints that a doomed language might leave on the language to which its community shifts. In the period immediately before death, a language is not likely to be potent or large enough to leave any prints. But long before that point, when the doomed language is still vibrant, and as its speakers start to shift to another, they may do so in sufficent numbers to influence the target language. If it leaves traces, it might be posthumuously recognized, which is important to historical linguists trying to discern past events. In the case with which we are concerned, the language out of which speakers shifted historically did not in fact die. That is not relevant: it left a nice set of traces! In all, four (Bantu) languages/dialects are involved: Chifundi, Vumba, Digo, and Segeju. Chifundi and Vumba are Swahili dialects. Digo is a southern Mijikenda dialect. Swahili and Mijikenda (thus the main three targets) are typologically similar and closely related. Segeju (of Tanzania, see following) is today a form of Digo, but most likely resulted from a language shift into Digo by a community once speaking Daisu, a member of the Central Kenya Bantu group (= Kikuyu, Kamba, Meru, etc). Some old Segeju on the

182

Derek Nurse and Martin Walsh

coast in 1935 still spoke Daisu (Dammann 1936-1937: 231-233). Chifundi, Vumba, and Digo each have some internal variation. As far as we know, Chifundi is only spoken today in four Kenya coastal villages between the Mkurumiji and the Ramisi Rivers, and in two villages on Wasini Island.1 It was formerly spoken more widely, to the north, for instance, at the old seventeenth century capital of Tumbe, now a ruin just south of the Mkurumiji River, and at Bogoa on Wasini Island, which its inhabitants abandoned for Mkwiro in the nineteenth century. Chifundi speakers also refer to themselves and their dialect as Shirazi, a label they also apply to the people of Aleni and Ngoa, villages to the west of the Shimoni Peninsula, where varieties of Vumba are spoken today. For these places, and the following, see the map. Vumba is spoken at Wasini on Wasini Island, in most villages on the Shimoni Peninsula, and in a string of coastal villages between approximately Aleni/Majoreni (Kenya) and the Boma Peninsula (Tanzania). Digo is spoken between just south of Mombasa (Kenya) and Tanga (Tanzania), mostly in the hinterland, but at some points Digo settlements come down to the coast. We are not clear exactly where Segeju is spoken but most likely on the Tanzanian coast between the border and Tanga Town. Daisu, the language from which Segeju originated by shift, is still used actively by several thousend people at Bwiti, 25 miles from the coast in northeastern Tanzania. All observers agree that, while there are several hundred people on the Shimoni Peninsula in southern Kenya claiming to "be" Segeju, there is no Segeju language in Kenya today. 2 Some observers (e.g. McKay 1975: 11-12) claim the same for Tanzania, that is, all those who claim to be "Segeju" in Tanzania speak only Digo and/or Swahili. We are not sure this claim is true: in the 1970s Nurse and Philippson collected from a young man who lived on the northern Tanzanian coast, and who identified himself as a "Segeju", a 1000 word-list that was similar, but by no means identical, to a word list collected from another informant living only a few miles away and giving "Digo" as his ethnic affiliation. It is unlikely that a young man of university age in the middle 1970s was the only one of his kind, and that he and all his kind have now (1990) passed away. Thus either we say Tanzanian Digo has considerable variety, or we say that both "Digo" and "Segeju" are spoken in Tanzania. The 1979 Kenya census registered 1,519 "Swahili/Shirazi" in Kwale District, that is, 1,519 people chose to identify themselves to the census takers as "Swahili" or "Shirazi" on the Kenya coast between Mombasa and the border with Tanzania. Since "Swahili/Shirazi" in this context is more or less synonymous with "Chifundi" plus "Vumba", there were ca. 1,519 people

Chifundi and Vumba: Partial shift, no death

Map 1. Southern Kenyan and Northern Tanzanian coast

183

184

Derek Nurse and Martin Walsh

who labelled themselves as Vumba or Chifundi in Kenya. Similar figures can be found in McKay (1975: 5-6). Given the rate of increase of the East African population, and that there are several "Vumba" villages on the northern Tanzanian coast, the total "Chifundi/Vumba" population is likely between 2,000 and 3,000. 120,024 people registered as "Digo" in Kenya (101,336) and Tanzania (18,688). Segeju numbers are not clear, as we will see below. The liberal use of inverted commas in the preceding paragraph is meant to express doubt about how to interpret these figures. In this area there is often a discrepancy between what people speak and how they identify themselves ethnically. For example, nearly 20,000 people identified themselves as "Segeju" in the last Tanzanian census (1969) that still referred to ethnicity, plus several thousand in Kenya. But in Kenya no one has spoken Segeju for at least several decades. In Tanzania we suggested above that there may be "Segeju" who speak a form of Digo sufficiently different from other forms of Digo to be called a separate dialect of Digo. Similarly, when one of Walsh's co-workers in Kenya interviewed 50 Vumba-speaking women in Shimoni, they all identified themselves as "Digo". Hence the figures cited above do not reflect accurately language affiliation, of which we in fact have no detailed breakdown. Our best guesstimate would be the number of Vumba (and possibly Chifundi) speakers ought to be slightly higher than the ethnic identification suggests, and that the Digo figure would be correspondingly lower. Although we have no way of knowing how many people speak Segeju in northern Tanzania, the fact that the informant who supplied the data to Nurse and Philippson was young would suggest that there is still an active Segeju speaking community, possibly several hundred or thousand strong. These discrepancies between language and ethnic affiliation appear to largely result from historical language shifts. The "Segeju" originally spoke Daisu. Most shifted to Digo, but some, either directly, or through Digo, moved to various forms of Swahili, including Chifundi and Vumba: hence people who claim to "be" Segeju, while being first speakers of one of the other tongues. While most Digo have retained their language, some shifted to Swahili, including Chifundi and Vumba: hence people who identify themselves as Digo while having Vumba (and, possibly, Chifundi) as their first language. 3 In recent centuries the main shift has been to Vumba, but we cannot exclude the possibility that in earlier centuries a shift to Chifundi also occured. Finally, as we see below, the Vumba once "conquered" the Chifundi. The Vumba were thus the dominant local force, and undoubtedly some Chifundi speakers shifted to Vumba. In general in these shifts, non-Swahili communities shifted to Swahili, and in particular to Vumba. 4

Chifundi and Vumba: Partial shift, no death

185

Even the modified figures of language affiliation above beg the question of what it means to say that a person is a speaker of one of these dialects/languages. This involves two issues: how well people speak their own first tongue, and the patterns of bi- or multi-lingualism in the area. How well people in this area speak their own first tongue not only depends on the usual variables (where they were born and raised, where they now work, educational level, male vs female, etc) but also on the general areal patterns of language use. Down to the beginning of the twentieth century the dominant cultural and linguistic norms in the area were those of Zanzibar and Mombasa. Mombasa has been the most important town on the coast since the beginning of the sixteenth century, and Zanzibar rose in stature during the nineteenth century. Small communities such as Chifundi and Vumba were much influenced by these external norms. Many people in the area travel often to Mombasa. During the second half of this century Standard Swahili has largely taken over the role of Zanzibar Swahili. All coastal communities are exposed to Standard daily, through school, the radio, newspapers, books, officials of the government, constant contact with outsiders, and members of their own communities who leave to find work elsewhere and then return. The result is a very mixed pattern of language use, and an increasing number of people who do not control their own dialect fully. This is most true of young people, adult males (who travel more than woman), and people in larger centers: women and more rural dwellers have more ability in their own tongue. This general situation affects primarily Chifundi and Vumba: in the early 1980s, one of the authors had trouble finding any one in northern Tanzania who spoke anything like a "pure" Vumba. Even Digo is not exempt, as the Digo have been in constant contact with Swahili communities for several centuries, and Islamized since at least the 19th century (Sperling 1985). This situation, plus the fact that the four communities have been coexisting in the area for at least four centuries, has also led to considerable bi-/multilingualism. Most or all Digo, and presumably the Tanzanian Segeju, speak their own language and one or more forms of Swahili. Many (younger, urban, male) first speakers of Chifundi and Vumba are also fluent in Standard and/or Mombasa. Swahili speakers tend to speak only forms of Swahili, whereas the non-Swahili tend to speak their own language plus a form, or forms, of Swahili.

186

Derek Nurse and Martin Walsh

2. Historical background The conventional history of the area under consideration is as follows. 5 A line of Swahili communities has been in place for over a millenium along the East African littoral. Stretching over a thousand miles from north to south, it was kept intact by strong economic, cultural, religious, and family ties. The archaeological record, supported by a fairly detailed oral tradition, indicates that Vumba communities (Vumba K^uu, see map) have existed in situ since at least circa AD 1400. While the archaeological evidence for Chifundi communities is harder to interpret, number of considerations make it likely that Chifundi settlements are equally old. The difficulty of distinguishing Vumba from Chifundi settlements at that remove, and the dispersed nature of the settlements plus the relative absence of stone/coral buildings in the area (T. Wilson p. comm) make it difficult to estimate the size of these early populations. But the very nature of the settlements and the absence of early reference to them makes it likely that they were not large, with populations probably hardly larger than today's. At a point unclear but most probably during the sixteenth century, the Digo moved down from the north and rapidly occupied the hinterland. Finally, during the late sixteenth, and early seventeenth, century, the Segeju (or Daisu) arrived from the north and settled in the hinterland. In the early decades of the seventeenth century, the Vumba community expanded and with the aid of Digo and Segeju conquered the Chifundi and other Shirazi, capturing many of their main villages, and reducing them to a subordinate role. The high point of Vumba power was reached in the rest of the seventeenth, and during the first part of the eighteenth, century, when they played an active economic role along the coast, and were the dominant political force in the region south of Mombasa. The later eighteenth, and the nineteenth, century was a period of decline, in which several factors seem to have played a role: an increasingly marginal economic and political role, as small communities were sidelined by the rising size and power of Mombasa and Zanzibar, internal political squabbles, attacks by the Maasai, several famines, and a number of epidemics. Vumba in the present century plays little or no role in the major affairs of the coast, considerable out-migration occurs, and the dialect itself is increasingly threatended by the spread and use of Standard. We know much less of the history of the Chifundi community, but it would appear to mirror that of Vumba, except that it started from a lesser political and economic base in the seventeenth century.

Chifundi and Vumba: Partial shift, no death

187

Although we know equally little about the history of the Segeju community (Sperling 1985), certain facts can be deduced. Initially, at the start of the seventeenth century, they and the Digo were employed by the Vumba in their subjugation of the Chifundi, suggesting a considerable demographic and military presence, and also an intermingling of their affairs with those of the Digo, with whom they live today in the same or adjacent villages. Judging by scattered reports, the Segeju still seem to have played an occasional military role down to the nineteenth century. They also played an active economic role in the trade networks leading from the coast to the interior. There seems to have been more Segeju presence in Tanzania than in Kenya, where their main settlement area was the Shimoni Peninsula. Unfortunate involvement in the political struggles between Mombasa and Zanzibar, and in the policies of the British colonial power, in the second half of the nineteenth century, led to the destruction of some (many?) of their coastal villages in Kenya. When the Segeju first arrived along the hinterland in the early seventeenth century, it is likely that they were still speaking Daisu, their original language, which even today is spoken by the remnant community at Bwiti, in northeastern Tanzania. There appear thereafter to have been two Segeju communities, one on the coast, the other up at Bwiti, constant contact between the two, and a leaking of people from Bwiti down to the coast. Parts of the coastal community were most likely bilingual in Daisu and Digo, their form of the latter being Segeju. We do not know exactly when the use of Daisu on at least the Tanzanian coast was given up, but the destruction of Segeju villages in the late nineteenth century, together with the facts that in 1935 only a few old people still spoke the language and today no coastal community uses it any more, would suggest a progressive loss during the latter decades of the last century and the first half of this century. We also do not know when or why the use of Segeju was abandoned on the Kenya coast, or even whether it in fact was ever used on that coast. Since the main area of Daisu settlement on the coast was rather in Tanzania, it is possible that Segeju as such never existed on the southern Kenya coast. On the other hand, because Segeju still seems to exist on the Tanzanian coast, and because a shift to Digo (= "Segeju"), or at least bilingualism in Digo, by the coastal Daisu in Kenya would have been probable during their centuries of co-existence, it is possible there was a Kenya Segeju. If it did exist, then - since economically Digo and Segeju were virtually indistinguishable, and since Segeju numbers in what is now Kenya were probably always less than in Tanzania - these factors, plus the dispersal of their separate identity resulting from the destruction of their villages, were probably responsible on the Kenya coast for the disappearance of Segeju at much the same time as that of Daisu.

188

Derek Nurse and Martin Walsh

The Digo are the largest of the four communities, but despite their size, seem to have played a somewhat secondary role to Vumba up to the present century. Because of their geographical presence in the entire hinterland from south of Mombasa to Tanga, and their adjacency to the Swahili trading communities of the coast, their affairs have long been inextricably tied to the economic and political fortunes of those communities. This connection to the coast led to their Islamization in the 19th century, which gave the Swahili communities yet another hold over them, as the centers of Islamic culture were firmly rooted in the Swahili communities. Mostly bilingual in one or other form of Swahili today, they have likely been so for several centuries. In summary, in non-linguistic terms, the two most significant communities from the early 17th, to the 19th, century were Vumba and Digo, the former because of their economic, political, cultural and religious power, the latter on account of their numbers and omnipresence. The Chifundi were subordinate to the Vumba, and the Segeju to both Digo and Vumba. This all implies that the norms of Vumba and Digo, in that order, are likely to have been dominant, and that any language shift would have been out of Segeju, Chifundi, and Digo, in that order, into Vumba and Digo, in that order. As we will see shortly, examination of the linguistic evidence only partly supports these hypotheses.

3. The linguistic picture Although the main targets of this investigation are Vumba and Chifundi, we now look at each of the four languages in turn to get a complete picture of the diachronic interactions between them.

3.1. Segeju As above, we use Segeju here to refer not to Daisu, but to the form of Digo apparently still used in Tanzania, and possibly formerly on the Kenya coast. Of the four tongues under examination, it is the one for which we have least data. This data derives from a word list of circa 800 items filled out in the 1970s by a Tanzanian Segeju from Monga village, followed by a short interview focussing on details of phonology and morphology. The main features of this data are:

Chifundi and Vumba: Partial shift, no death

189

- lexically, phonologically, and morphologically, this Segeju most resembles Digo (Nurse 1982a), specifically southern Digo (see below). - while most lexis is shared with Digo, a very few Segeju lexical items are shared with Daisu alone, and the phonology and/or geographical distribution of these items suggests a Daisu origin. In slightly more cases, items are shared by Segeju, Digo, and Daisu. 6 - of the 800 lexical items at least 150 are borrowed from Swahili. In a modified version of Swadesh's 100-word list at least 13% is borrowed from Swahili. As we see below, the phonology of Vumba and Chifundi is quite distinctive from that of the rest of Swahili: on this basis, one item in 800 is borrowed from Vumba or Chifundi. That is, Vumba and Chifundi are not the source of Swahili loans in Segeju. The majority of loans from Swahili are from the Zanzibar dialect Unguja or from Standard Swahili. - the Segeju informant was more cavalier in recording phonetic details than other Mijikenda informants completing the same list: so affricates were sometimes rendered as the equivalent fricatives, [1] and [r] were sometimes used interchangably, the voiced bilabial fricative was rendered differentially, etc. 15 years later it is hard to know whether this was the result of the informant's incomplete control of the language, or of his lack of experience in writing it down, or whether it mirrored the state of Segeju at that time. At any rate, it seems to suggest that the informant was less aware of norms than other Mijikenda informants. It could be noted here that a Digo linguist at the University of Dar es Salaam once referred disparagingly to Segeju as a "garbled" form of Digo. - the use of [1] and [r] just mentioned is of particular interest. All Swahili dialects are "1-languages", that is, the regular reflex of Proto-Bantu *1 (Guthrie's */d/) in inherited items is IM (or zero). With the exception of Digo and Segeju, the regular reflexes of this *1 in inherited items in all Mijikenda dialects were originally [r] (now /r/) in the context of front vowels, otherwise [1] (now /I/): statistically /l/ predominates. In Segeju and Digo, especially in southern Digo (the type of Digo spoken where there are or were adjacent Segeju communities), the distribution of IM and Irl is opaque, not following any obvious or regular pattern, but with a preponderance of Irl. In Daisu and its relatives in Central Kenya, the regular reflex of PB *1 in inherited words is Irl. This behaviour of Segeju and southern forms of Digo would seem to suggest a transfer of articulatory or allophonic habits from Daisu, as no other language in the whole area treats *1 in this way. Whether these habits were transferred directly into southern Digo, or initially into Segeju, and later into southern Digo, is not clear.

190

Derek Nurse and Martin Walsh

3.2. D i g o At all linguistic levels the Mijikenda dialects, including Digo, are relatively homogeneous. Digo has undergone certain minor innovations not shared by the northern Mijikenda (Nurse 1982a). As most of these are not attributable to Swahili influence we do not discuss them here. Swahili-induced features are: - lexicostatistically, Digo is the most deviant of the Mijikenda dialects. The lexical innovations in Digo are mainly the result of borrowing, from a variety of sources, most obviously from Swahili. In the modified version of Swadesh's 100-word list, roughly the same % derives from Swahili as in Segeju. In the total of circa 1,600 Digo items available to us several hundred are borrowed from Swahili. As for Segeju, most of these are visibly from Unguja or Standard Swahili. A very few (5?) are from Vumba or Chifundi. Thus Standard -tetema 'shiver, tremble' would regularly correspond to Digo -(hje(h)ema but Vumba and Chifundi -rerema\ in this case the actual Digo form is -rerema, a loan from Vumba/Chifundi. - an interesting phonological feature can be observed in Digo. Digo speakers, or at least some Digo speakers, have an abilility to manipulate the phonological code between Digo and Swahili. Numerous items in Digo were recorded in two forms, one a Swahili shape, the other the corresponding Digo shape: when asked about this, some speakers seemed to think it unimportant, and were often unable to identify the Swahili form as such. Some words known to be recent loans in Swahili also appeared in Digo, but in the corresponding Digo shape, that is, Digo speakers are easily able to transmute Swahili material phonologically into Digo. This is presumably the result of widespread and longstanding bilingualism in Digo and Swahili: - where details of Digo nominal or verbal morphology differ from the rest of Mijikenda most of the basic differences do not appear to result from borrowing from Swahili. A few details do originate in Swahili: thus, the replacement of the Mijikenda -kala 'be (copula, auxiliary)' by general Swahili -wa, and the loss of 3sg verbal subject w- (still present in the late nineteenth century). - there are certain small lexical and phonological differences 7 between northern Digo and the southern Digo spoken roughly from the Chifundi area in Kenya down into Tanzania. While some of these cannot be attributed to external influence, a few can. Thus a few lexical items present in southern, but not northern, Digo are shared with Segeju (and often Daisu), so were probably borrowed from Segeju. As pointed out in 3.1., a higher incidence of M in southern Digo points to a Segeju and/or Daisu source.

Chifundi and Vumba: Partial shift, no death

191

3.3. C h i f u n d i Chifundi and Vumba share a number of innovations which distinguish them from the rest of Swahili and make them more like Mijikenda/Digo (section 3.5.). In this section we look only at Chifundi-specific features, and in 3.4., at Vumba-specific features. The Swahili coastal dialects can be divided into Northern and Southern. Chifundi is the most southerly of the Northern Dialects, Vumba the most northerly of the Southern Dialects. Just because the two communities have sat at the Northern-Southern divide for centuries, both have naturally absorbed features from the other set. Chifundi is unique in having a number of specific features that can be attributed neither to Northern or Southern Dialects, nor to Digo or Segeju. As these are hardly likely to have developed in the last four centuries, when Chifundi has been hard pressed by the other communities in the area, this is a reason for thinking that Chifundi has a longer independent history than the archaeology would suggest (see 1.). Because of intrusion by Vumba, Digo, (and Segeju?) in the last four centuries, the Chifundi community is today split geographically into two: Funzi, 8 spoken between the Ramisi and Mkurumiji Rivers; what could be called Wasini, (maybe better "Mkwiro"to avoid confusion with the Vumba village?) spoken in two villages on eastern Wasini Island. For the Funzi and Wasini we have a certain time perspective, since Lambert (1958) is based on data collected in the 1920s, whereas Walsh has worked in the area in the last four years. Walsh reports the Wasini Chifundi claim to have no Digo or Segeju settled among them - whereas the Funzi villages are surrounded by Digo settlements. Relevant features specific to Chifundi, or with a distribution that reportedly differs from Wasini to Funzi, are: - palatalization of velars before front vowel (cf. cifyu 'knife', masicini 'poor ( < Arabic)', mce 'woman', -fyajia 'sweep', mjeni 'guest' with other Swahili kisu, maskini, mke, -fagia, mgeni). This palatalization is a regular feature of Mijikenda dialects (except the most northerly Mijikenda) including Digo and Segeju. It also occurs in Jomvu, a Swahili dialect north of Mombasa, also adjacent to the Mijikenda area. Lambert says it is more regular in Wasini than Funzi. - Lambert (1958: 16) describes the four anterior fricatives of Chifundi(/¡ ν, s, ζ) as being affricated. In general Mijikenda has more segments that are clearly affricated than Swahili: so Mijikenda ts = Swahili c, Mijikenda dz = Swahili j . 9 However, the four anterior fricatives of Swahili and Mijikenda correspond regularly ( S w a h i l i / ν, s, ζ = Mijikenda f , ν, s, ζ), so it is unclear how these Chifundi affricated sounds are to be explained. Walsh has observed

192

Derek Nurse and Martin Walsh

that today only Funzi has these sounds, whereas Lambert implies that in the 1920s they were present throughout Chifundi. - Lambert reports that the form of the 3pl in Funzi can be wa- or a-, whereas Wasini, Vumba, and all other Swahili has only wa-. All Mijikenda have a-, the result of regular loss of /w/. W-loss also appears spasmodically in other words, e.g. kaida 'habit' (cf. Standard Swahili kawaida, from Arabic).

3.4. V u m b a Within the Swahili spectrum, Vumba has a number of specific features, but none that is uniquely attributable to any Mijikenda.

3.5. Chifundi and V u m b a Chifundi/Vumba share a number of features with Mijikenda, features which set them off from the rest of Swahili: - a small set of core/content lexis that for reasons of phonology or geographical destribution must have been transferred from Mijikenda. We are not sure how extensive this set is, but in the 100-word list the following were noted: -ili 'two' (Swahili -wili, Mijikenda iri by regular w-loss), -osi 'all' (Mijikenda -osi, Swahili -ote), -aupe 'white' (other Swahili has -eupe, which shows the inherited initial vowel, whereas some northern (*) Mijikenda have innovated initial [a]), rombo 'breast' (occurs today in all northern Mijikenda (*), where all Swahili dialects have another word), wia 'song' (the Mijikenda word, where Swahili has a derivative of -imba 'sing'), and possibly words involving g-loss such as mongo 'back' and mbeu 'seed' (all Mijikenda have g-loss forms, where Southern Swahili has mgongo and mbegu). Some but not all of these items also occur in Mtang'ata, the Swahili dialect just to the south of Vumba. Möhlig (this volume, section 3.5.) points out that a few function words in Chifundi and Vumba are probably also tranfers from Mijikenda. - all Mijikenda dialects have a regular(*) labiovelarization process whereby sequences of /kw, gw, mw/ are realised as [kp, gb, qmw]. Although the process itself is not active in Chifundi or Vumba, they do show several lexicalized forms of it (also in Mtang'ata to the south): thus -bwa 'fall' (other Swahili -gwa), and a passive suffix -bwa (elsewhere -gwa). Lambert reports these cases as characterizing the Wasini, rather than the Funzi, dialect of Chifundi. - one verb morphological feature that characterizes Chifundi and Vumba is a -jambwa- 'not yet' morpheme, for instance Vumba kha-jambwa-(ku)renda

Chifundi and Vumba: Partial shift, no death

193

'he hasn't done yet'. This occurs in no other Swahili dialect, but throughout Mijikenda we find forms such as Giryama kha-dzangxve-kuhenda 'he hasn't done yet'. Mijikenda /dz/ corresponds regularly to Swahili and Vumba ///, and if we assume that Vumba mbw is a lexicalised form of the labiovelarization process, then only the Mijikenda /e/ and Vumba /a/ is irregular. In Mijikenda at least this is a compound, grammaticali sed, morpheme (-dza- 'recent past' < 'come', plus -ngwe- = ?). - the most significant feature of Chifundi and Vumba, however, is what has happened to PB *p and *t. Whithin the subgroup of Bantu languages ("Sabaki") to which Swahili and Mijikenda belong, one of the most significant isoglosses is one in which PB (and Proto-Sabaki) *p and *t lenited in Mijikenda (also Pokomo and Comorian), but were retained in Swahili (also Elwana and Mwani). Virtually all Swahili dialects have kept *p, *t even today, and in the few that have not, the shifts are minor and recent. Comparative evidence from Mijikenda, Pokomo, and Comorian strongly suggests that the lenition was a shared process that occured in the mid or late first millenium AD, and proceeded thus: Proto-Bantu *p *t

Proto-Sabaki, Swahili *p *t

Early Mijikenda, Pokomo/Comorian *φ *rh (= vl fricative r)

Later developments > β in Comorian, Digo > h in north Mijikenda > r in Comorian > h in all Mijikenda, some Pokomo

The last column is meant to indicate that later, after Mijikenda, Pokomo, and Comorian split up, the voiceless fricatives were retained down to the present only in (part or all of) Pokomo. The northern Mijikenda further weakened both fricatives to /h/, the southern Mijikenda, including Digo, weakened *rh to /h/ but voiced φ to /ß/. Segeju reflects Digo here. The result is regular sets of items such as PB and Swahili pa-, northern Mijikenda ha-, southern Mijikenda βα- 'Class 16 marker'; Proto-Bantu and Swahili -tatù, but Mijikenda -hahu 'three'. In Segeju and southern Digo this /h/ can delete (so also -au 'three'). Contrary to what happens in all other Swahili dialects, *p and *t have become /β/ and Irl respectively in Chifundi and Vumba. 10 So the reflex of the voiceless labial is identical (/ß/) in Digo, Chifundi, Vumba, (and Segeju). the reflex of the apical is different: Chifundi and Vumba Irl but Digo (and

194

Derek Nurse and Martin Walsh

Segeju) Ihl. It might be noted that the reflex of *p in Daisu is also /β/ (but of *t is /t/). We combed through nearly all the available Chifundi and Vumba data (Bakari 1985, Möhlig various, Lambert various, Nurse field notes) involving reflexes of *p and *t, and found the following rates of occurrence. Each of the two sets of percentages below is based on between 60 and 90 eligible items from all the sources, depending on which dialect or stop was involved: 11

Chifundi

Vumba

[ß] [p] mixed [β, ρ]

79% 16% 5%

65% 27% 8%

Reflexes of *t: [r] [t] mixed [r, t]

83% 10% 7%

71% 26% 3%

Reflexes of *p:

It should be noted that the data on which these percentages are based came from a range of sources: some older/some more recent, some oral/some written, some from narratives/some from elicitation of individual items, some from men/women, older/younger informants, etc. Thus the figures should be taken as a general indicator of frequency, not an exact measure of occurrence. Although Chifundi shows higher levels of the lenited continuants than Vumba, both dialects clearly have many more continuants than stops. The items containing the stops and continuants respectively are frequently the same in the two dialects. 12 The reason for the appearance of apparently nonlenited segments is not that they failed to undergo lenition diachronically, but rather, under pressure from Mombasa Swahili and Standard Swahili, that the two small communities are slowly now replacing their continuants by the stops found in the rest of Swahili, word by word. The obvious but difficult question here is: why did Chifundi and Vumba lenite at all? Five possible scenarios present themselves: i. Chifundi and Vumba are not Swahili dialects and never were. In other words, when the initial early split occurred between Swahili (also Elwana and Mwani: the conservative members) and Mijikenda (also Pokomo and Comorian: the leniting, innovating, members), Chifundi and Vumba went along with the leniting group.

Chifundi and Vumba: Partial shift, no death

195

The objection to this scenario is that, apart from this feature and the few others mentioned in 3.3. and 3.5., basically every other characteristic of Chifundi and Vumba is Swahili-like. So Chifundi and Vumba would have had to undergo massive borrowing at every linguistic level, which is possible but implausible. It is also not supported by non-linguistic factors, such as oral tradition: Mijikenda/Digo traditions do not claim that the Chifundi and Vumba were once Mijikenda, nor do the core Chifundi or Vumba traditions. ii. Chifundi and Vumba developed together with, or were heavily influenced by, Comorian before the Comorians left the mainland coast in the second half of the first millenium AD, a possibility advanced by Nurse (1982b, 1984-1985). The central basis for this claim is that from *p and *t, Comorian developed /β/ and Irl, just as Chifundi and Vumba. A secondary support comes from Comorian traditions which claim that they came from "Mrima": "Mrima" in Swahili refers to the northern Tanzanian coast and might be held to include the Chifundi and Vumba areas - but "Mrima" in Comorian refers to the East African coast in general. The objection to this scenario is that, while it is true there is considerable evidence that Comorian and Swahili in general (so including Chifundi and Vumba), have had long contact over the last millenium and do share certain linguistic innovations, there is no evidence, other than their identical treatment of *p and *t, of specific shared similarities or innovations between Comorian and Chifundi/Vumba. iii. Chifundi and Vumba developed together with, or were heavily influenced by, Upper Pokomo at some unspecified early point, which is similar to a hypothesis advanced by Möhlig (1984-1985). The only basis for this claim is that where Upper Pokomo has the (earlier) voiceless pair /φ, rh/, Chifundi and Vumba have the exact (later) voiced congeners /ß, r/ which could easily have developed from the Upper Pokomo situation by a simple voicing rule. The main objection to this suggestion is the same as that to the preceding: there is no evidence of other specific linguistic similarities of any kind between Upper Pokomo and Chifundi/Vumba. A secondary objection is non-linguistic: no oral tradition published or known to us even suggests it. iv. Despite the difference in the Chifundi/Vumba and Digo treatment of *t, it could be argued that the whole set of similarities (section 3.3. and 3.5.) between them points to lengthy and intense contact, with either this contact or a language shift from Digo (or pre-Digo) into Chifundi/Vumba being responsible for today's situation: we discuss borrowing vs. shift in section 4. This hypothesis is heavily supported by the known historical facts as set out in section 2. The Digo (or pre-Digo) arrived with /β/ and /rh/. /β/ presents no problems of interpretation, as it is common to Digo, Vumba, and Chifundi. The differ-

196

Derek Nurse and Martin Walsh

enee between the Chifundi/Vumba and Digo reflexes of *t can be reasonably explained: both /h/ and Irl derive via a single step by common phonological processes (lenition and voicing, respectively) from /rh/, a segment which in East Africa at least appears to be unstable and is highly marked in general (Maddieson 1985). That is, when Digo and Chifundi/Vumba first came in contact, Digo still had /rh/, the earlier stage. This leaves open the question of why they should have developed differently (section 4.). While this is the best of the four scenarios so far, a fifth should also be considered. v. In the first two paragraphs of this section 3.5., it will be noted that certain similarities (marked (*) in the text) between Mijikenda and Chifundi/Vumba are not Digo specific: the one phonological feature, labiovelarization, is reportedly more active in the more northerly Mijikenda, than in Digo, and a number of lexical items are attested in more northerly Mijikenda, not in Digo today. Furthermore, the Digo reflex /β/ of earlier *p is also attested in Duruma and Rabai, spoken to the north of Digo, whereas the far northern Mijikenda have /h/. Duruma/Rabai and Chifundi/Vumba are not today adjacent. So the possibility exists that the presence of these features in Chifundi/Vumba might derive from an earlier period of contact when some of today's more northerly Mijikenda communities lived further south, adjacent to Digo. Spear (1978) suggests that on the way to their present location, some of the more northerly Mijikenda communities followed a circuitous route which involved formerly being some distance south of Mombasa before moving further north. It is hard to judge this hypothesis. While the evidence for it is clear enough, it could easily be explained away. Labiovelarization might have been more active earlier in Digo, but be lessening under the ongoing influence of Swahili. Digo might earlier have had the lexical items, only to have replaced them more recently. And /β/ as the reflex of *p might have recently spread north from Digo, a large community, into Duruma and Rabai, smaller communities: Rabai, for instance, has both /h/ and /β/ as reflexes of *p in different lexical items. We see no other linguistic evidence of similarity between Digo and more northerly Mijikenda. We find the fourth hypothesis the most convincing overall explanation of the similarities between Digo and Chifundi/Vumba: we admit the fifth as a minor additional possibility.

Chifundi and Vumba: Partial shift, no death

197

3.6. Summary of the linguistic picture If we look at the preceding from the point of view of recipients, we see: (Tanzanian) Segeju. Segeju today appears to be a form of southern (i.e. Tanzanian) Digo. It has considerable lexical and limited verb-morphological material from Swahili: almost no evidence of specific material from Chifundi or Vumba, most Swahili material being from Unguja and/or Standard Swahili. It is possible that Segeju is unstable, since our informant showed variation in his rendering, although this might be due to other factors. Limited lexical and phonological evidence for Daisu material in Segeju. Digo. It has considerable lexical and limited verb-morphological material from Swahili: limited evidence of specific material from Chifundi or Vumba, most Swahili material being from Unguja and/or Standard Swahili, and thus, as in Segeju, fairly recent (nineteenth century onward). Southern Digo shows limited traces of material from Daisu, possibly mediated through Segeju. Chifundi. Considerable phonological evidence for convergence with Digo: reflexes of *p and *t (?), palatalization, lexicalized cases of labiovelarization, possibly affrication and limited loss of /w/. Also some (how much?) Digo lexis, and at least one verb-morpheme. The ongoing replacement of /ß, r/ by /p, t/ points to influence from more standard forms of Swahili. Lambert's data for the differences between the Wasini and Funzi dialects suggests that both have absorbed Digo material, but apparently differentially. Although the evidence of Digo influence is better today for Wasini (more palatalization and labiovelarization) than for Funzi (limited loss of /w/, and affrication - which is hard to substantiate), this may be the result of the loss of these features in Funzi, so is non-diagnostic. Vumba. Phonological evidence for convergence with Digo: reflexes of *p and *t (?), and lexicalized cases of labiovelarization. Also some (how much?) Digo lexis, and at least one verb-morpheme. The ongoing replacement of /β, r/ by /p, 1/ points to influence from more standard forms of Swahili, more so than in Chifundi. If we look at the preceding from the point of view of donors, we see: Daisu. Only mentioned briefly above is the fact that the earlier Daisu community (referred to as "the Segeju") had a considerable lexical, and some phonological, impact on the Mijikenda in general, in the period during and after the sixteenth century, as they moved down the coast from the north,

198

Derek Nurse and Martin Walsh

passing along the Mijikenda spectrum, and apparently settling among the Mijikenda (Nurse 1982a). Apart from, and later than, that, there is limited lexical and phonological evidence of Daisu material in Segeju, and in Digo, possibly mediated through Segeju. Daisu ceased being spoken on the coast between the last decades of the last, and the middle of the present, century. Segeju. Ambiguous evidence, as just mentioned, for some Segeju material in Digo: it might have passed into Digo from Segeju, or directly from Daisu. Digo. Strong evidence of Digo phonological and lexical material in Chifundi and Vumba, and limited verb morphological material. More obvious in Chifundi than Vumba. Possible that other Mijikenda dialects might have contributed earlier: that might be formulated as "earlier forms of Mijikenda might have contributed". Could Digo material have passed first into one or other of Chifundi and Vumba, and then into the other? Chifundi. No evidence that Chifundi has contributed material to any of the other communities, other than minimal lexical traces in Digo, which however might equally have come from Vumba. Vumba. As Chifundi, which is surprising, given that the Vumba community is represented as being the political, economic, cultural, and religious center of the area during the seventeenth and eighteenth centuries, and possibly into the nineteenth. If it were asked why Vumba, Chifundi, Digo and adjacent Mijikenda communities have /β/ as the reflex of *

η which is not shared by other Kalenjin dialects. (2) Part of the Terik vocabulary is related exclusively to the Elgon dialects and must have been acquired somewhere in the Elgon region. (3) The sequence l-W-l has been replaced by r-V-n in these three dialects. (4) Bong'om and Pok frequently spirantize initial p, which can also be observed occasionally with Terik speakers but never happens in other Kalenjin dialects. (5) There are lexemes which are confined to Terik and the other Elgon dialects and which do not occur in Nandi, e.g., murwaket 'snail', puntet 'nail', musempet 'sheep tail', puur 'sit, remain'. The following are among the features that Terik appears to have borrowed from Nandi (cf. Rottland 1982: 269-270; Roeder 1986: 56ff.): (1) The Elgon dialects show a vowel elision rule r-W-r > rr, i.e. ProtoKalenjin vowels have been deleted in these dialects when preceded and followed by r. Like Nandi, Terik does not show any trace of this rule, e.g., Proto-Kalenjin Nandi Terik Elgon dialect

*moororooc moororoocet moororoocet moorroocet

'frog'

(2) The Elgon dialects have reduced the Proto-Kalenjin vowel system from ten to seven short vowels. Terik, however, has the full set of ten short (and ten long) vowels. (3) Verbs of "Class 2" have a stem-initial vowel i in Nandi and Terik, in the Elgon dialects, however, there is no such vowel, although it can be morphophonologically reconstructed. (4) There are certain lexemes which are absent in the Elgon dialects, but which Terik shares with Nandi, e.g. caariiren 'lazy, weak', coot 'melt (itr.)', kireer 'grave', kwer 'beat'.

260

Bernd Heine

2. Socio-cultural change 2.1. The emergence of a new cultural orientation While a history of the Terik still remains to be written, Roeder's rich collection of oral traditions (Roeder 1986: 133-172) and her analysis provide sufficient evidence to support the linguistic hypothesis that Terik belongs historically to the Mt. Elgon region, i.e., that a substantial part of the modern Terik originate from the slopes of Mt. Elgon and reached their present homeland through migrations which are likely to have involved a period of common ethnic identity with the Bong'om. Such evidence is based on observations like the following (cf. Roeder 1986: 172): (1) The Terik share with the Kalenjin of Mt. Elgon the myth of origin according to which "we are people of Mt. Elgon". (2) "Elgon" is an important factor of Terik ethnic identity. (3) The eponymous foundation father of the Terik, called Terikin or Sabaot, is linked invariably with the Mt. Elgon region. (4) Together with Pok, Bong'om is the dialect most closely related genetically to Terik. Now, according to Bong'om traditions, "the people who later called themselves Terik were still Bong'om when they left Elgon and moved away in a southern direction" (Roeder 1986: 142). As we observed above (section 1.1), relations between the Terik and the Nandi in pre-colonial times were frequently strained, marked by mutual raids for cattle, land and women. This perspective is still alive among old Terik people, and the memory of hostile interactions and war to some extent influences their present-day attitudes vis-à-vis the Nandi. Younger Terik, on the other hand, have a different historical perspective. Among these, the view prevails that relations between Terik and Nandi have always been peaceful and friendly as they are at present. This suggests that stereotypes on inter-ethnic relations have changed drastically within a few generations, and that the cultural assimilation that the Terik experience today might not have been possible a century ago. There are a number of factors contributing to the emergence of a new ethnic orientation of the Terik. Perhaps the most important ones are: (a) Due to population pressure and the resulting overpopulation and dwindling agricultural potential in their own area, there is a gradual population movement among the Terik from their homelands of Sino and Masan into their new settlement region which they call "Nandiland" (see Map 2). This recent population movement can be seen as forming the major factor in inducing ethnic consciousness. As long as the Terik live in their home areas of

Dialect death: The case of Terik

Map 2. Terik territory

261

262

Bernd Heine

Sino or Masan, their ethnic identity is firm and stable, and they tend to consider their speech as representing "pure Terik", however Nandi-ized it may actually be (see below). Once they settle in the Nandiland region, however, their attitudes toward group cohesion change (Roeder 1986: 199). (b) This population pressure is caused mainly by the expansion of their western neighbours, the Luyia, especially the Logoli group of Luyia. While interethnic relations, most of all with the Tiriki Luyia, were friendly in the past, they have become hostile during the last decades. Luyia are felt to be a menace to Terik ethnic identity. A Terik man marrying a Luyia woman is considered to be lost for the Terik community: he is unlikely to maintain a Terik culture in his family, since his children are believed to fall victims to the dominant Luyia population and culture surrounding many Terik homes in Sino and Masan. 4 For fear of being absorbed by the Luyia, the Terik orient themselves more and more towards their eastern neighbours, the Nandi. Some Terik even claim that they abandoned their homeland and settled in Nandiland because they were afraid of the Luyia. (c) Part of the assimilation process is also a gradual change towards a Nanditype age set system (Roeder 1986: 205-210). The presence of a cyclical age set system is an essential characteristic of traditional Kalenjin social organization. While the Nandi have a system of seven age sets, Roeder found twelve sets to be nomenclaturally distinguished among the Terik, eight of which are of practial relevance. That a transition to the Nandi system is taking place is suggested by the fact that Terik living in Nandiland tend to name a smaller number of age sets than those living in the other parts of Terik country, i.e. in either Sino or Masan, as Table 1 shows.

Table 1. Number of age sets named by Terik people in and outside Nandiland, respectively Percentage of persons interviewed Number of age sets named 6 or fewer 7 8 9 or more Total Total number of persons interviewed

Outside Nandiland (Sino and Masan)

Nandiland

11 25 19 45

29 47 18 6

100

100

54

17

Dialect death: The case of Terik

263

(d) While initiation rites form one of the major means of introducing boys to the cultural values of Terik society, they are also a factor of change: the Terik distinguish between an initiation of the "Nandi way" and of the "Christian way". The latter is associated with the Luyia, with whom the modern Terik do not share very close relations, as we observed above, 5 while the former is considered as the only genuine, traditional initiation, and many consider it a "must" for any boy who wants to become a good Terik. Terik boys in Nandiland therefore undergo initiation together with Nandi boys and, hence grow up with the Nandi culture and consciousness (cf. Roeder 1986: 198). (e) For the modern Terik, the Nandi form a group of a different quality from other ethnic groups. For many of them, a transition from being a Terik to being a Nandi does not constitute a change in ethnic affiliation. For example, to settle in Nandi territory is not seen as affecting one's ethnic identity; it tends to be viewed as a change in the neighbourhood which requires, e.g., that one adapts one's pronunciation to that of the neighbours. Settling among the neighbouring Luyia, on the other hand, is more immediately regarded as a threat of one's "Terik-ness", which may result in the loss of one's ethnic identity (see above). (f) Compared to the Nandi, the Terik form a small minority, and their dialect is almost entirely a spoken medium; there are hardly any written documents in it. Nandi, on the other hand, has a semi-official status and an established orthography. There is a Bible translation, there are text-books for primary schools and regular radio programs in Nandi, of which the Terik can and do take advantage. (g) Assimilation to the Nandi is also enhanced by the growth of an increasing Kalenjin consciousness. Up to about 1940, groupings like Nandi, Keyo or Kony formed largely independent ethnic entities. Since then, a new kind of ethnicity has developed, uniting the various groupings which speak mutually intelligible varieties of the same language. For many Terik this certainly has been a contributing factor to eliminating the ethnic boundary that had existed between them and the Nandi and to develop a new kind of "super-ethnic" identity referred to by the label "Kalenjin".

2.2. R e a c t i o n s to c h a n g e Perhaps surprisingly, after all that has been said in 2.1, the Terik conceive their ethnic status and group cohesion as stable and static. This at least can be derived from a number of observations. For example, to the question "Have you ever heard about Terik joining another tribe?", 79% answered "no" and only 21% "yes". Similarly, only a minority (32%) were aware that other

264

Bernd Heine

people had ever become Terik, while more than two thirds of the consultants (68%) stated never to have heard anything like that happen (Roeder 1986: 179-180) While for many Terik the ongoing Nandi-ization process appears to take place unnoticed, there are others who are aware of it, or at least of some aspects of it. Their reactions range from regret to indifference to full approval, especially since it is a welcome alternative to Luyia-ization, which is regarded as a threat to their ethnic and linguistic autonomy.

3. Language behaviour Among the various linguistic features which have been mentioned in section 1.2, Roeder found that one is especially wellsuited to serve as a parameter for measuring the degree of Nandi-ization, i.e. the extent to which the speech habits of a given Terik have been influenced by the Nandi dialect of Kalenjin. This feature concerns the Nandi / : Terik η correspondence. Together with the other Elgon dialects, Terik has undergone a sound shift / > η (see 1.2), and some Terik speakers say that "pure Terik" is a language without I. Now, in the process of Nandi-ization, this shift is being reversed. Roeder found out that the more often a Terik speaker uses /, where others use n, the more his or her speech turns out to be Nandi-ized, and she proposes the following tripartite classification of Terik people: "Elgon speakers" (E), who use no more than ten times / in a sample of 100 words containing n. "Mixed speakers" (E/N), who use / more often that ten times within a sample of 100 words containing n, and more often than ten times η in a sample of 100 words containing /. "Nandi-ized speakers" (N), who use η no more than ten times in a sample of 100 words containing I.6 This classification is employed as a framework for describing the present state that dialect replacement has reached.

Dialect death: The case of Terik

265

3.1. A demographic profile In order to test to what extent areal factors contribute to the relative degree of Nandi-ization, the informants' home areas were divided into two zones: 7 Nandiland, i.e. the Terik area adjacent to Nandi territory, and non-Nandiland, i.e. the areas Sino and Masan, which do not border on the Nandi territory (see Map 2). As Table 2 suggests, there is a clear correlation between the degree of Nandi-ization and geography: While in Nandiland the "Nandi-ized" speakers (N) form the majority (53%), their number is insignificant in the remainder of Terik country (7%), where more than two thirds (69%) are "mixed speakers" (E/N). Note that the percentage of "Elgon speakers" (E) does not differ remarkably in the two zones. Table 2. Degree of Nandi-ization and geography (in percentages) Residence of informants Non-Nandiland Nandiland "Elgon speakers" "Mixed speakers" "Nandi-ized speakers" Total Number of informants

24 69 7

20 27 53

100

100

80

30

According to these quantitative findings, one may say that the closer a Terik lives to the Nandi territory, the less likely he or she is to be a "mixed speaker", and the more likely to be a "Nandi-ized speaker". This suggests that the relative degree of Nandi-ization depends on the intensity of contact Terik people have with Nandi. Roeder (1986: 76) volunteers the following observation: "The low percentage of "mixed" speakers in Nandiland is the result of pressure to assimilation the Terik are exposed to once they have settled in Nandiland." If the hypothesis of a dialect shift in progress is correct then one would expect that the largest number of "Elgon speakers" is found among old people and the largest number of "Nandi-ized speakers" among the younger generations. As Table 3, based on Roeder (1986: 77, 79), shows, this holds true with some limitations: the higher the age of the group of consultants chosen, the higher the number of "Elgon speakers", and the lower the number of both "mixed" and "Nandi-ized speakers" is. Accordingly, among the population of speakers over 60, "Elgon speakers" form the largest group (47%), while among the Terik under 30, two thirds are "mixed speakers".

266

Bernd Heine

Table 3. Degree of Nandi-ization and age (in percentages) Type of speaker

"Elgon speakers" "Mixed speakers" "Nandi-ized speakers" Total Number of informants

Age 0-30

0-60

Over 30

Over 60

11 66 23

19 60 21

43 43 14

47 37 16

100

100

100

100

73

91

37

19

Thus, a Terik speaker over 60 is statistically most likely to be an "Elgon speaker". A third variable that turns out to be relevant to a description of dialect shift is sex. As Roeder's survey suggests, women have a consistently higher rate of "Elgon speakers" than men. According to Table 4, 8 female "Elgon speakers" are more numerous than female "Nandi-ized speakers", whereas the male population exhibits the opposite proportion: the number of male "Nandi-ized speakers"(36%) is almost 20 times higher than that of male "Elgon speakers" (see Roeder 1986: 82).

Table 4. Degree of Nandi-ization and sex (in percentages; only consultants under 30 of age are considered) Women "Elgon speakers" "Mixed speakers" "Nandi-ized speakers" Total Number of informants

Men

19 68 13

2 62 36

100

100

31

42

This observation suggests that Terik women show a more conservative language behaviour than men: they have been less affected by Nandi-ization. On the basis of such demographic data we may say that the highest percentage of Terik classified as "Nandi-ized speakers" are likely to be found among young male Terik living closest to the Nandi, while the largest number of "Elgon speakers" are to be found among old women living farthest away from the Nandi.

Dialect death: The case of Terik

267

3.2. Attitudes and sterotypes In spite of the ongoing shift from an Elgon variety to Nandi, the Terik appear to have a remarkable loyalty to their own dialect. 78% (102) of the persons interviewed want their children to learn Terik as their first language and to use it in the family, as opposed to only 12% (15) who opted for Nandi,9 and the majority also want Terik to be the medium of instruction in primary education (Roeder 1986: 110). It remains unclear, however, what exactly the term "Terik" stands for in each case - considering the discrepancy alluded to above between linguistic competence and one's own beliefs about that competence. On the whole, Terik people appear to be largely unaware of the process of dialect shift (cf. Roeder 1986: 111, 112). This is evident, e.g., when the question was asked as to what constitutes "good Terik", the elicited responses to which were "Good Terik is what I speak now", "Good Terik is the language spoken by (true) Terik", or "Good Terik is the language spoken by people born in Terik country". Implicitly, however, there appears to be some awareness of this process. Thus, the statement "Good Terik is what I speak now" was made only by old people, and only young people suggested that in order to find "good Terik language" one should look for old people. While the majority of features named by Terik consultants as distinguishing "Elgon" from "Nandi" are extra-linguistic, some also relate to linguistic differences. Common examples are the phonological distinction η (Elgon) vs / (Nandi), as in nanet (Elgon) : laiet (Nandi) 'bag', or lexical distinctions such as Elgon wees : Nandi lapat 'to run'. Note, however, that the linguistic features that link the Terik with the Elgon dialects are not viewed by them as symbolizing continuity with their Elgon past but rather as an indication of their own linguistic identity as distinct from Nandi. Not infrequently, a remarkable discrepancy between self-assessment and one's actual behaviour can be observed. People very often do not speak the way they think they do. Roeder (1986: 92) cites the case of an 80-year old Terik man who she clearly classified as a "Nandi-ized speaker", i.e. with reference to almost all parameters he used Nandi-type, rather than Elgon-type linguistic features. Yet he considered himself to be a "good Terik speaker" as he belonged to the age set of Maina; he claimed that only the Nyonki, Maina and a few people of the Cuma age set were able to speak "good Terik". The vast majority of consultants considered their own Terik to be "good", and only relatively few said that they do not speak "good Terik". This selfassessment is at variance with the respective linguistic performance as dis-

268

Bernd Heine

cussed by Roeder (1986: 117) on the one hand, yet on the other it correlates directly with the degree of Nandi-ization as calculated on the basis of the ηΛ index introduced above. Thus, Table 5 suggests that "Elgon speakers" (as well as "mixed speakers") are very likely to consider themselves to be "good Terik speakers", while two thirds (65%) of the "Nandi speakers" consider themselves to be "bad Terik speakers" and only one third (36%) as "good Terik speakers". Table 5. Degree of Nandi-ization: self-assessment vs. actual linguistic performance (in percentages) Percentage of informants who claim to be

"Elgon speakers" "Mixed speakers" "Nandi-ized speakers" Total

Number of informants

"good Terik speakers"

"bad Terik speakers"

83 72 36

17 18 65

23 54 17

100

100

94

It would seem, however, that self-assessment correlates more strongly with age than with actual linguistic performance: the older a Terik is the more likely he or she is to claim to be a "good Terik speaker". As Table 6 suggests, there is a remarkable cleavage between people who are younger and older, respectively, than 30 years. This would seem to provide another example in favour of the claim that language behaviour is much more determined by extra-linguistic than by linguistic facts. Young people, even if they are clearly "Elgon speakers", tend to have doubts as to whether they speak "good Terik", while old people frequently claim to speak "good Terik" even if they use a highly Nandi-ized form of Terik (see Roeder 1986: 121). Table 6. Self-assessment and age (in percentages) Age

0-30 over 30

Number of informants who claim to be "good Terik speakers"

"bad Terik speakers"

55 90

45 10

Total percent

Total number

100 100

62 54

Dialect death: The case of Terik

269

4. Conclusions The data looked at suggest that the process of dialect shift from Terik to Nandi is the result of extra-linguistic forces. It would seem that population pressure and subsequent population movements form the ultimate cause of Terik dialect replacement: increasing infiltration of their western Luyia neighbours into Terikland is responsible for the Terik settling in the less densely populated Nandiland to the east. There they are exposed to a different but closely related culture, which they gradually adopt and pass on to their friends and relatives still living in the original Terik homelands of Sino and Masan. This process is enhanced by a number of factors, such as the cessation of the traditional Terik-Nandi hostility, the minority status of the Terik, and the emergence of a new, "super-ethnic" structure called Kalenjin, to which both Terik and Nandi belong on account of their historical and linguistic relationship. The result is a gradual cultural and linguistic shift from an Elgon affinity to Nandi. Roeder (1986: 224-225) comes to the conclusion that the death of Terik as a dialect of its own will not necessarily entail an extinction of Terik ethnicity. There are, in fact, many examples in Africa which suggest that language shift does not necessarily lead to a shift in ethnic identity. Although with the death of the oldest Terik generations a substantial part of the "Elgon speakers" will disappear, one cannot predict that Terik as a distinct linguistic form will also become extinct. While this is very likely to happen, it is conceivable that those extra-linguistic forces which gave rise to the process described in this paper may give way to other forces leading to the arrest of this process - or even a partial reversal of it. If we divide the shift from one dialect/language to another into (I) an initial, (II) an intermediate, and (III) a final stage then we may say that Terik has already passed Stage II. This becomes clear especially on the basis of lexical evidence of the following kind: (a) As the lexicostatistical data adduced by Rottland (1982: 257) shows, Terik has acquired higher percentages of cognates with the target dialect Nandi than it actually shares with its Elgon source: Terik shares higher percentages of cognates with Nandi (93) and Kipsiikis (95) than with the Elgon dialects Pok (86) and Sapiny (80) (see Table 7). (b) Whenever there is a lexeme in Terik which differs from the corresponding Nandi lexeme then the latter may be used as well. All Elgon-specific Terik words are in "free variation" with corresponding Nandi words, while there are many Nandi words in Terik which do not have a corresponding Elgonspecific "free variant" (cf. Roeder 1986: 88).

