IIIIIIIIIIIII'IIIIL LIBRARY 11 11111111111111 Michigan State 1293 00670 3668 University This is to certify that the dissertation entitled Bukavu Swahili: A Sociolinguistic Study of Language Change presented by Timothy Lloyd Wilt has been accepted towards fulfillment of the requirements for Ph.D. degreein Linguistics 442/ gajor professor Date Jig, /y /429 ,/' / MS U is an Affirmative Action/Equal Opportunity Institution 0—12771 LMSU . RETURNING MATERIALS: Place in book drop to remove this checkout from your record. FINES will be charged if book is returned after the date stamped below. BUKAVU SWAHILI: A SOCIOLINGUISTIC STUDY OF LANGUAGE CHANGE By Timothy Lloyd Wilt A DISSERTATION Submitted to Michigan State University in partial fulfillment of the requirements for the degree of DOCTOR OF PHILOSOPHY Department of Linguistics ABSTRACT BUKAVU SWAHILI: A SOCIOLINGUISTIC STUDY OF LANGUAGE CHANGE By Timothy Lloyd Wilt The primary research concern of this dissertation is how various historical and social factors have influenced the development of Swahili as spoken in Bukavu, Zaire. Hypotheses concerning variation in linguistic structures and social attitudes are tested by statistical analyses of the recorded speech of over one hundred Bukavu residents and of responses to an interview schedule. The linguistic variables used to focus the analysis are: pronunciation of the class 2 marker 1315; and the phoneme (j), realization of the copula (ni), and the frequency with which French lexical items are incorporated into one’s speech. Implicational scales indicate how interaction of vowel quality, syllable position with regard to stress, and lexicon affect realization of (j). The key independent variables are age and education but also studied are: network strength, degree of church involvement, occupation, sex, ethnic group, and evaluation of the importance of learning standard Swahili. Explanation of the Bukavu speech situation involves an integration of models of language contact and change: diglossia, koineization, continuum, shift and borrowing. It is concluded that socioeconomic conditions favored a continuum of Swahili varieties ranging from vernacular to. near-standard in prefindependent Zaire but that the greater accessibility of French and resultant lack of socioeconomic prestige associated with standard Swahili in post- independent Zaire have yielded a diglossic relationship between the vernacular--which continues to develop as a koine--and standard varieties. It is also concluded that: Thomason and Kaufmann (1987) are right to insist that understanding of a language’s social setting is a prerequisite for interpretation of its linguistic features, at least in a complex multilingual speech situation such as is found in Bukavu; that two types of prestige--linguistic as well as socioeconomic-~should be distinguished in the study of linguistic variation; and that understanding of the relationship between prestige of and exposure to linguistic varieties will facilitate explanation of differences between individuals’ speech patterns. Copyright by Timothy Lloyd Wilt 1988 For my number one (tlanguage) helper and friend Rukabuza Uwamaria ii ACKNOWLEDGMENTS Many people have enabled my work on this dissertation to be a positive—-generally enjoyable--experience. My first words of appreciation must be given to Professor Carol Myers Scotton who has persistently insisted that I resist intellectual inertia, encouraged diverse academic interests, frequently assisted me in obtaining financial support for my studies, and continued unofficially but actively as my dissertation adviser after leaving to direct the linguistics program at the University of South Carolina. I am grateful for Professor Grover Hudson’s willingness to take on the duties of being my official dissertation director, after Professor Scotton’s departure. His logistical assistance while I was in Zaire and his lessons on diplomacy while I was here are especially appreciated. My research in Bukavu was--most of the time--a pleasurable experience due to the assistance of many people. Francois Bushikwabo, Kambala Kimono, Buhendwa, and Masumbuko Lusamaki were very helpful as diligent research assistants. Ron Butler of African Christian Mission cheerfully made available the ACM print- shop and his linguistic expertise. Mark Huddleston of Pioneer Bible Translators provided as much friendship as disagreement and challenges concerning my linguistic work when he would come to town. My family is thankful for the joy in our Bukavu stay of having been able to frequently be with the families of Maria, Pascal, Anatole, Dative, and Mamas and Babas Kora, Nathan, Patrique, Sifa, and Teté and with David Aronson, Lisa Berry, Berthe, Francoise, and our faithful helper and friend Bazera. The field research would not have been possible without the financial support, which I here gratefully acknowledge, from Fulbright-Hayes Grant #0008640397 and National Science Foundation Grant #BNS-8619484. The latter grant also helped meet computational expenses back at MSU. Finally, I affectionately acknowledge the support of the Lloyd and Lois Wilt foundation which has specialized in providing me frequent emergency funding and bountiful long- term caring. iii Table of Contents List of Tables.......................................... vi List of Figures......................................... vii 0. Introduction........................................ 1 l. Perspectives on Language Contact and Change......... 12 Bukavu Swahili and the creole continuum model. 13 l Zairean Swahili varieties as "creoles"...... 14 2 Zairean Swahili and the continuum model..... 19 3 Hypotheses in light of the continuum model.. 22 Bukavu Swahili as a koine..................... 25 Diglossia and polyglossia models.............. 29 Fishnan’s expansion of the diglossia model.. 30 Timm's expansion of the diglossia model..... 32 Platt’s expansion........................... 33 Two Highs v. the vernacular Swahili Low..... 35 Specialization of function................ 36 Prestige.................................. 37 Literary heritage......................... 41 Formal acquisition of H................... 42 Standardization........................... 43 Stability................................. 44 Grammar................................... 45 Lexicon................................... 45 Phonology................................. 47 anguage shift and borrowing.................. 48 1.5 Summary....................................... 52 Notes.OOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOO. 54 Hl-‘HHHhfiHr-‘HHHHHHHHHHHH @QQQOI-wao-b bwuwwuuwuwuwuwwwuuwn 2. Methodology and preliminary analysis................ 62 The interview................................. 63 The interviewer............................... 65 The interviewee............................... 66 Recording the data............................ 67 Independent variables......................... 67 1 Age group................................... 68 2 Education................................... 71 3 Occupation.................................. 73 4 Sex......................................... 77 5 Ethnic group................................ 79 6 Network strength............................ 80 NNNNNNNNNNN OIUIOIUIUIWOIIFQNH iv NNNNN e e o o e 9Q 0009 moommacoooooaoq QQQO’OI O O 03 NH NNNNNNNNNN eeoeeeeoee gas the: O 3. HF‘QQQQUIIFODNH wwwwuuwwuwwwwwuwwuuw O O O O O O O O O O O O O I O O O O O O NNNNNNNNHHHHHHHHHHwH o o e e e e e «mason-butch. 4. concluaionOOOOOOO0.00000000000.0000000000000000 .7 Level of church involvement................. econdary studies of Bukavu Swahili........... ociolinguistic interview schedule............ Evaluation of Swahili varieties........... Questions concerning use and understanding Of swahilj-OOOOOOOOOOOOOOOOOOOOOOOOOOIOOOOOOO Questions concerning social patterns........ Dependent variables........................... Phonological variables...................... .1 .2 .3 1:"!8(018882prefix)..................... jLJlOOOOOOOOOOOOOOOOOOOOIOOOOOIOOOOOOOIOOOO Two approaches to variation............. Phonological environments and frequencies 97 Constructing an implicational scale..... 100 Morphosyntactic variable: (nil.............. 118 Environments with nearly categorical use of 'kOeeeeeeeeoeoeo000000.0000000000000so. Environments used in hypothesis testing... Ethnic language influence on realization Of the variable...OOOOOOOOOOOOOOOOOIOOOOO. Lexical variable: Hypotheses Hypothesis Hypothesis Hypothesis Hypothesis Hypothesis Hypothesis Hypothesis Hypothesis Hypothesis Hypothesis Hypotheses Hypotheses Hypotheses Hypotheses Hypothesis Hypothesis Hypothesis 7.90000... 1000...... 2.0...OOOOOOOOOOOOOOOOOOOO. 3.... 4000...... 5000...... 60.0.00... 10-110000. French frequencies....... Teatin‘ 0f HyPOthQBBSOeeeeeoeoeo0.000000000000000one Hypotheses concerning linguistic variables.... 0.00.00.00.0000.000.00.00... 12.0.00.00.00.0000000000000000000 1300...... 1400...... 15.1800... 19.2100... 22'2300000 24.250.00.00... 26.... 27.0000... 280.000.... 3.3 Discussion......... NotesOOOOOOOOIOOOOOOOOOOO General conclusions.... Theoretical implications.................... For further study........... Language policy issues..................... V esting of Hypotheses 15 to 28................ OOOOIOOOOOOOOOOOOIOOOO. .OOOOOOOOOIOOOOOOOO... OOOOOOOOOOOOOOOOOOOOOOO CO .OOOOOOOOOOOOOOOO. 120 122 117 131 132 143 145 145 148 155 157 160 161 162 163 165 168 169 173 173 177 179 182 183 185 187 188 202 206 206 210 220 223 Appendix A: A comparison of varieties of Swahili: Standard Coastal, Standard Zairean, and BukaVUOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOI. 225 Appendix B: Zairean Swahili features cited as deriving from local language influence but appearing in coastal Swahili dialects.. 235 Appendix C: Texts used in translation preference test.. 237 Appendix D: Interview schedule: language use and attitUdeBOOOOOOOOOOOOIOOOOOOOOOOOOOOOOOO0.. 245 Appendix E: Scoring for interview schedule............. 253 Appendix F: Individual (1) scores for each environment. 256 Appendix G: Individual (nil scores for each environment 261 Appendix H: Individual {nil scores for combined enVi-ron-entaOOOOD0.000000000000000000000000 266 Appendix I: Plot of w a- scores....................... 269 Appendix J: Independent variables and phonological and IQXical scorGSOOOOOIOOO0.00.00.00.00... 270 Appendix H: -Samples of recorded interview texts........ 273 ReferenceSOOOOOOOOOOOO0.000000000000000...OOOOOOOOOOOOOO 292 vi Table HHHO O O O O “NHH ram 004 cameo: NH NN MN NNNN MN 0 C O O N O H y... 2.12 2.13 2.14 LIST OF TABLES Caption Page Mother tongue language(s) according to age..... 9 Translation preferences according to education. 39 Diglossic lexical pairs........................ 46 French and L Swahili equivalents commonly used in Bukavu................................. 47 Transcribed interviewees according to age...... 69 Number of subjects at various levels of education...................................... 72 Subjects according to occupation and age group. 76 Number of subjects according to age and sex.... 78 Local language cognates........................ 79 Number of subjects according to ethnic group and age group.................................. 80 Environments considered in analysis of (j)..... 99 Implicational scales based on percentage of [-sonorant] forms of (j).......................102 Revised cross-over environments specified......104 Percentage of [-sonorant] pronunciations by age group on lower and upper ends of the implica- tional scale...................................106 Total of [-sonorant] vs. [+sonorant] occur- rences of (j) for three individuals............107 Coding of individual subjects' (j) scores......108 Occurrence of variants within environments.....108 Variation in environments specified in Table 2.7......................................110 [-sonorant] realizations of (j), according to segmental environment..........................112 Significance of difference between pre-stress (j) scores and scores in other environments with regard to stress... ..... ..................115 Effect of my presence on teachers’ scores......l47 Percentage of g; variants according to interviewer....................................147 Mean score of phonological and lexical vari- ables correlated with occupation...............150 Percentage of n; variants according to occupation.....................................150 Teachers’ phonological scores..................151 (wa-) scores in the oldest-age group according to education...................................l53 vii 3.7 3.8 3.9 3.10 3.11 3.12 3.13 3.25 3.26 3.27 3.28 3.29 3.30 (wa-) scores according to level of church involvement....................................154 Correlation of scores with educational level, according to age...............................156 Phonological scores for educated speakers, according to age group.........................158 Mean ~1j1ua score according to age group and educationO O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O O 158 Percentage of n; occurrences by age group and educational level O O O O O O O O O O O O O O O O O O O O O O O O O I O O O O 1 59 Evaluation of necessity of learning standard Swahili correlated with age group..............162 Correlation between phonological scores and subjects’ positive evaluation of the necessity of learning standard Swahili...................162 Scores according to sex........................164 Phonological variables correlated with sex, discounting those with a high level of church involvement....................................165 Percentage of g; variant by ethnic group.......166 Percentage of n; variant by ethnic group, with regard to education.......................167 Linguistic scores correlated with degree of church involvement.............................170 Median scores according to degree of church involvement....................................170 Church attendance correlated with phonological scores.........................................l71 Percentage of g; variant uses by degree of church involvement.............................172 Evaluation of Swahili varieties by age group...175 Evaluation of speakers’ educational and salary level according to age groups...........178 Number of times standard speaker was con- sidered as being from a certain area, by age ‘roupOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOOIBO Vernacular speakers evaluated as to origin.....181 Evaluation of speakers’ personality............183 Preferred educational language by age group....185 Comprehension of radio Swahili by age group....186 Understanding of church Swahili by age group...187 Self-evaluation of ability to use Swahili......188 viii Figure 1.1 1.2 1.3 LIST OF FIGURES Caption Page Full diglossia and complementary diglossia in Zaire, based on Timm (1981)....... 33 ' Application of Platt’s "multilingualism and polyglossia" scheme to eastern Zaire........... 35 Continuum situation/Language shift, present in pre-Independence eraOOOOOOOOOOOOOOOOOOOOOOOOOOO 53 Diglossia with standard opposed to koineization 54 Relative sequences of (j) environments.........103 Sequences rearranged for greater parallelism if differences between occurrences within age groups are not significant.....................103 Likelihood of [-sonorant] realization of (j)...117 Hierarchy of influences on linguistic choice...218 ix Introduction -This dissertation reports on a field study conducted in Bukavu, Zaire with regard to the structural features of the Swahili spoken there and the social framework within which these features have developed. Basic research questions were: --Is the Swahili spoken in Bukavu converging with a standard form of Swahili? --How have other languages, especially French, affected the vernacular? --How does education influence one’s Swahili speech performance? --How are Swahili varieties socially evaluated and how do these social evaluations correspond to speech performance? --Have there been diachronic changes in social evaluation of the Swahili varieties? The two main data bases used to test hypotheses based on these and related questions are: (1) the recorded and transcribed speech of over one hundred Bukavu residents and (2) responses to an interview schedule concerning language use and social attitudes toward varieties of Swahili speech. The subjects were selected through a stratificational sampling approach. That is, subjects were non-randomly selected according to their fit with various social categories which, through many sociolinguistic studies, have been shown to influence linguistic behavior--e.g., level of education, occupation, and sex. Analysis of the linguistic data involved the selection of phonological, morphosyntactic, and lexical variables, counting occurrences of the variants, and assigning scores based on the percentage of standard variants used. These scores were then correlated with the selected social variables to test the hypotheses concerning social influences on the diachronic development of Bukavu Swahili. I have primarily relied on two models of language contact and change to formulate the hypotheses and interpret the results of testing them: the continuum model used in studies of creoles (e.g., DeCamp 1971, Bickerton 1975, and Singler 1983) and koines (Siegel 1985) and the diglossia- model originally formulated by Ferguson (1959) and expanded by sociolinguists such as Fishman (1967) and Platt (1977). The study of variables from three different linguistic sub- systems (phonological, morphosyntactic,and lexical) is inspired by Thomason and Kaufman’s (1988) argument that focus on one linguistic subsystem may lead to a distorted view of the sociohistorical developments of a language. The results of testing the hypotheses indicate that pre-Independence conditions favored vernacular speakers’ acquisition of Standard Swahili and, perhaps, the gradual convergence of the vernacular with the standard; but, since Independence, the much greater accessibility to French and its almost complete displacement of Swahili as a language of socioeconomic advancement have greatly diminished the motivation for incorporating features of standard Swahili into one’s speech. In terms of the models used, conditions favoring a speech continuum between the vernacular and Standard Swahili varieties have been replaced by those favoring a diglossic relationship of Swahili varieties--and a polyglossic relationship between these varieties and other languages used in Bukavu. The difference between the pre- and post-Independence conditions is reflected in both the linguistic features of recorded subjects’ speech and their self-reports on their use, understanding, and evaluation of Swahili varieties. Before proceeding to the main body of my work, I shall sketch the historical background of Swahili in Bukavu. 1. Swahili The Bantu language Swahili, the most widely spoken African language, is one of the two official languages of Tanzania and Kenya, one of the four national languages of Zaire and a lingua franca used in at least seven other countries: Mozambique, Zambia, Burundi, Rwanda, Uganda, Somalia, and Southern Sudan. Originally the language of people living on the northern Kenyan and Somali coast in the ninth century (Nurse and Spear 1985:48-49), Swahili was spread by its trading and fishing speakers southward as far as Mozambique during the next four centuries (ibid:52) and remained the main coastal language up through the twentieth century. During the colonial era, the dialect of Zanzibar, administrative center for both Arabic and European powers, came to be considered the standard. Swahili’s nineteenth and twentieth century spread inland was expedited by traders from the coast seeking ivory and slaves. Swahili in Zaire dates from the second half of the nineteenth century when Swahili-speaking traders from the Tanzanian Coast established trade networks throughout the eastern Congo. Some linguists seem to believe that these traders spoke a "pure" coastal dialect, nearly equivalent to the one described in grammars based on Zanzibar Swahili, but that the purity was lost as Swahili spread throughout the Congo; the Swahili of the new Congolese speakers, according to this perspective, was simplified in its quick adoption as a lingua franca, and reshaped via influences of local mother tongues (e.g., Goyvaerts and Kabemba 1986:218-226; Haddad 1983:79; Polomé 1969:221-226; Burssens 1954:27; LeCoste 1948). However, I suggest an alternative explanation in section 1.1. The Swahili as spoken in the Congo came to be known, by outsiders at least, as Kingwana because of its association with the wsngwsna, a term originally referring to a "free, civilized person" from the east coast Swahili culture but later extended in the Congo to anyone working for the Arabo— Swahili traders (Goyvaerts and Kabemba 1986:217). However, there is a confusion in the literature in the use of "Kingwana": some use it to refer to the spoken Swahili in the Congo/Zaire most closely resembling the "good" Swahili of coastal dialects (e.g., Harries 1955:13; LeCoste:l954) but others use it to refer to the most pidgin-like forms of Swahili (e.g., Nida and Fehdereu (1970:152); Van den Eynde l944:6). I have never heard even the oldest of speakers spontaneously refer to their language as "Kingwana"; after prolonged probing on my part, one old man did say he spoke Kingwana but that it was the same as Swahili. Attitudes toward the varieties of Swahili spread throughout the Congo varied from those of the Protestant missionaries who produced literature in a Swahili that was "unique, and embodied local provincialisms and faithfully reproduced grammatical errors and peculiarities of vocabulary" (Deans 1953:79) to of the Catholic missionaries and Belgian administrators who essayed to use only "correct" Swahili, that is, Zanzibar Swahili as represented in grammar books: Ce "kingwana" n’est qu’un vil baragouin, ne tenant nullement compte des regles essentielles d’accord de la vrai langue Swahili.... Nous devons nous dresser contre la tendance a repandre ce "sabir" dans notre Colonie. (Van den Eynde 1944:6) Between these two extremes were the Protestant missionaries who worked towards synthesizing a "Congo Swahili" from the various areal dialects.1 This conflict between what the great majority of the Congo population speaks and what language policy-makers think they should be speaking continues to this day. .5 Furthermore, the policy-makers themselves remain divided with regard to what the proper forms of Swahili in Zaire should be. The most recent Protestant Bible translation (Habari N139; 1981) is consistent with previous ones that have attempted to make a compromise between the vernacular and "grammatical" Swahili. Catholic priests, on the other hand, continue to incorporate into their publications many morphosyntactic forms that are never used and lexical items that are not understood2 by vernacular Swahili speakers. A European priest responsible for the recent translation of the Bible into "Zairean Swahili" told me that this translation (to which no Zaireans contributed) was adapted from the Coastal Swahili but was still "correct"--unlike the Protestant translation which was "full of faults," "a bad Swahili," "the Swahili of the market, of the women" (incidentally, study of the variables examined in this dissertation indicates very little difference between women’s and men’s speech). Further complications are seen in that both the Catholics and Protestants use orthographic forms that are not used in secular publications and do not use vernacular grammatical forms that Egg recognized in the grammar published by the University of Lubumbashi (Nkiko et al. 1985). Regardless of conflicts, however, Swahili is one of Zaire’s four national languages (the use of each is basically determined on a geographical basis). In the three provinces where it is the dominant lingua franca, it is a medium of instruction for the first three years of primary school and is then, in principle, taught as a subject for the next three years. A standard, or at least non- vernacular, Swahili is also used extensively for oral and written religious instruction, on radio broadcasts, and, in Bukavu, in a two-page section of the otherwise French weekly newspaper.3 2. Bukavu‘ Bukavu, with a population of about 300,000, is the fourth biggest city in Zaire. It is the largest city and capitol of Kivu, Zaire’s easternmost province which borders Uganda, Rwanda, Burundi, and Tanzania. Originally inhabited by the Shi, the Bukavu area took on military importance for the Belgian colonizers in the early twentieth century because of its proximity to the borders of Rwanda which was controlled by the Germans until World War Two. With a mile-high altitude and on the southern edge of a 60-mile long lake surrounded by mountains, Bukavu’s scenery and temperate weather encouraged increased settlement by foreigners until it became the administrative center of Eastern Zaire as well as a vacation center for the colonialists and an educational center for their children. Coffee plantations, urban construction, and increased settlement by expatriates drew laborers from various areas and ethnic groups throughout the Eastern Congo. The first centre extra-coutumier for these laborers to live in was created in 1935; a second, in 1953. Until Independence, the Congolese were restricted to these areas, unable to venture into white areas without permits. Kadutu, the first centre extra-coutumier, remains the most densely populated area of Bukavu. The majority of subjects for my study are from this area; most of the rest are from Nguba, a smaller but still densely populated, low- income settlement developed since Independence in an area that once was reserved solely for Europeans. The two predominant ethnic groups are the Shi, the area’s original inhabitants, and the Rega, originating from south and west of the Shi territory. However, there is a large number of inhabitants from other ethnic groups: Goyvaearts et a1. list forty-four ethnic groups in Bukavu (1983:54). A diversity in ethnic languages and dialects corresponds with this diversity in ethnic groups. Thus, a lingua franca is of obvious necessity. Before Independence, there was little language spread from one ethnic group to another (Goyvaearts and Kabemba 1986:218ff) and French was restricted to a very small minority of Congolese educated within a Europeanized framework. Thus, the Swahili brought by the coastal traders and used by missionaries and colonialists became the inter-ethnic group language. While ethnic mother tongues are still widely used in rural areas where ethnic groups and geographical areas more clearly coincide, urbanization has led to decreasing ethnic it”! A 11514 31 TE (4) If. a l' language use and increasing Swahili use. Table 0.1, based on responses to the interview schedule explained in chapter 2, indicates how Swahili is replacing mother tongues in Bukavu. # of 2232 Subjects EL ggigg §E Sw+Fr EL+Sw+Fr Before 1950 28 86% 7% 7% 0% 0% 1950-1969 27 7% 44% 48% 0% 0% 1970-1975 29 3% 14% 72% 7% 3% Table 0.1 Mother tongue language(s) according to age. EL=Ethnic language, Sw=Swahili, Fr=French. Selected subjects were demographically similar across age— groups. (p < .001) While the great majority of subjects born before 1950 speak (or spoke) only their ethnic language with their parents, the large majority of those born after 1970 speak only Swahili with their parents. This does not necessarily imply that they learned the Swahili from their parents; rather, that even if the parents would prefer to use the ethnic language, the children prefer to use the inter-ethnic language. One friend’s comment on how his young children react when he speaks his ethnic mother tongue-~the numerically predominant one in Bukavu--is indicative of future developments: "When I speak Mashi, they laugh at me!" Notes 10 1. In 1934 and 1946, two conferences with the goal of synthesizing a Congo Swahili were held for translators. In the first conference, the varying areal dialects were sorted into two types: that which was rudimentary but constantly evolving ("the Ituri dialect") versus that which was "the best spoken in the region" ("the Lualaba dialect") (Deans 1953:80). 'The New Testament (along with much other religious literature) was later published in each of these two posited dialects but by the time of the second conference, Deans asserted: Many differences between the two dialects were found to have practically disappeared. Lualaba was willing to concede that Kingwana should have "d’s" and "1’s", and to relinquish many of the provincialisms which had divided previously. Ituri granted that it needed a better grammar, incorporating the Lualaba’s system of concords. (Deans 1953:81) The first sentence of this quote suggests a convergence of linguistic dialects but the next sentences indicate that it was more a convergence of translators’ attitudes towards the dialects that occurred. This apparent convergence of attitudes ultimately resulted in the publication of the Bible in "Congo Swahili" in 1960. Nida’s observation that "Literary products in the dominant dialect set the norm of usage, and other dialects tend to conform to this usage" was quoted by Deans as support for the hope that "The Scriptures in [Congo Swahili] will prove...to be of great value in encouraging linguistic unity..." (1953:82). However, though Congo Swahli was the dominant dialect in the minds of the missionaries supporting its use against the coastal standard or the "deteriorated" (Deans 1953:78) Kingwana, it was certainly not the dominant dialect spoken by the Congolese. Congo Swahili was, perhaps, less foreign than the coastal standard but it was still foreign. 2. This claim is based on vocabulary tests that I gave to about fifty Bukavu residents. They are not reported on in detail in this dissertation. 11 3. The Swahili section of the paper has been viewed as a vehicle for reaching the great majority of the population, especially the older people and rural dwellers who do not know French sufficiently (Muganza et a1. 1981:368). 4. Much of the historical background given here is based on Kasongo (1985) and Bizuru (1973). Chapter 1 Perspectives on Language Contact and Change Since the early part of this century, linguists have applied a variety of sociolinguistic labels to the Zairean Swahili vernacular, evaluated it in terms of the coastal standard, and made predictions about its future development. In this chapter, I consider these various perspectives in light of the sociopolitical history of Zaire, suggest how different models of language contact and change may be integrated to account for the interaction of the vernacular and standard Swahili and French, and offer hypotheses that test the validity of using these models to explain the Bukavu Swahili speech situation. The integrated framework which I propose necessitates viewing sociolinguistic models of di- and poly-glossia, the creole continuum, koineization, language shift and borrowing as pertinent to co-existent aspects of a sociohistorically unified language contact and change situation. Although the necessity of this integrated approach is implicit in many sociolinguistic studies, analysts tend to focus on one model of language contact and change and discuss whether it is appropriate for the analysis of particular speech communities. In studies of di-lpolyglossia, this focus on one model to the exclusion of others has led to static models of language contact situations (e.g., Timm 1981; Platt 1977; Fishman 1967). In studies of creoles and koines, it has led to much debate with regard to defining 12 13 the history and linguistic structure of a prototypical creole or koine and to the validity of applying this prototype to other language situations (e.g., Bickerton (1984) v. Singler (1984) and Goodman (1985) on creole origins and articles reviewed by Siegel (1985) on koines). Studies of Zairean Swahili have paralleled this reliance on an exclusive model of language contact and change. Most studies harmonize with the continuum model of creole studies. Only three (that I know of) consider Zairean Swahili in terms of koineization, and none have discussed it in terms of di-lpoly-glossia. The bulk of this chapter is devoted to showing how each of these models of language contact and change are relevant to the Zairean Swahili situation. Most attention is given to the di-/poly- glossic model since this has received so little attention with regard to the Zairean situation, yet has high explanatory value. 1.1 Bukavu Swahili and the creole continuum model In this section, I first consider why Polomé, the most frequently cited contemporary linguist with regard to Zairean Swahili, classified Lubumbashi Swahili (a variety very similar to Bukavu Swahili; see Appendix A) as a creole. I then consider other linguists’ predictions about Zairean Swahili’s development with regard to the continuum model. Finally, I indicate which of my hypotheses about the 14 development of Zairean Swahili are especially relevant to this model. The continuum model is particularly useful for explaining sociolinguistic dynamics of Zaire’s pre- Independence past, especially with regard to the effect of education on Swahili speech patterns. But, there must be recourse to other models for explanations of other aspects of both the pre- and post-Independence eras. 1.1.1 Zairean Swahili varieties as "creoles" Zairean Swahili varieties have been referred to as pidgins (Duran 1979:32; Hancock 1971:518; Nida and Fehderau 1970:152) and creoles (Fabian 1986:111; Manessy 1977:129; Polomé 1971:57). In Bukavu and Lubumbashi (the focus of Polomé’s studies), the use of Swahili as a mother tongue and in a variety of public situations coupled with its none isolating morphosyntactic system immediately rules out the appropriateness of the label "pidgin".1 However, the label "creole" merits more consideration. I shall consider it in terms of Polomé’s use since his argumentation is the most detailed. Linguistic features that he considers to indicate Lubumbashi Swahili’s status as "undoubtedly a creole" (1969a:233) are: 1. extensive lexical borrowing from a socioeconomically prestigious language that is not linguistically related to the languages on which the grammar is based; 2. morphosyntactic and phonological features that result from local language interference; 15 3. morphosyntactic simplification, especially with regard to the class concord and the tense systems; and 4. its use as a first languagez. He does not claim that the creole developed from a pidginized form of Swahili. Similarly, Fabian (1986:111) says that Lubumbashi Swahili is a creole that did not go through a pidginization stage. However, many creolists believe that only languages which have developed from a pidgin should be considered to be a creole (e.g., Ferraz 1983:120; Woolford 19833; Bickerton 1981:4; Sankoff 1980:198; Givon 1979:23).4 The social origins of the linguistic system are usually also definitive. Most creolists (e.g., the ones just cited in this paragraph) are concerned with those languages which have developed among enthnolinguistically diverse speakers who were permanently separated from their ethnic homelands to work in plantation colonies. While urban dwellers in the Congo were ethnolinguistically diverse, their displacement was not nearly as extreme as it was for the plantation creole speakers and the urban dwellers were generally able to travel back and forth between their work-sites and their home areas (Polomé 1969b). If these limitations on the meaning of "creole" are not accepted, the term becomes so broad that its usefulness is questionable. As I point out below, this is a basic problem with Polomé’s concept of a creole: though the features he discusses are found in prototypical creoles, they are also a. 16 found in languages that have been influenced by other types of contact situations. The first two features listed above are typical of creole situations in which the acrolect (target) language provides the bulk of the vocabulary and the ethnic languages of pidgin/creole speakers have greatly influenced the grammatical structures. However, the parallel between Zairean Swahili and typical creole situations is not as evident as Polomé implies. First, the extensive lexical borrowing from French that Polomé cites is mainly in reference to cultural or technical innovations! and thus no more an indication of creole status of the language than would be Japanese speakers’ incorporation of English words into their language. Even the use of French words that have a ready Swahili equivalent (see section 1.3.1.8) would not necessarily indicate creole status. The same phenomenon of "unnecessary" borrowing exists in situations where speakers of one language borrow from the language of another group of higher socioeconomic status if there is ongoing, heavy social contact between the two groups (as in, for example, Canadian French borrowings from English (Mougeon and Beniak 1987), and Ateso borrowings from Swahili (Scotton and Okeju 1973)). In these situations, speakers maintain their own distinctive grammatical systems in spite of the extensive borrowing. Second, several features of Zairean Swahili that Polomé ascribes to the influence of local languages were actually 17 features of coastal dialects that members of the nineteenth century trade caravans--many of whom settled down in Zaire-- were probably speaking when they came to Zaire. For example, Polomé claims that the "influence of the local [Zairean] languages is especially obvious in the shape of the class prefixes, as shown by....the replacement of syllabic [m] in classes 1 and 3 by 92;" (1969a:225). However, this was a common feature of all coastal Swahili dialects, according to Sacleux (1939:619). In fact, of nine syllabic or morphemic features claimed by LeCoste (1948) and/or Polomé (1969a) to be the result of local language influence, at least six were features of coastal Swahili dialects (see Appendix B). Furthermore, even the features that do appear in Zairean Swahili but not in coastal dialects are not necessarily due to local language interference. Considering the heterogeneity of ethnic and geographical origins of those in the large trade caravans, it seems more likely that reshaping of the coastal Swahili would already have been in process before the gradual incursion of the traders into Zaire. Goyvaerts and Kabemba say that the subordinates of the Arabo—Swahili trade caravan chiefs were "nombreux" and "originsires de plusieurs regions de l’Est africain et meme du zsire" and that it was these people who "plus que [leurs maftres],...étaient en relation svec les autochtones" (1986:225). The caravans were composed of hundreds of support personnel and would take months to cross from the m- '4 Lu... RI . 18 coast into Zaire; the expeditions in Zaire would often go on for years (Kajiga 1967:18; Tippu-Tip 1958). Thus, there would be extensive interaction between caravan members speaking diverse Swahili dialects before entry into Zaire and it would be from these auxiliaries more than from the leaders of the caravan that the Zaireans would be learning Swahili. Simplification of phonological and morphosyntactic features, Polomé’s third criterion for calling Zairean Swahili a creole, is often given as a characteristic of creoles but, again, this is a phenomenon that occurs in many other language-change situations, including one which we shall consider below: koineization. In conclusion to this section, it may be said that Polomé’s reasons for calling Lubumbashi (and, by extension, Bukavu) Swahili a creole are not convincing. However, it, like Kampala Swahili, is "creole-like" with regard to general social evaluation of it as not being a "good" or ”correct" language like the standard Swahili (Scotton 1968). However, the negative evaluation is mostly with regard to linguistic form rather than social identity of the speaker; it could not be said of the Zairean situation, as DeCamp says of the Jamaican one, that "The creole is inseparably associated with poverty, ignorance, and lack of moral character" (1971:26). A speaker who borrows extensively from French but uses vernacular Swahili forms will likely be judged as being of higher education and 19 economic status than the standard Swahili speaker even though his Swahili is considered to be "bad" (see results of testing Hypotheses 16 and 17). 1.1.2 Zairean Swahili and the continuum model Even if the Zairean Swahili vernacular varieties are not creoles, the continuum model of language contact and change developed in creole studies (e.g., DeCamp 1971; Bickerton 1975, Ferguson and DeBose 1977, Singler 1983) may be used to explain certain aspects of the vernacular’s relationship with the standard. The model posits "a socio- economically-oriented linguistic continuum" in which there is "a continuous spectrum of speech varieties whose extremes are mutually unintelligible but which also includes all possible intermediate varieties" (DeCamp 1971:28). The upper extreme is close to the standard language upon which the majority of the lexicon is based and spoken by "highly- educated leaders"; the lower extreme is the pidgin or creole varieties spoken by, for example, uneducated laborers. As Sankoff (1980) points out, the model is actually relevant to any language situation in which speakers of one language attempt to learn a second language.“ Siegel (1985:373ff) adapts the continuum model of creole studies to explain the development of a koine. Up through the 60’s, many linguists concerned with Swahili in Zaire believed that the vernacular varieties (in the low and middle ranges of the continuum) would be 20 replaced by "grammatical" Swahili (upper range of the continuum) through persistence in institutional use of the "correct" forms (Whitehead 1926:21-22, 1928:iii-iv; Deans 1953:81-82; Harries 1955:14; Hunter l956:ix; Natalis 1965:221; Kajiga 1967:19).7 Not until the late 60’s were there attestations that the vernacular forms of Swahili, far from disappearing, were "developing quite independently from the East Coast standard" (Polomé 1969:233) and that "a blind puristic adherence to East African forms by teachers and broadcasters will only result in the continued divergence between the Standard and the spoken varieties in Zaire" (Gilman 1976:16). One reason for this difference in predictions is that most of those in the first group were writing at a time when the socioeconomic and political conditions of Zairean Swahili speakers were quite different from when those in the second group were writing. Thus, they were not in speech situations that equally met the two conditions DeCamp (1971:29) listed as necessary for members of a speech community to move towards the standard: 1) The dominant official language is the same as the creole vocabulary base. 2) The social system provides for sufficient social mobility and sufficient corrective pressures from above. - French has been the official language in both eras (except for the brief stint of Flemish in the early part of the century). However, Swahili was an important administrative language, in both secular and religious 21 realms, in the pre-Independence era. Now, it is the national language of eastern Zaire, used as a medium of teaching in the first three years of school and, officially at least', taught as a subject in the following three years. When the mass media use Swahili, it is usually of the variety on the continuum’s "upper extreme". Religious institutions also continue to use a form of Swahili that is in the upper range. However, probably the most important corrective pressure mentioned by DeCamp-~"the necessity of learning a more nearly standard variety...in order to get a better job"--though present in the pre-Independence era is now rarely exerted on Zaireans as Swahili speakers. In the pre-Independence era, French was, as it is now, the language of the socioeconomic elite. But, there was much less opportunity than there is now for the Congolese to obtain an educational or occupational level in which French would be learned and used. Standard Swahili was the marker of a much more obtainable socioeconomic status; its speakers had an educational and, often, occupational status that was intermediate to that of the elite speaking French and that of the masses speaking ethnic languages and non-Standard Swahili. This era was one in which the church, upholders of standard Swahili (especially the numerically and economically predominant Catholics), had power nearly equal to the secular administration and business (Young and Turner 1985) and controlled health-care, educational and, often, 22 agricultural institutions as well as sacred ones. Thus, for the oldest speakers--and for them alone--there was a socioeconomic incentive for learning the standard. Since Independence, however, this pressure is much more apt to be applied with regard to French, the language of higher education and inter-regional and international communication (Hypotheses 5 and 24). 1.1.3 Hypotheses in light of the continuum model Several of the hypotheses guiding my research predict that the difference between pre- and post-Independence corrective pressures to learn the language variety at the higher extreme of the continuum will result in the continuum effect being observable only among the older subjects. Hypotheses 5,6 and 24 were formed with the assumption that opinions about the importance of learning Swahili in school would reflect socioeconomic pressures to do so: Hypothesis 5: Subjects’ positive evaluations of the necessity of learning standard Swahili will decrease from the oldest to youngest age groups. Hypothesis 6: Positive evaluation of the necessity of learning standard Swahili will positively correlate with standard scores. Hypothesis 27: From the oldest to youngest age groups, there will be a decrease in the perceived value of learning Standard Swahili. Paralleling this prediction of differences in attitudes concerning the importance of learning Swahili was the 23 prediction of differences in linguistic scores among age groups. It was predicted that while there would be evidence for a continuum effect, mainly due to education, among subjects of the old age group, there would be no such evidence for the middle and young age groups. The scores of the subjects in the old age group without formal education were assumed to be representative of the Swahili mesolects from which decreolization could take place. It was predicted that the linguistic scores of educated speakers in the old age group would indicate an acquisition of standard speech forms but that those of the middle and young age groups would indicate a maintenance of the mesolect forms of the old age groups speakers without. formal education: Hypothesis 2: Level of education will positively correlate with standard scores only among those who were educated before Independence, excluding middle and young age group scores of teachers and preachers. Hypothesis 3: Among the educated, standard scores will decrease from older to younger speakers, with the same exclusion as in Hypothesis 2. - Corollary A: Phonological and grammatical scores of middle and young age group speakers will not be different from the scores of the older, uneducated speakers. Hypothesis 4: Higher income will correlate with standard scores only in the old age group of speakers. These hypotheses could be summarized as follows: a. In general, there will be evidence for a continuum of Swahili speech varieties among those subjects who grew up before Independence but not among those who are younger. 24 b. Subjects’ evaluations of Swahili varieties will indicate a diachronic decrease in esteem of Standard Swahili and, thus, a decrease in the socio-- economic motivation to acquire it that is ecessary to sustain a continuum effect. Two groups of people were expected to provide exceptions to the predictions that those in the middle and young age groups would not be motivated to incorporate Standard Swahili into their speech: (1) those whose occupations encouraged use of the Standard (especially, teachers and preachers) and (2) those with a high degree of religious involvement: Hypothesis 1: Those whose occupation encourages use of the standard will show signs of movement towards the standard. Corollary A: The scores of teachers and preachers will be higher than the scores of all other subjects. Corollary B: Apart from those educated before Independence, only the scores of teachers and preachers will suggest a continuum effect by ranging from the low scores of others towards 100% use of standard forms. - Hypothesis 14: Those with a high level of church in— volvement will have standard scores higher than all other subjects, excluding teachers and preachers. In view of the diachronic and occupational limitations on the continuum model’s relevance to the Zairean Swahili situation, we must consider other models of language contact and change to achieve a comprehensive picture. I have argued for use of the continuum model while casting doubt on the appropriateness of the label "creole" for major Zairean 25 Swahili varieties. I shall now suggest that "koine" is more appropriate. 1.2 Bukavu Swahili as a koine Siegel (1985) offers an extensive review of the use of "koine" and "koineization" in the sociolinguistic literature. Like "creole", these terms have been loosely used in reference to various kinds of language mixing. He proposes the following, restricted definitions: Koineization is the process which leads to mixing of...language varieties which either are mutually intelligible or share the same genetically related superposed language. It occurs in the context of increased interaction...among speakers of these varieties. A koine is the stabilized composite variety which results from this process.... A regional koine usually develops as the lingua franca of a geographical area in which different dialects are spoken. It often becomes expanded in form and function to become a regional standard or literary language. (p. 376) Koines, according to Siegel, differ from pidgins/creoles in that the koines stem from language varieties that are much more similar typologically and develop much more slowly than pidgins. Koines may have three of the four characteristics listed by Polomé in his description of Lubumbashi Swahili (1969a), especially morphosyntactic simplification and use as a first language. Extensive borrowing from a non- genetically related language, the third characteristic given by Polomé, could also characterize "the expansion of function and form...[which] may occur in koineization after 26 the initial stage" (p. 376).9 Extensive incorporation of local languages’ morphosyntactic and phonological features into the Swahili variety would not be consistent with the koine model but I have pointed out the weakness of Polomé’s argument that this has occurred. Siegel quoted a historian’s description of the formation of the prototypical koine--koine Greek. With a simple change of geographical and cultural labels, this description could easily be applied to the formation of Zairean Swahili--assuming the validity of my perspective on its formation mentioned in section 1.1.1: The Zanzibar dialect spread rapidly as an official/literary/institutional language throughout East Africa, and it was spoken generally by educated Africans, though, probably, they still used their local dialect among themselves. Among the common people, one of the main centers for the growth of a mixed vernacular was, first, the mainland Mrima seaport and the caravans with which many of these speakers travelled inland and, later in Zaire, the cities Kisangani, Lubumbashi, and Bukavu. We hear complaints about the "impurity" of spoken Swahili in the Congo area as early as 190310. In this way the conditions are present for the formation of the Zairean Swahili koine, which was mainly Zanzibar Swahili but included many elements drawn from the coastal/caravan Swahili and some from local Zairean dialects. (my adaptation of the G. Thomson quote, in Siegel 1985:358, concerning the formation of koine Greek)11 The above description refers mainly to oral speech varieties but the koineization process may result in a recognized standard dialect that is used for literary purposes. There is some evidence that this is occurring-- though very slowly—-in Zairean Swahili on a literary level. 27 I shall suggest how this might be so by showing how tw0' grammars, one published about thirty years ago, the other five years ago, have been approaching the grammars underlying Zaireans’ oral Swahili use. In the introduction to this dissertation, I mentioned a "Congo Swahili" Bible produced by a group of Protestants. This translation was based on a "Congo Swahili Grammar" whose preface warned the reader that: i the language spoken just exactly as it is given in these lessons is not spoken by the African. You are learning a union dialect (and actually a literary dialect). (Hunter 1956:4) In this grammar, the East African Swahili orthography was adjusted to be more representative of the vernacular Swahili phonology, vocabulary adjustments were made (for example, the vernacular numerals were used instead of the coastal arabicized numerals), and the grammar was simplified-~but, Hunter assured his readers, still "correct". The degree of grammar simplification was limited. For example, though most Swahili speakers in the Congo distinguished only between human and non-human references in the class-concord system, Congo Swahili Grammar would use the more complex system of the East Coast. Hunter did not elaborate on the notion of "correctness" but seemed to mean that, in principle, though Congo Swahili Grammar would not contain all grammatical forms of the East Coast Swahili, it would contain 22$! forms found there.” For example, Congo Swahili Grammar would use only one of the three possible 28 coastal standard relative markers; (1) was used but not (2) or (3): l. m-tu a-na-ye-enda cl.l-person s/he-present-who-go 2. m-tu a-enda-ye cl.l-person s/he-go-cl.1 relative 3. m-tu amba-ye a-na-enda cl.l-person re1.marker-cl.1 s/he-present-go "A person who goes" Furthermore, Congo Swahili Grammar would not use :ggzi (see 4), by far the most common relative marker in Zairean Swahili, since it is not used in standard coastal Swahili to mark relative clauses. 4. m-tu mw-enyil' a-na-enda cl.l-person cl.l-rel.marker s/he-present-go The Nkiko et a1. (1983) grammar ventures even further. They recognize: the vernacular :55; suffix not only as a marker of habitual behavior when cooccurring with the present tense marker :9;;, but also as a marker of distant time when cooccurring with the past tense marker :11; or the future marker ;£§;; the possibility of two cooccurring objective ("indirect" and "direct") verbal prefixes (though this is one of the few morphosyntactic features which, apparently, appears in Shaba Swahili but not in Bukavu Swahili); and the diminutive nominal prefixes kg; (singular) and £3; (plural). Both the :95; tense/aspect markers and the diminutive prefixes are very common in the vernacular but did not appear in grammars before Nkiko et a1. 29 These grammars represent a description of Zairean Swahili that is much closer to the oral varieties than the Zanzibar-based grammars or the oversimplifications presented in various pre-Independence "guides" to Zairean Swahili (Fabian 1986 disparages European cultural biases represented in these guides). However, the gap between oral and written grammars is still wide. It remains to be seen if it shall be further diminished or if it shall continue in a manner describable in terms of the diglossia model to which we now turn. 1.3 Diglossia and polyglossia models Models of di-/poly-glossia have been overlooked in discussions of Zairean Swahili but may be used to explain much of the linguistic situation not covered by the continuum model. Much of the relationship between the vernacular and standard varieties of Swahili in Zaire may be described in terms of Ferguson’s classic definition of diglossia: a relatively stable language situation in which, in addition to the primary dialects of the language (which may include a standard or regional standards), there is a very divergent, highly codified (often grammatically more complex) superposed variety, the vehicle of a large and respected body of written literature, either of an earlier period or in another speech community, which is learned largely by formal education and is used for most written and formal spoken purposes but is not used by any sector of the community for ordinary conversation. (l972[1959]:245) 30 However, as already suggested in discussion of the continuum model, the importance of French as Zaire’s official language necessitates consideration of its influence on Swahili use as well. As sociolinguists following Ferguson pointed out, concepts which he presented in terms of the relationship between two dialects of one language within one speech community are also relevant to situations in which two or more languages are used by the same speech community. In this section, we shall first consider expansions on Ferguson’s model to provide a general overview of the functional relationships between languages in Zaire and then focus on the relationship between the Bukavu vernacular, standard Swahili, and French, using key points of Ferguson’s definition as a guideline. I follow Ferguson’s use of "High" (H) to refer to the "superposed" variety and "Low" (L) to refer to the variety of "ordinary conversation". 1.3.1 Fishman’s expansion of the diglossia model Ferguson defined diglossia as pertaining to varieties of the same language. Fishman (1967) expanded the concept of diglossia to include relationships between non- genetically related language varieties within a society. Considering bilingualism to be a "characterization of individual linguistic behavior" and diglossia to be a "characterization of linguistic organization at the sociocultural level" (34), he suggested that the following 31 language situations relevant to these two dimensions may exist (examples following each type are his): 1. Both diglossia and bilingualism--members of the society know two languages, each having communicative functions clearly distinct from the other. Example: Paraguay, where almost the entire population speaks both Spanish and Guarani. 2. Bilingualism without diglossia--occurs in "circumstances of rapid social change... of widespread abandonment of prior norms before the consolidation of new ones" (34). Example: dislocated immigrants in intragroup communication. 3. Diglossia without bilingualism--occurs when there is political unity but socioeconomic elitism. The elite use one language, the masses use a different one. Example: Pre—World War 1 European elites speaking French and the masses speaking a different language. 4. Neither diglossia nor bilingualism--one linguistic system is used in all situations. Example: Small, isolated and undifferentiated speech communities (no specific example is given). (Fishman 1967:31-36) On a macro-level, positing a dichotomy between French and all African languages, the third situation could be taken as characteristic of the pre—Independence era when the Belgian colonialists controlled the Congo with the assistance of a small group of French speaking Congolese evques. With the great increase in education since Independence, though, the first situation is characteristic of an ever-increasing sector of Zaire’s population. However, Fishman’s schema is developed only with regard to speech situations involving two language varieties. The 32 typical African situation is much more complex. A speaker may often choose: between African languages (for example, between an ethnic tongue and a lingua francs), between varieties of one African language (for example, between H and L Swahili), or between an African language and a European one. We shall now consider two other expansions of the diglossia model which enable the application of basic relationships discussed in Ferguson to multilingual situations. 1.3.2 Timm’s expansion of the diglossia model Timm (1981) points out that Fishman’s expansion of the concept of diglossia considered only complementary distribution of functions, thereby overlooking other important aspects of diglossia mentioned by Ferguson such as the literary heritage and prestige of H and the stability in the relationship between the H and L varieties. Timm suggests that the distinction between this new concept of diglossia as a complementary distribution of languages (Dca) should be kept distinct from full diglossia (Dr) and then shows how these two different types of diglossia may be hierarchically related to each other. As well as having the advantage of keeping the more strictly defined diglossia of Ferguson, Timm’s approach also allows for easier representation of language relationships in multilingual situations. Figure 1.1 represents an application of her framework to the Zairean speech situation. 33 D National Official Ethnic Lan uages Language Languages Lingala Regional Lan uages Swahili Kikongo Chiluba Figure 1.1. Full diglossia and complementary diglossia in Zaire, based on Timm (1981). D¢¢=Complementary Diglossia; Dc=Full diglossia; H=High language forms; L=Low. This diagram represents the complementarity of the official language, national languages and ethnic languages on a national scale but then allows for further distinction between varieties of each of these languages on a smaller scale. I represent Lingala as occupying one branch in diglossic complementary distribution to the other national languages, grouped together under the other branch, since the use of Lingala by the Zairean military and musicians extends its use throughout all regions of Zaire much more so than the other languages. 1.3.3 Platt’s expansion of the diglossia model Platt’s (1977) reaction to previous work on diglossia was based on his experience in Malaysia. Here, the complex 34 relationship between several different language varieties is more parallel to the Zairean situation than those of other sociolinguists responding to Ferguson’s model. Platt argued that Fishman’s model of bilingualism and diglossia could be expanded to enable a framework for the study of gglgllingualism and pglzglossia. He suggested that in complex language situations there are "M(edium)" varieties as well as H’s and L’s. The degree of highness, he said, may be determined by the domains in which the variety is used and by speakers’ attitudes about the varieties; basically, "the more exclusively a speech variety is associated with more publicly and formally oriented domains, the higher would be its status" (370). Thus, as in Fishman’s model of diglossia, the factors involved in positing di-lpolyglossic relationships are not nearly as restricted as in Ferguson’s original definition. Figuree 1.2 represents an application of Platt’s scheme to the speech situation in eastern Zaire. 35 Order of Status Speech Varietz . H1 H; M; M; M; L; L; L; L- French x Standard Swahili x ”Protestant" Swahili x Vernacular Swahili x Lingala x Mashi X Kirega x Kihavu x Indoubil Figure 1.2. Application of Platt’s "multilingualism and polyglossia" scheme to eastern Zaire. The ”L-" in this scheme refers to a language variety with very limited functional use such as a thieves’ jargon or a bartering language. "Indoubil", the example I have given of a Zairean L-, is a language for the young, often associated with delinquents, based on Swahili grammar but with extremely high borrowing from other languages (Italian, English, and Chinese as well as languages commonly spoken throughout Zaire) and with reassignment of meaning (often with regard to the domain of sex, money, or illegal activities) to more common words (Kasongo 1985). 1.3.4 Two Highs v. the vernacular Swahili Low We now return to Ferguson’s definition of diglossia but shall consider the H v. L contrast with regard to the two 36 Highs--French (H-F) and, especially, standard Swahili (H-S). Although, as indicated in section 1.3.3, the vernacular Swahili might better be labelled as an "M"--Middle--variety in consideration of its role in a multilingual situation, I shall use ”L"--Low--to refer to it since this is in keeping with Ferguson’s terminology and since the focus of the following discussion is on the relationship of the vernacular Swahili with the High varieties only. I shall focus on the contemporary Bukavu speech situation. It shall be seen that although the French H was more remote than the Swahili H in the pre-Independence era (section 1.1.3), the opposite is true now. 1.3.4.1 Specialization of'function: "in one set of situations only H is appropriate and in another only L" (p.3283‘) In Bukavu, the H-8 is almost exclusively used in religious literature and meetings, in the mass media and in school. H-F is also used in these domains but less than H-S in the religious realm and more than H-S in the secular domains. The L is used in ordinary conversations and day- to-day interactions. Ferguson observed that, in Arab countries, although the policy is to use only H in class-rooms, L is often used in explanation of H materials. A similar situation exists in Zaire in the early primary school years only. At this level, book lessons are written on the board in H-S (students usually do not have their own books and thus spend 37 a good deal of time copying from the blackboard to their notebooks), but the teacher often breaks into L during class presentations--especially if s/he is addressing students for emotive purposes such as reprimands or effective story telling. L expressions may also be used to aid students in the early acquisition stages of French. However, from the fourth year of primary school on, recourse to L is hardly ever made. In many secondary schools, it is against the rules to use Swahili (or any other African language) even during recess, let alone during class.15 In church services that I have observed in Bukavu, the leaders use H-S much more consistently than the primary school teachers but, of course, their speech situation is more controllable than in the class-room setting. Editors of the local paper’s Swahili section attempt to maintain H-- even "upgrading" the grammar of letters to the editor (Masumbuko: personal communication). 1.3.4.2 Prestige: "There is usually a belief that H is somehow more beautiful, more logical, better able to express important thoughts, and the like." (p. 330) With regard to the H-S/L distinction, this principle holds true for Bukavu speakers who have been to school but not for those without an education. When subjects, responding to the interview schedule discussed in section 2.5.1, were asked to judge the quality of different speakers’ recorded Swahili1°, only eight of seventy-seven (10%) with at least some formal education felt that the L 38 speaker’s Swahili was as good as or better than the H speakers’s; but, fourteen of eighteen (78%) without formal education felt that the L speaker’s was as good or better, mainly because it was more understandable. A translation preference test given to about sixty members (fairly evenly divided according to sex) of a Protestant church about fifteen miles from Bukavu gave further evidence that the H is positively evaluated only if it is understandable. Subjects were given two pairs of texts. One had the H Catholic translation of a Biblical text juxtaposed to a Protestant translation of;the same, passage which was lexically closer to L (through circumlocutions and use of local equivalents of H termslV) but used H syntax and orthograpy. (I shall refer to this Protestant translation as "M" for Medium, in keeping with Platt’s scheme discussed in section 1.3.3.) The other had a different text from the Protestant Bible juxtaposed to my translation of the same passage using L morphosyntax as well as L vocabulary (but the same orthography as the Protestant Bible); this nearly-Low version shall be referred to as L. The texts used for this test are given in Appendix C. Subjects were asked: "If it was up to you to choose the kind of Swahili with which the Bible would be written, indicate the passage that has the kind that you would choose. We aren’t asking you which Swahili is correct; we’re asking you to choose the kind of Swahili that you would prefer to read." The results, broken down according to educational level, are given in Table 1.1. 39 Number of subjects choosing: Education H v. M M v. p 1-3 years 6 5 6 6 4-6 years 8 14 12 12 7-12 years 4 18 11 12 Table 1.1. Translation preferences according to education. While there is a clear (p-L.001) preference of more educated subjects for M over H (and if the scores of subjects with apparently inconsistent responses are discounted, the 1-3 year educational level group of subjects has a percentage of subjects choosing M over H that is close to that of the other two groups13), there is no significant difference in the scores for M and L1”; however, there are important differences in the subjects’ stated reasons for choosing M or L. The typical reasons given for the M preference over H were that M was more understandable and that it was the Swahili of Zaire. In the words of one subject: "[H] is like Tanzania Swahili; we are Zaireans". Thus, if the H is 399 High it may be perceived as "foreign" rather than "beautiful" or "more logical".30 Although there was no significant numerical difference in the choice between the M and L versions, there was a telling difference in the subjects’ explanations of their choices. I focus on the men’s explanations since they tended to give more detailed and consistent responses (see footnote 14). 40 Those who chose M said that it could be understood by most Zaireans and that it was a better Swahili: "It’s Kivu Swahili." "It’s better grammatically." "It’s the Swahili everyone must understand even if they don’t understand Swahili." "The other is hardly Swahili." "It sounds good for reading." The sentence subject in all of these responses refers to the text. The focus is on style and how Swahili should be written more than on how easily the individual himself is able to read and understand the passage. Of the sixteen men giving reasons for their M preference, only three answered saying that for them personally it was easy to understand. In contrast, eleven of the fourteen who chose the L version explained their preference in the first person, reflecting more of a concern for their own interaction with the text than for stylistic notions about what a text should be. "I understand it well" was the typical response. One liked it because "I don’t need to have it translated [into my ethnic mother tongue]" and another because "it is translated", that is, a foreign Swahili dialect has been translated into the local Swahili dialect. In sum, the preference of M over H coupled with the subjects’ explanations of M versus L preferences indicates that M has made a compromise between H and L so that, for many, it avoids being perceived as a very foreign dialect at one extreme or as grammatically crude at the other extreme; however, the M version’s avoidance of many common, wide- spread L forms results in a lack of accommodation to the 41 many for whom understandability/readability minimizes concern for "correct" grammar. With regard to the H-F versus L distinction, speakers are not concerned which variety is "more logical" or "more beautiful" but which one enables a better education and socioeconomic prestige. Obviously, French is the preferred language in this respect. Thus, H-F may be opposed to H-S as well as L; the large majority of present-day students would prefer that Swahili not be used at all in education so that the benefits of knowing French could be maximized (see .test of Hypothesis 27 and concluding section of chapter 3). 1.3.4.3 Literary heritage: "There is a sizable body of written literature in H which is held in high esteem by the speech community." (p. 238) Considering the whole of Swahili speaking Africa, there is, of course, a great deal of written literature in Swahili: recordings of poetry and legends centuries old, poetry and novels, scientific journals, magazines, newspapers, and, probably most extensive, religious literature. But, H Swahili’s classical literature is of little interest to the average Zairean (even the well- educated one)“ and little, if any, secular, contemporary literature is available in Swahili.“ Contemporary Zairean literature intended for an (inter-)national audience will be in French. Swahili textbooks are scarce for teachers, let alone students. Thus, the only significant literary material in ~ 42 Swahili is religious. This results in a psychologically powerful association between Sunday worship and a Sunday language and, perhaps, a stage of the process in which "H fades away and becomes a learned or liturgical language...not used actively in the community" (Ferguson 1959: 339).33 1.3.4.4 Formal acquisiton of H: "L is learned in...the ’normal’ [informal] way.... H is chiefly [learned] .by the means of formal education." (331) This is true in Zaire as well, with regard to both H languages. However, materials for teaching H are much less available in Zaire than they are in the situations discussed by Ferguson: at the commencement of the past school year, the local paper lamented the fact that there were, regardless of the language in which they were written, only "200,000 manuels pour 8 millions d’éléves!" (Jua 18 Octobre, 1986: 7). The emphasis on Swahili in the classroom has fluctuated throughout Zaire’s history. Before Independence, Swahili was the main medium of education with only a a very small minority having the opportunity to have more than a few years of schooling and study in French." From 1960 (the year of Zairean Independence) to 1974, however, only French was to be used as a medium for education in Zaire. Then, from 1975 to 1984, an African national language was to be used in the first two years, with French taught as a subject. The third year was to be a transitional one in which much more French was to be used than in the previous 43 years. From the fourth to sixth year of primary school, Swahili would only be taught as a subject. In 1984, a decision was made to move back towards increasing the use of French from the first year on: Swahili would still be used in most classes in the first two years but French would be taught as a subject from the first year on and could be used as a medium in courses such as mathematics. Interviews with and observation of primary school teachers in the Bukavu area indicate that there are insufficient educational materials and knowledge of H Swahili grammar for teachers to confidently teach H Swahili in any detail (Masumbuko 1987). Not until one year before the latest policy decision to reduce the use of Swahili in schools did Zairean linguists produce a grammar (Nkiko et a1. 1983) intended as an aid in teaching Swahili; the distribution of this grammar was very limited. 1.3.4.5 Standardization: "There is a strong tradition of grammatical study of the H....an established norm for pronunciation, grammar, and vocabulary which allows variation only within certain limits. The orthography is well established and has little variation." (pp. 331-332) There have been several attempts to write grammars for Congo/Zairean Swahilias but there is certainly not the uniformity, even in orthography, of the archetypal diglossic situation. Several grammars of "Zairean" Swahili are simply, apart from pedagogical exercises, abridgements of grammars based on Zanzibarean Swahili. Section 1.2 44 discusses two grammars proposing a standard that differs from the coastal Swahili standard.- H-F, of course, has the "strong tradition of grammatical study". Educators of Bukavu’s Pedagogical Institute (post-secondary school), though, lament a continual decline in the quality of their students’ French that parallels a steady deterioration of the Zairean public school system. 1.3.4.6 Stability: "Diglossia typically persists at least several centuries.... The communicative tensions which arise...may be resolved by the use of relatively uncodified, unstable, intermediate forms of the language...and repeated borrowing of vocabulary items from H to L. .(p. 332) Stability is the key characteristic distinguishing a diglossic situation from a continuum one. If L speakers do not have adequate motivation or opportunity to use the H in their daily speech, there will not be intermediate varieties between the L and the H; the H/standard will continue as the language of elite groups and/or a special language for formal situations and the L will develop on its own terms. As mentioned above, the question of whether or not this is so in Zairean Swahili has received various answers in the past and is central to my study. Hypothesis 3 and its corollary A predict that the speech of most middle and young age-group subjects will reflect a movement away from the patterns of older subjects with formal education to those of older subjects without formal education: 45 Hypothesis 3: Among the educated, standard scores will decrease from older to younger speakers, except those of teachers and preachers. Corollary 3A: Phonological and morphosyntactic scores among middle-aged and young speakers will not be significantly different from the scores of the old, uneducated speakers. In other words, it is predicted that the relationship between H-8 and L, though destabilized through pre- Independence institutional pressures to acquire H-S, is returning to a stable state in which a diglossic distance between H-8 and L is maintained. 1.3.4.7 Grammar: "H has grammatical categories not present in L and has an inflectional system of nouns and verbs which is much reduced or totally absent in L." (p. 333) This distinction between H and L is also found between creoles and their associated target languages (section 1.1.1) and between koines and the varieties from which they derived (1.2). Appendix A gives an extended list of grammatical differences between Zairean Swahili H and L. 1.3.4.8 Lexicon: "[There exist] many paired items, one H one L, referring to fairly common concepts fre— quently used in both H and L, where the range of meaning of the two items is roughly the same, and the use of one or the other immediately stamps the utterance or written sequence as H or L." (p.334) The focus here is on words referring to "fairly common concepts". In the case of specialized terminology, there would be no L term (except when referring to items particular to the L’s region/culture for which there might 46 be no H term) and the L speaker would have to either borrow from the H or another language or use a circumlocution. A long list illustrating this paired reference to common concepts in the Zairean Swahili situation could be drawn up but I shall offer just a few examples of words with varying grammatical functions (Table 1.2). English Eguivalent High L95 after (temporally) baada nyuma before ( " ) kabla mbele relative clause amba— -enye marker give -pa -patia try -jaribu -pima sell -uza -uzisha _buy -nunua -uza3' village kijiji mugini lake ziwa bahari37 bicycle baisikeli kings Table 1.2. Diglossic lexical pairs. An interesting lexical phenomenon in the Bukavu speech situation (and most other multilingual areas where a world language such as French or English is an official/national language) that cannot be accounted for within the diglossic framework as conceived by Ferguson is the use of words borrowed from the official language for non-specialized references that have L equivalents (Table 1.3). 47 English Eguivalent French Egg but mais ‘ lakini village village mgini hat chapeau kofia marriage mariage ndoa persevere/endure supporter ~vumilia cause trouble déranger -fanya fujo live vivre -ishi Table 1.3. French and L Swahili equivalents commonly used in Bukavu. The correlation between subjects’ use of French, core lexical items and overall incorporation of French into one’s Swahili is briefly discussed in section 2.8.3. At this point, I will simply observe that although in many respects French is a co-H with standard Swahili, the use of French words is characteristic of L speech whereas the use of H Swahili words is not. While French words such as those listed above can be used in ordinary conversation without any change in register, H Swahili words cannot. 1.3.4.9 Phonology: "The sound systems of H and L constitute a single phonological structure of which the L phonology is the basic system" and "if ’pure’ H items have phonemes not found in ’pure’ L items, L phonemes frequently substitute for these in oral use of H" (335,336). Phonological differences between H and L Bukavu Swahili are given in Appendix A. I will not attempt a theoretical 48 discussion of whether the H or the L is the basic phonological system. There is one H phoneme that is not found in the ’pure’ L of Bukavu: H /h/ is usually not pronounced. A lengthened vowel occurs instead, as in (1).30 (1) E L hapendi a:pendi "he doesn’t like to" This rule in conjunction with L’s tendency to have word- final {-e] in negative verbs, rather than H’s {-i], results in a phonemic contrast occurring in the third person singular between the short vowels of subjunctives (2) and lengthened vowels of present negatives (3): 1! la. (2) atake atake he shall want to (3) hataki a:take he doesn’t want to" 1.4 Language shift and borrowing So far, I have not given the motivation for the choice of particular linguistic variables to test the hypotheses. For example, Hypothesis 3a predicts that morphosyntactic and phonological scores of the middle and young age group subjects with formal education will not be significantly different from the old age group subjects without formal education; however, Hypothesis 3b predicts that the lexical scores will be different. Why the comparison of scores for one linguistic component should work out differently than for those in another component has not been explained. I 49 shall now do this in terms of Thomason and Kaufman’s (19873”; hereon, ThK) models of language contact and change. T&K discuss language contact and change at a higher level of abstraction than any of the models previously presented. They argue that the two models of "Borrowing" and "Language Shift" may be used to cover any language change situation. The sociohistorical relationships of the speakers of one dialect/language to those of another (or others) will determine which of these changes a language will undergo. In TaK’s model of language shift, a speech community will eventually lose its first language for the sake of acquiring a second language. This shift occurs since knowledge of the second language will offer advantages that could not be had without it. In this shift, the first language’s lexicon will only slightly interfere with that of the second language but there may be heavy interference on the phonological and morphosyntactic levels. ' The creolists’ continuum model is used to account for language shift from a diversity of linguistic backgrounds present in one speech community to the standard. It predicts that this interference would be evident in varying degrees among different speakers: the degree of exposure to the standard and of socioeconomic motivation to incorporate the standard into their speech would be inversely related to the degree of interference. 