270

Bernd Heine

Table 7. Lexicostatistic counts for some Kalenjin dialects (in percent, 100-wordlist of Swadesh; Rottland 1982: 257) Nandi 96 90 77 93 77

Kipsikis 90 78 95 79

Pok 91 86 78

Sapiny 80 80

Terik 78

Päkot

Heine and Rottland (1980) comment the present situation of Terik in the following way: . . . by putting Terik within the Elgon group we refer to a form of Terik which is largely extinct. If the "Nandiisation" of Terik continues we may in the end no longer be able to retrace its genetic origin because all Elgon features have disappeared. Then, strictly speaking, Terik will no longer have a place in a genetic tree. This is where the principle of divergency reaches its limits. (Heine Rottland 1980: 48)

In Heine (1971a) and Rottland (1977, 1982), Terik was classified as being genetically most closely related to the Elgon branch of Kalenjin. The same type of phonological, morphological and lexical evidence could be used, however, to show that Terik does, in fact, belong to the Nandi-Markweta branch - the more so since the application of the technique of lexicostatistics would support the latter hypothesis. 10 It would seem that neither of these hypotheses adequately describes the historical status of Terik - we simply lack a model which accounts for both divergent and convergent language development. In this respect, the Terik case resembles that of Ma'a (Mbugu), which also combines material from two different linguistic sources. There are, however, at least two factors which suggest that Terik presents some problems which do not exist to the same extent in the Ma'a case. First, Ma'a exhibits a relatively clear division between lexicon and grammar: most lexical items belonging to the basic vocabulary can be traced back to a Southern Cushitic stock, while the grammatical morphology derives entirely from Bantu. Such a clear-cut pattern does not seem to exist in Terik, where the Elgon vs Nandi division cuts across the linguistic levels like phonology, morphology and lexicon in a seemingly unpredictable way. Second, compared to Ma'a, Terik appears to present a more dynamic instance of language interaction, both with reference to diachrony and to synchrony. Compared to the relatively stable pattern characterizing the Ma'a/

Dialect death: The case of Terik

271

Mbugu situation, the Terik case presents a prototypical continuum ranging from an "Elgon" type of speech at the one end to Nandi at the other. Diachronically, this continuum can be observed in the process of dialect shift from the erstwhile Terik to Nandi, while synchronically it is manifested, for example, in the differing degrees of Nandi-ization that the various demographically defined segments of the Terik population exhibit, as we saw in section 3.1.

Notes 1. I wish to express my gratitude to Eithne Carlin, Matthias Brenzinger, Franz Rottland, and Derek Nurse for valuable comments and corrections. 2. For reasons see Roeder (1986: 31, fn. 2). 3. The term "Tiriki" must not be confused with "Terik". Although the two terms are historically related, they refer to two quite different peoples. 4. A number of negative stereotypes and rumours have spread among the Terik. It has been claimed, for example, that the Luyia try to kill Terik livestock by littering plastic bags which are swallowed by the latter (Roeder 1986: 200). 5. Part of the training during the seclusion period of the initiation is the practice of shooting with bows and arrows, where Luyia form the imaginary target. 6. For more details, see Roeder (1986: 73). 7. Roeder (1986: 75) actually distinguishes three zones, but for the present purpose we may use a more reduced distinction. 8. This table is based on interviews with persons under the age of 30 only, since the number of women above 30 was too small to be included in a quantitative analysis. 9. Another 6% opted for English, and 4% each for "Kalenjin" and Swahili (Roeder 1986: 106-107). 10. "It is worth noting that lexicostatistics, which according to its theoretical foundations should always lead to a genetic classification, leads to erroneous conclusions in the case of Terik. This means, that even the 'basic vocabulary' used in lexicostatistics can be influenced to such a degree by contact that it no longer reflects divergency. Apparently, this happens only between languages with a high degree of mutual intelligibility, which is the case for Terik and Nandi" (Heine - Rottland 1980: 48).

References Baumann, Hermann (ed.) 1975

Die Völker Afrikas und ihre traditionellen Kulturen. Part 1. Wiesbaden: Franz Steiner.

272

Bernd Heine

Heine, Bernd 1969

"Zur Frage der Sprachmischung in Afrika", Zeitschrift der Deutschen Morgenländischen Gesellschaft, Supplement 1 (3): 1104-1112. 1971a "Kalenjin glottochronology", Mila 11 (2): 51-55. 1971b "On genetic relationship and the case of the 'Nilo-Hamitic' languages", Journal of Eastern African Research and Development 1 (1): 19-28. Heine, Bernd - Franz Rottland 1980 "Divergency, convergency, and the case of Terik", Transafrican Journal of History 9 (1/2): 42-51. Huntingford, G.W.B. 1950 A comparative study of the Nandi dialects of East Africa. [Unpublished MS.] Köhler, Oswin 1955 Geschichte der Erforschung der nilotischen Sprachen. (Afrika und Übersee, Beiheft 28.) Berlin: Dietrich Reimer. 1975 "Geschichte und Probleme der Gliederung der Sprachen Afrikas", in: Hermann Baumann (ed.), 141-374. Otterloo, Roger van 1979 A Kalenjin dialect study. [Unpublished MS, Nairobi] Roeder, Hi Ike 1986 Sprachlicher Wandel und Gruppenbewußtsein bei den Terik. (Sprache und Geschichte in Afrika, Beiheft 7.) Hamburg: Helmut Buske. Rottland, Franz 1977 "Zur Gliederung der südnilotischen Sprachen", Zeitschrift der Deutschen Morgenländischen Gesellschaft, Supplement 3 (2): 1483-1494. 1982 Die südnilotischen Sprachen. Beschreibung, Vergleichung und Rekonstruktion. (Kölner Beiträge zur Afrikanistik 7.) Berlin: Dietrich Reimer.

Language shift among the Suba of Kenya Franz Rottland and Duncan Okoth Okombo

1. Introduction This paper summarizes some aspects of a case study based on (published) oral history and on the results of a sociolinguistic survey carried out in 1985.' In terms of Sasse's threefold division (in this volume) it deals with the External Setting and the Speech Behaviour but not with the Structural Consequences. Regrettably then, it belongs to the numerous studies which cannot fully contribute to Sasse's model, but we accept his label "defective" for it only in the sense of "incomplete", since the absence of data for SC does not invalidate any observations on External Setting and Speech Behaviour. 2 The observations made here are restricted to the Kenyan situation, although the Suba/Luo settlement area extends across the border into Tanzania. Field work in Tanzania, however, was begun much later and is still in progress. 3 In Kenya, there were according to the 1979 Population Census 59.668 Suba (Abasuba) of whom 58,650 lived in South Nyanza Province, i.e., in the Suba areas indicated on the map, viz. Rusinga, Mfangano, Gwassi, Kaksingiri, Suna West and Suna East locations as well as Muhuru peninsula and adjacent territory in South Kadem. 4 The numerically dominant population of South Nyanza are the Luo (speaking Dholuo, a Western Nilotic language) who according to the same census number 646,486 in South Nyanza and 1,955,845 in Kenya. The 1979 census was the first to offer the choice of "Suba" as an ethnic label. Very probably the people who identified with that label in 1979 had previously been listed as Luo. There are several mutually intelligible Bantu dialects spoken in the Suba areas. Although there is little linguistic justification to do so (cf. below), for the purposes of this paper we refer to these dialects as the Suba language or just Suba. 5 This language has never received any official recognition; it has never been described let alone standardized or appeared in print. Dholuo on the other hand has a well developed literature; throughout South Nyanza

Language shift among the Suba of Kenya

275

(including the Suba areas) it is the main language of the Christian churches and (with Swahili and English) the medium of instruction in schools. The Suba/Luo situation thus involves two ethnically and linguistically unrelated groups (Bantu vs. Nilotic) with a marked imbalance in size of population and in language status. Not surprisingly, there is agreement in the relevant literature about an ongoing process of Luoization of the Suba which includes the (actual or foreseeable) disappearance of their language. Before looking at this process in detail it seems appropriate to point out some of the difficulties we find when trying to define the Suba as a group.

2. Suba identity Both from the angles of external distinctiveness and of internal cohesion it is very problematic to define the Suba. 6 The origin and history of the term Suba are also quite unclear. As opposed to the Luo their common denominator can be seen in the fact that they speak a Bantu dialect or are at least of Bantu-speaking ancestry and that their traditions of migration are different from those of the Luo. That does not distinguish them from nearby Bantu groups who do not accept the Suba label. Thus, the southern Kenyan Suba (of Muhuru and Suna) are in many respects closer to Tanzanian groups (e.g. Kuria) than to the northern Suba groups. This refers to language, to circumcision (which does not take place in the northern groups and in the South is arranged in conjunction with the Kuria), and to migrational traditions including references to eponymous ancestors. Although all Suba dialects are mutually intelligible, there is a relatively wide north-south rift between them - too wide to allow for the assumption of one original Suba language (proto-Suba), and further research will have to go in search of northern links for northern Suba and southern links for southern Suba (Rottland n.d.). In sum, it is not possible to establish any objective linguistic basis for the notion "Suba", neither diachronically nor synchronically. The same applies to cultural features such as circumcision, which only takes place in Muhuru and Suna. According to some informants it was abandoned under Luo pressure in the other areas, but there is reason to assume that for the majority of the northern Suba it never existed. 7 The migrational traditions are also very divergent and basically may be divided into north-western and southern origins. As for the people who accept the Suba label for themselves, the range of application varies almost with each individual, points on the scale being my

276

Franz Rottland and Duncan Okoth Okombo

clan, my local group, all Bantu groups in South Nyanza, all Bantu, all non-Luo (Rottland - Okombo 1986). Suba identity can therefore not be established by reference to their heterogeneous Bantuness alone. What is equally important (and what - as a reaction - may have created their actual subjective identity) is the process of Luoization which all Suba groups have undergone, although in varying degrees. The Suba may therefore be tentatively defined by reference to both a common Bantuness and a common experience of Luo domination. The former distinguishes the Suba from the Luo and the latter from neighbouring Bantu.

3. Luoization of the Suba The process of Luoization of the Suba as revealed by oral history has been outlined by Ogot (1967) and described in more detail as well as documented by both Ayot (1977, 1979) and Kenny (1977, 1978, 1982). Ogot states that the Luo "colonization of South Nyanza was more or less complete by the third generation back, about 1850". The "important historical process" from then on was the assimilation of "non-Luo elements", especially in the northern Suba areas (Ogot 1967: 216 f.). How this assimilation was achieved in pre-colonial times remains largely obscure. Cohen (1983) has proposed an intricate model for Luo/Bantu interaction in Busoga, which we will only refer to here. Some more straightforward suggestions are made by Ayot (1977: 166 ff.), according to whom the channels of assimilation were initially trade and intermarriage. "The Abasuba people, being fishermen, usually took dried fish to the Luo and exchanged it for grain." It was thus "essential for the Abasuba groups to learn Dholuo". As for marriage, "the Abasuba girls married in Luo land were compelled to practice only Luo customs and give up their own", while Luo girls marrying into Suba communities "carried Luo traditional ways of life with them without much opposition". Still according to Ayot, the shift towards the Luo language began with the adoption of Luo proper names, a feature which has been observed elsewhere as an early sign of language shift (Dressier 1988: 188). It extended to practices and their Luo designations and gradually spread to all spheres of life. Ayot makes it clear, however, that the process of assimilation was relatively slow before colonial factors entered into it. He describes in detail how colonization - resting on the three pillars of administration, formal western education and Christianization - was invariably linked to mastering the Luo language. To give an example: Obar, who was fluent in Dholuo, was in 1900 appointed the first colonial chief of Rusinga and Mfangano. He consolidated

Language shift among the Suba of Kenya

277

his control by dismissing the traditional council of elders and replacing it by young men who spoke Dholuo. By conducting his meetings in Dholuo he excluded the monolinguals. Other chiefs Ayot refers to had before their appointment been messenger, teacher or house boy and thus learned Dholuo (plus Swahili, but the function of Swahili is not made clear by Ayot). In their exercise of power they followed Obar's example. Also on the lower levels of administration nobody could obtain a government job without knowledge of the Luo language. Christianity and western education were also only obtainable via Dholuo since both spread from mission stations which were founded in Luo Land and which concentrated on the medium of Dholuo. The account given by Ayot was largely supported during a number of informal interviews which we conducted. The synonymy of modernization and Luoization was brought to a point in the following statement: "My father decided to become a Christian, to donate land for a school to be built upon, and to speak Dholuo in the family." The sociolinguistic process which Ayot refers to or implies in his account of Suba assimilation is the well known one of an increasing one-sided bilingualism gradually leading to language shift. The main motivating factor seems to have been a result of colonization, viz. the imposition of Dholuo (and, generally speaking, Luo identity) as the only medium of upward social mobility. The other apparently important factor, intermarriage, is rather a means than a cause, since we still need an explanation for the fact that, linguistically, the Suba girls lost in the process and the Luo girls won. 8 Ayot, whose final point of reference is the year 1940, basically conceives the assimilation process as a completed one (thus implying language death). The term Luo-Abasuba which he uses for the present Suba can be interpreted as meaning "Luo of Suba descent". This claim was the starting point for our survey of language knowledge, usage, and attitudes among the Suba.

4. Survey results The survey was limited to the Suba areas of South Nyanza (cf. map) and, furthermore, was concentrated on persons claiming to be Suba. 9 The statistically evaluated sample was 583.

278

Franz Rottland and Duncan Okoth Okombo

4.1. Competence and usage Taking Luoization as a point of departure an important statistical result of the survey is the existing strength of the Suba language. 88.1% claimed competence in Suba (99.1% in Dholuo). We assume, however, that such competence must be ranged on a wide scale since only half of the sample claimed to have learned Suba as their first language. This relatively low figure may be partly due to a misunderstanding of the question (learning understood as formal acquisition), but it nevertheless seems to point to a high number of speakers to whom Suba has not been transmitted in the family and who, by implication, have a restricted Suba competence. Crosstabulation by "generation" (ages of 50+, 25-49, and below 25) has revealed a steady decrease in the acquisition of Suba as the first language. With the exception of one area (Muhuru, cf. below), less than a quarter of those under the age of 25 have learned Suba as their "mother tongue". Multilingualism is predominant in the sample. Only 15 persons (2.6%) were monolinguals (2 in Suba, 13 in Dholuo), almost 20% were bilingual and the rest claimed to know three or more languages. Among the multilinguals the prevalent combination was Suba + Dholuo + Kiswahili. As can be seen below, however, multilingualism does not affect the dominant-recessive relationship of Dholuo and Suba, since the domains of any other languages than the two are marginal ones. The distribution of languages over various domains shows the expected decrease of Suba usage and the corresponding increase of Dholuo usage when we start with the family and proceed to wider areas of communication. We indicate only three points along this line: 43% have Suba as the "main language in the family" (44% Luo, 12% "mixed"), whereby 50% use Suba with the father; 32% use Suba with neighbours, and 17% at the market. It is only in the wider areas that Kiswahili is used to a noteworthy extent (e.g., 23% with the administration). Again crosstabulation by age shows a similar gradient but a preference for Dholuo in all domains with younger persons.

4.2. Attitudes The survey contained several questions on ethnic symbols, language prestige, and the future of the Suba language. In summarizing the statistical results it can be said that the Suba language comes out most persistently as a symbol of Subaness. Among a given choice of three ethnic symbols, "origin" was accepted by 95%, "language" by 75%, and "home" by 23%. 10 Two thirds of the sample

Language shift among the Suba of Kenya

279

also named language as the main feature differentiating Suba from Luo. The answers to both questions were largely independent of a respondent's claimed competence, i.e., a respondent might define his Suba identity by reference to a linguistic competence which he did not possess. It seems that language is subjectively conceived of as a feature inseparable from the group but not from the individual member of that group. Asked to arrange four given languages in order of prestige 11 the majority were in favour of the following order: 1st Suba, 2nd Kiswahili, 3rd Dholuo, 4th English (English being far behind). What is particularly surprising in this ranking is the low position of Dholuo which is in marked contrast to its high frequency of usage (distribution over domains) and thus its high general utility as compared to, e.g., Kiswahili. It also contrasts with the initial high prestige of Dholuo about which Ayot reports. A tentative explanation for this apparent contradiction is offered in the final part of the paper. Except again in the Muhuru area (cf. 4.3) two thirds of the sample regarded Suba as a disappearing language, regretted the decline but in their majority thought it could be prevented. The suggested measures of prevention were mainly formal and administrative ones, such as standardization, literary development ("writing books"), and formal teaching (50%), only 15% were of the opinion that counteraction should begin in the family. The majority view on these points is a realistic one. Parents frequently complained that bringing up young children in Suba puts them at a disadvantage when entering school.

4.3. Relative strength of areas For a number of our survey questions there are considerable differences between the Suba areas in terms of statistical figures. To give only one example, we obtained the following percentages for Suba as the first language learned: Muhuru 74.6, Kaksingiri 57.0, Gwassi 45.5, Suna 41.7, Mfangano 39.3, Rusinga 8.3. What is important here is less the actual figures than the degree of variation, 12 the reasons for which are not altogether clear. Partly they may be connected with the accessibility of areas and their position relative to the main stream of Luo expansion (in a southern direction, across the Kavirondo Gulf and/or around it). 13 Tentatively we interpret the variation in temporal terms, i.e., in terms of different dates for the initial Luo impact or of its duration. The figures obtained do not always range the areas in the same order but it can safely be said that in terms of Suba linguistic vitality Muhuru is by far the strongest and Rusinga by far the weakest area. The tendencies are basically the same in each area (e.g. decrease of Suba competence with

280

Franz Rottland and Duncan Okoth Okombo

younger people), but following our temporal interpretation we might say that Muhuru is perhaps 70 years "behind" Rusinga and that Rusinga hardly shows any longer signs of Suba decline since it has almost reached the end point.

5. Conclusions In regard to Suba competence and usage the following summary will suffice: The South Nyanza Suba groups are characterized by a massive language shift - past and present. The stage of language death has been reached at least on Rusinga island if not elsewhere, but a synchronic comparison of the areas shows considerable unevenness. What needs further discussion are questions of attitude and motivation. In Rottland - Okombo (1986) we have talked about an (emotional, not practical) Suba "revival" among younger people. From our present position this seems to be a slight overstatement. Nevertheless, on the affective or emotional scale Suba ranks highest and Dholuo surprisingly low. This may be explained by reference to observations made with immigrants, e.g., in Australia (Johnston 1963, 1965) and the U.S.A. (Fishman 1965). 14 With these immigrants (and their offspring over several generations) the appreciation of the original first language (e.g., Polish in Australia) increases almost at the same rate as the competence in it declines. In other words, there is a growing discrepancy between subjective and objective assimilation, and this discrepancy is probably based on a causal relation between the two streams of assimilation. The particular situation in South Nyanza has an additional aspect: The motivation to learn Dholuo is in present-day Kenya not quite the same as it was in colonial days. At least in regard to education Dholuo is for the Suba still a necessary means of upward social mobility. Competence in Dholuo is also indispensable in the South Nyanza context. On the other hand, while in colonial times the only accepted identity was an ethnic one and a Suba label was not handed out, independent Kenya has created a national context in which it would - at least theoretically - be possible for the Suba to participate directly and to dispense with their Luo affiliations. Whether an awareness of this situation has led to a lower prestige of Luo and a higher of Suba remains open. Objectively, however, the assimilation process continues because it was created administratively and is maintained administratively. It could only be changed in the same way.

Language shift among the Suba of Kenya

281

Notes 1. Survey work was conducted jointly by both authors, it was sponsored by the German Research Foundation (DFG) through the Research Unit (SFB 214) "Identities in Africa" of the University of Bayreuth. It consisted mainly in the application of some 700 questionnaires on demographic factors, language competence, language behaviour and language attitudes, supplemented by formal interviews and the collection of word lists. 2. The confinement to External Setting and Speech Behaviour had in our case - and will have in most other cases - practical reasons: lack of time and of competence in Suba. We hope to be at least aware of the consequences for the evaluation of the survey: Competence is measured according to an all-or-nothing principle (you either speak Suba or you don't), and all language names given are of the same homogeneous vagueness (we have no idea of the reality somebody refers to when he claims to speak Suba of Luo). 3. Research on the Tanzanian part of the Suba/Luo situation was begun in 1989 and is conducted by Mechthild Reh (again within the framework of SFB 214; cf. note 1). 4. We left out a former Suba stronghold, Gembe, since we assumed that Luoization had been entirely completed there and thus the area offered little for the observation of an ongoing shift. 5. Since we use the language prefix for Dholuo we should do so for Suba, but there is a choice of two prefixes, eki- being used in the south and olu- (oru-) in the north. Furthermore, in the north dialect names are preferred, e.g., orukune or olumwa (-mwa being a Luo word stem for non-Nilotes). 6. Cf. Kenny (1977: 277), calling the Suba "spuriously unified under a common name which subsumes a number of lineage-based groups of diverse geographical and temporal origins". 7. Kenny (1982: 19) reports that in Kaksingiri and Gwassi circumcision and clitoridectomy were practiced "until the impact of the Luo became pervasive", while no record of such practices is contained in Ayot (1977, 1979). 8. We do not have such an explanation. There is at least some indication that the dominance of Dholuo in intermarriage is also prevalent in Tanzania (M. Reh, p.c.) and thus seems to be independent of the particular sociocultural pattern that developed in Kenya. 9. We considered the labels "Suba", "Luo-Abasuba", and "Jamwa" as equivalent for these purposes (the last mentioned being a Dholuo term for non-Nilotic non-Luo). It should be mentioned here that the sample contains certain imbalances (coverage per area, sex, and age ratio), due to a basically rule-of-thumb approach we had to follow. The survey is fully described and evaluated in Franz Rottland, Doris Schmied, and D. Okoth Okombo "A sociolinguistic survey among the Suba of Kenya"(ms.).

282

Franz Rottland and Duncan Okoth Okombo

10. "Circumcision" was initially given as a fourth symbol and scored very high in Muhuru and Suna. It consquently had to be withdrawn since circumcision is not practiced in the other areas. 11. The prestige criterion given by the interviewers was the relative strength of a respondent's wish to be fully competent in a given language, referring to himself or to his/her children. 12. Figures for Rusinga are less reliable in our survey, since we did not cover the island systematically (which we did in most other areas). There is no doubt, however, that Rusinga is the weakest area. Heine (1980: 77) who conducted interviews on the island in 1968 equally obtained quite low figures for Suba competence. 13. Kenny (1977: 278), e.g., says that "the Luo spread overland (cf. Ogot 1967) whereas the Basuba came to their present homes by canoe along the shores of Lake Victoria". Evans-Pritchard (1949: 29) has it that the southward migration of the Luo "seems to have been across the gulf ( . . . ) and not round it". 14. The information was taken from Fishman (1972).

References Ayot, Henry Okello 1977

Historical texts of the lake region of East Africa. Nairobi: Kenya Literature Bureau

1979 A history of the Luo-Abasuba of Western Kenya from Nairobi: Kenya Literature Bureau.

A.D.-1760-1940.

Cohen, David W. 1983

"The face of contact: A model of a cultural and linguistic frontier in early Eastern Uganda", in: Rainer Voßen - Marianne Bechhaus-Gerst (eds.), 339-355.

Dressier, Wolfgang U. 1988

"Language death", in: Frederick J. Newmeyer (ed.), 184-192.

Evans-Pritchard, Edward E. 1949

"Luo tribes and clans", Rhodes-Livingstone

Journal 7: 24-40.

Fishman, Joshua A. 1965

"Language maintenance and language shift in certain urban immigrant environments: The case of Yiddish in the United States", Europe Ethnica 22: 146-158.

1972

The sociology of language. An interdisciplinary social sciences approach to language in society. Rowley, Mass.: Newbury House Publishers.

Heine, Bernd 1980

"Language and society", in: Bernd Heine - Wilhelm J.G. Möhling, 59-78.

Language shift among the Suba of Kenya

283

Heine, Bernd - Wilhelm J.G. Möhlig 1980 Geographical and historical introduction. Language and society. Selected bibliography (Language and Dialect Atlas of Kenya 1) Berlin: Dietrich Reimer. Johnston, Ruth 1963 "A new approach to the meaning of assimilation", Human Relations 14: 295-298. 1965 Immigrant assimilation: A study of Polish people in Western Australia. Perth: Paterson Brokensha. Kenny, Michael G. 1977 "The relation of oral history to social structure in South Nyanza, Kenya", Africa 47 (3): 276-287. 1978 Basuba historical narratives. (Cyclostyled, 193 pp.) 1982 "The stranger from the lake: A theme in the history of the Lake Victoria shorelands", Azania 27: 1-26. Newmeyer, Frederick J. (ed.) 1988 Linguistics: The Cambridge survey, Vol. IV Cambrigde: Cambridge University Press. Ogot, Bethwell A. 1967 History of the southern Luo, Vol. I. Nairobi: East African Publishing House. Rottland, Franz n.d. Linguistic correlates to the ethnic notion of Suba: A diachronic survey. (MS). Rottland, Franz - Duncan Okoth Okombo 1986 "The Suba of Kenya: A case of growing ethnicity with receding language competence", Afrikanistische Arbeitspapiere 7: 115-126. Voßen, Rainer - Marianne Bechhaus-Gerst (eds.) 1983 Nilotic studies. Proceedings of the international symposium on languages and history of the Nilotic peoples, Cologne, January 4-6, 1982. (Kölner Beiträge zur Afrikanistik 10) Berlin: Dietrich Reimer.

175 years of language shift in Gweno Christoph

Winter

Gweno is a member of the Kilimanjaro Bantu languages in North Eastern Tanzania. Guthrie (1967-1971) defined these languages as group 60 in his zone E. Outside this group, the language most closely related to Kilimanjaro Bantu is Dabida, E 74a, spoken in the Taita Hills in Kenya. The Kilimanjaro Bantu languages are chiefly the various Chagga idioms (E 62a-c) spoken on the slopes of Kilimanjaro, but they also include Rwo (E 61) spoken on the slopes of Mt. Meru, as well as Kahe (E 64) and Kuma (E 63) spoken in two river oases on the plains below Kilimanjaro, and Gweno (E 65) in the northernmost part of the North Pare Hills. The centre and south of the North Pare Hills, the Middle Pare Hills, and the South Pare Hills are the home of another Bantu language, Pare. Guthrie defined it as number 22 of his zone G. The other members of this group are Tubeta (G 21), Sambaa (G 23), and Bondei (G 24). The language outside this zone G which is most closely related to Pare, is Sagala, E 74b, spoken in the Taita Hills in Kenya. Pare falls into two dialect groups, Asu in the South and Middle, and Athu in the North Pare Hills. Athu shares, besides some lexemes, a number of phoneme pronunciations with Gweno rather than Asu, the most outstanding one being /th/ in Athu and Gweno for /s/ in Asu. It must be admitted at this point that the existence and geographical distribution of these two dialects, Asu and Athu, seems to be far from reliably established. The chief authority for the respective assumptions made here is a note made rather in passing by Tucker and Bryan (1974: 192) that Athu is the dialect of North Pare and Asu the dialect of Middle (and by implication also of South) Pare. Their research dates back nine to fifteen years before the year of its publication. My own cursory experience in North and South Pare (1986-1990 and 1991 respectively) is, however, (a) that the pronunciation /th/ prevails all over the North and also is encountered in many parts of the South while /s/ occurs here, too, in some parts; (b) that in spite of this geographical preponderance of /th/, it appears to be accepted by almost every speaker that the pronunciation should be /s/, and that in support of this appreciation of the standard the translation of the New Testament is adduced

286

Christoph Winter

Map 1. The Gweno Paramountcy, ca. 1760 and the colonial division: Ugweno, Usangi

175 years of language shift in G w e n o

287

which appeared about twenty years ago and is widely known, and which employs the grapheme /s/; (c) that this appreciation of the standard has also been influenced in the last two decades or so by the vast increase both of contacts with other peoples outside their mountain fastnesses and of the spread of Kiswahili (a language which ordinarily employs /s/) among them, which have not only exposed them to interethnic teasing for being lispers but made them sensitive to this, too; and (d) that speakers generally deny knowledge of dialectal differences within the Pare language, instead explaining manifest pronunciation differences as the idiosyncrasies of individuals. Obviously, they are acutely in the grip of a shift in pronunciation from /th/ to /s/. There also are indications of the possibility that in parts of the South an earlier, but also recent, shift from /s/ to /th/ occurred. Until more thorough research clarifies this confusing situation, I therefore incline for the present purpose, where this whole question is but of minor importance, to assume Tucker and Bryan's note as correct. An outline of the history of the relationship between the Gweno and Pare languages can be reconstructed from the research on Pare political history done in 1965-1966 by Kimambo (1968, 1969) and additional data from the reports of early German missionaries in North Pare. Such an historical outline based on Kimambo has already been published by O'Barr (1973: 29 ff., 42 ff., 53 ff.) but my own differs from Kimambo and O'Barr in regard to the chronology of the events. Kimambo ordered his oral historical data according to numbered generations back into the past, and (1969: 39) introduced the notion that a generation ran to approximately 30 years. My own research leads me instead to assume a mean duration of only 16 years for one of Kimambo's "generations". My reasons are the following: Kimambo reports two historical links with the history of the Chagga. One (1969: 139) occurred in the 3rd "generation" back, the other (1969: 58) in the 13th "generation" back. For the first instance, Chagga history provides the clear witness of a contemporary document showing that this event occurred in 1885. It was the death of the Gweno paramount chief Sereki in a raid ordered by the Chagga chief Rindi of Mochi (Schanz 1913: 47). The other instance, namely the installation of Okongi/Ukonyi as chief of Keni in southern Rombo by the Gweno paramount chief Righiria and the interference in this by the Chagga chief of Mamba, Maraviti (Schanz 1913: 7), can be rather securely dated to about 1712 on the basis of the comprehensive chronology which I worked out for Chagga history, taking into account several hundred items of oral history data relevant for chronology, the linking of such information to the history of the named age-sets, and the "clock-work" of the Chagga age-set cycle which I have been able to recon-

288

Christoph Winter

struct (Winter 1977). Two further links with Chagga history which Kimambo (1969: 136) reports for the 7th/8th and the 12th "generations" back, are much vaguer in substantial information, but at least they make good sense in terms of Chagga history when chronologized in my way, whereas they make no such sense when we follow Kimambo's chronologizing. Finally, a link with Sambaa history which Kimambo (1969: 112) reports for the 7th "generation" back - certain auxiliaries coming from the Sambaa chief Kimweri - turns out on my chronological reckoning to be acceptable (involving a time around 1820), while Kimambo (leading back into some period such as 1755-1785) is forced to assume that an anachronism has vitiated this particular tradition. But obviously, a period of just 16 years is too short to reflect the duration of a real generation. I therefore consider that the Gweno genealogies reported by Kimambo, while proposing father-son relationships, often in fact involve brother-brother relationships. Indeed, Kimambo's genealogy of the earliest rulers, viz. those of the Shana clan, in combination with a traditional naming system whereby a first son was given the name of his paternal grandfather, indicates - unobserved by Kimambo - that among those nine rulers there were probably three pairs of brothers. Although very little is known about the Gweno age-set system, there are reasons, and Kimambo himself points out some of them, to believe that its working was very similar to the Chagga age-set system (cf. also Fuchs 1902a: 226; 1902b: 370). From the latter we know that a ruler's reign was in earlier times often limited to the eight years of the political dominance of his age-set or the 16 years of dominance of his super-set. When the next age-set (or super-set) succeeded to dominance, the ruler normally had to hand over his throne to his younger brother who was the leader of the new ageset. A first son might succeed to the rule at the earliest 24 years after his father's accession. It seems likely that these circumstances also applied to the Gweno system, thereby accounting for these rather short "generations". The 16 years assumed here should be understood as the statistical average of the duration of the leadership periods of eleven rulers over a total period of 173 years prior to 1885. This allows the assumption of a maximum of only eight real generations, and requires the assumption of at least two cases of same-generation succession, within this period, but it remains difficult to decide which of the ten successions were of this kind. Now, following Kimambo's oral history data (and my chronology), it appears that originally Gweno was the language in the whole of the North Pare Hills, and - at least from the early 18th century - Asu was the language of the south (and Middle) Pare Hills. Then, Asu began to encroach on the Gweno area, initially and partly through immigrations from the South Pare

175 years of language shift in Gweno

289

Hills, but later predominantly through a shift of Gweno speaking groups to bilingualism and then to exclusive use of Asu. As these Gweno speakers adopted Asu, they carried phonological, lexical and other Gweno-isms over into Asu, thereby producing the new dialect of Athu. The Gweno of the North Pare Hills appear to have constituted an influential unified political unit with a paramount chief ruling mostly by his ritual powers over some 10 sub-chiefdoms each of which was centred on a different local clan. This constitution may have begun about the last quarter of the 15th century under the leadership of the Shana clan of ironworkers, but around the middle of the 17th century it clearly got into full swing when the Suya clan usurped the paramountcy (Kimambo 1969: 50 ff.). The Gweno paramount chiefdom built its influence around a cycle of its own of male youth initiations, first at home in its paramountcy, and then gradually penetrating the whole of Chagga, from early beginnings in the second half of the 16th century, until by the middle of the 17th century all of the Kilimanjaro Bantu region was involved, and it kept this regional leadership right into the early 20th century. But its more manifest domination was centred on its ability to acquire cattle for the products of its iron industry. Wealth in cattle served to regale and keep followers, thereby increasing military dominance and security. Then, and to an even greater extent, cattle served to pay for acquiring ivory from specialized elephant hunters in the neighbourhood, and later all over the region. This helped to monopolize contacts with the ivory-buying caravan-traders who mediated trade with coastal towns. The cloth and beads obtained in exchange from these traders served to reward the loyal followers and to help acquire more ivory. Trade profits and levies in kind and labour imposed on the followers rewarded the leaders for their organizational accumen and efforts. However, the Gweno monopoly of the caravan trade contacts was early challenged by Chagga chiefdoms which had originally arisen as its dependencies, but which later strove for independence. After the usurpation of the Gweno paramountcy by the Suya clan its grip on Chagga became shaky, and from about 1720 the Chagga chiefdom of Mamba broke this monopoly for good. The later process of language shift appears to have its earliest roots in the fact that the Suya paramount chief of Gweno, Shimbo, who ruled in about 1750, invited an Asu ritual expert to come with his relatives and stay with him in Gweno (Kimambo 1968: 24; 1969: 137). These immigrants developed into their own clan, the Sangi, with a sub-chiefdom of their own which enjoyed a slightly special status under the paramountcy, although in all matters of state

290

Christoph Winter

and initiation rituals, and probably also linguistically, these Sangi became thoroughly Gweno-ized. However, in the first quarter of the 19th century, the powers of the paramountcy were weakened, presumably because of the ascendence of the Chagga chief Orombo, and the sub-chief of the Sangi, Munga, claimed independence from the Suya paramount chief, Minja II. Munga appears to have been encouraged to do so by the fact that his sub-chiefdom was uniquely well placed geographically to engage in the growing trade with long-distance caravans from the coast. He is also reported to have argued the Sangi independence on the basis of the slightly special status of his chiefdom in the realm. Although Munga did not succeed in obtaining his independence, yet he brought in auxiliary troops from South Pare, many of whom he settled permanently next to his sub-chiefdom as a buffer against the adjoining Gweno in the north, creating thereby a new sub-chiefdom. Now, with a view to safeguard the existing system of political checks and balances, it was stipulated that these Mbaga were never to be allowed to become integrated into the Gweno system of youth initiation. Instead, their only choice was to send their youths home to South Pare for this purpose (Kimambo 1969: 108-116). This status as foreigners in Gweno made the Mbaga stick to their Asu identity also in terms of their home language. During the following few decades several different developments converged to increase the status and importance of Asu as against Gweno in North Pare. First, the Sangi chiefdom and with it the Mbaga settlement did obtain independence from the paramountcy while at the same time thriving even more from its foreign trade (Kimambo 1969: 117, 122). This appears to have strengthened Asu identity and Asu speech in this group. Our earliest linguistic documentation from Pare dates from this period. It is a small vocabulary of just over 40 words collected by the explorer Carl Claus von der Decken in 1861-1862 (Decken 1887-1888). The majority of lexemes in this vocabulary are Asu, but about half a dozen are Gweno. Unfortunately, von der Decken's transcription is too incompetent to allow tracing pronunciation differences between Asu and Athu. Moreover, there is no information provided regarding the circumstances of the collection, apart from his general travel account which indicates that von der Decken passed through Asu speaking locations in Middle Pare as well as visiting the Sangi chiefdom in Gweno. Johnston collected a small list of Gweno words when he stayed at Tubeta in 1884, but neither this list nor his travel account (Johnston 1886) reveals anything of relevance for our discussion.

175 years of language shift in Gweno

291

Secondly, the sub-chiefdoms in the south of North Pare were cut off from direct contact with the northern part of North Pare where the paramount chief resided, by the intervening chiefdom of the Sangi (Kimambo 1969: 117). Thirdly, while these southern sub-chiefdoms were also being drawn more and more into interregional trade where an Asu ethnic element from the South predominated, they fourthly also began striving for political independence and eventually obtained it factually (Kimambo 1968: 29; 1969: 118 ff.). In consonance with their increasingly Asu-orientated identity during this period they, too, shifted to Asu speech. The next step occurred in the period from 1885 to 1900. In 1885, as a consequence of the prolonged period of drought in East Africa which had started in 1882, and due to peculiar political circumstances in Kilimanjaro at this time, the whole of Pare, but the northernmost, still Gweno speaking parts in particular, became the object of protracted raiding for cattle by most Chagga chiefdoms and the Arusha (Kimambo 1969: 134, 139 ff.). As a result the Gweno paramountcy collapsed definitely, the Gweno people became utterly impoverished, and the Gweno speaking area was depopulated. In 1900, the total population of this area was estimated to be no more than 1000 (Fuchs 1901: 421). While those parts which had an Asu identity, or had meanwhile adopted it, were able to make up for their losses through their trade opportunities, the remaining Gweno lacked this chance because of the inaccessibility of their mountain fastness. Still, those Gweno-speaking parts anywhere near the Sangi-Mbaga chiefdom now began re-orientating themselves economically towards the Sangi-dominated, Asu-speaking markets. After 1900, a new factor began to influence the struggle between Gweno and Athu. This was the Lutheran Mission which decided to concentrate on the use of Athu in their evangelizing activities in the whole of North Pare. How did this decision come about? In 1894 when the first Lutheran missionaries arrived from Kilimanjaro looking for a place to build a station in North Pare, they were not yet aware of the complicated language situation there and assumed that concerning the language of the Wasangi and Wagweno, it appears to be rather closely related to Kichagga, because not only did our Wachagga seem to be able to communicate very well with the local people but we found ourselves that many words are the same or very similar in both languages. Presumably the relationship with the language of the Wasambaa is yet closer, however, because the Wasangi are supposed to descend from the Wasambaa. And, indeed, the sound of the language reminded me very much of that of Kisambaa. (Althaus 1894: 452)

292

Christoph Winter

These missionaries had entered North Pare on foot from the north, passing first through Gweno speaking areas. But they finally headed for the Sangi chiefdom because they found that the Sangi chief Makoko had overpowered the Gweno chief Ndoile (Althaus 1894: 441), and the missionaries considered it important to settle near to the worldly power. Circumstances prevented them, however, from setting up their new station immediately. In fact, six years passed before they returned to North Pare to stay. When they finally moved in, in 1900, the missionaries had a much better command of the Chagga dialect of Wunjo, appreciated the closeness between Chagga and Gweno, and the difference from these of Athu, and reconsidered their earlier choice of location for the new station, especially since also the Gweno chief Ndoile had recovered politically. Their new choice fell on Shighatini in southern Gweno country, just a little north from the Sangi border, mainly because the plot itself appeared the most suitable. They were surprised, though, and a little disappointed to discover that the language in this area of southern Gweno was not Gweno but Athu (Althaus 1900: 432 ff., 457). They had to decide for Athu as the language of evangelization, although those of their early adherents who came from somewhat farther afield, were mostly Gweno speakers and often learnt their Athu, gladly, at the mission station (Fuchs 1901: 421). The missionaries consoled themselves with the consideration that their learning Athu would help them later to expand into Middle and South Pare, as it did, and that Athu was anyhow gaining ground against Gweno. They also expressed hope that they might later turn to Gweno, too, if they should be able to set up another station further north, something that never happened (Fuchs - Rother 1903: 364). In 1913, the mission visitator, Weishaupt (Weishaupt 1913: 126), reported population estimates of 7000 Gweno and 8860 Athu in North Pare but underscored the importance of the Athu language, adding that the Gweno language was perhaps going to die out. According to the census of 1967, these populations in the two administrative divisions of the North Pare Hills had risen to 24,018 in Ugweno and 25,942 in Usangi. While they were virtually all speakers of Athu, and a high percentage were also fluent in Swahili, some part of the people of Ugweno were still also speakers of Gweno. Gweno was still a dying language. The Gweno language had been afforded a new spell of life by the introduction of the policy of "indirect rule" after 1926. The Gweno paramountcy, which had been in decay since 1885, was reconstituted by the British. Although the chiefdoms of the Sangi and Mbaga, and the other Athu-speaking chiefdoms in the South of North Pare, were excluded from it, the rescuscitation of the paramountcy nevertheless helped Gweno-speakers to recover

175 years of language shift in Gweno

293

their ethnic pride and identity. This situation lasted until the abolition of all chieftaincies by the independent government of Tanganyika in 1963. In 1972, in a dialect survey of the Kilimanjaro Bantu languages which I conducted together with Dr. Wilhelm Möhlig, we collected a 600-word list of Gweno. The informants were two secondary school students 17 years old. They did hardly less well than Chagga dialect informants of similar age and background regarding lexical amnesia or interference from Swahili. These students were boarders who had regularly spent long periods away from home. In 1989, however, primary school students in their home area, an area of Ugweno where the Gweno language is still known, were not able to write down in Gweno a local folk tale. Those who were able to do so, were - with one exception - only grown-ups from age 30 upwards. Nowhere in Chagga dialect areas did I meet with a similar situation. I was told in that Gweno community that since some 20 years ago the children were no longer exposed to Gweno, and that instead they were brought up in Athu and Swahili only. It appears that this communal decision to abolish Gweno for the new generation was connected with the decision of the Tanzanian Government around 1967 to emphasize school education in the national language, Swahili. A 30 years old evangelist admitted, however, in 1989 that he would still on occasion mix his preaching and praying in Athu with passages in Gweno for the benefit of elderly people, especially women, among the congregation. So far, we have considered three processes here: (1) The development of bilingualism of Gweno and Asu; (2) the development of a new Asu dialect, Athu, on a Gweno substratum; and (3) the dropping of Gweno, and the sole retention of Athu, by bilingual Gweno communities. There is a fourth process to which I must refer before closing this contribution. This is the gradual Athu-ization of the Gweno language. The 600-word list of Gweno which we collected in 1972, contains a rough average of 15% items which are identical with those in a list of Athu collected at the same time. Some 2% among these are due to parallel borrowing in both languages from Swahili. The lack of further comparative material from Asu prevents me, unfortunately, from determing among the rest which quantities are to be accounted for (a) by borrowing from Gweno into Athu, (b) by borrowing from Athu into Gweno, and (c) by independently occurring sameness of lexical shape. However, special research on one semantic field, viz. kinship terminology, which I did in 1989, leads me to the following observations. Original Gweno kin terms which survive, are only those for members of the common household: Father, Mother, Siblings (opposite sex sibling, same sex elder sibling, same sex younger sibling), Husband, Wife, Son's Wife, Husband's Father,

294

Christoph Winter

and Husband's Mother. The term for Child is the same in both languages. All the terms for relatives living in other households show the influence of Asu: The term for Father's Sister is a uniquely Gweno composition, apparently rather old, consisting of the respective Asu term augmented by a Gweno prefix. Borrowed from Asu are the terms referring to Grand-parents and Grand-children, Mother's Brother and his Nephew/Niece, Cross-cousins, Wife's Parents and Daughter's Husband, and all Brothers- and Sisters-in-law. Isolated data on kin terms which was collected in 1972 in connection with the 600 word list, suggests that this borrowing of kin terms from Asu is not a very recent development. Moreover, the substitution of Asu terms for original Gweno ones has also resulted in adopting the Asu semantic structure of the kinship terminology system. This change altered the predominant structural pattern from Omaha III, or "Shona", to bifurcate-merging, or "Iroquois". The bifurcatemerging structural pattern is produced by a simple merging rule which may be formulated thus: "Let any person's sibling of the same sex as himself (or herself), when a link to some other relative, be regarded as equivalent to that person himself (or herself) directly linked to the said relative." By this merging rule, a male speaker's Brother's Child, and a female speaker's Sister's Child, will terminologically be his or her "child", and the speaker himself or herself will be that "child's" "father" or "mother". Hence there will be "uncle" and "aunt" terms only for Mother's Brother and Father's Sister, and "nephew/niece" terms only for a male speaker's Sister's Children and for a female speaker's Brother's Children. The Cousins (of 1st degree and others classed with them) will be terminologically differentiated: Parallelcousins, i.e. the children of those uncles and aunts who are called "fathers" and "mothers", will be terminological siblings, and only the Cross-cousins, i.e. the children of "uncles" and "aunts" called as such, will be given special "cousin" terms. A further Asu rule specifies that "uncle" and "nephew/niece" be just one reciprocally used term, and likewise "aunt" and "nephew/niece". The Gweno kinship terminology previous to the change was of the Omaha III structural type. This means that in addition to the merging rule just described which will produce a bifurcate-merging pattern, there also was in operation a skewing rule which may be formulated thus: "Let a man's sister, as a linking relative, be regarded as equivalent to that man's daughter as a linking relative." By this skewing rule, a man's Sister's Son or Daughter will be regarded as terminologically equivalent to his Daughter's Son or Daughter, i.e. they will be the same as his "grand-children", terminologically, and hence he, their Mother's Brother (MB), will be the same as a "grand-father" to them, terminologically. His son, i.e. MBS, will also be a terminological

175 years of language shift in G w e n o

295

"grand-father", and will call his matrilateral Cross-cousins "grand-children". The Mother's Brother's Daughter (MBD), however, will not be a terminological "grand-mother", but a "mother", and she will accordingly call her matrilateral Cross-cousins "children". Thus, in an Omaha III pattern there are no special "cousin" terms, because Parallel-cousins are "siblings" and Crosscousins are "grand-fathers", "mothers", "children" and "grand-children". Neither is there a special "uncle" term, because Father's Brothers are "fathers" and Mother's Brothers are "grandfathers". For Father's Sister, however, there is a term "aunt", and there may be special "nephew/niece" terms for her to call her Brother's Children. (Apparently the latter were no longer present in Gweno due to a previous Omaha II pattern, or an incipient development into an Omaha IV pattern.) These are just the most outstanding consequence of this skewing rule. In structural terms, the change in the Gweno kinship terminology achieved neither more nor less than just to abolish this skewing rule and to return to the simple bifurcate-merging pattern without this extra rule. The structural patterns of kinship terminologies are, however, not features unconnected with the rest of a culture. Rather, they are the expressions, in the sphere of language, of certain arrangements in the sphere of social organization. An Omaha III pattern such as the Gweno had before the change, tends to develop in a society where marriage is used to establish long-lasting networks of affinal ties between large numbers of clans which are corporately organized and multilocally spread over a large area, the purpose of this network being the underpinning of the political influence of these clans in an extensive political unit whose leader wields more ritual than worldly power. We can easily see that this kinship pattern was suited to the kind of political state organization which the Gweno had developed since the 15th century. A bifurcate-merging structural pattern, such as developed through the change, expresses a political horizon which hardly transcends the immediate neighbourhood of a couple of villages. Clans are not corporate units, but rather the individual households and families are the bearers of political influence within the small commonwealth, and political marriages are contracted within a narrow circuit of relatives, such as Mother's Brother's Daughter. This situation was increasingly reflected by the political circumstances in Gweno after the collapse of the paramountcy in 1885. This structural pattern has, however, also been traditional among the Asu who never had a state with a paramountcy. Their politically fragmented existence appears, however, to have turned out an asset to them ever since the growth of trading opportunities in the 19th century favoured the economic initiative of individual households over the heaviness of a centralized bureaucracy.

296

Christoph Winter

By the way, it was only with great difficulty that the lost Gweno kin terms and the original structural pattern could be recovered in 1989. During a long unsuccessful search for competent informants, I was directed to, among others, a 75 years old Asu ex-teacher who has been living in Ugweno for many years, and who was considered a great expert on the history of Ugweno, but he knew very little of the Gweno language. Finally I happened upon a pair of brothers, one aged 71 and the other 96. During two hours of intensive brain-racking the younger one came up one by one with the lost lexemes but had to rely on his older brother's explanations as to their proper kin referents. This observation indicates that in this community and/or family the use of the traditional Gweno Omaha III system of kinship terminology was discontinued in the period between about 1910 and 1925. We have seen that the process of Gweno losing ground to Asu in North Pare began with the introduction of Asu speaking Mbaga settlers in about 1808-1824, and that a succession of further circumstances has made the process continue through the following 175 years until now, when it may be nearing its end. On another occasion I have described the death of the language of the Aasáx, a hunter-gather tribe in the neighbourhood of Kilimanjaro (Winter 1979). Aasáx as a living language disappeared within the much shorter period of about 27 years between 1892 and 1919. Its disappearance was achieved in two stages. The first stage began in 1892 when famine-stricken Maasai moved into the Aasáx settlements, demanded to be fed and - since this occupation lasted for years - prohibited the use of their own language by the Aasáx. The result was that the Aasáx language was definitely abolished as a home language and remained only as the men's work language while hunting in the bush, or - after a shift in economic pursuits - while herding their cattle in the bush. The second stage followed in about 1919, when the Maasai prevented the Aasáx from continuing with their cattle-keeping economy. The Aasáx then realized that their existence as a separate ethnic group had become impossible because it lacked a viable economic basis. Hence they decided to disband and individually to join surrounding ethnic groups which still had their economic bases. In both stages, the cause was a coercion which originated outside the group, hence a cause of the "push" type. In the case of Gweno, causes of the "push" type are not readily discernible. We may perhaps mention as a "push"-type cause the fact that the Mbaga settlers who first introduced Asu into North Pare were by a constitutional agreement not permitted to participate in the Gweno initiation rites. They were thus forced to maintain strong links with their place of origin and its

175 years of language shift in Gweno

297

initiation rites, by which reason they upheld their Asu identity and Asu speech in Gweno country. We may also mention as a "push" factor that the Tanzanian government imposed Swahili as the school language, thereby causing Gweno communities bilingual in Gweno and Athu to drop the use of Gweno, so as to ease the learning burden of their young, schooling generation. For the rest, and this means for most decisions which made up the whole process of language shift in North Pare, it appears that not "push" but "pull" factors are accountable. "Pull"-type causes are considerations such as that Athu speakers had better opportunities in trade with caravans, that Athu identity was more prestigious and offered more political security, and that only a command of Athu opened the road to modern education in the mission schools and to jobs in the modern sector. It seems to me that it was this preponderance of "pull" over "push" factors which was responsible for the comparatively slow progress of the language shift from Gweno to Athu.