50 Thus, the hypotheses predict that evidence for shifting from vernacular to Standard Swahili will be found only among members of the oldest-age group for whom the socioeconomic conditions of the shift were present-—except for those whose occupation encourages use of the standard. Furthermore, they predict that evidence of this shift will be found in the phonological and morphosyntactic components where a high degree of variability is expected due to interference from the vernacular. The degree of interference in the old age group of subjects is expected to be inversely proportional to the degree of educational and occupational exposure to the standard. In a borrowing situation, speakers are influenced by speakers of a different language group but they retain primary loyalty to their first language. The first elements to be borrowed are lexical and this borrowing may be . extensive.. Borrowing from another language’s phonological or morphosyntactic systems is likely to occur only "if there is strong long-term cultural pressure from source-language speakers on the borrowing language group" and there is "extensive bilingualism among borrowing-language speakers over a considerable period of time" (p.56). If the vernacular Swahili of Bukavu is being changed through borrowing, speakers could be borrowing from one of several sources: French, Lingala, predominant ethnic languages or even English. French would be the most readily available source because of its extensive use in education 51 and inter-regional communication. Lingala has spread throughout Zaire as the language of popular musicians and the military. English is taught in secondary school and spoken by residents recently come from Tanzania. Swahili speakers could be drawing from the lexicon of these languages but all the while maintaining and perceiving vernacular Swahili as a dialect distinct from standard Swahili.u This being the case, one would expect a significant, structural linguistic gap between the local variety and the standard Swahili with little evidence for "a continuous spectrum of speech varieties" (DeCamp 1971:28) between the two. In the following chapters, I shall analyze only lexical borrowing from French. However, in terms of TaK’s principles mentioned above, the "cultural pressure" from French speakers and the increasing number of residents who are bilingual in vernacular Swahili and French suggest that there might also be borrowing of French phonological and morphosyntactic features. Indeed, there is anecdotal evidence that such borrowing is occurring in both components. Various linguistics students at Bukavu’s teaching institute have told me that it is common for secondary school girls to substitute the French back-velar /r/ for the Swahili flapped one.’1 Polomé suggests that there has also been borrowing from French syntax into Zairean Swahili vernaculars33: 52 ...the practice of literal translation has led to the transference of some French patterns into this variety of Swahili [in Lubumbashi], especially in its written form, e.g. unakuwa katika hatari” on the model of "tu es en danger" (’you are in danger’), fanya ile plaisir na miye, word for word ’fais-moi ce plaisir’34 (’do me this favor’). It went even as far as taking over idiosyncracies of Belgian French, like the postpositional use of ’avec’ with certain verbs, e.g. ’du fil pour coudre avec’ (’thread to sew with’), rendered faithfully as uzi ya kushona nayo35. (1969a:231) 1.5 Summary In the pre-Independence era several sociolinguistic conditions were favorable for a continuum situation in which speakers would move away from a koine Swahili towards the standard. Swahili was used extensively in education. The Catholic church, very influential in secular as well as religious affairs, supported use of Standard Swahili. Furthermore, the indigenous population had very limited access to the relatively high levels of education and the work that necessitated use of French . Thus, hypotheses predict that variation in phonological and morphosyntactic scores along with responses to an interview schedule concerning language attitudes will give evidence for a continuum, language shift situation in the old age group. Figure 1.3 represents this situation. 53 French (H1) ' Standard Swahili (H2) Lexical items borrowed from: Lingala, {L——“*~ £é/////’Local languages Vernacular Swahili Figure 1.3. Continuum situation/Language Shift, present in pre-Independence era. With the radical sociopolitical changes stemming from Independence, though, education has become available to increasingly higher proportions of the population and at increasingly higher levels. This has resulted in a perception of French as being much more accessible than it was before Independence and as essential for socioeconomic self-advancement. At the same time, the socioeconomic importance of Swahili is practically non-existent, except for the small group of religious leaders and, perhaps, primary school teachers who must use it as a medium of instruction. Thus, hypotheses predict that there will not be a continuum effect evident in the middle and young age groups; rather, they predict that linguistic scores will indicate a polyglossic situation in which the structural gaps between the koine and the standard remain the same for the middle and young age groups, regardless of educational level, as they were for the old age subjects without formal 54 education. The main difference between these groups will be that the younger subjects have borrowed from French more extensively because of their greater exposure to it in school. This situation is represented in Figure 1.4. Standard Swahili (H3) French (H1) Lingala (M) Local languages (L) L E X'I C.A L I T.E.M S l I / l 1/ 3V1? ; 8V2 4—9 Koine Swahili 8V, a‘————> 8V. Figure 1.4. Diglossia with standard opposed to koineization. SV=Swahili vernacular. thes 1. Duran (1979:148) contrasts standard Swahili structures with the isolating morphosyntactic constructions typical of pidgin Swahili in west—Central Kenya. To his list of examples, I add the equivalent, typical Bukavu constructions (morpheme by morpheme glosses are mine): a. Standard: si-mw-on-i lsg.negehim-see-pres.neg.concord Bukavu: si-mw-on-e lsg.neg-him-see-pres.neg.concord Pidgin: mimi hapana ona yeye I/me no see he/him "I do not see him." 55 b. Standard: si-ja-mw-ona ls.neg.-neg.impf—him-see Bukavu: si-ya-mw-ona ls.neg.-neg.impf-him-see Pidgin: mimi bado ona yeye I/me not.yet see he/him "I have not yet seen him." c. Standard: u-si-je you-not-come Bukavu: u-si-kuye you-not-come Pidgin: hapana kuja no come "Don’t come." Duran claims that "an equally characteristic feature of pidgin form is the use of the invariable predicative particle iko, derived from the Standard Swahili locative particle -ko, to form locative, predicate sdjectival, and possessive constructions" (p. 148). However, all his examples are in the third person, singular with animate subjects. Thus, they do not illustrate the invariability of this morpheme. In Bukavu, :59 has the same functions as those mentioned by Duran but it is conjugated according to person, number, and certain classes. 2. This last criterion is given in a different article: (Polomé l969b:912). 3. Although she does not explicitly make this claim, her whole discussion concerning the conditions under which a pidgin will creolize implies that all creoles develop from pidgins. 4. Not all creolists hold to this restriction. Gilman states: There has been a tendency to consider that JC [Jamaican Creole] is more complex than West African varieties of recent origin, and to believe that the first creole spoken in Jamaica was very pidgin-like... No evidence has ever been adduced to support this view, which directly contradicts the findings of the present study, nor are early citations of creole notably more pidgin-like than the present varieties. The idea is probably a carry-over of the old doctrine (inherited from the 19th century when there had been little 56 systematic study of any pidgin or creole) that creoles result when populations adopt pidgins as their native language. Muhlhausler suggests that, though most creoles have developed from pidgins, there are some that have developed from jargons. In this situation, "repair is needed at all levels, i.e., there is need for a natural phonetological and semantactic systems as well as a possible pragmatic system" (1980:32). He gives as examples of this: West Indian English Creole (1980:32), Hawaiian Pidgin English, and Unserdeutsch of former German New Guinea (l986:8). Similarly, Thomason and Kaufman, in discussing languages "in which the availability of the [Target Language] is so limited that the shifting speakers have successfully acquired only the vocabulary of the [Target Language], but little or none of its grammar" gives as examples: those creole languages that did not develop directly from fully crystallized pidgins. Some creolists do not believe that such languages exist; but certainly, as we will argue, there are a number of creoles for which a definite pidgin stage is not attested, and whose structure can be accounted for under a hypothesis of extreme [sic] unsuccessful acquisition of a [Target Language]. Hawaiian Creole English, Mauritian Creole, Seychelles Creole, and most of the Caribbean creoles fall into this category. We call this ’abrupt creolization’, and its products correspond to what Bickerton has termed ’early-creolized creoles’... (MS(pub1ished l988):71) 5. After stating that "Obviously [the French loanwords] are mostly related with the effects of the colonial administration and the partial Westernization of urban life", Polomé gives as examples of loans the words for "hospital," "bill" (or "receipt"), "hat," and "matches" either in the French orthography or in a orthography that represents Swahiliized pronunciation (1969a:15-16). 6. Sankoff observes: The problems posed for linguistic analysis by the variable linguistic materials we deal with as creolists are not, by their nature, different in degree or kind from the linguistic data encountered in other speech communities. (1980:139) and, citing several references in support, claims: That learners’ productions come to resemble a scale of increasingly closer approximations to the target code has been demonstrated in studies of the efforts of 57 foreign workers to learn German...and Dutch... of English traders to learn Navajo...and of Spanish immigrants to learn English in the United States. (1980:142) 7. In this era, various labels were applied to the vernacular forms. They mostly always were negative, viewing the variety as an illegitimate dialect with no rules: "mutilated form [of Swahili]" (Stapleton 1903:s) "[a language in which there is a] disparition des regles" (anon. n.d: 6); "un langage qui emprunte au Kiswshili la majeure partie de son vocabulaire et la plupart de ses fbrmes" (Labeye 1928:3); une langue relacbée (van de Weyer and Quets 1929:3); "un kiswahili impur....simplifié...assez abatardie" (anon. 1938: 9); "...une langue mixte...un mélange dialectal" (Verbeken 1944:3); "related more closely to ’kitchen’ Swahili or ’KiSettla’ than to genuine dialects" (Harries 1956: 50); developed from coastal Swahili speakers’ "constant necessity for speaking down to the people...[which] inevitably resulted in a deterioration of the Swahili language" (Harries 1955:12). 8. When I asked the principal of one of Bukavu’s leading schools to take me to classes beyond third grade in which Swahili was being used, he took me to one room where, according to the official schedule, Swahili was being taught. The teacher, however, excused herself saying that she wasn’t capable of teaching it. We went to another class where the teacher, scheduled to be teaching religion in Swahili, said it would be okay to observe. It was obvious, upon entering, that he had not been teaching the religion class, but a different course in French. However, he pulled out a religion text book and began reading from it and asking mechanical questions. 10. This would be Stapletons’s reference to the "mutilated form" of Swahili spoken in the Congo (1903:s). 11. The actual quote is: the Attic dialect spread rapidly as an official language throughout the Aegean, and it was spoken generally by educated Greeks, though they still used their local dialect among themselves. Among the common people, one of the main centres for the growth of a mixed vernacular was the Peiraieus, the seaport of Athens, inhabited by Greeks from all parts of the Mediterranean. .We hear complaints about the "impurity" of spoken Attic as early as the fifth century 8.0... In this way the conditions were created for the formation of the Hellenistic Koine, which was mainly Attic but included many elements drawn from Ionic and some other dialects. (G. Thomson quoted in (Siegel l985:358)) 58 12. One exception to this apparent principle is the use of the -ak- verbal suffix rather than the standard hp; prefix to mark habitual aspect (Hunter l956:xx; this also appears in English-Swahili Grammar 1985 which stems from the Hunter work). However, I have found no instance of the use of this suffix in the Congo Swahili Bible. 13. [i] varies with [e] in the final vowel position. 14. The page number at the end of each quote which follows the sub-headings in 1.3.4.1ff refers to the page in the Ferguson 1958 article from which the quote was taken. 15. This ban is regardless of whether the variety is High or Low. 16. The question asked concerning each variety was simply "Do you think this person’s Swahili is good, bad, or average?". 17. However, the M Protestant version also differed from the H Version in that it attempted to make explicit some ideas that its translators felt were implicit in the original Biblical texts. The degree to which subjects evaluations were affected by this difference in translational approach is not known. The assumption of the test is that they were positively evaluating the Protestant text more than the other because it used lexical items and morphosyntactic features that were more comprehensible to those accustomed to the vernacular. 18. A puzzling aspect of these figures is that the choice of H decreases as educational level increases. I believe that the reason for this is due to the fact that the L text is several lines longer than the H text rather than to the degree of highness variable since 8 subjects (all women: 4 with 1-3 years of education, 3 with 4-6 years, and 1 with 8 years) who chose H over M all chose L over M because, they said, the L was more easily understood. For all other subjects, the choice of H over M correlated with the choice of M over L. Furthermore, a separate vocabulary test (not given to these subjects) indicated that the H would be less understandable because of its lexical choices. I believe these 8 subjects may have indicated a preference for the H versions simply because it was seven lines shorter than the M version. For those who had already labored over comparing the disguised Mark passage (the two versions being of practically identical length) and who probably had domestic chores waiting to be done, the shortness of the H version may have made it seem more preferable. If the choices of these eight subjects are discounted, the indication of M over H preference is even more marked 59 than in Table 1.2 and the inverse ratio between education and preference for M is not so extreme: Namber of subjects choosing: Education M v. M 1-3 years 2 (29%) 5 (71%) 4-6 years 5 (26%) 14 (74%) 7-12 years 3 (14%) 18 (86%) 19. Choices could have been biased by some subjects’ recognition of the Protestant text as being in the style that they were most accustomed to. In a test given to a group of women in a Catholic Bible study, most chose the Catholic version. However, I attempted to compensate for this bias by disguising the New Testament passage; I changed personal and geographical names and modified the plot to make the story read like a narrative taking place in contemporary Bukavu. Subjects could not have seen the Protestant Old Testament text since it has not yet been published and they would have been very unlikely to have read the same text in the new Catholic Bible. 20. This situation seems to contrast with that described in Ferguson in which many speakers of a language involved in diglossia characteristically prefer to hear a political speech or an expository lecture or a recitation of poetry in H even though it may be less intelligible to them than it would be in L. (1959:330) However, there may be a greater concern for understandability with regard to the Bible than in the situations mentioned in this quote: test results concerning Hypothesis 28 indicate that understanding of church Swahili is higher than of radio Swahili. To know if there is really a contrast between the situation described by Ferguson and the Bukavu Swahili situation, one would also have to know if the "many people" Ferguson refers to is regardless of: educational level, frequency of exposure to H, and domain in which the H is used. 21. This is probably due to the average Zairean’s lack of empathy with the cultural history and values represented by the literature as well as the lack of exposure to the literature in school. 22. I could find no secular books written in Swahili, in any of Bukavu’s book stores. 60 23. It is unsurprising, in this light, that, in Bukavu’s leading school the only course apart from Swahili Grammar taught in Swahili after the third grade is Religion. 24. Kaplan reports on the small numbers educated before Indepence and the steady increase in those having a secondary education since Independence: During the colonial period education was the province of the religious missions... Among them the Roman Catholics were the most important. The primary aim of education was to train Africans for tasks that would be useful to European enterprises. Usually these were positions of a clerical or lower level administrative nature. Secondary education was limited to Africans who were to serve as clergyman, teachers, or army petty officers. No Africans attended universities or studied in Europe until the 1950’s.... In the mid-1970’s secondary education.... had doubled since the mid- 1960’s. (1978:176-77) 25. Van Spaandonck (1965) lists ten grammars dating from 1902 to 1965 and six "guides" to conversation written within that time period. An update of this list would include Natalia (1965), Annicq (1967), Kajiga (1967), Taabu (1975), Heylen (1977), Nkiko et al. (1983). Rwanika et a1.(l986). 26. Note how L -uza has taken on the opposite meaning of H -1128 e 27. In H, bahari means "ocean; sea" 28. The lengthening of vowels may also be accompanied by a rise in tone, in forming the negative at least. 29. The page numbers I use are based on a pre-publication manuscript, not on the recently published book to which I have not had access. 30. Subjects’ explanations of their evaluations of the standard and vernacular Swahili varieties (chapter 3, discussion of Hypotheses 16-19) indicate that, generally, they do indeed perceive the Standard as a variety distinct from the one they are accustomed to. 31. I have recordings of two school girls who do so. I also witnessed this practice in the speech of a sixth grade neighbor girl; her parents did not speak French in the home--and rarely outside it--but she has been going to "l’école des Belges", where no Swahili is used, since first grade. There are no Bantu languages, in the Bukavu area at least, which could be a source of this sound. Also with regard to this borrowing, the story is told of a Bukavu priest called Pere Mgpuri "Father Good" not 61 because of his righteous deeds but because of his using the French /r/ in Swahili words such as the one for "good". 32. ThK suggest that morphosyntactic borrowings are likely if phonological ones are made and, perhaps, to the same degree. ' 33. The literal gloss, not provided by Polomé, of this expression is: u-na-kuwa katika hatari you-pres.-become in danger "You become in danger." 34. However, contrary to Polomé’s claim, the second example' is not a "word for word" transference: Fanya ile plaisir na miye. Fais ce plaisir avec moi do that favor with me "Fais-moi ce plaisir." "Do me this favor." 35. The literal gloss, not provided by Polomé, is: uzi ya ku-shona na—yo thread of to-sew with-it Chapter 2 Methodology and Preliminary Analysis The main question underlying my field research was a diachronic one: is the Bukavu vernacular Swahili converging with or diverging from the standard? In order to address this question, the primary methodological concern was gathering tape-recorded speech from subjects belonging to different age groups. Within these age groups, subjects were classified according to other variables commonly considered in sociolinguistic research: education, occupation (which will indicate income level and socioeconomic prestige), ethnic group, and sex. They were also classified according to variables that have received less attention: subjective evaluation of Swahili speech varieties, opinion of the necessity of knowing the standard, network strength, and degree of involvement with a church. With an interest in getting an overview of the linguistic aspects of the Swahili speech situation, I studied phonological, morphosyntactic, and lexical variables. This choice was made giving special heed to Thomason and Kaufman’s observation that different components of the linguistic system may be affected in different ways depending upon the type of contact situation speakers of a language are involved in (section 1.4). The following sections describe aspects of my research methodology, explain the choice and subdivision of various 62' 63 independent variables, offer further hypotheses to be tested with regard to the influence of these independent variables on linguistic behavior, give an overview of the dependent variables selected, and briefly describe computational procedures used to test the hypotheses. 2.1 The interview All analyzed speech was recorded in interview situations with the interviewee told in general terms that the interviewer was interested in having stories and discussions in Swahili concerning the way of life in Bukavu. As in most sociolinguistic research of this type, the interviewer’s ideal was, generally, to "divert attention away from speech, and allow the vernacular to emerge" (Labov 1972:209) since the vernacular is assumed to be the truest indicator of language development (Rickford 1987:150ff). The unnatural speech situation of recorded interviews makes the eliciting of natural speech somewhat difficult but the interviewer tries to compensate for this by involving the interviewee in discussions or stories that interest the speaker enough to concentrate on the ideas rather than the speech. The setting of the interview also can influence the naturalness of the speech. Most of the interviews for this study were conducted in the interviewee’s home, generally with other family members or friends in attendance. Labov 1972 and others, e.g., Bickerton 1975, Singler 1983, have found this group setting to be effective in eliciting 64 natural speech. The Zairean interviewers, all long-time residents of the Bukavu area, used the vernacular to conduct the interview and asked questions that encouraged lengthy responses by the interviewees. Since the interview setting was generally constant, I assume that the scoring of linguistic behavior is valid on a relative basis though it may not be an absolute indication of natural linguistic performance. That is, I assume that if Subject A has a phonological score that is 10% higher than what it would have been without the tape-recorder in front of him, Subject B will also have a score 10% higher, all other factors being equal. Of course, other factors will not always be equal. For example, subjects educated before Independence may be more prone to use Standard forms while being interviewed than subjects educated after Independence. But even if this is so, the greater propensity to monitor one’s speech will indicate that a greater value is placed on the use of Standard forms. Questions were asked about matters to which the 'subjects could be expected to respond with ease. Elderly interviewees were often asked to compare life as it is today with how it was when they were young (physical changes in Bukavu, the youth, material welfare, customs, etc.). For old age-group subjects and the older ones of the middle age- group, a good question for eliciting a lengthy narrative concerned what happened in Bukavu when two different rebel armies fought there. This was a politically safe question 65 since one of the few, fairly legitimate claims to glory of the present "president" (who allegedly wins nearly 100% of the votes in elections which pit him against no one) in Zaire is that, under him, peace came to a country that was being ruined by political divisions in the 60’s, the time of these rebellions. Questions to the younger interviewees often concerned school and entertainment. Different questions were asked to different age groups to encourage spontaneity and ease of conversation. However, this creates a problem in dealing with lexical variation since the topic of a discussion or narrative will affect the lexical choices made. While I assume that the lexical scores (explained in section 2.8.3) used for testing the hypotheses offer a reliable, though approximate, figure, my future analyses of the data will involve studying how test results will be affected by more rigid control of topic. 2.2 The interviewer In twenty-six interviews, the speech situation was made less natural by the presence of a person who differed from the others in attendance by nationality, race, and, usually, socioeconomic status: myself. I was interested in whether or not my presence would significantly affect the speech patterns of the interviewees. If incorporation of standard features into one’s speech is positively evaluated by subjects, one would expect them to attempt to make their speech more standard to favorably impress the outsider (such 66 ‘"upgrading" of one’s speech has been commonly reported: e.g., Labov 1972:43ff, Russell 1980, Rickford 1987)1. The following hypotheses will be tested before the main hypotheses listed in section 1.5. . HYPOTHESIS 7: The absence or presence of a foreign interviewer will be a weak predictor of linguistic performance if a Zairean is present to co- conduct the interview. HYPOTHESIS 8: The absense or presence of a foreign interviewer will be a weak predictor of linguistic performance even if a Zairean is not present to co- conduct the interview. HYPOTHESIS 9: If either hypothesis 7 or 8 is false, the foreigner’s presence will significantly affect only those whose occupation would encourage use of the standard and those who were educated before 1960. 2.3 The interviewee A random sample of Bukavu residents that would enable efficient testing of my primary hypotheses was impossible, at least with my resources. Thus, I asked research assistants to seek out people according to various criteria. For example, I would ask an assistant to find one person each who met the following criteria: 1. Age: born before 1940 Ethnic group: Shi Education: at least 3 years 2. Age: born around 1955 Ethnic group: Regs Education: university graduate 3. Age: born after 1960 Ethnic group: not Shi or Regs Occupation: teacher 67 The assistant would find people corresponding to the descriptions, come back to me with a list of the people he found, and then, after discussion of the various ones on the list, go back to interview the one(s) I selected. With this sampling method, of course, the conclusions drawn from my analyses technically apply only to the selected subjects. But, they may, I believe, also be taken as a rough indicator of sociolinguistic tendencies in Bukavu and, perhaps, in other urban Zairean Swahili areas as well. 2.4 Recording the data 107 interviews, averaging around forty-five minutes each, were transcribed by Zairean Swahili speakers. Every transcription was double checked by an assistant specially instructed with regard to how to check the transcriptions. The transcription work was by far the most demanding task of the research, time-wise, financially, and mentally. It took six to ten hours to transcribe a forty-five minute tape'. After the transcription, another assistant and I typed the texts onto computer diskettes. We coded all French words so that they could be easily isolated for the study of lexical borrowing. A program giving frequency counts of all words in each text greatly facilitated the accurate, efficient scoring of each linguistic variable. 2.5 Independent variables The following variables are listed in the order of the importance that I attributed to them from the outset of the 68 study. The variable listed last--leve1 of church involvement (section 2.3.6)--was one that I had not considered at the outset; but, while recording the data, I came to believe that it might have more influence on scores than most of the other variables. 2.5.1 Age group Since diachronic trends in Bukavu Swahili are of special interest in my research, a key independent variable is age. Since there are no reliable records of Swahili as spoken in the early part of the century in Bukavu (or Zaire, in general), we must rely on apparent time analyses in which differences in the linguistic performance of the speakers of different age levels are assumed to reflect actual linguistic changes that have occurred through time. Labov’s Martha’s Vineyard and New York City studies (1966;1972), are the most well-known sociolinguistic works with apparent time used as an analytical tool. Rickford says that "Unfortunately,...distributions or correlations involving different age groups are rarely presented in studies of variation in pidgin-creole-speaking communities" (1980:176; his emphasis) but approvingly notes Gillian Sankoff’s work on Tok Pisin as an exception. Sankoff (1980:69-70) observed, for example, changes in stress and morphosyntactic function of a Tok Pisin lexical item indicating future tense by comparing linguistic behavior of children who were native speakers of Tok Pisin 69 with adults who were not. The historical situation of her subjects was similar to that of those in my sample. Most of those in the oldest age group grew up knowing an ethnic tongue as a first language and the lingua franca as a second. However, for those in the younger age groups, the lingua franca became a mother tongue. Sankoff says: Though not many languages are presently undergoing a change toward being, for the first time, native languages, the kinds of intergenerational differences observable among these speakers can be seen in other speech communities, and differences among groups need not be categorical to be signficant (1980:71). Rickford (1969) and Bickerton (1981) are other creolists that have looked to subjects of different generations for evidence of diachronic changes'. In her study of Mombasa Swahili, Russell (1980) also found linguistic behavior to be different between age groups. She then broadened the perspective on this factor by showing how differences in educational level and sense of in-group security were associated with the differences in age of her subjects (1980:134ff). In my study, the speakers of the transcribed interviews were assigned to one of three age groups (Table 2.1): Number of ngpp p233 Interviewees Oldest before 1950 - 34 Middle 1950-68 45 Youngest 1969—76 27 Table 2.1. Transcribed interviewees according to age. 70 Although several in the 0(1dest) age group were born outside of the Bukavu area (this to reflect the influx of workers from various geographical regions and ethnic backgrounds to the centres extra—coutumiers (see introduction)) during the first half of the century, all but one of those in the M(iddle) and Y(oung) groups were born in Bukavu‘. There were two main considerations in grouping the ages as in Table 2.1: population movement patterns and use of Swahili in education. Apart from members of the original ethnic group of the Bukavu area, most of the 0 subjects were not born in Bukavu. They were either brought at a young age by their parents or came as young men to work or look for work. Though there were relatively large population increases in Bukavu after the second world war (Kasongo 1985:12), the migration was controlled by the Belgian authorities who, in principle, would not allow people to move to Bukavu unless they either had work there or could find it within a short period of time. In contrast, those of the M age group were brought up and learned their Swahili during a time when the population of Bukavu exploded due, mainly, to migration from rural areas of eastern Zaire: Avec l’accession de notre pays a l’indépendance, en 1960, toutes les plantations sont abandonnées, les forces vives des collectivités affluent en masse vers la ville dans l’espoir de faire fortune et d’acquérir les biens laissés par le colonisateur suite aux promesses fallacieuses leur faites par les politiciens 71 de 1’époque....,§g§avugggi, en 1958, était peuplée 4e 48,263 habitants va voir sa population passer a 134,861 habitants en 1970, soit un accroissement démographique de 1701 en 12 ans. (Mpenda 1933: 118,119; my emph.) Those in the Y group are of school age. 2.5.2 Education The general policy concerning use of Swahili as a medium of education was different for each of these age groups. Until 1962, Swahili was the main medium of education for the first three years of primary school. In 1962 (the approximate time from which most of those in the M group would be starting schoolS), the policy changed to exclusive use of French as a medium. From 1975 on (the time when the oldest of the Y group were starting school), the policy reverted back to using Swahili as a medium. Thus, we may look for the effects of education on the speech of the subjects with regard to Swahili as a medium of instruction. (More information on use of Swahili for instruction is in sections 1.3.1.4.) The age groups subdivided according to educational level are given in Table 2.2. 72 Age Years of education Q5222 Q 1;; fi;§ 1:3 10-12 13-15 16+ 0 15 4 8 2 4 1 1 M 3 2 6 10 18 4 3 Y 1 1 12 11 3 - - Table 2.2. Number of subjects at various levels of education. Secondary school begins in the seventh year and ends in the twelfth. Undergraduate studies are for three years after secondary school.“ Years given are in terms of highest level achieved; thus, one who took 14 years to finish secondary school would still be considered in the "IO-12" group. The educational level of the M and Y subjects is concentrated in the middle levels more than for the 0 age 'group for three reasons. First, two of the five major hypotheses directly concern the effect of advanced levels of education on Swahili speech habits and two other major hypotheses, concerning occupation, indirectly concern this. Second, education is highly valued (Young and Turner 1985) and one can expect increasingly higher proportions of residents to have a high school education. Using statistics found in Kabanga et al. (1983) and Nzongola (1983), one can estimate (quite roughly) that at least 70% of the secondary school age residents in Bukavu were going to high school in 1976’. Third, as my field research progressed, it became evident that not until one reached the secondary level of education did lexical borrowing from French proliferate. Since this borrowing was of special interest to me, I wanted 73 to have a sizable sample of speakers from this educational category. Although the role of Swahili in the curriculum has received much attention (for example, articles in Whiteley 1974a and Polomeand Hill 1980), the effect of exposure to Standard Swahili in education on vernacular speech patterns has received little attention. Russell (1980) found that subjects with formal education, in comparison with those without, would make a "greater shift [from the vernacular] towards ’Standard’ Swahili" when talking with her. However, she does not give much attention to the influence of education on normal vernacular speech. 2.5.3 Occupation According to Labov, "It appears that a person’s own occupation is more closely correlated with his linguistic behavior--for those working actively--than any other single social characteristic" (1972:45). The main reason for this is the obvious link between occupation and socioeconomic prestige/class; it is the speech behavior of the socioeconomic elite which is commonly considered the standard variety and, thus, on which those of lower socioeconomic status may attempt to model their linguistic behavior with varying degrees of success (for example, with regard to English sociolects in the United States and 74 England: Labov 1966; Trudgill 1974: 34-56; with movement from Creole to standard varieties: DeCamp 1971; Bickerton 1975; Samarin 1980; with regard to choice between English, representing socioeconomic prestige, Swahili and ethnic languages: Scotton 1972,1977; Parkin 1974; Barton 1980).“ At the same time, non-standard varieties may be maintained as an expression of in-group loyalty and/or socioeconomic isolation from groups speaking the (near-)standard variety. (for example, with regard to English dialects: Labov 1980; Milroy and Margrain 1980; with regard to Creoles: Wurm 1977; Craig 1980; d’Offay de Rieux 1980; with regard to Swahili- speaking areas: Russell 1980; Scotton 1982). Since accessibility to French has greatly increased since Independence and since it is French, not Swahili, that now represents socioeconomic prestige, my hypotheses predict that socioeconomic status, as represented by occupation, will not correlate with incorporation of Standard Swahili features into one’s speech. They predict that occupation will correlate with linguistic scores only with regard to a classificatory dichotomy between jobs which encourage use of Standard Swahili (teaching and preaching) and those which do not (all others held by subjects who were tested): Hypothesis 1: Only those whose occupation encourages use of the standard will show signs of movement towards the standard. That is, a. the standard scores of teachers and preachers will be higher than the scores of all other subjects. b. the phonological and morphosyntactic scores of all 'other subjects will be clustered at a relatively low level. 75 c. only the scores of teachers and preachers will be spread from the low levels of others to the high, near- standard levels. Subjects other than students were classified as belonging to one of four occupational categories: 1. Manual labor and other low income occupations 2. Teachers 3. Skilled labor 4. Administrative 5. Preachers Included in the first group were: soldiers, manual laborers (such as house workers and night watchmen), an animatrice (one who is paid, very little, to dance in support of governmental policies and alleged accomplishments), a machinist, taxi or truck drivers, vendors, craftsmen, mechanics, and a hairdresser. Though those in the second group would be no more economically advanced than those in the first group, their work might encourage them to use a nearer-to-standard Swahili. Those in the third group included: nurses and mid-wives with no more than a secondary school education, a pharmacist, an inspector of hospitals, a notary, and governmental secretaries. Those in the fourth group included a technical supervisor, a highly educated executive secretary and a highly educated nurse in private (and thus economically advanced) industry, a banker, and a wealthy business person. The two members of the fifth group 76 were both Protestant preachers. The number of subjects assigned to each category are given in Table 2.3. Occupational Category ManL Tch Sle Adm Prch 1 2 2 4 5 A g 0 29 Z 4 3 1 e G M 19 7 9 2 1 I‘ o H Y - - - - P Table 2.3. Subjects according to occupation and age group. The concentration of subjects in the first category reflects the lop-sided economic structure of Bukavu where, according to one survey, the unemployed outnumber the employed (Nzongola 1983:302) and the large majority of employments are of low-paying manual labor and "micro- commerces [qui]...permettent a ceux qui en pratiquent de ne pas mourir de faim" (Nzongola 1983:300). There were two main difficulties in having representatives of the higher occupational cells. One was the lack of contacts that my assistants, of lower socioeconomic background, had with those of higher socioeconomic background. A second was the difficulty of arranging interviews even when contacts were made. This was due in part to the busy schedules of the potential subjects; but it also seemed to be due to the fear that a recorded 77 interview might somehow be used against them. Mobutu has been very successful in causing a general fear of rocking the political boat and those of the socioeconomically higher class are especially aware of the precariousness of their positions, often corruptly obtained (Young and Turner 1985 discuss this phenomenon in detail); Mutamba and Kabemba (1983), professors at the Bukavu Teachers’ College discuss similar problems of conducting interviews with those of higher socioeconomic class. 