References Althaus, Gerhard 1894 "Untersuchungsreise nach dem Ugueno-Gebirge", Evangelisch-Lutherisches Missionsblatt (Leipzig), 440-443, 449-453. 1900

"Meine Reise nach Nord-Pare", Evangelisch-Lutherisches Missionsblatt (Leipzig), 430-435, 454-459. Decken, Carl Claus von der 1887-1888 "Wörterverzeichnisse aus dem Ki-Dschagga und Pare", Zeitschrift für afrikanische und oceanische Sprachen, 1: 72-76. Fuchs, Hans 1901

"Nachrichten von der Station Schigatini in Nord-Pare", EvangelischLutherisches Missionsblatt (Leipzig), 420-423. 1902a "Einige Bilder aus dem Leben der Wapare", Evangelisch-Lutherisches Missionsblatt (Leipzig), 224-227. 1902b "Hindernisse der Arbeit in Schigatini", Evangelisch-Lutherisches Missionsblatt (Leipzig), 370-372. Fuchs, Hans - Paul Rother 1903

"Nachrichten aus Schigatini in Nordpare", Missionsblatt (Leipzig), 361-365. Guthrie, Malcolm

Evangelisch-Lutherisches

1967-1971

Comparative Bantu: An introduction to the comparative linguistics and prehistory of the Bantu languages. (4 Vols.) Farnborough: Gregg. Johnston, Harry H. 1886

The Kilima-Njaro expedition. London: Kegan Paul.

298

Christoph Winter

Kersten, Otto (ed.) 1871 Baron Carl Claus von der Decken's Reisen in Ost-Afrika in den Jahren 1862 bis 1865. Leipzig: C.F. Winter. Kimambo, Isaria N. 1968 "The Pare", in: Andrew Roberts (ed.), Tanzania before 1900. Nairobi: East African Publishing House, 16-36. 1969 A political history of the Pare of Tanzania c. 1500-1900. Nairobi: East African Publishing House. O'Barr, William M. 1973 "The Pare of Tanzania", in: Mark A. Tessler - William M. O'Barr David H. Spain, Tradition and identity in changing Africa. New York: Harper & Row, 17-100. Schanz, Johannes 1913 Mitteilungen über die Besiedelung des Kilimandscharo durch die Dschagga and deren Geschichte (Baessler-Archiv, Beiheft IV). Leipzig: B.G. Teubner. Tucker, Archibald Ν. - Margaret A. Bryan 1974 "The 'Mbugu' anomaly", Bulletin of the School of Oriental and African Languages, 37:(i): 188-207. United Republic of Tanzania 1969 1967 Population Census: Vol. 1, Statistics for enumeration areas, Dar es Salaam: Central Statistical Bureau. Weishaupt, Martin 1913 Ostafrikanische Wandertage: Durch das Gebiet der Leipziger Mission in Deutsch-Ostafrika, Leipzig: Ev. luth. Mussion. Winter, J. Christoph 1977 Self-actualization in an African society: Its impact upon the development of Christianity among the people of Old Moshi, Kilimanjaro, during the period from 1870 to 1970. [Habilitationsschrift, Cologne.] 1979 "Language shift among the Aasáx, a hunger-gatherer tribe in Tanzania: An historical and sociolinguistic case-study", Sprache und Geschichte in Afrika 1: 175-204.

Part III

A survey on language death in Africa Gabriele

Sommer

1. Introduction The present survey on language death in Africa is one of the results of a research project on minority languages and language death in East Africa which had been carried out at the Institute of African Languages at the University of Cologne between 1988 and 1990. Being aware of the rather poor data situation on extinct languages in Africa in general, it was decided to provide - as a first step - an overview on the data available on this topic. The following documentation is the outcome of this investigation, which provides a very general and - at various points preliminary survey on those languages, known to us, that are either extinct, in the process of extinction or threatened by extinction. Without trying to be exhaustive, the reference survey in its present form is a synthesis of information from various sources and is hoped to be useful as a data-base for future research. Remaining mistakes or misinterpretations of the published and unpublished material are due to the present author. The conduct of such a survey would have been impossible without the help and encouragement of various scholars working on the languages concerned. I would like to thank all those who have contributed to this data collection in one way or the other. First of all I would like to express my gratitude to Bernd Heine and Matthias Brenzinger of the Institute of African Languages in Cologne who have initiated and encouraged this undertaking and who have commented extensively especially on those entries concerning languages distributed in East Africa. Very valuable and most recent information was obtained through responses to a questionnaire (see below) sent to more than one hundred colleagues working on African languages either as linguists or anthropologists. More than 30 completed copies of the questionnaire returned to us provided a most recent and first hand source of information. My special thanks go to

302

Gabriele Sommer

Monika Esser (Bayreuth University) for her help with the compilation of the final draft of the manuscript. The survey mainly contains - often outdated - information by general handbooks on African languages and other published sources on extinct languages in Africa. As a starting point - and being aware of the fact that minority languages whose speakers are numerically few tend to be affected by language death more easily than larger linguistic groups - it was decided to restrict the information contained in the survey to those speech communities whose number of speakers does not or is said not to exceed 500 speakers (if no other evidence indicates its status as a language threatened by extinction). Language names and variants were differentiated as detailed as possible, distinguishing between self-applied or foreign names, name of the language and name of the ethnic group. In the analysis of the information available the main emphasis was put on the status-categories established (extinct, in the process of extinction, threatened by extinction), the replacing language and the reasons for language shift. A language was regarded as "extinct" if several sources explicitly and independently reported its extinction - fortunately sometimes confirmed by recent empirical data surveying the actual language use and sociocultural background. Languages have been understood as being "in the process of extinction" if a restricted language use or competence in the abandoned language could be deduced from the sources. The languages belonging to this category seem to make up the most promising field of investigation because the sociolinguistic and cultural background of the actual shift situation is still accessible.

2. Questionnaire (1) Name of the language (1.1) Self-applied language name (auto-glossonym) (1.2) Variants of the language name (glossonyms) (1.3) Self-applied name (autonym) (1.4) Name applied by other groups (xenonym) (1.5) Name of the ethnic group (ethnonym) (1.6) Name of river, town, administrative region (toponym) (2) Language classification (3) Language status (3.1) Extinct

A survey on language death in Africa

303

(3.2) In a process of extinction (3.3) Threatened by extinction (3.4) Name of replacing language and/or adjacent people (4) Speakers (4.1) Number of people with a good command of the language (4.2) Total number of people with some knowledge of the language (4.3) Number of people belonging to the same ethnic group (5) Area of distribution (6) Reasons for giving up the language / further information (7) Bibliographical references

3. List and abbreviations of handbooks on which many of the references are based Hb 1 : Grimes, Barbara F. (ed.) 1984 Languages of the world. Ethnologue. (10th edition.) Dallas, Texas: Wycliffe Bible Translators. Hb 2: Mann, Michael - David Dalby (eds.) 1987 A thesaurus of African languages. London - München - New York - Paris: Hans Zell Publishers - Saur. Hb 3: Tucker, Archibald N. - Margaret A. Bryan 1956 The non-Bantu languages of North-Eastern Africa. London: Oxford University Press. Hb 4: Bryan, Margaret A. 1959 The Bantu languages of Africa. London: Oxford University Press. Hb 5: Forde, Daryll (ed.) 1956 Linguistic survey of the northern Bantu borderland. Vol. 1. London: Oxford University Press. Hb 6: Westermann, Diedrich - Margaret A. Bryan 1952 The languages of West Africa. London: Oxford University Press. Hb 7: Guthrie, Malcolm 1953 The Bantu languages of Western Equatorial Africa. London: Oxford University Press.

304

Gabriele Sommer

4. Survey of languages A-Z Agara?iwa (1.1) Agara ? iwa. (1.2) Hb 1 (p. 225) mentions Chishingyini, Ashaganna, Ashinginai and Chisingini as other variants of the language name. Hb 2 (p. 85) lists - besides 1.1 - Cishingyini, Ashinginai, Auna and Salka (see Hansford Bendor-Samuel - Stanford (1976: 102) who add the variants Mawuchi and Ngwsci). (2) Plateau group, Benue-Congo (Niger-Congo). (3.1) According to Hb 1, the language is extinct. Hansford - Bendor-Samuel Stanford (1976: 173) remark that Ashaganna is "possibly extinct". (5) Nigeria, North West State, Kontagora Division (Hb 1). According to Hansford - Bendor-Samuel - Stanford (1976): Kwara State, Borgu Division, Agwarra and Babana Districts, Sokoto State, Yauri Division. (7) Hansford - Bendor-Samuel - Stanford (1976). Ajawa (1.4) (1.5) (2) (3.1)

Äjanci. Äjäwa. North Bauchi, Chadic (Afro/Asiatic). According to Skinner (1974), the language became extinct between 1920 and 1940. (3.4) It was replaced by Hausa (another Chadic language). (5) Nigeria, Bauchi State. (7) Skinner (1974). Ake (1.1) Ake (Hb 2, p. 90). (1.2) Hb 1 and Hb 2 give Akye and Aike as other variants of the language name. (2) Benue, Benue-Congo (Niger-Congo) according to Hansford - BendorSamuel - Standford (1976). (3.3) Probably threatened by extinction. (4.1) Hb 1 (p. 221) estimates the total number of 300 or fewer speakers (reference: SIL*). (5) Nigeria, Plateau State, Lafia Division, Assaikio District. (7) Hansford - Bendor-Samuel - Standford 1976. * Summer Institute of Linguistics

A survey on language death in Africa

305

Akie (1.1) According to Rottland (1982: 25), the autoglossonym (and ethnonym, see 1.5.) is ákiék. This form corresponds with akiék (okiek) used in Kenya. (1.4) Mósiro. (1.5) Ayiek (Hb 2, p. 108), see 1.1. (2) Kalenjin, Southern Nilotic (Nilo-Saharan). (3.2) Probably dying out, i.e. Akie is being replaced by Maasai (accompanied by the process of changing their subsistence base from huntinggathering to agriculture). Maguire (1948) describes Southern Nilotic ("Nandi-speaking" in his terminology) hunter-gatherer groups who lived in the Maasai Plains in Tanzania at the beginning of this century. Some of them are probably identical with or closely related to Akie. He mentions - among others - "Mósiro Dorobo" (Mósiro being a clan name). Maguire uses Mósiro as a collective name for groups who call themselves Mósiro, Médiak or Kisankasa. He remarks concerning the language use of these Mósiro (Maguire 1948: 10): "The language of the Mósiro is dying, as any language except Masai tends to do in the Masai country." (3.4) Maasai and/or Swahili (according to Nurse, p.c.). Maguire comments on the Mósiro that they all spoke Maasai (sometimes imperfectly). (4.1 and/or 4.3) Nurse and Smith (in Blackburn n.d.) estimate fewer than 1,000 speakers. Nurse comments: "Numbers of speakers difficult to estimate, because of various factors: no official figures available: variations in habitat (basically village vs. basically bush); variations in language affiliation (Akie/Maasai vs. only Akie); limited geographical knowledge on part of informants . . . minimum a few hundred, maximum a few thousand (?). Limited impression we got was that at least among village-associated Ndorobo, younger people were less fluent in Akie and more fluent in Maasai and/or Swahili." Rottland (1982: 25) reports that A.N. Tucker found a small group of Akie near Mkomazi (100 km northeast of Kijungu) in 1965 and that he, Rottland, met a group near Kijungu at the Handeni-Kondoa road in 1976. Rottland assumes that there exist other groups in this area between the above mentioned locations. (5) Tanzania, southern part of Arusha Region. According to Nurse and Smith in Blackburn n.d.: "Makami, Kibaya to Ruvu River . . . " and " . . . the southern part of the Maasai Steppe, exact extent unknown, roughly (east) Handeni/Ruvu River across to around Makami, then

306

(6) (7)

Gabriele Sommer

somewhat further west, then south to somewhere south of Kibaya." Rottland (1982: 25) remarks that the Akie live in small groups south of Arusha among Maasai speakers. See 3.2. Blackburn (n.d.); Maguire (1948); Rottland (1982). Lars Christian Smith and Derek Nurse have unpublished material on the Akie.

Alagwa (1.1) (1.3) (1.4) (1.5) (1.6) (2) (3.3)

Alagwa. Alagwa, Alagmo (for male sg.). Waasi. Waasi is a name applied to Alagwa by other groups (cf. 1.4). The village Kolo is their centre (Mous, p.c.). Southern Cushitic, Cushitic (Afro-Asiatic). Hb 2 suggests that the language has already become extinct. Ehret (1980: 12) does not share this view but suggests that Alagwa is at least in a gradual process of extinction. Cf. his remark on the status of Alagwa: " . . . with still many speakers is thought to have relatively few younger speakers and may thus possibly be on the way to extinction." Mous (p.c.), nevertheless, regards Alagwa as a language threatened by extinction. According to his data, all the children play with Rangi children (cf. 3.4) at school and learn better Rangi than Alagwa. (3.4) Alagwa is being replaced by Rangi. (4) Hb 1 (p. 273) estimates 13,000 speakers of Alagwa. Mous (p.c.) reports: "The Alagwa people live in tarafa [section] Kolo in the kata [village] Soera (3/4 Alagwa), Changaa (2/3 Alagwa), Kolo (1/3), Thawi (1/5). The estimations of the percentage of Alagwa living in these villages are from the local authorities in Kolo, the katibu tarafa Kolo [secretary of the section Kolo], The number of people in these villages is according to the 1988 census: Kolo 7666 * 1/3 = 2555; Soera 5807 *l/3 = 2903; Changaa 9582 * 1/5 = 1916; Thawi 9212 * 1/3 = 3071. Total 10445. So there are about 10,000 Alagwa. I have the impression that all Alagwa people have at least a fairly good command of the language. The language is not in a process of extinction. The original settlement is Kolo but there is a large number of Rangi living among the Alagwa in Kolo and in the other areas." (5) Tanzania. According to Mous (p.c.): Wilaya Kondoa, tarafa Kolo, kata Kolo: village of Kolo, not in Bulisa, Achwi. Kata Soera: part of Bukulu, Humáy, Kwadino, not in Soera. Kata Thawi: Thawi juu, not in Thawi

A survey on language death in Africa

(6)

(7)

307

madukani, Sakami. Kata Changaa: Chambwembeni, not in Chololo, Kwamafunche, Túmbelo, Bubuchangaa. According to Mous there is no area which is purely Alagwa. Nearly all Alagwa know Rangi. Alagwa children speak Rangi among each other. Rangi is the dominant language. The cultural identity of the Alagwa is diminishing. Very few people remember the oral history of the Alagwa. Ehret (1980); Elderkin (1976); Tucker - Bryan (1966); Whiteley (1958). An unpublished Alagwa wordlist can be found at the Department of Foreign Languages and Linguistics at the University of Dar es Salaam. M. Mous (University of Leiden) has unpublished field notes on Alagwa.

Amdang (1.) Doombos - Bender (1983: 54) suggest that simí amdangtii 'mouth of the people' is the self-applied language name. (1.2) Several variants of the language name are mentioned in the literature. Due to a confusion in the nomenclature, however, these names do not always refer to the same idiom. Hb 2 mentions Simi Amdangti, Mimi, Mima and Biltine as variants of the glossonym Amdang, the latter of the four also being referred to as "Jungraithmayr's Mimi". Weimers (1971: 848) does not mention "Jungraithmayr's Mimi" but distinguishes between two distinct Mimis both classified as belonging to the Maban subgroup of Nilo-Saharan: 1. Mimi as represented by vocabulary from Gaudefroy-Demombynes and 2. Mimi as represented by vocabulary from Nachtigal. Hb 1 (p. 170) only mentions one Mimi according to the authors also referred to as Andangti, Mututu or Mime. The following data refer to the language called Amdang by Hb 2 (p. 74) with the corresponding ethnonym Mimi. (1.3) See 1.5. (1.5) Mimi. Doornbos and Bender (1983: 54) remark that Mimi is the ethnonym for the groups living in Chad and Mima for those living in the Sudan. The self-applied ethnonym (in the Biltine area) is - according to them - Amdang. The Biltine Mimi are divided into 1. mountain Mimi and 2. plains Mimi. (1.6) Biltine. (2) Hb 2 classifies Amdang as a single unit but also refers to another classification: Fur (Nilo-Saharan). (3.2) Hb 2 comments: " . . . almost extinct (3.4) The authors mention that almost all the "ethnic" Mimi are Arabicspeaking today.

308

Gabriele Sommer

(4.1 and 4.2) Doornbos - Bender (1983: 54) estimate 5,000 speakers of Mimi (in the Wadai-Darfur area?). They comment: "Today the Mimi of Chad and Sudan number approximately 50,000, of whom only a minority from among the Biltine Mimi (in 1962 estimated at 17,000 people) speak the Mimi language. The 1956 Population Census of Sudan listed only eight Mima speakers. Thus the overwhelming majority of this group speaks Arabic." (5) Chad. Hb 1 (p. 170) suggests that Mimi is spoken north of Biltine in East Central Chad and possibly also in the Sudan. According to Doornbos - Bender (1983: 54): " . . . in Chad, (1) North of Biltine near Jebels Baträn and Agán; (2) scattered throughout Wadai as nomadic pastoralists; in Sudan, (3) in Darfur, colonies near Woda'a and Fafa; (4) in Kordofan in Abu Daza district on the border with Darfur, and at Magrür north of Bara in central Kordofan." (6) "Sudanization". Cf. Doornbos - Bender (1983: 43): "The concept of Sudanization will be introduced to explain the disappearance of some languages and the impending disappearance of others, synchronization of dialect clusters in the direction of one dialect, decline of the degree of multilingualism in favor of the lingua franca Arabic, and finally Arabic monolingualism among the younger urbanized members of a few of the language groups." (7) Doornbos - Bender (1983); Greenberg (1972).

Animere (1.3) Animere. Anyimere according Hb 1 (p. 274); Hb 6 (p. 70) and Hb 2 (p. 63f.). (1.5) Animere. (2) Ka group of Togo-Remnant, Kwa (Niger-Congo). (3.2) Accroding to Heine (1968: 74) the language is going to be extinct "in a few decades". (3.4) The Animere-speaking group Heine investigated some twenty years ago spoke Adele (another Togo-Remnant language) as their first language and were bilingual in Twi and Adele, the latter also being the language of the neighbouring people. (4.1 or 4.2) Heine gives no exact figure but mentions that "only elderly people [in the Animere village Kecheibi] speak Animere among themselves". Hb 1 mentions 250 speakers. (4.3) See 4.1 and 4.2. Heine (1968) says that only the members of the clans Nkwantá and Kontrô speak and have some knowledge of Animere.

A survey on language death in Africa

(5) (6) (7)

309

Ghana, Kecheibi and Kunda villages. Language shift to Adele, bilingualism in Twi and Adele. Heine (1968).

Argobba (1.2) The following variants of the language name appear in the literature: Argobba and Arbobbinya. (1.4) Demoz (p.c.) suggests that the xenonym Argobbañña used by Amharas. (2) Ethio-Semitic (Afro-Asiatic). Fivaz - Scott (1977: 70) classify Argobba as a dialect of Amharic, Southern Ethiopie, Ethio-Semitic. (3.2) Demoz suggests that the language is in a process of extinction. This corroborates that of Lipsky ([1967]: 57) who says: "Argobba was spoken in an area north of Addis Ababa and in a few communities south of Harâr, but no more than several hundred persons speak the language, which in the north is being replaced by Amharic and in Hârargë by Gällä." (3.4) Argobba is being replaced by Amharic or Galla, i.e. Oromo, according to Lipsky ([1967]: 57). (4.2) No figures are available besides the rather high number of 2,500 to 3,000 speakers suggested by Hb 1 (p. 178). Its reference is probably to Weimers (1971: 770). Lipsky ([1967]: 57) mentions that only a few hundred speakers are left (see 3.2). (5) Ethiopia, between Käsäm and Borkänna rivers on the Shoan escarpment. (6) Demoz mentions the pressure from Amharic, the national language in Ethiopia, as a reason for language shift. (7) Lipsky (1962 [1967]).

Asa (1.1) Afoku Aasáx, Asá. (1.2) Further names or variants: Asax, Asa, Asa(x), Aramanik. (1.4) Nurse (in Blackburn n.d.) states that the names mentioned under 1.2 are xenonyms. (1.5) Hb 2 gives Asa as an ethnonym. Cf. Winter (1979: 192): "Aasáx means 'hunters' (pi.) generally as well as being the tribal name. The singular is Asiuk."

(1.6) Cf. 1.2: Aramanik. (2) Southern Cushitic (Afro-Asiatic).

310

Gabriele Sommer

(3.1) According to Winter (1979) and Ehret (1980, as quoted in Hb 2) the language has been extinct since 1976. However, since Smith (in Blackburn n.d.) mentions the number of about 100 people who have some knowledge of the language, this statement has yet to be verified. (3.2) It is not clear if there are any speakers left today (cf. 3.1). (3.4) Adjacent people and name of replacing language: Maasai (Eastern Nilotic). (4.1 to 4.3) According to Winter (1979), the last speaker died in 1976. Nurse (p.c.) gives further information which partly contradicts Winter's statement: "In an area with Komolo roughly as its center we unearthed three levels of Aasa-ability: 1. Poor: the speakers or their parents represented the last generation to speak the language. In general, they had not used it regularly for years, or had only heard it and never used it regularly: they could recall something but not much. Such people exist in Komolo, Loiborsoit, and doubtless other places also. 2. "Speak well": we did not meet any of these (probably very few) individuals, because they were off seat on business when we arrived, but we were assured that they exist. 3. Unknown: Because of the rough terrain, we could not reach one group said to be Aasa-speaking . . . : since they lived far away from villages, and were said not to be mixed with Maasai, there is a good chance of reasonably fluent speakers." As mentioned above, Smith gives the approximate number of 100 people who have some knowledge of the language. (5) Tanzania, area centred around Komolo and Loiborsoit, Arusha Region. (6) The reasons for language shift to Maasai are given by Winter (1979) who distinguishes four stages in the development of Aasáx contraction. Concerning the gradual assimilation of Aasáx to the "dominant" Maasai culture and language, Dimmendaal (1989: 24) comments on the attitude of Maasai lodgers towards the language of their "hosts" during a rinderpest epidemic: "The Maasai lodgers would not permit the Aasáx to speak their own language . . . in their presence. The result was that small children no longer learned Aasáx, while older children unlearned it." (7)

Blackburn (n.d.); Dimmendaal (1989); Fleming (1969); Winter (1979).

Azer (1.1) Hb 2 (p. 50) lists Azayr as self-applied name. (1.2) Other variants of the language name are: Azer, Ajer, Azjer. (1.3.) See 1.5.

A survey on language death in Africa

311

(1.4) Hb 2 mentions Azer as a foreign ethnonym. (1.5) Masiin is an ethnonym according to Hb 2. (2) Mande. Weimers (1971: 772) for example classifies Azer as a dialect of Soninke. For a discussion about classification see Monteil (1939). (3.1 or 3.2) Azer has died out in Mauretania and is probably in a process of extinction in Mali (no data available for the latter area of distribution). According to Monteil (1939), Azer was already in a process of extinction in Mauretania in the 1930s, i.e. at the time when the article was written. In the last century it had still been in use according to older sources. Norris (1970-1971: 99) comments on the status of Azer in Mauretania: "Azayr, . . . , has now effectively vanished" and " . . . the last speaker of it in Wädän, 'Abdullah b. Bäbü of the Idaw al-Hâjj, died at the beginning of the century." Or " . . . it is no exaggeration to say that this language, a dialect of Soninke, is dead as far as Mauretania is concerned." He remarks that there are only people left who have a remnant knowledge of Azer ("TiiMt contains a few speakers with a limited vocabulary. . . . Even if it is now impossible to find anyone whose comprehension is other than superficial. Hassäniya-speaking merchants from Tisfrit residing in Nouakchott can however furnish interesting details of vocabulary." He concludes that this remnant knowledge of Azer shows " . . . the profound influence of Azayr on everyday life in parts of Mauretania, particularly in the regions of Tagänit and Adrar". (3.4) Azer has been replaced/is being replaced by Hassaniya, i.e. the variant of Arabic spoken in Mauretania. Kastenholz (p.c.) assumes that Azer is probably being replaced by Manding in Mali. (4.2) Norris (1970-1971: 99) estimates "a few speakers" of Azer in Mauretania, i.e. some families near Tirenni who retained their language. Monteil records that in the 1930s there were only some semi-speakers of Azer left. As mentioned in 3.4 there are "some" Hassaniya-speaking merchants who preserved a superficial knowledge of Azer. (5) Mali. Mauretania, especially the regions of Tagänit and Adrär; also near Tirenni, south of the capital of the Western Hawd. (6) Azer-speaking people gave up their language in favour of Hassaniya after Azer had lost its former importance as a lingua franca, i.e. was replaced by either Mande or Songhay. The reasons for language shift to Hassaniya are not clear. Perhaps it was due to the increasing importance of Hassaniya in the area. (See Monteil 1939: 215.) (7) Monteil (1939); Norris (1970-1971).

312

Gabriele Sommer

Β anta General remark: Our data on Banta are not taken from the handbooks but from an article by Dalby (1963a) where he provided more detailed information on this language. (1.2) Banta. (1.5) Banta. (2) Mel languages, West-Atlantic (Niger-Congo). Hb 2 (p. 36) classifies Banta as a dialect of Temne. Cf. also Fivaz - Scott (1977: 98): Temne subgroup, Mel group, Southern branch, West-Atlantic. (3.2) Dalby comments on the status of Banta: "The Banta language, still spoken by a few old people in the Mende-speaking chiefdoms of Banta and Banta-Mokelle, is in fact an isolated dialect of Temne, with noticeable similarities to the Yoni dialect, and modified through close contact with Mende (and perhaps with Bullom)." (3.4) Banta is being replaced by Mende (see also 3.2). (4.2) Only a few speakers left according to Dalby (1963a). (5) Sierra Leone. According to Dalby (1963a): " . . . the area now occupied by the Banta is only a short distance inland from Sherbro Island, (See map in Dalby 1962.) (7) Dalby (1962, 1963a).

Basa Kontagora (1.1) Basa Kontagora seems to be an autoglossonym according to the information provided by the handbooks: Hb 2 (p. 84); Hb 1 (p. 223). Hb 1 mentions Bassa Kontagora as the only known variant. We assume that what Blench (p.c.) calls Tubasa is identical with Basa Kontagora. He mentions Tubasa as self-applied language name. (1.5) Blench suggests Basa or Bassa as ethnonyms. (1.6) Kontagora is a place name, Basa Kontagora can therefore be regarded as a toponym. (2) Plateau 1A or Western Plateau, Benue-Congo (Niger-Congo). Blench regards Basa Kontagora as a member of the Basa-Kamuku language group. (3.1 or 3.2) According to Hb 1: extinct. Cf. also Hansford - Bendor-Samuel Stanford (1976: 173) who regard Bassa-Kontagora as being possibly extinct. Blench, who investigated the Basa-Kamuku language group in 1987, regards Basa Kontagora as a language being in a process of extinction.

A survey on language death in Africa

(3.4) (4.2) (4.3) (5) (7)

313

Hausa is the replacing language. Blench estimates 10 speakers. According to Blench: circa 30,000 people. Nigeria, northwestern part, northeast of Kontagora. Hansford - Bendor-Samuel - Stanford (1976).

Bati (1.1) Bati according to Hb 2 (p. 130); Hb 1 (p. 149); Hb 4 (p. 15) and Hb 7 (p. 36). (2) Sanaga group, Bantu (A 60 according to Hb 2) Niger-Congo). (3.2 or 3.1) Probably dying out, if not already extinct. (4.1 to 4.3) Hb 4 and Hb 7 report "a small group of people" who speak or spoke the language. (5) Cameroon, around Monatele, "Center South Province" (according to Hb 1). Hb 7 further suggests: " . . . along the west bank of the R. Lihoua, near its confluence with the R. Sanaga". (7) Johnston (1919-1922).

Bayso (1.1) Bayso, Baiso. (2) Lowland East Cushitic (Afro-Asiatic). (3.3) The language is spoken by a small group of people who often also have a knowledge of other languages such as Wolayta (an Omotic language) or (Guijji) Oromo. According to Fleming (1974), most of the Bayso speakers do not know Amharic, the most important language in Ethiopia. Because of their relatively isolated homeland on Gidicho Island and the small number of speakers, their language might be threatened by extinction. (4.2) Hb 1 (p. 179) estimates a number of 500 speakers. (5) Ethiopia, Gidicho Island in Lake Abaya. It is also spoken in a number of villages located along the western shore of Lake Abaya. Cf. Hayward (1978: 540): "Everyone testified to the dry and rocky conditions prevailing on Gidicho, and it seems clear that the increasing exodus to the western shore has been motivated by necessity. Moreover recent changes in Ethiopian rural administration have strongly encouraged the resettlement programme." (6) The information (mainly linguistic) provided by Hay ward (1978) and Fleming (1964) suggests a relatively stable condition concerning the

314

(7)

Gabriele Sommer

linguistic development of Bayso - despite their isolated location and small number of speakers. Fleming (1964); Hayward (1978).

Bedik (1.1) Hb 2 (p. 34) suggests that oNik is the self-applied language name. (1.2) Several other variants can be found in the literature. According to Hb 1 (p. 250): Bedik, Basari du Bandem(b)a, Tenda Bande, Tendanka and Tandanke. (1.3 and 1.5) Hb 2 suggests Bë Dik. (2) Cf. Fivaz - Scott (1977: 101): Tenda-Konyagi subgroup, Eastern Senegal Guinea group, Northern branch, West-Atlantic. (3.2) The classification in this category is only tentative. It stems from Hb 1 (p. 250) where it is suggested that what is called Tandanke "may be dying out". (4.2) 1,000 speakers according to Hb 1. (5) Senegal, southeastern part (according to Hb 1).

Beeke (1.1) (1.2) (1.5) (2) (3.3)

Ibeeke according to Hb 2 (p. 125). Beeke, Beke (Hb 2). Babeeke (Hb 2). Bira-Huku group, Bantu (Niger-Congo). There are only a few speakers of Beeke recorded in the literature. This might indicate that the language is threatened by extinction. (4.2) The only data concerning the number of speakers can be found in Hb 1 (p. 287), Hb 5 (p. 85) and Hb 4 (p. 102): two villages of Beeke-speakers only. (5) Zaire. According to Hb 5 (p. 85): the villages of Ibeke on the AvakubiIrumu road, km. 388, and the village of Ibeke south of River Ituri.

Berti (1.1) The only name mentioned by Hb 2 is Berti. It is understood as an autoglossonym. The authors of Hb 1 suggest that Berti might be a dialect of Zaghawa, a view not shared by Hb 2 which lists it as a separate language.

A survey on language death in Africa

315

(1.5) The entry in Hb 3 (p. 52) implies that Berti is regarded as an ethnonym: "A dialect ? of Zaghawa is spoken by the Berti in the Sudan (2) Zaghawa, Saharan branch (Nilo-Saharan) according to Fivaz - Scott (1977: 7). (3.2 or 3.1) In a rapid (?) process of extinction. Cf. Hb 2 (p. 30). Doombos Bender (1983: 49) regard the language as being extinct. (3.4) The replacing language is Arabic. According to Hb 1 (p. 257): "All the Berti people have adopted Arabic." (4.2) Spoken by "very few speakers" according to Hb 1. Following Doornbos - Bender (1983) we have to assume that no speakers of Berti are left today. (5) Sudan, in the Tagabo Hills, Dar Fur and in Kordofan according to Hb 3. (See map in Doornbos - Bender 1983: 42.) (6) "Sudanization" (see entry Amdang.) (7) Doornbos - Bender (1983); MacMichael (1920). Bete (1.2) Bete. (2) Unclassified. (3.2) According to Hansford - Bendor-Samuel - Stanford (1976: 52) Bete is dying out. (5) Nigeria, Gongola State, Wukari Division, Ayikiben District, Bete town. (7) Hansford - Bendor-Samuel - Stanford (1976). Beygo (1.2) Doornbos and Bender (1983: 56) seem to derive the language name from the ethnonym. (1.4) They mention Bégonga (applied by the Fur), Bégukung (by Tama) and Bégogé (used by Daju) as foreign names. (1.5) While Hb 5 (p. 68) regards Baygo as ethnonym (and further adds the variants Bego, Beigo, Beko, Beygo and Béogé), Doornbos - Bender (1983) mention Beygo. (2) Daju group, Eastern Sudanic (Nilo-Saharan). (3.2 or 3.1) Cf. Doornbos - Bender (1983: 56): " . . . it is probable that the Beygo language, which had only two speakers according to the 1956 Population Census, is extinct." (5) Sudan; according to Doornbos - Bender (1983: 56): "Beygo live southeast of Nyala."

316 (6) (7)

Gabriele Sommer

"Sudanization" (cf. entries Amdang and Berti). Doornbos - Bender (1983).

Birgid (1.2) Several variants of the language name are found. Hb 1 (p. 257), for example, mentions Birked, Birguid, Birkit, Birqed, Murgi and Kajjara. (1.3 and 1.5) Hb 2 (p. 31) and Hb 3 (p. 77) and MacMichael (1918: 43) regard Murgi as the self-applied name/ethnonym. (1.4) Hb 3 suggests that Kajjara is a xenonym used by the Fur (see also MacMichael 1918: 42). MacMichael (1918) adds the xenonyms Kagarügei (used by the Dägu) and Kajargei (used by the Baiqo). (2) Nubian language group, Eastern Sudanic, (Nilo-Saharan). (3.2) The comment mentioned in Hb 3 (p. 77) implies that Birgid is actually in a process of extinction: "Most Birked now speak Arabic, and their own language is becoming extinct." Thelwall (1977: 197) emphasizes that he met only a few speakers of Birgid in 1969 ("The Birgid language is on the verge of extinction, and I was only able to locate a speaker of the language with some difficulty."). His informant and the remaining speakers were said to be over 80 years old at that time. (3.4) Arabic. (4.2) There were only a few speakers of Birgid left in the 1960s. (5) Sudan; according to MacMichael (1918: 42): " . . . to the east of Jebel Marra between Jebel El Haraiz and Dar Rizeiqät (Baqqâra) . . . There is also a small but long-established colony of Birqed a day's journey to the northeast of El Fasher at Turza. At the beginning of the XlXth century there were also some Birqed in Wadai . . . ". Cf. also Hb 3 (p. 77) and Hb 1 (p. 257). (6) "Arabization". (7) MacMichael (1918); MacMichael (1920); Thelwall (1977); Ushari Ahmed Mahmud (1974). Bir(r)i (1.1) Viri. (1.4) Zande Biri. Cf. also Santandrea's remark (1950: 55): "Birri is in all likelihood their Zande name, slightly mispronounced . . . ". (1.5) According to Boyd (p.c.): viri. (2) Cf. Santandrea (1966: 82f.): "Birri stands as a 'central' language, both geographically and linguistically, in the Mbomu-Bahr el Ghazal area,

A survey on language death in Africa

317

showing relationship to the Momvu-Makere etc. (according to de Calonne), the Moru-Madi-Lugbara-Abukaya-Logo and the Bongo group, including the Kresh subgroup (with Aja). Moreover there exists some relationship - too extensive to be casual - between Birri and the Ndogogroup, and, though in a lesser degree, the Feroge-group as well." Santandrea further remarks that the above mentioned Birri should not be confused with (Belanda)-Bviri which is sometimes also spelt Biri and which belongs to the Ndogo group. Birri has to be classified as a member of the Central Sudanic branch of Nilo-Saharan. (3.3) According to Boyd the language is threatened by extinction. (3.4) Biri is replaced by Zande. Boyd (p.c.) comments: "All speakers (even women) are bilingual in Zande. The Biri are treated as a clan by the Zande, and live alongside monolingual Zande speakers." (4.2) Boyd further says that no figures are available. He estimates a maximum of 2,000-3,000 speakers. Cf. Santandrea (1950: 55): "It is difficult to give an idea of their actual number, owing to lack of official data; . . . 4500 Birri are said to be living in Rafai's area. To judge from what happened to other local tribes, they are presumably reduced by now to about a third. At Deim Zubeir I succeeded in counting eight tax-payers." (5) Central African Republic; in a few villages north of Zemio according to Boyd. Cf. Santandrea (1950: 55): "Their home is in the OubanguiChari, beyond the River Warra, a northern tributary of the Mbomu, their neighbours being the Kare (south and east), the Bodo (or Vodo), Sere, Tagbu, Golo (west), not to speak of the omnipresent Zande and Bandia." (6) According to Boyd the distribution and retention of Biri language use in the area north of Zemio can be explained as follows: "The pocket has held out against Zande domination for nearly two centuries (even today: association for the defense of Biri language and culture recently formed). But its continued existence depends on the cultural and economic isolation of the Haut Mbomou province. If young people are drained towards the towns (where the Biri are not numerous enough to form quarters), the language will not be passed on." (7) Santandrea (1950); Santandrea (1966). Boyd has unpublished material on Biri.

318

Gabriele Sommer

Bobe (1.2) Several variants of the language name are found in the literature: Hb 1 lists Bobea, Wovea, and Bota (the latter is the official form according to Ardener 1956). Hb 2 adds Bubia and Woweta. (1.3 and 1.5) Hb 2 refers to the phonetic transcription [uouea] as ethnonym. Ardener (1956: 9) remarks that Wovea is "the term by which the people call themselves to-day". (2) Duala, Bantu (Niger-Congo) according to Fivaz - Scott (1977: 106). (3.3) The rather small number of speakers is the only indication that Bobe is threatened by extinction. Further information is needed to verify this assumption. (4.2) Hb 1 mentions 600 speakers. Cf. also Ardener (1956: 14). (5) Cameroon. Cf. Ardener (1956: 12): "The Wovea occupy the largest island of the small Pirate group in Ambas Bay ("Bota Island") and the village on the opposite coast ("Bota Land"). The village of Mondali (Mondole) on the coast of the Bimbia promontory is also related to them. Formerly the Wovea also occupied the two large islands in Ambas Bay . . . ". (7) Ardener (1956).

Bodo (1.2) The only glossonyms (which might belong to 1.1) are Bodo, Bodo and Boy. (2) Kari language group, Bantu (according to Hb 4). Fivaz and Scott (1977: 106) suggest: Bahr-el-Ghazal group, Bantu (Niger-Congo). (3.2 or 3.3) Seen in comparison with the other languages and dialects of the Kari group this language might already be in a process of extinction, too. See 3.4. (3.4) According to Santandrea (1948: 95) Bodo is mutually intelligible with Bukur (i.e. Boguru) and Huma (i.e. Homa). The author also states (concerning the Bodo language itself): "Very little information can be expected from a tiny community, only a few of whom know anything of their mother tongue, having been for over 150 years, at least, completely under the influence of neighbouring foreign tribes. . . . the Bodo stand at present midway between the Azande and the Ndogo group, partaking almost equally of both." This indicates a process of extinction, absorption - also linguistically - by "the neighbouring foreign tribes".

A survey on language death in Africa

319

(4.2) Hb 4 comments (p. 87): "Number : Few". Santandrea (1948: 95) states: "Some general information. This was gathered at Dem Zubeir. There are here 21 Bodo tax-payers, but they all say that their fellow-tribesmen are more numerous in French territory, mostly in Gubere's area, on the Kere, etc., up the Goangoa . . . , where they live close to the Sere, Tagbu, and Golo." (5) Cf. Hb 5 (p. 126), Hb 1 (p. 257) and Hb 4 (p. 87): Sudan (in the neighbourhood of Dem Zubeir and in the River Warra-River Kerre area (see Santandrea 1948: 79). According to Hb 1: Sudan and Central African Republic. (7) Santandrea (1948).

Bolem (1.2) Besides the autonyms listed under 1.3 Hb 1 mentions the Mmani, Bome (Bum, Bom) and Kafu as dialects of their Northern Bullom (Bome probably being a separate language). One has to add that the data given in Hb 1 (p. 251 ) cause some sort of confusion. The newer data provided by Cameron - Iverson (n.d.) clarify the statements mentioned in Hb 1. (See below.) The authors of Hb 1 mention a language called Northern Bulom which according to them is nearly extinct. They also mention a language called Sherbro (p. 252) which is (according to them) not mutually intelligible with Bullom or Krim and obviously not threatened by extinction. We therefore first of all take into consideration their data that deal with Northern Bullom. The other handbooks use both language names as synonyms. Hb 6 (p. 12f.) differentiates between several sections of Bullom or Sherbro: Bullom who live in the Bome River and are known as Bome or Bum or Bomo; those who live in the Krim River area and who are known as Krim, Kirim or Kittim; another section of Bullom/Sherbro is called Mampa or Mampwa and this denotes the Bullom of Sherbro who are known as Shiba, a corruption of the English sea-bear (this name is obviously a xenonym; cf. 1.4). Concerning a further differentiation on the dialect level Hb 6 comments: "The 'Bulom ' [obviously Northern Bullom] and 'Mampa ' [i.e. Southern Bullom] of Koelle appear to differ; the dialect of Bum is closely akin to that of Mampa. There may be other dialects, as the class system of Bulom, according to Koelle, differs from that given by Sumner, as does the vocabulary to some extent." In general Hb 6 distinguishes between Bulom, Sherbro and Krim. This distinction agrees with the data suggested by Cameron - Iverson which have been verified by their

320

(1.3)

(1.4)

(1.5) (2)

(3.2)

(3.3)

Gabriele Sommer own field research. They treat the languages Bullom So (i.e. Northern Bullom), Sherbro (i.e. Southern Bullom) and finally Krim as distinct languages. Their cover name used for this language group is either Sherbro or Bullom languages (the latter is their preferred variant). They mention Peninsula Sherbro, Shenge Sherbro, Ndema Sherbro and Sitia Sherbro as dialects of the language Sherbro. According to HB 2 (p. 36) and the other handbooks: Butam, Bulom, Bolon, Bullom, Bullun, Mmani, Mandingi, Mandenyi, Sherbro, Serbro, Sebora, Sherbroo, Shenge, Jèqgé, Sitia, Kafu, Bolom and Mnani. Hb 6 states that Bulom (with the variants Bolom, Bullom and Bullun) and Sherbro are synonyms. Kim or Kimi are given as self-applied names (cf. 1.5) for Krim. Hb 6 (p. 12f.) mentions Mampa and Amampa as xenonyms used by Temne and Vai speakers. The name Shiba is used by the Vai for the Bulom of Sherbro (belonging to Southern Bullom). Some of the variants listed under 1.3 are also ethnonyms. Hb 6 gives Kim or Kimi as self-applied names for Krim. According to Hb 1 : Mel group, Southern branch of West-Atlantic (NigerCongo); Hb 1 groups Northern Bullom as belonging to the Kis(s)iTemne group of Southern branch, West-Atlantic. Their so-called Southern Bullom belongs to the Mel group; Hb 6 suggests that Bullom belongs to a language group called Kis(s)i-Landoma. Cameron - Iverson classify Bullom So or Northern Bullom as belonging to the Kisi-Temne subgroup and Sherbro being a member of the Mel subgroup. Both of them belong to the Southern branch of West-Atlantic (see above). Hb 1 (p. 251) classifies Northern Bullom as "nearly extinct". According to the newest data provided by Cameron - Iverson the language Krim is actually in a process of extinction. Cf. their comment: "Our visit to the Krim area confirmed what sources had told us, that the majority of the Krim people no longer speak the Krim language." And: "We were informed . . . of a remote area where a pocket of people including children still speak the Krim language." Considering the figures given in Hb 6 (concerning Southern Bullom/ Sherbro) and keeping in mind that Hb 1 mentions a high number of speakers for Southern Bullom, this variety of the language might not be threatened by extinction. Nevertheless, considering the general comments of Cameron - Iverson, the whole language group consisting of Northern Bullom, Southern Bullom and Krim has to be classified as being threatened by extinction (although several thousand people still seem to speak these languages). Cameron - Iverson comment: "It [their

A survey on language death in Africa

321

survey] has revealed that Sherbro is declining due to advancement of Mende (as well as Temne and Krio) in nearly all but the most remote Sherbro areas." The authors conclude that the Sherbro in general are conceding to the surrounding languages. (3.4) Hb 1 only remarks that the (Northern) Bullom intermarry with Temne and Susu. Cameron - Iverson give further details. In the Lungi area " . . . very few Sherbro speakers were found. Those that were found were from the northern Bullom So speaking area in Samu chiefdom. The predominant languages there are Susu and Temne." In the Western Area Peninsula they " . . . were told of five coastal villages, at the most, that were considered by the inhabitants to be Sherbro villages". The predominant language there is Krio. In Samu chiefdom there are many Bullom speaking areas in remote coastal areas. In Shenge Sherbros are concentrated on the coast, Mende being the predominant language there and Temne and Krio being spoken as well. On Sherbro Island Sherbro is spoken in the interior, whereas Mende and Temne is used on the coast. (4.2) No exact figures are available in Hb 1 concerning Northern Bullom. Here the number of speakers is estimated at 175,000 for Southern Bullom and 12,800 for Krim. Hb 6 suggests 167,200 speakers of Bullom or Sherbro and 44,600 for Krim (the latter figure based on government sources of 1945). Cameron - Iverson do not give any definite numbers but regard the figures given in Hb 1 as being too high. They mention a "relatively small number" of Sherbro speakers and further comment referring to numbers given in the 1978 version of the SIL publication: "The 1978 Ethnologue states the population of Sherbro speakers to be 40,000. From the information we have gathered in the many Sherbro speaking areas we visited, we believe that this is a very high estimate. However, at this time we cannot make an educated guess as to the true number of Sherbro speakers." (5) Sierra Leone; along the coast from the Guinea border to the Sierra Leone river according to Hb 1 (Northern Bullom); Hb 6 mentions Sierra Leone, Sherbro District and coastal villages of Sierra Leone peninsula as areas of distribution for their Bullom/Sherbro. Cameron - Iverson distinguish between the following "Sherbro or Bullom Languages speaking areas": All of them are spoken in Sierra Leone. Bullom So is spoken in the northwest, north of Lungi. Sherbro (and its dialects) are distributed as follows: Peninsula Sherbro is spoken on the coast, southeast of Freetown, Shenge Sherbro in the northwest (to Rotifunk) and east of Shenge. Ndema Sherbro is spoken on Turtle Islands and south of the peninsula.

322

(6)

(7)

Gabriele Sommer Sitia Sherbro is distributed in most of the peninsula area, Sitia Sherbro between Mattru in the north and the coastal area and Krim in the east of the Sitia area and also along the coast. "Three villages in this remote area of Lake Kwako" were said to be Krim-speaking. Cameron - Iverson comment in their sociolinguistic analysis: " . . . in the area of language usage, their [Sherbro] kindheartedness causes them to acquiesce rather than to insist on using their own language when trading with people of another language. This explains why on the main routes there is such a strong presence of Susu, Krio or Mende. We were told that in the case of tribal intermarriage, the Sherbro spouse (whether husband or wife) would learn the language of the one they married. The children would also learn the language of the non-Sherbro parent." The authors further remark: "Yet the trend [of language decline] was already evident over sixty years ago when Sumner (1921: vii) reported on the intermingling of Sherbro and Mende . . . ." Cameron - Iverson (n.d.); Koelle (1854); McCulloch (1964); Migeod (1926); Sumner (1921).

Bonek ( 1.2) The most frequently used variants of the language name, at least in older sources, seem to be Bonek, Bonek, Ponek or Poneck. Some sources, i.e. Hb 1 (p. 148) or Weimers (1971: 785, 776), mention Bonek as a dialect of Banen. Hb 2 also mentions the variants otaqg atomb and otomp as further variants of the glossonym Tuotomp which they use as language name. (1.6) Hb 2 (p. 129) regards Poneck as a toponym. (2) Bantu (Niger-Congo). Fivaz - Scott (1977: 47) classify Bonek as Banen, Basa group, Bantu. (3.2) We tentatively classify the language as being in a process of extinction. We can only refer to the comments in the older handbooks. Cf. Hb 5 (p. 29) or Hb 4 (p. 12): "Spoken only by the village elders, (3.4) According to the same sources Bonek is being replaced by Banen. (4.2) The only figure mentioned in the sources is not more than 50 speakers according to Hb 5 (p. 29). The authors comment: "This figure is an estimate given by the village chief who said that the younger inhabitants now speak Βαηεη." (5) Cameroon. According to Hb 4 and Hb 5: in the village of Bonek on the Ndikiniméki-Bafia road, "just within the limits of Bafia S.D.".

A survey on language death in Africa

(7)

323

Th.C. Schadeberg (University of Leiden) has some unpublished material on Bonek.

Bong'om (1.1) Rottland (1982) gives as /ρο:ηόηι/ self-applied language name. (1.2) Other variants of the glossonym were also recorded: Bong'om, qoma, Bong'omek (see Hb 2, p. 108). Rottland (1982) further adds Ng'oma and Ong'om as variants. (1.3

and probably /po:qóm/.

1.5)

/po:qómek/\

/po:qómi:n/,

/po:qómi:ntét/

and

(2) Southern Nilotic, Nilotic (Nilo-Saharan). (3.2) No definite statements are possible. Nevertheless, Rottland's remarks might imply that Bong'om is in a process of extinction. (3.4) Bong'om is replaced by Luhya according to Brenzinger (p.c.). (4.2) Rottland (1982: 24) estimated five hundred Bong'om-speaking families, i.e. circa 2,500 speakers, in the 1970s. (5) Kenya, southwest and northwest of Bungoma town, mainly around the hills Kapchei, Luchu, Sang'alo and Kimukuni. Also in scattered settlements in Luhya-speaking areas. (7) Rottland (1982).

Boni (1.3) Aweera (see 1.5). Cf. also Tosco (this volume): "Heine refused the denomination "Aweer(a)", claiming it to be generical (meaning simply "hunter"), and adopted the term "Boni", which is today generally used in linguistics, even if "Boni" is in its turn derogatory (being the Somali denomination of various low-caste groups; . . . ". (1.4) Boni, waBoni (which is, according to Hb 2, an ethnonym), Sanye (a term which has also been used to refer to the neighbouring Dahalo and Afaan Oromoo (dialect), cf. Hb 2 p. 14), Waata. (1.5) According to Heine (1982a: 11): "There are two ethnonyms commonly in use for the Boni: Aweer and Boni. While Aweer . . . is used by many Boni people, Boni is a Swahili word (wa-b'oni) which is gradually being adopted by the Boni themselves." Heine (1982a) lists several dialects in his monograph on Boni dialects. (2) Lowland East Cushitic (Afro-Asiatic). (3.3) Probably threatened by extinction.

324

Gabriele Sommer

(3.4) Cf. Heine (1982a: 12): "The Boni have a relatively high rate of multilingual competence . . . Most men and the majority of women know some Swahili which, especially in the northern parts of Boni country, is strongly influenced by the Bajun (Tikuu) dialect of Swahili." And (1982a: 17): "In addition to Swahili, there are two more second languages of major importance, Oromo (Orma dialect) and Somali." Hb 3 states that "Sanye", i.e. Boni, "is fast disappearing in favor of Galla". Hb 1 (p. 203) comments: "Reported to be largely assimilated, and highly bilingual in Somali and/or Swahili." (4.2) Daniel Stiles (in Blackburn n.d.), referring to his unpublished data based on seven field trips between 1978 and 1981 mentions the number of 2,000 Boni speakers in Kenya and probably 1,000 in Somalia. Hb 1 estimates 5,000 speakers. Heine (1982a: 11) estimates between 2,000 and 4,000 speakers. (5) Kenya and Somalia. Cf. Heine (1982a: 11): "The Boni people inhabit the hinterland of the northern Kenya coast as well as the adjacent parts in southern Somalia . . . Administratively, the vast majority of them live in the Lamu District of Coast Province, Republic of Kenya." (6) The authors writing on the Boni agree that the subsistence base of this people is changing or has changed from hunting-gathering to agriculture. Whether this change is accompanied by a language shift is not clear. Cf. Heine (1982a: 11): "Formerly hunter-gatherers who were forced by the administration to adopt a farming existence, most of them [Boni] nowadays live in settlements stringed along the main road between Mkunumbi and Kiunga." Heine further suggests that the population movement in connection with a change in subsistence base has given rise to new dialect distinctions (which could be seen as an argument against the assumption that a process of extinction is taking place or is going to take place in the near furture). (7) Blackburn (n.d.); Heine (1977); Heine (1982a, 1982b); Tosco (this volume). Boro (1.2) Boro. (2) The classification of Boro is not clear. While some authors classify it as a member of the Togo-Remnant group of Kwa (Niger-Congo), Heine (1968: 300) treats Boro as an unclassified unit. (3.1) According to the comments in Heine (1968: 300) and Köhler (1975: 338), Boro is an extinct language which died out in the last century.