2.5.4 Sex Anecdotal observations that secondary school girls often replace the Swahili flapped /r/ with the French back velar one raise the question of whether women’s behavior differs from men’s in other respects. Their reputation for incorporating the French back velar /r/ of the prestigious French language suggests the possibility that the same may hold true for women in Bukavu as holds true for women in Britain and America: [In] urban dialect surveys carried out in Britain and America....allowing for other factors such as social class, ethnic group and age, women constantly use forms which more closely approach those of the standard variety or the prestige accent than those used by men. (Trudgill l974:91) This would probably be true only for those who have sufficient exposure to French through education or frequent contact with formally educated people. One would not expect older women with no formal education to attempt to emulate 78 the French accent or incorporate French core lexical items into their speech. Thus, the following hypotheses are to be tested: HYPOTHESIS 10: When educational levels are similar,< M and Y women will have a higher rate of borrowing from the French lexicon than men. HYPOTHESIS 11: M and Y women will not have a higher rate of standard Swahili pronunciations and morphosyntactic structures than men, of similar educational level. Hypothesis 11 is made with the assumption that Swahili is not associated with socioeconomic prestige in the Middle and Young age-groups and, thus, there will not be the motivation to incorporate Standard Swahili features into one’s speech that there is to incorporate French features. These hypotheses do not concern women from the 0 group since only an extremely small minority from that group had the same educational opportunities as men. The breakdown of subjects according to sex is given in Table 2.4. Age Sex 95222 Male Female 0 34 4 M 29 19 Y 12 15 Table 2.4. Number of subjects according to age and sex. 79 2.5.5 Ethnic group With the exception of mother tongue interference in the Swahili of those in the 0 group (the majority of whom, unlike those in the M and Y groups, did not have Swahili as a second--or first--mother tongue) and a few lexical borrowings from the numerically predominant group, there was no reason to believe that ethnic background of the urban dwellers would significantly affect their linguistic behavior in speaking the language adopted for inter-ethnic communication. With regard to ethnic language parallels to the Swahili variables considered in this study (to be explained in section 2.8), the languages that I know of have much in common. In all languages studied, the consonant in the class 2 morpheme, which marks plural references to humans, is [-sonorant] [ba-J or [ba-l. Their cognate equivalent of (j) is {-sonorant] (in the following examples of Table 8 are left where there are not clear cognates): "water" reflexive "name" "come" "not.yet" "be.full" Swahili maji -ji- jina -ja -ja- -jala Mashi amighi -gi- igina -i§ha -Ql- -yugjula Rega magi - i- igina -i§ala Rwanda amagi - i- igina -uga -u§ura Table 2.5 Local language cognates. Comparisons between ethnic languages with regard to the (ni) variable will be discussed in section 2.9.2.3. Incorporation of French into Swahili would most likely be related to 80 ethnic group only if there was an imbalance among ethnic groups in their average educational levels.10 Thus, the following hypothesis: HYPOTHESIS 12: Ethnic group membership will be a very weak predictor of scores for all examined linguistic variables. Subjects were coded as belonging to one of three groups (Table 2.5): Corresponding Number of Code Ethnic group subjects Explanation O M Y 1 Shi 17 16 13 Members of the ethnic group that is numerically predomin- ant and lived in the Bukavu area before the incursion of non-regional traders, mis- sionaries, and colonialists 2 Rega 9 10 7 The second most populous ethnic group 3 Others 14 15 13 Includes at least twenty-two other ethnic groups Table 2.6. Number of subjects according to ethnic group and age group. 2.5.6 Network Strength Recent work by British linguists (e.g., Milroy and Milroy 1985 and Russell 1980) has linked the degree of strength of social networks to the likelihood of the maintenance of vernacular norms: the denser the social networks of an individual the more likely s/he is to maintain the vernacular. A simple method for scaling network strength is given in Milroy and Margrain (1980:51): 81 Each informant is assigned a score on a scale ranging from 0-5. One point is assigned for each of the following conditions he or she fulfills: 1. Membership of a high density, territorially based cluster. 2. Having substantial ties of kinship in the neighbourhood. (More than one household, in addition to her or his own nuclear family). 3. Working at the same place as at least two others from the same area. 4. The same place of work as at least two others of the same sex from the area. 5. Voluntary association with work-mates in leisure hours. This applies in practice only when conditions three and four are satisfied. Milroy and Margrain concluded their paper by suggesting that their hypothesis that strong social networks will positively correlate with the preservation of vernacular norms should be tested in urban settings of different world areasu . Accordingly, the following hypothesis is made: HYPOTHESIS 13: Subjects Network Strength scores will negatively correlate with the phonological and morpho- syntactic scores and positively with the lexical scores. Scores for testing this hypothesis were based on responses to an interview schedule in which questions were asked to see how much the subject met the five criteria given above. The interview schedule was generally given much later than the taped interview with the result that only sixty-seven of the 106 transcribed subjects were scaled for network strength. Only twenty-five of these sixty-seven had network strength scales of less than 5. 82 2.5.7 Level of church involvement Before doing the field research, I had expected pastors’ Swahili to be much closer to the standard than the average. speaker’s because of the pastors’ professional use of near-standard Swahili in both study of Christian literature and occuational communication with the laity. I did not expect this church Swahili to affect anyone else. Thus, -my HYPOTHESIS 1 mentioned only pastors with regard to the religious domain. However, while recording the data of interviews on the field, I came to believe that this hypothesis was too narrow. Thus the following hypotheses were to be tested also: HYPOTHESIS 14: Those with a high level of church involvement will have standard scores significantly higher than all other subjects, excluding teachers and preachers. I considered a person to have a "high level of church involvement" if s/he attended church more than once a week. Those who went once a week or less would represent the typical layperson while the one who went more often would probably have a position(s) of responsibility in the church or at least above average participation in the church and thus be more prone to adopt aspects of linguistic behavior associated with the church. The majority of those rated as having a high level of involvement were Protestants. 2.6 Secondary studies of Bukavu Swahili While recording and transcribing the speech of a variety of Bukavu residents was the main focus of my field 83 research activity, supplementary studies were done that concerned: subjects’ reports on language attitudes and use, use of Swahili in the schools, and comprehension of the standard versus the vernacular. Only the first of these studies will be discussed in any detail in this dissertation. The study of language attitudes and use was based on an interview schedule which is discussed in the next section. 2.7 Sociolinguistic interview schedule The sociolinguistic interview schedule was composed of three main parts: (1) evaluation of three different oral Swahili varieties, (2) questions concerning the subject’s use and understanding of Swahili, and (3) questions about the subject’s social networks and language choice in various social relationships. The interview schedule and the transcription of the recordings on which the evaluations of part 1 were based are given, along with translations, in Appendix D. The administration of the interview schedule took about forty-five minutes. Ninety-eight interviews were done, 67 of the subjects having been already interviewed for the purpose of having their speech recorded. All but a few of the interview schedules were administered by my research assitants; I was the sole interviewer for only two interviews. Vernacular Swahili was always used by the interviewer. 84 2.7.1 Evaluation of Swahili varieties The procedures used to elicit evaluation of oral Swahili varieties was very close in both methodology and substance to that briefly reported on in Scotton’s (1973) study of Ugandan Swahili. Three texts were written concerning the description of a post-card which had four different scenes from Bukavu (these texts are given in Appendix D). I had intended to have the only significant differences between the three texts be non-referential, formal ones: phonological, morphosyntactic, and lexical. The first text was in the typically Bukavu vernacular but without any French words. The second was in standard Swahili structures without any French. The third had the same linguistic structures as the first except that there were many French words (the frequency of French words was higher than most, but not all, of the transcribed interviews of Bukavu speakers). After having returned from the field research, it was pointed out that the vernacular speaker incorporating little French was the only one whose text called for him to say that he did not know the location of one of the pictured scenes. It is conceivable that such a discrepancy could have influenced subjects’ evaluations of the speaker. It might have been taken as an indication that he was not from the Bukavu area; but, he, more than any of the speakers, was rated as being from the Bukavu area. It might also have 85 been considered an indication that he was not as well educated, though the likelihood of linking recognition of a certain location with educational level is questionable. However, it is a weakness of the test that this discrepancy did exist. The texts, each read by a different male speaker after practice runs, were recorded--and later played to subjects-— 5J1 the order presented in the previous paragraph. In none (xf the subjects’ evaluations was there a response such as '“this person’s speech sounds unnatural". Subjects were instructed about the testing procedure, were assured that only their opinion--not a "correct answer"--was desired, listened to the first text, and were then asked five questions about the speaker and the Swahili of the text. The playing of a text and asking of questions was repeated for the next two texts. The questions (given 1114Appendix D) elicited the subjects’ opinions on each speaker’s Swahili quality, personalityu , salary, level of 'education, and origin. Hypotheses underlying the questions were: HYPOTHESIS 15: All groups will evaluate the standard speaker as speaking the best Swahili and the one who mixes in French as having the worst Swahili. HYPOTHESIS 16: There will be a decline in the number of subjects positively evaluating the quality of standard speaker’s Swahili, from the oldest to youngest age groups. HYPOTHESIS 17: The one who mixes in French will not be as negatively evaluated by the young-age group as he is by the old-age group. 86 HYPOTHESIS 18: Evaluation of the vernacular speaker not incorporating much French will be the same for each age group. HYPOTHESIS 19: In all age groups, the vernacular speaker borrowing from the French lexicon will be most positively evaluated with regard to level of salary and level of education. HYPOTHESIS 20: From the old to young age group, there will be a significant increase in the number of subjects who positively evaluate the speaker who borrows from the French lexicon, with regard to level of salary and level of education. HYPOTHESIS 21: There will be no significant difference between evaluations of the standard speaker and the vernacular speaker who does not borrow from French with regard to level of income and level of education, except in the old-age group where the standard speaker will receive more positive evaluation. HYPOTHESIS 22: The standard speaker will be evaluated as an outsider more often by the middle and young-age groups than by the old-age group. HYPOTHESIS 23: All groups will judge the two vernacular speakers to be from the Bukavu area. HYPOTHESIS 24: In the old-age group, positive personality evaluation will be highest for the standard speaker. HYPOTHESIS 25: In the M and Y group, positive personality evaluation will be highest for the vernacular speaker who does not borrow from the French. Support of hypotheses 16, 20, 21, 24, and 25 would ixniicate that the prestige attached to the standard has diminished and, thus, that there is less motivation to learn axui speak the standard than there used to be among speakers educated before Independence. Support of hypotheses 22 and 23 would indicate that a standard speaker would be considered as geographically distanced from the Bukavu 87 resident; this, along with dense social network ties would :reenforce adherence to the vernacular norms. Support of hypotheses 17, 19 and 20 would indicate that the borrowing (xf French lexical items is perceived of as a socioeconomic jprestige marker more so than speaking standard Swahili. Support of hypothesis 25 would indicate that French ‘borrowing is perceived as reflecting a speaker’s quest of prestige more than as a necessity stemming from lack of corresponding terms in the vernacular Swahili. 2 .7.2 Questions concerning use and understanding of Swahili Subjects were asked: which language(s) they thought should be used in the primary school education; if they thought. it necessary to know standard Swahili; if they could easily use Swahili whenever they wished to; and how they rated their understanding of Swahili when speaking with their friends, when at work, when listening to the radio, and when in church. Answers to these questions were to test the following hypotheses: HYPOTHESIS 26: From the O to Y groups; there will be a decrease in the perceived value of learning Standard Swahili. HYPOTHESIS 27: In the M and Y groups, selfeva1uation of the understanding of the standard will be lower than in the 0 group. HYPOTHESIS 28: Self-evaluation of the ability to use Swahili will not differ among age groups. 88 Support of Hypothesis 26 (along with Hypothesis 5) would indicate that only French is perceived as important enough to motivate the effort necessary to learn a language well. Support of the last two hypotheses would indicate satisfaction with the vernacular variety of Swahili and, thus, lack of motivation to acquire the standard. Trudgill’s (1974:96ff) study of Norwich subjects’ self- reports on linguistic behavior indicated that subjects may over—estimate their use of standard forms if they favorably accept general social/institutional norms but under-estimate their use if they do not. The former situation would reflect the overt prestige of the standard form while the latter situation would reflect the covert prestige of the vernacular. If the vernacular-standard situation in Bukavu parallels that of Norwich and if Hypotheses 5, 26, and 27 are true, Hypothesis 27 could actually be a corollary of Hypothesis 26. Middle and young age group subjects could self-report themselves as having a lower understanding of the standard because of the lower prestige attached to it. 2.7.3 Questions concerning social patterns In the last section of the interview schedule, questions were asked which would enable scaling subjects’ Network Strength according to the criteria given in 2.5.6. Subjects were also asked which language they used with: their parents, children, friends, and work or school mates. 89 This information enabled testing of the following hypothesis: HYPOTHESIS 29: Swahili is replacing ethnic languages as the mother tongue of the great majority of Bukavu residents. Table 0.1 in the introduction to this dissertation shows that this hypothesis was supported. 2.8 Dependent variables Hypotheses 1-15 were all tested by doing statistical analyses of the effect of the independent variables discussed above on selected phonological, morphosyntactic, and lexical variables present in the subjects’ recorded speech. Sociolinguists studying inherent variation have generally accepted Labov’s criteria for selection of the linguistic variables. The variables, he stated, should: occur frequently, be structurally integrated into a larger system of functioning units, have asymmetric distribution over a wide range of social strata, and be salient but fairly immune from conscious distortion (1972:8). The variables I have selected easily meet the first three of these criteria; whether or not they meet the fourth vaguely stated criterion is more questionable but they at least are not conscious to a degree that would invalidate the study.13 The dependent linguistic variables selected are discussed in this section. Tests of the first 15 90 hypotheses, done with regard to these variables, are discussed in chapter 3. Following Labov, I use parentheses rather than slash marks in reference to a linguistic variable: for example, (j) rather than /j/. "Whereas // means that internal variation is to be disregarded as insignificant, () indicates that this variation is the prime focus of study" since the variants "show significant differences in their distribution and carry sociolinguistic information" (Labov 1972:11). Temporal and financial limitations on a study such as this necessitate focusing attention on a few variables with the assumption that the same sociohistorical factors affecting these variables may account for variation patterns in other variables. The two phonological variables that I have chosen--(w) in the class 2 (human, plural) morpheme and (j)--differ in that the former occurs in a only one morpheme while the other occurs in several. I chose to study pronunciation of the class 2 morpheme (312; because I knew that for clergymen this variable was almost exclusively realized with standard pronunciation in the social setting and frequently by them in a secular setting. Also, residents employed by foreigners would often use the standard pronunciations. However, I also knew that the non-standard pronunciation was very common. Just how common--and the effect of various social groupings on its realization--was a matter for study. 91 The (j) variable was chosen mainly because it was obvious from the outset that its pronunciation varied in several different segmental and lexical environments and, thus, enable a study of more complex interaction between phonological and social environments. The morphosyntactic variable--the copula (pi) is of interest because of its presence as a variable throughout Swahili-speaking Africa (reported in, e.g., Duran 1979 and Scotton n.d.) and because it affords study of the interaction of substratal (ethnic languages) versus innate/universal linguistic influences. Study of the lexical variable--incorporation of French into one’s speech--is relevant to an area of much contemporary sociolinguistic research, the study of social influences and linguistic restrictions on borrowing from and code-switching to a second language. I do not consider incorporation of Standard Swahili lexical items into vernacular Swahili since it is obvious from cursory inspection of the texts that this is a much less dynamic source of lexical items than French. 2.8.1 Phonological variables For the study of phonological variation, I analyzed pronunciation of the class 2 marker (gig; and of the sound represented in Standard Swahili orthography by 1. For both variables, sonorance is the key feature in question. The first is [+sonorant] in Standard pronunciation but commonly 92 {-sonorant] in the Bukavu (and, generally, Zairean) vernacular; on the other hand, (j) is [-sonorant] in Standard pronunciation but frequently [+sonorant] in the vernacular. 2.8.1.1 The class 2 marker w a- The class 2 morpheme w a- of Swahili marks plurality of human or anthropomorphized referents. It, like other class markers, is prefixed to nominals, as in (1) and (2) and to adjectives modifying a class 2 nominal, as in (3). It may also occur in either the subject or object prefix slot, as in (4). In each of the following examples, other class markers are also used to illustrate the structure of the Swahili (Bantu) class system. Orthography of the illustrations represents common vernacular pronunciations (thus, I write (w)a- as ba-. (1) mu-tu "man" bg-tu "men" ki-tu "thing" bi-tu "things" (2) u-le "that person" bg-le "those persons" (3) mu-toto mu-dogo "young child" bg-toto bg-dogo "young children" ka-toto ka-dogo "small child" tu-toto tu-dogo "small children" (4) bg-zazi bg-li-ba-ach-i-ak-a . cl.2-parent c1.2-past-cl.2—leave-applicative- -remote(?)-indicative "The parents left them there." 93 In (4) the first class 2 marker prefixed to the verb is co- referential with the subject bazazi. The second class 2 marker refers to a different group of human subjects (the children of the parents); a reflexive marker would have been used if the referents in the object slot were the same as the those in the subject slot. The [-sonorant] consonant, [b] or [b], of the vernacular class 2 morpheme alternates with the standard’s [+sonorant] [w]. Although the vernacular [-sonorant] may be either a stop or a fricative, I, for ease of reference, will use b to refer to it. Though [wa-] occurs in all written literature, is used almost categorically by media professionals, and is taught in the schools as correct, [ba—l remains a salient marker of vernacular speech. The (3);; scores were the easiest, most straightforward to calculate of any of the variables studied. In recording the data, it could be seen that there was rarely a switch in form within the same clause.14 Thus, counts could be based simply on the form of the mandatory subject prefix of a clause’s main verb.15 I counted the prefix only when attached to a verb in the present tense since this was generally the most frequently occurring tense and would practically always occur with (w)a- at least 30 times per each subject’s text. 2.8.1.2 (j) In Bukavu, the standard [-sonorant] (j) is generally pronounced as a voiced, alveopalatal affricate, [d ]; 94 sometimes the friction after the stop gives way to a near semi-vocalic quality--[dY]. The non-standard [+sonorant] variant is pronounced as the high, unrounded semi-vowel [y]. For purposes of scoring, I have counted elided consonants as, [+sonorant] since they apparently represent the end of a weakening process ([d‘]-9[d!]-9[y]-9[fl]) occuring in some of the (j) environments. The scoring of the variable (j) was more complicated than that for (gig; since various lexical as well as phonological environments affect the form. In this section, I present a general analysis of the environments and frequencies of the various forms and variation between age groups. It will be seen that an implicational scale can be constructed to relate the various environments. From this analysis, it becomes evident that focus on mo(j)a "one" and (1122 "know" may be fruitful for purposes of the more detailed analyses that will be involved in testing the various hypotheses. Each of these words: (1) occur/s frequently, (2) has a high degree of variability in its pronunciation, (3) differs from the other in phonological environment, and (4) has developed differently diachronically. Before we consider the data, I shall sketch the theoretical background to use of implicational scales in the study of variation. 95 2.8.1.2.1 Two approaches to variation Linguistic variation may be studied with focus on social groups (as in, for example, Labov (1972), Cedergren and Sankoff (1974), Sankoff (1980), and Guy (1980)) or on the individual (as in, for example, DeCamp (1971), Bickerton 1971, DeCamp (1971), and Bailey (1973) and LePage (1980)). The working assumptions of these two groups of researchers are quite different. Those who focus on the social group assume that individuals within a speech community share a "group grammar" which enables them to know the probability of a certain variant’s occurrence in various environments. There may be rules that have persisted for centuries since variation is an inherent part of any grammatical system. Those who focus on the individual argue that there is no group grammar but only individual grammars which stem from one person’s daily interaction with others. One individual in a speech community may have a set of rules hat is different from another in the same speech community. Variation is not a stable part of a grammatical system. Mental calculation of the probability of a certain variable’s use is viewed as an absurdity (Bickerton 1971; similarly, Trudgill 1981:160)1‘. So, instead of variation being represented as part of the grammatical system, it is viewed as a transition between categorical rules. Thus, rather than probabilistic rules, implicational scales are developed: Insofar as lects differ in terms of phenomena which can be placed on an implicational scale, they can be 96 uniquely designated by the point at which they occur on such scales. The phenomena whose presence or absence characterizes them implies the presence or absence of other phenomena lower on the scale and is implied by, but does not imply, phenomena located higher up on the scale. (Bailey 1973:28) I construct rough-grained implicational scales for the (j) phonological variable in section 2.8.1.2.3. In this section, it is shown how the implicational scale may be useful in sorting data and grouping environments which affect the frequency of a variant’s occurrence. Furthermore, a study of Appendices F and G indicates that over-all variation in several environments is the reflex of variation in grouped individuals’ speech but that the speech of each individual shows variation in only a few environments; however, the limited number of occurrences in some environments for individual subjects prohibits a firm conclusion. Guy (1980:12) says that linguistic behavior which can be described by implicational scales is to be expected in situations "where there is independent evidence of historical or ongoing change" and not in more stable linguistic situations such as in American English where final stOp deletion has persisted for a long period of time (he does not say how long) for a wide range of social groups. This points to perhaps the most important difference between the two approaches to the study of variation. Those who focus on the individual have focused on heterogeneous speech communities where there has been, or 97 is, intensive contact between speakers of different language and socio-cultural backgrounds. Those who have focused on the group have focused on much more homogeneous situations in which long-term, clear-cut social stratification is involved and there is much less linguistic diversity between groups. 2.8.1.2.2 Phonological environments and frequencies The two basic segmental environments affecting sonorance of the (j) variant are: the vowel before which it occurs and its syllabic position in the word. (J) occurs only in the initial position of CV syllables. The occurrence of (j) after a syllabic nasal consonant (as in nSjita "road") is not considered in the following analysis since it is categorically pronounced as [-sonorant]. In Swahili, the penultimate syllabe of a word always receives primary stress. For brevity’s sake, I will state (j) environments in terms of whether (j) occurs before, after, or in "stress(-ed position)" to indicate the position of the syllable in which (j) occurs with reference to the stressed syllable. Two other factors affecting the sonorancy of (j) are the morphological function and, especially, the frequency of occurrence of the lexical item containing the variable. During my frequency counts of the variants, it became evident that a quite misleading picture of the variation 98 would result if only the segmental environments were considered. For example, if all lexical items in which (j) occurs before /i/ word finally are lumped together, the [-sonorant] variant occurs 50% percent of the time. However, this would obscure the much higher [-sonorant] frequency if ma i is not counted: in words other than 921; it occurs 96% of the time but in ma(j)i it occurs only 20% of the time. Several other examples of this type of distinction will be seen below. Thus, in setting up environments, I often counted specific, frequently occuring words or morphemes as separate environments while grouping together less frequent words or morphemes in which (1) occurs in segmentally identical environments.17 The environments that were ultimately distinguished are given in Table 2.7. In the next section (2.8.1.2.3), I construct an implicational scale which suggests how these environments should be grouped and offer an explanation of what has determined the shape of the scale. 99 Total oc- % realized as Environment currences [-sonorant1 .(j)i before stress 1209 100 ex. -jifunza "study (vb.)" -ji- reflexive vb. prf. (j)i in stressed position 376 98 ex. -jibu "answer" (j)i after stress maji "water" 385 20 all other occurrencesu 245 96% ex. mubeleji "Belgian" mukaaji "resident" kijiji "village" (j)e in stressed positionu jembe "hoe" 39 77% -jenga "build" 236 42% -(j)a word or Aux finally¢° -ja- "negative imperfect" 550 7 (verb stem prefix, unstressed) -kuja "come" 2472 8 -ngoja "wait 188 15‘ (-)mdja "one" 3096 16 (j)a before stress 252 99 ex. jamaa "family" . (j)a in stressed position ki/vi-(j)ana "youth(s)" 349 61 ma(j)ani "grass; leaves" 151 20 -(j)ala/-jaza "be full/fill" 87 8 all other occurrences 139 89 ex. ajabu "amazing phenomenon' (J)u in stressed position21 -jua "know" 2636 sex (j)o before stress ex. jogoo "rooster" 4 100% (j)o stressed ex. mujomba "uncle" 61 18% -koyola "urinate"22 (j)o after stress fujoH "problem; disturbance" 237 74% Table 2.7.' Environments considered in analysis of (j). 100 2.8.1.2.3 Constructing an implicational scale The first step in constructing an implicational scale was to arrange in ascending order the environments according to percentage of [-sonorant] realizations of the variant. Since a basic concern of my research is to look for diachronic developments in Bukavu residents’ speech, I grouped subjects according to age, as discussed in section 2.5.1, and constructed the scales based on the overall percentages for each age group. Ithill be seen in the following discussion that the scales for the middle and youngest-age groups are more similar to each other than to the oldest-age group’s scale. This is not necessarily due to age differences alone. The subjects in the oldest-age group were fairly evenly split between those without any formal education and those with at least some. Also, most oldest-age group subjects who were not Shi were not born in Bukavu. In contrast, the large majority of subjects in the middle and youngest-age groups were born in Bukavu (regardless of ethnic group membership) and have some formal education. The reasons for these disparities are discussed in sections 2.5.1 and 2.5.2. There may have been greater similarity between the oldest and other age groups’ scales if the disparities had been factored out. However, since the implicational scales are used primarily as a heuristic tool rather than a means of testing the major hypotheses and since I assume that social N 101 conditions encouraged greater heterogeneity of Swahili varieties among pre-Independence Bukavu Swahili residents than among Bukavu natives growing up around Independence and later, I settle for this rough comparison between age groups. In the testing of Hypotheses reported on in chapter 3, the handling of these disparities is much more refined. It was assumed that building an implicational scale according to age group would facilitate constructing scales for individuals within those groups as well as provide a general impression of whether or not the scales were similar between age groups. This assumption proved to be valid for each age group. The percentages of [-sonorant] occurrences for each environment were arranged in ascending order. The resulting scales were then compared. The results are given in Table 2.8. 102 YOUNG MIDDLE OLD 1. -jala (0) -ja- "not yet" (09) -kuja 2. -ja- "not yet" (2) -jala (11) -ja- "not yet" 3. -kuja (5) -kuja (11) majani 4. (-)moja (8) (-)moja (14) maji 5. -ngoja (11) -ngoja (16) -ngoja 6. maji (11) -jo— stressed (17) -jenga 7. -jua (56) majani (26) moja 8. fujo (78) maji (38) -jua 9. ja/stress (97) -jua (38) kijana 10. ja/__stress (100) -jenga (65) fujo 11. ji/stress (100) kijana (69) ja/stress 12. ji/__stress (100) fujo (72) ji/stress 13. ja/stress (93) ji/stress__ 14. ji/stress__ (97) ja/__stress 15. ja/__stress (99) ji/__stress 16. ji/stress (100) 17. ji/__stress (100) Table 2.8. Implicational scales based on percentage of [-sonorant] forms of the (j) variable (percentage is given in parentheses). Environments were not listed for an age group if they occurred less than 30 times in that age group. When the order of environments is compared between age groups, based on Table 2.8, there are several ordering discrepancies as can be seen in Figure 2.1 in which lines connect identical environments. For example, -ja1a is #1 for the youngest-age group’s ordering of environments but #2 for the middle-age group; thus, in Figure 2.1, a line connects the "Young" #1 with "Middle" #2. This line crosses over the line connecting the "Young" #2-- -ja— --with "Middle" #1-- -ja-. Each line crossing over AAAAAAAAAAAAA mmwoqqmwnuwwwaa QQGMQNNDGHQOOQQ vvvvvvvvvvvvvvv 103 Young: 1 2 3 4 5 6 7 8 9 10 11 12 Middle: 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 Old: 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 Figure 2.1. Relative sequences of environments.‘ Numbers correspond with environment listed in row of Table 2.7 for relevant age group. Lines connect identical environments. another represents a sequential disagreement between age groups. However, when differences between environmental occurences of the [~sonorant] variant are checked for significance (chi-square tests at .05 level), the sequences between age groups can be seen to be much more parallel if one age group’s ordering of environments with insignificant differences between them is rearranged to parallel another age group’s ordering in which the differences are significant (Figure 2.2). For example, in Figure 2.1 there Young: 1 2 3 4 5 6 7 8 9 10 11 12 Middle: 2 1 3 4 5 7 63‘ 8 9 10 11 12 13 15 14 16 17 Old: 2 1 4 5 3 7 6 8 9 10 11 14 13 12 15 Figure 2.2. Sequences rearranged for greater parallelism if differences between occurrences within age groups are not significant. is a cross-over between environments 1 (—(j)ala) and 2 104 (-(j)a-) for the Young and Middle age groups. But, since there is not a significant difference between the scores of the two environments in the Middle age group, the ordering of the two environments may be reversed. As a result, there are no cross—overs involving these environments between any of the age groups. With this revision, there are no cross-overs between the young and middle age groups but their orderings contrast with that of the old age group’s. Already this indicates justification for hypotheses which predicted that significant differences in linguistic behavior would be found mainly in the old age group as opposed to the other two. The environments between which cross—overs occur and percentage of the [-sonorant] realizations for each age group are given in Table 2.9. 91g Middle Young maji (10) (-)moja (14) (-)moja (8) -nsoJa (15) -nsoja (16) -ngoja (11) majani (10) majani (26) (-)moja (26) jo/stress (17) maji (38) maji (11) -jenga (21) -jua (38) -jua (56) -jua (29) -jenga (65) Table 2.9. Revised cross-over environments specified. (percentage of [-sonorant] pronunciations are in parentheses.) Two aspects of Table 2.9, in conjunction with Figure 2.2, are particularly noteworthy. First, the lack of cross-overs occuring on the extreme edges of the percentile. 105 ranges indicates a long-term stability in these areas: the non-standard [+sonorant] variant of (j) has long been preferred in -(j)a1a, -(j)a-, and -ku(j)a and the Standard [-sonorant] variant has long been preferred before /i/ except in p51;. Second, the young age group’s scores continue in the diachronic direction of change apparent in the old and middle age group switch-overs concerning the two most frequently used words -mo(j)a and -(j)ua. This observation then leads to the notice of an apparent "great divide" in the diachronic development of (j). The line connecting the Older and Middle-age groups’ -mo ' a scores represents a diachronic decrease in standard, [-sonorant] scores for mo(j)a and all words beneath it on the middle age group’s implicational scale; however, the line connecting these age groups’ :jpg scores represents a diachronic increase in [- sonorant] scores for -(j)ua and all words above it on the middle-age group’s implicational scale (Table 25191 106 Age Group Is difference Environment g;_ Middle Young Significant (.05): Upper end ji/__stress 98 100 100 No ji/stress 95 100 100 Yes ja/__stress 97 99 100 Yes ja/stress 77 93 97 Yes fujo 72 72 78 No jua 29 38 56 Yes Lower end moja 26 14 8 Yes kuja 6 112: 5 Yes -ja- (neg.) 7 93' 2 Yes -jala 11 0 No Table 230 ‘Percentage of [-sonorant] pronunciations by age group on lower and upper ends of the implicational scale. "Lower end" has lowest [-sonorant] scores; "Higher end" has highest [-sonorant] scores. So far, the implicational scale has been discussed only in terms of overall scores for each age group. Analysis of the tables in Appendix G indicates that the overall scores offer a good summary of individual subject’s scores as arranged by environment on an implicational scale. However, the correlation between individual’s implicational scales with their corresponding age group’s general implicational scale is inversely related to age. That is, a parallel between an individual’s and an age group’s sequencing of environments is most likely to occur for young age group subjects and least likely to occur for subjects of the oldest age group. Moreover, the percentage of variants within one environment also decreases with each generation. Tables 2.11 and 2.12 exemplify how this count was done. In Table 2.11 , the exact counts of (j) variants are given 107 for three subjects, each from a different age group". Here, for 5 l 3 3 i 3 E Z 9 2 19 ll 12 13 1! 1‘15 l§ v - o-t H: H: H - M: 9-4 - - H M - - 3—o 1H n H No 0-13 H: M - - 5-31 - - 1-1 - - - 2-o ns-o 0 - 2- 2 3' l 46- I - l- 0 9- 1 4-0 23-0 - 8-0 6-0 ll-O 41-0 29-0 Table 2.11. Total of l-soeoraetl (before dash) v. [+sonorant] (alter dash) occurrences 01 (j) (or three individuals. 8- Sebject; Y-froe Young age groan; nsfroe nlddle age group; oslroe Oldest age group. Under- lined eeebers represent a particular environeeet. example, the Y subject did not use the word —(j)ala "be full" (Environment #1), used —ku a "come" (Environment #3) eleven times, pronouncing (j) as [+sonorant] each time, and used -(j)ua "know" (Environment #8) thirteen times, pronouncing (1) as {-sonorant] nine times and as [+sonorant] four times. These numerical scores were then coded:_"C" if there was a categorical realization of one variant 33d the variant occurred more than five times; and "M" if there was "mixing" of variants-~more than one variant was used (no minimal frequency limit). Thus, scores were assigned as in Table 2.12. 108 § .1. .3. .31 i. .5. Q 1 11 2121113131115. A C+ C+ C+ M-_ I C+ C+ C+ M+ C+ C+ M+ C- C- P M+ M+ M+ C- c- Table 2.12 Coding of individual subjects’ (j) scores. C=Categorical(only one variant used; M=Mixed (both variants occur); + = [+sonorant] realization is favored; - = [-sonorant] realization is favored. Blank spaces indicate that there were not enough occurrences of the variants in the environment to be considered for the tally. Scores for each age were then gained by totalling the numbers of C’s and M’s for each age group. The result is given in Table 2.13. Old Middle Young Both variants (M) 112 (41%) 125 (36%) 43 (29%) One variant only (C) 163 (59%) 227 (64%) 103 (71%) Table 2.13. Occurrence of variants within environments specified in Table 2.6, by age group. p< .001. Table 2.13 indicates that there has been a decrease in mixing of variants from the oldest to youngest generation. Another tally of those in the old age group was done to see if there was a significant difference between the scores of those with formal education and those without. There was not (p:>.50). Also, scores of only subjects with formal education were compared cross-generationally. There was no significant difference between this tally and that represented in Table 2.13. 