A survey on language death in Africa

(5) (7)

325

Ghana, in the areas Worawora and Tapa, east of Asukawkaw. Heine (1968); Köhler (1975).

Buga (1.2) Hb 1 (p. 261) mentions the language names Mangaya, Mangayat, Mongaiyat. The authors state that Mangaya or Bug might be grouped together with the languages Feroge and Indri which Hb 3 (p. 22) lists as separate languages belonging to the same language group (Feroge subgroup of Sere-Mundu). (1.3) Hb 1 (p. 102) claims that Buga is an autonym whereas other sources call this language Mangaya (obviously a xenonym). Hb 3 mentions the self-applied name Buga (vs. the xenonym Mangaya) and distinguishes between the "local variants" Buga, Bugwa, Buka and Bukwa. (1.4) Mangaya. (1.5) Buga (Hb 2), Bug (Hb 1). (2) Sere-Mundu languages (Hb 3), Eastern branch (Hb 1), AdamawaEastern (Niger-Congo). (3.3) The small number of speakers might suggest that Buga is in a process of extinction. (4.2) Hb 1 (citing Weimers 1971) agrees with the figure given in Hb 3 (p. 22) and assumes the number of 300 speakers of Buga. (5) According to Hb 1 : West Sudan, in a small area around Raga.

Burji (1.1) Hb 2 (p. 15) suggests d a a j e as self-applied name. (1.2) Several other variants of the glossonym occur in the sources. The frequently used and best known variant seems to be Burji. Hb 1 (p. 179) and Weimers (1971: 788) add Bembala, Burjinya, Bambara and Bambala. (2) East Cushitic, Cushitic (Afro-Asiatic). (3.3) Our information is restricted to the situation in Marsabit District, Kenya, where the Burji are less numerous than in Ethiopia (cf. Heine - Möhlig 1980). Concerning the situation in Ethiopia Hb 1 comments: "Many of the speakers in Ethiopia are older." For Kenya, Heine - Möhlig (1980) assert that the Burji are gradually giving up their language in favour of the Borana dialect of Oromo. Referring to the situation in Moyale town, Heine - Möhlig remark that one no longer finds any monolingual Burji there.

326

Gabriele Sommer

(3.4) The Burji living in Kenya speak Borana (Oromo) today. (4.2) Heine - Möhlig (1980: 57) mention 1,000 Burji speakers in Kenya. The estimations given in Hb 1 (p. 179) are rather high: 15,000-20,000 Burji in Ethiopia and 10,000 or fewer in Kenya. (5) Ethiopia (south of Lake Chamo) and Kenya, in the high altitude areas of Marsabit, especially around urban centres like Marsabit and Moyale. Heine - Möhlig remark that the Burji originally emigrated from Ethiopia to the Marsabit District at the beginning of this century. They soon grew in number and became an important economic factor in the area. (6) Cf. Heine - Möhlig (1980: 72): "Of the four possible patterns of language knowledge [in Moyale], only three occur: there were no monolingual Burji; all knew Borana, either as first or second language. The distribution of mother tongues is more or less complementary in that Burji is spoken as a mother tongue by people over twenty years old, whereas the population under twenty years have Borana as their mother tongue (first language) . . . ". The motivation for this language shift is not clear. The Burji themselves mention that they have adapted to Borana since they came to settle in the "Borana area". Mixed marriages between Burji and Borana often occur. (7) Heine - Möhlig (1980); Sasse - Straube (1977). Burunge The following data are provided by Mous (p.c.) who collected some material on Burunge during a three day visit to Goima, 9-11 July 1989. (1.1) Buruqge. (1.2) Buruqga iso 'Burunge language'. (1.4) Waburunge (Swahili name applied by the neighbours of Burunge). (2) Southern Cushitic, Cushitic (Afro-Asiatic). (3.3) Mous classes Burunge as a language that is threatened by extinction (see 6). (3.4) Swahili, Rangi. (4.2) Mous remarks that the number of speakers of Burunge is "small": According to the 1988 census, the number of speakers is approx. 19,000 {Rata Goima 15374*2/3=10249; Kata Chemba 7813*3/4=5860 Kata Candama 15717*1/8=1965; Kata Mrijo 7806*1/8=976; Total 19050). (5) Tanzania, tarafa (division) Goima, wilaya (district) Kondoa. In kata (subdivision) Goima and kata Chemba the majority (two thirds and three quarters, respectively, according to local authorities) is Burunge. In kata Candama and kata Mrijo only a small number is Burunge.

A survey on language death in Africa

(6)

327

Mous mentions that Burunge is threatened by extinction. Children also know Swahili and, to a lesser extent, Rangi. The use of Rangi and Swahili increases because of an increase in contact between the Rangi and the Burunge, the latter of which are less numerous today ("Since the number of speakers [of Burunge] is small there is more and more contact with Rangi people for which KiRangi and KiSwahili is used"). Mous also states that intermarriages between Burunge and Rangi frequently occur which result in the assimilation of Burunge to Rangi culture and language ("Originally Burunge were more numerous than Rangi and the decline in number is due to intermarriage with Rangi."). Mous further comments that the neighbours of the Burunge are the Rangi, Gogo, and the Sandawe people. The languages of contact are Swahili and Rangi (the latter used in conversation with Rangi people). Contact with Sandawe (at Farqwa, a small place in the southwest), and occasionally also with Maasai and Datoga occurs less frequently. The Burunge are not in contact with their closest relatives, the Alagwa and Gorowa, because the Rangi live in an area in between.

Buso (1.2) Going through the sources which, once again, contain only meagre information and do not allow any differentiation between self-applied and foreign names one finds Bus(s)o as the only glossonym. (2) Boa group, Adamawa branch, Adamawa Eastern (Niger-Congo). (3.2) Taking into account the really small number of speakers a process of extinction can be assumed. (3.4) The comment of Hb 3 (p. 42) where it is stated that the language use of Buso is limited to the caste of smiths in Bousso town suggests an extensive language contact with speakers of other languages spoken in this town. The whereabouts of these languages are unknown. (4.2) Hb 1 (quoting Weimers 1971) agrees with the figure mentioned in Hb 3 where the number of speakers is estimated between 40 and 50. (5) Chad, Bousso town.

Buy (1.1) (1.2) (1.3) (1.5)

Hb 2 (p. 35) suggests Buy as an autoglossonym. Hb 1 (p. 196) states Kobiana, Cobiana, Ubôi and Buy as glossonyms. Hb 2 lists U-Bôi as autonym. Several authors (see Hb 2) see U-Bôi and Cobiana as ethnonyms.

328

Gabriele Sommer

(1.6) Hb 2 mentions Kobiana as toponym. (2) Eastern Senegal Guinea group, Northern branch, West-Atlantic (cf. Fivaz - Scott 1977). Hb 1 further comments that Kobiana is closely related to Banyum and Kasanga. Hb 2 mentions Konjiagi-Biafada as subclassification. (3.2 or 3.3) The classification of Buy in one of these categories is mainly based on the small number of speakers. (4.2) Hb 1 and Hb 6 (p. 15) both estimate the number of speakers at 300. This corroborates Weimers (1971: 824). (5) Distributed in Guinea Bissau (near Banyum according to Hb 1) and Senegal (the latter according to Hb 2).

Cena In general, data about this language are rather scanty. The only source (see 7) is an article by Wolff (1974-1975) who emphasizes that traces of what he calls Cénà can only be identified as a substrate of Glavda. Other remnants might be found in some other languages related to Glavda such as Guduf or Paduko (the former two spoken in Nigeria, the latter spoken in the north of Cameroon today). Cf. W o l f f s comment (1974-1975: 209) that can be understood as an answer to 6 as well. We treat it as a general remark because these seem to be hypothetical statements about a language no longer existing and of which no records exist at all: "Unbeantwortet bleibt auch hier die Frage nach der Identität der als Substrat durchscheinenden originären Sprache der Cena: Waren sie vor der Übernahme des Glavda bereits Gudufsprecher, oder sprachen sie vor ihrer Emigration aus Cikicfe dort eine dem Guduf nahestehende Sprache, die möglicherweise ebenfalls als Substrat im Cikide-Dialekt des Guduf Spuren hinterlassen hat?" W o l f f s statement about the role of Paduko might suggest that the Cénà speakers gave up their "original" language and another one as well, i.e. Guduf. (Guduf-speakers invaded the area where Cénà speakers formerly lived). His sources for this assumption are oral traditions, his conclusions drawn from the interpretation of these sources being: "Danach wären Paduko einst unmittelbare Nachbarn der Cikicfe und Cena gewesen - vielleicht waren sie sogar Träger derselben Sprache, deren Aussterben durch die Gudufeinwanderung vor etwa 16 Generationen eingeleitet wurde." (1.2) The only name available is Cena. Wolff (1974-1975: 204) also mentions language names formerly used in the literature that might refer to what he calls Cena: Chikide and Chinine (the latter has to be regarded as a toponym).

A survey on language death in Africa

329

(2)

The classification of Cena is not clear at all. The languages that might be closely related to it, thus mentioned by Wolff are: Glavda, Guduf and Paduko all belonging to the Biu-Mandara group of Chadic (AfroAsiatic). (3.1) This language died out several generations ago. (3.4) The Cena speak Glavda today. Wolff suggests that they might have spoken Guduf after they lost their "original" language and before they adopted Glavda. (5) Nigeria, area around Gwoza, especially Chinine village east of Gwoza Hills and Cena hill. (6) Invasion of Guduf-speaking groups in Cena-speaking areas "about 16 generations ago". (7) See entry Gwara (Wolff (1974-1975).

Chamo (1.2) Chamo. Chamo does not appear as a distinct entry in Hansford - BendorSamuel - Stanford (1976: 113, 173) but is treated as a member of "Kuda-Chamo" (see entry Kudu in our survey). (2) Northern Jos, Plateau, Benue-Congo (Niger-Congo). (3.1) While Hansford - Bendor-Samuel - Stanford (1976) remark that the language (i.e. Kuda-Chamo) is "almost extinct", Shimizu (1982b: 167) mentions that there were no speakers of Chama (and Ningi, see separate entry) left when he visited the area between 1973 and 1975. We therefore regard the language as being extinct. (5) Nigeria, Bauchi State, Bauchi Division, Ningi District (cf. Hansford Bendor-Samuel - Stanford 1976). (7) Hansford - Bendor-Samuel - Stanford (1976); Shimizu (1982a); Shimizu (1982b).

Coptic (1.2) Coptic; several dialects are mentioned in the literature: Sahidic, Bohairic, Fayumic, Achmimic, Sub-Achmimic and Middle Egyptian. These dialects had been in use as languages of literature (Literatursprachen) over the centuries. Cf. Krause who states that the dialects mentioned above had all been in use during the 3rd and 4th centuries, whereas Sahidic was mainly used in the 5th and 6th centuries. Bohairic was finally established as the liturgical language of the Coptic Church in the 11th century. The latter dialect is still in use as the liturgical language of

330

Gabriele Sommer

the Egyptian Coptic community today. The term "Coptic" is regarded as a mutilated Greek form used by the Arabs; nevertheless, the form gibtith is also documented. (2) Egyptian (Afro-Asiatic). (3.1 and 3.4) Several statements about the actual date of extinction occur: the various dialects were used successively as colloquial forms of speech and/or as high variants in the literature. Thus the exact date of extinction cannot be fixed. Coptic was gradually replaced as the language of every day use after the Arabic conquest of Egypt in 641 A.D. Only Bohairic has survived as liturgical language until today. Junge (1984) holds the view that the Egyptian language had been replaced gradually by Arabic in the 10th century. He also mentions that this language " . . . wird von den letzten Sprechern in der 1. Hälfte des 13. Jh. in grammatischen Abrissen niedergelegt." Concerning the history of the Coptic language, Stork (1981) suggests that it co-occurred with Demotic as the language of the Christian community in Egypt as early as the 3rd century A.D. He further mentions: " . . . im christlichen Nubien, das kirchenrechtlich zwar unter der Jurisdiktion der ägyptischen Kirche stand, aber staatliche Unabhängigkeit besaß, war Koptisch neben Nubisch und Griechisch in Gebrauch . . . ". The Greek language did not have a strong influence on Coptic although there are a lot of Greek loanwords in Coptic (Thissen, p.c. For a different view see Krause 1980: 731). Vycichl (1930) states that he still met some last speakers of Coptic in the thirties of this century. (4.3) Because of their minority status in Egypt today (adherence to Christianity in a Muslim country) the official government in Egypt does not publish data about the actual number of people belonging to the Coptic community. (5) Egypt; there are Coptic quarters in the big cities such as Alexandria (religious centre, residence of the Pope) or Cairo and in smaller towns all over the country. (6) Dominance (political and religious) of Egyptian Arabs for centuries. (7) Several grammatical descriptions of Coptic translations of ancient Coptic texts exist. Krause (1980); Junge (1984); Stork (1981). Dahalo (1.1) Nurse (1986), one of the sources for the following data, proposes nu. ma guho:ni as a self-applied language name. He mentions Elderkin (1974) as reference. According to Tosco (this volume), Dahalo refer to their

A survey on language death in Africa

331

language as gúhoni 'the language of the people', while the term dáako, sglt. mase, dáakoto, is also understood but might denote only a section of the people (see 1.3 and 1.4). (1.3 and 1.5) According to Nurse (1986), Dahalo is one of several autonyms/ethnonyms. Tosco, however, notes that the term "Dahalo", i.e. d'aháálo, sglt. mase, d'ahááloto, is understood though never used by the Dahalo but is regarded as a xenonym applied by the Swahili. (Cf. Tosco's comments: " . . . the Dahalo pretend that it means 'slave' in their own language" and "The Dahalo we met always referred to themselves as dááko (sglt. mase, dáákoto) . . . ". Nurse suggests that the Dahalo prefer guho gárima:ni 'people of the outside' or guho gwits:o 'little people' as self-applied names. Tosco found that only the latter is actually used. According to Nurse, Dahalo informants list Dahalo and Sanye as names for two groups of Dahalo that formerly existed. Concerning the latter term, he remarks (1986: 268): "In Swahili and European traditions there is historical allusion to 'Sanye/Sanya', but as this term refers indiscriminately to all small gathering and hunting groups of the coast, it is of little specific help for Dahalo history." (See also 1.4 and Elderkin's comment 1974: 2: "It seems that the word Sanye is best dropped, . . . "). (1.4) Various xenonyms occur: Bon (used by the Somali), Waata (by the Orma), M-Sanya/M-Sanye (by the Swahili), M-Kore (by the Lower Pokomo) and Juwaano (a term applied by the Aweera, i.e. Boni). Concerning the names for Dahalo used by Aweera informants, Nurse remarks that Aweera informants also spoke of two groups (of Dahalo), dako and denk. The former term corresponds to one of the names used by the Upper Pokomo, i.e. Mudahalo and M-Dako, respectively. Tosco adds "Dahalo" as a xenonym used by Swahili (see above). (2) Southern Cushitic (Afro-Asiatic). The occurrence of a click sound in Dahalo has led several authors to the assumption of a strong Khoisan adstratum (concerning this question see Ehret 1980). Tosco proposes a subclassification of Dahalo in the East Cushitic subgroup of Cushitic. (3.2 or 3.3) Nurse (in Blackburn n.d.) assumes that Dahalo is not far from being linguistically extinct. Tosco (this volume) concludes that it has to be regarded as an "endangered" language, possibly in a process of extinction. (3.4) Elderkin (1974) remarks: "The vast majority of Dahalo speakers are bilingual to a greater or lesser extent in Swahili. Some are also competent in Aweera and Pokomo." Tosco describes the process of language replacement of Dahalo by Swahili (in its Northern dialect form).

332

Gabriele Sommer

(4.2) While Nurse (1986: 269) estimated less than 500 speakers some years ago, he assumes a figure of approximately 200 speakers today (Nurse, p.c.)· Tosco states that there are approx. 400 people who speak Dahalo today (see his discussion on the various estimations found in the literature while statistics dealing with the number of Dahalo-speakers are totally lacking). (5) Kenya; according to Tosco: Lamu District of Coast Province. Cf. Nurse (1986: 268f.): "Today the Dahalo live in a small area on the northern Kenya coast, just above the Tana River, in a triangle roughly bounded by Mokowe, Witu, and Kipini." (6) Nurse (1986: 269) and others emphasize a change in subsistence base from hunting-gathering to agriculture: "Until recently they lived by hunting, fishing, and gathering . . . Since the government discouraged hunting, they practice some agriculture or work on other people's shambas." Tosco remarks that the economic shift mentioned above is one reason for language shifting in the Dahalo case. He continues to emphasize that the sociolinguistic situation of the Dahalo, being regarded as a so-called "low caste people", is another important factor for language replacement. In mixed marriages, the children are brought up in Swahili today. He comments: "In our opinion, a major role in the decadence of the Dahalo language has been played by the proximity to the Dahalo area of such important centres of the Swahili people as Lamu and, in the past, Witu." (7) Blackburn (n.d.); Ehret (1980); Ehret - Elderkin - Nurse (1989); Elderkin (1974); Nurse (1985, 1986); Tosco (this volume). Dama General remark: Our main source in this case is an article provided by Dalby (1963b) who collected his material in 1961. (1.2) Dama. (1.5) Dama is also used as ethnonym. (2) Mande. Dalby suggests that the ancestors of the Dama were Vai who spoke a language that was referred to locally as the Dama language and described as "a mixture of Kono and Vai", two other languages belonging to the Mande group. (3.1) Extinct. Cf. Dalby's comment: "There is evidence of a Mande language, or dialect, having become extinct within living memory, . . . ". (3.4) Mende. Cf. Dalby (1963b: 50): "Mende is now the only language spoken in the chiefdom" and (1963b: 51): "It is thus clear that Dama was

A survey on language death in Africa

333

already extinct in some parts of the chiefdom by about 1900; since a language would normally take at least one lifetime to die out, Mende must have begun to replace Dama during the early part of the nineteenth century, or before." (4.2) Dalby suggests that no speakers of Dama are left today but nevertheless remarks: "Although the Dama language is now extinct, a few of the older inhabitants in Kpandebu and in Giema were able to quote a number of words from the language, which they remembered from their childhood." According to him the last person who was able to speak Dama properly (in the 1930s) was an old woman living in Lopa (8 km from Giema). Dalby further comments: "The Chiefdom speaker was certain that the language had died out in Giema before 1920, and an old man there (about 65 years old) remembered the language being spoken in the town when he was very young, but only by old people." In the other areas, i.e. at Kpandebu and at Sembehun (near Giema) the language was said to have died out 60 years ago (at Kpandebu) or had become extinct before the oldest inhabitants were born. (5) Sierra Leone, southeastern part, mainly in the Dama chiefdom (i.e. Kpandebu) and in the chiefdom town of Giema. (See map in Dalby 1962.) (6) Dalby (1963b: 53) mentions that the language shift of former Dama speakers to Mende was mainly caused by an early Mende dominance (probably used as a second language at first) in the whole area: "The ousting of the Dama language, by Mende, is yet a further indication that the language-area of Mende has expanded considerably during recent centuries. Mende now represents the most extensive unified languagearea in Sierra Leone, . . . At the present time, Banta [see separate entry in this survey], Bum and Krim [see entry Bolem] are all on the verge of extinction, and Mende has taken their place." (7) Dalby (1962, 1963b). "Degere" General remark: The information on "Degere" is based on data provided by Walsh (1990) who collected his material in the time between August and November 1987. (1.4) See 1.5. (1.5) Degere (adegere) which is a Mijikenda ethnonym; Vuna (avuna) which means 'harvest people' (see Walsh 1990). (2) Southern Cushitic, Cushitic (Afro-Asiatic).

334

Gabriele Sommer

(3.1) Following the hypotheses presented by Walsh (1990), we have to assume that the former language of the Degere is extinct. Walsh assumes that their former language might belong to the Southern Cushitic branch of Cushitic (cf. his remarks about a relation to the Taita Cushites; concerning the latter, see separate entry in our survey). Cf. Walsh (1990: 2): "There are a number of indications that the Degere may be descended from an earlier stratum of population, possibly Southern Cushitic in origin. Considerable interest therefore focuses upon their history and language, especially if further research confirms the suggestion, that some Degere have retained elements of their earlier speech." (3.4) Degere speak Mijikenda (Bantu) today. (4.3) Walsh (1990) estimates that there are no more than one or two thousand Degere people today. (5) Kenya/Tanzania border area, in the immediate hinterland of the East African coast; in the south of Kwale District in Kenya and in the north of Tanga District in Tanzania. According to Walsh, the principal area of the Degere is along and near the rivers Mwena and Umba, west of the main road between Kenya and Tanzania and the border posts of Lungalunga and Holoholo. Some Degere are said to live at Kilibasi, i.e. in Duruma territory. (6) Change in subsistence base and the low social status of Degere are mentioned as main reasons for an assumed language shift to Mijikenda. Cf. Walsh (1990: 2): " . . . Degere have abandoned their former way of life and become sedentary cultivators. Through intermarriage they have adopted the customs, language, and to a large extent the identity of their Mijikenda neighbours." And: "This process of assimilation is evidently incomplete, and the Degere are still looked upon and regarded with suspicion: further cause for concealing their true identity." (7) Walsh (1990). Deti (1.1) Hb 2 (p. 169) gives Detikhwe as autoglossonym. (1.2) Hb 1 and Hb 2 also list other variants of the name such as Deti, DetiKhwe, Teti, Tietie, Kwaa-Khoe. Hb 1 (p. 144), furthermore, mentions the dialect names K'ere-Khwe and Tsh'erekhwe. Voßen (p.c.) lists Deti as language name. (1.4) Voßen gives Mateti or Bateti as xenonyms. (1.5) Voßen mentions Deti as ethnonym. (2) Central Khoisan (Khoisan).

A survey on language death in Africa

335

(3.2 or 3.1) Voßen suggests that the language is "threatened by extinction" (according to the small number of speakers, see 4.1, mentioned by him we regard the language as being already in a process of extinction). (3.4) Tswana has replaced or is replacing the language. Voßen comments " . . . like other eastern Khoe (central Khoisan) groups, the Deti were surrounded for hundreds of years by Tswana-speaking peoples and have adopted most of their cultural values over time." (4.1) Voßen estimates between 5 and 10 speakers today. (4.3) He also suggests a higher figure for those who belong to the same ethnic group: 1,000-2,000. (5) Botswana, Central District, on the western edge of Makgadikgadi Pans, along the Botletle River in villages such as Xhumo and Rakops. (See map in Voßen 1984: 22). (6) See 3.4. (7)

Voßen (1984); Winter (1981).

Dongo Ko (1.1) Hb 2 (p. 103) mentions dbqgo ko as autoglossonym. (1.2) Hb 1 (p. 288) mentions the variant Donga; Hb 3 (p. 27) mentions Dongo which according to them is also an ethnonym. Hb 5 which gives more details about the language status calls it Dongoko. (Cf. Hb 5, p. 113f., where it says: "There are four dialects named after the chiefs: (a) Ndakala; the chief boasts that his is the only chefferie which has retained its independence, and that this is considered to be the purest dialect... (b) Deso; this chefferie has for some time been under Logo domination and the dialect has been influenced accordingly . . . ; (c) Sirika; the same remarks apply as for ( b ) . . . ; (d) Sirika/Djama diverges more violently than the other dialects, due to the fact that in the past they were subjugated by their southern neighbours the Mangutu (1.5) The various handbooks mention dòqgò nyó as ethnonym (sg. cfòqgò gà)· (2) Hb 2 claims it belongs to the Eastern branch, Adamawa Eastern (NigerCongo); Hb 3 groups it as a member of an "isolated language group Mba". (3.3) The comments found in Hb 5 (p. 113) suggest that the language is threatened by extinction: " . . . the Dongonyo are threatened with extinction". (3.4) Nothing definite is known about a replacing language; the neighbouring peoples seem to have had some influence on the speech community

336

Gabriele Sommer

(cf. 1.2 and Pasch 1986: 19: "Zande, Mangbetu und Logo sind die Nachbarsprachen, . . . "). (4.1 and 4.2) The figures found in the literature vary to some degree; Hb 1 (p. 288) estimates 5,000 speakers, following Weimers (1971). Hb 3 records the decreasing number of 4,870 speakers in 1949; Hb 5 comments: "Statistics in 1948 put them at 5,600 strong; by 1949 the figure had fallen to 4,870." Pasch (1986: 19) refers to these figures and remarks: "Die Zahl der Sprecher lag 1949 bei 4,850, war aber abnehmend (5) Zaire, Haut-Zaire Region, east of Watsa (Hb 1), Territoire Faradje, in the neighbourhood of Makolo (Hb 3), on Obi Kibali, the mountain peak in the southeast corner of Faradje Territory, near the Mangutu of Watsa Territory (Hb 5). Cf. also Pasch (1986: 19): "Es [Dongo Ko] wird östlich von Watsa in der Umgebung von Makoro ('Makolo' nach Tucker/Bryan 1956: 27) zu beiden Ufern des Kibaliflusses gesprochen." (See map in Pasch 1986: 419). (7) Pasch (1986). Dulbu (1.1) Hb 2 (p. 19) lists Dulbu as autoglossonym. This would appear to be the only name found in the literature anyway. (2) Jarawan Bantu, Benue-Congo (Niger-Congo) according to Hansford Bendor-Samuel - Stanford (1976). (3.3) Nothing definite can be said about the status of the language. The number of speakers might indicate that the language is threatened by extinction. (4.2) The only information available is the figure of 80 speakers mentioned in Hb 1 (p. 226). See also Hansford - Bendor-Samuel - Stanford (1976: 63). (5) Nigeria, Bauchi State, Bauchi Division, Zungur District. (7) Hansford - Bendor-Samuel - Stanford (1976). Dungi (1.2) Hb 1 (p. 226) and Hb 2 (p. 85) list several language names without differentiating their self-applied or foreign origin. These are: Dungi, Dingi, Dwingi and Dunjawa. (2) Western Plateau, Benue-Congo (Niger-Congo) according to Hansford Bendor-Samuel - Stanford (1976).

A survey on language death in Africa

337

(3.3) Dulbu might be threatened by extinction. (4.2) Hb 1 (p. 226) mentions the number of 310 speakers (a figure from 1949). (5) Nigeria, Kaduna State, Saminaka Division, Kauru District, Dungi town. (7) Hansford - Bendor-Samuel - Stanford (1976). Ega (1.2) Hb 1 (p. 199), from which we have taken most of the information here, lists - besides Ega - the language names Diés and Egwa. (1.5) Hb 2 (p. 59) mentions Ega as ethnonym. (2) Bia language group, Volta Bandama, Ewa (Niger-Congo) according to Hb 1. (3.2) According to Hb 1 the language will soon be extinct ("Nearly extinct"). (3.4) Cf. Hb 1 where it says: "All Ega claim to speak Dida, and Ega does not seem to be known well by younger generation." (4.2) Hb 1 refers to the figure of 67 speakers taken from a 1975 census. (5) Ivory Coast, Southern Department, Dies Canton of Divo Sub-Prefecture and Dies Canton of Guitry Sub-Prefecture. Egyptian Nubian General remark: The information presented in this entry refers to two articles by Rouchdy (1989a, 1989b) whose data are restricted to those varieties of Nubian spoken in Egypt (as opposed to those distributed in the Sudan). The data were collected during visits to the area in 1979, 1981, and 1986. (1.1) Matoki (spoken by the Kenuz in Egypt), Dongolawi (spoken by the Kenuz in the Sudan), Fadicca (spoken by the Fadicca in Egypt), Mahasi (spoken by the Fadicca in the Sudan). (1.2) Egyptian Nubian (as opposed to those varieties spoken in the Sudan). (1.5) Kenuz, Fadicca. (2) Eastern Sudanic (Nilo-Saharan). Cf. Rouchdy's comment (1989: 92): "Both Matoki and Fadicca belong to the Nile Nubian language group and are closely related linguistically." (3.3 or 3.2) Egyptian Nubian has to be regarded as being at least threatened by extinction (see 6). (3.4) The language is being replaced by Arabic. (4.2) Referring to Voegelin and Voegelin (1977) and other sources, Rouchdy mentions the estimations of 50,000 to one million Nubians. Cf. Rouchdy (1989a: 101): "Egyptian census data cannot be used to determine the

338

(5)

(6)

(7)

Gabriele Sommer

number of Nubian speakers since the population is divided into the Muslim and Christian categories, and Nubians are all Muslims, like most other Egyptians." Egypt; Egyptian Nubians were located at the Nile between Aswan and the Sudanese border before the construction of the High Dam in Egypt in the 1960s. 45,000 to 50,000 Nubians were resettled in an area 3-10 kms away from the Nile. This area is called "New Nubia" and is the present location of Egyptian Nubians. Migration of Nubian men to cities like Cairo and Alexandria for decades in order to look for wage labor. The resettlement of Nubians in "New Nubia" (see 5) led to an increase in contact with Arabic-speakers. The higher prestige of Arabic (the official language, language of education, language of mass media and of religion) has caused a decrease in Nubian language use and (mainly in urban contexts) in language competence. Young urban Nubians tend to speak better Arabic than Nubian today (cf. Rouchdy's comments (1989a: 96): "The urbanités attach less importance to their language as a symbol of ethnic-group membership than do the rural dwellers." And (1989a: 97f.): "In urban areas Nubian women are more educated. They go to school and speak Arabic fluently . . . The children grow up speaking Arabic and understanding and generating some Nubian, although faulty in terms of traditional Nubian grammar."). Rouchdy also describes the phenomenon of domain restriction of Nubian language use (1989a: 100): " . . . Arabic is taking over function by function, while Nubian is becoming more specialized to topics related to family occasions." Nubians have developed a negative attitude towards their language and often prefer to use Arabic (cf. Rouchdy (1989b: 266): "The decreased use of Nubian is indeed due to the dominance of the Arabic language . . . the Nubian attitude toward their language is a determining factor as to their willingness in the future to speak Nubian or alternatively to opt for Arabic. This attitude will play a major role in the persistence of the Nubian language or the linguistic tip in favor of Arabic . . . "). Rouchdy (1989a, 1989b).

Ehobe Belon (1.1) Hb 2 (p. 127) lists Ehobs Belon as autoglossonym. (1.2) Hb 2 and Hb 5 also suggest the name of Balondo or Babndo (cf. 1.6). Weimers (1971: 845) mentions Mbo as a cover name for several dialects - one of these is Balondo.

A survey on language death in Africa

339

(1.5) Hb 2 adds the ethnonyms Mlón and Bèlón. (1.6) Hb 5 (p. 24) suggests in a footnote that Balondo is the administrative name. (2) According to Hb 2 the language is a member of A 1, Bantu (NigerCongo). Hb 5 lists it as a member of the "Mbo cluster". (3.3) The small number of speakers might indicate that the language is threatened by extinction. (4.2) Hb 5 estimates the number to be 263 speakers. (5) Cameroon, according to Hb 5: S.D. Nkongsamba to the south of Mount Nlonako. El Hugeirat (1.2) El Hugeirat, El Hagarat, El Hujerat, Al-Hajarät (see 1.6). Hb 1 comments: "Perhaps a dialect of Hill Nubian". (1.4) Hb 2 (p. 30) mentions El Hugeirat as xenonym and toponym. (1.6) See 1.4. Weimers (1971: 800) suggests that El Hugeirat is a place name. According to Jakobi (p.c.) it is an Arabic toponym meaning 'rocks' or 'stones'. (2) Nubian, Eastern Sudanic (Nilo-Saharan) according to Hb 1. (3.3) Despite the varying contributions regarding the numbers of speakers the language is classified as being threatened by extinction. (4.2) Hb 1 mentions two figures taken from different sources. Nothing is known about the reliability of either. According to Voegelin and Voegelin (1977) the number of speakers can be estimated at appox. 200. Another estimation is taken from internal SIL sources from 1978 where 1,000 speakers are recorded. Hb 3 (p. 76) mentions 202 speakers of El Hugeirat. (5) Sudan, West Kordofan on El Hugeirat Hills. (7) MacDiarmid - MacDiarmid (1931). Eliri (1.6) Eliri seems to be a toponym. No other language names are recorded in the literature. Cf. Hb 1 (p. 258), Hb 2 (p. 54), Hb 3 (p. 69), Weimers (1971: 800). (2) Talodi, Kordofanian (Niger-Kordofanian) according to Hb 1. (3.3) The language might be threatened by extinction. Cf. Seligman (19101911: 168): "The Eliri have mixed with the 'Arabs' below to some extent and now only inhabit a small village high up on the jebel, a

340

Gabriele Sommer

few houses close to the Lafofa village, and a settlement at the base of the jebel. A few Eliri men speak the Lafofa dialect, though none of the latter people profess to understand the Eliri dialect and only 2 or 3 mixed marriages are recorded." (4.2) The only figure available is given in Hb 1: 400 speakers. This figure is obviously from Weimers (1971: 800). (5) Sudan, Southern Eliri range. (See map in Seligman 1910-1911). (7) Seligman (1910-1911).

Elmolo (1.2) Elmolo, Gura pawa, Dehes, Ldes according to Heine - Möhlig (1980: 57). (1.3 and 1.5) Sobania (1988: 46) explains that Gurapua is a self-applied ethnonym (cf. also Heine - Möhlig 1980: 175). (1.4) Lides, Dehes, Hereny and Ngimoile are foreign names for the Elmolo applied by Samburu, Rendille, Dasenech and Turkana, respectively (see Heine - Möhlig 1980 and Sobania 1988). (2) East Cushitic (Afro-Asiatic). (3.1) Extinct, although one speaker might still be alive (Heine, p.c.). (3.4) According to Brenzinger (p.c.) Elmolo was replaced by "Elmolo-Samburu". (4.2) Heine - Möhlig (1980) mention 5 speakers of Elmolo, all of them over the age of 50 in the 1970s. Heine (p.c.) assumes that all, with the exception of one, have died since then. (4.3) Tablino (p.c.) estimates 538 speakers belonging to the same ethnic group, 246 of them living in Marsabit District. These figures are taken from the National Kenya Census of 1979. (5) Kenya, Loiyangallani and Elmolo Bay on the southeastern side of Lake Turkana, Marsabit District (Heine, p.c.). (6) Heine - Möhlig (1980: 177) describe the process of language replacement of Elmolo by Samburu as follows.: "The causes of language replacement are complex. Although this process started hardly more than fifty years ago oral traditions record hardly any details about its beginnings. There can be no doubt that it was an intensive cultural and economic contact with the Sampur that ultimately led the Elmolo to no longer use their own language." Although no major economic or other changes took place in the Elmolo society that could explain the motivation for language change, the "brief historical account" shows, however,

A survey on language death in Africa

(7)

341

that a certain correlation exists between the process of language shift and socio-economic developments in Elmolo society. Heine (1975-1976, 1980); Heine - Möhlig (1980); Sobania (1988).

Fumu (1.1) Ifumu might be an autoglossonym (cf. Hb 2, p. 135). (1.2) Other variants of the language name are Fumu or Mfumu. (2) Teke group, Bantu (Niger-Congo). (3.1 or 3.2) Following the statements given in Hb 4 and Hb 7, the language is nearly extinct (the assumed date of extinction given in one of these handbooks goes back to the time before the 1950s). Cf. Hb 7's comment: "This language is described in the work of Calloc'h, but at the present time no trace of the language can be found. It was presumably spoken to the north of Brazzaville by a small group now extinct." Hb 4 mentions: "This language was believed to be extinct, but speakers of it were discovered in 1957." The actual status of the language has therefore yet to be confirmed. (4.2) Hb 4 (p. 30) assumes 200 speakers. Weimers (1971: 803) also mentions 200 speakers. (5) Congo, "around Franceville" (Hb 4), cf. also 3.1 and 3.2. (7) Calloch'h (1911).

Gafat (1.2) Gafat, Gafatinya (Hb 1, p. 180). (1.6) Bender (1976: 8) remarks that Gafat is also the name of a village. (2) Ethio-Semitic (Afro-Asiatic). Cf. Fivaz - Scott (1977: 290) who classify the language as Northern Gurage, Southern Ethiopie, Ethio-Semitic: "The status of Gafat . . . within Northern Gurage and its relationship with other Ethiopie groups is not clear. Its alleged extinct state is also not entirely accepted by all scholars." (3.1) All the handbooks (Hb 1 p. 180, Hb 2 p. 12f„ Hb 3 p. 136) report the language as being extinct. Lipsky ([1967]: 57) comments: "For all practical purposes, Gafat is extinct." We will therefore ignore their further comment that Gafat is a "nearly" extinct language. This decision is supported by the fact that the author does not mention any domains where Gafat is supposed to be still in use today. Bender (1976: 8) also regards Gafat as an extinct language. Cf. also Fivaz - Scott (1977: 290).

342

Gabriele Sommer

(3.4) According to Hb 3 and Bender (1976) Gafat was replaced by Amharic as early as the second half of the last century. (5) Ethiopia, South Blue Nile area. According to Hb 3: "south of Gojam on the Blue Nile". Bender (1976) says that Gafat is spoken in the village Gafat in the Begemidir Province. (7) Bender (ed.)(1976); Lipsky (1962 [1967]). Galke (1.2) The only glossonym mentioned in the literature is Galke. Lacroix (1962) also mentions "dáí; Boyd (1974) reports that Pormi is the name of a closely related language or dialect. (2) According to Hb 1: Adamawa, Adamawa Eastern (Niger-Congo). (3.3) Although the data given in Hb 1 might imply that the language is in the process of extinction, Boyd (p.c.) regards Galke as a language "probably in danger of disappearing". We therefore classify the language as being threatened by extinction. (3.4) Boyd (p.c.) reports that the speakers of Galke are bilingual in Fulfulde (like those of other languages of this group such as Kali, Dama or the Pam-Mono cluster). (4.2) Hb 1 (p. 152) simply says: "Only a few speakers left". Boyd (p.c.) reports that Galke and its closely related languages are spoken by "very small groups of people". (5) Cameroon, Tcholliré. (See map in Boyd 1974.) (6) Speakers of Galke-Pormi, Kali, Dama, and the Pam-Mono cluster are bilingual in Fulfulde (see 3.4). Cf. also Boyd (1974: 15): " . . . le galke: langue parlée dans un quartier de la ville de Tcholliré (Département de la Bénoué). Autrefois, elle était parlée sur une aire plus grande, comme en témoigne le nom de rivière, Mayo Galke, donné par les Peuls puisque les habitants de la région appartenaient à cette éthnie. Une enquête a été menée aussi sur la langue pormi, parlée dans un autre quartier de Tcholliré et considérée par les locuteurs comme un proche dialecte du galke, Γ intercompréhension étant complète." (7) Boyd (1974); Lacroix (1962). Gamo (1.1) Shimizu (1982a: 128f.) suggests tì-gamo as self-applied name. (1.2) Several names occur in the literature: Shimizu says that Butu is the official name; Hb 2 (p. 86) mentions Buta and Butu; Hb 1 (p. 224)

A survey on language death in Africa

343

and Weimers (1971: 809) treat what they call Buta(wa) (according to Hb 2 a xenonym) as a dialect of Butu-Ningi (Hb 1) or Gure-Kahugu (Weimers 1971). (1.3 and 1.5) Shimizu lists -dòò + gamo, pl. à-ndi + gamo as self-applied ethnonyms. (1.4) Butawa (Hb 2) and according to Shimizu: butancii (glossonym) and bùmbùtuu, pl. butaawaa (ethnonym) as applied by Hausa speakers. (2) Northern Jos, Plateau, Benue-Congo (Niger Congo). (3.2) According to the sources Gamo is in a process of extinction: Cf. Hb 1 : " . . . the use of the language is dying out . . . " or "Nearly extinct" and Shimizu (1982a: 128): "Of 32 Butu settlements given by the Chief of Bura, the language is now said to be still in use only in the village of Kurmi." (3.4) The Gamo seem to speak Hausa, according to Shimizu (1982a). (4.1 and 4.2) Varying numbers and statements are found: Hb 1 mentions "few" speakers of the Butu-Ningi cluster. Shimizu (see 3.2) suggests that only one village is still Gamo-speaking. Weimers (1971: 809) mentions 5,000 speakers of their whole dialect cluster called Gure-Kahugu. (4.3) In Hb 1 it says: "There are about 15,000 members of the ethnic group, but the use of the language is dying out." (5) Nigeria, several villages (see Shimizu 1982a: 128); main centre is Bura, the only Gamo-speaking village seems to be Kurmi. (7) Shimizu (1982a). Gana (1.1) According to Shimizu (1982a: 110) the autoglossonym is Gànà. (1.3) Gànà (see 1.5). (1.4) Xenonym unknown: Shimizu (1982a) mentions the fact that Gànà or the Gànà are "unknown outside". (1.5) Gànà. (2) Northern Jos, Plateau, Benue-Congo (Niger-Congo). (3.3 or 3.2) Gerhardt (p.c.) - being aware of the data published in Shimizu (1982a) - suggests that Gànà is at least threatened by extinction. (3.4) Gànà is being replaced by Hausa or Gura (a Jarawan Bantu language). (4.2) The information provided by Shimizu implies that the number of speakers is small. (5) Nigeria, Plateau State, at Bàbban Gànà, Rùkoro and Yààgi (Yargi). (6) Cf. the article by Shimizu (1982b: 167) where he explains that a whole group of languages called "Northern Jos languages" are in a process

344

(7)

Gabriele Sommer of extinction. The informants of Kudu, Takaya, Si, Gana, Ziriya and Sheni (see separate entries in this survey: Kudu, Rishi, Sheni, Taura or Ziriya) he interviewed seem to be the last speakers "remembering the languages". Whereas most of these languages are no longer used (replaced by Hausa or Jarawan Bantu) or - in case of Chamo and Ningi (see separate entries) - are no longer remembered, Gamo (another language belonging to this group, see separate entry) is still spoken in one village. Shimizu (1982a, 1982b).

Ganjule (1.2) The only glossonyms used are Ganjule (frequently used) and Ganjawle. (2) East-Ometo (according to Hb 1), North Omotic, Omotic (Afro-Asiatic). (3.2) The only data available are to be found in Hb 1 (p. 180) which suggests that Ganjule is "nearly extinct". (4.2 and 4.1) Hb 1: very small group of speakers, possibly 50 men. This reference is probably based on Bender (1976: 8) who mentions the same figure. (4.3) The authors of Hb 1 suggest that there are only a few speakers left and all of them are bilingual. The number of people belonging to the same ethnic group is unclear. (5) Ethiopia, on the only inhabited island in Lake Chammo. (7) Bender (ed.)(1976). Geez (1.1) Bender (1976: 9) proposes gihz as self-applied language name, derived - like the ethnonym - from the root g r z 'emigrate, be free'. (1.2) Geez, Ge ? ez, Giiz. (1.5) According to Bender (1976): ^agafozyan. (2) Ethio-Semitic, Semitic (Afro-Asiatic). (3.1) Extinct. Died out as spoken language between the 10th and 12th century and has survived as a liturgical language in Ethiopia until today. It remained the almost exclusive literary language until the 19th century and is still taught in church schools. Because of this use in church ritual and the revered historical chronicles it is considered a sacred language by Ethiopian Christians today (cf. Lipsky [1967]). (3.4) Geez was replaced by Amharic in the south and west and by Tigrinya and Tigré in the north.

A survey on language death in Africa

(5) (7)

345

Ethiopia. Bender (ed.)(1976); Lipsky (1962 [1967]).

"Gemsbok Nama" (1.2) "Gemsbok Nama" according to Hb 2. Winter (1981: 339) mentions Kaokofeld-"Strandläufer", Hai/Aom, Cwaga and Kedi (it is not clear whether these are synonyms for "Gemsbok Nama" or names for distinct idioms). (2) Central Khoisan (Khoisan). (3.1) Hb 2 (p. 167) regards "Gemsbok Nama" as being already extinct. Several members of Köhler's Khoe-Khoe group (as quoted in Winter 1981: 339) have to be regarded as extinct languages (see separate entries Kora and Xiri). (5) South Africa according to Hb 2. (7) Winter (1981).

Gatame (1.1) Bender (1976: 9) mentions gat'ame as self-applied name. (1.2) Hb 1 (p. 181) uses the glossonyms Kachama, Haruro and Aruro. Hb 2 adds the variants Gatama and Gatsama. Weimers (1971: 827) suggests that Kachama is a closely related language or dialect of Koyra. Bender (1976: 9) explains that Haruro means 'boat' in Welamo. (1.3) Hb 2 (p. 27) mentions Gatame as self-applied name. (1.5) Gatame. (2) Ometo, North-Omotic (Afro-Asiatic) according to Bender (1976). (3.3) According to the small number of speakers given in Hb 1 and the high rate of bilingualism among them, the language might be threatened by extinction. (3.4) Hb 1 suggests that the speakers of Gatame are bilingual with Walamo (Wolaytta). (4.2) Hb 1 estimates 500 speakers. This refers to an estimation suggested by Bender (1976: 9). (5) Ethiopia, according to Hb 1. Bender states that the Gatame-speakers are located "on south island of Abaya (Lake Margarita)". (7) Bender (ed.)(1976).

346

Gabriele Sommer

Gomba General remark: The only data available are those provided by Hieda (p.c.). One does not find a language of this name in the handbooks which were used for reference. (1.2) Hieda does not differentiate between self-applied or foreign name. He only mentions Gomba. (2) The classification is unclear. (3.1) Hieda suggests that the language has already died out. (3.4) He mentions a language shift from Gomba to the neighbouring Bume language (which belongs to the Nilotic language group). It was accompanied by cultural assimilation to the Bume people ("... the people merged into a sub-clan of the Bume."). (4.2) Hieda suggests that no speakers of Gomba have survived. The last speakers of Gomba died a few generations ago (cf. 4.3). (4.3) According to Hieda: " . . . there are a few persons who are belonging to the ethnic group which spoke the language till recently." (5) Ethiopia, Labok village along the Omo River in the Southern Omo Province. (7) No published material available. Hieda collected his data during a field research trip to Ethiopia from December 1988 to March 1989.

Guanche (1.2) (2) (3.1) (3.4) (5)

Guanche (cf. entry in Hb 1, p. 162). Berber (Afro-Asiatic). Extinct. Died out in the 16th century. Replaced by Spanish. Canary Islands.

(6)

Spanish invasion and oppression led to the final extinction of Guanche.

Gubi (1.2) Hb 1 (p. 228) mentions Gubi, Guba or Gubawa as glossonyms. The authors list Gubi and Guru as dialects of this language. (1.6) Hb 2 (p. 114) suggests that Gubi is a toponym. (2) Jarawan Bantu, Bantoid, Benue-Congo (Niger-Congo) according to Hansford - Bendor-Samuel - Stanford (1976: 75f.). (4.2) The only data available is to be found in Hb 1: 300 speakers. (5) Nigeria, Bauchi State, Bauchi Division, Zungur District.

A survey on language death in Africa

(7)

347

Hansford - Bendor-Samuel - Stanford (1976).

Gule (1.2) Hb 1 (p. 259) and Hb 2 (p. 76) list Gule, Fungi, Hameg and Fecakomodiyo as language names. (1.3) Anej (which according to Bender (p.c.) is a variant of Hamej). (1.4) Hb 3 (p. 83) suggests that Hameg is a xenonym used by Arabs, and that Fecakomodiyo is a name applied by "the people of Sillok". Gule seems to be a foreign ethnonym (according to Hb 2). (1.5) Gule (cf. 1.4). (2) Koman (Nilo-Saharan). (3.2) Bender's data about a process of extinction corroborates the information given in Hb 3 (p. 83) where it says "the language is dying out". (3.4) Gule or Anej was replaced or is being replaced by Arabic. (4.1) Bender suggests that among the six elderly speakers of Gule he met in 1979 there was only one speaker left who had a good command of the language. This speaker was over 80 years old then. The youngest of them could only recall little of the language. (4.2) Bender estimates between six and a dozen speakers. (4.3) He also estimates a number of about 1,000 people belonging to the same ethnic group. (5) Sudan, Eastern part, Jebel Gule. (6) Bender suggests that what he calls "Arabic-Islamic imperialism" was a reason for giving up the language. (7) Bender (1983). Gwara (1.1) Wolff (p.c.) mentions Gwàrà as self-applied language name for an idiom that he classifies as a dialect of Margi. According to Wolff Gwàrà can only be identified as a "substrate" in Margi today. He also calls it Màrgyi Gwàrà. None of these glossonyms occurs as a separate language entry in the handbooks. (1.4) Wolff's comments on the etymology of the language name Margi (19741975: 201): "Es dürfte in diesem Zusammenhang interessieren, daß ein altes Lamangwort für "Nicht-Moslem(s)" màrgi lautet." This might imply that Margi is a foreign name. (2) Bura language group, Biu-Mandara, Chadic (Afro-Asiatic). (3.2) Wolff (p.c.) suggests that Gwàrà is in a process of extinction.

348

Gabriele Sommer

(3.4) Most of the speakers of Gwàrà are multilingual with Margi and Kanuri. The younger speakers are giving up their language in favour of Hausa. (4.2) Taking into consideration the remarks found in Wolff (1974-1975 and p.c.) one has to assume that the number of speakers is rather small, and possibly decreasing. (5) Nigeria, Borno state. According to Wolff (1974-1975: 187) the groups of people speaking Gwàrà live at Màdsbè, Bà:dlà, Podlàq, Gúlvú, Gyòmoqi, Gázál, Mbalàqgflaq and together with speakers of the Margi dialects Izge and Kanuri in Bi:tà and Tákombaré. In general the Gwàrà live on the northern border of the Màrgyi Bábál territory. (6) Migration and dominance of the neighbouring groups are the cause of loss of the language according to Wolff (p.c.). (7) Wolff (1974-1975).

Gweno (1.2) Gweno. (1.5) Gweno. (1.6) Ugweno. (2) Kilimanjaro Bantu (E 65), Bantu (Niger-Congo). (3.2 or 3.3) We classify Gweno as belonging to this category according to Winter's comment [this volume]: "We have seen that the process of Gweno losing ground to Asu in North Pare began with the introduction of Asu speaking Mbaga settlers in about 1840, and that a succession of further circumstances has made the process continue through the following 150 years until now, when it may be nearing its end." (4.2) Census figures (1967) report an increased number of people (i.e. Gweno and Asu) in Ugweno (24,018) and Usangi (25,942), the administrative divisions of the North Pare Hills. Winter's comments, nevertheless, imply that there are only a few Gweno speakers left today: "While they [Gweno] were virtually all speakers of Athu, and a high percentage were also fluent in Swahili, only some part of the people of Ugweno were also speakers of Gweno." And: "During a long unsuccessful search for competent informants [on Gweno kinship terminology], I was directed to, among others, a 75 year-old Asu ex-teacher who was living in Ugweno for many years, and who was considered a great expert on the history of Ugweno, but he knew very little of the Gweno language. Finally I happened upon a pair of brothers, one aged 71 and the other 96. During two hours of intensive brain-racking the younger one came up

A survey on language death in Africa

(5) (6)

(7)

349

one by one with the lost lexemes but had to rely on his older brother's explanations as to their proper kin referents." Tanzania, north eastern part, northernmost part of North Pare Hills. Immigration of Asu speakers in Gweno territory; internal and external factors that led to a gradual dominance of Asu and their language over Gweno; increase in importance of Asu language (favored by missionaries since 1900) and Swahili (promotion of Swahili through the Tanzanian government in recent times); domain restriction of Gweno language (or rather dialect) use. Winter [this volume]: "I was told [in 1989] in that Gweno community that since some 20 years ago the children were no longer exposed to Gweno, instead they were brought up in Athu and Swahili only." And: "A 30 year-old evangelist admitted, however, in 1989 that he would still on occasion mix his preaching and praying in Athu with passages in Gweno for the benefit of elderly people, especially women, among the congregation." Cf. also Winter's summary [this volume] on the reasons for language shift of Gweno to Asu (i.e. the development of a new Asu dialect, called Athu, with a Gweno substratum). Winter (this volume).

Gyem (1.1) Shimizu (1982a: 122f.) mentions that i-gyem is the self-applied language name. (1.2) The glossonyms found in the literature are - besides Gyem - Gyema, Gema, Gyemawa, Gemawa. Shimizu (1982a: 122f.) suggests that Gyam is the official and best known name. (1.3 and 1.5) Shimizu lists ù-gyem, pl. à-gyem as self-applied ethnonyms. ( 1.4) Shimizu mentions gyemancii (glossonym) and bàgyemii, pl. gyemaawaa (ethnonym) which are used by Hausa speakers. (1.6) According to Shimizu, Gyam is the name of the main settlement, i.e. a toponym. (2) Northern Jos, Plateau, Benue-Congo (Niger-Congo). (3.2) The remarks of Shimizu indicate that Gyem or Gyam is already in a process of extinction. His comments (1982a: 122f.) concerning language status suggest that Gyem is only spoken in Gyam, a settlement situated five miles south of Jenge on the way to Mara. He also claims: "Another group of people, who are called ù-gwà, pl. à-gwà in Gyem, formerly speaking the same language as Gyem, now speak only Hausa." (3.4) Gyam or Gyem is being replaced by Hausa.