109 Another tally was done to consider environments only if there were 10 or more chances and to count an environment as . "categorical" if there were 5% or less realizations of one variant. These limitations were made in view of comments elsewhere in the sociolinguistic literature concerning representativeness of samples and percentage of errors in real speech. Guy (1980:20) concluded that "above 10 tokens [in a factor group] there is 90% conformity with expected pattern, whereas below 10 tokens only 63% of the relationships are as expected". A statement four sentences later suggests that, in this quote, he meant to say "10 tokens or more" rather than "above 10 tokens". The limitation "5%" accords with the margin of error frequently accepted in statistical studies. It is also close to the average of the two percentage figures Labov (1972:203) cited in his argument against the "myth" of "the ungrammaticality of everyday speech": When rules of ellipsis are applied, and certain universal editing rules to take care of stammering and false starts, the proportion of truly ungrammatical and ill-formed sentences falls to less than two percent (Labov 1966b). When nonacademic speakers are talking about subjects they know well--narratives of personal experience--the proportion of sentences that need any editing at all in order to be well-formed drops to about ten percent. I have received confirmation of this general view from a great many other linguists who have worked with ordinary conversation. With this adjustment, the gross scores showed no significant difference (p) .50) between the Old and Middle age groups. The Young age group’s score is lower than the 110 Middle’s (p< .05) but less than the Old with the probability only less than .20. However, when the difference between scores of those with and without education in the Old age group are considered, there is evidence of decline in percentage of environments in which variation occurs from the Old to the Young educated subjects. Within the Old-age group, both variants are found in 44K% of the cells for those with formal education but in only 19% of the cells for those without formal education (p<:.01). Thus, among subjects with formal education, as Table 2.14 shows, there seems to be a steady_decrease in the occurrence of both variants within a cell from the old to young age groups such that the young age group’s categorization of rules approaches that of the old age group’s subjects who had no formal education. Old Middle Young Old (+ED) (+ED) (+ED) (-ED) <, 5% variation 30 (44%) 68 (37%) 23 (25%) 10 (19%) > 5%1variation 38 (56%) 114 (63%) 7O (75%) 42 (81%) Table 2.14 Variation in environments specified in Table 2.6, by age group. Cells counted only if 10 or more chances. p<:.001. whether or not scores of Old subjects without formal education are counted. It is also of interest to note how close the percentages of Table 2.13 are to those in Table 2.14, where cells were counted with much fewer entries and no percentage restriction on mixing. However, the greater restriction on what should be counted for Table 2.14 results in observation of the split between educated and non-educated subjects’ 111 scores in the old age group (a difference between these two sub-groups’ linguistic behavior was predicted in several of the hypotheses). In contrast to what Guy’s observation, quoted above, would lead us to expect, there is no significant difference between the overall totals of the two tables even though the one is based on counts of cells with fewer environments. Furthermore, restriction to consideration of environments in which there are 10 or more chances results in less environments exhibiting variation rather than more. The decrease from the old to young age groups in mixing of variants within specific environments and in the irregularity of scaling patterns suggests that the phonological variation is part of a process of historical change leading to categorical rules. The differences in scores between groups indicates that the middle and younger generations had to choose between the more stable patterns of those without formal education and the less stable patterns of those with formal education. Although Table 2.14 indicates a degree of categorization of rules for the Young that approaches that of the Old without formal education, TablelLlO indicates that the choice of variant has moved towards the [+sonorant] favored by the Old without formal education in one set of environments but towards the [-sonorant] favored by the Old giph formal education in another set of environments. 112 Apart from considerations of social evaluation of speech varieties, why does the split occur in this way? Segmental environment, stress, and lexical environment may all have an in influence. The segmental environment of (j) may be taken as the starting point for an explanation of (j)’s realizations. Table 2.15, based on the conglomerated scores of all subjects.indicates the likelihood of (j) being {-sonorant] (standard pronunciation) when segmental environment alone is considered. % of [-sonorant] Total Environment occgrrenceg (j) occurrences __i 85 2215 __o 63 302 __e 47 275 __u 38 2636 __a 19 7284 Table 2.15. [-sonorant] realizations of (j), according to segmental environment. Figures are based on conglomerated scores of all subjects. Environments are in descending order of % of {-sonorant] occurrences. Only one natural grouping of environments listed in Table 2J5 is possible when the frequencies are considered: (j) is much less likely to be [-sonorant] when followed by a [+low] vowel than when followed by a [-low] vowel. The percentage of [-sonorant] occurrences gives no evidence for a [+front] versus [-front] distinction or a [+high] versus [~high, -low] distinction. 113 The most extreme difference in segmental scores is that between the "__i" and "__a" environments. Perhaps an explanation of this could be construed in terms of the common Swahili CV pattern (Appendix A, section 1.3 gives examples where standard non-CV patterns are commonly realized as CV in Zairean Swahili vernaculars). The sonorization of (j) before li/ would result in near- acoustical identity of the two segments. This could lead to the CV pattern [yi] becoming V only, through the elipsis of [y]. This in turn would lead to homonymony of previously distinct morphemes as illustrated in 1 and 2. 1. a-na-ji-ona a—na-zi-ona a-na-i—ona he-pres.-himself-see "He sees himself." 2. a-na-_i-ona he-pres.-ip-see "He sees it." In the sonorization of (j) before /a/, however, an acoustical difference between the segments is maintained and, thus, the CV syllable pattern is protected. However, when (j) scores are broken down as in Tables _2.8 and 2.10 above, it is clear that factors other than segmental environment must be considered. For example, (j) is [+sonorant] more frequently when it occurs before /i/ in ma(j)i "water" than when it occurs before /a/ if the syllable is in pre-stress position. Furthermore, there is a great difference in (j) scores within the segmental environment "__a". If (j) occurs before /a/ in word or Aux final position, its highest {-sonorant] score is 26% in the 114 oldest age group, 16% in the middle, and 11% in the youngest. Before stress, though, the lowest score of any age group is 97%. This indicates that the weakening process ([d§]~9[d¥] [y]~’[D]) is most apt to occur in the syllabically weak post-stress position. This is the syllable position in which vowels are commonly elided when followed by a word beginning with a vowel, not only in Bukavu Swahili (5) but in other Bantu languages such as Kinyarwanda (6). (5) kweny’ unaneda (kwenye unaenda) place.where you.are.going "where you are going" (6) Umugab’ arway’ umutwe. (Umugabo arwaye umutwe) man is.sick head "The man has a headache." In Bukavu Swahili, the elision of a final syllable" is 'fairly common in fast speech. In verbs, the final syllable {-ka] , a tense/aspect marker39, is frequently dropped and apparently evident only by the maintenance of stress in the preceding syllable as illustrated in (7).3° (7) ye’ aliisha kufa njoo mwe’ alinifundisha’ (yeye aliisha kufa njoo mwenye alinifundishaka he already die so who he.taught.me kazi ya mbabula. work of heater "He already died; it’s him who taught me how to make charcoal heaters." On the other hand, the [-sonorant] variant is more likely to be retained in stressed or pre-stress positions. 115 Further evidence for the weakness of syllable final position may be found in consideration of the difference between (j) scores for p91; "water" and the scores for other environments in which (j) occurs before /i/. The post- stress (j) in p21; is, for any age group, [-sonorant] at least 62% less frequently than is (j) before /i/ in stressed or pre-stress position (footnote 19 suggested an explanation for the discrepancy in (j) scores between ppji and other words in which (j) occurs before /i/ in post-stress position. Finally, with regard to stress, whether (j) occurs before /i/ or /a/, it always has the highest [-sonorant] (standard) score in pre-stress position than in any other environment. Table 2L16 indicates whether the differences are significant, according to environment and age group. Significant difference (p<..05) from pre-stress (j) score? Environment Oldest Middle Ybungest stressed (j)i No* No -- post-stress (j)i No Yes -- other (j)i Yes Yes ~- stressed (j)a Yes No* No post-stress (j)a Yes Yes Yes othe (j)a Yes . Yes Yes Table 2.16. Significance of difference between pre- stress (j) scores and scores in other environments with regard to stress. 3 : p < .10 The third major influence on sonorance is lexical environment. Apart from kijana "youth", pajapi "leaf; grass", and :jglg "fill", the (j) before /a/ in stressed 116 position is [+sonorant] in 93% of the middle age group’s occurrences and 97% of the youngest age group’s (Table 2.8, above). In the stressed position of -'ala, however, it is [+sonorant] in only 11% of the middle-age group’s 'occurrences and 0% in the youngest-age group’s. The other two isolated lexical items with (j) in stressed position are between these extremes. Frequency of a lexical item’s occurrence in discourse may have some effect on these differences: the isolated lexical environments occur much more frequently (at least, in my data base) than any one word in the environments that group words according to stress. In harmony with this suggestion, are the observations (based on Table 2.7 above) that: p21; "water" occurs much more frequently than any other word having (j) before /i/ in post-stress position; and, of the two words counted for the "(j)a in stressed position", the most frequently occuring word is the one in which (j) is most frequently [+sonorant]. However, inspection of lexical items having (j) occur before other vowels and in different stress positions shows that the factor of frequency by itself cannot account for the differences in realization of (j). For example, ppjg "one" occurs much more frequently than kpjg "come" but (j) is [-sonorant] more frequently in the former than the latter. The relative strength of the influences of segmental environment, stress, lexical item, and frequency of occurrence is difficult to determine because of the uneven 117 frequencies with which the various environments occur. For example, the occurrence of (j) only in stressed position before /u/ and the relatively low occurrences and limited environments of (j) followed by mid vowels seems to prohibit conclusive findings with regard to the influence of stress and how particular lexical items affect sonorance. However, based solely on study of (j)’s realizations before /i/ and /a/, the interplay of the various influences might be represented as in Figure 2.3. Likelihood of {-sonorant] realization Greatest Postion relative i Quality of to stress following vowel Pre-stress L High, Front e x Stressed i c o Post-stress n Low Least Figure 2.3 Likelihood of [-sonorant] realization of (j). The middle range [—sonorant] scores for (j) in :jpg "know" (29% in the oldest-age group, 38% in the middle, and 56% in the youngest) are consistent with this scheme since (j) is in stressed position, high, but not fronted. Similarly, combined score for jgppg "hoe" and :jgpgg "build", the two words in which (j) realizations were counted before /e/ is 118 in the middle range—-47%; the (j) occurs in stressed position and the vowel is [+front] but not [+high]. A question rising from this analysis is: is it in environments with these middle range scores--where linguistic environment does not have as clear-cut an effect on realization of the variable--that social evaluation of language varieties may have the most effect on its realization? 2.8.2 Morphosyntactic variable For testing the hypotheses on a morphosyntactic level, I concentrated on the variation between the use of pi and :59 as copulas. Very little attention has been given to this variation though differences between vernacular and standard uses of these copulas are found throughout Swahili speaking Africa (Gilman 1979; Duran 1979; Scotton MS). In Standard Swahili, :59 is one of three copula forms used only in locative references. Use of the three forms is determined, as 1-3 illustrate, by how precisely the speaker refers to the location of the referent in subject position. Both :pp and :32 (the latter always used to refer to an entity enclosed within a larger entity) are used to refer to locations that are more precisely designated than that referred to by :53. (1) Kitabu ki-po hapa mezani. "The book is right here on the table." (2) Kitabu ki-ko huku mezani. "The book is somewhere here on the table." 119 (3) Kitabu ki-mo mfukoni. "The book is in the bag." The copula pi is used in non-locative environments: (4) Kile ni kitabu changu. "That is my book." (5) Kile kitabu ni kizuri. "That book is good." (6) Furaha ni mwalimu. "Furaha is a teacher." In the Bukavu vernacular, :52 is generally used as the sole "locative" form but it may be used in more than a locative sense as examples given in the discussion below will show. If one asks a Bukavu resident about the difference between the two copulas :39 and pi, he will likely be told that they are used interchangably. Based on my two years of residence in Bukavu before doing field research for this dissertation, I had accepted this. layperson’s analysis and felt that :59 occurred much more frequently than pi. I would probably have accepted Gilman’s claim that "all types of ZS [Zairean Swahili] use a single copula [:52] with both noun and locative" (1979:103), dismissing exceptions as code switches to the standard. However, it is clear that both the lay analysis and Gilman’s claim are false: there are two forms of the copula and the choice of copula is affected by the environment in which it occurs. I have yet to arrive at an accurate, economical, and comprehensive statement of the environments affecting which form is used. For this dissertation, I will consider their 120 occurrences in three environments which, my preliminary analysis has indicated, represent low, middle, and high range Standard scores. Before discussing these environments, though, I shall mention environments in which there is nearly categorical use of non-standard :59, in Bukavu Swahili. 2.8.2.1 Environments with nearly categorical use of :52 From the outset of my analysis of Bukavu Swahili, it was clear that there were certain environments in which :59 was used almost exclusively: locative predications (7a), first or second person subject (8a,9a), positive possession (10a; standard forms are used in the negative counterparts), and negated copula (11a). (The last two structures are illustrated, even though the standard form does not have pi in it, since, in Bukavu Swahili, the copula :32 is used rather than the standard form.) The standard forms (the "b" counterparts in 7 to 12) would be heard in these environments very rarely. (7) a. Furaha i-ko mu nyumba. Furaha 3sg.-be in house b. Furaha yu-mo nyumba-ni. Furaha 3sg.-be house—in "Furaha is in the house." (8) a. Ni-ko muzee sana. I-be old very. b. Mimi ni mzee sana I be old very "I am very old" or "I am a very old man". 121 (9) a. Tu-ko barega. We-are Rega. b. Sisi ni Warega We ' be Rega "We are Rega." (10) a. Bi-ko na batoto tano. They-be with children five b. Ba-na watoto tano. They-with children five "They have five children." (11) a. Ha-i-ko muzuri zaidi Neg.-it-is good more b. Si nzuri zaidi Not good more "It isn’t better." (12) a. Ha-i-ko bibi yangu Neg.-she-is wife my b. Yeye si mke wangu She not wife my "She is not my wife." Apart from these environments, though, the rules governing selection of pi vs. :59 are far from clear. Furthermore, there are several examples of variation in nearly identical environments--at a clausal level or lower, at least (there may be discourse restraints on their use). For example, one person said both 13 and 1431 and another said both 15 and 16. (13) baba na mama ni bakristu. father and mother be Christians "Father and Mother are Christians." 122 (14) Maman i-ko mukristu kabisa. Mother she-be Christian completely "Mother is a real Christian." (15) batu balikimbia kwa juu ni kitu kyenyi people they.ran for reason be thing which kilifika, ni kitu kigeni. arrived be thing foreign "People ran since it was something which had arrived, it was something foreign." (16) watu walijua kwamba barebelle biko people knew that rebels they-be kwa juu iko kitu kyenyi kilijulikana for reason it-be something which it.was.known depuis lors. since that.time "People knew that the rebels were around since it was something that had been known for a that time." 2.8.2.2 Environments used in hypothesis testing The environments selected to test the hypotheses in chapter 3 were done so on the basis of a preliminary analysis which indicated that :59 is much more likely to be used than pi when the complement is not a Noun Phrase but that a simple [-NP] vs. [+NP] dichotomy between complements would be too general, especially with regard to the range of environments that could fall within the [+NP complement] category; here, Lexical, syntactic, and/or semantic differences between structures correlate with a wide range: in frequencies of use of the copula in its standard form. So, while keeping [-NP complement] as the environment representing low range pi scores, I have chosen two environments that involve nominal complements that represent .: 123 middle and high range pi scores. (These environments do not include all occurrences of a copula with a nominal complement.) The middle-range scores occur in environments in which the subject is [+Human] and the complement refers to the occupation, ethnic group, nationality, or religious commitment of the subject. Both the subject and complement are filled by a simple NP (pronoun or noun not modified by a relative clause though it may be modified by an adjective or prepositional phrase). I will refer to this environment as "[eeSimpleJ". The high-range scores occur in environments which are more syntactically complex than those of the middle-range scores. The copula is preceded or followed by a NP modified by a relative clause (17-19) or is preceded or followed by an untensed clause (20). (17) pi inchi ingine yenye ilikuwa ilibatuma. be country other which it.was sending.them "It is an other country which was sending them." (18) Ile yote ilikuwa inapitikana tu ni mafungu that all it.was happening only be vacation tu. only "The time when all that was happening was vacation." (19) Maisha ni mtu mwenye kuyua fulani huyu. life be person who know someone this.one "The one with a good position in life is the person who knows a certain someone." (20) Kazi yangu tu pi kushinda nazunguruka. work my only be to.conquer going.around "My work is just going from here to there." 124 For the sake of testing the hypotheses, these structures shall be lumped together and considered as one environment which shall be referred to as "[urComplex1".32 There is no overlap between the middle and high-range environments but there could be overlap between the high- range and low-range environments. (21), for example, is (21) ile imbo yenye aliimba ni muzuri. that song that he.sang be good "That song he sang is pretty." [upComplex] since the subject noun is modified by a relative clause but it is also [-Nom Comp] since the complement is an adjective. In such cases, the complement was scored as occurring in both environments. 2.8.2.3 Ethnic language influence on realiztion of the variable The influence of ethnic languages on the distribution of ipii variants in Bukavu Swahili is worth investigating because of the many parallels between ethnic language and vernacular Swahili uses of the copula. Furthermore, though, as noted above, linguists have pointed out the vernacular use of :39 in non-standard ways, none have considered how ethnic languages may have affected this extension. Here, I shall indicate some of the parallels but an in-depth study shall remain for another time. Many languages spoken in Eastern Zaire use a copula that is parallel in structure to the Swahili :52 and somewhat parallel in function: the copula is prefixed by a class marker agreeing with the 125 subject and it is used in both locative and non-locative references. However, it is difficult to find ethnic language structural and functional parallels to the Swahili pi, In the introduction to this section, four environments which have practically categorical realization of the :52 variant were given. The first was when the reference to the subject is with regard to its location. 1-5, below, show verb forms of languages spoken in Eastern Zaire that parallel the :52 structure and are used in locative reference; all have a class marker prefixed to the copula morpheme. 1. SS: Wa-tu wa-ko nyumbani. cl.2-entity cl.2-be at.house "People are at the house." 2. BS: Ba-tu bi-ko ku nyuma cl.2—entity cl.2.be at house "People are at the house." 3. Mashi: Aba-tu ba-ali omu buarho cl.2-entity cl.2-be in canoe "People are in the canoe." 4. Rwanda: Aba-ntu ba-ri mu nzu. cl.2-entity cl.2-be in house "People are in the house." 5. Rega: Ki-tabo ki-li mu numba cl.7-book cl.7-be in house "The book is in the house." In the other three environments, as 6-16 show, the ethnic languages have a structure which parallels the Bukavu Swahili form, ungrammatical in Standard Swahili: 126 First and Second Person 6. SS: Mimi ni mzee sana. tNi-ko I be old very *1sg.-be "I am very old." 7. BVS: Ni-ko muzee sana. lsg-be old very "I am very old." 8. Kirega: Ni-li mwigisya. lsg-be teacher "I am a teacher." 9. Kibembe: U-le m’mikeca. ng-be teacher "You are a teacher." 10. Mashi: Rhu-li banji. lpl-be many. "We are many." Possession 11. BVS: Baba leo i-ko na franga. Father today Ss-be with money "Father has money today." 12. Kirega: muntu e-li na matuzi man 3s-be with ears "a man has ears" 13. Kibembe: A-le na Wamwitu 3s-be with spirits "He has evil spirits" Negated copula 14. BVS: Ha-i-ko bibi yangu. neg-3s-be wife my "She isn’t my wife." 15. Kirega: Kasongo nt-eli mwigisya Kasongo neg-be teacher "Kasongo isn’t a teacher." 16. Kibembe: Kasongo ta-le m’mikeca. Kasongo neg—be teacher. "Kasongo isn’t a teacher." 127 Another non-standard use of :99 is in forming the present continuative. This was not mentioned in the introduction since it would vary with the use of a Standard verb affix, not with pi. This use of :59 as an aspect marker also has parallels in the local ethnic languages (17- 19): Present continuative 17. BSV: Ni-ko naenda Is-be going "I am going." 18. Kibembe: Ni-le miuiya Is-be going "I am going." 19. Mashi: ba-li bakweeba orhueeshe omu miishi. 3p-be throwing nets into water "They are casting their nets into the water." Ethnic language equivalents of -k0 are also found in environments where the vernacular Swahili varies between pi and -k0 forms. For example: gefore podifiers 18. BVS: Maisha ya leo i-ko ngufu. life of today 3s-be hard. "Today’s life is hard." 19. Kirega: Kalamo ka 1elo ke-li kabibo. Life of today 3s-be hard. "Today’s life is hard." 20. Kibembe: Hukyo a-le wasona.33 That.person 3s-be bad "That person is bad." 128 gefore a Noun 21. BVS: I-ko mwalimu. 3s-be teacher "He is a teacher. 22. Kirega: e-li mwigisya 3s-be teacher "He is a teacher." 23. Kibembe: a-le m’mikeca 3s-be teacher "He is a teacher." 24. Mashi: Ogu gu-li mukoo gwaani this 3s-be blood mine "This is my blood." However, in languages close to the Bukavu area it is difficult to find forms that parallel the Swahili pi which, unlike -ko, does not have an agreeing class marker prefixed to it (8). 25. SS: Mu-alimu w-angu ni Maria. cl.l-teacher cl.l-poss. be Maria "My teacher is Maria." 26. BS: Mu-alimu w-angu ni Maria. cl.l-Teacher cl.l-poss. be Maria. "My teacher is Maria." The only obvious parallel that I know of is the Kinyarwanda _i that is restricted to use in the third person, as in 27 and 28. (27) Umu-alimu w-angye ni Maria. tya-ri cl.l(3sg.)-teacher cl.l-poss. be Maria icl.l-be "My teacher is Maria." (28) Wowe u-ri Maria. *ni You 28g.-be Maria. *be "You are Maria. 129 It may be that linguistic universals also affect realization of the (ni) variants. Bickerton (e.g., 1981) has argued that in prototypical creole languages innate linguistic tendencies are most likely to be realized in speech since the speakers for whom a pidgin became a mother tongue had no clear grammatical paradigms as models for acquisition and thus made many grammatical distinctions based on innate linguistic tendencies. In an expansion of Bickerton’s approach, Scotton (n.d.) argues that universal linguistic tendencies are likely to surface not only in creoles but "in any situation in which the language learner is not constrained by already existing rules" (p.21) and that: It seems likely second language situations can be considered alongside creole development as instances of such situations, as well as cases of dialect divergence from a standard if the socio- psychological conditions do not favor the standard as a prescriptive force. (p.21) The main source of data for her argument is from the speech of Nairobi non-standard Swahili speakers who extend -ko beyond its standard use as a locative copula and use it in "expressing a state--and especially a temporary or acquired state" (p.5). But she also points to similar uses of locative copulas in both creole and non-creole languages and suggests that the reason for this extended use may be due to universal linguistic tendencies. Scotton cites many examples of this extension that are similar to those I have given in section 2.8.2.1. Most of 130 her examples have [-Nominal] complements. This accords with the statistical distribution of :59 versus pi forms in Bukavu Swahili. However, before developing her arguments that the extension of :59’s function reflects a universal tendency, she dismissed the influence of substratal (ethnic) languages on the vernacular Swahili uses: There is no evidence [for sub-stratal influences]. Further, since these speakers have different first languages, even if one of their languages showed such a construction, it would be an unlikely source. (p. 11) As shown by the examples in this section of parallels between the Bukavu Swahili :59 and local ethnic languages’ copula, such a statement could not be made in the Zairean Swahili situation. However, the ease of finding substratal parallels to :59 but difficulty of finding parallels to pi raises the question of why the 1 vs. -ko distinction should be realized as it is in Bukavu Swahili. It could be that a paradigm was brought to the Bukavu area by outside speakers and is being variably assimilated by Bukavu residents. It would be interesting to see if there are Tanzanian ethnic languages, as well as Zairean ones, in which this distinction is made and which could have influenced the development of the pi/:59 distinction in the nineteenth century during the westward movement of Swahili. In want of such evidence, one could argue in support of Scotton’s approach. 131 2.8.3 Lexical variable -- French frequencies The lexical scores assigned to subjects were based on how much they used French in their Swahili discourse; High scores represented relatively low use of French and vice versa. The scores were calculated by dividing the total number of characters (key strokes) in the transcribed text of a subject by the number of unique French words (a word would not be counted more than once) used and then dividing this by 10 so that the scores would range between 1 and 100. This resulted in scores ranging from a minimum of 6 to a maximum of 9934. It is recognized that this is a very gross score that does not distinguish between lexical borrowings (such as xiiip "city" or présidentlprezidanti "president" for which most Bukavu residents have no ready equivalent in Swahili”) and code-switches (in which the speaker uses a French term, phrase, or sentence even though s/he could have used a Swahili equivalent) and that does not allow for differences in topics which could affect the lexical items being used. I have not attempted to do this because of the great amount of time it would have consumed. But, of course, a future, more fine-grained analysis would be of interest. The high correlation between the scores based on this formula and scores based on switches between the Swahili and French equivalents of "but" indicate that this formula results in a good ball-park figure, at least: the Spearman correlation coefficient was .7596 (p=.000). 132 1. In his study concerning the Guyanese creole continuum Rickford found that: The effect of the expatriate interlocutors is particularly marked; not only are two of the four subcategories in sample D significantly different...from their counterparts..., but the remaining two represent categorical non- usage, giving us the misleading impression that Ustad never uses the basilectal forms therein. (1987:159, his emphasis) 2. I paid the equivalent of about twenty dollars per forty-- five minutes of transcribed tape. My main transcribers were able to earn as much as the equivalent of one hundred dollars a week -- quite an incentive considering the fact that the one with the highest regular salary was earning the equivalent of about fifty dollars a month. 3. Rickford reports: My own recent research on the Guyanese continuum (Rickford 1979) supports the view that the relationship between variation and change is not necessarily nor always symmetric, at least not if variation is interpreted as indicating ongoing change. With respect to morphological variation in the pronouns..., the age distributions suggest that change is in progress for the two main social classes, in some subcategories... But the situation across the different generations is also fairly stable in other subcategories. (1980:176) Similarly, my findings indicate that, diachronically, (j) is being pronounced more as [-sonorant] in some environments, as [+sonorant] in others, and has been fairly stable in others. Bickerton argues that the pidgin of older speakers may be somewhat different from how they spoke it in their youth but that, if there is change, it would be towards more complex forms: Although it is widely held that an individual’s speech changes little after maturity has been reached, this may not necessarily be true of second languages or 133 contact languages. Pidginization is a process, not a state, and it is therefore possible that at least some of our subjects may now speak differently from the way they spoke when they first arrived in Hawaii, even though the vast majority were already adults at that time. However, one thing is certain. If their version of HPE [Hawaiian Pidgin English] has changed in those intervening years, then it must be more complex in structure and less subject to idiosyncratic or ethnic- group variation than it was in the years that immediately followed their arrival. It is unthinkable that after several decades of life in a community that was steadily becoming more integrated their version of HPE should have grown less complex or more idiosyncratic. We must therefore assume that either their HPE now adequately represents the HPE of the early pidginization period, or that the latter was even more primitive and more unstable than the versions they use today. (1981:8) 4. The one exception, in the M group, arrived in Bukavu when he was four years old. 5. The reason for including the oldest of the M group (born 1950-1953) in this group rather than the 0 group is that those with high education still received their education in a post-colonial context, unlike almost all of those in the 0 group: 6. All subjects who had more than 12 years of education achieved this after Independence. At the time of Independence there were only very few Zaireans with a post— secondary education. 7. Kabanga et al. (1983:349) say that 67,470,9tudents were in secondary school in the Kivu province in 1976 and that 49% of these (=33,060) were in Bukavu. According to statistics cited in Nzongola (1983:291), there were, at that time, 100,170 residents in Bukavu of the age 0-19. Taking eight-nineteenths of this figure as roughly representative of those of secondary school age (12-19), we have a total of 42,176. Thus, the 33,060 enrolled students would be 78% of the secondary school aged residents. This might have to be diminished to allow for students who had come from other areas of Kivu to study only. Whether or not this group was counted in the study cited by Nzongola is unknown. 8. Hemphill found that there were several problems in teaching Swahili in Kenyan primary schools: lack of interest in teaching it, lack of materials and effective methodology, and "a great difference" between the regional, vernacular forms of Swahli and the "more or less literary variety that is found in textbooks" (1980:463). These parallel problems 134 mentioned in Masumbuko (1987) concerning teaching Swahili in Zaire. 9. Another important reason for the correlation between occupation and linguistic behavior is due to the extent to which an occupation puts one into contact with others from different socioeconomic backgrounds. Studies of British and American dialects have indicate that the more diverse one’s occupational contacts are, the less likely one is to hold to a particular sub-group’s vernacular norms (Labov 1972:433- 69; Labov 1980; Milroy and Milroy 1985). 10. Scotton found this to be true with regard to language attitudes and use in Kampala (1972:75). 11. "One scholar (Lewis 1973) has pointed out that [dense] network structure has been reported as characteristic of the urban poor in Africa, the Americas and Europe. he suggests that it is universal because its basis is economic; the maintenance of strong solidary relationships is necessary for survival. It should certainly be possible to test out these results by replicating the study in other speech communities." (Milroy and Margrain 1980:69) 12. Only this, of the five areas listed here, was not asked about in Scotton’s (1973) study. 13. Of course, whether or not a variable meets the fourth criterion "fairly immune from conscious distortion" seems to be a quite subjective judgement. To my knowlege, no one has tried to make this criterion more precise. Discussions with Bukavu residents indicate that the main-~perhaps the most conscious difference between the vernacular and standard varieties of Swahili--are lexical ones: the safi "clean" aspect of the standard speaker is his use of Swahili words which the vernacular speaker seldom uses; the vernacular speech will consist of either alternative Swahili forms (e.g., jambo, never standard hujambo, for a greeting) or French borrowings (e.g., magasain or magazini, never standard duka for "store"). However, relative immunity from conscious distortion may be inferred by seeing which variables are changed the most in response to a less natural speech situation. A Kruskal-Wallis one—way ANOVA test of the significance of my presence at an interview indicated that the pronunciation of the class 2 prefix was affected much more than pronunciation of (j) or the amount of French borrowings or code switches. Significance of my presence was at the .1054 level with regard to pronunciation of the second class prefix but only .4517 and .4108, respectively, with regard to the other two. 14. In the two examples of intra-clausal switches that I have found (though the search was not comprehensive), the 135 referent referred to by 99: is different from that referred to by wa-: ba-nachukia wa-tu cl.2,they-hate cl.2-entity "they hate people" (subject 38) ...juu kuko ba-zazi mingi ba-nazala since there.are cl.2-parent many cl.2,they—bear wa-toto ya-bo, ba-toto ya-bo cl.2-child poss.-cl.2, cl.2-child poss.-cl.2 ba-nakuya huku. cl.2-come hear. "since there are many parents who bear their children and then their children come here." (subject 99) 15. I did contrast scores based on counts of the verb subject prefixes with those of nominal prefixes. Though the 'scores for the nominal prefixes would almost always be higher than, if not equal to, those for the verbal prefix, the correlation between the two scores was so high that it did not seem necessary to consider both (Kendall correlation was .7391 and Spearman correlation was .8508; correlation between other closely related variables, such as the pronunciation of (j) in moja and in 199 were not nearly as high). Why would there be a difference at all if there was rarely an intra-clausal switch? Mainly because of the nominals appearance in the object position with a non-third person plural subject. 16. However, there is evidence from cognitive psychology for such mental calculation of frequency (Hasher and Zacks 1984; Sampson 1987). 17. I did not make such a distinction if it was clear from the transcription of texts that there was little variation regardless of frequency. So, for example, the much more frequently occuring -jifunza "study" was included in the same environments as jirani "neighbor" since there was nearly categorical pronunciation of (j) as [-sonorant] when it occurred before /i/ in pre-stress position. 18. Note that the second example involves a suffix used to form a noun from a verb: mu-kaa-ji class l-reside-er "resident" The second is the high form of the word for "village"; the more common, vernacular form is mugini. Thus, just the fact 136 that the speaker uses the high lexical item would predict that the high pronunciation of (j) would be used. Most occurrences of (j)i in post-stress position were either as nominalizers or in the word kijiji. 19. The two words listed in this environment are the only ones that occurred frequently enough in the texts to be worth counting. 20. I consider the negative, imperfect marker -ja- "not yet" as AUX-final in keeping with Barrett-Keach’s analysis of the Swahili verb "as having a bipartite internal structure [in which] Aux, or perhaps its successor, Infl, constitutes a morphological unit" (1986: 560). She has two main arguments. First, "all o-form [pronominal clitics] appear in the same position with respect to their hosts-- consitutent finally" (e.g., kwacho "with it" and amba-cho "relative marker-it") (p. 560). Second, analyzing the verb as consisting of two morphological units does away with the necessity of having two stress rules, one assigning primary stress to the word’s penultimate syllabe and another, more complex, rule assigning secondary stress near the beginning of the word. The syntactic positioning of -(j)a- gives further support to her argument: it is always the final element in what she calls the Aux constitutent (as in 1) and the syllable before it always receives stress (as in 2). (1) a. kisu [Auxki-Bl-Cho-][V-k8te] knife [cl.7-neg.-rel.] [cut] "a knife which does not cut" b. *kisu [Auxki- -si- ja- -cho- ][v-kata] * H N N -0110- -Ja- knife [01. 7- -neg.-neg.imprf. -rel.-] [cut] " -rel. -neg.imperf.- " c. kisu amba-cho [Auxha-ki-ja-llv-kata] knife rel.-cl.7 neg-cl.7-neg.impfr. cut (2) a. si-ja-enda ls.neg.-neg.imprf.-go "I have not yet gone" b. ha-tu-ja-enda neg.-1pl.-neg.imprf-go "we have not yet gone" It can be seen from the examples used in (2) that the ungrammaticality of (1b) is not due to the redundancy of cooccuring negation markers but, rather, to the consecutive occurrence of two Aux final morphemes in post-stress position. 137 My statistical analysis indicates an additional argument for Barrett-Keach’s approach: the low occurrence of [-sonorant] pronunciations of -(j)a- statistically groups it with other words that contain constituent final :jp and distinguishes it from the words in which jp occurs before the primary stress--but must be analyzed as part of the same morphological unit that contains that stress. 21. Except for 199 "because; concerning", -'ua "know" is the only word in this environment that occurred in the conversation of the subjects frequently enough to be considered for analysis. See footnote __ for discussion of the juugyplu variant; it was not counted in this analysis since the computer program I was using did not enable me to tell whether one of these forms was a phonological variant of the other or a homonym with the other. 22. The Standard form is -kojoa but this form never occurred in any of the texts. 23. This is practically the only word in the texts that contained (j)o in the post-stress environment. 24. The middle age group’s 6th and 7th row environments are switched not in consideration of other age group’s orderings but so that there will not be an interruption of the consecutive ordering of environments concerning 111 before /a/. 25. Here there is a significant increase from the old age group’s score to the middle age group’s but from the old age group to the young age group there is a significant decrease. 26. The difference between the old and middle age groups’ scores is not significant. 27. The subjects chosen for the examples are simply the first (and, thus, youngest) within each age group as listed in the Table in Appendix G. 28. Or, at least, elision of the final consonant. The elision often occurs in words in which there is vowel harmony between the final and penultimate vowel. Of my two main transcribers, one would represent the elision as applying to the whole final syllable; the other as applying to the final consonant. 29. Nkiko et al. (1983) claim that this morpheme is used in reference to distant tense. But I have not found this to sufficiently explain its distribution in the subjects’ texts. 138 30. An elision process somewhat parallel to this occurs in Kinyarwanda. When someone is calling out another person’s name, stress may be placed on the first syllable and the last syllable is dropped. Fore example, one may call out to a person whose name is normally pronounced [ka-i~’te-si] by saying [’ka-i-te]. 31. This speaker uses pi the five times the subject is plural "Christians" and i59 the three times the subject is singular. However, that this is not a rule that can be generalized for other speakers is seen, for example, in the speech of another subject who uses pi in both contexts: a. Ndiyo naye ni mukristu vile vile. "Yes, and she is a Christian also." b. Sisi sote ni wakristu. "We all are Christians." 32. The question of why pi should be used much more frequently in this environment than in the [+Simple NP] environment is yet to be resolved. Of special interest in considering this problem is the observation that semantic restrictions on the lexical items in the subject and complement positions seem to affect the frequency with which pi is used. A hypothesis to be tested in further research is: if the relationship between the subject and complement is definitive/essential/permanent, pi is more likely to be used than if the relationship is accidental/possibly temporary. If this is validated, it would indicate that the reason structures which I have called "[+Complex NP]" usually have pi as the copula is not simply because of their syntactic structure but because these structures are most apt to be used when statements of permanent identity, relationship, and/or definition are being made. Apparent support for this hypothesis stems from observation of how the frequency of pi varied according to environments isolated in my preliminary analysis. .For example, the frequency of pi was even higher than in the [+NP complex] environment (92%; in this footnote, percentage marks after a stated environment will refer to the frequency with which pi, rather than :59 was used) when the copula occurred before the interrogative pronouns nini and nani (95%) and in identity statements of the type "An X is an X" (94%). In a and b, typical examples of the former environment, the question of definitive identity of the subject referent is in focus rather than accidental attributes that would be asked for by questions such as "How (Namna gani) may the phenomenon in question be realized in a particular setting?" or "When (Wakpti gani) has this phenomenon been realized?" (a) (b) In c and d, of subject 139 Kuogopa ni nini? To.fear be what "To be afraid is what?" or "What does it mean to be afraid?" Siyasikia kabisa ile hamu ya I.haven’t.yet.felt completely that desire of kusema--ya kujua kama Mungu ni nini. to.say of to.know that God be what "I haven’t yet really felt the desire to say-- to know what God is is." examples of the latter environment, the equation and complement is most obvious as the subjects use tautologies: (e) (d) Tabu ni tabu. problem be problem "Problems are problems." Mutu ni mutu. person be person. "A person is a person." In e, there is a near—tautology, though the complement restricts the meaning of the subject: (e) Based on th explain the Yangu ni yangu mi’ peke...ma’ake byako byote mine be mine me alone meaning yours all ni byako we’ peke. be yours you alone "My things are mine alone...in other words, your things are yours alone. e use of pi in these environments, one may predominance of pi realizations in sentences such as f‘and g which have structures included in the [urComplex] action refe identity of environment and in sentences such as h where the rred to in the main clause is dependent on the the referent in the conditional clause. (f) Maisha ni mtu mwenye kuyua fulani huyu. life be person who know certain this.person "The means of making a living is by a person knowing a certain someone." 140 (g) Kitu kinapiga batu surtout ni namna ya thing it.hits people above.all be way of maisha. life "The thing which affects people most of all is their way of making a living." (h) Kama ni mwizi wa busiku banamupiga’ mipanga if be thief of night they.hit.him machetes "If it is a thief in the night, they beat him with machetes." The permanent/definitive vs. temporary/non- definitive/accidental distinction also accounts for why there is also a high realization of the copula as pi when the copula occurs before references to a familial relationship (87%) or before a proper name (89%). When people say of another person that "He is my father" or "She is my aunt", they are talking of essential, permanent ‘ relationships that only under unusual circumstances can be altered. Likewise, when one gives the name of himself or another ("My/his name is __"), there is, generally, an essential, permanent identity between the clause’s subject and complement. The complement is realized as pi much less frequently when it links the subject with his occupation (35%)--a relatively accidental/temporary/non-essential attribute of the subject. The person is identified in terms of what position or occupation they now hold; but it would be easy to conceive of him as having a different one or having this one for a relatively short period of time. The higher frequency of pi realizations when referring one’s ethnic group or nationality (69%) than in environments when referring to one’s occupation is consistant with the previous observations based on the permanent/essential vs. non—permanent/accidental distinction. Nationality and, especially, ethnic group, are, generally, permanent aspects of one’s identity--unlike one’s occupation. Assuming the validity of this semantic distinction, one may explain why the same speaker uses pi in i but -ko in j. ' 141 (i) Kusema akiwa mwanamuke anatumika ni to.say if.she.be woman she.works be comptable tu. Lakini banamuke habakuwa accountant only but women they.weren’t na makazi mingi mingi ile wakati ya kibeleji. with jobs many many that time of Belgian "That is to say, if a woman worked, it was only as an accountant, but women didn’t have a lot of jobs during the Belgian period." (j) Moya iko ingénieur pilote, mwengine iko one he.be engineer pilot other he.be inspecteur. inspector "One is a pilot’s engineer; another is an inspector." In both i and j, the complement refers to an occupation. However, in i the occupation is equated with, or defining, "a woman’s work" whereas in j the occupation is given as an accidental attribute of the persons being talked about. k and 1, from the same speaker, give an even clearer contrast between the definitive/permanent vs. non- definitive/temporary contrast. (k) Epsi nilizaliwp ni gpa ppp Bukavu. place I.was.born be here here Bukavu "The piace where I was born is right here, Bukavu." (l) Sawa awa waloko ile tano benyi biko like these little.ones that five who they.are apa badogo badogo... Niko un peu here small.ones small.ones I.be a little condamné kuvivre nao vile mpaka condemned to.live with.them like.that until banakuwa nafika ku ile age yenye iko yulu they.are arriving at that age which it.be high kiloko na pale banatoka. little and there they.leave "Like these five little ones who are here, very young... I’m rather condemned to live with 142 them until they arrive at the somewhat higher age where they can leave." In k, the locative complement defines the subject. The link betweeen the subject and complement is regardless of time-- thirty years ago and thirty years from now the "place where I was born" will always be "right here, Bukavu". In 1, though, the locative complement refers to a temporary location of the subjects (little children) that was different before they moved in with the speaker and will be different again when the children are older. The effect of the accidental/temporary vs. essential/permanent distinction on choice of copula is similar to that in Spanish where estar is used "with an adjective to indicate a state or condition of the subject, which may be temporary, accidental, or variable" but 999 is used "with an adjective to express an essential quality or characteristic of the subject that is relatively permanent" (Turk and Espinosa 1983:50-51). However, in Spanish this distinction is generally relevant only when the complement is an adjective; in Bukavu Swahili, the distinction is also relevant when the complements are nominal. 33. In Kimbembe, the copula is often omitted. I do not know what rules govern whether or not it is explicitly realized. 34. The scores of "99" are the only ones that don’t represent the exact figure (rounded to the nearest 1) derived from the given formula. They would have been even higher but it was not necessary to represent these in a more precise way since the subjects with these scores can be easily categorized: three are among the oldest of the old age group and have no formal education. The other is a pastor. 35. I make this claim on the basis of a translation test that I gave to about thirty people of varying age and educational levels (including no formal education). Whereas some French words that are often used in Swahili discourse were very easy for them to translate (for example, ppip "but"), ville and rais were seldom correctly translated to their Swahili equivalents. CHAPTER 3 TESTING OF HYPOTHESES Introduction In this chapter, each hypothesis mentioned in chapters 1 and 2 is tested. First, the method used to test the hypothesis is described, then the results are given. Since the results concerning one hypothesis often relate to those of other hypotheses, discussion of the results is postponed until the end of the chapter except with regard to Hypotheses 7-9, concerning the effect of my presence at interviews. These are discussed first since the results indicate which scores should be discounted in testing the other hypotheses. While .05 or less is most frequently the level of significance set in testing hypotheses, I do not automatically dismiss results with a higher level of significance. This is especially true when testing involves the use of correlation ratios (in which "if X and Y have some sort of curvilinear relationship, the value of r will underestimate the true degree of relationship") (Glass and Hopkins 1974:83) or a Scheffe Range test (making two— way comparisons of scores between each of three or more groups) in which the smaller number of scores and method of calculation often result in probabilities larger than .05 though the overall probability of contrast in scores between groups may be much less than .05. Since the phonological 143 144 and lexical scores were evaluated with the same testing measures, they are often grouped together in the discussion of results. that: Test results concerning linguistic variables indicate --pbonological scores are most affected by differences in education and age group and by degree of church involvement; --morpbosyntactic scores are most affected by ethnic group membership; and --1exica1 scores are most affected by level of education, regardless of age. --Network strength affects morphosyntactic and lexical scores. --Sex makes no difference in scores. Concerning language attitudes, it is found that: --there is a positive correlation between attitudes about the necessity of learning standard Swahili and phonological scores. --Both the belief that learning Standard is important and the phonological scores have decreased diachronically. --There is a diachronic increase in the perception of French as essential, and Swahili as irrelevant, to academic and socioeconomic advancement. Nevertheless, --The great majority of subjects of all age groups rated the speaker who incorporated French words into his Swahili as "arrogant"; this is in harmony with the finding that those with the highest Network Strength scores incorporate less French into their speech than those with lower Network Strength scores. Discussion of how the test results affect evaluation of the models mentioned in chapter 1 is, for the most part, kept for the final chapter. 145 Because of the sampling methods used, the results of the following tests cannot be claimed to be valid for any more than the group of subjects whose scores are used. Furthermore, it is possible that the lack of strict correspondence of educational levels between subjects of the various age groups and the greater heterogeneity of oldest- age group subjects’ geographical origins could interfere with the accuracy of conclusions drawn from tests concerning the relationship between age and other factors. However, it is believed that the reasons for having these differences, discussed in chapter 2, are legitimate and that the results, in keeping with the pilot-project nature of this research, may at least be considered to be indicative of general trends in the Bukavu speech situation and of interesting areas for future research. 3.1 Hypotheses concerning linguistic variables In this section, all hypotheses concern factors which affect the scores of the dependent variables presented in section 2.8. 3.1.1 HYPOTHESES 7-9 Hypothesis 7: The absence or presence of a foreign interviewer will not affect a subject’s linguistic performance if a Zairean is present to co-conduct the interview. Hypothesis 8: The absence or presence of a foreign interviewer will not affect a subject’s linguistic performance even if a Zairean is not present to co- conduct the interview. 146 Hypothesis 9: If either hypothesis 7 or 8 is false, the foreigner’s presence will affect only those whose occupation would encourage use of the standard or those who were educated before 1960. Test: Linguistic scores were correlated with the interviewer variable. The values were: (1) only a Zairean present to conduct the interview; (2) both a Zairean and myself present to conduct the interview; or (3) only myself present to conduct the interview. Result: At the .10 significance level in testing phonological scores, my presence did correlate with an increase in the 1919: score (that is, an increase in the frequency of standard [+sonorant] pronunciations) if I was not present with another Zairean: interviewer uppn (wp-) score 1. Zairean only 12 2. Zairean + me 13 3. Only me 35 Since two of the seven subjects I interviewed alone were teachers and since other hypotheses had predicted that teachers’ and preachers’ scores would be higher than those of others, the test was redone to exclude the teachers. The result indicated that my being the sole interviewer did not significantly affect the scores of the five other subjects. 80, tests of all variables were done again within the occupational group of teachers. Table 3.1 indicates that my being alone could significantly alter their speech patterns. 147 Mean (wa-) . Mean mo(j)a Mean -(j)ua Mean French Interviewer score (.001) score (.0000) score (.0000) score (.0072) Zairean 12 ' 4 9 15 present Me only 91 83 97 25 Table 3.1. Effect of my presence on teachers’ scores. Numbers in parentheses indicate probability. With regard to (pi), the scores of those interviewed by a Zairean and me were significantly (p=.05) different from the other groups with regard to the [-Nominal] and [epSimple] environments. However, unlike the 1919: scores, the percentage of standard usages was actually i999; than in the other groups (Table 3.2). Enviroment Zairean only Zairean + me Me only -Nomina1 36 27 32 eeSimple 59 30 67 Table 3.2. Percentage of (ni) variants according to interviewer. ' There was also a significant difference (.05) between the scores of those for whom I was the sole interviewer and the scores of the others in the [upComplex] environment, the former being higher. However, as with the i919: scores, the scores of the two teachers whom I interviewed alone were significantly (.001) higher than those of the teachers I did not interview by myself. Discounting the two teachers’ scores, there was no significant difference between groups in this environment. My presence had no significant effect on the lexical scores . 148 Conclusion. The large gap between the scores of teachers who were interviewed by me alone and of those who were not indicates a qualitative switch from a low to high variety. While other subjects may have somewhat modified their linguistic behavior if I was the sole intervewer (see, for example, the discussion of the (wa-) variable plotted according to occupation) the modification was not nearly as extreme as that of the two teachers. This is probably due to the fact that the teachers could achieve the high variety with more ease than the other subjects who would occasionally use high forms, probably at more self-conscious moments, but most frequently use the low forms. Since the presence of an expatriate would not be typical of most speech situations and since we want the speech behavior we analyze to be as normal as possible, I exclude the scores of the two teachers I interviewed by myself for the testing of the hypotheses, except in Hypothesis 1b and lo where the analysis of a histogram enables easy isolation of their scores. The ipii scores of those interviewed by a Zairean and myself are also discounted in the [-Nominal] and [upSimple] environments. 3.1.2 HYPOTHESIS 1: Only those whose occupation encourages use of the standard will show signs of movement towards the standard. That is, a. the standard scores of teachers and preachers will be higher than the scores of all other subjects. 149 b. apart from those educated before Independence, the scores of all other subjects will be clustered around a comparatively low score. c. apart from those educated before Independence, only the scores of teachers and preachers will suggest a continuum affect by ranging from the low scores of other subjects towards 100% use of Standard forms. . Test of 1a. The dependent variable scores were correlated with the five occupational groups: (1) manual laborers, (2) teachers, (3) skilled laborers, (4) administrators, and (5) preachers. Teachers were subdivided according to level at which they teach to see how this affected scores. Results. This hypothesis was stated too generally. As Table 3.3 shows, of teachers and preachers, only the latter group’s scores are significantly higher than other groups scores with regard to both pronunciation and rate of borrowing. With regard to pronunciation, the preachers are in a high-scoring group and the other four occupations are in a low-scoring group. With regard to the French score, the preachers are grouped with manual laborers in contrast to the group of teachers, skilled laborers, and administrators. 150 Occupational Group Variable ML T SL A P (wa-) 14* 12* 13* 3* 96 mo(j)a 17* 4* 15* 6* 93 -(j)ua 32* 9* 41 30 95 French 45 15* 14** 6** 75 Table 3.3. Mean score (of possible 100) of phonological and lexical variables correlated with occupation. ML=manual laborer, T=teacher, SL=skilled laborer, A=administrator, P=preacher. *zscore significantly different from preachers’ score; *tzscore significantly different from preachers’ and manual laborers’ scores. P=.05 except for italicized scores which are at .10 level. The (ni) scores show similar results with regard to the pastors; they have the highest scores in each of the three environments. However, it can again be seen that the hypothesis was wrong in lumping together teachers and preachers since the two groups are actually on opposite ends of the scoring range (Table 3.4). Occupational Group Environment 55 I 95 5 g [-Nomina1] . 34 22 33 39 84 [epSimple] 62 25 73 [n.Complex] 90 85 94 84 98 Table 3.4. Percentage of pi variants according to occupation. Scores are not given if there were not at least 10 chances. Explanation of code given in Table 3.3. In testing Hypothesis 1a, teachers were not subdivided according to the level at which they taught. It might be 151 expected that since the primary school teachers use Swahili as a medium and teach it as a subject, they would have higher scores in pronunciation than the high school teachers. This was not the case, however, as Table 3.5 indicates. Variable Score (of 100) Teaching level Subject (wa-) mo(j)a -(j)ua Primary 41 3 3 8 40 0 0 14 85 3 0 4 Mean: 2 1 9 Secondary 39 37 8 11 3 18 10 6 Mean: 23 9 9 Table 3.5. Teachers’ pronunciation scores. There was no significant difference between these two groups with regard to the morphosyntactic scores. The use of French, though, is less among the primary teachers (mean=19). This could be due to the fact that they use French less in their occupation than do the secondary teachers (mean=91), who use it to teach all classes. HYPOTHESIS 1b and 10 Test. Scores of dependent variables were plotted according to occupational groups at 2.5 point intervals (Appendix I). Those with scores above a low cut-off point were further categorized according to whether or not I was the sole interviewer and their age, educational level, and 152 religious orientation. Some chi-square analyses were done to test for significance of how these subjects were grouped according to this additional categorization in contrast to the subjects with low scores. Plots were not done for the ipii variants since the scores do not vary as widely within a specified environment as they do for the other variables and there are not as many individual scores with sufficient entries for each environment as there are for the other variables. Results. The plotting of the phonological scores were very similar for each variable so only the plot for the (wa-) variable will be discussed here. The plot of the French scores will not be discussed in much detail since the key independent variable affecting the range of French scores is education rather than occupation. Plot of (w)a- scores (shown in Appendix I) 66% of the subjects have a (wa-) score of 6% or less and the median score is approximately 1. Obviously, there is a large cluster around "a comparatively low score" as the hypothesis predicts. However another attempt to generalize about the twenty-six subjects whose scores are in this continuum-like range must be made since only four are teachers--two having been interviewed by me alone--and two are preachers. The three major factors that I have isolated are: my being the sole interviewer, education for those in the Old- 153 age group, and religious orientation. First of all, four of the five people whom I interviewed one-on-one are here: for example, the administrator with a score of 13 (the only administrator scoring over 1). Secondly, ten of the eleven subjects who scored over 6 999 are in the old-age group had formal education; the only one without formal education had the lowest score (8) of the eleven. Table 3.6 breaks down scores in the old-age group with respect to those with education and those without. (wa—) score Education Above 6 Below 6 Some 11 5 None 1 11 Table 3.6. (wa—) scores in the old-age group according to education (p=.01). Finally, at least seven of the 13 in the middle-age group scoring above six had a high degree of involvement in a Christian church’; information on church involvement was not obtained from four of these subjects. Table 3.7 breaks down scores of all subjects from whom information on the degree of church involvement was obtained. 55% of those 154 Church (wa-) score involvepent Above 6 Below 6 High 12 10 Low - 13 31 Table 3.7. (wa-) scores according to level of church involvement. (p'< .05) with high church involvement scored above 6 whereas only 29% of those with average or less involvement scored over 6. A chi-square test shows this to be significant at the .05 level. The only subject with a score over 65 that is not a teacher or preacher is the only Moslem in the old-age group who was interviewed. As preachers are stereotyped as speaking ppii "clean" Swahili so too are the small group of Moslems in the Bukavu area. One of the middle-age group speakers with a comparatively high (wa-) score was asked several weeks after his recorded interview if he could say why his Swahili incorporated more standard forms than most speakers. He attributed his speech patterns to his frequent association with Moslem friends. If we discount subjects who may be categorized as being interviewed by me alone, both old and educated, and/or influenced by patterns of speech associated with religion, only 4 of the 26 subjects scoring above 6 remain’. 155 Plot of French scores The plot of incorporation of French into conversational Swahili indicates that the pastors and manual laborers have the highest scores; that is, they incorporate French into their speech less than those of the other occupations. However, the only ones who scored over 40 and who had more than a primary school education are the pastors. 3.1.3 HYPOTHESIS 2: Level of education will positively correlate with standard scores only among those who were educated before Independence. Test. For each age group, phonological and lexical scores were correlated (Pearson Correlation) with educational level. Rather than grouping subjects together within three-year time slots (e.g, rather than putting one subject with one year of education, another with two years, and a third with three into the same group and coding it as "1") as was necessary for some other tests, the highest level in number of years for each subject was correlated with the scores; thus, the correlation could be much more precise. Preachers’ scores were excluded since the tests done for Hypothesis 1 indicated that the effect of their occupation overrode the effects of education. However, teachers scores were kept (except those for whom I was the sole interviewer). Results. As Table 3.8 indicates, this hypothesis received only limited support at a .10 signifcance level. 156 Old Middle Young Variable Coef. Prob. Coef. Prob. Coef. Prob. (wa-) .2623 .081 .1021 .263 .1686 .205 mo(j)a .1617 .188 .0387 .405 .1436 .242 -(j)ua .2570 .078 .0966 .274 .1331 .258 French -.7533 .000 -.7529 .000 -.4420 .013 Table 3.8. Correlation of scores with educational level, according to age. For the phonological scores, there is a significant, positive correlation between i919: and :iji9p scores and education only in the old-age group. This gives further support to the validity of the observation made in the discussion of Hypothesis 1 that formal education of subjects in the old-age group was a key contributor to the apparent continuum effect of the phonological score plots. With regard to ipii, the old-age group was the only one in which educated subjects had significantly higher scores than non-educated subjects and even this difference was only in the [~Nominal] environment; the scores among uneducated speakers was 15% (of 84 total occurrences) as opposed to the educated speakers’ 40% (of 88 total occurrences). The negative correlation between education and lexical scores is very high for all subjects: the higher the education, the more likely one is to incorporate lexical items from French, the main language of education, into his Swahili. 157 3.1.4 HYPOTHESIS 3: Among the educated, standard scores will decrease from older to younger speakers. corollaries: a. Phonological and morphosyntactic scores among middle and young-age group educated speakers will not be different from the scores of the old, uneducated speakers. b. Lexical scores among the middle and young-age group speakers will decrease from the scores of the old, whether uneducated or educated, in favor of French. Test. First, the median phonological and lexical scores of subjects with at least one year of formal education were compared according to age groups. Then the scores of those with at least one year of formal education in the middle and young-age groups were compared with those with no education in the old—age group. Preachers and teachers whom I alone interviewed were not counted. Results. There is support for the hypothesis and its corollaries with regard to phonological scores but not for the morphosyntactic or lexical scores. Table 3.9 indicates that there is a decline among educated speakers between the old-age group’s phonological scores and either the middle or young-age group’s scores at a probability of Scheffe score differences no greater than .15. The middle and young-age groups’ scores are always much closer to each other than to the old-age group. 158 Mean Score Variable Probability gig Middle Young (w)a- .15 21 12 7* mo(j)a .10 24 12 9* (j)ua .05 58 42* 41 Table 3.9. Phonological scores for educated speakers, according to age group. *=score is significantly different from the old-age group score at stated level. Corollary (a) was supported by the i919: and mo(j)a scores at a .10 significance level; that is, there was no significant difference between the scores of middle and young-age group educated speakers and old-age group uneducated speakers. However, there was not support for this from the scores for the :(j)ua variable (Table 3.10). Age Group Mean -(j)ua score Old, uneducated 14 Middle, educated 42* Young, educated 41* Table 3.10. Mean -(j)ua score according to age group and education. *zscore is significantly different from old, uneducated group’s score (p=.05). The young, educated speakers’ pronunciation of -(j)ua was significantly different from that of the old-age group’s uneducated speakers (Table 3.10) but not from that of the old-age group’s educated speakers (Table 3.9). However, the opposite is true with regard to pronunciation of mo(j)a: the young, uneducated speakers’ pronunciation was significantly different from the old-age group’s educated speakers but not the uneducated speakers. a. 159 The hypothesis was not supported with regard to Lpii. Only in the [upSimple] environment was there a significant decrease in the educated subjects’ scores from the old to younger age groups‘| but even here there was no significant difference between the old and young-age groups’ scores. Also, with regard to educated subjects only, there was no significant difference in scores in the [-Nominal] environment and an increase in scores in the [erComplex] environment. In all three environments, there were, contrary to the prediction of Corollary (a), significant increases in pi scores from the old uneducated group to the groups of middle and young educated speakers (Table 3.11). -Educated +Educated +Educated Environment Old Middle Young —:_ [-Nominal] 15 (84) 35 (389) 37 (134) [ueSimple] 36 (14) 56 (84) 65 (31) [epComplex] 90 (122) 90 (711) 96 (120) Table 3.11. Percentage of pi occurrences by age group and educational level (total number of chances given in parentheses). Corollary (b) falsely predicted that there would be a significant increase in the educated subjects’ incoporation of French from the old to the younger age groups but correctly predicted that there wold be a significant (.05 level) increase from the old, uneducated speakers to the middle and young-age groups’ educated speakers. 160 3.1.5 HYPOTHESIS 4: Higher income will correlate with standard scores only in the old-age group of speakers. Test. For both the old and middle-age groups, mean phonological and lexical scores of the five different occupational groups (see discussion of Hypothesis 2) were compared. The scores of the adminstrators, by far the most affluent subjects, were then compared with the other subjects lumped together into one group. Finally, the occupational groups were compared discounting the scores of those with high church involvement. The hypothesis could not be adequately tested with regard to ipii since there were not sufficient chances in the higher income level values for satisfactory testing. However, tests were done in the middle-age group to compare the high salaried administrators’ scores with those of the other subjects in the two environments for which there were sufficient chances. Results. The test in which the scores of subjects with non-administrative jobs were lumped together and compared with those having administrative jobs produced the same results as the other tests. Apart from the scores for preachers, already discussed, there was no significant difference in scores except with regard to the French scores which showed that the administrators incorporated more French lexical items than those in the other group (p=.08 and .05 in the old and middle-age groups, respectively; see 161 discussion of the plotting of scores for testing Hypothesis 1). Thus, for the phonological scores, education, age and religious orientation had more effect on the outcome than level of income. Tests of the ipii variants indicated that the adminstrators’ scores in the middle-age group were significantly higher (p=.05) in the [-Nominal] environment but not in the [upComplex] environment. There are not enough chances to enable comparison in the [eeSimple] environment. Thus, there is not solid support for the hypothesis with regard to the morphosyntactic variable. 3.1.6 HYPOTHESIS 5: Speakers’ positive evaluations of the necessity of learning standard Swahili will decrease from the old to young age groups. Test. A comparison was made of each age group’s mean response to the interview schedule’s question: Unawaza ni lazima kwa mtu kujua kiswahili bora? "Do you think it is necessary for a person to know standard Swahili?" 1. Hapana, hata kidogo. No, not at all. 2. Si lazima lakini ni muzuri. It’s not necessary but it is good. 3. Lazima kabisa. It’s very necessary. Results. The hypothesis was supported by a decline in scores in which the young-age group’s score significantly differed from the old-age group’s score at the .05 level (Table 3.12). 162 Age group Mean §core Old 2.7917 Middle 2.5556 Young 1 2.2500* Table 3.12. Evaluation of necessity of learning standard Swahili correlated with age group. *=score is significantly different from that of the old-age group. 3.1.7 HYPOTHESIS 6: Positive evaluation of the necessity of learning standard Swahili will positively correlate with standard scores. Test. Phonological and lexical scores were correlated (Pearson) with the subjects’ opinion on the necessity of learning standard Swahili (based on same question and scale as given with regard to Hypothesis 5). Results. As Table 3.13 indicates, the hypothesis was supported with regard to the phonological variables. Correlation Variable Coefficient Probability (w)a- .2069 .05 mo(j)a .2479 .02 Table 3.13. Correlation between phonological scores and subjects’ positive evaluation of the necessity of learning standard Swahili. However, the correlation between lexical scores and attitude about the necessity of learning standard Swahili was not significant (probability was greater than .20). The difference between groups in the use of pi had a .10 163 probability in the [upComplex] environment but was insignificant in the others. 3.1.8 Hypotheses 10 and 11 HYPOTHESIS 10: When educational levels are similar, middle and young-age group women‘ will have a higher rate of borrowing from the French lexicon than men. HYPOTHESIS 11: Middle and young-age group women will have a higher rate of standard phonological and morphosyntactic farms than men. Test. An initial correlation test indicated that there were no signifcant imbalances between male and female subjects with regard to educational level. So, an overall correlation test was first done. Then educational levels were grouped according to whether the subjects’ educational level was nil, primary, secondary or university. Correlations were done considering each educational level. Tests were then done for all educated speakers with those having a high level of church involvement factored out. Finally, a correlation between sex and opinion on the necessity of learning standard Swhaili was done since the hypotheses were formed with regard to the perceived prestige of the standard. Results. Hypothesis 10 was not supported but Hypothesis 11 was. As Table 3.14 indicate, the women’s phonological scores were actually lower than those of the men and there was no significant difference in the lexical scores. However, at 164 Male scores Female scores Variable Prob9bility Mean Min. Max. Mean Min. Max. (w)a— .0835 12 0 92 5 0 47 mo(j)a .0562 14 0 84 6 0 47 -(j)ua .6812 44 0 100 48 0 100 French .6043 28 4 99 31 5 77 Table 3.14. Scores according to sex. less than a .10 probability there is a difference in the scores for i919: and mo(j)a. When there was a breakdown according to educational level for these to variables, though, it was found that the only group in which-there was a significant difference between sexes was the secondary school one (p=.013 for 1919: and .026 for mo(j)a. There was no significant difference between sexes if the subjects had either more or less than a secondary school education. Difference between sexes with secondary education was also found in two of the ipii environments. As in the phonological scores, there was no significant difference among the other groups in any of the environments. However, in the secondary group, there was in the [-Nominal] and [eeComplex] environments (p=.05 and .02, respectively). Again, the women’s scores were lower than those of the men. When scores were redone to discount those with a high level of church involvement, the probability of the difference in 1919: scores increased almost 10 percent but that of the p9iji9 score remained nearly the same, dropping a little (Table 3.15). Again, the women’s scores are lower than the men’s. 165 Significance Male scores Female scores Variable level Mean Min. Max. Mean Min. Max. 1.1.- .1781 9 ' o 58 4 o 47 mo(j)a .0490 11 0 84 3 0 21 Table 3.15. Phonological variables correlated with sex, discounting those with a high level of church involvement. The influence of church involvement was even clearer with regard to the ipii scores. When those at a "4" level of church involvement were discounted, no environment showed a significant difference between male and female subjects-- whether only those with a secondary school education or all educated speakers grouped together were considered. The correlation between sex and evaluation of the necessity of learning Swahili was not significant (probability was .7209). 3.1.9 HYPOTHESIS 12: Ethnic group membership will not effect subjects’ scores. Test. First, a check was done to see if there were imbalances in the distribution of ethnic group membership with regard to other independent variables. Ethnic group scores were then analyzed (one-way) with regard to the dependent variables. Results. This null hypothesis was clearly supported with regard to the phonological variables (no correlation having less than a .30 probability). 166 However, there are significant (.001) differences between groups in all (ni) environments (Table 3.16). Environpent Shi Rega Others [-Nominal] 39 (319) 25 (142) 36 (288) [ueSimple] 69 (70) 40 (30) 57 (61) [neComplex] 94 (559) 91 (271) 88 (424) Table 3.16. Percentage of pi (v. -k0) by ethnic group (total chances are in parentheses). The Shi speakers’ scores are consistently higher than the Rega speakers. As mentioned above, there was an imbalance of educational levels among the three groups. Tests for Hypothesis 3 showed that there was a significant difference between the scores of old, uneducated speakers and middle and young educated speakers. Thus, the tests for ethnic group differences were redone first to compare only the scores of old, uneducated speakers and then to discount the scores of those without formal education. In both tests, significant differences between groups remained in all environments.o If the influence of ethnic language is key to the differences here, one might expect the uneducated speakers to reflect this most clearly. However, as Table 3.17 indicates, this is not so. Another interesting aspect of 167 Shi Rega Others Environment +ED -ED +ED -ED +ED —ED [—Nomina1] 42 13 25 25 ' 4O 13 [upComplex] 94 95 90 94 88 84 Table 3.17. Percentage of pi (v. :59) variant by ethnic group, with regard to education. Chances in the [-Nominal] environment averaged about 222 for educated speakers and 28 for uneducated. Chances in the [upComplex] environment averaged about 377 for educated and 41 for uneducated. this table is that, while the [-Nominal] scores of the Shi and "Other" speakers remain basically the same in both the uneducated and educated groups, there is a score increase of more than 25% from the latter to the former. The testing of Hypothesis 11 indicated that apparent differences between sexes in the ipii environments among those with no more or less than a secondary education were eliminated when differences in degree of church involvement were taken into consideration. So, the ipii tests concerning ethnic groups were redone to allow for this factor. First, the level of church involvement was held constant, counting only scores of those with average church attendance (once a week). Differences between groups remained in both the [-Nominal] and [neComplex] environments ([eeSimple] could not be tested because of insufficient chances).7 Then the scores were retallied with those of above average attendance excluded. Differences between groups remained (p=.001) but differences between the Shi and Rega subjects diminished.8 168 Thus, the differences between ethnic group scores are diminished but still exist even when differences in sex, church involvement, and education are factored out. The difference in lexical scores was significant at a .009 level. However, the correlation with other independent variables showed an imbalance in the distribution of ethnic groups with regard to education and occupation. The effect of both, especially education, on the lexical variable is so clear from the tests of Hypotheses 1-3 that it can be assumed that it is this imbalance rather than sociocultural attributes of various ethnic groups to which the difference in lexical scores may be attributed. 