350

Gabriele Sommer

(4.2) The only figure available can be found in Hb 1 (p. 229): 100 speakers. (5) Nigeria, Bauchi State, Bauchi Division, Lame District (see 3.2). (7) Shimizu (1982a).

Haal (1.1) Hb 2 (p. 35) mentions Haal as autoglossonym. (1.2) Hb 1 (p. 249, 196) also mentions the language names Kasanga (mainly used variant), Cassanga, Kassanga and I-Hadja. (1.3) Ihaja, Ihaje, I-Hadja. (1.5) Ihaja, Ihaje, I-Hadja, Kassangue, Cassangue. (2) Nun, Northern, West-Atlantic (Niger-Kordofanian) according to Hb 1. Fivaz - Scott (1977: 157) suggest: Kobiana-Banhum subgroup, Eastern Senegal Guinea group, Northern branch, West-Atlantic. (3.3) According to the small number of speakers we classify the language as being threatened by extinction. (4.2) Cf. Hb 1: 400 total. Weimers (1971) estimates 420 speakers. (5) Senegal and Guinea Bissau, in the latter area (cf. Hb 1 p. 196): "remnant living near Felupe, northwest, sparsely populated border area". Weimers (1971: 821) suggests Portuguese Guinea, i.e. Guinea Bissau as area of distribution. Hb 2 (p. 35) suggests Senegal and Guinea.

Hadza (1.1) Hb 2 (p. I l l ) uses Hadza. (1.2) Several variants of the language name are found in the literature: Hatsa, Hadzapi (according to Elderkin (p.c.) -pi is a plural suffix), Kindiga, Kangeju, Tindiga, Wakingdiga, Watindega, Bali. Hb 1 (p. 267) mentions the dialects Kondoa and Merilu. (1.5) Hb 3 (p. 162) mentions Hadzapi or Hadzabi as ethnonyms. (2) Single unit (Khoisan); see Fleming (1986). For a different view see Tucker (1967). (3.3) Despite the varying numbers of speakers that can be found in the literature one might assume that the language is threatened by extinction. (4.2) Hb 1: 2,000; Hb 3: "probably only a few hundreds." Weimers (1971: 811) suggests 200 speakers only. Elderkin (p.c.) mentions the estimation of 900 Hadza speakers in the late 1970s. (5) Tanzania, some distance northwest of the Sandawe, southeast of Lake Victoria (cf. Hb 1); "Kondoa and Mbulu Districts, near Lake Eyasi,

A survey on language death in Africa

(7)

351

mainly to the west" (Hb 3). According to Elderkin (p.c.) the Hadza are located mainly to the east and south of Lake Eyasi. Lars Christian Smith has unpublished material on Hadza. Fleming (1986), Tucker (1967).

Hamba (1.2) The following statements are based on unpublished information provided by Moser concerning a language called Hamba. (1.4) Moser mentions the ethnonyms Hamba and Mahamba as applied by neighbouring peoples. (1.5) He suggests Hamba might be an ethnonym. (2) Moser does not mention any classification of what he calls Hamba. (3.1) Moser suggests that Hamba is most probably extinct. (3.4) He mentions neighbouring languages such as Mwera, Ngindo, Matumbi, Ndonde, Yao and Swahili as replacing languages. (5) Tanzania, northwest of Nachingwea, south of Mbemkuru River. (6) Moser comments: "The Hamba were hunters and gatherers who were resettled and scattered in the 1950s; when their hunting-and-gathering area was planned to be used for ground-nut-plantations. Though that project failed for climatical reasons, the Hamba vanished or were assimilated by neighbouring groups. Already before parts of them had been assimilated by invading Mwera, Ndonde, Ndendeule and Ngindo."

Haraza (1.4) See 1.6. (1.6) Haräza or Jebel Haraza is an Arabic toponym from which the xenonym was derived. (2) Nubian, Eastern Sudanic, (Nilo-Saharan). (3.1) The language is extinct. Cf. Bell (1973: 76, 1975: 1): "Haraza Nubian is, in fact, extinct as a spoken language. However, a few words could still be remembered by several elderly men." And: " . . . Haraza Nubian, a language which is now extinct, but which was once spoken around Jabal al-Haräza in the northern Kordofan". (3.4) Haraza was replaced by Arabic. (4.2) As the language died out several decades ago no speakers are left. The number of people belonging to the same ethnic group is unknown. (5) Sudan, about 200 kilometres west of Khartoum at Jebel Haraza. (6) "Arabicization."

352 (7)

Gabriele Sommer

Bell (1973, 1975).

Holma (1.2) (1.4) (1.5) (2)

(3.2) (3.4) (4.1) (4.3) (5) (6) (7)

Holma. Holma is also used by neighbouring peoples. Holma. Blench (p.c.) does not give any classification of Holma. Hansford Bendor-Samuel - Stanford (1976: 137) mention Holma as a dialect of Nzangi which belongs - according to them - to North-Eastern, BataSubgroup, Bata-Gude group, Central (Chadic). Holma is in a process of extinction according to Blench (p.c.). Blench mentions Fulfulde as replacing language. He suggests 4 speakers. According to Blench: approx. 5,000 people belonging to the same ethnic group. He remarks that the ethnic identity of Holma remains very strong. Nigeria/Cameroon border near Sorau. Blench mentions the prestige of Fulfulde as reason for language shift. Hansford - Bendor-Samuel - Stanford (1976); Meek (1931).

Homa (1.2) The only language name (and/or ethnonym) found in the literature is Homa. (2) Kari group, Bantu (Niger-Congo). (3.3) The small number of speakers might indicate that the language is threatened by extinction. (4.2) Cf. Hb 4 (p. 87): few. This entry corroborates with Hb l ' s information (p. 259): few speakers. (5) Sudan, around towns of Mopoi and Tambura. Isuwu (1.2) Several variants of the language name can be found in the literature: Subu, Isubu, Bimbia, Su, Isu. (1.5) Hb 2 (p. 128) Isuwu as an ethnonym, see Ardener (1956: 9). (2) Duala group, Bantu (Niger-Congo). (3.2 or 3.3) Varying statements are found: Hb 2 (p. 128) suggests that Isuwu has become nearly extinct (cf. their comment on languages like Isuwu (Hb 2, p. 8): " . . . language which is extinct or virtually so, i.e. which is

A survey on language death in Africa

353

no longer being transmitted to the younger generation." Hb 1 (p. 160) suggests a relatively high number of speakers (a number that goes back to 1982 and therefore is the most recent one). This implies that Isuwu has to be classified as belonging to category 3.3. Hb 5 (p. 21) mentions fewer speakers and several factors that might indicate a process of extinction (high death rate, migration, contact to Duala speaking peoples). Hb 4 ' s statements (p. 7) about the status of the language are ambiguous concerning the estimated number of speakers. If one takes the smaller number of speakers a classification as belonging to 3.2 seems to be adequate. (3.4) According to the statement found in Hb 5 ("The male speakers of the language are mostly fishermen who are in close touch with Duα/α-speaking peoples. There is a strong blood relationship between the people of Bimbia and those of Douala.") one might assume bilingualism with Duala. (4.2) Hb 5: 652 speakers (cf. their comment on the reliability of this statement: "This was of course in 1921. Since then a high death-rate and emigration to Victoria have greatly reduced this community. An elder of the tribe estimated their numbers in October 1949 at less than 500 although he admitted that accurate figures would be hard to come by owing to the exodus to Victoria." Weimers (1971: 88) obviously takes this figure as a basis for his estimated 652 speakers, too. Hb 4: Number: under 500 or 140 (according to their different sources). Hb 1: 800 speakers (referring to their own sources, SIL 1982). (5) Cameroon, west of Douala and east of Victoria along coast, South West Province (Hb 1), "two small enclaves on the Atlantic coast, east and west of Victoria in Bakwiri Terriroty. Bimbia, Likolo, Wuna gombe and Wunabile are three of the localities concerned." (Hb 5): "B.C. [i.e. British Cameroons], on the extreme south coast of Bimbia promontory . . . " (Hb 4). (7) Ardener (1956).

Izora (1.1) Hb 2 (p. 86) mentions Izora as a self-applied language name (cf. Shimizu 1979: 267). (1.2) Hb 1 (p. 225) suggests Chokobo and Cikobu as further variants of the language name. (1.3) Azora.

354

Gabriele Sommer

(1.4) Shimizu (1979) reports that the Hausa names for Izora are cokobancii (glossonym) and cokobaawaa (ethnonym). (1.5) Azora. Shimizu (1979: 267) regards nda zòrà, pl. à-zòrà as self-applied ethnonyms. (2) Northern Jos, Plateau, Benue-Congo (Niger-Congo). (3.3) Nothing specific is known about language status. Considering the small number of speakers it might be classified as being threatened by extinction. (4.2) The only figure available is given in Hb 1 (p. 225), Weimers (1971: 791) and Hansford - Bendor-Samuel - Stanford (1976: 59): 424 or 425 speakers. This figure is from 1936. (5) Nigeria, Plateau State, Jos Division, Jere District (northeastern part according to Shimizu 1979, roughly the area between Gambo River and Limoro-Cikobo Forest Reserve). (7) Hansford - Bendor-Samuel - Stanford (1976); Shimizu (1979).

Jelkuna (1.1) Hb 2 (p. 49) mentions Jelkuna as autoglossonym. (1.2) Hb 1 (p. 281) and Hb 2 suggest that Ble is another variant of the language name. It is not clear whether this is a xenonym. Weimers (1971: 798) mentions Ble as a dialect or closely related language of Dyula but does not mention the language name Jelkuna in this context. Ble might be another variety of the so-called Ligbi dialect cluster according to Kastenholz (p.c.; cf. entry Ligbi). Concerning this problem of classification see Prost (1968). (1.3) (1.5) (2) (3.2)

Dyala Nu and Dyala are listed as autonyms/ethnonyms. Dyala Nu, Dyala. Mande. Considering the small number of speakers given in several handbooks one might assume that Jelkuna is in a process of extinction. (3.4) Hb 1 (p. 281) only mentions bilingualism with Jula ("The people say that everyone speaks Jula"). Hb 1 (p. 283) and Weimers (1971: 798) regard Jula as a synonym for Dyula. Nothing is known about a replacing language. (4.2) Hb 1: 538. Hb 6 (p. 35): 200 or 500 speakers (according to their various sources). (5) Burkina Faso (according to Hb 2); according to Hb 1: village of Blédougou, west of Banfora, near the town of Sindou. Hb 6: "in Haute Volta, Cercle Bobo Dioulasso, among the Soninke north of Banfora."

A survey on language death in Africa

(7)

355

Prost (1968).

Jimi (1.2) The only language name found in the literature is Jimi (see Hb 1, p. 232). (2) Eastern, Southern Bauchi, Western, Chadic (Afro-Asiatic) according to Hansford - Bendor-Samuel - Stanford (1976). (3.3) The small number of speakers might indicate that the language is threatened by extinction. (4.2) The only figures available can be found in Hb 1 (p. 232): 400 speakers and Hansford - Bendor-Samuel - Stanford (1976: 97): 250 speakers. (5) Nigeria, Bauchi State, Bauchi Division, Ganjuwa District, Jimi village. (7) Hansford - Bendor-Samuel - Stanford (1976).

Ju (1.2) Ju seems to be the only known language name. (2) Guruntum subgroup, Eastern, Southern Bauchi, Western, Chadic (AfroAsiatic) according to Hansford - Bendor-Samuel - Stanford (1976). (3.2 or 3.3) The rather small number of speakers might indicate that the language is not only threatened by extinction but is already in a process of extinction. (4.2) The only figure available can be found in Hb 1 (p. 232): 150 speakers. (5) Nigeria, Bauchi State, Bauchi Division, Zungur District. (7) Hansford - Bendor-Samuel - Stanford (1976).

Kakolo (1.1) Kàgorokân. Hb 2 (p. 49) mentions Kakolo Qaqo as autoglossonym. (1.2) Hb further mentions Kagoro and "Fula-Bambara" as other variants of the language name. (1.5) Kàgoro. (2) Manding subgroup of Mande. Vydrin (p.c.) suggests "North-Western branch of Mandeng" for subgrouping. (3.3) Kakolo is threatened by extinction. (3.4) It is (according to Vydrin who refers to Creissels 1986 and DNAFLA 1983) being replaced by Bamana (i.e. Bambara) and Soninké (Saracollé).

356

(5)

(6) (7)

Gabriele Sommer

Mali, Kulikoro and Kayes Provinces, mostly in Kaarta-Biné, i.e. area between Baoulé and Darrouma rivers. Vydrin lists four villages (at least) populated by Kakolo speakers: Sebekoro, Sikoro, Kulukono and "Jankute Karta". In Wanina village Kakolo people live together with Bamana (Bambara), and in Mara (not far from Nioro) and Turunkunbé they coexist with Soninké (Sarakhollé). Vydrin (referring to DNAFLA 1983) further mentions the following areas of distribution: Kakolo is still spoken in Séféto arrondissement (i.e. in Kaarta-Biné); three villages are mentioned: Siramisen, Nahadji and Jungunte (i.e. "Jankute Karta")· There are also Kakolo villages surrounded by the Bamana population: Sebekoro, Missira, Tiokumbougou (Beledougou region), Waaro, Kamiko, Sirado, Madina-Kagoro (Banamba cercle), Kakolodougou-Ségala (Nyamina arrondissement)·, at least in one of them, KakolodougouSégala, the Kakolo language is being replaced by Bamana, and in Lambidou village, the Kakolo abandon their language for Soninké. Creissels (1986: 1) mentions the villages Kourougué, 40 km south of Diéma, and probably also Foulabougou (40 km southwest of Kourougué). Expansion of Soninké and Bamana languages, migration of Kakolo to urban centres such as Bamako. Bird (1982); Creissels (1986); DNAFLA (1980, 1983); Galtier (1980).

Kande (1.1) Hb 2 (p. 133) seems to use Okande as self-applied language name. (1.2) Other variants of the language name are found: Hb 1 (p. 185), Hb 4 (p. 25) and Hb 7 (p. 64) use Kande and Kanda. (2) Tsogo group, Bantu (Niger-Congo) according to Fivaz - Scott (1977: 155). (3.2) According to the various sources the language is "nearly extinct" or "almost extinct". (4.2) The sources either merely comment: "very small group of speakers" or give no figures at all. (5) Gabon, Ogooue-Ivindo Province, west of Booue (according to Hb 1). The other sources simply mention that it is spoken "near Booue".

Kazibati (1.2) Hb 2 (p. 101) and Hb 3 (p. 40) mention several variants of the language name. These are Kasiboti, Ganzibati, Hadjibatie. (1.3) Hb 2 suggests KazißAti as autonym.

A survey on language death in Africa

357

(1.5) KazißAti is given as ethnonym. (2) According to Hb 2: member of the Baya-Ngbaka-Manzi group, Eastern branch, Adamawa-Eastern (Niger-Congo). (3.3) Due to the small number of speakers the language is tentatively classified as belonging to the category of languages that are threatened by extinction. (4.2) An estimated figure of 365 speakers is mentioned in Hb 3. This estimation corresponds to the figure given in Hb 5 (p. 104). (5) Zaire, according to Hb 3 in Territoire Watsa, near Makolo (Ara village). See map in Costermans (1938: 179). (7) Costermans (1938).

Kiballo ( 1.2) The only variants of the language name that can be found in the literature are Kiballo and Kiwollo. (2) Western Plateau, Benue-Congo (Niger-Congo) according to Hansford Bendor-Samuel - Stanford (1976). (3.3) The language might be threatened by extinction. (4.2) The only figure available can be found in Hb 1 (p. 234) where the authors estimate 500 speakers. Hansford - Bendor-Samuel - Stanford (1976: 107) mention 335 speakers (a figure from 1949). (5) Nigeria, according to Hb 1: Kaduna State, Saminaka Division, Kauru District. (7)

Hansford - Bendor-Samuel - Stanford (1976).

Kidie Lafofa (1.1) Hb 2 (p. 54) mentions Kicfie Lafofa as autoglossonym. This statement agrees with the autonym mentioned in Hb 3 (p. 70). The dialects of this language are Jebel el Amira and Jebel Tekeim (which according to Hb 2 and Hb 3 are toponyms; cf. 1.6). (1.5) According to Hb 3: αηιεε lafofa. (1.6) Dialect names: Jebel el Amira, Jebel Tekeim. (2) Talodi, Kordofanian (Niger-Kordofanian). (3.3) The rather small number of speakers might indicate that the language is threatened by extinction. (4.2) In Hb 3 the following figures are given: Lafofa: 501 speakers; Jebel el Amira dialect: 60 speakers; Jebel Tekeim dialect: 45 speakers. Hb 1 (p. 261) gives higher figures such as 2,450 speakers of Kicfie Lafofa

358

(5)

(7)

Gabriele Sommer

including 2,000 Lafofa, 250 Jebel el Amira and 200 Jebel Tekeim. The authors do not mention their source for these data. Sudan, according to Hb 1: Central Eliri range and on two hills to the south and east; according to Hb 3: in Lafofa village and nearly settlements in central Eliri, and on hills Umm Shatta and Takei. Cf. Seligman (1910-1911: 170). MacDiarmid - MacDiarmid (1931); Seligman (1910-1911).

Kimanteney (Quara) (1.1) Hb 2 (p. 16) mentions Kimanteney as authoglossonym. Bender (1976: 10) lists kmanta 'people' as self-applied name. (1.2) Several other variants of the language name are found in the literature: Kemant, Kimanta, Qemant, Kimantinya, Chemant, Kamant. Hb 2 mentions East Kimanta and West Kimanta as dialects. Hb 1 (p. 183) adds Qwera (Quara, Kara of Kwarasa), Kayla (Falasha, Felasha or Yihudi) and Qemant as dialect names. Hb 1 mentions that the dialect Quara is extinct. This "dialect" is mentioned in Hb 2 (p. 16) as a separate language called Beta Israel (ethnonym). The authors list Fälasa, Falasha as xenonyms and Qwara, Quara, Hwara, Yihudi and Kayla as other variants of the language name. Gamst (p.c.) proposes that "Quara, Quarinya, spoken by a few Falasha in Quara (to the west of Qemantland), is nearly the same language as Qemant." We are therefore treating both "languages" together in this survey, which contradicts Bender who regards Kemant and Felasha as two separate languages. (1.3) Gamst mentions Këmant and Këmantnây as self-applied names; Hb 2 mentions Kimanta. (1.4) Cf. 1.2 where we mentioned Fälasa and Falasha as xenonyms for a "dialect" of Kemant. Hb 1 adds that the Falasha are also referred to as "Black Jews". Bender (1976: 8) treats Felasha (from Geez falas 'immigrant') and Yihudi as variants of the glossonym Felasha while he mentions that Kwara or Qwara is the name for a region northwest of Lake Tana. (1.5) Gamst suggests Qement or Qemantiya as ethnonyms. Hb 2 mentions Kimanta. (2) Central Cushitic, Cushitic (Afro-Asiatic). (3.1) Cf. Hb 1 (p. 183) which claims that Quara, a dialect of Kemant, is already extinct. Bender (1976: 8) remarks concerning Felasha: " . . . it is reported that a Cushitic language is still used in religious services". (3.3) According to Gamst, Kemant is threatened by extinction.

A survey on language death in Africa

359

(3.4) The different sources agree in the assumption that probably all or most speakers of Kemant (and Quara) are bilingual with Amharic. Bender (1976: 8) reports that Felashas now speak Amharic. (4.1) Gamst estimates a few hundred speakers as having a good command of Kemant. (4.2) He estimates about 1,000 speakers. Hb 1 gives the total number of 17,000 speakers of what is called Qimant. (4.3) Gamst estimates a number of 25,000 people belonging to the ethnic group of Kemant (who are according to him a division of the Agaw peoples). (5) Ethiopia, according to Hb 1: North of Lake Tana, Central Begendir Province, northwest, also Sudan, Gamst suggests the countryside extending to the north and south of the east-to-west Gonder-Aykel road as well as the east of Gonder and the west of Aykel as areas of distribution. (6) Gamst mentions as reason for language shift of Kemant to Amharic: Acculturation towards Amhara culture, including Christianization. (7) Bender (ed.)(1976); Gamst (1969).

Kinare (1.2) (1.6) (2) (3.1)

(5) (6) (7)

Kinare, "Okiek of Kinare" (/akié:k pa kínárel. Cf. Rottland (1982). Kinare is also a place name. Southern Nilotic (Nilo-Saharan). Extinct. Cf. Rottland's comment on language use of some elderly men in Kinare (1982: 25): "Ihre Eltern hatten jedoch als honigsammelnde Okiek in den Wäldern um Kinare gelebt. Sie selbst benutzten die Sprache ihrer Eltern nicht mehr, konnten aber auf Verlangen ein Vokabular und einige Sätze aus der Erinnerung hervorholen." Kenya, eastern slope of Rift Valley, near Kinare village. Adoption of Kikuyu culture and language. Rottland (1982).

Kinuku (1.2) Kinuku is the most frequently used name (found in Hb 2, p. 86; Hb 1 and Weimers 1971: 823). Other variants used in the literature are Kinugu or Kinuka. (2) Western Plateau, Benue-Congo (Niger-Congo).

360

Gabriele Sommer

(3.3) The language has to be classified as being threatened by extinction. There are, however, no reliable data available. (4.2) The only figure to be found is an estimation of 500 speakers of Kinuku (Hb 1, p. 234). Cf. also Hansford - Bendor-Samuel - Stanford (1976: 108): 460 speakers (a figure from 1949). (5) Nigeria, Kaduna State, Saminaka Division, Kauru District. (7) Hansford - Bendor-Samuel - Stanford (1976).

Kiong (1.1) Hb 2 (p. 82) mentions ki- όη as self-applied name. Hb 1 (p. 234) mentions Kiong as the most frequently used variant. (1.2) Other variants of the glossonym are: Kiong, Akayon, Akoiyang (cf. Weimers 1971: 768), Okonyong, Okoyong, okoqoq and Äkäyöfl. (1.4) According to Hb 2 that variant okoyoq is a foreign glossonym and ethnonym applied to the Kiong by the Efik. (2) Upper Cross, Delta-Cross, Cross River, Benue-Congo (Niger-Congo) according to Hansford - Bendor-Samuel - Stanford (1976). (3.2) According to the statement found in Hb 1 (p. 234) the language is "nearly extinct". This view is confirmed by Cook (p.c.) who comments: " . . . kiòq, an Upper Cross language north of Efik which is on the verge of extinction." Cf. also Hansford - Bendor-Samuel - Stanford (1976: 108): "Spoken only by old people, younger generation speak Efik." And Cook (in Williamson 1973: xxxvi): "The Kiòq language is rapidly dying out and being replaced by Efik. In most Kiòq areas the language is still spoken only by people of sixty years or older; the younger people speak Efik better than Kiòq, bilingual at least for several generations". (3.4) Hb 1 mentions that the younger people today speak Efik. (5) Nigeria, Cross River State, Calabar and Akampka Divisions. For details see Cook (in Williamson 1973: xxxvii). (7) Hansford - Bendor-Samuel - Stanford (1976); Williamson (ed.)(1973). Kir (1.1) Kir is given as the self-applied language name by Hb 2 (p. 19). (1.2) Other variants recorded in the literature: Hb 1 (p. 234) mentions KirBalar and Kirr (Hb 2 treats Balar as a dialect of Kir). Hb 2 lists among others - Kirr as possible glossonym. (2) Boghom, Eastern, Southern Bauchi, Western, Chadic (Afro-Asiatic) according to Hansford - Bendor-Samuel - Stanford (1976: 108).

A survey on language death in Africa

361

(3.3) The small number of speakers might indicate that the language is threatened by extinction. (4.2) Hb 1 (p. 234) mentions the figure of 360 speakers (referring to SIL sources from 1971). (5) Nigeria, Bauchi State, Bauchi Division, Zungur District. (7) Hansford - Bendor-Samuel - Stanford (1976).

Kole (1.2) Hb 2 (p. 128) lists Kole as a dialect of Duala but remarks that Dieu Renaud (eds.)(1983) suggest it is a distinct language. This view agrees with findings in Weimers (1971: 824) and in Hb 1 (p. 147). In the latter the language is called Bakolle. Other variants of the glossonym are Bakole, Bakole and Bakolle. Hb 4 (p. 7) and Ardener (1956: 9) mention Κόΐέ and Bakolle as language names. (2) Duala, Bantu (Niger-Congo). (3.3) The small number of speakers might suggest that the language is threatened by extinction. (4.2) The only figures available are those mentioned in Hb 1 (300 speakers) and in Weimers (1971: 824): 307 speakers. Cf. Ardener (1956: 15). (5) Cameroon, around Bamusso, South West Province. (7) Ardener (1956); Dieu - Renaud (eds.)(1983).

Kooki (1.2) Only two variants of the language name are found in the literature. One is Olukooki and the other one is Kooki (Hb 2, p. 142; Hb 4, p. 106; Weimers 1971: 826). (2) Bantu (Niger-Congo). Hb 4 suggests the following classification: InterLacustrine group. (3.2) Hb 4 suggests that Kooki is dying out. (4.2) Hb 4 only mentions "very few speakers" of the language. (5) According to Hb 4: Uganda; in a small enclave east of Mbarara.

Kora (1.1) Hb 2 (p. 167) mentions Cora as autoglossonym. (1.2) Other variants of the language name are found: !Kora, Koran, Koraqua or !ora. Hb 3 (p. 160) lists Koran(n)a, too. (Cf. also Weimers 1971: 867).

362

Gabriele Sommer

(1.5) Hb 3 suggests IKorana as ethnonym. (2) Hb 1 (p. 254) classifies the language as belonging to Central Khoisan (Khoisan). (3.1) Hb 2 indicates - without comments - that the language is extinct. Hb 1 mentions that the Kora are nomads and are slowly dying out. (4.2) Hb 1 mentions 50 speakers of the language according to the various sources. Hb 3 (p. 160) notes that there are "perhaps not more than 50 individuals" who speak the language. (4.3) Hb 1 (p. 304) mentions 10.000 members of this ethnic group. (5) South Africa (according to Hb 1 possibly also Botswana); Hb 3 suggests Griqualand West and the area around Kimberley as areas of distribution. Kore (1.2) No other language names are known; nowhere is the information about this language very detailed. Our source, which seems to be the only one, is an article by Heine - Voßen (1979). (2) Central Maa, Eastern Nilotic (Nilo-Saharan). (3.1) Kore is extinct; the last speaker, who served as a consultant to Heine and Voßen, died between 1980 and 1988 (Heine, p.c.). (3.4) According to a remark by Dimmendaal (this volume) Kore is being replaced by Somali today: " . . . the ethnic Kore all nowadays speak Somali as their first language and Swahili is used as a lingua franca" and "The Kore claim to have spoken Maasai before they actually took up Somali as their primary language." (4.2) Only two speakers (over 60 yrs.) in 1979. There are no speakers left today. (5) Kenya, Lamu island off the Kenyan coast and around Mokowe on the mainland facing this island. (7) Dimmendaal (this volume): Heine - Voßen (1979). Kpan (i.e. Bissaula and Eregba) (1.2) Under the cover name Kpan (cf. Hb 1, p. 235; also Yorda, Ibukwo, Kpwate, Hwaye, Hwaso, Nyatso, Kpanzon, Kpanten, Ikpan, Akpanzhi, Abakan, Nyonyo), two dialects (or languages) are mentioned that are probably threatened by extinction or have already died out. Hb 2 mentions Bissaula, a Western Kpan dialect, as being extinct. This statement is confirmed by the information given in Hb 1 (p. 235). Weimers (1971: 800) suggests what he calls Eregba - and lists as a separate language

A survey on language death in Africa

363

(cf. Hb 2's classification of Eregba as an Eastern Kpan dialect) - is "probably extinct". Hansford - Bendor-Samuel - Stanford (1976: 112) regard Bissaula as being extinct. Fivaz - Scott (1977: 280) mention that Eregba is possibly extinct. (2) We treat Bissaula and Eregba as dialects of Kpan. Kpan is classified as Central Jukunoid, Benue-Congo (Niger-Congo) by Hansford - BendorSamuel - Stanford (1976: 112). (3.1) As mentioned above, Bissaula and Eregba can be classified as being extinct. This statement, nevertheless, has to be verified. (5) Nigeria, Gongola State, Wukari Division, Kato Bagha, Wukari, Suntai, Gayan, Gindin Dutse, Likam Districts (areas mentioned by Hb 1 and Hansford - Bendor-Samuel - Stanford (1976) to indicate the area of distribution for the whole Kpan group). (7) Hansford - Bendor-Samuel - Stanford (1976). Kpati (1.2) No other glossonym besides Kpati is found in the literature. Cf. Hb 1 (p. 235), Weimers (1971: 827). (2) According to Hb 1 and Weimers (1971), Kpati has to be classified as belonging to the Benue-Congo branch (Niger-Congo). Hb 1 subclassifies it as Grassland Bantu. Cf. Fivaz - Scott (1977: 168): Ngemba subgroup, Grasslands group, Bane, Bantoid, Benue. (3.1) Hb 1 suggests that Kpati is extinct. (5) Nigeria, Gongola State, Wukari, Takum Divisions according to Hb 1. Fivaz - Scott (1977) suggest Cameroon as area of distribution. Kreish language group (1.2) "Kreish" languages (cf. Doornbos - Bender 1983). Hb 3 (p. 13) uses Kresh, Kreish or Kpala, Kpara as language names. (1.4) Hb 3 (p. 130) comments: "This name [Kresh] is regarded by all sections of the Kresh as a foreign name; most of them acknowledge the name Kpala." (1.5) Kresh, Kredj, Kreich, Kreish (Hb 3). (2) Fivaz - Scott (1977: 169) give the following classification: Kreish group, Central Sudanic, Chari Nile (Nilo-Saharan). For details see Doornbos - Bender (1983). (3.3) Doornbos - Bender (1983: 45f.) report: " . . . under the aegis of enculturation to Sudanization and, historically as a result of slavery, a number

364

Gabriele Sommer

of languages are under a threat of extinction: this applies most of all to the so-called "Kreish" languages belonging to the Sara-Bongo-Bagirmi group, which are spoken by communities settled in the extreme South of Darfur Province and the Northern part of Bahr el-Ghazal Province. Today these people are settled predominantly in trade centers in the north and have discarded their ethnic and linguistic stigmata as far as they can, engage in prayer and trade with their neighbors on an equal basis, and teach their children Arabic." (3.4) Arabic? (5) Sudan. Cf. 3.3 and map in Doornbos - Bender (1983: 42). (6) "Sudanization". See entries Berti or Beygo. (7) Doornbos - Bender (1983). Kudu (1.2) The only name available is Kudu. Concerning Kudu we refer to Shimizu (1982a: 132f.). Hb 1 (p. 235) and Hansford - Bendor-Samuel - Stanford (1976: 113) mention "Kuda-Chamo", also called Kudawa (see entry Chamo). (2) Northern Jos, Plateau, Benue-Congo (Niger-Congo). (3.2) According to Shimizu (1982a: 132): "This language seems to be almost entirely extinct." (3.4) Shimizu does not explicitly mention any replacing language(s). It is probably Hausa - if one compares the situation with other members of the Northern Jos group mentioned in the same article. (4.2) Shimizu (1982a) remarks that he met only a few speakers who were able to speak or at least remembered Kudu. "At Jììmi in the Kudu country an old man gave me a short wordlist of this language . . . " and "At Ningi, however, I was able to meet another old man (5) Nigeria, a few speakers were found living in Ningi. (7) Shimizu (1982a). Kupto (1.1) According to Leger (p.c.) Kùttò seems to be the self-applied language name (and ethnonym cf., 1.5). (1.2) The most frequently used name is Kupto, occasionally also Kubto. (1.3 and 1.5) Kùttò. (2) Hb 1 (p. 236) classifies the language as belonging to the Southern group of Bole-Tangale, Plateau-Sahel, Chadic (Afro-Asiatic). (3.3) Leger suggests that the language is threatened by extinction (cf. 6).

A survey on language death in Africa

365

(3.4) Kupto is being replaced by Hausa and Kanuri. About the latter, Leger is not quite sure but merely adds that recent migrations of Kupto-speaking groups have exposed them to the influence of Kanuri-speakers. (4.2) Leger estimates between 5,000 and 6,000 speakers. (4.3) He suggests that there are 6,000 people belonging to the same ethnic group. (5) Nigeria, Eastern Bauchi/Western Bornu State (North-East). (6) Leger mentions "recent migrations [of Kupto-speakers to the area of Hausa and Kanuri-speaking groups] due to the flooding after the opening of the Dadin-Kowa dam on the Gongola river" as one reason for language replacement and remarks that since then the Kupto have been divided into two separate groups who live approximately 10 to 15 km from each other. (7) Hansford - Bendor-Samuel - Stanford (1976); Hoffmann (1971); Leger (1983-85, 1990, 1991); Newman (1977).

Kwadi (1.2) Several variants of the language name are found in the literature (Hb 2, p. 169; Hb 1, p. 139). These are - besides Kwadi - Cuepe, Cuanhoca, Mucoroca, Koroca, Bakoroka and Makoroko. (2) Khoisan. (3.1) Voßen (p.c.) remarks that Kwadi is probably extinct. Cf. also Winter (1981: 347): "Das Kwadi in Angola dürfte inzwischen ausgestorben sein." (3.4) Hb 1 (p. 139) mentions 15,000 Kwadi which probably refers to the members of the ethnic group. (5) Angola. (7) Winter (1981).

Kwadza (1.1) Hb 2 (p. 28) uses the variant Rwadza as glossonym. (1.2) Hb 2 also mentions Qwadza as language name. (1.3 and 1.5) Kw'adza is given as autonym and ethnonym in Hb 2 and by Ehret (p.c.). (1.4) Ehret mentions Ng'omvia as xenonym. (2) Southern Cushitic, Cushitic (Afro-Asiatic).

366

Gabriele Sommer

(3.2) The comments in Hb 2 ( " . . . reported as almost extinct . . . ") and by Ehret ("possibly a few very old speakers left") imply that the language is in a process of extinction. (3.4) Ehret mentions CiGogo as the replacing language. (4.1) Ehret suggests that in the late 1970s there were only 4 men and 1 woman who knew Kwadza. (4.2) Ehret further mentions a few middle-aged people who may have some knowledge of Kwadza: "I was told about 1970 that Fweda's daughter, then c. in the 40's, had some knowledge [of Kwadza]." (5) Tanzania, Dodoma District, at Bankolo (cf. Ehret: "all actual speakers are probably now dead") and near Isolo (cf. Ehret: "a couple of speakers may still be alive"). Elderkin (p.c.) reports that there are some Kwadza speakers living in the area of Itiso and Solwu. Kwegu (1.1) Bender (1976: 10) states that toko kwe:goi is the self-applied name. (1.2) Several variants of the language name are found in the literature: Hb 2 (p. 109) lists jiidi and Yidi (which are, according to Bender (1976) xenonyms used by Mursi and Bodi speakers, respectively), Hb 1 (p. 182) further mentions Yidinich, Yidinit (singular form of Yidi, according to Bender) and Muguji. The latter is also mentioned in Dimmendaal (1989). (1.3 and 1.5) Hb 2 suggests that Kwegu is an autonym and ethnonym. Bender mentions Muguji as an ethnonym. (1.4) Nidi, Yidi (see 1.2). Bender adds Kwegi and mugu.ji 'apo 'Muguji mouth' as foreign names applied by Maji and Hamer, respectively. (2) Surma group, Eastern Sudanic, Chari-Nile (Nilo-Saharan) according to Bender (1976). (3.2) Nearly extinct. Cf. Dimmendaal's comment (1989: 18): " . . . the smallest groups of Kwegu, of sometimes ten or twenty or even single individuals, have often forgotten their own language and speak only Bodi. The Kwegu language is nearly extinct as a result of the shifting language solidarity." (3.4) Bodi and/or Mursi. Cf. Dimmendaal (1989: 17): "When the Kwegu are among themselves, they speak their own language, but they are also fluent in either Mursi or Bodi, depending on whom they are in closer contact with." (4.2) One figure can be found in Hb 1 (p. 182) where 300 speakers of Kwegu are estimated. This figure dates back to 1982. Dimmendaal (1989: 17)

A survey on language death in Africa

(5)

(6)

(7)

367

remarks that there are circa 500 Kwegu today who are dominated by the far more numerous neighbouring Mursi and Bodi (approx. 6,000 each). Bender (1976: 10) estimated approx. 250 Kwegu, the Muguji according to him living in a separate village "south of main Kwegu concentration" - consisting of circa 100 people. Ethiopia, "scattered in villages along the banks of the Omo River in the extreme southwestern corner of Ethiopia" according to Dimmendaal (1989). Domination of neighbouring Bodi and Mursi, low social status of Kwegu, loss of ethnic group identity, division of language use, "typical for hunter-gatherer societies interacting with pastoral and agricultural people", intermarriage with neighbouring peoples. Cf. Dimmendaal (1989: 17) who, concerning the latter point, remarks: "The Kwegu are diminishing in number because of intermarriage, and they are beginning to lose their identity as a separate ethnic group, a fact of which they are very much aware. The domains in which either Kwegu or Mursi and Bodi are spoken are beginning to overlap, a situation abetted by intermarriage. The contexts in which Kwegu is used among adults are decreasing, and the acquisition of Kwegu as a first language becomes less frequent because it strongly depends on the social network of which the parents are part." Bender (ed.)(1976); Dimmendaal (1989).

Kwisi (1.2) Kwisi, Mbundyu, Kwandu. (2) Bantu (Niger-Congo)? (3.2 or 3.1) Köhler's short remark on Kwisi (see 4.2) implies that the language is in a process of extinction or is already extinct. (4.2) Köhler quotes Westphal (1965) who proposed that Kwisi is only spoken by a few elderly people. (5) Angola, southwestern part, west of the Kuvale area. (7) Köhler (1975); Westphal (1965). Ligbi General remark: We will neglect the information given in the handbooks (Hb 2, p. 49; Hb 1, p. 200 and Weimers 1971: 834) in this case in favour of the data provided by Kastenholz (p.c.). These data are based on the literature mentioned under 7 and the results of his field research in 1989.

368

Gabriele Sommer

(1.1) Kastenholz suggests several self-applied language names varying according to the various areas of distribution: Wela or Jogo (in some villages or towns in the Bondoukou area), Numu or Ligbi (in Tehini and one village in Bouna and Kinéta). Jelü (in four villages in the Mbéngué area), Atumfuo, Wela, Ligbi and Jogo (in Ghana). Kastenholz further suggests that Ble might be another variety of the so-called Ligbi cluster (see entry Jelkuna in our survey) distributed in Burkina Faso. (1.4) Soyo-Vanye, i.e. 'Red Mande' (by Kulango, in Bondoukou and Bouna areas), Celle or Jeri (by Senufo and Dyula, in Mbéngué), Ndwera (by Abrong for Wela and Muslim Ligbi, in Ghana), Ntoore and Atumfuo (by Abrong for Numu and Ligbi (Jogo) in Ghana). (1.5) He also specifies the several ethnonyms according to their area of distribution: Wela and Numu (in Bondoukou), Numu and Ligbi (in Tehini and Bouna/Kinéta), Jelü (in Mbéngué), Numu, Wela and Ligbi (in Ghana). (2) Mande. According to Fivaz - Scott (1977: 176): Hwela-Ligbi, Northern, North-Western, Mande. (3) General remark: Kastenholz suggests that Ligbi has to be specified according to its status in the several areas of distribution. (3.1) The language is extinct in the area of Nassian, Ivory Coast where the last speakers died in 1979. (3.2) In the area of Bondoukou the language seems to be in a process of extinction due to Kulango dominance. Wela and Numu (the latter being blacksmiths) seem to have a low social status and adapt to the more prestigious language Kulango. (3.3) Ligbi is threatened by extinction, at least in the other areas in the Ivory Coast. This seems to be less certain for the varieties spoken in Ghana. (3.4) Kastenholz suggests Kulango and Bondoukou-Dyula (a Manding dialect) as replacing languages. (4.1) Kastenholz estimates 600 to 700 speakers at most in the Ivory Coast and circa 5,000 in Ghana (cf. Persson - Persson 1980). (4.3) These are considerably more, no demographic data being published. Kastenholz further comments: "The Numu, Ligbi and Wela who shifted to the Kulango language regard themselves as non-Kulango; those who shifted to Dyula (the lesser case, always connected with Islam) tend to regard themselves as Dyula but know about their origin." (5) Ivory Coast (Bondoukou, Bouna and Mbéngué areas) and Ghana (Banda/Wenchi area). (6) Kastenholz further explains that the language shift of Numu and Wela to Kulango was motivated by the fact that Kulango has become what

A survey on language death in Africa

(7)

369

he calls an "egalitarian" language after the Abrong rulers had shifted to Kulango. Kropp-Dakubu (1976, 1988); Persson - Persson (1980).

Lorkoti (1.2) Lorkoti and Lorokoti (concerning the latter term see Spencer 1973). (2) Maasai, dialect of Maa language (South Maa), Eastern Nilotic (NiloSaharan). (3.1) Extinct (according to Heine, p.c.). (3.4) Lorkoti was replaced by Samburu (North Maa, Eastern Nilotic, NiloSaharan). (4.2) Spencer (1973: 204) mentions the number of 210 tax-payers in 1958 for his "Laikipiak Ndorobo" consisting of seven hunter-gatherer groups including the Lorokoti. Heine (1981: 32) found one group of "Dorobo" (called l-orkóti) in the 1980s who still spoke a Maasai dialect (cf. his comment: "The speech of l-orkóti elders, . . . , differs markedly from Sampur.") His Lorkoti informant was 60 years old then. (5) Kenya, northern part. According to Spencer (1973: 201), the Lorkoti claimed their territory were the Ngelesha and Amaya areas. Spencer (1973: 204 and 1973: 151) reports that the ethnic Lorokoti live on the Leroghi Plateau today. (6) Transition of economic activity from hunting-gathering to pastoralism motivated the language shift. (7) Heine (1981); Spencer (1973).

Lufu (1.2) The only language name used in Hb 1 (p. 237), which is the only source reference for the language, is Lufu. (2) Hb 1 does not mention any specific classification, i.e. the authors regard the language as being unclassified. (3.1) Hb 1 notes that the language is extinct. (5) Nigeria, Gongola State, Wukari Division, Lufu Division.

"Lugha ya zamani" (1.2) Lugha ya zamani 'the old language'. (2) Dialect of Swahili, Bantu (Niger-Congo). Cf. Nurse (1984-1985: 228f.): "In all essential respects, this [lugha ya zamani] is Mtang'ata [a Swahili

370

Gabriele Sommer

dialect, spoken along a 20 kilometer stretch of coast just south of Tanga, from Kivindani in the north to Kigombe in the south]". (3.1) Cf. Nurse (1984-1985: 228): " . . . some old people have a tenuous grasp of what they call the lugha ya zamani ('the old language'):" (5) Tanzania, according to Nurse (1984-1985: 228f.) the people who "have a tenuous grasp" of this language or dialect, live north of Mtang'ata, from Kivindani to some miles north of Tanga, and south of Mtang'ata, from Kigombe to Saadani, today. Nurse assumes that in the nineteenth century and before Mtang'ata the lugha ya zamani was probably spoken from just north of Tanga Town to just south of Dar es Salaam. (7) Nurse (1984-1985).

Luri (1.2) The only language name that appears in the literature (Hb 2, p. 19 and Hb 1, p. 237) is Luri. (2) Chadic (Afro-Asiatic). Hansford - Bendor-Samuel - Stanford (1976: 123) leave the language unclassified. (3.3 or 3.2) The rather small number of speakers indicates that the language is at least threatened by extinction. (4.2) The only figure available is the estimation of circa 30 speakers given in Hb 1. (5) Nigeria, Bauchi State and Division, Zungur District. (7) Hansford - Bendor-Samuel - Stanford (1976).

Mampoko (1.2) The other language names found in the literature are Makumu and Mobele. Cf. Hb 2 (p. 137). Some sources suggest that Mampoko is not a distinct language but a dialect of Ngiri (cf. Hb 1, p. 295 and Weimers 1971: 840). (2) Bantu (Niger-Congo). Hb 4 (p. 37) classifies Mampoko as a member of the Ngala group; Fivaz - Scott (1977: 51) regard Mampoko as a member of Loi-Ngiri, Bangi-Ntumba, Bantu. (3.1) The only information available concerning the status of the language is provided by Hb 2 which suggests that the language may be or is already extinct. The authors do not comment on the particulars of extinction. (4.2) Hb 1 - indicating the number of speakers of the whole Ngiri cluster consisting of seven dialects at least - estimated 6,000 speakers in 1977.

A survey on language death in Africa

(5)

371

Zaire. Hb 1 (p. 295) referring to the general area of distribution of Ngiri mentions Equateur Region, southern part of Bomongo Zone, between Oubangi River and Zaire River.

Mangas (1.2) The only glossonym available is Mangas (cf. Hb 2, p. 19; Hb 1, p. 238). (2) Chadic (Afro-Asiatic). Hb 1 subclassifies the language as follows: Boghom, Southern Bauchi, Plateau-Sahel, Chadic. Cf. Hansford - BendorSamuel - Stanford (1976: 124): Boghom, Eastern, Southern Bauchi, Western, Chadic. (3.3) We tentatively class the language as being threatened by extinction. (4.2) Hb 1 (p. 238) suggests 180 speakers. (5) Nigeria, Bauchi State and Division, Zungur District. (7) Hansford - Bendor-Samuel - Stanford (1976).

Mboa (1.1) Hb 2 (p. 115) uses the glossonym Mboqa. (1.2) The other variant of the language name if Mboa. The following data are provided by Gerhardt (p.c.) who refers to Strümpell (1910). (2) Cf. Fivaz - Scott (1977: 190): Jarawa subgroup, Nigerian group, Bane, Bantoid, Benue (Niger-Congo). The authors (Fivaz - Scott 1977: 281) remark that Mboa might also be classified as Adamawa-Eastern. (3.1 or 3.2) The only information available is mentioned by Struck-GrossLichterfelde in Strümpell (1910: 450) who recorded some remnant speakers in 1908 living in Baya-speaking territory. ("Mboa ist ein ziemlich isolierter Sprachrest inmitten des Bayagebietes westlich von Kunde auf dem rechten Lom-Ufer gelegen . . . ".) According to Gerhardt, who refers to Voeltz (p.c.)., this language is extinct today. (5) Cameroon, west of Kunde. (7) Strümpell (1910).

Mbondo (1.2) We have only very little information about this language. The only available data are mentioned in Hb 2 (p. 137) where the authors do not give any comments. No other language name is known. (2) Bantu, Benue-Congo (Niger-Congo) according to Hb 2. (3.1) Hb 2 indicates that the language is already extinct. (5) Zaire.

372

Gabriele Sommer

Mboq (1.1) Hb 2 (p. 129) mentions Lumboq as self-applied language name. (1.2) Other variants of the language name are also found. These are: (ta)Ka(a)loq , (a)-Mboq or Dimboq. (2) Bantu (Niger-Congo). Hb 7 (p. 33) suggests Bafia as subclassification. (3.3 or 3.2) Although the estimations given in Hb 7 and Hb 4 (p. 13) might indicate that the language is already in a process of extinction, the figure mentioned in Weimers (1971: 816) contradicts this assumption. (4.2 and/or 4.3) Whereas Hb 7 suggests that Lumboq is spoken "by a handful of people" and Hb 4 (p. 13) estimates circa 50 speakers of the language, Weimers (1971: 816) gives the estimated number of 50,000 speakers of what he calls b-Kaalong (which might be the total number of people belonging to the ethnic group). (5) Cameroon, according to Hb 7 and 4: on the south bank of the River Mbam just below its confluence with the River Noun. Meroitic (1.2) No other language names besides Meroitic are found in the literature. (1.6) The name Meroitic is derived from the name for the capital of the Meroitic Empire, i.e. Meroe. (2) The classification of this extinct language remains to this day unclear. It is usually classified as belonging to the Eastern subgroup of Chari-Nile (Nilo-Saharan) or as a member of the Afro-Asiatic language family. (3.1) All sources mention this language as being extinct (cf. Hb 2, p. 5; Weimers 1971: 848). The language existed, i.e. was in use, approx. between the 8th century B.C. and the 4th century A.D. Written sources of Meroitic are recorded from between the 2nd century B.C. and the 4th century A.D. (3.4) There exist no written records of Meroitic after the 4th century A.D. It was probably used as a spoken language after the 4th century A.D. but nothing is known about the actual language use at that time or about a replacing language in the nuclear area (Kerngebiet) of Meroitic. In Nubia Meroitic was used as an official language and as a ritual language (Kultsprache) and was replaced by Nubian. People in Nubia began to use Nubian script when Nubia became Christian (Bechhaus-Gerst, p.c.).

A survey on language death in Africa

(5)

(7)

373

Sudan; the nuclear area (also called "Meroe island") was the area between Atbara and the Nile. In Nubia Meroitic was used as an official language only. Hofmann (1981).

Mongoba (1.2) Hb 3 (p. 40) suggests that Mongoba is the language name. (1.3 and 1.5) Hb 2 (p. 101) and Hb 3 mention Moqgoba or Muqgußä as self-applied names/ethnonyms. (2) According to Hb 3 (p. 40): Banda-Gbaya-Ngbandi languages. (3.3) The information available is very scanty. We classify the language as being threatened by extinction on account of its small number of speakers. (4.2) In Hb 3 (p. 40) a number of circa 50 speakers is given. (5) Zaire, according to Hb 3: Territoire Watsa, near Aribi (Arebi).

Muskum General remark: We do not refer to the data mentioned in the handbooks but concentrate on the information provided by Tourneux (1977). (1.2) Mùskûm. (1.5) Muskum or Mouskoun. (1.6) Mouskoun. (2) Chadic (Afro-Asiatic). Tourneux (1980: 238,243) suggests that Muskum should be grouped together with Musgu-Vulùm and Mbárá in a "groupe des langues tchadiques baptisé Musgu". (3.1) Cf. Tourneux (1977: 13f.): "C'ést à Muskum qu'en companie de géographe Christian Seignobos nous avons rencontré pour la première fois en 1976 un vieillard du nom de Lèpèkê qui s'est présenté en disant qu'il connaissait une langue ignorée de tous les autres villageois, le mùskûm." When Lépékê had come back to Muskum in 1950: "II constate à son retour que plus personne ne parle mùskûm à Muskum." (See 6). (3.4) Cf. Tourneux's comment (1977: 15): "Petit à petit le mùskûm s'est trouvé évincé au profit du vlùm, ... ". (4.2) One old man in 1976. Cf. also this man's comment about the number of speakers of Muskum (Tourneux 1977:15: "il y aurait encore, . . . un autre ex-locuteur de mùskûm, . . . qui résiderait à Holom, au nord de Muskum."

374 (5)

(6)

(7)

Gabriele Sommer Chad. Muskum was spoken in the city of Muskum (Mouskoun), "situé à dix kilomètres au nord de Katoa, sur la rive tchadienne du Logone". See map in Tourneux (1980: 230). Cf. Tourneux's comment (1977: 14f.): "En fait, la disparition de cette langue était prévisible de longue date. Adolf Friedrich zu Mecklenburg passant par là en 1911 signale qu'il a trouvé à Muskum une langue qui n'est plus parlée que par les vieux. Il s'agit très certainement de la même langue que celle dont nous avons pu recueillir un dernier écho. . . . en 1937, nous pensions que cette langue était totalement inconnue." The informant Lèpèkê described the situation at the time when Muskum still existed as follows: " . . . it y a eu une époque où l'on ne parlait que le mùskûm à Muskum, et il n'y avait pas d'intercompréhension entre le mùskûm et les langues voisines." And Tourneux further comments: " . . . D'après Lèpèkê, c'est à la génération de son père que le mùskûm a commencé à régresser. A cette époque, les hommes se sont mis à prendre femme à Gouaye, village où l'on parle vlùm (mulwi)." Tourneux (1977, 1980).

Nagumi (1.2) Nagumi, Bama and Mbama (cf. Hb 2, p. 115 and Weimers 1971: 853). (2) Jarawa subgroup, Nigerian group, Bane, Bantoid, Benue (Niger-Congo) according to Fivaz - Scott (1977: 199). (3.1) According to Gerhardt (p.c.) the language has already become extinct. Gerhardt refers to Strümpell (1922-1923). (4.2) Strümpell reports one old man at the beginning of this century. (Cf. Strümpell (1922-1923: 69): "Auch die Sprache der Nagumi ist verklungen. Ein blinder Greis wurde mir schließlich in Alhadschin Galibu zugeführt, der nur die wenigen Worte anzugeben vermochte, die in das Wörterverzeichnis aufgenommen sind.") (5) Cameroon, at the Benue river, north of Garoua. (7) Strümpell (1922-1923).

Napore (1.2) Napore. Also Niporen, Nipori, Nyipori and Poren (cf. Hb 3, p. I l l and Bryan 1945: 194). (2) Dialect or closely related language of Karimojong, Eastern Sudanic (Nilo-Saharan). (3.1) Extinct according to Heine (p.c.).

A survey on language death in Africa

375

(2.4) Karimojong (Dodos dialect). (5) Uganda, northeastern part, Dodos country, Napore Hills. (7) Bryan (1945). N d a m of Dik (1.1) (1.2) (1.5) (2) (3.3) (3.4)

Bag digään. Ndam (of Dik). masc. sg. digë, fem. sg. digä, pl. comm. digo. Somrai group, East Chadic (Afro-Asiatic). According to Broß (p.c.) the language is threatened by extinction. The neighbouring groups are Tumak, Sibine, Miltu, Boa, Niellim (both dialects). (4.2) Broß estimates less than 1,000 people. (5) Chad. (7) Broß (1988). Ngbee (1.2) Hb 4 (p. 87) regards li-Ngbee as the language name. Besides liNgbee, Weimers (1971: 857) suggests Lingbe and Mangbele as variants of the glossonym. Considering the fact that Weimers estimates a rather high number of speakers for his liNgbee we assume that he uses this glossonym as a cover term. (1.5) Hb 4 (p. 87) suggests Ngbee (Mangbele) as ethnonyms. (2) Bantu, (Niger-Congo). Hb 4 (p. 87) regards li-Ngbee as a member of the Kari group, Bantu. (3.2) Leaving aside the information in Weimers (1971) who assumes that there are thousands of speakers of liNgbee and mainly emphasizing the information given in Hb 4, we classify the language as already being in a process of extinction. Cf. Hb 4 (p. 87): "This language was still in existence at Gombari a few years ago, but now Mangbetu or Mayogo is spoken." Hb 4 mentions two areas of distribution. In one of these areas (Terr. Niangara) the language was only spoken by a few elders. The second area is at Gombari (cf. their comment cited above) with a really small number of speakers (see 4). (3.4) Ngbee (formerly?) spoken at Gombari is being replaced (see 3.2) by either Mangbetu or Mayogo. (4.2 and/or 4.3) Contradicting figures can be found in the sources available (see also 1.2). Weimers mentions 30,000 speakers of liNgbee or Mang-

376

(5)

Gabriele Sommer

bele. Hb 4 (p. 87) reports the following figures: only a few elders in "Terr. Niangara" and circa 30 speakers at Gombari. Zaire; "Terr. Niangara" and at Gombari according to Hb 4.

Ngbinda (1.2) The variants of the language name found in the literature are: (Bu) Ngbinda, Bangbinda (Hb 1, p. 263; Weimers 1971: 857). (1.5) Hb 4 (p. 87) mentions 'Bangbinda as ethnonym. (2) Kari group, Bantu (according to Hb 1 and Hb 4) (Niger-Congo). (3.3) The data available suggest that the language is threatened by extinction. (4.2) Hb 1 and Hb 4 mention "few" speakers of Ngbinda. (5) Sudan and Zaire; according to Hb 4: "in a few scattered spots in the Uele region, and at Kotele village 20 km. from Buta".

Nimbari (1.1) Hb 2 (p. 98) seems to use Nimbari as self-applied language name. At least there are no other glossonyms mentioned. (1.2) Hb 1 (p. 171) further suggests the variants Nyamnyam, Niamniam and Bari. (1.4) See 1.5. (1.5) Cf. Strümpell (1922-1923: 68f.): Bari or Nimbari " . . . soll der einheimische Name dieses von den Fulbe Niam-Niam genannten Völkchens gewesen sein, . . . ". (2) According to Fivaz - Scott (1977: 208): Adamawa, Adamawa-Eastern (Niger-Congo). (3.2) One source available dealing with the status of the language is Hb 1 (p. 171) where it says: "Does exist in Nigeria", whereas the language seems to be extinct in Chad and Cameroon: "May be none in Chad. Now extinct in Cameroon". Cf. Strümpell (1922-1923: 69): "Wie die Niam-Niam untergegangen sind [in the Mandara area], so ist auch ihre Sprache verklungen. Nur wenige Männer findet man, die sich ihrer noch erinnern." (4.2) Hb 1 (p. 171) gives an estimation of 120 speakers in Nigeria. (5) Nigeria, probably no longer in Chad and Cameroon. (7) Strümpell (1922-1923).

A survey on language death in Africa

377

Ningi (1.2) Ningi. The language is subsumed under the topic Buta-Ningi in Hansford - Bendor-Samuel - Stanford (1976). (2) Northern Jos, Plateau, Benue-Congo (Niger-Congo). (3.1) Hansford - Bendor-Samuel - Stanford (1976: 57) treat the language as being in a process of extinction. Shimizu (1982b: 167) did not find any Ningi speaker in the 1970s. We therefore regard the language as being extinct (see entry Chamo). (5) Nigeria, Bauchi State, Bauchi Division, Ningi District according to Hansford - Bendor-Samuel - Stanford (1976). See also map in Shimizu (1982a: 98). (7) Hansford - Bendor-Samuel - Stanford (1976); Shimizu (1982a, 1982b).

Nyang'i (1.2) Nyang'i, Nyangiya, Nyangeya (see 1.6). (1.3) Nyaqi, sg. nyaqiát. (2) See entry So. Heine classifies Nyang'i as a member of the Kuliak languages, i.e. as an isolated unit. (3.2 or 3.1) In the 1970's Heine (1976: 2) commented: "The Nyang'i language is rapidly dying out; Nyang'i is nearly extinct now." (3.4) Younger people speak Dodos nowadays (Heine 1976: 2). (4.2) As early as in the 1970's, Nyang'i was only spoken by less than 100 people, all of them over 40 years then. In March, 1983, Heine (p.c.) visited the Nyang'i area again and found the language to be spoken by only 8 people, all over the age of 50. (5) Uganda, southern ranges of the Nyangeya Mountains along the boundary of Karamoja and Acholi Districts. (7) Heine (1974-1975, 1976).

Old M f e n g u (1.2) The information available in this case is very scanty. A language called "Old Mfengu" is only mentioned in Hb 4 (p. 154) and does not appear elsewhere in the literature. Besides Old Mfengu we also found the glossonym Fingo (given in Hb 4, p. 153) which might be a synonym for Old Mfengu. (2) Hb 4 classifies the language as a member of the "Nguni group". Nguni is a cover term used for a group of languages consisting of Zulu, Xhosa,

378

Gabriele Sommer

and Swazi and their dialects. Cf. Weimers (1971: 859) where he adds: " . . . although these [Nguni languages] are largely mutually intelligible, they have generally been considered three 'languages' and Nguni has rarely been called 'a language' . . . " . Fivaz - Scott (1977: 67) classify Old Mfengu as follows: Swati, Nguni group, Bantu. (3.1) The only information available is Hb 4 ' s statement on the status of their Old Mfengu: " O l d Mfengu', now extinct, but known from older writers, belonged to the Nguni group." The authors do not, however, elucidate who these "older writers" are or what they had to say about Old Mfengu. (4.2) See 3.1. (5) Hb 4 does not give any information about the area of distribution. We therefore can only refer to the general area of distribution for the socalled Nguni group. Most of the languages of the Nguni group (which consists of even more languages according to the classification given in Hb 2 (p. 165) than that given by Weimers) are spoken in South Africa, Lesotho, Swaziland and Mozambique. Omo-Murle (1.2) Omo-Murle, "Acient Mûrie". Cf. also Hb 3 (p. 88). (2) Dialect (?) of Murle, Didinga-Murle group, Eastern Sudanic (NiloSaharan). (3.2 or 3.1) According to Dimmendaal (1989: 26) Omo-Murle is almost extinct. ("This offshoot of the Murle seems to be practically extinct, at least linguistically.") (3.4) The Omo-Murle speak Nyangatom as their first language today; i.e. they have adapted linguistically to the neighbouring Nyangatom pastoralists. (4.2) According to Dimmendaal (1989: 26) who refers to Tornay (1981) there are only a few women left who remember Omo-Murle today. (4.3) Although there are no figures available concerning the number of ethnic Omo-Murle we can deduce from Dimmendaal's comments that the ethnic group of Omo-Murle does exist today: "The descendants of originally Murle-speaking people still exist as an ethnic group . . . ". (5) Ethiopia, southwestern part, in the vicinity of the Nyangatom who live along the Kibish River. Dimmendaal remarks that the "ethnic OmoMurle" live in the border area between Sudan and Ethiopia, i.e. on the Boma Plateau. (6) For details concerning the process of cultural and linguistic adaptation of the Omo-Murle to the pastoral Nyangatom see Tornay (1981). (7) Dimmendaal (1989); Tornay (1981).

A survey on language death in Africa

379

Omotik (1.2) The most frequently used variant of the language name seems to be Omotik. Rottland (1982) uses Omotik as language name and ethnonym. (1.3 and 1.5) According to Heine - Möhlig (1980) o-laamoot-i, pi. laamoot is an ethnonym; Omotik (see 1.2). (2) Southern Nilotic (Nilo-Saharan). (3.2) According to Heine - Möhlig (1980: 76) Omotik is becoming extinct: "Their language [Omotik] is rapidly dying out and will probably be extinct within one or two decades." Rottland (1982) agrees with this assumption. (3.4) Omotik is replaced by Maasai. (4.2) Cf. Heine - Möhlig (1980: 76): "The total number of Omotik speakers is probably less than fifty, all above the age of forty years and most of them women . . . " . Rottland (1982) suggests that Omotik speakers are even less numerous, i.e. approx. ten women and one man, all of them over 50. (5) Kenya, Narok District. According to Heine - Möhlig (1980: 76) the last speakers of Omotik can be found around Entasekera, Lemek and Lolgorien on the borderland of Maasai country. Heine - Möhlig further mention that the original homeland of the Omotik was on Mount Ngulot to the south of River Amala (Mara) on the southern fringes of the Mau forest. (6) Heine - Möhlig (1980) mention three main reasons for the language shift of Omotik speakers to Maasai: 1. change from hunter-gatherer economy to pastoralism; 2. Omotik mixed culturally and economically with Maasai and 3. Maasai gradually became the first language for younger Omotik. Cf. also Rowland's comment (1982: 26): "Die soziale Lage der Omotik gleicht der der Okiek von Kinare [see separate entry]; sie existieren nicht mehr als Gruppe, haben für alle Lebensbereiche eine andere Identität angenommen und geben von ihrer ursprünglichen Sprache und Identität nur noch auf Verlangen Kunde. Sie sind oder waren mit Maasai verheiratet, leben in Maasai-Siedlungen und sind praktisch Maasai. Von ihrer Kenntnis der Omotik-Sprache wissen oft nicht einmal mehr die Enkel." (7) Heine - Möhlig (1980); Rottland (1982).

380

Gabriele Sommer

Ongamo (1.1) Heine - Voßen (1975-1976) mention òqg'àmóni, pl. òqg'àmò as selfapplied name. (1.3) See 1.1 and 1.5. (1.5) See 1.1 and 1.3. (2) The language is classified as a member of the Eastern Nilotic, Eastern Sudanic (Nilo-Saharan). Concerning the discussion about genetic classification see Heine - Voßen (1975-1976: 8If.). (3.2) Cf. Heine - Voßen (1975-1976: 81): "Das Ongamo gehört zu den im Aussterben begriffenen Restsprachen Ostafrikas." Möhlig (p.c.) also mentions that Ongamo is only spoken or better known by some elderly people today. (3.4) The younger Ongamo people speak Chagga today (Heine - Voßen 19751976). (4.2) Heine - Voßen (1975-1976) estimate approx. 10 settlements where circa 700 Ongamo live. They regard the estimation of 1,000 speakers (as suggested by Hb 3, p. 112 and also Hb 1, p. 270) as being too high. (5) Tanzania, northeastern slopes of Kilimanjaro in the three localities Ubetu, Reha and Kahe between the valleys Makuku and Machima near the Kenyan/Tanzanian border. (6) Cf. the comments in Heine - Voßen (1975-1976: 83) that suggest a language shift of Ongamo speakers to Chagga: "Die Ongamo-Sprache wird nur noch von älteren Menschen gesprochen; die jüngere Bevölkerung wächst mit Dschagga als Muttersprache auf." (7) Heine - Voßen (1975-1976).

Poko (1.2) Bapoko (Hb 2, p. 128). Hb 2 and Hb 7 (p. 24) give several other variants of the language name such as Naka, Puku, Poko, and Bapuku. (2) Bantu (Niger-Congo); Hb 7 classifies the language as belonging to the Bube-Benga subgroup. Weimers (1971: 884) suggests that Bapoko is a dialect of baTanga. Cf. also Fivaz - Scott (1977: 237) who classify Poko-baTanga as members of the Bube-Benga group, Bantu. (3.3) We classify the language as being threatened by extinction in consideration of the small number of speakers. (4.2) The only information available in Hb 7 is the comment that Poko is spoken "by a very small group". Weimers (1971), however, estimates 2,000 speakers of baTanga. Ardener (1956: 15) reports 713 Tanga-Poko.

A survey on language death in Africa

(5) (7)

381

Adams (1907) only mentions that the Bapuku are not very numerous and that there are only a few villages inhabited by Bapuku. Cameroon, "along the coast of Cameroun, near Grand Batanga" according to Hb 7. Adams (1907); Ardener (1956).

Putai (1.2) The sources available mention Putai and Marghi West or West Margi as glossonyms. (2) Chadic (Afro-Asiatic). Hansford - Bendor-Samuel - Stanford (1976: 125) include Putai under the term "Margi language cluster" which is classified as Western, Bura group, Central, Chadic. (3.2) According to Hb 1 (p. 242) the language is dying out, i.e. "nearly extinct". (3.4) Hb 1 suggests Kanuri as the replacing language. (4.2 and 4.3) Hb 1 mentions "few speakers" but gives no figures. It is further commented: "Language dying out in favor of Kanuri, but ethnic population large". (5) Nigeria, Borno State, Borno Division, Margi District. (7) Hansford - Bendor-Samuel - Stanford 1976.

Rishi General remark: our main source in this case is Shimizu (1982a). The language is not mentioned at all in the other sources. Concerning the language name Shimizu (1982a: 112) remarks that " . . . the term rishi is a Hausa-word, which comes from the original *ri-si". (1.1) Shimizu (1982a) mentions ì-sì as autoglossonym. (1.2) He suggests Rishi as the glossonym that is used as the official name for the language. (1.3 and 1.5) Shimizu further mentions ù-ndu rishi 'person of Rishi' as selfapplied ethnonym. (1.4) Shimizu mentions rusancii (as glossonym) and bàrùshii, pl. rusaawaa (ethnonym) as foreign names used by the Hausa. (1.6) Rishi is also given as the name for the main settlement of the speakers of i-sì. (2) Northern Jos, Plateau, Benue-Congo (Niger-Congo). (3.2 or 3.1) Cf. Shimizu's comments on the status of this language. He states that his informant " . . . was an old man, and was said to be the

382

Gabriele Sommer

only surviving bàrùshii, who was still able to speak the language." He nevertheless speaks of two villages where "former ì-sì speakers are found", not mentioning whether they still call themselves Rishi. (3.4) Shimizu suggests that ì-sì was replaced by Hausa. (4.2) One old man in 1974. (4.3) Not known, but probably some. Cf. Shimizu's comment (1981a: 112) that "The Rishi live isolated in the Gura speaking area." (5) Nigeria, three villages in the north of Jos (Rishi, Yelwan Rishi and Sabon Kaura) according to Shimizu (1982a: 89). (7) Shimizu (1982a). Sarwa (1.1) Hb 2 (p. 24) seems to regard Sarwa as a self-applied name. (1.2) Hb 1 (p. 160) also mentions Sharwa, Sarua and Saroua as variants of the language name. (1.5) Hb 3 (p. 42) suggests that Sarwa, Sarua and Saroua are variants of the ethnonym. (2) Fivaz - Scott (1977: 226) classify Sarwa as a member of the Somrai group, Sahel group, Chadic (Afro-Asiatic). (3.3) The small number of speakers of Sarwa might indicate that the language is threatened by extinction. (4.2) The only figure found in the literature stems from Hb 3 (p. 42) where an estimation of circa 400 speakers is proposed. Weimers (1971: 875) and Hb 1 (p. 160, 171) also mentions this figure. (5) Chad (according to Hb 2); Hb 1 suggests Cameroon as another area of distribution. According to them Sarwa is spoken in south central Chad, near Miltou and northeast of Dourbeye in North Province, Cameroon. Segeju (1.2) "Kisegeju" refers and referred to a form of Digo spoken on the coast today. It is also used today to refer to what used to be called Daisü (not to be confused with Daisu, a seven-vowel language mentioned in Hb 2, p. 149). (1.5) Wasegeju (which is an ethnonym that is also used by several people who are not Kisegeju-speaking; see 4.3). (2) Mijikenda dialect, Bantu (Niger-Congo). (3.1 and 3.2) There are no Kisegeju-speakers left in Kenya today according to Nurse (p.c.). There might still be some speakers of Kisegeju in Tanzania.

A survey on language death in Africa

383

(3.4) Kisegeju has been or is being replaced by Digo and/or Swahili. (4.2) No speakers in Kenya. (4.3) In the Tanzania Census of 1969 there were 18,688 people recorded who called themselves Wasegeju. The number of those ethnic Wasegeju is probably smaller in Kenya according to Nurse (p.c.). (5) Kenya, Tanzania, in the coastal border area (southeastern Kenya and northeastern Tanzania). (6) Nurse (p.c.) assumes that two language shifts took place that led to the final extinction of Kisegeju, i.e. (in the Kenyan case) their cultural and linguistic absorption by the Digo: 1. Daisü adopted Segeju at first (this assimilation process occurred in the first half of this century) and 2. then gave up Segeju in favour of Digo and/or Swahili. (7) Nurse (1982). Seki (1.1) Hb 2 (p. 132) suggests Seki as autoglossonym. (1.2) Several variants of the language name can be found besides Seki. These are Sekiyani, Sekiyani, Shekiyana, Bulu and Sheke (cf. Hb 2, p. 132; Hb 1, p. 186 and Weimers 1971: 875). (2) Kele group, Bantu (Niger-Congo) according to Fivaz - Scott (1977: 227). (3.2 or 3.1) The sources agree that Seki is spoken by a small group of people only and that the language is "almost extinct" or spoken by "an almost extinct tribe" (cf. Hb 7, p. 59). The actual source seems to be Hb 7 which is referred to in Hb 4 (p. 24) and perhaps also in Hb 2 where the language is treated as being already extinct (Hb 2, p. 132). (4.2) If one follows the suggestion of Hb 2 (see 3.2/3.1), no speakers of Seki exist nowadays. (5) Gabon, northwest coast around Cocobeach (cf. Hb 1); Hb 4 and Hb 7 suggest "the coast of Gabon to the north of Libreville" as area of distribution. Shabo (1.1) Unseth (1984) mentions Shabo and Shako as synonymous variants of the language name. He prefers Shabo to avoid confusion with an Omotic language called Shako. (1.4) Mekeyir, Mikair and Mekeyer are variants of a xenonym used by the Majang.

384

Gabriele Sommer

(1.5) Probably also Shabo. (2) Several attempts have been made to classify Shabo (cf. the discussion presented in Unseth 1984). Unseth claims that Shabo is not - like the neighbouring Majang language - a member of the Surma subgroup of East Sudanic (Nilo-Saharan) but rather he leaves it unclassified. (3.3) Dimmendaal (1989: 18) claims that Shabo is a language that is threatened by extinction. (3.4) Shabo are bilingual in their own language and Majang. Cf. Dimmendaal's comment (with reference to Unseth) that all the Shabo speak Majang while only a few of the Majang speak Shabo. (4.2) Unseth (1984) was able to interview some younger Shabo-speaking people (approximately 20 years old). These people had some remnant knowledge of Shabo. Whether there were still (older) people who had or have a good command of the language is not clear. (5) Ethiopia, southwestern part. The Shabo live scattered among the Majang. According to Hoekstra, as quoted in Unseth (1984: 2), the Shabo live 10 to 20 miles north of Godare. (6) Dimmendaal (1989: 18) mentions Majang dominance and the assimilation of Shabo to Majang as the main reason for the above mentioned language shift. Cf. his following comment: "The Shabo have become subject to, and have come to be dominated by, the Majang after having been saved by them from slave-traders. . . . " . Unseth (1984: 2) mentions that the Shabo may represent a former hunter-gatherer population in the area. (7) Dimmendaal (1989); Unseth (1984).

Shanga (1.2) Several language names are found in the literature. Our main source in this case is not the information provided by the handbooks (that treat Shanga and Tyenga as synonyms) but an article by Platiel (1982) where she distinguishes between two peoples called Tyanga and Shanga who speak closely related languages. Although she recently (p.c.) mentioned Tyenga as an ethnonym for speakers of both languages we find it more convincing to continue treating them as separate entities. We therefore present the data on Tyanga or Tyenga in a separate entry. Concerning variants of the language name(s), Hb 2 (p. 51) mentions Shanga, Shonga, Kyenga, Kenga, Tyenga, Ceqga, Tchanga. Hb 1 (p. 243) adds Shangawa and Shongawa.

A survey on language death in Africa

385

(1.3 and 1.5) Platiel (1982) suggests that Shanga is not only a glossonym but also the self-applied name for the people who speak Shanga. (2) According to Platiel (1982): Mana-Busa group of Mande. Platiel (1982: 78) mentions that more than 30% of the Shanga vocabulary is in fact Hausa today. (3.2) Platiel suggests that the language is already in a process of extinction. Cf. Platiel's comments on the situation of Shanga (1982: 78): "De même que le tyanga, le shanga est en voie de disparition et même d'une force beaucoup plus dramatique encore. A ma connaissance, il n'y a plus qu'un seul village où le shanga soit encore quotidiennement parlé par une population qui, de même chez les tyanga, est déjà absoluement bilingue." (3.4) Platiel (1982: 78) continues to explain the situation of bilingualism of Shanga speakers who give up their language in favour of Hausa: "Cette disparition du shanga au profit du hausa qui, à l'origine, avait sans doute les mêmes causes, s'est trouvée aggravée ces dernières années par la mise en place d'un immense barrage qui, en inondant une bonne partie de la vallée du Niger, a provoqué le déplacement de nombreuses populations". (5) Nigeria, Sokoto State, Shanga district; according to Platiel (1982: 71): "Les shanga, pour leur part, sont plus circonscrits sur la rive droite du Niger et ne se trouvent qu'au Nigèria, dans la canton de Shanga, proche de Yelwa-Yauri." (6) See 3.4. Expansion of the Hausa language; literacy campaign in Hausa; influx of other ethnic groups in Shanga ethnic territory after the dam on the Niger river was built. (7) Platiel (1982).

Shau (1.1) lì-Shàù. (1.4) According to Shimizu (1982a) shawancii (glossonym) and bàshàyii, pl. shawaawaa (ethnonyms) are used as foreign names for Shau. (1.5) Shimizu (1982a) mentions ù-shàù, pl. à-shàù as self-applied ethnonyms. (2) Northern Jos, Plateau, Benue-Congo (Niger-Congo). (3.2) The language is in a process of extinction, i.e. nearly extinct. (3.4) Hausa is the replacing language? (4.1) One old man in 1975. (4.2) Shimizu (1982a: 118) remarks: "In the evening of 21st February 1975, the chief of Shau sent for one of about 20 old men, who could still

386

(5) (7)

Gabriele Sommer

speak the language. . . . After the interview the older and middle-aged people said that they still knew and could recognise the words this old man was giving me, but that they themselves were no more able to produce them upon request. Younger people had no ideas at all about this language." And: "In the Shau land there is another group of people known as Pàka at Durugù, Sàgeerè and Pàka near Pingel. They do not speak their language any more, but some of them said that it was the same as the Shau language." Nigeria, in Shau (main settlement) and in eight other Shau settlements. See map in Shimizu (1982a: 98). Shimizu (1982a).

Sheni As a source we mainly refer to an article by Shimizu (1982a) who provides detailed information about Sheni (official name). (1.2) Several other variants are mentioned by Hb 2 (p. 86) and Hb 1 (p. 243): Shani, Sheni, Shaini and Asennize (regarded as self-applied ethnonym). (1.4) Shimizu mentions sheenancii (glossonym) and sheeni, pi. sheenaawaa (ethnonyms) as xenonyms used by the Hausa. (1.5) Asennize (cf. 1.2). (2) Northern Jos, Plateau, Benue-Congo (Niger-Congo). (3.2 or 3.1) According to the comments mentioned by Shimizu the language is in an actual process of extinction or is already extinct. (3.4) Sheni has been/is being replaced by Hausa. (4.2) Shimizu mentions two old men who were still able to remember Sheni in 1974. He comments: " . . . apparently the language is no longer spoken by these gentlemen [his informants], for they had some difficulty in producing words upon request". (5) Nigeria, Kaduna State, Saminaka Division, Lere District according to Hb 1 (p. 243); according to Shimizu (1982a): Gurjiya, situated to the east of Ganji, 72 km from Jos. Other Sheni settlements are Shemare and Tudu. (7) Shimizu (1982a). ShiChopi (i.e. Lenge) (1.1) Hb 2 (p. 166) seems to suggest Jicopi as self-applied variant of the language name. Lenge is treated as a synonym for Jicopi. In other sources Lenge is described as an "archaic form" of ShiChopi. Their

A survey on language death in Africa

387

primary source (Earthy 1931: 477) suggests that "The ki-Lenge dialect is allied to Ji-TJopi, but presents some differences from both Ji-TJopi and Ji-Thonga." Lenge, nevertheless, whether it is a dialect, a synonymous name for ShiChopi or an "archaic form" is of interest in our context. The various language names given in this paragraph seem to cover all dialects of the ShiChopi dialect cluster which - as a whole - has a large number of speakers today (cf. Hb 1, p. 216). (1.2) Hb 2 lists Jichopi, Shicopi, Tchopi, Copi and Lenge. Hb 1 (p. 216) further mentions Txopi. (2) Chopi group, Bantu (Niger-Congo). Cf. Fivaz - Scott (1977: 68). (3.2 or 3.1) According to Earthy (1931) Lenge was in an actual process of extinction in 1931. Whether some speakers still speak this "dialect" or "archaic form" today is not clear. We therefore list it in this category. Cf. comments in Hb 4 (p. 157) and Earthy (1931: 477): "Lenge is an archaic form, now disappearing, but still known to older women." And: " . . . the old ki-Lenge language, which is fast dying out, being spoken only by some of the older people (5) Mozambique, according to Hb 1 (concerning the whole Shicopi cluster) on the southern coast, above Limpopo River; Earthy (1931: 477) suggests south of Gazaland in former Portuguese East Africa as area of distribution for Lenge. (7) Earthy (1931). Shirawa (1.2) The only language name given in the sources is Shirawa. This language is mentioned in Hb 1 (p. 243) and Weimers (1971: 877) but not in the other sources. (2) Plateau-Sahel, Chadic (Afro-Asiatic) according to Hb 1. Cf. Fivaz Scott (1977: 74): Ngizim group, Western, Chadic. (3.1) Hb 1 (the only source available) suggests that Shirawa is extinct. (5) Nigeria, Bauchi State and Division, Darazo District. Singa (1.2) Singa, luSinga, liSinga are found as variants of the language name. (2) Nyoro-Ganda group, Bantu (Niger-Congo) according to Hb 1. (3.1) Sources where information on this language is available are Weimers (1971: 878), Hb 1 (p. 280) and Hb 4 (p. 109). They assume that the language is "perhaps extinct" or simply "extinct".

388

Gabriele Sommer

(3.4) Cf. Hb 2 (p. 109): "Whiteley, however, . . . states that the people on this [Rusinga island] and the neighbouring islands now speak Luo (nonBantu), but that there seems to be a small group of people still speaking an archaic form of Ganda." (5) Uganda, Rusinga island.

So (1.1) According to recent information provided by Heine (p.c.) the selfapplied name for those idioms spoken by the Tepes, Kadam and Napak is So. (1.2) The following variants of the language name are found in the literature: Hb 1 (p. 280), Hb 2 (p. 111) and Hb 3 (p. 93) mention Tepes or Tepeth as glossonyms. See also 1.5. (1.3) See 1.5. (1.4) Heine (p.c.) suggests that é-tepes-it, pl. ηί-tepés is a xenonym for the Tepes living on Mount Moroto used by the Karimojong. The corresponding Karimojong names for Kadam and Napak are é-kadamá-it, pl. ηί-kadamá and é-tepes-ít, pl. ηί-tepés, respectively. The Pokot names used for the So, i.e. the Tepes and Kadam, are tapasíáac and kádamá, respectively. (1.5) In most sources (i.e. Hb 2, p. I l l ; Hb 1, p. 280; Hb 3, p. 93 and Heine 1976) Tepes or Tepeth has been used as an ethnonym. Heine (p.c.) now suggests so as the self-applied, i.e. vernacular, name for the speakers of Tepes, Kadam and Napak. The self-applied names of these three idioms are tepes, sg. teps-át, kacfam, sg. kadm-át and yó γ ΐοηί, sg. neBtoiji, respectively. (2) Contradicting classifications of So have been proposed. The classification given in the handbooks (cf. Hb 1, p. 280 or Hb 2, p. I l l ) refers to Greenberg (1963) who classified So as a member of Eastern Sudanic, Chari-Nile (Nilo-Saharan). Heine (1976 and p.c.) claims that So and its closest related languages form an isolated unit he called the Kuliak language group. (3.2) Heine mentions that the language is now in a process of extinction, i.e. the number of speakers is gradually decreasing. He furthermore comments: " . . . the So language is likely to die out in the near future . . . " and " . . . on Mount Moroto and Mount Napak So is spoken exclusively by old people." (3.4) The remaining So speakers tend to replace their mother tongue either by Pokot (especially those living on Mount Kadam) or Karimojong

A survey on language death in Africa

389

(primarily the So living on Mount Moroto). According to Heine, all of the So speakers have Karimojong at least as a second language. (4.1) While earlier sources (cf. Hb 3, p. 93) estimated "a few hundred" So speakers, Heine suggests that the number of speakers has very much decreased by now. Cf. the following comment: " . . . the total number of people with a good command of So does not exceed 500, and is in fact estimated to be much lower." (4.3) Heine estimates between 2,000 and 5,000 people belonging to the ethnic group today although the exact number is unknown. These are figures also found elsewhere in the literature (cf. Hb 1, p. 280 which estimates 5,000 speakers). (5) Uganda, Eastern part, Moroto District, southern Karamoja. According to Heine the So people occupy three extinct volcanic masses, i.e. Mount Kadam (where the Tepes live) to the south, Mount Moroto (where the Kadam are located) to the north, and Mount Napak (location of the Napak) to the west. (6) Heine mentions the gradual replacement of So by either Karimojong or Pokot as a main reason for the decline of the So language. Cf. the following comment: "One reason for this decline is that Karimojong, a second language for most So, has for many become their first language, thus on Mount Moroto and Mount Napak So is spoken exclusively by old people. Even on Mount Kadam, where both the So language and traditions have enjoyed a much stronger status than in the other two areas . . . , the language is losing ground to both Karimojong and Pokot, the latter of which here has become an important second language." (7) Heine (1976; to appear).

Sogoo (1.1) /sokô:/ according to Rottland (1982: 25) or Sogoo as proposed by Heine (1973). (1.3 and 1.5) /sokô:/. (1.6) /sokô:/. (2) Southern Nilotic (Nilo-Saharan). (3.3) The actual status of the language is not clear. We classify Sogoo as being threatened by extinction. (4.2) Rottland discovered a settlement consisting of circa ten round huts inhabited by Sogoo speakers in the 1970s. (5) Kenya, southern Mau forest between the Amala and Ewas Ng'iro rivers. (7) Heine (1973); Rottland (1982).

390

Gabriele Sommer

Southern Khoisan General remark: Most of the languages of Southern Khoisan are extinct or nearly extinct today (Winter 1981: 340). Due to the lack of further reliable data we only list the following language or dialect names. Cf. also Winter's comments on the status of !ga-!ne, Maluti and //xegwi (1981: 342): "Das !ga-!ne wurde um 1932 von Anders (1934/35) aufgenommen. Es handelt sich um die spärlichen Erinnerungen zweier alter Männer von einer Sprache, die sie seit fünfzig Jahren kaum mehr gesprochen hatten." Or: "Mit 'Maluti' ist eine Handvoll Wörter gemeint, die sich in den übersetzten Texten zur Mythologie der Maluti-Buschleute von Orpen 1874 verstreut finden." And: "Die drei bekannt gewordenen Aufnahmen des //xegwi . . . sind lautlich zu unterschiedlich, um sie als Repräsentanten eines einzigen Dialekts auffassen zu können, obwohl die Gesamtsprecherzahl Anfang der 1950er Jahre weniger als drei Dutzend Personen betrug." (1.2) "Oudtshoorn", "Boesjesman", '.khua'i, /nusan, 4=nusa, /kham, "Katkop", "Strontbergen", "Prieska", "B. early", "Sneese", !gä !ne, ? Khuai, "Bushman", "Maluti", //kü/ ? e, "Boshof", =j=unkwe, //q!ke, //xogwi, //kxigwi, k/ ? eg wi, =t=khomani, /quhci, /nusa, =t=kaurure//nai, auni, Ki/hasi, ^kê^kê, "Khakhea", "Masarwa". (2) Southern Khoisan (Khoisan). (3.1 or 3.2) Extinct or nearly extinct. (4.2) Cf. Winter's remark (1981: 347) where he states that approx. 2,000 speakers of all the surviving Southern Khoisan languages can be estimated while most of the above mentioned languages have died out during the last 150 years. (5) See map in Winter (1981: 333). (7) Winter (1981).

Suba (1.2) The only glossonym appearing in the literature is Suba. (2) Bantu, Benue-Congo (Niger-Congo). (3.2) Taking into consideration the various remarks on the so-called "Luoization" of Suba speakers in Kenya and Tanzania as mentioned in Heine - Möhlig (1980: 76), one has to assume that the Suba language is gradually becoming extinct. But see also the more detailed observations on language use as presented in Rottland - Okoth Okombo (1986: 117f.): "One might assume . . . that the Suba language has almost died out by now, and this is in fact what historians referring to the Suba

A survey on language death in Africa

391

have said or implied. It is, however, far from being true and the picture of the spread of Suba competence shows considerable variation. We can distinguish 3 types of Suba areas, viz. (a) strong areas (Mfangano, Muhuru), where you find a strong and functioning Suba speech community throughout; (b) median areas (Gwassi, Kaksingiri, Suna), where the standard of Suba competence is not uniform, on average lower, and where the vitality of a Suba speech community is proportional to its lack of accessibility; (c) weak areas (Rusinga and probi. Gembe), where Suba is the "secret language" of a few old people." (3.4) Heine - Möhlig (1980) and Rottland - Okoth Okombo (1986) mention a high degree of assimilation to the Luo language, i.e. Luo often becomes the mother tongue of Suba speakers, at least in certain areas. Cf. Rottland - Okoth Okombo (1986: 118): "Almost all persons who call themselves Suba speak Luo, either as a first or as a second language." Cf. also Hb 1 (p. 207) which estimates a rather high number of "Suba" (obviously referring to the ethnic group): "Population is generally bilingual or monolingual in Luo." Heine - Möhlig (1980) remark that the Suba form a minority group among the Luo in South Nyanza District, Kenya, where they are gradually replacing their mother tongue by Luo. (4.2) Cf. Heine - Möhlig (1980: 76): "Their number [Suba] is not known - the Kenya Population Census does not distinguish them but rather classifies them as Luo." (4.3) As mentioned above the rather high estimation of 17,000 Suba speakers in Tanzania and 59,668 speakers in Kenya as suggested in Hb 1 (p. 207) describes the Suba ethnic group rather than the actual number of speakers. (5) Tanzania, Kenya (South Nyanza District, especially in Rusinga, M f a n g ' ano Islands, Kaksingri, Gwassi, Suna Locations on the opposite mainland). (6) Cf. the following comments taken from Heine - Möhlig (1980: 76) who assumed a gradual process of language shift in both Tanzania and Kenya: "The Suba gradually being assimilated culturally and linguistically by the Luo. . . . linguistically this process leads towards the replacement of the mother tongue by Luo." According to Heine - Möhlig this is certainly the case in Tanzania. Several stages of language use as observed in Tanzania are described: " . . . while the Simbiti in the south have hardly been affected by this process ["Luo-ization"], other Suba groups such as Sweta and Surwa seem to have passed the first stage of "Luo-ization"; . . . Luo is spoken as a second language by most people with the exception of a few older persons, in particular old women.

392

(7)

Gabriele Sommer

The Kine are slightly more "Luo-ized". . . . Luo is spoken throughout, but Suba is still the first language . . . ". The following reasons for the above mentioned language shift are given (Heine - Möhlig 1980: 78): "The beginnings of language shift go back to a time when Swahili had not yet spread in the Nyanza region. Luo became the linguistic medium of communication between Suba and Luo and a case of assimilating bilingualism emerged where the minority gradually became absorbed linguistically and culturally, by the surrounding majority." Heine - Möhlig (1980); Rottland - Okoth Okombo (1986).

Tagbu (1.2) Hb 1 (p. 264) and Hb 2 (p. 101) mention Tagbo and Tagba as variants of the language name. (1.5) Hb 2 (p. 101) suggests that Taßu is an ethnonym. (2) According to Fivaz - Scott (1977: 36): Ndogo-Feroge group, AdamawaEastern. Hb 3 classifies Tagbu as a member of the Ndogo-Sere cluster. (3.3) The small number of speakers might indicate that the language is threatened by extinction. (4.2) Hb 1, Hb 3 and Weimers (1971: 883) estimate 100 or fewer speakers of Tagbu. (5) Mainly Sudan (cf. Hb 2 and Weimers 1971), Hb 1 mentions: mainly in Zaire and Central African Republic and "a few" in the Sudan. (6) Cf. Santandrea's (1961: 2) comment: "Tagbu. A tribe recently discovered - as such - by the author, . . . Their only known settlement is near Mboro Mission, at Ngua, with 12 T.P. [tax payers] is all! A few individuals can still be met at Dem Zubeir and Raga, and scanty groups, on the way either to extinction or to absorption into the big Zande melting pot, are scattered here and there in the Central African Republic, Obo Territory." (7) Santandrea (1961).

Taita C u s h i t i c (1.1) The self-applied name of these languages is not known. (1.2 and 1.6) Ehret - Nurse (1981) - our main source in this case - suggest Taita Cushitic as language name. They distinguish - according to the interpretation of their reconstructed linguistic material - between Taita Cushite A (TC-A) and Taita Cushite Β (TC-B). They suggest these names mainly because of the lack of any synchonically available data

A survey on language death in Africa

393

besides Cushitic loanwords in two Bantu languages belonging to the Chaga-Taita subgroup. Ehret - Nurse (1981: 125) remark: "The region [Taita Hills] was one of even greater ethnic diversity in the past, a homeland for not only a variety of Bantu settler populations but of several distinct Southern Cushitic communities as well. Loanwords in the present-day Bantu languages, Dawida and Saghala, reveal the former presence of at least three separate and different Southern Cushitic languages, one probably persisting down to quite recent times before its speakers were finally absorbed into the various Bantu communities." As the name Taita Cushitic refers to the area of distribution of these Bantu languages that show remnant features of formerly existing Cushitic languages, it can also be understood as a toponym. Merritt (1975: 29ff.) who analyzed oral traditions of the Taita peoples also assumes that "two pre-Taita peoples", usually classified as Southern Cushites, formerly inhabited the Taita Hill area. Concerning the names for these peoples he comments: " . . . the names by which they are remembered vary and are often interchanged one for the other; . . . The names include Bisha, Wasi, Sikimi, and Nyamba". (2)

Southern Cushitic, Cushitic (Afro-Asiatic). Cf. Ehret - Nurse (1981: 126): " . . . their next nearest linguistic relatives are the diverse Cushitic languages of Ethiopia and the Horn". (3.1) Both Cushitic languages are extinct. Linguistic features of TC-A and TC-B are only traceable through loanword evidence in the Bantu languages spoken in the Taita Hill area today (i.e. Dawida and Saghala and their several dialects). (3.4) The date of extinction of TC-A and TC-B, i.e. their absorption into Bantu languages, is not clear. Ehret - Nurse (1981: 140f.) suggest: "The Taita Cushites . . . could have reached the Taita area conceivably as early as the second millenium B.C. . . . " and "the Taita Cushite presence would have been of some centuries standing by the period A.D. 500-1000". These interpretations seem to be very tentative indeed as dating procedures in historical linguistics generally are. They therefore have to be understood as mere estimations in our context. (5) South-eastern Kenya, Taita Hills "located at a crossroads of population movement through eastern Kenya and far northeastern Tanzania" (Ehret - Nurse 1981: 125). The Bantu languages Dawida and Saghala where Southern Cushitic loanwords can be found are spoken on the main range of peaks of the Taita Hills (Dawida) and Teri Valley, Taita Hills (Saghala).

394

(6)

(7)

Gabriele Sommer

Cf. the comments of Ehret - Nurse (1981: 125) that indicate a long period of multilingualism in the Taita Hill area. How, when and why the Cushitic languages were finally absorbed by the neighbouring Bantu languages is not clear at all (Ehret - Nurse 1981: 125f.): "The sorts of word-borrowing that took place [between Southern Cushitic and the neighbouring Bantu languages and also a Ma'a-speaking community] reflect several long-term situations of extensive bilingualism, or more probably multilingualism, and indicate that at least two of the Southern Cushitic societies at some point in time formed a significant, probably majority proportion of the Taita population, in all likelihood constituting the original pre-Bantu agricultural settlement of the hills." Ehret - Nurse (1981); Merritt (1975).

Taura (1.1) Shimizu (1982a: 114) - our main source in this case - suggests u-cìlà àtàkàyà 'language of Takaya people' as self-applied name. (1.2) The official name and well-known one is either Taura or Taurawa (cf. Hb 1, p. 244 or Weimers 1971: 884). (1.3 and 1.5) Shimizu (1982a) suggests à-tàkàyà as ethnonym. (1.4) He mentions taurancii (glossonym) and bàtàurii, pl. tauraawaa (ethnonyms) as the names applied by the Hausa. (2) Northern Jos, Plateau, Benue-Congo (Niger-Congo). (3.2 or 3.1) Varying statements are found concerning this point. While Weimers (1971: 884) estimated 810 speakers in the early seventies, Shimizu (1982a: 114) remarks that only a few speakers of Taura existed around 1975 ("She [his informant] was said to be one of the few women who can still speak the language, . . . "). Hb 1 (p. 244) finally regards this language as being already extinct. (3.4) Hausa is the replacing language. (4.2) Only a few speakers of Taura in the 1970s. (5) Nigeria, according to Hb 1: Bauchi State and Division, Lame District. Shimizu mentions the following "Taura settlements": Taura, Tulu, Nasaraawò and Kulfana. (7) Shimizu (1982a). Tenet (1.2) Tenet (see 1.5), also Irenge and Tenat according to Hb 3 (p. 88). (1.5) Tenet according to Dimmendaal (1983: 335).

A survey on language death in Africa

395

(2)

Didinga-Longarim-Murle group, Surma, Eastern Sudanic (Nilo-Saharan) according to Dimmendaal (1983). (3.3) Tenet seems to be threatened by extinction. Cf. Dimmendaal (1989: 26): " . . . the Tenet language is clearly threatened because its use is tending to become more and more restricted socially". (3.4) Lopid is the replacing language. Cf. Dimmendaal (1983: 340): "Alles deutet darauf hin, daß sich ein allmählicher Sprachwechsel zum Lopid hin vollzieht." And Dimmendaal (1989: 26): "At home the Tenet speak either Tenet or Lopid; all ethnic Tenet seem to be fluent in both languages, (5) Sudan, southern part, Lopid (Lafid) Mountains. (6) Cultural and linguistic assimilation to Lopid (see Dimmendaal 1983: 339ff. and 1989: 26): "Their shifting language allegiance is probably not due to the negative prestige of Tenet but, rather, to the importance of Lopid iron-working and, by consequence, of the Lopid language." (7) Dimmendaal (1983, 1989). Terik (1.1) /té:rik/ according to Rottland (1982). (1.2) Nyang'ori (official name). (1.3) / t é r i k / , /terikék/,

sg. /teríki:n/,

/terikùntét/.

(2) Elgon group, Southern Nilotic (Nilo-Saharan). (3.2) We tentatively classify Terik as being in a process of extinction. (3.4) Rottland (1982: 24) only suggests that the cultural and linguistic influence of Nandi is very strong which might lead to a gradual assimilation to the Nandi within a few generations. (5) Kenya, "Nyang'ori Location" on the southwestern end of Nandi territory (from Serem in the northeast to the Kisumu-Kakamega road in the southwest). There are some Terik groups living in Luhya-speaking areas. (7) Heine (this volume); Roeder (1986); Rottland (1982). Tijanji General remark: It is not clear whether Tijanji/Ajanci is identical with or related to Ajawa/Ajanci (see separate entry). (1.1) Hb 2 (p. 87) regards Tijanji as the self-applied name. (1.2) Several variants of the language name are found in the literature: Janji, Anafejanzi, Jenji, Ajanji (cf. Hb 1, p. 232). (1.3) Hb 2 (p. 87) regards Ajanji and Anafejanzi as self-applied ethnonyms.

396

Gabriele Sommer

(1.5) See 1.3. (2) Northern Jos, Western Plateau, Benue-Congo (Niger-Congo) according to Hansford - Bendor-Samuel - Stanford (1976: 93). (3.3) Tijanji might be threatened by extinction (see 4). (4.2) The only estimations available are found in Hb 1 (p. 232), Weimers (1971: 815) and Hansford - Bendor-Samuel - Stanford (1976): 360 speakers (a figure from 1950). (5) Nigeria, according to Hb 1 : Plateau State, Jos Division, Rukuba District. (7) Hansford - Bendor-Samuel - Stanford (1976).

Togoyo ( 1.2) The language name mentioned in the various sources is either Togoyo (mainly used) or Togoy (cf. Hb 1, p. 265). (2) According to Fivaz - Scott (1977: 241): Ndogo-Feroge group, Eastern branch, Adamawa-Eastern. (3.2 or 3.1) Various statements can be found concerning the status of the language. These might indicate that the language has become extinct recently. While Weimers (1971: 887), Hb 3 (p. 22) and Hb 1 (p. 102) mention that the language is "almost extinct", Hb 2 (p. 102) indicates that Togoyo has already become extinct. (5) Sudan, according to Hb 1: west Sudan in a small area around Raga. (7) Santandrea (1969).

Tyanga General remark: cf. entry Shanga in this survey. As mentioned already Tyenga is usually regarded as a synonym for Shanga, Shanga being more often used as glossonym in this case (cf. Hb 2, p. 51; Hb 1, p. 243). Platiel (1982) treats Tyenga as a separate language from Shanga although both have to be regarded as closely related languages belonging to the same subgroup (see 2). (1.1) Tyanga (or Tyenga), actually an ethnonym, is used as glossonym as well in the literature. Whether both terms were indeed identical is difficult to find out today mainly because only the ethnic group seems to have survived while Tyenga has been replaced by Hausa. (1.2) For the other variants of the language name see entry Shanga. (1.5) According to Platiel (1982) Tyanga is an ethnonym. (2) Mana-Busa group of Mande.

A survey on language death in Africa

397

(3.2) According to Platiel (1982) the language is in a process of extinction. The language Tyanga seems to have already died out in Niger: cf. Platiel (1982: 76): " . . . les populations qui se disent tyanga et qui, d'ailleurs, conservent très fortement le sentiment de leur appartenance à ce groupe, affirment ne plus parler leur langue depuis au moins deux générations. A l'heure actuelle, tous les tyanga parlent le hausa, mais ils prétendent n'avoir adopté cette langue que récemment et qu'auparavant, ils parlaient le zerma." (3.4) Tyanga was either replaced by Hausa (in Niger, Nigeria and Benin) or Zerma (primarily in Niger). (4.2) Platiel (1982: 77) mentions several villages that are inhabited by Tyanga speakers in Benin while only four were found in Nigeria. (4.3) The number of those who call themselves Tyanga or Tyenga while only speaking Hausa or Zerma today is considerably higher. (5) Benin (in several villages), Nigeria (in four villages on the right bank of the Niger next to the Benin border), no longer in Niger. (6) Cf. entry Shanga in this survey. One has to assume that Hausa dominance is the main reason for the replacement of Tyenga by Hausa. (7) Platiel (1982).

Ware (1.2) The only glossonym mentioned in the literature is Ware (cf. Hb 1, p. 272 and Weimers 1971: 894). Cf. also Hb 4 (p. 114): "Nothing further is known of the Ware, who in Johnston's time were living on one of the islands in the Kavirondo Gulf, (2) Hb 1 classifies Ware as a member of the Ragoli-Kuria subgroup of Bantu, Benue-Congo (Niger-Congo). (3.1) The only sources that mention a language called Ware are Hb 1 and Weimers (1971). Both suggest that the language is "perhaps extinct". (5) Tanzania.

Wetu (1.2) The most frequently used glossonym seems to be Wetu. Hb 2 (p. 70f.) mentions in a footnote that Wetu was classified as one member of a dialect cluster called Rumbuk Jur in the Sudan Census of 1962 (1955). (1.5) Hb 2 also remarks that Wetu is obviously an ethnonym. Hb 3 (p. 12) mentions in a footnote: "These [Wetu] may be the people referred to by Schweinfurth, Junker, and others as Mittu, although the vocabulary

398

Gabriele Sommer

of Mittu collected by Schweinfurth is almost identical with Morokodo. No trace of any people called Mittu can be found at the present day." (2) According to Fivaz - Scott (1977: 11), Morokodo-Mittu is classified as belonging to Bongo-Baka, Bongo-Bagirmi, Central Sudanic (NiloSaharan). (3.2 or 3.1) The only reference concerning the status of the language in this case is Hb 3 (p. 12) where Wetu is regarded as a language that is "almost extinct". (4.2) Hb 3 (p. 12) mentions "five families". (5) Sudan.

Weyto (1.1) Gamst (p.c.) suggests Wäyto as a language name and comments: "Today, the Wäyto have no other name for themselves except for 'Wäyto'. The word 'Wäyto' may well be an appellation given to the Wäyto by the Amhara, perhaps with a similar etymological root as "Watta", the name given to the hippopotamus hunters of Lake Zwai, much further south of the Lake Tana home of the Wäyto. In all, the word 'Wäyto' may well originally have been a xenonym that became an ethnonym among the Wäyto people." (1.2) Weyto is the only name mentioned in Hb 1 (p. 184). (1.5) Dimmendaal (1989) uses Weyto as ethnonym, too. (2) Hb 1 proposes that Weyto should be regarded as "Unclassified. Possibly was Eastern Sudanic", i.e. (Nilo-Saharan). Dimmendaal, nevertheless, comments (1989: 18) on the genetic classification of Weyto: " . . . their former language [Weyto] probably belonged to Cushitic", i.e. (AfroAsiatic). Gamst (p.c.) remarks on the classification of Weyto that " . . . it can be assumed that if the Wäyto did not speak Amharic 200 years ago, their language must have been Agäw, the original tongue of all northwestern Ethiopia". (3.1) According to Hb 1 and Dimmendaal (1989): extinct. Cf. also Gamst's comment on the status of Wäyto: "In the late 1700s, James Bruce of Scotland, who spoke Amharic, visited the Wäyto and said that they had a language quite distinct from Amharic. He said the Wäyto were socially quite distinct from their neighboring Ethiopians and held in "utter abhorence" by them as being ritually unclean (Bruce 1790, iii:403). By the time of the research of Eugen Mittwoch, about 1900, the aboriginal language of the Wäyto was extinct. No trace of the older native Wäyto language was to be found and all Wäyto spoke Amharic (Mittwoch

A survey on language death in Africa

399

1907). In my 1964-1965 field research among the Way to, I found no surviving native words, not even relating to their hunting and fishing worktasks. All that was found were specialized words, for technological artifacts and for certain animals and plants in the Tana environment. All Wäyto today speak Amharic as the sole language. An exception is that some who have been to school also speak English. And a number of Arabic words and phrases from the Quran and concerning other Islamic religious matters are sometimes used by informed Wäyto adults." (3.4) Hb 1: "People now speak Amharic-based jargon." Cf. Dimmendaal (1989: 18): "The ethnic Weyto live in an area surrounded by speakers of Amharic which they use as their primary and first language (See also 3.1.) (4.3) While there are no speakers of Weyto left today, Dimmendaal talks about "ethnic Weyto" which implies that some members of the ethnic group remained after the people no longer spoke their former language. Gamst, referring to this publications of 1979 and 1984, reports that in the mid-1960s, perhaps some 2,000 persons were considered by nonWäyto to be Wäyto. But not all Wäyto would admit before a stranger to be Wäyto. (5) Ethiopia, Lake Tana Region. Gamst mentions the area along all of the shores of Lake Tana, but especially the southwestern shores as areas of distribution. He adds that as the Wäyto became tenant agriculturalists, some Wäyto moved somewhat inland from these shores. (6) Dimmendaal holds the view that the hippopotamus-hunting Weyto gave up their earlier language after having come into contact with surrounding Amharic-speaking peoples. Gamst remarks that the reasons for giving up the language are unknown. (7) Bruce (1790); Dimmendaal (1989); Gamst (1965, 1979, 1984); Mittwoch (1907).