3.1.10 HYPOTHESIS 13: Subjects’ Network Strength scores will correlate negatively with the phonological and morphosyntactic scores and_positively with the lexical scores. Test. The basis of assigning network strength scores is given in section 2.5.6. Phonological and lexical scores were correlated (Pearson) with the linguistic variables. Results. The correlation between network strength and the phonological variables’ scores was not significant but the correlation with the morphosyntactic and lexical scores was significant at the .10 level”. Factors that we have considered already have more of an effect.10 169 3.1.11 HYPOTHESIS 14: Those with a high level of church involvement will have higher scores than other subjects. Tests. First, scores of the linguistic variables were correlated (Pearson) with degree of church involvement. Then Scheffe tests concerning the linguistic variables which had significantly correlated with church involvement were done to see at what probability the scores of those with the highest degree of church involvement were shown to be significantly different from others. The tests were done including all subjects for whom there was information on their degree of church involvement (except for the two teachers I interviewed by myself), then redone to first exclude the only person with a score of "0" for church involvement11 and then exclude his score and the scores of the two preachers. Results. The initial correlation test indicated that there was no significant correlation of either the phonological or the lexical scores with the degree of church involvement. However, the test was redone to exclude the effect of the sole subject in the "0" group. This resulted in a significant positive correlation of all three phonological scores, but not the lexical score, with degree of church involvement (Table 3.18). 170 Correlation Number of Variable Coefficient subjects Probability (w)a- .3115 63 .006 mo(j)a .3338 65‘ .003 -(j)ua .2136 65 .044 French .0346 65 .392 Table 3.18. Linguistic scores correlated with degree of church involvement. Table 3.19 shows how the median scores reflected this positive correlation. The two preachers’ scores were Median score Variable 1(2) 2(6) 3(35) 4(22) (w)a- 3 5 9* 25 m0(j)a 6 3* 12* 28 —(j)ua 23 33 37 52 Table 3.19. Median scores according to degree of church involvement with 4 highest (numbers in parentheses indicate the number of subjects in each category). *=score is significantly different from group 4’s score at .15 level. included in those given for Table 3.18 and 3.19 because of the obvious link between their occupation and degree of church involvement. However, the tests were redone to see how strong an effect their presence had on the correlation ratios and Scheffe tests. The mean scores for group 4 dropped to 17 for i919:, 21 for mo(j)a, and 48 for iiipp. The correlation coefficients were lower for all phonological scores but still within the .10 significance level range. However, even at a .25 level, the Scheffe test showed a significant difference between 4 and another group’s score 171 only for mo(j)a where the contrast was between it and the lowest scoring group. Finally, the scores were done to discount both uneducated and educated subjects of the old—age group for - ' us, and old, uneducated subjects for mo(j)a and 1919: since the testing of Hypothesis 3 had shown these groups to be significantly different from the middle and young educated subjects in the respective environments. If there were, for example, more old, educated speakers among those with higher church attendance than in the lower attendance groups, the apparent correlation between linguistic scores and church attendance might actually be due to this imbalance of age and education distribution. The result of this retest was that the correlation of church attendance with -1j)ua fell below the .10 probability level but remained low with 1919: and mogjla (Table 3.20). Correlation # of Environpent Coefficient Probability Subjects (w)a- .3600 .006 49 mo(j)a .3453 .007 s 51 -(j)ua .2033 .107 39 Table 3n20. Church attendance correlated with phono- logical scores. Scores of pastors and age/education groups that affected scores in Hypothesis 3 tests were factored out. The hypothesis was also supported by the (n1) results. results. There is no significant difference between groups in the [neComplex] environment (the presence of the two pastors’ scores does not make a difference). However, there ‘ 172 is a significant difference in the two other environments with scores positively correlating with degree of church involvement, even if the pastors’ scores are excluded (Table 3.21). In the [upSimple] environment, a Degree of Church Involvement Environment 0-2 9 i Pastors [-Nominal] 11 (28) 33 (269) 40 (209) 84 (25) [erSimple] 45 (11) 53 (38) 62 (39) 73 (15) [neComplex] 91 (74) 92 (358) 91 (432) 98 (52) Table 3.21. Percentage of pi (v. :59) uses by degree of church involvement (numbers in parentheses refer to total chances). The two pastors’ scores are not included in the group 4 scores. one-to-one comparison of scores (such as the scores of those in the 4--above average--group to those in the 2--below average--group) do not reveal significant differences. In the [-Nominal] environment, such a comparison reveals a significant difference only if the subjects with 0-2 levels are grouped together and their scores opposed to those with 3-4 (or 3,4, and pastors’) grouped together. When ethnic group is kept constant and pastors’ scores are excluded, there is little evidence that church attendance significantly affects the morphosyntactic scores. For the Rega and "Other" ethnic groups, there is no significant difference between scores when level 4 of church attendance is opposed to level 3, in either the [-Nominal] or [npSimple] environment. In the Shi group, there is a significant difference in the [-Nominal] but not in the 173 [epSimple] environment. There were insufficient occurrences to compare level 3 and/or 4 scores with level 1 and/or 2 scores.. Thus, the hypothesis was valid with regard to phonological scores, has only weak support from the morphosyntactic score and was not valid with regard to the lexical score. 3.2 Testing of Hypotheses 15 to 28 All of the tests for hypotheses 15 to 28 are based on tallies of subjects’ responses to the interview schedule (section 2.4ff) with chi-square tests of association done to check the level of significance. Each subject was classified as being in one of the three age groups used to test several of the preceding hypohteses. The assigned significance level is p=.10. 3.2.1 Hypotheses 15-18 HYPOTHESIS 15: All groups will evaluate the standard speaker as speaking the best Swahili and the one who mixes in French as having the worst Swahili. However, HYPOTHESIS 16: There will be a decline in the proportion of subjects positively evaluating the quality of the standard speaker’s Swahili from the oldest to youngest age groups. HYPOTHESIS 17: The one who mixes in French will not be as negatively evaluated by the young-age group as he is by the old-age group. HYPOTHESIS 18: Evaluation of the vernacular speaker incorporating little French will be the same for each age group. 174 Test. Scores for testing these hypotheses are based on the subjects’ responses to the question asked for each taped Swahili speaker: Kiswahili yake ni namna gani? How is the Swahili of this speaker? I. mbaya sana "very bad" 2. mbaya "bad" 3. kadiri "okay" 4. mzuri "good" 5. mzuri sana "very good" The subjects’ responses were counted to see how many times each speaker received the lowest score of the three and how many times he received the highest score. Ties were not counted. Results. Table 3.22 gives the result of this tally. Speaker Evaluation 1 g 9 Ybung-age group Highest score 3' 22 2 Lowest score 12 1b 20 Middleeage group Highest score 0 27 1 Lowest score 9 0 17 Oldest-age group Highest score 7c 18 0 Lowest score 2 3d 24 Table 3.22. Evaluation of Swahili varieties by age group. Speakers: 1=Vernacular without French, 2=Standard Swahili, 3=Vernacular with French. a: one had no education and two had the lowest level of those who had been in school (four years, primary school). b: no education. 0: six of seven had no education. d: two had no education. Hypothesis 15 is clearly supported since, combining the totals of all age groups, the standard Swahili speaker is most positively evaluated by 67 of 71 subjects and the speaker who mixes in a lot of French is most negatively evaluated by 61 of 64 subjects. However, Hypothesis 16 is not supported. There is no significant difference between the age groups’ evaluations of the standard speaker. In fact, the group with the highest percent of negative evaluations were in the oldest age group. Two of these negative evaluators, like the one in the young-age group, had no formal education; one of these dismissed it as "like the Swahili of one who has studied a lot" and the other said that it was "very 176 difficult to understand." In general, though, subjects positively evaluated the Swahili as 999i "clean" or 99 sarufi "according to the grammar". The contrast between the groups’ evaluations of the vernacular speaker who mixed French was significant with p: .001. Whereas none of the sujbects in the old-age group gave the speaker a positive rating, one in the middle-age group and two in the young-age group did. Two of these subjects explained their choice by saying that the one mixing in French was speaking Swahili in the contemporary manner: kiswahili yote ya sasa ni ya kuchanga na francais ao anglais "all Swahili nowadays is one of mixing in francais or anglais".12 Thus, Hypothesis 17 is supported. Hypothesis 18 is not supported because of the significantly high proportion of older speakers who evaluated the vernacular speaker as speaking the best Swahili. All of the speakers in this group giving a high rating were over 60 years old and six of the seven had no formal education. Even the educated one (five years of primary school) explained his choice in terms of understandability rather grammaticality: the standard speaker, he said, spoke a Swahili that was nguvu sana "very difficult" which "the one hasn’t studied wouldn’t be able to understand" but the vernacular speaker who did not mix in French spoke a Swahili which kusikilika "is understandable". His comments reflect the perspectives of the other subjects in this group. If the scores of these subjects were not 177 counted, there would be no significant difference between groups’ responses. 3.2.2 Hypotheses 19-21 HYPOTHESIS 19: In all age groups, the vernacular speaker borrowing from the French lexicon will be most positively evaluated with regard to level of salary and level of education. HYPOTHESIS 20: From the old to young age group, there will be an increase in the number of subjects who positively evaluate the speaker who borrows from the French lexicon, with regard to level of salary and level of education. HYPOTHESIS 21: There will be no difference between evaluations of the standard speaker and the vernacular speaker who does not borrow from French with regard to level of income and level of education, except in the old-age group where the standard speaker will receive more positive evaluation. Tests. Scores were based on subjects’ responses to the following questions. What salary do you think [the recorded speaker] receives per month? 1. 0-1999 zaires (080-20) 2. 2-3999 (0320-40) 3. 4-6999 " (0841-70) 4. 7-ll,999 " (0870-120) 5. 12,000+ " (03120+) What level of education do you think he has? 1. primary school, 1-3 years 2. " 4-6 years 3. secondary school, 1-4 years 4. " " 5-6 years 5. university Total scores for each speaker represented the amount of times that he was given a higher or lower score than any of the other speakers. Ties were not counted. 178 Results. As the tallies in Table 3.23 indicate, Hypothesis 19 is clearly supported. However, Age Group Evaluation Oldest Middle Youpgest Re: salary level D V/-F H 0 O 0 L 25 23 31 StS H 2 2 2 L 2 O 1 V/+F H 24 22 22 L 1 1 1 Re: educational level V/-F H 0 0 0 L 25 23 31 .StS H 2 2 2 L 2 0 1 V/+F H 24 22 22 L 1 1 1 Table 3.23. Evaluation of speakers’ educational and salary level according to age groups. H: number of times speaker was evaluated as having the highest level; L=number of times he was evaluated as having the lowest level. Speakers: V/-F=vernacular without French, StS=standard Swahili, F/+F=vernacular with French. Hypothesis 20 is not supported since there is no significant difference between age groups in the proportion of subjects who evaluated speaker 3 most highly. Hypothesis 21 is not supported with regard to both the main prediction and expected exception. It predicts that among the middle and young aged subjects there will be no significant difference in their evaluations of the two speakers who do not use French. However, the great majority of both groups believed that the vernacular speaker had the 179 lowest level of salary and education. This is evident from Table 3.23; also, in a count of scoring that excluded speaker 3, highest scores were given to the standard speaker by 41 (87%) of 47 subjects with regard to salary and by 100% of 54 subjects with regard to education. Thus, there was no significant difference between the responses of subjects in these age groups and those of the old-age group who also rated the standard speaker highest (73% of 22 subjects with regard to salary and 93% of 27 subjects with regard to education). 3.2.3 Hypotheses 22 and 23 HYPOTHESIS 22: The standard speaker will be evaluated as an outsider more often by the middle and young- age groups than by the old-age group. HYPOTHESIS 23: All groups will judge the two vernacular speakers to be from the Bukavu area. Test. Scores were based on subjects’ responses to the question: Where does the speaker come from? 1. Outside of the Kivu province 2. From Kivu, but not Bukavu 3. Bukavu To test hypothesis 22, responses (1) and (2) were grouped together to signify the standard speaker’s being evaluated as an "outsider" and opposed to (3) which would signify his being evaluated as an "insider". However, the differences between age groups in their choices between (1) and (2) if they considered the speaker to be an outsider were also found to be of interest and are discussed below. 180 Result. The hypothesis were unsupported, not because there was no significant.difference between the groups but because a much higher proportion (p=.02) of the 919-age group viewed the standard speaker as an outsider (Table 3.24). Outside of Kivu but Age group Kivui_i not Bukavu Bukavu .Young 9(26%) 9(26%) 16(48%) Middle 0(00%) 10(31%) 22(69%) Old 3(09%) 18(56%) 11(35%) Table 3.24. Number of times standard speaker was considered as being from a certain area, by age group. Also of interest in the subjects’ evaluations is the difference in age groups with regard to just how much of an outsider the standard speaker was considered if he was not thought to be from Bukavu. The majority of both the old and middle-age group subjects (no significant difference) believed that the speaker was probably from the southern area of Kivu where people are stereotyped as speaking a "Swahili of the Arabs"--a more standard Swahili. However, half of the younger subjects (a difference from the other groups significant at the .001 level) perceiving the speaker as an outsider felt that he must be even farther removed from the Kivu scene. While some of them guessed that the speaker was from Shaba, others felt, in the words of one subject, "inaonekana hakuzaliwa mu Zaire (it’s obvious that he wasn’t born in Zaire)". Such subjects, then, seem to 181 perceive the standard speaker as speaking a foreign dialect, not just a diglossic church variety or a dialect particular to a different region in the country. As Table 3.25 indicates, Hypothesis 23 is also unsupported. The majority of those in the young and Percent of subjects evalu- ating speaker as from Bukavu Speakep Old Middle Young 1 81 44 38 3 16 44 44 Table 3.25. Vernacular speakers evaluated as to origin. Number of subjects: old=31, middle=32, young=34. ' ' middle-age groups believed the vernacular speaker who did not use French to be from outside of Bukavu while the large majority of those in the old-age group guessed the opposite. On the other hand, the majority of all groups guessed that the third speaker was not from Bukavu with the number of those in the old group significantly larger than those in the other groups. Several subjects stated that the first speaker sounded like someone from the numerically predominant local ethnic group (he was) and guessed that he was from one of the outlying villages of Bukavu. There were also several subjects who claimed that the third speaker who mixed in French did not sound like someone who grew up near Bukavu. They were correct. Though the speaker had lived in Bukavu for several years, he was born and raised in Lubumbashi. 182 Just what effect this recognition of accents had on subjects’ responses to other questions cannot be known for sure. With regard to the one who mixed in French, there would seem to be little effect since few, if any, subjects identified him as being from Lubumbashi and explanations of their evaluations generally focused on his use of French. The recognition of the villageois accent of the other vernacular speaker, though, could have been a key reason for subjects’ rating him relatively low in education and salary. A follow-up study would, of course, need to address this problem; the best way to do this would be to record the same speaker, born and brought up in Bukavu, giving the text in each of the three varieties, play one variety only to one large group of subjects, then compare results with each of two other, socially similar, groups of subjects each of which hear one of the two other varieties. This would also be a much more time-consuming and expensive test. 3.2.4 Hypotheses 24 and 25 HYPOTHESIS 24: In the old-age group, positive personality evaluation will be highest for the standard speaker. HYPOTHESIS 25: In the middle and young-age groups, positive personality evaluation will be highest fer the vernacular speaker who does not borrow from the French. Test. Scores were based on subjects’ responses to the question: Do you think this person: 1. is arrogant? 2. is average? 3. is humble? 183 Results. Table 3.26 indicates that the majority of subjects in each age group chose the vernacular speaker not using French as having the most pleasing personality. Rating by Speaker Age group i g 9 Old , Most pleasing 10 6 0 Least pleasing 0 2 25 Middle Most pleasing 16 7 0 Least pleasing 0 1 23 YOung Most pleasing 18 6 0 Least pleasing 1 2 29 Table 3.26. Evaluation of speakers’ personality. However, the numerical differences between the subjects’ evaluations of him and the standard speaker are not significant, either within groups or between groups.- Thus, neither hypothesis is supported. The most clear-cut aspect of the subjects’ evaluations is their very low rating of the speaker who mixed in French. 3.2.5 HYPOTHESIS 26: From the oldest to youngest age groups, there will be a decrease in the perceived value of learning standard Swahili. Scores used to test this hypothesis were based on subjects’ answers to the questions: 184 Do you think it is necessary for a person to know standard Swahili? 1. No, not at all. 2. It’s not necessary but is good. 3. Very necessary. Here in Bukavu teachers use much Swahili in the first two years of school. They then teach Swahili grammar for the next four years. Do you think it is good like that? Or do you think it wold be better to use and to teach French only from the very beginning? Or do you think it would be better to use and to teach Swahili more? 1. Continue as is done now. 2. French from the beginning. 3. More Swahili. Responses to the first question have already been discussed with regard to the testing of Hypothesis 5; it was found that there was indeed a significant decrease from the old to young subjects in the evaluation of the necessity of learning standard Swahili. Here, I focus on responses to the second question. Before testing for the significance of differences between age groups, I tested to see if, in the old-age group, educated subjects chose choice (3) more than uneducated subjects or if those with a high level of church involvement chose (3) more than those with average or less involvement. Neither of these variables made a significant difference in the responses. So, the chi-square test included the responses of all subjects. Results. Responses to this question gave further support to Hypothesis 27 (p=.001). As Table 3.27 shows, the 185 Language option Age Group 1 g 9 Old 5 (16%) 8 (25%) 19 (59%) Middle 3 (09%) 18 (56%) 11 (34%) Young 2 (06%) 31 (94%) 0 (00%) Table 3.27. Preferred educational language by age group. Language options: 1=Continue as is now, 2=French only, 3=More Swahili. Percentage figures represent proportion of subjects within an age group found in the particular column (these are not the percentages on which the significance test are based). majority of older subjects felt that there should be even greater use of Swahili in the educational system but the majority of the middle and young-age groups feel just the opposite--that Swahili should be dropped and only French should be used. Moreover, this opinion is increasingly prevalent: the large proportion of young subjects believing this is significantly larger than that of the middle—age group. 3.2.6 HYPOTHESIS 27: In the middle and young-age groups, self-evaluation of the understanding of the standard will be lower than in the oldest age group. Test. Scores were based on subjects’ responses to the question: Are you able to understand kiswahili without problem: a. when you are with friends? b. at work? c. on the radio? d. in church? Responses were marked on a scale of one to four: (1) only a little, (2) a fair amount (599ipi), (3) much, or 186 (4) everything. For testing of this hypothesis, the subjects’ evaluations of their understanding of Swahili on the radio and in church were of particular interest. Results. The hypothesis as stated is not supported since there is no significance difference between the old and middle-age groups’ responses. There is, though, a difference (p=.001) between the young-age group’s responses and those of the other groups. As Table 3.28 indicates, the majority of subjects in each age group evaluated themselves as being on the low end of the understanding scale with regard to the radio. However, the proportion of the youngest group is much smaller than that of the other groups in the highest category and is larger than the other groups in the second from lowest category. Little Some Much All Age Group 1 2 3 4 “as“ M Oldest 4 (13%) 13 (42%) 9 (29%) 5 (16%) Middle 4 (13%) 15 (47%) 8 (25%) 5 (16%) Youngest 4 (12%) 19 (56%) 10 (29%) 1 (3%) Table 3.28. Comprehension of radio Swahili by age group. Percentages refer to portion of subjects within an age group (not the percentages that were used to calculate significance level). Even if the scores of those who are not yet in their teen years are discounted, thus elimininating two of the lowest scores and seven of the next to lowest scores, there is still a significant difference between the youngest and 187 other age groups due mainly to the low number of youths claiming the highest level of understanding. The hypothesis is also unsupported with regard to church Swahili. Here, as Table 3.29 indicates, there are 1 2 3 4 Age group Little Some Much All Young 1 (3%) 5 (15%) 12 (35%) 16 (47%) Middle 1 (3%) 0 ( 0%) 11 (33%) 21 (64%) Old 0 (0%) 2 ( 6%) 15 (48%) 14 (46%) Table 3.29. Understanding of church Swahili by age group (percentages represent proportions within an age group. actually more young and middle aged speakers who rate themselves in the highest category than older speakers. Still, though, more youngsters rate themselves on the lower half of the scale than do subjects from the other groups combined. Tallies were done counting the scores of only the eighty-six subjects with average church attendance but the results were basically the same. 3.2.7 HYPOTHESIS 28: Self-evaluation of the ability to use Swahili will not differ among groups. Scores for testing this hypothesis were based on the question: Can you use Swahili in a way that pleases you-- to say anything, anytime you want to use it? 1. many times, no. 2. many times, yes. 3. always, yes. The difference between groups’ responses was significant at the .001 level. 188 Results. The hypothesis was unsupported because of the oldest group’s comparatively low selféevaluation! While the majority of those in both the middle and young-age groups evaluated themselves at the highest level, only a small minority of the oldest age group did (Table 3.30). 1 2 3 Age group Many times,no Many times, yes piyays Oldest 5 (16%) 17 (55%) 9 (29%) Middle 1 ( 3%) 11 (33%) 21 (64%) Young 0 ( 0%) 15 (44%) 19 (56%) Table 3.30. Self-evaluation of ability to use Swahili. Percentages refers to proportion of subjects within an age group. In the testing of Hypotheses 2 and 3 correlating scores of linguistic variables with age and educational level, it was seen that the phonological scores of those in the oldest age group who had formal education were significantly higher than those of subjects in the other age groups. Thus, the testing of Hypothesis 25 was redone to consider only those with a formal education. The retest did not affect the significance of the difference among groups: a majority of those educated in the oldest age group evaluated themselves as not being able to use Swahili in a way that pleased them, at least some of the time. 3.3 Discussion The most basic assumption underlying my methodology and formation of hypotheses was that age would make a difference in Bukavu residents’ linguistic performance and language 189 attitudes—-that one may observe diachronic developments in the Swahili spoken there. The testing of hypotheses has shown this to be a legitimate assumption and that, together, age and education are key variables in accounting for differences among Bukavu native residents in both linguistic patterns and language attitudes. However, the differences between the speech of subjects educated before Independence and those educated after Independence are not as clear cut as the hypotheses had predicted. Phonologically, as the hypotheses predicted: (1) there is a positive correlation between linguistic scores and level of education only in the old-age group, (2) the scores decrease (that is, the pronunciations become less standard) from the old, educated subjects to the young educated subjects and (3) in two environments mo(j)a and 1919:, the scores of the middle and young aged educated subjects were not significantly different from those of the old, uneducated subjects. However, scores in the :11199 environment do not accord with the hypotheses’ predictions. Though the middle and young-age educated subjects’ scores were less than the old, educated subjects’, they were greater than the old, ppeducated subjects’. This is in keeping with the observation made in section 2.9.1.2.2 that there is an apparent split in the diachronic development of the variable 1;); in one set of environments there is increasing 190 [+sonorant] pronunciation and in other environments there is increasing [-sonorant] pronunciation, perhaps after an initial period of oscillation between the standard pronunciation of old, educated speakers and the non-standard pronunciation of old, uneducated speakers. While the hypotheses concerning the effects of age and education were fairly accurate concerning the phonological variables, they had little validity with regard to the morphosyntactic variable. It was true, as hypothesis 2 had predicted, that the level of education correlated positively with standard scores in, and only in, the old—age group. However, the educated subjects of the youngest age group had scores as high or higher than those of the oldest age group and, thus, their scores also represented a divergence from the old, uneducated speakers rather than, as predicted in the hypotheses and true of the phonological scores, an equality with them. Thus, with regard to morphosyntax, the younger subjects might seem to be influenced in the direction of the standard, by example of their educated elders and/or by instruction. However, it is important to realize that the only environment in which the young, educated subjects’ scores were pignificantly higher than those of the old, educated speakers is the pi [upComplex] environment. This is an environment on the side of the semantic/syntactic scale that most favors use of pi regardless of age group, regardless of education. Since the young subjects have moved beyond the 191 old, educated subjects only in this environment, it cannot be firmly concluded that their scores reflect a movement resulting from exposure to the standard. Rather, it could reflect ongoing clarification of the morphosyntactic and semantic distinctions between environments mentioned in section 2.9.2. The fact that there is an increase in standard scores from the old, uneducated to the middle and young educated may reflect a bridging of the gap between differences between speakers stemming from ethnic mother tongue influence and differences stemming from an educated- uneducated dichotomy. Here, it may be recalled that when the ethnic groups’ use of 1pii in the [-Nominal] environment was measured in terms of education (Hypothesis 12), the scores of the Shi and "Others" were relptively the same for both the educated and uneducated; but, in each of these groups, there was a sharp increase from the scores of the uneducated to those of the educated. Lexically, level of education was clearly a key factor in predicting subjects’ incorporation of French into Swahili. Only two other factors, occupation and network strength, were found to significantly affect the lexical scores. Pastors and manual laborers incorporated French into their speech less than the other occupational groups. The manual laborers also had a lower average level of education than the other groups so their low French use could be 192 attributed to their relative lack of exposure to French in school. This is not so for the pastors, though. The pastors had a secondary school education, thus as much exposure to French as the other subjects, but they incorporated French into their speech much less than those with similar levels of education. Their avoidance of French, then, need not be stated negatively (lack of higher education). It may be stated positively in two respects. First, they are loyal to 99ii Swahili--a Swahili as free as possible from "contamination" by French. This loyalty is evident in their relatively high phonological and morphosyntactic scores as well as lexical scores. .Second, Swahili is for the pastors, more than any other occupational group, their occupational language. The sermons and the literary tools of their trade—-the Bible and theological literature-—are in Swahili. Thus, they are apt to have a much larger, active Swahili vocabulary than other groups who are more likely to switch between French and Swahili (see Fabian (1985) for discussion of manual laborers’ switching between French and Swahili) or use French almost exclusively in their occupations. Network strength scores correlated negatively with lexical scores. That is, as the density of one’s local social ties diminishes, the likelihood of incorporation of French into one’s speech increases. This may be explained in part by the test results concerning Hypotheses 24-25: the speaker who mixed French words in with his Swahili was 193 evaluated as "arrogant" by the large majority of subjects in all age groups. Those with close social ties would stand to lose favorable personality evaluation by their peers if they appeared to be using French more than necessary. However, those with less dense social ties may feel that they have less to lose in terms of personality evaluation and more t0' gain in terms of being perceived as being of a higher academic, if not financial, status than those who use French less (the concept of making language code choices in terms of social costs and benefits to the speaker is discussed in detail by Scotton (1972)). The connection between education, occupation, network strength and lexical scores is clearly evident in consideration of the administrators. Since their educational level is so high they are prime candidates for elite occupational positions. These positions involve frequent travel and/or contact with people from outside the Swahili speaking area of Zaire--a lessening of local network strength. This necessitates their use of French in work- related oral communication as well as in their paper work. Thus, education, occupation, and social contacts that supercede dense, local networks all necessitate the administrators’ thinking in French; Swahili plays a relatively insignificant role. It is little wonder, then, that the French lexicon is so readily available and used when they converse in Swahili. 194 Church attendance and ethnic group membership are two other variables that influenced linguistic choices. The former affected phonological and morphosyntactic scores; the latter affected only morphosyntactic scores. Church attendance was found to positively correlate with scores in two of the three phonological environments and in two of the three 1pii environments. This was true whether or not the pastors were included in the scores of the most frequently attending group and when signficant differences between age and educational levels were factored out. Apparent differences between sex in linguistic scores were eliminated when those with the highest level of church involvement were factored out.13 As mentioned in section 1.1.2, DeCamp pointed out that for decreolization--movement away from a vernacular to the standard dialect--to occur, there needs to be sufficient institutional pressure. The key institution for providing this pressure is the school. But, the results of the preceding tests indicate that, though the schools did influence students to incoporate Standard features into their speech in the pre-Independence era, this influence has greatly diminished. In contemporary life, this pressure has dropped off, leaving only the church as a source of significant institutional pressure. Frequency of attendance at church correlates with higher phonological and morphosyntactic scores in certain environments but it does not correlate with lexical choice 195 between French and Swahili alternatives. The vocabulary of the sacred generally does not carry over into the secular; the educated subject with higher church attendance is thus as prone to resort to French as the one with lower church attendance. The secular-sacred distinction is indicated in the results of testing Hypothesis 27, based on responses to the interview schedule. In all age groups, the large majority of subjects (82-97%) claimed a high understanding of Swahili in church but only a minority (32-45%) of each group claimed the same with regard to Swahili used on the radio. The effect of ethnic group was evident only in the morphosyntactic component. The effect remained even after differences of age, education, and level of church involvement were factored out. Thus, it might seem reasonable to assume that the effect of ethnic languages has persisted in the morphosyntactic component. However, as pointed out in section 2.9.2.3, a comparison of ethnic languages’ use of the copula does not clearly show this to be the case. For example, both Mashi and Kirega, the languages of the numerically predominant ethnic groups, have a form paralleling :59 that is used before adjectives so it is unclear why the Shi speakers would use pi more frequently than the Rega speakers. The two languages form relative clauses differently: Mashi by word order, Kirega by use of an unconjugatable verb somewhat similar to the vernacular pi 196 (exemplified in 2.9.2.3). If the ethnic language influence was to be manifested in the vernacular Swahili morphosyntax, it would seem most likely that the Rega speakers would use pi more often than the Shi speakers since there is a morphological parallel in their language. However, the opposite is true. An in-depth study of this problem is beyond the scope of this dissertation. Tests of Hypothesis 11 indicated that there is no significant difference between sexes with regard to any of the variables except for pronunciation of mo ' a in which female subjects’ scores are lower than those of the male subjects. This may indicate that female speakers are no more concerned with standardizing their Swahili speech than male speakers. It may also indicate that since it is not socioeconomically advantageous to know Swahili (except, perhaps, for those with a high level of church involvement), the question of prestige is irrelevant to the contemporary Swahili speech situation. Testing for the degree of standardness in French linguistic structures might produce quite different results. As mentioned in section 2.5.4, it is school girls who have been stereotyped as bringing the back velar French /r/ into their speech. It might seem that this apparent phonological innovation would be paralleled by differences in lexical incorporation of French into Swahili; but this was not found to be the case. 197 Responses to the interview schedule concerning language attitudes and use indicated that, diachronically, there is decreasing importance placed on learning standard Swahili but increasingly conficent self-evaluation of ability to use Swahili; however, evaluation of different Swahili varieties has remained relatively stable with regard to their signalling of educational and economic status. Perhaps the most important diachronic change in language attitudes concerns the importance of knowing standard Swahili. From the oldest to youngest age groups there is a steady decline in the percentage of subjects who believe that it is necessary to know standard Swahili. Accordingly, while the majority of old subjects felt that more standard Swahili should be used in the classroom, the majority of middle and young-age groups--with a sharp increase from the former (56%) to the latter (94%) felt that only French should be used in the classroom. The reasons subjects gave for their opinions reflect how the socioeconomic circumstances have changed over the generations. Subjects in the oldest age group believed that it was important to learn and use Swahili in school for the sake of communication with Swahili speakers of other areas and in situations other than personal interaction: Ni vema kusikiwa na kila ye "It’s good to be under- yote asemaye kiswahili. stood by whoever speaks Swahili." 