X a m t a q a (i.e. Abergelle A g a w ) General remark: This entry is rather tentative because of the scant information available. Hb 2 (p. 15) suggests that there is only one dialect of their Xamtaqa that is threatened by extinction, namely Abergelle Agaw. Hetzron (1976) seems to call this "dialect" Xamta of Abergelle. (1.2) While Hb 2 suggests that Abergelle Agaw, Ç iraça Agäw or TcheratzAgaw are synonyms Hetzron uses a different nomenclature (cf. Hetzron 1976). (2) Cushitic (Afro-Asiatic).

400

Gabriele Sommer

(3.1) The only information available is provided by Hetzron (1976: 35): "Eastern Agaw, with Xamta of Abergelle, . . . but it is not certain that there are any Xamta speakers left today." (3.4) Cf. Hetzron (1976): "They [Xamta Abergelle] are also bilinguals, also speaking Tigrinya (Xamta) . . . ". (5) Ethiopia. (7) Hetzron (1976).

Xiri (1.1) Hb 2 (p. 167) regards Xiri as the self-applied glossonym. (1.2) Other variants are mentioned in the various sources (cf. e.g. Hb 1, p. 255): Grikwa, Griqua, Gry, Cape Hottentot, ixrikwa and Xirikwa. Hb 3 (p. 160) seems to regard Grikwa as glossonym and ethnonym (see 1.5). (1.5) Grikwa or Griqua. (2) Central Khoisan (Khoisan). (3.1) Although Hb 1 tentatively suggests that Xiri is "nearly" or "possibly extinct", we classify the language as being already extinct. We herein follow the comment taken from Hb 3 (p. 160): "This language is now extinct, and it is not known whether any individuals survive who still can speak it." (5) South Africa; according to Hb 3: " . . . in Griqualand East, i.e. in southern Natal around Kokstad, in the east of the Kei, in the Karroo, and in the western Cape perhaps as far as Oudshoorn".

Yaaku (1.1) The self-applied language name is Yaaku according to Heine (p.c.). (1.3 and 1.5) Yaaku, sg. Yakunte. (1.4) Mogogodo, Mokogodo and Mukogodo have been used as synonyms for Yaaku by several authors (see 3.4). (2) Cushitic (Afro-Asiatic). (3.2 and 3.4) Cf. Heine (1975: 32): "The recent history of the Yaaku provides an example of language replacement - a process that is presently nearing completion." And: " . . . all Yaaku are Maasai-speaking, while there are no monolingual Yaaku. Only a small part of the population knows Yaaku. . . . A knowledge of Yaaku is virtually non-existent within the younger population". Yaaku has been replaced by a dialect of Maasai since the 1930s. This dialect which is called Mukogodo has been de-

A survey on language death in Africa

401

scribed in Heine - Brenzinger (1988). In addition to the former bilingual situation of the Yaaku assimilating to the Maasai, the use of the lingua franca Swahili has increased in this area. Cf. Heine (1975: 34): "The gradual extinction of Yaaku coincides with the expansion of Swahili as a lingua franca." (4.2) Heine estimated less than 200 speakers of Yaaku in the 1970s. There are only 10 Yaaku speakers left today who have some command of the language according to Brenzinger (p.c.). (5) Kenya, eastern part of Mukogodo Division, Laikipia District, about 30 miles north of Mount Kenya. (6) See also 3.2/3.4 and Heine (1975). The Yaaku have gradually assimilated to the Maasai. This not only led to the adoption of their language as the following comment shows (Heine 1975: 28): "They [the Yaaku] are one of the five "Dorobo" groups inhabiting the Mukogodo Division which all depended on hunting and gathering before they turned to pastoralism in the course of the last decades." (7) Heine (1975); Heine - Brenzinger (1988).

Yashi (1.2) Yashi is the only glossonym that appears in the various sources. (2) Benue, Benue-Congo (Niger-Congo) according to Hansford - BendorSamuel - Stanford (1976: 163). (3.3) We class the language as being threatened by extinction in consideration of the small number of speakers (see 4.2). (4.2) Hb 1, p. 246, (the only source where an estimation of the number of speakers is given) suggests 400 speakers (referring to SIL sources). (5) Nigeria, Plateau State, Akwanga Division, Mama District. (7) Hansford - Bendor-Samuel - Stanford (1976).

Yei (1.1) ShiYéí. (1.2) Hb 2 (p. 162) lists several language names: Ci Yei, Yei, Yeei, Yeye, Yeyi, Koba and Kuba. (1.3) Voßen (p.c.) mentions ShíYéí as autonym which probably corresponds to Hb 2's entry Ciyei. (1.4) Makoba. (1.5) Voßen claims that BaYei, BaYeyi and BaYeei are ethnonyms.

402 (2) (3.2) (3.4) (4.2) (4.3)

(5)

(7)

Gabriele Sommer

According to Hb 2: R 40, Bantu (Niger-Congo). Cf. Fivaz - Scott (1977: 253): Yeye group, Bantu. Voßen classifies it as being in a process of extinction. Yei is replaced by a dialect of Tswana. Voßen states that no figures are available. He nevertheless mentions the figure of 20,000 persons belonging to the same ethnic group (vs. those who have a good command of the language). According to Voßen: in and all around the Okavango Delta, Ngamiland, North-West District, Botswana; also in East Caprivi, Namibia. (See map in Voßen 1988: 38.) Voßen (1988).

Zaramo (1.2) Zaramo. (2) Bantu (Niger-Congo). (3.3 or 3.2) According to Heine (p.c.) the language is at least threatened by extinction. (5) Tanzania; in and around Dar es Salaam. (6) Urbanization?

Zenaga (1.2) Zenaga and Znaga (cf. Hb 2, p. 11 and Hb 1, p. 215). (2) Berber (Afro-Asiatic). (3.3) Contradicting statements are mentioned in the literature. Hb 2 mentions those that regard Zenaga as already becoming extinct and those that regard Zenaga as already becoming extinct and those that deny this. We therefore classify Zenaga as being threatened by extinction. The data provided by Hb 1 do not contain any information concerning the status of Zenaga. (4.2) Hb 1 mentions the rather high number of 16,000 speakers of Zenaga who are Muslim, i.e. Arabicized Bedouin Berbers. (5) Mauretania, between Mederdra and the Atlantic coast according to Hb 1.

A survey on language death in Africa

403

Ziriya (1.1 and 1.5) Shimizu (1982a:108ff.) - the only source available where detailed information on this language is given - mentions Zìrìyà as name for the language and the people. (2) Northern Jos, Plateau, Benue-Congo (Niger-Congo). (3.2) The information provided by Shimizu indicates that the language is nearly extinct. Cf. Shimizu's comment: "These two old men [his informants] were still able to remember something of their own language (3.4) Zìrìyà is being replaced by Hausa. (4.2) Only a few speakers seem to have some knowledge of Zìrìyà nowadays. (5) Nigeria; according to Shimizu (1982a): several villages such as Zìrìyà (5 km south of Yaagi near Chokobo), Kéré (two miles east of Rahaman Bauchi) and probably Wurno. (7) Shimizu (1982a).

References Adams, P. Gustaf A. 1907 "Die Banoho und Bapuku in Kamerun", Anthropos 2: 1022-1028. Ardener, Edwin 1956

Coastal Bantu of the Cameroons. (Ethnographic Survey of Africa, Western Africa 11.) London: International African Institute. Baumann, Hermann (ed.) 1975

Die Völker Afrikas und ihre traditionellen baden: Franz Steiner Verlag.

Kulturen. Vol. 1. Wies-

Bell, Herman 1973

"An extinct Nubian language in Kordofan", Sudan Notes and Records 54: 73-80. 1975 "Documentary evidence on the Haraza Nubian language", Sudan Notes and Records 56: 1-35. Bender, M. Lionel 1983

"Remnant languages of Ethiopia and Sudan", in: Bender (ed.), 336354. Bender, M. Lionel (ed.) 1976 1981

The non-Semitic languages of Ethiopia. East Lansing, Michigan: African Studies Center, Michigan State University. Peoples and cultures of the Ethio-Sudan borderlands. Michigan, East Lansing: African Studies Center, Michigan State University.

404

Gabriele Sommer 1983

Nilo-Saharan language studies. Michigan State University: African Studies Center. Bird, Charles Stephen 1982 The Dialects of Mandenkan. Bloomington: Indiana University African Studies Center. Blackburn, Roderic Hall n.d. East African hunter-gatherers: a survey. [Unpublished material.] Boyd, Raymond 1974 Etude comparative dans le groupe Adamawa. Paris: Société d'Etudes Linguistiques et Anthropologiques de France. Broß, Michael 1988 Materialien zur Sprache der Ndam von Dik (Rep. Tschad): Untersuchungen zur Phonologie und Morphologie. [Unpublished Μ.Α. thesis, Marburg University.] Bruce, James M. 1790 Travels to Discover the Source of the Nile, 1768-73. (5 vols.) Edinburgh: G. and J. Robinson. Bryan, Margaret A. 1945 "A linguistic no-man's land", Africa 15: 188-205. 1959 The Bantu languages of Africa. London: Oxford University Press. Calloc'h, J. 1911 Vocabulaire Français-Ifumu (Batéké). Paris: Librairie Paul Geuthner. Cameron, Barbara A. - J. Iverson n.d. A sociolinguistic survey of the Bullom languages of Sierra Leone. [MS.] Costermans, B.J. 1938 "De Kazibati", Kongo-Overzee 4: 177-184. Creissels, Denis 1986 "Le système prédicatif du kagoro", Mandenkan 11: 1-16. Dalby, T. David P. 1962 "Language distribution in Sierra Leone (with map)", Sierra Leone Language Review 1: 62-67. 1963a "Banta and Mabanta", Sierra Leone Language Review 2: 23-25. 1963b "The extinct language of Dama", Sierra Leone Language Review 2: 50-54. De Lavergne de Tressan, M. 1953 Inventaire linguistique de l'Afrique Occidentale française et du Togo. Dakar: Institut Fondamental d'Afrique Noire. Dieu, Michel - Patrick Renaud (eds.) 1983 Atlas linguistique de l'Afrique Centrale, inventaire préliminaire: le Cameroun. Yaounde: Agence pour Co-opération Culturelle et Technique and Cerdotola.

A survey on language death in Africa

405

Dimmendaal, Gerrit J. 1983 "Tenet (Surma): Ein Fall partiellen Sprachwechsels", in: Voßen Claudi (eds.), 331-344. 1989 "On Language Death in Eastern Africa", in: Dorian (ed.), 13-31. DNAFLA (Direction Nationale d'Alphabétisation Fonctionelle et de la Linguistique Appliquée) 1980 Dialectes mandenkan - mandenkan bolofaraw. [Roneotyped.] Bamako - Paris: DNAFLA and Agence pour Co-opération Culturelle et Technique. 1983 Dialectes manding du Mali. [Roneotyped.] Paris: Agence pour Coopération Culturellle et Technique and DNAFLA. Doornbos, Paul - M. Lionel Bender 1983 "Languages of Wadai-Darfus", in: Bender (ed.), 42-79. Dorian, Nancy C. (ed.) 1989 Investigating obsolescence: Studies in language contraction and death. (Studies in the Social and Cultural Foundations of Language 7.) Cambridge: Cambridge University Press. Earthy, E. Dora 1931 "A Chopi (ΤΣορί) love song and a story in Ki-Lenge", Africa 4: 475-482. Ehret, Christopher 1980 The historical reconstruction of Southern Cushitic phonology and vocabulary. (Kölner Beitäge zur Afrikanistik 5.) Berlin: Dietrich Reimer. Ehret, Christopher - Derek Nurse 1981 "The Taita Cushites", Sprache und Geschichte in Afrika 3: 125-168. Ehret, Christopher - Edward Derek Elderkin - Derek Nurse 1989 "Dahalo lexis and its sources", Afrikanistische Arbeitspapiere 18: 5-49. Elderkin, Edward Derek 1974 The phonlogy of the syllable and the morphology of the inflected word in Dahalo. [Unpublished M.A. thesis, Nairobi.] 1976 "Southern Cushitic", in: Bender (ed.), 278-297. Fivaz, Derek - Patricia E. Scott 1977 African languages. A genetic and decimalised classification for bibliographic and general reference. Boston: G.K. Hall & Co. Fleming, Harold Crane 1964 "Baiso and Rendille: Somali outliers", Rassegna di Studi Etiopici 20: 35-96. 1969 "Asa and Aramanik: Cushitic hunters in Masai-Land", Ethnology 8: 1-36. 1986 "Hadza and Sandawe genetic relations", Sprache und Geschichte in Afrika 7 (2): 157-187.

406

Gabriele Sommer

Forde, Daryll (ed.) 1956 Linguistic survey of the Northern Bantu borderland. Vol. 1. London: Oxford University Press. Galtier, Gérard 1980 Problèmes dialectologiques et phonographématiques des parlers mandingues. [Roneotyped; Ph.D. thesis, Université de Paris VII.] Paris. Gamst, Frederick 1965 Travel and research in northwestern Ethiopia. (Notes for Anthropologists and Other Field Workers in Ethiopia 2.) Addis Ababa: Institute for Ethiopian Studies, Haile Sellassie I University. 1969 The Qemant. New York: Holt, Rinehart and Winston. 1979 "Wayto ways: Change from hunting to peasant life", in: Hess (ed.), 233-238. 1984 "Wayto", in: Weeks (ed.), 852-856. Ganslmayr, Herbert - Herrmann Jungraithmayr (eds.) 1977 Gedenkschrift Gustav Nachtigal 1874-1974. (Veröffentlichungen aus dem Übersee-Museum Bremen, Reihe C, 1). Bremen: Übersee-Museum. Greenberg, Joseph Harold 1963 The languages of Africa. Bloomington: Indiana University. 1972 "On the identity of Jungraithmayr's Mimi", Africana Marburgensia 5 (2): 45-49. Grimes, Barbara F. (ed.) 1984 Languages of the world. Ethnologue. (10th edition.) Dallas, Texas: Wycliffe Bible Translators. Guarisma, Gladys - Suzanne Platiel (eds.) 1980 Dialectologie et Comparatisme en Afrique Noire. Paris: SELAF. Guthrie, Malcolm 1953 The Bantu languages of Western Equatorial Africa. London: Oxford University Press. 1959 Revised classified list of Bantu languages. [Unpublished MS.] Hansford, Keir - John Bendor-Samuel - Ron Stanford 1976 An index of Nigerian languages. (Studies in Nigerian Languages 5.) 5-204. Accra: Summer Institute of Linguistics. Hayward, Richard John 1978 "Bayso revisited. Some preliminary linguistic observations, I", Bulletin of the School of Oriental and African Studies 41 (3): 539-570. Heine, Bernd 1968 Die Verbreitung und Gliederung der Togorestsprachen. (Kölner Beiträge zur Afrikanistik 1.) Berlin: Dietrich Reimer. 1973 "Vokabulare ostafrikanischer Restsprachen. Teil II", Afrika und Übersee 57 (1): 38-49.

A survey on language death in Africa

1974-1975 1975 1975-1976 1976 1977 1980 1981 1982a 1982b

to appear Heine, Bemd 1988 Heine, Bernd 1980

Heine, Bernd 1975-1976 1979

407

"Tepes und Nyang'i. Zwei ostafrikanische Restsprachen", Afrika und Übersee 58 (3-4): 263-300. "Notes on the Yaaku language (Kenya)", Afrika und Übersee 58 (1-2): 27-61, 119-138. "Bemerkungen zur Elmolo-Sprache", Afrika und Übersee 59 (4): 278-299. The Kuliak languages of eastern Uganda. Nairobi: East African Publishing House. "Bemerkungen zur Boni-Sprache (Kenia)", Afrika und Übersee 60 (4): 242-295. The non-Bantu languages of Kenya. (Language and Dialect Atlas of Kenya, 2.) Berlin: Dietrich Reimer, 175-178. "Lorkoti Dorobo, a Maasai dialect", in: Hofmann (ed.), 31-46. Boni dialects. (Language and Dialect Atlas of Kenya 10.) Berlin: Dietrich Reimer. "Language and culture south of Lake Turkana", in: Deutsche Forschungsgemeinschaft (ed.). Recent German research on Africa: Language and culture. Part 1, German Africanistic Research. Boppard: Harald Boldt, 98-105. The So language of Eastern Uganda. [MS.] Matthias Brenzinger "Notes on the Mukogodo dialect of Maasai (Kenya)", Afrikanistische Arbeitspapiere 14: 97-131. Wilhelm J.G. Möhlig Geographical and historical introduction. Language and society. Selected bibliography. (Language and Dialect Atlas of Kenya 1.) Berlin: Dietrich Reimer. Rainer Voßen "Zur Stellung der Ongamo-Sprache (Kilimandscharo)", Afrika und Übersee 59 (2): 81-105. "The Kore of Lamu: A contribution to Maa dialectology", Afrika und Übersee 62: 272-288. Thilo C. Schadeberg - Ekkehard Wolff (eds.) Die Sprachen Afrikas. Hamburg: Helmut Buske.

Heine, Bernd 1981 Hess, R.L. (ed.) 1979 Proceedings of the fifth international conference on Ethiopian studies, Session B. Chicago: University of Illinois at Chicago Circle. Helck, Wolfgang - Wolfhart Westendorf (eds.) 1984 Lexikon der Ägyptologie. Volumes 3 and 5. Wiesbaden: Harrassowitz. Hetzron, Robert 1976 "The Agaw languages", Afroasiatic Linguistics 3 (3): 31-75.

408

Gabriele Sommer

Hoffmann, Carl 1971 "Provisional check list of Chadic languages", Chadic Newsletter. (Special Issue, January 1971.) Hofmann, Inge 1981 "Meroitisch", in: Heine - Schadeberg - Wolff (eds.), 301-304. Hofmann, Inge (ed.) 1981 Festschrift zum 60. Geburtstag von P. Anton Vorbichler, 1. Teil. (Beiträge zur Afrikanistik 11.) Wien: Afro-Pub. Jacquot, André - Irvine Richardson 1956 Report of the western team. Linguistic survey of the Northern Bantu borderland. Vol. 1. London: Oxford University Press. Johnston, Harry Hamilton 1919-1922 A comparative study of the Bantu and semi-Bantu languages. (2 vols.) Oxford: Clarendon Press. Junge, Friedrich 1984 "Sprache", in: Helck - Westendorf (eds.), Vol. 5: 1176-1211. Jungraithmayr, Herrmann - Henri Tourneux (eds.) 1990 Etudes Tchadiques. Verbes Monoradicaux. Paris: Librairie Orientaliste Paul Geuthner. Köhler, Oswin 1975 "Geschichte und Probleme der Gliederung der Sprachen Afrikas", in: Baumann (ed.), 141-374. Koelle, Sigismund Wilhelm 1854 Polyglotta Africana. London: Church Missionary House. Krause, Martin 1980 "Koptische Sprache", in: Helck - Westendorf (eds.), Vol. 3: 731-737. Kropp-Dakubu, Mary Esther 1976 "On the linguistic geography of the area of ancient Begho", in: Trutenau (ed.), 63-91. 1988 "Mande languages: Ligbi", in: Kropp Dakubu (eds.), 160-162. Kropp-Dakubu, Mary Esther (ed.) 1980 West African language data sheets, Vol. 2. Leiden: West African Language Society and Afrika Studiecentrum. 1988 The languages of Ghana. London: Kegan Paul International. Lacroix, Pierre-Francis 1962 "Note sur la langue Galke (™dáí)", Journal of African Languages 1 (2): 94-121. Leger, Rudolf 1983-1985 Unpublished field notes, collected with the support of the Deutscher Akademischer Austauschdienst. 1990 "Monoradical verbs in Kupto", in: Jungraithmayr - Tourneux (eds.), 15-18.

A survey on language death in Africa

1991

409

"Grammatische Analyse einer Herkunftserzählung der Kupto", FAB - Frankfurter Afrikanistische Blätter 3: 78-93. Lipsky, George A. 1962 Ethiopia. Its people its society its culture. New Haven: Hraf Press. [1967] [Reprinted.] MacDiarmid, P.A. - D.N. MacDiarmid 1931 "The languages of the Nuba Mountains", Sudan Notes and Records 14: 149-162. MacMichael, H.A. 1918 "Nubian elements in Darfur", Sudan Notes and Records 1: 30-48. 1920 "Darfur linguistics", Sudan Notes and Records 3: 197-216. Maguire, R.A.J. 1948 "Il-Torobo", Tanganyika Notes and Records 25: 1-27. Mann, Michael - David Dalby (eds.) 1987 A thesaurus of African languages. London - München - New York - Paris: Hans Zell and K.G. Saur. McCulloch, M. 1964 Peoples of Sierra Leone. (Ethnographic survey, western Africa 2.) London: International African Institute. Meek, Charles Kingsley 1931 Tribal studies in Northern Nigeria. Vol. 1. London: Kegan Paul, Trench, Trubner & Co. Merritt, E. Hollis 1975 A history of the Taita of Kenya to 1900. Ann Arbor, Michigan: University Microfilms International. Migeod, Frederick William Hugh 1926 A view of Sierra Leone. London: Kegan Paul, Trench, Trübner & Co. Mittwoch, Eugen 1907 "Proben aus dem amharischen Volksmund", Mitteilungen des Seminars für Orientalische Sprachen zu Berlin 10 (2): 185-241. Möhlig, Wilhelm J.G. - Franz Rottland - Bernd Heine (eds.) 1977 Zur Sprachgeschichte und Ethnohistorie in Afrika. (Neue Beiträge afrikanistischer Forschungen.) Berlin: Dietrich Reimer. Monod, Théodore (ed.) 1939 Contributions à l'étude du Sahara Occidental. Vol. 2. (Publications du comité d'études historiques et scientifiques de l'Afrique Occidentale Française, série Β, no. 5.) Paris: Larose. Monteil, Charles V. 1939 "La langue azer", in: Monod (ed.), 213-341. Newman, Paul 1977 "Chadic classification and reconstructions", Afroasiatic Linguistics 5 (1): 1-42.

410

Gabriele Sommer

Nurse, Derek 1982

"Segeju and Daisu: A case study of evidence from oral tradition and comparative linguistics", History in Africa 9: 175-208. 1984-1985 "A historical view of the Southern dialects of Swahili", Sprache und Geschichte in Afrika 6: 225-251. 1985 "Dentality, areal features, and phonological change in Northeastern Bantu", Studies in African Linguistics 16 (3): 243-279. 1986 "Reconstruction of Dahalo history through evidence from loanwords", Sprache und Geschichte in Afrika 7 (2): 267-305. Pasch, Helma 1986 Die Mba-Sprachen: die Nominalklassensysteme und die genetische Gliederung einer Gruppe von Ubangi-Sprachen. (Sprache und Geschichte in Afrika, Beihefte 6.) Hamburg: Helmut Buske. Persson, Andrew - Janet Persson 1980 "Ligbi", in: Kropp Dakubu (ed.), [no page numbers]. Platiel, Suzanne 1982 "Notes de recherche sur le Tyanga et le Shanga", Mandenkan 4: 71-93. Prost, André 1968 "La langue blé", Bulletin de l'Institut Fondamental d'Afrique Noire 30 (3): 1256-1270. Roeder, Hilke 1986 Sprachlicher Wandel und Gruppenbewußtsein bei den Terik. (Sprache und Geschichte in Afrika, Beihefte 7.) Hamburg: Helmut Buske. Rottland, Franz 1982 Die südnilotischen Sprachen. Beschreibung, Vergleichung und Rekonstruktion. (Kölner Beiträge zur Afrikanistik 7.) Berlin: Dietrich Reimer. Rottland, Franz - Duncan Okoth Okombo 1986 "The Suba of Kenya: A case of growing ethnicity with receding language competence", Afrikanistische Arbeitspapiere 7: 115-126. Rouchdy, Aleya 1989a "'Persistence' or 'tip' in Egyptian Nubian", in: Dorian (ed.), 91-102. 1989b "Urban and non-urban Egyptian Nubian: Is there a reduction in language skill?", in: Dorian (ed.), 259-266. Santandrea, Stefano 1948 "Little-known tribes of the Bahr el Ghazal Basin", Sudan Notes and Records 29: 78-106. 1950 "Gleanings in the Western Bahr el Ghazal", Sudan Notes and Records 31: 54-64. 1961 Comparative outline grammar of Ndogo-Sere-Tagbu-Bai-Bviri. (Museum Combonianum 13.) Bologna: Editrice Nigrizia.

A survey on language death in Africa 1966 1969

411

"The Birri language", Afrika und Übersee 49: 81-105, 196-234. Note grammaticali e lessicali sul gruppo Feroge e sul Mundu (Sudan). Napoli: Istituto Universitario Orientale. Sasse, Hans-Jürgen - Helmut Straube 1977 "Kultur und Sprache der Burji in Südäthiopien. Ein Abriß", in: Möhlig - Rottland - Heine (eds.), 239-266. Sebeok, Thomas A. (ed.) 1971 Linguistics in Sub-Saharan Africa. (Current Trends in Linguistics 7.) The Hague: Mouton. Seligman, Brenda Z. 1910-1911 "Note on the languages of the Nubas of Southern Kordofan", Zeitschrift für Kolonialsprachen 1: 167-188. Shimizu, Kiyoshi 1979 "Five wordlists with analyses from the Northern Jos group of Plateau languages", Afrika und Übersee 62 (4): 253-271. 1982a "Ten more wordlists with analyses from the Northern Jos group of Plateau languages", Afrika und Übersee 65 (1): 97-134. 1982b "Die Nord-Jos-Gruppe der Plateau-Sprachen Nigerias", Afrika und Übersee 65 (2): 161-210. Skinner, A. Neil 1974 "A note on the North Bauchi language group", Africana Marburgensia 7 (1): 47-50. Sobania, Neal 1988 "Fishermen herders: Subsistence, survival and cultural change in Northern Kenya", Journal of African History 29 (1): 41-56. Spencer, Paul 1973 Nomads in alliance. London: Oxford University Press. Stork, Lothar 1981 "Ägyptisch", in: Heine - Schadeberg - Wolff (eds.), 149-170. Strümpell, Kurt 1910 "Vergleichendes Wörterverzeichnis der Heidensprachen Adamauas", Zeitschrift ßr Ethnologie 42: 444-488. 1922-1923 "Wörterverzeichnis der Heidensprachen des Mandara-Gebirges (Adamaua)", Zeitschrift für Eingeborenen-Sprachen 13:47-74, 109149. Sumner, A.T. 1921 A handbook of the Sherbro language. London: Crown Agents for the Colonies. Thelwall, Robin 1977 "A Birgid vocabulary list and its links with Daju", in: Ganslmayr Jungraithmayr (eds.), 197-210. Tornay, Serge 1981 "The Omo-Murle enigma", in: Bender (ed.), 33-60.

412

Gabriele Sommer

Tourneux, Henry 1977

"Une langue Tchadique disparue: le Muskum", Africana gensia 10 (2): 13-34.

Marbur-

1980

"Les dialectes du musgu (Tchad-Cameroun)", in: Guarisma - Platiel (eds.), 229-246. Trutenau, H. Max J. (ed.) 1976

Languages of the Akan area: Papers in Western Kwa linguistics and on the linguistic circle of Accra, Vol. III. (Mitteilungen der Basler Afrika Bibliographien 14.) Basel: Basler Afrika Bibliographien.

Tucker, Archibald Norman 1967

"Fringe Cushitic. An experiment in typological comparison", Bulletin of the School of Oriental and African Studies 30 (3): 655-680.

Tucker, Archibald Norman - Margaret A. Bryan 1956

The non-Bantu languages of North-Eastern University Press.

Africa. London: Oxford

1966

Linguistic analyses. The non-Bantu languages of North-Eastern ca. London: Oxford University Press.

Afri-

Unseth, Pete 1984

Shabo (Mekeyir). A first discussion of classification and vocabulary. [Unpublished MS.]

Ushari Ahmed Mahmud 1974

The phonology of a dying Nubian language: Birgid. [Unpublished M.A. thesis, Institute of African and Asian Studies, University of Khartoum.]

Voegelin, Charles F. - F.M. Voegelin 1977

Classification and index of the world's languages. New York: Elsevier North Holland.

Voßen, Rainer 1984

1988

"Studying the linguistic and ethno-history of the Khoe-speaking (Central Khoisan) peoples of Botswana, research in progress", Botswana Notes and Records 16: 19-35. Patterns of language knowledge and language use in Ngamiland, Botswana. (Bayreuth African Studies Series 13.) Bayreuth: Bayreuth University.

Voßen, Rainer - Ulrike Claudi (eds.) 1983

Sprache, Geschichte und Kultur in Afrika. Vorträge, gehalten auf dem III. Afrikanistentag. Köln 14/15. Oktober 1982. Hamburg: Helmut Buske.

Walsh, Martin 1990

The Degere: Forgotten hunter-gatherers of the East African coast. [MS.]

A survey on language death in Africa

413

Weeks, R.V. (ed.) 1984 Muslim peoples: a world ethnographic survey. (2nd edition.) (2 Vols.) Westport, CT: Greenwood Press. Weimers, William Everett 1971 "Checklist of language and dialect names", in: Sebeok (ed.), 759900. Westermann, Diedrich - Margaret A. Bryan 1952 The languages of West Africa. London: Oxford University Press. Westphal, Ernst Oswald J. 1965 "Linguistic research in S.W.A. and Angola", in: S.W.A. Wissenschaftliche Gesellschaft (ed.), Die ethnischen Gruppen in Südwestafrika. (Wissenschaftliche Forschung in Südwestafrika, 3. Folge.) Windhoek: S.W.A. Wissenschaftliche Gesellschaft, 125-144. Whiteley, Wilfred Howell 1958 A short description of item categories in Iraqw (with material on Gorowa, Alagwa and Burunge). (East African Linguistic Studies 3.) Kampala: East African Institute of Social Research. Williamson, Kay (ed.) 1973 Benue-Congo comparative wordlist. Vol. 2. Ibadan, Nigeria: West African Linguistic Survey. University of Ibadan. Winter, Jürgen Christoph 1979 "Language shift among the Aasáx, a hunter-gatherer tribe in Tanzania", Sprache und Geschichte in Afrika 1: 175-204. 1981 "Die Khoisan-Familie", in: Heine - Schadeberg - Wolff (eds.), 329374. Wolff, Ekkehard 1974-1975 "Sprachwandel und Sprachwechsel in Nordostnigeria", Afrika und Übersee 58 (3-4): 187-212. 1979 "Sprachkontakt und Ethnizität", Sprache und Geschichte in Afrika 1: 143-173.

Appendix Addresses Bearth,

Thomas

Weinbergstr. 27, CH-8703 Erlenbach, Switzerland Bechhaus-Gerst,

Marianne

Afrikanistik II, Universität Bayreuth, Postfach 10 1 2 5 1 , D - 8 5 8 0 Bayreuth, Germany

414

Gabriele Sommer

Bender, M. Lionel Dept. of Anthropology, Southern Illinois University, Carbondale, Illinois 62901, U.S.A. Bendor-Samuel, John Summer Institute of Linguistics, 7500 W. Camp Wisdom Road, Dallas, Texas 75236, U.S.A. Blackburn, Roderic H. Kinderhook, N.Y. 12106, U.S.A. Blench, Roger 15, Willis Road, Cambridge, CB1 2AQ, England Boyd, Raymond CNRS-LACITO (LP 3121), 44, Rue de l'Amiral Mouchez, F-75014 Paris, France Brenzinger, Matthias Institut für Afrikanistik, Universität zu Köln, Meister-Ekkehart-Str. 7, D5000 Köln 41, Germany Broß, Michael Professur für Afrikanische Sprachwissenschaften, Praunheimer Landstr. 70, D-6000 Frankfurt-Hausen, Germany Cameron, Barbara Α. Lutheran Bible Translators, Inc., 303 N. Lake Street, Box 2050, Aurora, Illinois 60507-2050, U.S.A. Capo, Hounkpati B. C. Dept. of Linguistics and Nigerian Languages, University of llorín, P.M.B. 1515, llorín, Nigeria Cook, Tom L. Oude Varkenmarkt 32, NL-2311 VR Leiden, The Netherlands Demoz, Abraham Linguistics Department, Northwestern University, Evanston, Illinois 60208, U.S.A. Ehret, Christopher Dept. of History, U.C.L.A., Los Angeles, California 90024, U.S.A.

A survey on language death in Africa

415

Elderkin, Edward Derek S.O.A.S., Malet Street, London WC1E 7HP, England Gamst, Frederick Dept. of Anthropology, U/Massachusetts, Harbor Campus/Building 2, Boston, MA 02125, U.S.A. Gerhardt, Ludwig Seminar für Afrikanische Sprachen und Kulturen, Universität Hamburg, Rothenbaumchaussee 5, D-2000 Hamburg 13, Germany Hansford, Gillian 14 Branksome Ave., Shirley, Southampton, SOI 5NY, England Heine, Bernd Institut für Afrikanistik, Universität zu Köln, Meister-Ekkehart-Str. 7, D5000 Köln 41, Germany Hieda, Osamu 3-7-18, Mayumi, Ikoma, Nara, Japan Jakobi, Angelika Afrikanistik II, Universität Bayreuth, Postfach 10 12 51, D-8580 Bayreuth, Germany Kastenholz, Raimund Institut für Afrikanistik, Universität zu Köln, Meister-Ekkehart-Str. 7, D5000 Köln 41, Germany Leger, Rudolf Professur für Afrikanische Sprachwissenschaften, Praunheimer Landstr. 70, D-6000 Frankfurt-Hausen, Germany Lepionka, Larry 1405 Newcastle Street, Beaufort, South Carolina 29902, U.S.A. Mendel, Daniela Seminar für Ägyptologie, Universität zu Köln, Meister-Ekkehart-Str. 7, D5000 Köln 41, Germany Meyer-Bahlburg, Hilke Seminar für Afrikanische Sprachen und Kulturen, Rothenbaumchaussee 5, D-2000 Hamburg 13, Germany

416

Gabriele Sommer

Möhlig, Wilhelm J.G. Institut für Afrikanistik, Meister-Ekkehart-Str. 7, D-5000 Köln 41, Germany Mous, Maarten Department of African Linguistics, University of Leiden, P.O. Box 9515, NL-2300 RA Leiden, The Netherlands Moser, Rupert Haus der Universität, Universität Bern, Schlösslistr. 5, CH-3008 Bern, Switzerland Niyibizi, Schadrack Musée Royal de l'Afrique Centrale, Section Linguistique, Β-1980 Tervuren, Belgium Ntahomvukiye,

Hilaire

25a, rue d'Heuval, B-1490 Court-St. Etienne, Belgium Nurse, Derek Dept. of Linguistics, MUN, St. John's, Newfoundland A1C 587, Canada Platiel,

Suzanne

12, rue Antoine Chantin, F-75014 Paris, France Prof. Rongier Université du Benin, Linguistics Department, Lomé, Togo, c/o Rowe, Jenny, B.P. 1525, Lome, Togo Schadeberg, Thilo Afrikaanse Taalkunde, RUL, Postbus 9515, NL-2300 RA Leiden, The Netherlands Skinner, Neil 2215 Eton Ridge, Madison, Wisconsin 53705, U.S.A. Tablino, Fr. Paul P.O. Box 10, Marsabit, Kenya Thissen, H.-J. Seminar für Ägyptologie, Universität zu Köln, Meister-Ekkehart-Str. 7, 5000 Köln 41, Germany Unseth, Pete P.O. Box 6779, Addis Ababa, Ethiopia

A survey on language death in Africa

417

Voeltz, Erhard FLSH, Université de Bangui, Boîte Postale 1037, Bangui, Central African Republic Voßen, Rainer Afrikanistik II, Universität Bayreuth, Postfach 10 12 51, D-8580 Bayreuth, Germany Vydrin, V. Department of Oriental Studies, Petersburg University, Universitetskaya nabereznaja 7/9, 199034 Petersburg, Russia Walsh, Martin P.O. Box 99187, Mombasa, Kenya Wolff, Ekkehard Seminar für Afrikanische Sprachen und Kulturen, Universität Hamburg, Rothenbaumchaussee 5, D-2000 Hamburg 13, Germany

Names of scholars

Abdulaziz, Mohamed H., 165 Abubakar, A.M., 105 Adams, P. Gustaf Α., 381 Akida, Hamisi, 95 Ali, Mohammed Nuuh, 146, 152 Althaus, Gerhard, 297-298 Andersen, Roger W., 26, 133 Appel, René, 3 Ardener, Edwin, 318, 352-353, 361, 381-382 Austin, P., 127, 129, 131 Ayot, Henry Okello, 276-277, 279, 281 Azuma, Shoji 55

Bakari, Mohamed, 194 Batibo, Herman, 4, 83, 86, 90-91, 95, 108 Bavin, Edith L., 118 Bearth, Thomas, 413 Bechhaus-Gerst, Marianne, 372, 414 Bell, Herman, 351-352 Bender, M.Lionel, 307-308, 315-316, 341-342, 344-345, 347, 358-359, 363-364, 366-367, 414 Bendor-Samuel, John, 304, 312-313, 315, 329, 336-337, 346-347, 352, 354-355, 357, 360-361, 363ff„ 370371, 377, 381, 396, 401, 414 Beniak, Edouard, 44-45, 54 Bernstein, Janice, 46 Besha, Ruth M„ 101, 108 Bird, Charles Stephen, 356 Blackburn, Roderic Hall, 305, 309310, 324, 331-332, 414 Blanc, Michel H.A., 61

Blench, Roger, 312, 314-315, 352, 414 Bloomfield, Leonard, 130 Blount, Ben G„ 130 Bolinger, Dwight, 54 Bourhis, R.Y., 132 Boyd, Raymond, 316-317, 342, 414 Brenzinger, Matthias, 3, 12, 24, 31, 51, 83, 88, 213, 236, 250, 271, 301, 323, 340, 401, 414 Breu, Walter, 8 Broß, Michael, 375, 414 Bruce, James, 398-399 Bryan, Margaret Α., 103, 139, 145, 153, 285, 287, 303, 307, 336, 374375 Busse, J., 100 Butz, Carla, V Byarushengo, Ernest R., 110 Bybee, Joan L„ 125 Bynon, Theodora, 157, 179

Cameron, Barbara Α., 319-322, 414 Campbell, Lyle, 15, 22, 60, 62, 70, 119, 129 Capo, Hounkpati B.C., 414 Carlin, Eithne, 250, 271 Chamungwana, W.H. Shedu, 112 Chanler, William Astor, 219, 250 Chum, Haji, 209 Claudi, Ulrike, 132 Cohen, David W„ 276 Cook, Tom L„ 360, 414 Costermans, B.J., 357 Creissels, Denis, 355-356 Cronk, Lee, 215, 219-223, 225, 228, 250

420

Names of scholars

Curtin, Romero, 117

Dalby, David, 93, 303, 312, 332-333 Dammann, Ernst, 182 Darwin, Charles, 85 Daryll, Forde, 303 Dawkins, R.M., 31, 65-66 Decken, Carl Claus von der, 290 Demoz, Abraham, 309, 414 Denison, Norman, 32, 49, 69, 89 Dieu, Michel, 361 Dimmendaal, Gerrit, 3, 12, 24, 76, 83, 88, 90, 92, 95, 120, 125-1126, 13Iff., 223, 251, 310, 362, 366-367, 378, 384, 394-395, 398-399 Doornbos, Paul, 307-308, 315-316, 363-364 Dorian, Nancy C„ 4, 8, 12, 15, 25-26, 52-53, 54, 60, 64, 70, 77-78, 89, 118, 123, 130 Dressler, Wolfgang, 8, 9, 14-15, 25, 70, 77, 89, 92, 109, 112, 276 Dundas, K.R., 219, 250 Dyson, W.S., 238, 252

Earthy, E. Dora, 387 Ehret, Christopher, 92-93, 137, 140, 144-145, 152-153, 214, 306-307, 310, 331-332, 365-366, 392ff„ 415 Elderkin, Edward Derek, 140, 145, 148, 152-153, 307, 330ff„ 350-351, 366, 415 Emeneau, Murray B., 54 Esser, Monika, 302 Evans-Pritchard, Edward E., 282

Fergusen, Charles Α., 91 Fishman, Joshua Α., 10, 280, 282 Fivaz, Derek, 309, 312, 314-315, 318, 322, 328, 341, 350, 356, 363, 368, 370-371, 374, 376, 378, 380, 382383, 387, 392, 396, 398, 402

Fleming, Harold Crane, 310, 313-314, 350-351 Fuchs, Hans, 291-292 Fuchs, V.E., 238, 252 Gal, Susan, 8, 63 Galaty, John G„ 144 Galtier, Gérard, 356 Gamst, Frederick C., 73-74, 358-359, 398-399, 415 Gerhardt, Ludwig, 371, 374, 415 Giles, Howard, 3, 10, 132 Goodman, Morris, 31 Graham, Eric, 141 Greenberg, Joseph H„ 85, 214, 308, 388 Grimes, Barbara F., 303 Gross, E„ 95 Guarisma, Gladys, 157 Guenther, Mathias, 144 Gumperz, John J., 10 Guthrie, Malcom, 285, 303 Gutkind, A.E., 103 Hamers, Josiane F., 61 Hansford, Gillian, 415 Hansford, Keir, 304, 312-313, 315, 329, 336-337, 346-347, 352, 354355, 357, 360-361, 363ff„ 370-371, 377, 381, 396, 401 Haugen, Einar, 48 Hay ward, Dick, 76-77, 313-314 Heine, Bernd, 4, 12, 24, 27, 76-77, 83, 92, 117-120, 123-129, 132, 139140, 157-158, 213, 217ff„ 221, 224227, 229, 236, 238, 240-244, 246247, 250ff„ 258, 270-271, 301, 308309, 323-326, 340-341, 362, 369, 374, 377, 379-380, 388-392, 395, 400ff„ 415 Heine, Ingo, 236, 250 Hetzron, Robert, 399-400 Hieda, Osamu, 346, 415

Names of scholars Hill, C.P., 103, 106 Hill, Jane H„ 9, 14, 22, 52, 74 Hill, Kenneth C„ 52 Hinnebusch, Thomas, 209 Hobley, C.W., 214, 251 Höhnel, Ludwig von, 238, 250 Hoffmann, Carl, 365 Hoffmann, Inge, 373 Huntingford, G.W.B., 214, 258 Ingram, David, 13, 105 Itandala, Buluba, 89 Iverson, J„ 319-322 Jacobs, Alan, 219-220 Jakobi, Angelika, 339, 415 Johnston, Harry Hamilton, 290, 313 Johnston, Ruth, 280 Junge, Friedrich, 330 Karttunen, Frances, 54 Kashoki, Mubanga, E., 112 Kastenholz, Raimund, 311, 354, 367368, 415 Kaufman, Terrence, 4, 27, 31-32, 38, 49ff., 65, 131, 199-205 Kaye, K„ 13 Kenny, Michael G„ 276, 281-282 Kesisoglou, I.I., 65 Khamis, Said Ahmed Mohamed, 104 Khamisi, Abdu M„ 105ff. Kihore, Yared M„ 108 Kimambo, Isaria N„ 287-291 Kite, Yuriko, 54 Köhler, Oswin, 258, 324-325, 345, 367 Koelle, Sigismund Wilhelm, 322 Konstantinopulos, Christos G., 23 Kratz, Corinne Α., 32 Krause, Martin, 329-330 Kropp-Dakuba, Mary Esther, 369 Labov, William, 10

421

Lacroix, Pierre-Francis, 342 Lambert, H.E., 191-192, 194, 207ff„ 219 Leger, Rudolf, 364-365, 415 Légère, Karsten, 4, 83, 88, 95, 104, 112

Lepionka, Larry, 415 Lester, Arthur, 217 Lester, Barbara, 217 Lipsky, George Α., 309, 341-342, 344345 Loving, Ed, 152

MacDiarmid, D.N., 339, 358 MacDiarmid, P.A., 339, 358 Mackey, William F., 223, 251 MacMichael, H.A., 315-316 Maddieson, Ian, 145, 196 Maganga, Clement, 104 Maguire, R.A.J., 130, 214, 305-306 Mann, Michael, 93, 303 McCulloch, M„ 322 McKay, W.F., 182, 184, 207-208 Mecklenburg, Adolf Friedrich zu, 374 Mekacha, Rugatiri D.K., 108 Mendel, Daniela, 415 Menn, Lise, 23, 26, 61, 133 Menovsöikov, G.A., 32 Merritt, E. Hollis, 393-394 Meyer-Bahlburg, Hilke, 416 Midgeod, Frederick William Hugh, 322 Miehe, Gudrun, 165 Miller, Christopher, 44 Mittwoch, Eugen, 398-399 Mkude, Daniel, 108 Mochiwa, Zacharia S.M., 88, 112 Möhlig, Wilhelm J.G., 5, 157-158, 179, 192, 194-195, 205, 208-209, 293, 325-326, 340-341, 379-380, 390ff„ 416 Moffett, J.P., 103, 112 Mohan, Peggy, 32

422

Names of scholars

Molnos, Angela, 85 Monteil, Charles V., 311 Moser, Rupert, 93, 351, 416 Mougeon, Raymond 44-45, 54 Mous, Maarten, 92, 306-307, 326-327, 416 Mpaayei, John ole, 123, 132 Muntzel, Martha C„ 15, 22, 60, 62, 70, 119, 129 Murdock, George Peter, 214 Muysken, Pieter, 3, 32, 48 Mwaruka, Ramadhani, 88, 95, 112 Myers-Scotton, Carol, 5, 35ff„ 39, 40, 42, 44, 46, 54-55, 95, 109, 118, 151, 208 Neuman, Paul, 365 Neumann, A.H., 214, 219 Nichols, Johanna, 133 Nishimura, Miwa, 43 Niyibizi, Schadrack, 416 Ntahomvukiye, Hilaire, 416 Nurse, Derek, 5, 83, 92-93, 95, 145, 153, 182, 184, 189, 194-195, 197, 208-209, 271, 305-306, 309, 330ff., 369-370, 382-383, 392ff., 416 Nyerere, Julius, 93, 96 O'Barr, William M„ 287 Ogot, Bethwell Α., 276, 282 Ohannessian, Sirarpi, 112 Okeju, John, 44, 54 Okoth Okombo, Duncan, 3, 83, 92, 276, 280-281, 390ff. Otterloo, Roger van, 258 Palmer, Frank, 74 Pasch, Helma, 336 Persson, Andrew, 368-369 Persson, Janet, 368-369 Philippson, Gérard, 182, 184 Platiel, Suzanne, 384-385, 396-397, 416

Polomé, Edgar C„ 91, 95, 103, 106 Poplack, Shana, 44 Prost, André, 354-355 Pugh, Stefan N„ 44

Raa, Eric ten, 221, 251 Reh, Mechthild, 25, 281 Renaud, Patrick, 361 Rilling, Art, 140ff. Rindler-Schjerve, Rosita, 9, 15 Roeder, Hilke, 255, 259-260, 262-271, 395 Romaine, Suzanne, 48, 60-61 Rongier, Prof., 416 Rother, Paul, 292 Rottland, Franz, 3, 24, 83, 92, 250, 258-259, 269ff„ 275-276, 280-281, 305-306, 323, 359, 379, 389-392, 395 Rouchdy, Aleya, 337-338 Rubanza, Yusuf T., 112

Sankoff, David, 44 Santandrea, Stefano, 316-319, 392, 396 Sapir, Edward, 32, 54, 125 Sasse, Hans-Jürgen, 4, 8, 15, 17, 23, 26, 59-60, 63, 68-69, 109, 119, 140141, 273, 326 Schadeberg, Thilo C., 323, 416 Schanz, Johannes, 287 Scherrer, Carol, 239 Schlieben-Lange, Brigitte, 70 Schmidt, Annette, 9, 70, 118, 128, 130 Schmied, Doris, 281 Scott, Patricia E„ 309, 312, 314f, 318, 322, 328, 341, 350, 356, 363, 368, 370-371, 374, 376, 378, 380, 382383, 387, 392, 396, 398, 402 Seligman, Brenda Z„ 339-340, 358 Sengo, Tigiti S.Y., 95 Sharpe, H.B., 222

Names of scholars Shimizu, Kiyoshi, 329, 342ff„ 349350, 353-354, 364, 377, 381-382, 385-386, 394, 403 Singler, John, 54, 131 Skinner, A. Neil, 304, 416 Smith, Lars Christian, 306, 351 Sobania, Neal, 236, 239ff„ 251, 340341 Sokoine, E., 107 Sommer, Gabriele, 7, 85-86, 92-93, 95, 250 Spear, Thomas, 196 Spencer, Paul, 214, 217, 219-220, 238ff., 250, 369 Sperling, D„ 185, 187 Stanford, Ron, 304, 312-313, 315, 329, 336-337, 346-347, 352, 354355, 357, 360-361, 363ff„ 370-371, 377, 381, 396, 401 Stiles, Daniel, 143, 324 Stork, Lothar, 330 Straube, Helmut, 326 Strümpell, Kurt, 371, 374, 376 Sumner, A.T., 322 Tablino, Fr. Paul, 340, 416 Taylor, T.M., 132 Teniäev, E.R., 68 Thelwall, Robin, 316 Thissen, H.-J., 416 Thomason, Sarah Grey, 4, 27, 31-32, 38, 49ff„ 65, 109, 131, 199-205, 208 Tiersma, Peter Meyes, 132 Tornay, Serge, 378 Tosco, Mauro, 4, 24, 32, 48, 89, 95, 137, 144, 323-324, 330ff. Tourneux, Henry, 373-374 Trudgill, Peter, 15, 17 Tsitsipis, Lukas, 8, 14, 17, 26, 64, 77 Tucker, Archibald N„ 123, 132, 139, 145, 153, 285, 287, 305, 336, 350351

423

Unseth, Pete, 383-384, 417 Ury, William, 40

Voegelin, C.F., 63, 337, 339 Voegelin, F.M., 63, 337, 339 Voeltz, Erhard, 371, 417 Voßen, Rainer, (Vossen) 92, 117-120, 123-129, 132, 250, 334-335, 362, 365, 380, 401-402, 417 Vycichl, W„ 330 Vydrin, V., 355-356, 417

Walsh, Martin, 5, 83, 92, 95, 184, 191, 202, 207, 333-334, 417 Weinreich, Uriel, 26, 61 Weishaupt, Martin, 292 Weimers, William Everett, 307, 311, 322, 325, 327-328, 336, 338-341, 345, 350, 353f, 359-363, 367, 370, 372, 374ff„ 378, 380, 382-383, 387, 392, 394, 396-397 Westermann, Diedrich, 303 Westphal, Ernst Oswald J., 367 Whiteley, Wilfred H., 31, 103, 209, 388 Williamson, Kay, 360 Wilson, Thomas H., 186 Winter, Christoph J„ 24, 83, 92-93, 288, 309-310, 335, 345, 348-349, 365, 390 Wodak-Leodolter, Ruth, 9, 14 Wolff, Ekkehard, 328-329, 347-348, 417 Wolff, John, 32, 47, 55 Woodburn, James, 139, 145, 153 Worthy, G.G., 218-221, 226

Zaborski, Andrzej, 137, 140, 152 Zador, Paul, 32

Subject index

Acculturation, 73, 218, 223, 359 Age-set system, 262, 288 Assimilation language/linguistic Α., 48, 69-70, 76, 89, 101, 132, 137, 144, 147, 15 Iff., 203, 205, 276-277, 310, 324, 327, 383, 391-392, 395, 401 language/ linguistic unassimilation, 48, 109, 143, 147ff„ 152 Assimilation cultural Α., 100, 144, 204, 241, 260, 262-263, 265, 280, 310, 327, 334, 346, 351, 384, 391, 395 assimilating power, 88 Attitudes towards languages, 3-4, 10, 14-15, 17, 40, 62, 86, 267, 277ff„ 281, 310, 338 A. towards local languages, 108 Attitudes social Α., 24, 139, 141, 144, 262 Attitudes towards hunters, 217 Autonomy linguistic Α., 264 ethnic Α., 264

Bilingual(ism), (multilingual(ism)), 4-5, 17, 33ff„ 39, 45, 47ff„ 53ff„ 60-61, 63-64, 68, 75, 86, 88, 90-91, 103, 141, 148-149, 151, 185, 187188, 190, 203-204, 241, 277-278, 289, 293, 308-309, 317, 324, 331, 342, 344-345, 348, 353-354, 359360, 384-385, 391-392, 394, 400 bilingual setting (multilingual), 10, 12, 25, 63, 131, 401

bilingual speech community, (multilingual), 13ff„ 40, 50, 61, 109, 119, 146, 293, 298 Borrowing, 16-17, 26-27, 31-34, 38-39, 43ff., 50-51, 53, 60, 64-65, 68-69, 91, 109ff., 121, 144ff„ 151, 153, 162, 171-172, 189-190, 195, 200206, 223, 246, 252, 259, 293-294, 394 massive/ extensive B„ 49, 195, 208

Change, socio-cultural C., 260 Chronology, 287 Classification genetic (language) C., 137, 207, 255, 258, 380 Code-mixing, 91 Codeswitching, 5, 3Iff., 40, 70, 74, 90-91, 109ff„ 118 Communal decision, 293 Communication yield, 4, 224 Comparative method, (historical C.m.), 158 Competence, (language/ linguistic C.), 4, 18, 26, 34, 52, 59, 62-63, 69, 76, 86, 89, 91, 95, 117-118, 227, 267, 278-282, 302, 324, 338, 348, 391 Consciousness language C„ 202 ethnic/group C„ 94, 99, 101, 260, 263 Contact cultural C„ 60, 64, 131, 161, 177178, 185, 187, 195-196, 201-202, 204-205, 218, 224, 239, 241, 287, 312, 327, 340, 366

426

Subject index

economic (based) C„ 218, 241, 340 caravan trade C , 289, 297 Contact language C„ 15ff„ 22, 25, 31-32, 51, 53-54, 59, 60, 69, 74-75, 101, 168, 223, 327, 338, 353, 399 Contact-induced change, 5, 15, 26, 3839, 43, 54, 59, 65, 75, 200 Continuum, 62-63, 64, 78, 90, 99, 118, 271 Contraction, (language C.), V, 8, 68, 95, 118-119, 124., 127-128, 131, 310 gradual C„ 117 sudden C„ 117 Convergence, (language development, cf. divergence), 131, 197-198, 270 Creolization, 21, 25, 59, 131, 200 Cultural values, 263 Cultural/ethnic maintenance, 51, 94, 262

Death bottom-to-top D., 22-23 gradual D„ 22-23, 61 radical D„ 22-23, 119 sudden D. 22 Decline (of language), 8, 142, 280 Dialect cluster, 256 Dialectological method/analysis, 158, 162-163, 169 Dialectometry, 157, 160, 172 Diglossia, 90-91, 131 Divergence, (language development, cf. convergence), 270f. Domain, (social), 4, 10, 14-15, 45, 9091, 101, 105ff„ 130, 168, 278-279, 338, 341, 349 home language, 290, 296 work language, 296

Dominance political/economic/ cultural D., 176ff., 288-289, 348-349, 367368, 384 socio-political D., 50, D. pattern, 86 Dominant community, 21, 218 D. ethnic group, 100 D. local force, 184 D. political force, 186 sociolinguistically D., 49 Economic initiative, 295 Economic shift, (change), 87, 143 Ethnic affiliation, (group a.) 92, 101, 137, 177, 182, 184, 263, 280 Ethnic boundary, 263 Ethnic denomination, 137 Ethnic language, (85-96, 99-109, 112 Ethnic orientation, 260 Ethnic self-denomination, 139 Extra-linguistic forces, 269 Genealogical model, 255 Genetic affiliation, 160, 271 Genocide, 3, 22 Group cohesion, 262f. Hostility inter-ethnic H., 256, 262 Hunter-gatherer, 137, 213, 215, 217221, 223-224, 236, 239, 250, 296, 305, 324, 351, 367, 369, 379, 384 Identity ethnic I./ethnicity, 4, 14, 24-25, 27, 39ff„ 50-53, 64, 85, 89-90, 92, 94-95, 99, 101, 104, 111, 118, 176-177, 184, 187, 207ff„ 215, 239, 260, 262-263, 269, 275ff„ 279-280, 290-291, 297, 307, 334, 352, 367, 379 cultural I., 92, 144

Subject index Indirect rule, 292 Initiation, 263, 271 I. ceremony/rite, 91, 256, 263, 290, 296f. Innovation, (linguistic I.), 126, 190191, 195, 199 lexical I., 181 Intelligibility, 99, 258, 263, 271, 273, 275, 318-319, 378 unintelligble, 88, 242 Inter-ethnic communication, (conversation), 36, 41, 86, 251 Interference, 5, 14, 32, 38-39, 53, 60, 65, 69, 90, 128-129, 131, 133, 198, 204-205, 287 Islamization, 185, 188, 201

Kinship terminology, 293-296

Language acquisition incomplete L.a., 132 Language affiliation, 184, 186, 305 Language decay, 11, 15, 18, 24ff„ 27, 52-53, 59, 61, 64, 68, 73, 75-76, 77, 130 Language expansion, (language spread), 85, 87-88, 92, 99, 106, 112, 186, 199, 204, 262, 279, 282, 287, 333, 391, 401 Language loyalty, 3-4, 14, 33, 52, 111, 267 Language maintenance, 23, 32, 38, 43ff., 53, 73-74, 88, 94, 118, 241 Language obsolescence, 3, 16, 24-25, 59-60, 62-63, 73, 75, 77, 91, 130, 165 Language prestige, 86, 107, 278f., 282 L. of high social p., 176, 224, 280 L. of local p., 86ff. L. of low p., 109, 280 L. of national p., 86, 93, 95

427

L. of regional p., 86ff. L. of special p., 86 Language shift, V, 3, 13, 22, 31-32, 38, 43, 48-49, 5Iff., 83, 87, 89, 103, 107-108, 111-112, 119, 140, 143, 146, 181-184, 187-188, 195, 200, 203-208., 213, 215, 224, 240-241, 250, 269, 276ff„ 280-281, 289, 297, 309, 324, 333-334, 346, 366, 383, 392, 395 cause of L.s., 85-86, 89 motivation for L.s., 142, 309, 311, 326, 349, 352, 359, 368-369, 379, 384 primary L.s., 13, 18 processes of L.s., V, 33, 100, 223, 341, 391 dialect shift, 265ff., 269, 271 cultural shift, 269 Language spread see language expansion Language status, 275, 335-336, 349 Language stratification, 157, 167, 169, 174, 176f. Language use/usage, (patterns of L.u.), 3-4, 43-44, 86, 105, 107, 130, 165, 185, 206, 221, 223-224, 277ff., 302, 305, 317, 338, 344, 367, 372, 391 Language abandoned L., 4, 13, 18, 22, 63, 118, 227, 302 target L„ 4, 13, 18, 63-64, 73, 75, 111, 119, 133, 146, 181, 204f. Language dominant L, (1. dominance), 14, 18, 22, 34, 36, 41, 50, 73, 87-88, 109, 112, 119, 131, 143, 224, 281, 307, 310, 333, 338, 397 larger/powerful L., 85. local dominant L., 107 main L., 275 smaller/weaker L. see minority language

428

Subject index

Language first L., 100, 118, 184-185, 213, 228, 267, 278ff., 308, 326, 362, 367, 378-379, 389, 391-392, 399 second L„ 90, 41, 141, 251, 326, 333, 389, 391 primary L„ 4, 13-14, 18, 26, 119, 226, 241, 362, 399 secondary L„ 4, 13-14, 18, 26 Language National L„ 42, 47, 83, 86, 93, 96, 293, 309 official L„ 86, 141, 338, 372f. Language, reduced L., 125 Language, source L., 129 Language tribal/local L. see ethnic L. Levelling, 129 Lexical amnesia, 293 Lexicostatistics, 258, 270f. Lingua franca, 87-88, 93, 118, 221, 240, 251, 308, 311, 362, 401 Linguistic affiliation, 179 Linguistic classification, 256 Linguistic identity, 207, 267 Linguistic performance, 267f. Linguistic trace, 77, 92-93, 178, 181, 197ff., 202, 206, 209, 227, 270 cultural trace, 95 Local elite, 207

277, 322, 326, 327, 332, 334, 367, 379 Migration 21, 25, 260, 281, 324, 348, 353, 356, 365, 392 immigration, immigrants, 176, 178, 280, 288-289, 349 migration traditions, 275 Minority language, (smaller/ weaker language), 7, 14, 32, 85, 30If. Minority, (small community, small (ethnic) group), 3-4, 8, 10, 14, 24, 87, 89, 101, 107, 143, 185-186, 194, 196, 199, 201, 238, 256, 263, 308, 318, 391 Mission, 291-292, 297 Mixed language, "Mischsprache", 23, 152

Maintenance, see language maintenance, cultural/ ethnic maintenance Marginality, (social M.), 144 Marked (categories), 119, 121, 123, 126, 130-131 unmarked, 119, 123, 128, 130, 132, 206, 209 Marriages inter-ethnic M., (intermarriage), 83, 90, 118, 141, 143, 178, 218f„ 222, 224-225, 239, 252, 256, 276-

Oral history, 287-288, 307 Oral traditions, 195, 220, 260, 275, 328, 340, 393 Origin genetic O., 270 Outcast people, 143

Mixed origin, 208 Mixed speakers, 264ff„ 268 Modern education, 297 Monolingual community, 90 Monolingual(ism), 14, 27, 35, 78, 90, 141-142, 203, 226, 240, 277, 308, 317, 325-326, 391 Mother tongue, 53, 99, 104, 106-107, 119, 130-131, 140, 142, 213, 215, 219, 224, 238, 241, 251, 326, 388, 391 Multilingual(ism) see bilingual(ism) Myth of origin, 260

Paradigm levelling, 124 Petrification, 120, 124, 128, 130 Pidginization, 17, 25, 92, 117, 119, 121, 125, 130ff„ 200

Subject index Political history, 287 Population pressure see pressure Power, 65 cultural P., 204 economic P., 204 political P., 204 social P., 204 Pressure external P. (political P./social P./ economic P./cultural P.), 4, 10, 14, 21-22, 24, 38, 49, 101, 224 demographic P., (population P.), 8788, 260, 262, 269 P. from other languages, 109, 223, 275, 309 strong cultural p., 202f. Prestige, 65 (cf. language prestige) high P., 101 negative P., 395 neighbour's P., 21, 89, 101, 201, 204, 368. Principle of divergency, 270f. Psychological effects, 144 "puir'-type causes, 297 "push"-type causes, 296f. Reduction, 4, 15, 18, 20, 26, 52, 60, 68-69, 77, 85, 91-92, 117, 119-120, 125, 128, 130-131 Relation inter-ethnic R., 260, 262 Relexification, 27, 47-48, 53, 151 Replacement dialect/ language R„ 18, 20, 27, 49, 53, 119, 132, 143, 179, 190, 197, 200, 205, 232, 236, 240-241, 264, 269, 331-332, 340, 365, 389, 391, 397, 400 Revitalization, 3, 17, 20-21, 27, 59 Revival language R„ 280 Self-assessment, 4, 160, 267f.

429

Self-identification, 4, 99, 132, 182, 184, 207-208, 215 Shift in economic pursuits, 296 Simplification, 14-15, 18, 26, 62, 68, 91-92, 118-119, 200 Small community see minority Social organization, 262, 295 Sound shift, 259, 264 Speaker fluent S„ 52-53, 61-62, 77 formerly-fluent S„ 77-78 full S., 17, 21, 60, 62, 69-70, 73 imperfect S., 8, 18, 23, 25-26, 6Iff., 78 near-to-full-S., 78 nearly fully competent S., 62 non-fluent S„ 123, 128 non-full S„ 123 rusty S„ 23, 26, 61ff. semi-speaker S., 15, 17-18, 21, 26, 53-54, 60-61, 63-64, 69-78, 109, 117-122, 129-130, 141, 311 terminal/last S., 17-18, 26, 65, 77, 199, 346, 362 weak S., 62 Speech behaviour, (language/ linguistic behaviour), 4, 8, 10-14, 45, 89, 107, 118, 264, 266ff„ 273, 279, 280, 302 Speech community, 3-4, 8, 10, 14, 18, 21-25, 27, 60-61, 63-64, 73ff., 86, 91, 213, 224, 251, 335, 391 Spread see language expansion Status (cf. language status) (socio-)economic S., 44, 101, 217 ethnic S., 263 historical S„ 270, 290 minority S„ 269, 330 semi-official S., 263 social S„ 26, 139, 334, 367-368 Stereotypes, 224, 267, 271 Stigmatization, 4, 63, 132 Stratum, 157, 167-168, 174-179 Structural decay, 129

430

Subject index

Subordinate position, 201 Substratum/substrate, 23, 25, 31, 92, 131, 201, 229-230, 232, 241-242, 245-246, 328, 347, 349 S. influence, 5, 18, 74, 200, 204, (cf interference) Superiority, (cultural/military S.), 87, 201, 203, 224 Superstrate, 20If.

Threatened languages, 85, 90 Transition period of transition, 213, 220f. Transmission language T., 13ff„ 18, 21-22, 24, 59, 61-62, 69, 74, 278, 353 irregular (language) T., 200 (language) T. strategies, 13, 18, 63

Index of languages (and variants)

Aasáx see Asa Abakan see Kpan Abergelle Agaw see Xamtaqa Achmimic see Coptic Adegere see "Degere" Äkäyöh see Kiong ?áfo gúhoni see Dahalo Afoku Aasáx see Asa AFRICAN PIDGIN LANGUAGES; 129 (cf. Fanagalo; Swahili - Kenya Pidgin Swahili) Agara?iwa, , 304 ?agaíazyan see Geez Agaw Languages, 68 (cf. Cushitic) Aike see Ake Ajanci see Ajawa Ajanji see Tijanji Ajawa, , 304 Ajer see Azer Akayon see Kiong Ake, , 304 Akie, , 88, 92, 132, 258, 305f. (cf. Kalenjin dialect cluster) Akié:k pa kínáre see Kinare Akiék see Akie Akoiyang see Kiong Akpanzhi see Kpan Akye see Ake Al-Hajärat see El Hugeirat Alagmo see Alagwa Alagwa, , 92, 102, 306307 Albanian, "Tirana Albanian", 21 Aleut see Mednyj-Aleut Amampa see Bolem Amdang, , 307-308

Amharic, , 68, 73ff. "Amharization", 73 Amu, < K e n y a > , 147, 152 (cf. Swahili dialects) Amee lafofa see Kidie Lafofa Anafejanzi see Tijanji "Ancient Murle" see Omo-Murle Andangti see Amdang Anej see Gule Animere, < G h a n a > , 308-309 Anyimere see Animere Arabic, 69, 162, 171, 192, 205, 207, 209 Aramanik see Asa Arbore, , 76-77, 238 Argobba, , 309 Argobbinya see Argobba Argobbmña see Argobba Armenian, 67 Old Α., 67 Modern West Α., 67 Aromunian, < Greece > , 14 Aruro see Ggtgme Arusha, , 88, 92 Arvanitika, < G r e e c e > , 8, 11, 17, 21, 23, 26-27, 61-62, 68-72, 74-75, 77-78 "Arvanitism", 78 Asa, (Aasáx, Aramanik) , 88, 132, 214, 296, 309-310 Asax see Asa Asennize see Sheni Ashaganna see Agara?iwa Ashinginai see Agara?iwa Asu, (Middle, South Pare), 88, 285, 288-291, 293-297 (cf. Pare)

432

Index of languages (and variants)

Ateso, 44 Athu, (North Pare), 285, 290-293, 297 "Athu-ization", 293 (cf. Pare) Atumfuo see Ligbi Auna see Agara?iwa AUSTRALIAN LANGUAGES, 118, 127-128, 131, AUSTRALIAN LANGUAGES, EASTERN, 131 (cf. Kamilaraay, Dyirbal, Warlpiri) Aweera see Boni Azayr see Azer Azer, , 310-311 Azjer see Azer Azora see Izora Ayiek see Akie Äjäwa see Ajawa

B'oon see Boni Babeeke see Beeke Β ahi, , 89, 93 Baiso see Bayso Bako(l)le see Kole Bakoroka see Kwadi Bakole see Kole Balar see Kir Balendru see 'Bahema Bali see Hadza Balondo see Ehobe Belon Balto-Finnic dialects, 44 Balondo see Ehobe Belon Bama see Nagumi Bambala/Bambara see Burji Banen/Banen see Bonek Bangbinda see Ngbinda Banta, , 312 BANTU LANGUAGES; (NIGERCONGO, 256), 5, 23, 31, 38, 40, 49ff„ 103, 137, 145, 153, 179, 181, 256, 270, 273, 275 "Bantuness", 276 Kilimanjaro Bantu languages, 285 Proto Bantu, 189

Proto Eastern Bantu, 170 Proto North Eastern Coastal Bantu, 161 Proto-Sabaki, 193 Bapoko/Bapuku see Poko Baraawa, (ChiMiini), 147, 151, 153 (cf. Swahili dialects) Bari see Nimbad Basa Kontagora, , 312-313 Basari du Bandem(b)a see Bedik Basque, < S p a i n > , 21 Bassa (Kontagora) see Basa Kontagora Bateti see Deti Bati, < Cameroon > , 313 Ba Ye(e)i/Ba Yeyi see Yei Baygo see Beygo Bayso, , 313-314 Bedik, < Senegal > , 314 Beeke, < Z a i r e > , 314 Beko see Beygo Bembala see Burji Bena, , 90 Bende, , 102-103 Berti, < S u d a n > , 314-315 Bete, , 315 Beta Israel see Kimanteney Beygo, < S u d a n > , 315-316 BeDik see Bedik Bilin, 74 Biltine see Amdang Bimbia see Isuwu Birgid, < Sudan > , 316 Bir(r)i, cCentral African R e p u b l i o , 316-317 Birguid/Birkit/Birqed see Birgid Bisha see Taita Cushitic Bissaula see Kpan "Black Jews" see Kimanteney Ble see Jelkuna and Ligbi Bobe, , 318 Bobea see Bobe Bodo, < S u d a n > , 318-319

Index of languages (and variants) "Boesjesman" see Southern Khoisan Bohairic see Coptic Bolem, , 319-322 Bolom/Bolon/Bom/Bome/Bomo see Bolem Bon see Dahalo Bondei, , 88, 91, 104, 108-112, 285 Bonek, , 322-323 Bong'om, , 258ff„ 323 (cf. Kalenjin dialect cluster) Bong'omek see Bong'om Boni (Aweera), , 89, 137-144, 145-146, 148, 152, 323-324 Bonek see Bonek Boro, , 324-325 "Boshof" see Southern Khoisan Bota/Boy see Bodo Breton, , 8, 70, 74, 77 (Bu)Nghinda see Ngbinda Bubia see Bobe qBug see Buga Buga, < Sudan >, 325 qBugwa/Buk(w)a see Buga Bul(l)om/Bullom/Bullun see Bolem Bulu see Seki Bulom/Bum see Bolem Bungu/Wungu, 102 Burji, , 325-326 Burjinya see Burji Burunge, , 102-103, 326327 Buruqa iso/Buruqge see Burunge "Bushman" see Southern Khoisan Buso, , 327 Busso see Buso Buta(a)(wa(a)) see Gamo Buta-Ningi see Gamo and Ningi Butancii/Butu(-Ningi) see Gamo Buy, , 327328 Bàgyemii see Gyem

433

Bàrùshii see Rishi Bàshàyii see Shau Bàtàurii see Taura Bèlón see Ehobe Belon Bégogé/Bégonga/Bégukung see Beygo Bùùmbutuu see Gamo B~og digään see Ndam of Dik Bodo see Bodo Cacaopera , 22 Cape Hottentot see Xiri Cassanga/Cassangue see Haal CAUCASIAN LANGUAGES, 67 Celle see Ligbi CELTIC LANGUAGES, 25 (cf. East Sutherland Gaelic) CENTRAL AMERICAN LANGUAGES, 132 (cf. Otomi) Ceηga see Shanga Cena, , 328-329 Chaga-Taita see Taita Cushitic Chagga, (Kichagga, Chaga), , 87-88, 92, 108, 285, 287-293; Chagga dialects: Wunjo, 292 Chamo, , 329 Chamus, (Camus), , 119121, 127, 132, 243-244 (cf. Maa) Chemant see Kimanteney Chifundi, (Chirazi), , 83, 88, 181-212 (cf. Swahili dialects) Chikide see Cena ChiMiini, (Baraawa Swahili), 153 Chinese Pidgin Russian, 131 Chinese, 67-68 Chinine see Cena Chirazi (Chifundi), , 157, 160, 163, 168-175, 177ff. (cf. Swahili dialects) Chishingyini/Chisingini see Agara?iwa Chivumba, (Vumba), , 83, 88, 95, 160, 163-167, 169, 171177, 179, 181-212

434

Index of languages (and variants)

Archaic Vumba, 175-176 (cf. Swahili dialects) Chokobo see Izora Chwaka, 83, 157, 160-168, 170-178 Archaic Chwaka, 178 (cf. Swahili dialects) Cikobu see Izora Cishingyini see Agara?iwa CiYei see Yei Cobiana see Buy Cokobaawaa/Cokobancii see Izora Comorian, 193ff. Copi see ShiChopi Coptic, , 22, 329-330 Cuanhoca/Cuepe see Kwadi CUSHITIC LANGUAGES, 23, 31-32, 49ff„ 76-77, 130, 137, 145-149, 238, 242-243 EASTERN CUSHITIC LANGUAGES, 137, 153, 213, 238, 249 Galaboid - Western group of Omo-Tana branch of E.C., 238 SOUTHERN CUSHITIC LANGUAGES, 103, 137, 145, 153, 270 Proto-South-Cushitic, 145, 152153 CENTRAL CUSHITIC, 68 (cf. Agaw Languages) Cwaga see "Gemsbok Nama" Cora see Kora "Cira'ca Agäw see Xamtaqa cfaafe see Burji Dá(á)ko/D'aháálo see Dahalo Dabida, , 285 Dahalo, , 4, 24, 32, 48-49, 83, 89, 137-155, 330ff. Daisu, (Central Kenya Bantu), 181182, 184, 187-190, 194, 197ff„ 208 Coastal D„ 199 (cf. Segeju) Dako see Dahalo

Dama, , 332-333 Dasenech, (Reshiat, Galab, Marille, Shangilla) , 236, 238, 251 Dawida see Taita Cushitic "Degere", , 333334 Dehes see Elmolo Dendereko, , 89, 91 Denk see Dahalo Deti, , 334-335 Deti(-)Khwe see Deti Di 'gë see Ndam of Dik Digo, < Kenya/Tanzania>, 88-89, 9192, 104, 160, 163-167, 169, 172173, 181-212 (cf. Mijikenda) Southern Digo, 197 Archaic Digo, 177 Pre-Digo, 195) Dimboi7 see Mboq Dingi see Dungi Diés see Ega Doe, , 89, 92, 102, 104 Donga see Dongo Ko Dongo Ko, , 335-336 do'qgo' nyo' see Dongo Ko Dongolawi see Egyptian Nubian Dòò + gamo see Gamo Dravidian, 54 Duala see Kole Dulbu, , 336 Dungi, , 336-337 Dunjawa see Dungi Duruma, , 160, 163-164, 169, 171-172, 177, 196 (cf. Mijikenda) Dwingi see Dungi Dyala (Nu) see Jelkuna Dyirbal , 9-10, 11, 70, 77, 118, 128, 128, 131

Index of languages (and variants) East Sutherland Gaelic (Scottish Gaelic), 8, 11, 25-26, 52, 70, 7778, 118 Ega, , 337 Egwa see Ega Egyptian Nubian, , 337-338 Egyptian, 7 Ehobe Belon, , 338-339 El Hagarat see El Hugeirat El Hugeirat, , 339 El Hujerat see El Hugeirat Elgon branch, (E. dialects), , 258-259, 267, 269-270 (cf. Kalenjin dialect cluster) "Elgon speakers", 265-266, 268 Eliri, , 339-340 Elmolo, , 3, 12, 24, 27, 7677, 83, 132, 213, 236-252, 340-341 Elmolo-Samburu, , 213, 236252, 340 Elwana, , 140, 145, 193-194 English, 27, 36-37, 40-46, 48, 55, 86, 95, 106-107, 131, 141, 153, 233ff., 271 Eregba see Kpan Etruscan, 7 EUROPEAN MINORITY LANGUAGES, 8-9 E-kadamá-it/E-tepes-ít see So Fadicca see Egyptian Nubian Fälasa/Falasha see Kimanteney Fanagalo, , 129, 132 Fayumic see Coptic Fecakomodiyo see Gule Felasha see Kimanteney Feroge see Buga Fingo see Old Mfengu Fipa, , 90 French, 25, 44-45 "Fula-Bambara" see Kakolo Fumu, , 341

435

Fungi see Gule Funzi, , 191-192, 197, 208 (cf. Swahili dialects)

Gaelic, 74, 77, 123, 130 (cf. East Sutherland Gaelic) Gafat, , 341-342 Gafatinya see Gafat Galke, < Cameroon >, 342 Gamo, , 342-343 Gana, , 343-344 Ganjawle see Ganjule Ganjule, , 344 Ganzibati see Kazibati Gat(s)ama see Gstame Ge'ez see Geez Geez, , 344-345 Gema(wa) see Gyem "Gemsbok Nama", < South Africa>, 345 German, 27 German dialect of Sauris , 69 Gatame, , 345 Gibtith see Coptic Giiz see Geez Giryama, , 143, 153, 193 (cf. Mijikenda) Gogo, , 87-88, 90, 92 Gomba, , 346 Gorowa, , 102-103 Gothic, 7 Greek, (Standard Modern Greek), 25, 66, 68ff., 72 Asia Minor Greek, 31, 65 Cappadocian varieties of A.M.G., 65 Ulaghatch, 65ff. Grikwa/Griqua/Gry see Xiri Guanche, , 346 Guba(wa) see Gubi Gubi, , 346-347 Gule, < Sudan >, 347

436

Index of languages (and variants)

Gura pawa see Elmolo Gure-Kahugu see Gamo Guru see Gubi Guho gwi'ts:o/Guho gárima.ni see Dahalo Gwara, , 347-348 Gweno, , 24, 83, 285-298, 348-349 "Gweno-ized", 290 Gyem, , 349-350 Gyema(wa)/Gyemancii see Gyem GI-?I-Z see Geez GAT'ame see Gatame Ha, , 90 Haal, , 350 Hadjibatie see Kazibati Hadza, (Hadzapi, Tindiga, Kindiga), , 89, 93, 102, 107, 350-351 Hadzabi/Hadzapi see Hadza Hagarat see El Hugeirat Hai//?om see "Gemsbok Nama" Hamba, , 93, 351 Hameg/Hamej see Gule Hangaza, , 87, 90 Haraza, (Haräza), , 351-352 Haruro see Gatame Hatsa see Hadza Haya, , 87-88, 90, 96, 104, 107 Hebrew, {cf. Ivrith), 17 Hehe, , 87, 90 Hereny see Elmolo Hittite, 7 Holma, , 352 Homa, , 352 Hungarian in Austria, 8 Hwara see Kimanteney Hwaso/Hwaye see Kpan I'-gyem see Gyem Γ-si' see Rishi

I(-)Ha(d)ja/I(-)Haje see Haal Ibeeke see Beeke Iberian, 7 Ibukwo see Kpan Ifumu see Fumu Ikizu, , 102-103 Ikoma, , 102 Ikpan see Kpan INDIC LANGUAGES, 54 INDOEUROPEAN LANGUAGES, 66 Indonesian, 32, 47, 55 Indri see Buga Iraqw, , 107 Irenge see Tenet Isanzu, , 90, 102 Issenye, , 102 Isu(bu) see Isuwu Isuwu, , 352-353 Italo-Albanian, 8, 21 Ivrit, (Modern Hebrew), 17, 22 Izora, , 353-354 Janji see Tijanji Japanese, 43, 55 Javanese, 32, 47, 55 Peranakan J., 47-48, 55 Pribumi (native) J., 47 Jebel el Amira see Kidie Lafofa Jebel Haraza see Haraza Jebel Tekeim see Kidie Lafofa Jelkuna, < Burkina Faso>, 354-355 Jelü see Ligbi Jenji see Tijanji Je ri see Ligbi Jewish, 251 Jiji, , 91, 102 Jimi, , 355 Jita, , 88 Jomvu, , 191 {cf. Swahili dialects) Ju, , 355 Jungraithmayr's Mimi see Amdang Juwaano see Dahalo

Index of languages (and variants) Jog o see Ligbi

k/?egwi see Southern Khoisan Kachama see Gstame Kadam see So Kafu see Bolem Kagarügei see Birgid Kagulu, , 91 Kahe, , 92, 285 Kajargei/Kajjara see Birgid Kakolo, < M a l i > , 355-356 Kakolo Qaip see Kakolo Kalenjin (dialect cluster), 218, 220, 250, 255, 258ff., 262-265, 269ff. (cf. Nandi-Markweta branch: Nandi, Kipsiikis, Keyo, Tuken, Marakwet; Elgon branch: Sapiny, Kony, Bong'om, Terik, Pok; Okiek: Kinare, Sogoo, Akie: Päkot) Proto-Kalenjin, 259 Kamant see Kimanteney Kamba, < K e n y a > , 89, 181, 250 Kami, , 91 Kamilaraay, , 127, 131 Kande, < G a b o n > , 356 Kangeju see Hadza Kaokofeld- "Strandläufer" see "Gemsbok Nama" Kara see Kimanteney Karanga, , 102 Karre dialect of Somali, 146 Kas(s)anga see Haal Kasiboti see Kazibati Kassangue see Haal "Katkop" see Southern Khoisan Kayla see Kimanteney Kazibati, (Kaziß-ti), < Z a i r e > , 356357 Kadam see So Kedi see "Gemsbok Nama" Kemant, , 23, 73, 77 Kenga see Shanga Kenuz see Egyptian Nubian

437

Kerewe, , 88 Keyo, < K e n y a > , 256, 263 (cf. Kalenjin dialect cluster) Këmant(nây) see Kimanteney "Khakhea" see Southern Khoisan KHOISAN LANGUAGE FAMILY, 3 KHOISAN LANGUAGES, 137 Ki/hasi see Southern Khoisan Kiballo, , 357 Kitfie L a f o f a , < S u d a n > , 357-358 Kiko, , 103 Kikuyu, (Gikuyu), < K e n y a > , 181, 233, 250 Kim/Kimi see Bolem Kimanta see Kimanteney Kimanteney, (Quara), , 358-359 Kimantinya see Kimanteney Kimbu, , 88, 91, 102 Kinare, < K e n y a > , 258, 359 (cf. Kalenjin dialect cluster) Kindiga see Hadza Kinugu/Kinuka see Kinuku K i n u k u , , 359-260 Kiong, , 360 Kipsiikis, (Kipsigis), < K e n y a > , 256, 258, 269-270 (cf. Kalenjin dialect cluster) Kir, , 360-361 Kir-Balar see Kir Kirim see Bolem Kirr see Kir Kisii, , 102 Kittim see Bolem Kiwollo see Kiballo Koba see Yei Kobiana see Buy Kole, < Cameroon > , 361 Konongo, , 88 Kony, < K e n y a > , 258, 263, 270 (cf. Kalenjin dialect cluster) Kooki, < U g a n d a > , 361

438

Index of languages (and variants)

Kora, , 361-362 Koran/Koran(n)a/Koraqua see Kora Kore, (Nkore), < K e n y a > , 3, 24, 83, 88, 117-135, 244, 362 (cf. Maa) Kornea see Kwadi Koyra see Ctetame Kpala/Kpara see Kreish language group Kpan (i.e. Bissaula and Eregba), , 362-363 Kpanten/Kpanzon see Kpan Kpati, , 363 Kpwate see Kpan Kredj-/Kreich see Kreish language group Kreish language group, < S u d a n > , 363-364 Krekonika, , 23, 27 Kresh see Kreish language group Krim see Bolem Kuba see CiYei Kubto see Kupto Kuda-Chamo see Kudu and Chamo Kudawa see Kudu Kudu, , 364 Kuma, , 285 Kupto, , 364-365 Kuria, (Kurya), , 88, 90 Kutu, , 91, 104 Kw'adza see Kwadza Kwaa-Khoe see Deti Kwadi, < A n g o l a > , 365 Kwadza, , 88, 92, 365366 kwadza see Kwadza ?Khuai see Southern Khoisan Kwandu see Kwisi Kwara(sa) see Kimanteney Kwasi, , 88 Kwaya, , 91 Kwegu, , 366-367

Κ were, , 89, 92, 104 Kwisi, < A n g o l a > , 367 Kyenga see Shanga Kàgóro (Kan)/Kàkólo see Kakolo Κίάη see Kiong Kùttò see Kupto Krmanta/Kimanteney see Kimanteney Kòlé see Kole

Langi, , 91-92 Latin, 25 Ldes see Elmolo Lega, , 89 Lenca, , 22 Lenge see ShiChopi Li(-)Ngbe(e) see Ngbee Li-Shau see Shau Ligbi, , 367ff. LiSinga see Singa Logoli (Maragoli), < K e n y a > , 256, 262 (cf. Luyia) Lorkoti, (Lorokoti) < K e n y a > , 369 Lufu, , 369 Luganda, < U g a n d a > , 44 Lugha ya zamani, , 369370 Lumboq see Μόοη Luo, (Dholuo), , 87-88, 92, 130, 256, 273-283 "Luoization", 275-278 Luri, , 370 LuSinga see Singa Luyia, (Luhya), < K e n y a > , 256, 262f., 269, 271 Luyia sub-groups: Logoli (Maragoli), 256, 262; Tiriki, 256, 262, 271 "Luyia-ization", 264 Lwidakho ("Luyhia variety"), 45-46 (La-)Κα(α)Ιοη/(La)Mboq see Mboq

M-Dako/M-Kore/M-Sanye

see Dahalo

Index of languages (and variants) Maa Language; 32, 119-126, 132, 213, 219-221, 223-224, 230, 249ff. (cf. Kore, NILOTIC, Maasai) South Maa - Maasai, 119, 221, North Maa - Samburu, Chamus, 119, 221, 241, 244 Proto-Maa, 231-232, 244-245 Maasai, , 27, 87-88, 92, 117, 119-127, 129, 132-133, 214, 218, 221 (cf. Maa; Mukogodo-Maasai) Machinga, , 88, 102 Mahamba see Hamba Mahasi see Egyptian Nubian Makoba see Yei Makonde, , 87-88 Makoroko see Kwadi Makua, , 91 Makumu see Mampoko Maltese, 153 "Maluti" see Southern Khoisan Mambwe, , 102 Mampa see Bolem Mampoko, , 370-371 Mampwa/Mandenyi/Mandingi see Bolem Mangas, , 371 Mangaya(t) see Buga Mangbele see Ngbee Manx (Gaelic), 17 Marakwet, (Markweta), , 258 (cf. Kalenjin dialect cluster) Marghi West see Putai "Masarwa" see Southern Khoisan Masiin see Azer Mateti see Deti Matoki see Egyptian Nubian Matumbi, , 87, 91, 104 Mawuchi see Agara?iwa Ma?a, (Mbugu), 21, 23, 26, 31, 38, 49ff„ 270-271, (Maa, Mbughu, sic!, 101.) Mbama see Nagumi

439

Mbo see Ehobe Belon Mboa, < Cameroon >, 371 Mbondo, , 371 Mboip see Mboa Mbundyu see Kwisi Mbunga, , 102-103 Mbungwe, , 92, 95, 102103 Mbaij, , 372 Media Lengua, (Quechua/ Spanish), , 27, 48 Mednyj-Aleut, (Aleut/ Russian), , 27, 31, 5051 Mekeyer/Mekeyir see Shabo Menomini, (American Indian), 130 Meroitic, , 372-373 Meru, , 181, 219 Mexicano (Modern Nahuatl), , 9, 52, 74, 77 Malinche Mexicano, 52 Mfumu see Fumu Michif, (Cree/ French), (American Indian), , 27 Middle Egyptian see Coptic Mijikenda; 160, 163-179, 189-190, 192ff„ 196 (cf. Duruma, Digo, Giryama, Rabai) Southern M„ 164, 166-170, 171172, 174, 177-179, 181 M. dialects/ M. languages/ M. language group, 162, 190, 198 Proto Mijikenda, 162, 171 Mikair see Shabo Mima/Mime/Mimi see Amdang Mittu see Wetu Mlon see Ehobe Belon Mmani/Mnani see Bolem Mobele see Mampoko Modern Greek, 17 Mogogodo/Mokogodo see Yaaku and Mukogodo-Maasai Mongaiyat see Buga

440

Index of languages (and variants)

Mongoba, , 373 Morokodo see Wetu Mouskoun see Muskum Mtanga'ata, , 192 (cf. Lugha ya zamani and Swahili dialect) Mucoroca see Kwadi MuDahalo see Dahalo Mugu:ji 'apo/Muguji see Kwegu Mukogodo-Maasai, , 213 (cf. Yaaku) Murgi see Birgid Musgu see Muskum Muskum, , 373-374 Muqgußä ' see Mongoba Mwani, , 194 Mwera, , 88, 93 Mósiro see Akie Moqgoba see Mongoba

Nagumi, , 374 Nahuatl see Mexicano Naka see Poko Nandi, , 214, 218-219, 255271 "Nandi-based", 214 "Nandi-ization", 264-268, 271 Nandi-Markweta branch, 258, 270 (cf. Kalenjin) Napak see So Napore, , 374 Nata, , 102 NATIVE AMERICAN, 10 Νda zòrà see Izora Ndali, , 88 Ndam of Dik, , 375 Ndema Sherbro see Bolem Ndendeule, , 93 Ndengereko, , 104 Ndonde, , 92, 102 Ndwera see Ligbi 11 dai see Galke Ng'oma see Bong'om Ng'omvia see Kwadza

Ngbee, , 375-376 Ngbinda, , 376 Ngimoile see Elmolo Ngindo, , 91, 93 Νgiri see Mampoko Ngoni, , 87, 108 Ngulu, , 89, 91 Ngurimi, , 102 Ngwxi see Agara?iwa pidi see Kwegu ipma see Bong'om Niamniam see Nimbad NILOTIC; (NILO-SAHARAN FAMILY, 255); 120, 125-128 PROTO-NILOTIC, 133 EASTERN NILOTIC, 120, 122, 125-126, 128, 213, 238, 249 Bari group, 126, 128 Non-Bari branch, 120 Maa cluster, 119-120 SOUTHERN NILOTIC, 126-127, 251, 255, 258 WESTERN NILOTIC, 273 Nilyamba, , 88, 90 Nimbari, , 376 Ningi, , 377 Niporen/Nipori see Napore Northern Bul(l)om see Bolem Norwegian, 48 Ntoore see Ligbi Nu:ma guho:ni see Dahalo Numu see Ligbi Nyahoza, , 103 Nyakyusa, , 87f„ 90, 100, 104 Nyamba see Taita Cushitic Nyambo, , 88 Nyamnyam see Nimbari Nyamwezi, , 87ff„ 93 Nyang'i, , 377 Nyang'ori see Terik Nyangeya/Nyangiya see Nyang'i Nyatso see Kpan

Index of languages (and variants) Nyag'i Nyiha, 100 Nyipori Nyonyo

see Nyang'i (Niha), , 88, 91, see Napore see Kpan

O'qg'ámóni see Ongamo O-laamoot-i see Omotik Occitan, 70 Okande see Kande "Okiek of Kinare" see Kinare Okiek, , 32, 258 (cf. Akie, Kalenjin dialect cluster) Oko(n)yong/okoyoiyokoq3q see Kiong Okuma, , 92 Old Mfengu, < Southern Africa>, 377-378 Olukooki see Kooki Omo-Murle, , 378 Omotik, , 132, 379 Ong'om see Bong'om oyk see Bedik Ongamo, , 88, 380 Oromo, , 137, 148, 153 (cf Waata) otaqg/atomb/otomp see Bonek Otomi. cCentral America>, 132 "Oudtshoorn" see Southern Khoisan Pangwa, , 91-92 Pare, , 50, 91-92, 285, 287, 290 ( c f . Asu -Middle, South Pare; Athu - North Pare) Pate, , 160, 209 (cf. Swahili dialects) Peninsula Sherbro see Bolem Pidgin, 131 (cf. Chinese Pidgin Russian; AFRICAN PIDGIN LANGUAGES) Pimbwe, , 102-103 Pogolo, , 87, 91 Pok, , 258ff„ 270 (cf. Kalenjin dialect cluster)

441

Poko, , 380-381 Pokomo, , 140, 143, 145, 169, 193-194, 207 Upper (Northern) Pokomo, 179, 195 Lower Pokomo, 153 Pokot, (Päkot), , 258, 270 (cf. Kalenjin dialect cluster) Polish in Australia, 280 Pone(c)k see Bonek Poren see Napore Pormi see Galke "Prieska" see Southern Khoisan Proto-Bantu, 189 Proto-Eastern-Bantu, 170 Proto-North-Eastern-Coastal- Bantu, 161 Proto-Sabaki, 193 Proto-South-Cushitic, 145, 152-153 Proto-Kalenjin, 259 Proto-Maa, 231-232, 244-245 Proto-Mijikenda, 162, 171 Proto-Nilotic, 133 Proto-Suba, 275 Proto-Swahili, 171 Puku see Poko Putai, , 381 Ρ3:ηόιη/Ρó:ηόηιêk see Bong'om Qemant( iya )/Qement/Quara/Quarinya see Kimanteney Quechua, 32, 48 Qwadza see Kwadza Qwara/Qwera see Kimanteney Rabai, , 196 (cf. Mijikenda) Rendille, , 118, 236, 238 Rimi, , 251 Rishi, , 381-382 Romani dialects, (English/ Armenian/ Spanish), 27 English Romani, 31 Rufiji, , 89, 92, 104

442

Index of languages (and variants)

Rumbuk Jur see Wetu Rungu, (Lungu), , 102 Rungwa, , 102 Rusancii see Rishi Russian, 32, 44, 49-50 Rwo, , 285

Safwa, , 88, 91 Sagala, (Saghala), , 285 (cf. Taita Cushitic) Sahidic see Coptic Salar, (Turkic language), 67 Salka see AgaraPiwa Samburu, , 24, 118-122, 127, 132, 214, 217, 221, 227, 231234, 236-243, 245 (cf. Maa) Sandawe, , 221 Sangu, , 91 Sanye, , 139-140 (cf. Boni and Dahalo) Sapiny, , 258, 269 (cf. Kalenjin dialect cluster) Sar(o)ua see Sarwa Sarwa, , 382 ß-Thonga/ß-Tfopi/ßcopi see ShiChopi fèqgé/Sebora see Bolem Segeju, , 88, 92, 102, 181-212, 382-383 Seki, , 383 Sekiyana/Sekiyani see Seki Serbro see Bolem Shabo, , 383-384 Shaini see Sheni Shako see Shabo Shambala, (Sambaa, Kisambaa), , 88, 90, 100, 285, 288, 291 Shanga, , 384-385 Shanga(wa) see Shanga Shani see Sheni Shashi, , 88, 102-103 Shau, , 385-386

Shawancii see Shau Sheenancii/Sheeni see Sheni Sheke/Shekiyana see Seki Shenge (Sherbro) see Bolem Sheni, , 386 Sherbro(o) see Bolem Shi, , 89 Shiba see Bolem ShiChopi (Lenge), < Mozambique >, 386-387 Shicopi see ShiChopi Shirawa, , 387 ShiYéí see Yei Shona, , 45-46 Shonga(wa) see Shanga Siculo-Italian, 153 Sikimi see Taita Cushitic Singa, , 387-388 Sitia (Sherbro) see Bolem Siu, 209 (cf. Swahili dialects) Sizaki, , 88 "Sneese" see Southern Khoisan So, , 388-389 Sogoo, , 258, 389 (cf. Kalenjin dialect cluster) Sokô: see Sogoo Somali, , 118-119, 128129, 131, 133, 137, 139, 142, 146147, 153, 251 Somali dialects: Karre dialect, 146 Southern Somali dialect, 148 Sonjo, , 88, 102 Southeastern Tzeltal, 22 Southern Bul(l)om see Bolem SOUTHERN KHOISAN, , 390 Spanish, 25, 32, 48, 52, 132 "Strontbergen" see Southern Khoisan Suba, , 3, 24, 83, 88, 92, 102, 273-283, 390ff. Proto-Suba, 275 Subi, , 88 Subu see Isuwu

Index of languages (and variants) Sukuma, , 87-90, 93, 104, 106, 108 Sumbwa, , 88, 91, 108 Sumerjan, 7 Surma languages, 126 Surwa, , 88, 92 Su see Isuwu Swahili, (KiSwahili), 36-37, 41-47, 49, 83, 86-96, 99-112, 118, 121, 128-129, 131, 137-153, 181-212, 271, 275, 277ff„ 287, 292-293; "Swahili-ness", 206 Swahili dialects: 171 (cf. Amu; Baraawa; ChiVumba; Chirazi Chifundi; Jomvu; Funzi; Wasini; Mtang'ata; Siu) Standard S„ 91, 163-168, 172, 174175, 178, 185, 189-190, 192, 194, 197ff„ 201, 209 Unguja, 163ff., 170, 175, 189-190, 197-198, 201, 209 Urban Zanzibar varieties/Kimji, 91, 105 Zanzibarian Swahili, 168 Kimtangata, 91 South Pemba, 91 KiHadimu, 91, 104-105 KiMakunduchi, 91, 105, 209 Kikae , 91, 93, 105 KiTumbatu, 105 Southern Tanzanian S., 104 Northern S., 143, 145, 147, 151 Northern Kenyan S. (cf. Pate Tikuu), 158, 163, 165, 170-173, 175 Central Kenyan Swahili - Mvita, 158, 160, 163-176, 185 Archaic Mvita, 167-168 Southern Kenyan Swahili, 158, 163, 166-169, 171-172, 174-179 Proto-Swahili, 171

443

Kenya Pidgin Swahili, 132 Sweta, , 88, 92 Simí amdangti see Amdang Seki(yani) see Seki Soyo-Vanye see Ligbi

Tagba/Tagbo see Tagbu Tagbu, < Sudan, Zaire, Central African RepubliO, 392 Taita Cushitic, , 392ff. Tandanke see Bedik Tapasíáac see So Tasmanian, 22 Tatog, (Datoga/Barabaiga), , 90, 107, 251 Taturu, (Nyaturu), , 102, 107 Taura, , 394 Taurancii/Taurawa see Taura Tabu see Tagbu Tchanga see Shanga Tcheratz-Agaw see Xamtaqa Tenat see Tenet Tenda Bande/Tendanka see Bedik Tenet, , 394-395 Tepes/Tepeth see So Terik, , 3, 83, 255-272, 395 (cf. Kalenjin dialect cluster) "Terik-ness", 263 Teti see Deti Ti'-gamo see Gamo Tijanji, , 395-396 Tikuu, 160, 169, 209 (cf. Swahili dialects) Tindiga see Hadza Tiriki, , 256, 262, 271 (cf. Luyia) Tletle see Deti Togoy see Togoyo Tongwe, , 88, 102 Toro/Nyoro, , 89 Tschopi see ShiChopi Tubasa see Basa Kontagora

444

Index of languages (and variants)

Tubeta, < K e n y a > , 285, 288 Tuken, (Tugen), < K e n y a > , 256 (cf. Kalenjin dialect cluster) Turkana, < K e n y a > , 236, 240, 246, 252 Turkish, 31, 65-68 Tuotomp see Bonek Txopi see ShiChopi Tyanga/Tyenga see Shanga Toko kwe:goi see Kwegu Te:rik see Terik U(-)bôi see Buy U-ci'la àtàkàyà see Taura U-gyem see Gyem U-ndu rishi see Rishi U-sháú see Shau Uto-Aztecan, 11 Vidunda, , 102 Vinza, < Tanzania > , 102 Viri see Bir(r)i Vuna see "Degere" Waasi see Alagwa Waaf see Boni and Dahalo Waata, < K e n y a > , 137, 139 (cf. Oromo) Wakingdiga see Hadza Wanda, , 102 Ware, , 397 Warlpiri, , 118 Wasi see Taita Cushitic Wasini, < K e n y a > , 191-192, 197 (cf. Swahili dialects) Watindega see Hadza Wela see Ligbi West Margi see Putai Western Kpan see Kpan Wetu, < S u d a n > , 397-398 Weyto, , 398-399 Wovea/Woweta see Bobe Wunjo, 292 (cf. Chagga dialects)

Xamta of Abergelle see Xamtaqa Xamtaqa, (Abergelle Agaw), , 399-400 Xiri, < South Africa>, 400 Xirikwa see Xiri Yaaku, < K e n y a > , 3, 12, 24, 27, 83, 88, 132, 137, 146, 213-236, 249, 400-401 Yao, , 91 Yashi, , 401 Yeei see Yei Yei, < Botswana/Namibia>, 401-402 Yeke, , 93 Yeye/Yeyi see Yei Yidi(nich)/Yidinit see Kwegu Yihudi see Kimanteney Yorda see Kpan Yó γ ίοηί see So Zaghawa see Berti Zande Biri see Bir(r)i Zaramo, , 88-89, 92, 104, 112, 402 Zenaga, , 402 Zinza, , 87-88, 91 Ziriya, , 403 Znaga see Zenaga Zulu, < S . A . > , 129

Clicks: /?auni /rju hci Musa Musan /kham

see Southern Khoisan see Southern Khoisan see Southern Khoisan see Southern Khoisan see Southern Khoisan

//η !ke see Southern Khoisan /fküJRe see Southern Khoisan //kxigwi see Southern Khoisan //xogwi see Southern Khoisan

Index of languages (and variants)

jkaurure//nai see Southern Khoisan f k ë f k ë see Southern Khoisan fkhomani see Southern Khoisan jnusa see Southern Khoisan fcmkwe see Southern Khoisan

.'gä.'ne see Southern Khoisan IKhuai see Southern Khoisan !Kora(na) see Kora lora see Kora ,'xrikwa see Xiri

445

m

Hounkpati B. C. Capo

m A Comparative Phonology of Gbe m

1991. 24 χ 16 cm. XXIV, 238 pages. Cloth. DM 138,ISBN 3-11-013392 X (Publications in African Languages and Linguistics 14)

m m m m m m m m m m m m m m m

(Foris Publications • Berlin · New York)

The Gbe language unit, spoken in Ghana, Togo, Bénin and Nigeria, comprises more than fifty lects including such well-known members as Eue, Fon, Gen-Mina and Aja. This book presents the first comparative study of this unit and proposes an internal classification based upon the author's firsthand investigation of nineteen present-day lects. On the basis of this documentation, a reconstruction of the major features ofProto-Gbe segmental phonology, including its phoneme inventory and major syllable structure constraints and phonological rules, is proposed. This monograph will be of interest not only for its specific contributions to linguistic reconstruction in West Africa, but also for its innovative approach to comparative methodology, which draws upon some of the results of contemporary phonology including distinctive feature theory. One of the author's major conclusions is that the Proto-Gbe phoneme system included nasal vowels but no nasal consonants, these arising as variants of oral consonants before nasal vowels, leading to the present-day system of "paired" oral and nasal consonants. The present monograph constitutes an important new addition to the field of Gbe studies and West African comparative linguistics, and will provide a valuable reference work and resource for future researchers.

mouton de gruyter Berlin · New York

Michael Clyne (Editor)

m

Pluricentric Languages

Differing Norms in Different Nations

rrrr

1991. 23 χ 15,5 cm. VI, 481 pages. Cloth. D M 198,-

m

(Contributions to the Sociology of Language 62}

m

m m m m m

m

ISBN 3-11-012855-1

This collection of original contributions deals with 16 languages which are national languages in several different countries. The different norms and different ways the languages are used in each of the countries are investigated. Each national variety of the language becomes a national symbol, though the language also unites different nations and cultures. The languages dealt with here are Arabic, Armenian, Chinese, Dutch, English, French, German, Hindi/Urdu, Korean, Macedonian, Malay/Indonesian, Pacific Pidgin Englishes, Portuguese, Serbo-Croatian, Spanish and Swedish. The articles deal with the differences in structures and norms, the differences in status of these structures and norms in the various countries, and the convergent or divergent planning that is taking place. A typology and theoretical framework for the description of pluricentric languages is developed. Comparisons are made and concluding findings are summarized.

m m m m m

mouton de gruyter Berlin · New York