198 Tusionekane kama watu wenye kupima Swahili. Lakini tuki— juwe kabisa. Ni vizuri kujua luga mingi. Wanafunzi wangekuwa na akili zaidi. "[So that] we won’t be seen as people who dabble with Swahili but as ones who know it completely." "It’s good to know many languages." "Students would be more intelligent." However, the youngsters’ explanations of their choice of "French only" for education indicate their perception of French as essential, and of Swahili as irrelevant, to future prosperity and the possibility of wide-spread communication- -with non-Africans as well as Africans from different areas:1‘ Juu nitajifunza kiswahili hata nyumbani. Kiswahili tunakijua tangu kuzaliwa... Kiswahili haina mafaa kwa masomo. Tunapashwa kuendelea mbele sasa. Swahili si luga ya progre. Hatuna lazima ya kujifunza Swahili bora sababu hatuna kazi nayo. Wakati tunaona watoto wa wakubwa wa inchi wanasema Francais tangu mwanzo wa masomo, tunafurahi sana na tunaona kama ile njo maendeleo. Ni lugha ya lazima sana. "I can learn Swahili at home." "We know Swahili from birth..." "Swahili has no value for studies." "We need to advance. Swahili isn’t a lan- guage of progress." "We don’t need to study standard Swahili be— cause we don’t need it for anything." "When we see that the country’s leaders’ speak French from the beginning of their studies, we are very happy and see that that is progress."15 "[French] is essential" 199 Ingesaidia kwa kujifunza "[Knowing French] would masomo. help out in our studies." Hatupashwe kubakia hapa Buka- "We don’t need to just vu tu fasi ya Swahili. Tuna- stay here in Bukavu, a ombwa kujua Francais zaidi Swahili area. We need juu ya maisha. to know French more to get along in life." Sababu ya kusoma vitabu us "[We need French] to kusumulia na wageni. read books and speak with foreigners."1° Kama unasumulia na bazungu "It’s French only if you ni francais tu. talk with foreigners." Kiswahli kitapotea mu inchi Swahili will disappear yetu. Ni Franqais njo itaba- in our country. It’s kia luga ya eshima. French which will remain as the prestige language. As predicted, all age groups evaluated the vernacular speakers who borrowed from French as having a higher level of education and income than the vernacular speaker who did not and the standard speaker. However, counter to the prediction of Hypothesis 20, there was no significant increase in this positive evaluation of the French borrower from the old to young-age groups. When these results are compared with those concerning the importance of learning Standard, it can be seen that, while the access to French signals intellectual and economic prosperity to all subjects, only the old subjects felt that standard Swahili was important to know. Though the old subjects’ reasons for this opinion, quoted above, refer mainly to greater communication abilities, there is also a socioeconomic reason for their 200 perception of standard as essential. In the era of their youth, French was, as it is now, the language of the socioeconomic elite. But, there was then much less opportunity for the positions held by French speakers to be obtained. Standard Swahili represented a much more obtainable socioeconomic status intermediate to the French elite and the masses who generally spoke an ethnic language. Their era was one in which the church (especially the numerically and economically predominant Catholics), upholder of Standard Swahili, had power nearly equal to the secular political administration and business (Young and Turner 1985) and controlled health-care, educational, and often agricultural institutions as well as the sacred ones.1' Thus, for the oldest speakers--and for them alone-- there was a socioeconomic incentive for learning the standard; the middle and young speakers realize how much the times have changed and how little they will benefit socioeconomically from knowledge of the standard. Though the old subjects considered standard Swahili worth learning, they, less than any other age group believed that the taped speaker using the standard could be from Bukavu; the majority felt that he must be from the area in south Kivu stereotyped as a place where inhabitants speak a more nearly standard Swahili variety. The middle and young- age groups evaluated the standard speaker as from Bukavu more frequently. Some said that he spoke in a way that one could hear in Bukavu but more of the subjects believed that 201 he learned to speak Swahili either by frequently travelling to areas where the standard is spoken or by making an unusual effort to learn the standard. While some guessed that the speaker was from Shaba, others felt, in the words of one subject, "ipponekana hakuzaliwa mu Zaire" "it’s obvious that he wasn’t born in Zaire". Such subjects, then, seem to perceive the standard speaker as speaking a foreign dialect, not just a diglossic church variety or a dialect particular to a different region in the country. The final aspect of the testing to be commented on here is the subjects’ self-evaluations of their ability to understand and use Swahili. Hypothesis 27 predicted that old-age group subjects would rate themselves higher than the others with regard to understanding the standard. Hypothesis 28 predicted there would be no significant difference between age groups’ evaluation of their ability to use Swahili as they pleased. Both hypotheses were wrong. The self-evaluation of the old-age group subjects was basically the same as that of the middle-aged subjects with regard to understanding Swahili on the radio (both groups had higher self-evaluations than the young subjects) but it was lower than that of the middle-aged and young-aged subjects with regard to understanding Swahili in church. Furthermore, the old subjects evaluated themselves lower than the middle-aged and young-aged subjects with regard to their ability "to use Swahili in a way that pleases you--to say any thing, any time you want to use it". The results 202 were the same even if, in the old-age group, only the scores of those with formal education were considered. This outcome indicates that those who were educated before Independence, when much more importance was placed on the learning of Swahili, may have felt, to some degree at least, the "desperate linguistic anxiety" which DeCamp (1971a:26) says is typical of those on the higher end of a creole continuum situation "proclaiming the superiority of their own ’standard’...while nursing inward doubts about whether [the variety they use] is really sufficiently standard". Though the old subjects did not claim that they speak in a superior way, their be1ief in the necessity of learning standard Swahili and using it in the classroom does give witness to the importance they place on knowing the standard. Their perception of the gap between their linguistic performance and desired competence causes them to rate themselves more negatively than the younger subjects who realize that they do not speak the standard but are not bothered since they can use Swahli for the situations in which they need it--situations which are irrelevant to intellectual or economic status. Notes 1. Note how close this score is to that of the administrators’. One might take this as indicating the teachers’ attempt to imitate the administrators’ prestigious 203 incorporation of French into the Swahili. I believe, though, that it is simply a reflection of the necessity of using French in the day-to-day work world.much more so than in the other occupations. 2. A subject was considered to have a high degree of involvement in the church if, in the interview schedule discussed in section 2.6, he reported himself as attending church more than once a week. 3. One of these frequently visits, for extended periods of time, close relatives who live along Lake Tanganika in a region stereotyped, at least, as an area where safi Swahili is spoken. 4. Old--7l% (17 occurrences); Middle--56% (84 occurrences); Young--65% (31 occurrences). 5. As stated in section 2.5.4, the hypotheses do not concern women from the old-age group since only an extremely small minority from that group had the same educational opportunities as men. 6. Tests of educated speakers only were also done within age groups. Significant differences remained in the [—Nominal] environment for the middle and young-age group subjects and in the [upComplex] environment for all age groups. Only the middle-age group had enough chances to test with regard to the [npSimple] environment. There was no significant difference between ethnic groups. Environment Shi Regp cherg Old . [~Nominal] 37 (76) 25 (12) 30 (26) [ueComplex] 92 (104) 95 (22) 64 (31) Middle [-Nominal] 35 (120) 25 (85) 43 (195) [neSimple] 57 (35) 45 (20) 56 (41) [urComplex] 92 (241) 90(195) 85 (207) YOung [-Nominal] 45 (72) 24 (25) 30 (27) [upComplex] 95 (99) 89 (19) 96 (28) Percent of_pi occurrences (total chances in parentheses). 7. In the [upComplex] environment, the difference between Shi and Rega subjects disappeared. Both had 94% pi scores (130 chances for the Shi subjects; 16 for the Rega). 204 8. Shi Rega Others Environment All -HCA pii :599 All :599 [-Nominal] 39. 30 25 25 36 34 [urSimple] 69 62 40 36 57 57 [eeCOmplex] 94 93 91 95 88 88 Percent of pi variants by ethnic group. Scores with all subjects counted compared with scores when High Church Attenders (HCA) are excluded. 9. Correlation coefficient equals .1689; p=.091. 10. While there is no significant difference between groups in the [-Nominal] environment, those with a Network Strength Score (NSS) of "3" have the highest pi scores (p=.001) in the two other environments. Network Strength Score Environment , i 2 3 g ; . 5 [-Nominal] 45 (ll) 40 (53) 41 (49) 36 (84) 38 (318) [epSimple] 50 (2) 56 (9) 89 (19) 50 (16) 57 (60) [erComplex] 86 (14) 94 (69) 97 (115) 91 (127) 92 (602) Percentage of pi (v. -ko) by Network Strength Score (explained in section 2.5.6. (Numbers in parentheses refer to total chances.) Those with NSS of "1" (only one subject) and "2" have lower pi scores than those at the "3" level but, when the scores of subjects with NSS of 1-3 are contrasted with those with scores of 4-5, the lower level NSS still have significantly higher scores (p = .02 in the [urSimple] environment and .10 in the.[upComplex] environment. 11. The subject was an educated member of the oldest age group. His high scores put the average for the "0" church involvement group--of which he was the only member--way above the average scores of the other groups (63 v. the next highest 25 for 1w)a- and 71 v. the next highest 28 for mo(j) . 12. The social factors associated with each of these three subjects are quite diverse, though each has at least a primary school education. The middle-aged one is a college graduate who teaches math and physics in high school. The young ones are a girl who is a senior in high school and a boy-chauffeur "truck-driver’s assistant". 205 13. Comparing scores by sex (and keeping educational levels constant), differences were found among those with a secondary school education in the three (ni) environments and the (w)a- and mo(j)a phonological environments. Differences were eliminated in all but the m01j2a environment when the high church attenders were factored out. 14. The following quotes are not necessarily in the exact grammatical/lexical forms that the subjects used. The interviews were not recorded; rather, the interviewer wrote down the subject’s comments as the interview progressed. 15. This is a fascinating observation, perhaps stated cynically, in that it underscores the typical hypocrisy of Zairean policy makers who produce regulations for the masses that they have no intention of keeping for themselves since it would interfere with their socioeconomic self- advancement; a national language should be used in the early education of all Zairean children, it is decreed, and the decreers send their children to French—only schools, preferably in another country: By the 1970’s schools became much more uneven in quality as they grew in number; opportunities for the child increasingly depended on access to the best schools... The top segment of the politico-commercial class began sending its off- spring to Europe for education, guaranteeing that their technical qualifications would be far superior to the sons and duaghters of humble strata. (Young and Turner 1985: 135) 16. Note here the implication that if you are going to read, the books are going to be in French. This goes along with our observation in section 1.3.4.3 that religious literature is practically the only literature available in Swahili. 17. Young and Turner (1985:112) point out how status was often determined by one’s association with the church: It was the missionaries, especially the Catholics, who through the school, their multiplex social connections with Zairians, and their function as moral custodians of "civilization," who often played the decisive role in conferring status. Chapter 4 Conclusion 4.1 General conclusions In light of the historical context of the development of Swahili in Zaire, models of language contact and change, and the results of testing the hypotheses, we may conclude that any potential for shift away from the vernacular towards standard Swahili that was present in the pre- Independence situation has given way to a situation which encourages the maintenance of a diglossic gap between the standard and vernacular varieties. Ferguson’s diglossia model was developed with regard to language varieties that had coexisted for "several centuries" (1959:332). Later expansions of his model often referred to language contact situations of relatively short duration such as those resulting from large-scale immigration or urbanization. In these expansions, focus was on the functional relationships of languages in these di- lpolyglossic situations. Little attention was given to the linguistic repercussions. In this study of Bukavu Swahili, we have been able to observe linguistic repercussions as well as social functional allocations. The linguistic variables studied, though few, have enabled an initial perspective on how features of different linguistic components have been affected by social forces in Bukavu’s complex speech situation. 206 207 We have seen that the Low variety (Bukavu Swahili vernacular) has gone from a relationship to the genetically- related High form (Standard Swahili) that could be appropriately described within the continuum model to a polyglossic relationship in which the Low variety of Swahili continues to develop in terms of the koine model but, in general, without its speakers attempting to model their speech on the High variety. This is due to the greatly increased accessibility of a High non-genetically related language which is associated with greater socio-economic and intellectual prestige than the High genetically related variety. The decrease of the institutional pressure for advancement towards the standard is seen in the decrease from old to young age groups of both the percentage of subjects who believe that it is necessary to know standard Swahili and the percentage of subjects who believe that standard Swahili should be used as a medium of instruction. The linguistic components most clearly reflecting this decrease in motivation to acquire the standard are the phonological and the lexical. In the phonological scores, there was a significant positive correlation between value placed on learning standard Swahili (a value which decreased diachronically) and the standard scores. However, even here it may be seen that the decrease in socioeconomic motivation to know and use standard Swahili has not resulted in a complete abandonment of the pre- up 208 Independence speakers’ advance along the continuum towards the standard. In the mo(j)a environment, the {-sonorant] (standard) scores of the middle and young educated subjects decreased from those of the oldest-age group who had formal education and were not significantly different from those of the oldest, uneducated subjects. However, in the -(j)ua environment, the scores of the middle and young educated subjects, though lower than those of the oldest, educated subjects, have increased from those of the oldest subjects without formal education. With regard to the lexicon, borrowing from French continues now as it did before Independence: level of education correlates positively with the quantity of French borrowings. The spread of education will result in the increased use of French lexical items in vernacular Swahili. The incorporation of French into the vernacular is somewhat slowed, though, by social restraints. The one who borrows heavily from French is perceived as arrogant, and those with high network density-~those whose behavior is subject to the close scrutiny of family and peers living in proximity-~will borrow from French less frequently than those with lower network density. In the morphosyntactic scores, while ethnic group has the most clear-cut effect, the effect of education is also apparent. In two environments, the ni scores are not significantly different between the oldest and youngest educated subjects but the scores of the middle and youngest- 209 age groups’ educated subjects are significantly higher than the scores of the old subjects without formal education. In the third environment, the middle and youngest-aged educated subjects’ scores are even higher than the old, educated subjects’; but it is not clear whether this is due to the continued effect of education or to a tendency to make categorical semantic distinctions between environments that would continue regardless of education. Thus, the 1pi1 and -(j)ua scores indicate that waves of change favoring the Standard that were set in motion in the pre-Independence era continue to have an effect, in spite of the decrease in institutional pressures which initially set off those waves. Even though there is now a lack of socioeconomic motivation to acquire the standard, the standard continues to be respected as "good" Swahili. The use of Swahili in the church coupled with sporadic use of standard in the classroom provides a pressure to maintain a passive knowledge of standard if only within limited domains. Subjects perceive church Swahili as easier to understand than the standard Swahili used by the media; the content of religious Swahili is more limited and the exposure to it more frequent. The lack of sufficient pressure to move from the vernacular to the standard encourages continuation of the koineization process. A koine variety may result from resolutions of educated versus non-educated differences in 210 speech that developed in pre-Independence (e.g., (j) realized as {-sonorant] in certain environments and [+sonorant] in others). It may also result from increased contact, stemming from greater mobility since Independence, between people of neighboring ethnic groups (whose mother tongues seem to have influenced the degree to which one of the (ni) variants is used) and between speakers of different Swahili speaking regions. Part of the koineization process is also reflected in the grammar books discussed in section 1.2. Their incorporation of wide-spread vernacular elements into their descriptions reflects a process opposite to that represented by the continuum model in which language development is from low forms to the H maintained by grammars: in the two Zairean grammars language development is from the H maintained by Zanzibar-based grammars toward the prevalent L forms. 4.2 Theoretical implications To study language variation and change in Bukavu Swahili, I have relied on models taken from studies of creoles, koines, and di-/poly-glossia. If there is any originality involved in this respect, it is not in the invention of a new model or the radical revision of an existent one. Rather, it is in the integration of various models to provide for a comprehensive perspective on a complex but sociohistorically unified language situation. 211 As I noted in chapter 1, there is often an attempt to explain the sociolinguistic relationships of language varieties in terms of one model exclusively. While there do seem to be paradigmatic instances of a speech situation which can most efficiently be described in terms of one model only, those, such as the one described in this dissertation, involving complex relationships between language varieties may benefit from integrated references to several paradigms. My argument that a pre-Independence continuum relationship between the Bukavu Swahili vernacular and Standard Swahili has given way to a diglossic situation has been supported by reference to the diachronic change in subjects’ evaluation of the importance of knowing Standard Swahili and having it used in school. This is in harmony with Thomason and Kaufman’s (n.d./1988) argument that "we must investigate the social setting before we can predict, or interpret historically, the areal diffusion of linguistic features" (p.61). My interpretation of the social setting of Swahili use in Bukavu is that, in Thomason and Kaufman’s terms, a "language shift" situation has given way to a "borrowing" situation. Each situation, they argue, is always manifested by the same pattern: a borrowing language may have a strong influx of lexical items but structural interference from the language borrowed from will be relatively weak. The opposite is true in the language shift situation. 212 If analysis of the Bukavu speech situation were based solely on the study of linguistic structures, it would be hard to draw any conclusion from this scheme. One could argue that the many non-standard features of the Swahili vernacular provide evidence that trends to incorporate Standard features (such as [-sonorant] (j) in certain environments) are only matters of borrowing from the Standard. However, one could also argue that the many non- standard features are the result of strong interference of vernacular patterns while the Standard is being acquired and that the observed trends to incorporate Standard features are actually indications that a shift is going on. The Swahili speakers’ resort to French, rather than Standard Swahili, might seem to indicate that there is no shift; but, it might also be argued that there is really a hybrid language that speakers are-shifting towards--a language that is characterized by Standard Swahili structures but extensive French lexical items. The tangled nature of such an assessment based soley on consideration of linguistic features gives evidence that Thomason and Kaufman are right to insist that understanding of the social setting of a language situation is a prerequisite for intepreting its linguistic features. Apart from direct questioning of subjects about their evaluation and use of Standard and vernacular Swahili forms, the social setting of Bukavu Swahili has been studied by analyzing differences in linguistic performance in terms of 213 differences in social groupings. Though this approach is not new, some of the findings may be of contemporary interest. First of all, it seems that exposure to a linguistic variety influences language variety as much as-—if not more than-~one’s assessment of a language variety’s prestige, of of the most common explanations in sociolinguistic literature of language choice. The great majority of subjects who evaluated taped Swahili varieties considered the vernacular speaker who frequently used French lexical items to be at a higher educational and salary level than the speakers who spoke the vernacular without French or the standard Swahili variety. This indicates that the prestige associated with French is high. However, the key factors indicating the extent to which speakers will incorporate French into their speech are those which indicate the extent to which they have been exposed to French--education and occupation--rather than how much prestige they place on it. The importance of exposure may be illustrated by considering the two subjects who were pastors. Both rated the vernacular speaker using French as highest in educational level and as high, if not higher, than the others in salary level but the pastors’ use of French lexical items was much lower than those of similar educational levels. This may be explained in part by the fact that their jobs necessitate both written and oral use 214 of Swahili much more than, say, a company secretary or a banker who must constantly use French during the work day. Similarly, all three primary school teachers evaluated. the vernacular speaker who frequently used French lexical items in his speech as having a higher economic status than the other speakers and having as high, if not higher, educational status. However, their own use of French lexical items was considerably less than that of the two secondary school teachers who, unlike the primary school teachers, use French categorically in the class-room. The secondary school teachers also had a higher average level of education (and thus more exposure to French) than the primary school teachers. The factor of "sheer exposure" is considered briefly by R. A. Hudson (1980:71) with regard to the effect of mass media on one’s speech. He laments the fact that little study has been done in this area. C. M. Scotton (personal communication) is presently doing research that is concerned with how borrowing (and code-switching) is affected by the speaker’s exposure to the language borrowed from. Exposure to various linguistic varieties is obviously related to one’s social network. The less dense one’s social network ties, the more likely it is that one will be exposed to different linguistic varieties-~as well as influenced by them. Though my study of the effect of network strength has depended solely on individuals’ self- reports, the results are consistent with those predicted by 215 Milroy and Margrain (1980) in the morphosyntactic and lexical components: those with the highest Network Strength Scores (NSS) have less frequent use of both standard morphosyntactic forms and French lexical items than those with lower NSS. However, there is no significant correlation between network density and phonological performance--the linguistic component with which Milroy and Margrain’s 1980 study is primarily concerned. In their study, the Belfast residents with lower NSS were likely to have more frequent day-to-day encounters with speakers having standard pronunciation associated with overt socioeconomic prestige. This would not be so for the Bukavu resident, though, who, no matter how loose his network ties in the Bukavu area, would not be likely to hear standard Swahili pronunciations unless in a religious setting or with a church person who has above- average church involvement. The reason for this is not that there is insufficient prestige associated with the Standard but that the prestige associated with the Standard is--for want of a better word-- diglossic rather than socioeconomic. That is, the prestige of Standard Swahili is based on evaluations of how language should be spoken with regard to notions of correctness rather than on considerations of socioeconomic status. This distinction between types of prestige is evident in subjects’ evaluations of the Swahili varieties recorded for the interview schedule. The great majority evaluated 216 the Standard speaker as having a better Swahili than the others but also evaluated the vernacular speaker frequently borrowing from French as having a higher level of education and salary. The socioeconomic importance of French and general socioeconomic irrelevance of Standard Swahili gives little motivation for most speakers to incorporate Standard forms into their speech in spite of their admiration of them. Apart from speakers brought up in the pre- Independence era, for whom the Standard could mark socioeconomic prestige as well as diglossic, the speakers whose speech is most affected by the Standard are those whose involvement with the church encourages focus on the variety of diglossic prestige. In section 2.5.4, Trudgill was quoted concerning the effect of sex on linguistic behavior: "women...consistently use forms which more closely approach those of the standard variety or the prestige accent than those used by men (1974:91; my emphasis). It is unclear whether or not the second of the underlined phrases is intended to be a paraphrase of the first. In his 1983 discussion of "sex, covert prestige and linguistic change", Trudgill opposes dialects of covert prestige--that is, non-standard dialects that are associated with belonging to an in-group—-with those of overt prestige. Statements such as the following indicate that he does equate the overt prestige dialect with the standard dialect: 217 ...we have been able to demonstrate that it is possible to obtain evidence of the ’covert prestige’ associated with non-standard varieties, and that, for Norwich men, working-class speech is statusful and prestigious. The clear contrast [of male subjects’ scores] with scores obtained by female informants...indicates that women are much more favorably disposed towards [middle class] standard forms. (ibid., 177) The study of Bukavu Swahili, however, indicates that the distinction between covert (in-group) and overt (associated with non-in-group norms) is insufficient. There are also two types of overt prestige: diglossic and socioeconomic. The prediction of my Hypothesis 11-—that female subjects in the middle and youngest-age groups would have a significantly higher rate of standard Swahili pronunciations and morphosyntactic structures than male subjects with similar educational levels-~may have been wrong not because of the lack of prestige associated with standard Swahili but because the prestige is diglossic rather than socioeconomic. The large majority of female subjects rated the Standard Swahili as speaking the best Swahili but, since the prestige of speaking the variety well has nothing to do with socioeconomic prestige, it does not affect their linguistic behavior. The relevance of this distinction to studies such as Trudgill’s is that unless it is clear whether it is diglossic or socioeconomic prestige (or both) that is being talked about, the discussion of social motivations influencing linguistic variation may be muddled. Trudgill says: 218 Why it should have been possible to obtain this sort of evidence [based on self-evaluations of linguistic behavior] of covert prestige from Norwich speakers but not from New York speakers [with regard to (r) pronunciation] it is difficult to say. This may be due to the fact that [working class] speakers in Britain have not accepted [middle class] values so readily or completely as [working class] speakers in America. (ibid., 178) In terms of the distinction I am proposing, the reasons for this difference may be that in New York the prestige associated with middle class speech patterns is primarily socioeconomic (or both socioeconomic and diglossic) whereas in Britain it is primarily diglossic. That is, in New York City the variable under consideration may be primarily perceived as a marker of socioeconomic prestige, but in Norwich, the women may over-report their standard uses of certain variables--while men under-report--primarily in light of diglossic norms, with socioeconomic prestige being secondary or irrelevant. The influences on linguistic choice of exposure, socioeconomic prestige and diglossic prestige may be hierarchically related as indicated in Figure 4.1 Diglossic Prestige Socioeconomic Prestige E X P O S U R E Figure 4.1. Hierarchy of influences on linguistic choice. The lowest layer "exposure" is the basic determinant of linguistic choice--the sine qua non. The likelihood of 219 using a particular variant is directly related to the frequency of exposure to that variant, regardless of the prestige associated with it. Then, socioeconomic prestige will generally have more influence on linguistic behavior than diglossic prestige. Thus, in the Bukavu situation, Swahili speakers will borrow from the socioeconomically prestigious French rather than than the diglossically prestigious Standard Swahili. However, where socioeconomic prestige is not perceived as a factor and when strictly linguistic forces are not decisive, diglossic prestige may combine with exposure, to influence one’s speech patterns. This may explain why educated speakers increasingly pronounce (j) in :199 "known" as standard {-sonorant] but (j) in 9919 "one" as non- standard [-sonorant]. As pointed out in section 218.1.2, phonological conditions for sonorization of (j) in 9919 seem fairly clear but this is not the case for :199. Thus, subjects’ exposure to standard (j) in school and church may be sufficient to influence their pronunciation in favor of the standard, in spite of the lack of socioeconomic prestige associated with it. The same principle may also explain why the educated subjects of the middle and youngest—age group have continued to use standard pi at a frequency similar to that of the subjects who were educated before Independence, rather than reverting to the lower frequency levels of those without education in the oldest-age group as they did in pronunciation of (w) and 991119. 220 4.3 For further study Because of this study’s broad scope, there are, of course, many areas touched on that could be studied in much more detail. A narrower, more in-depth focus on a particular issue could be accompanied by more refined methodological tactics. Much more work on the morphosyntactic variable (ni) remains to be done. Follow-up work would involve two key issues. First, it should clarify how ethnic languages and innate tendencies have interacted to determine which variable is most likely to be used in which environments. Second, it could give further insight into the nineteenth century westward movement of Swahili from the Tanzanian coast. This study would necessitate study of both pi/:59 variation in Swahili and of ethnic language equivalents in areas throughout Tanzania as well as in Zaire. Areal survey techniques would have to be developed to provide this historical and regional perspective. Two main methods would be used to gather the data: 1) spontaneous interviewing with questions asked that are likely to produce many statements with present tense copulas; and 2) translation questions assuring examples of occurrences in the many different relevant environments. The elicitation of ethnic language equivalents could probably proceed with a straightforward translation exercise. 221 With my present data base, I would like to study how the use of class markers has been developing. In the Bukavu vernacular, the basic concord system used is certainly more complex than the animate-nonanimate dichotomy that is often claimed to characterize Zairean vernaculars. My preliminary analyses have indicated that here, as with (ni), there is a good deal of variation according to both grammatical environment and speaker. Study of this variable might be more indicative of diachronic developments in the morphosyntax than the study of (ni), which is complicated by the question of the degree to which ethnic language and/or innate tendencies have influenced its development. With regard to the lexicon, the study of the incorporation of French into the vernacular Swahili may be expanded by considering: (I) phonological distinctions between borrowing and code switches; (2) borrowing of core lexical items that have commonly used vernacular equivalents; (3) how borrowing differs between subjects who have a high degree of education and those who do not; (4) how subjects’ lexical scores used in testing of this dissertation’s hypotheses would be affected if topic is more controlled; and (5) how the scores would be affected if switches were discounted (assuming a fairly reliable method for distinguishing between switches and borrowings could be worked out). With regard to social factors studied in this research, further research could focus more on middle-aged and young 222 subjects who have no formal education and on those in all age groups with low levels of church involvement. In this dissertation, the focus was on Middle and Young age subjects with formal education for two main reasons. First, hypotheses predicted that an advance toward acquisition of standard features by those educated before Independence would be discontinued by those educated after Independence because of new social attitudes viewing French as the sole High worth acquiring. Second, a primary education, at least, is increasingly available to Zaireans and statistics indicated that a large majority of Bukavu children do receive at least a primary education. If those with formal education are not acquiring standard speech patterns, it is doubtful that those without formal education are doing so. However, it would be interesting to concentrate more on those without formal education to see whether developments in their linguistic behavior and social attitudes parallel those observed in this study or whether they are diverging. This might also indicate how rural dialects are developing since the educational levels are much lower in the rural areas. Although my hypotheses predicted that pastors’ sociolinguistic behavior and attitudes would differ from most other subjects, I did not consider degree of church involvement for the laity to be an important factor until well into my research. However, statistical tests have indicated that this is a significant factor. The large 223 majority of the subjects selected for this study attend church once a week and several attend it more often. How representative this proportion is of the Bukavu population in general is not known. Methods for selecting subjects were biased towards choosing those who attend church at least once a week. A sociolinguistic study of those with low levels of involvement would enable a more conclusive evaluation of the effects of this factor. It is unfortunate that the sociolinguistic influence of an institution as socially and, often, economically and intellectually important as the church--especially in third world countries-~has received so little attention in the sociolinguistic literature. Finally, this study should be complemented by other regional studies in Zaire with particular regard to the koineization process: are the various regional dialects converging in a manner that will ultimately produce a "Zairean Swahili"? Or are the vernaculars of urban centers developing independently of each other? 4.4 Language policy issues It is hoped that this dissertation may be of interest to Zairean linguists and language policy makers, especially concerning the policy of language use in primary education and the forms of Swahili to be used in mass communication for citizens who do not have adequate recourse to French. With regard to education, key issues are students’ lack of 224 socioeconomic motivation to know standard Swahili—-coup1ed with their high motivation to acquire French--and teachers’ frustrations with the program for use of Swahili in the school (Masumbuko 1987; licence thesis done under my direction). With regard to mass communication, the translation-preference test reported on in section 1.3.4.2 is relevant to secular communication as well as religious, especially in light of subjects’ rating themselves higher in understanding of "church Swahili" than "radio Swahili" (Chapter 3, Hypothesis 28). If subjects prefer Lower Swahili versions of literature for a domain in which they have relatively frequent exposure, they would probably also prefer Lower versions for literature concerning domains in which they have had much less exposure. Appendix A A Comparison of Varieties of Swahili: Standard coastal (SC), Standard Zairean (SZ)1, and Bukavu (B), It may be assumed in reading the following that every feature cited as a characteristic of Bukavu Swahili is also a feature of Lubumbashi Swahili, unless a footnote indicates otherwise. Descriptions and/or texts of Lubumbashi Swahili on which I base my comparison are found in Polome (1969), Fabian (1982), and Mwamba (1985). Also, where there is no difference between SC and 82, I will simply refer to "standard". 1. Phonology Vowels, consonants, and syllable patterns are presented in this section. 1.1 Vowels All varieties have a five vowel system: i, e, a, o, u. 1.2 Consonants All phonemes listed in the chart are, according to Hinnebusch and Mirza (1979), found in the standard coastal dialect. If a phoneme is circled, it does not appear in any of the Zairean varieties. There are no phonemes that occur in the Zairean varieties but do not occur in the standard coastal variety. Parentheses around a phoneme indicate varying pronunciation of it in both the Lubumbashi and Bukavu varieties. The Zairean equivalents of the circled sounds and the variants of the sounds in parentheses are listed below the chart. 225 226 Bila- Labio- Inter- Alve- Alv- Pala- bial dental dental olar pa . tal Velar Glottal stop - voiced p , t (k) + voiced b d (8) affricate - voiced c + voiced (J) fricative - voiced f 6' s s ' (h) + voiced v 3 z X nasal , m n n 9 cluster mb mv nd nj ng lateral l flap (r) glide (w) y In Zaire: SC [57 merges with /s/ (hadifli "story" becomes hadisi). SC /3/ merges with /z/ (9999i "sin" becomes zambi). SC /X/ merges with /g/ (1999 "language" becomes 19g9). SC lg] merges with /ng/ (99999 "cow" becomes ngombe). 2227 Variant pronunciations in both B (and L): (g) (j) (b) (r) (v) Phonetic Realizations It], [9] [9]! [j], [h], [r], [u], [k] [y], [0] [I] [l] [b] Especially in the Enziseoesot-2!a see sec. 2.8.1.2 final syllable Ease!!! pifkla 'cook' a-na-enda-(kla 'he often goesII 3sg.-pres.-go-habitual final syllable of Iain text all all class 2 prefix kidofglo 'little' aofjla 'one' (hlapana 'no' -fora(h)i 'be happy' auzu(r)i 'good' -(r)udia 'retutn to' 1.3 Syllable patterns (ea-)toto 'children' tu-(va-lona 'to see thee' CS and 28 syllabe patterns are the same. have common variants. common but not necessarily categorical. Slender! +voeel +vovel _-front -front [ mm] {H mm] [Wolves -front -fron i/__u cl.9llo pfx. bu-Icl.l4 pfx. Exaeple Stan!g agou -kaa -kofloa atu -aeka nchi nova usiku Q augulu 'leg' -kala 'sit' -tohola 'cough' auto 'person' -aauta 'arise' inchi 'country' ialwa 'dog' busiku 'night' Both 8 and The patterns listed as B and L are I. 228 2. Morphosyntactic differences In the following, I shall present dialectal differences as if they are categorical but this is simply for brevity of presentation. There is, of course, variatibn in this area of linguistic performance as well as in pronunciation. I am simply presenting what are probably the most common tendencies of vernacular speech. It may be pointed out that, in spite of the-important differences between B and standard listed below, the varieties have more similarities than differences. For example, use and formation of the following verbal categories are the same in both varieties: simple past, narrative, present and future tenses; causitive; prepositional/applicative; passive; stative; and general verb structure. 2.1 Class distinctions While Standard has 15 noun class distinctions, B commonly has only 7: when the reference is to a human, mg;(s.)/b§;(pl.) are used; when to non-humans, 9/5523; to form diminutives, §§:/EE;; to form the augmentative singular, lg; is used as an augmentative but the plural is formed by adding ma- to the singular augmentative. 229 2.2 Locative prepositions . Where Standard will use the nominal suffix 2g; to indicate location, B will use 52 or 33. Standard: nyumbani "at/in/to the house" B: ku nyumba "at/to/etc. the house" mu nyumba "in the house" 2.3 Copulas Standard uses one set of copulas exclusively for predications concerning location (:59, :29, or 2mg, depending on the specifics of the location) and only the copula pi for all other predications. In B (there are no reliable studies of this variation in other Zairean vernaculars), only :52 of the first set is used but the rules distinguishing its use from the use of g; are much more complex than in Standard. See section 2.8.2 of main text. 2.4 Tense and Aspect 2.4.1 Progressive B uses :52 before a verb in present tense apparently to signal either present progressive or immediate present-- or both. I haven’t yet been able to study the significance of this use of :52 in detail, though it is a fixed feature of B. 230 For the past progressive, B prefixes -na- to the main verb whereas Standard uses -ki-: Standard: nilihuwa nigihitaji "I was needing" B: nilikuwa nigghitaji 2.4.2 Perfective In B either the present or past tense marker is used instead of the standard’s :23; used to signal positivee perfective. Generally, if the idea would be stated in the present perfect tense in English, B will use :li; (past tense marker): Standard 9 Nigganza kazi. Nilianza kazi. "I have started work." Aggkwenda Goma. Alienda Goma. "He has gone to Coma." If the idea would be stated in the present tense in English, the -na- tense marker is used. So, verbs of state that would necessitate -me- in the standard take -na— in B: Standard B Wamglewa. Wagglewa. "They are drunk." Tumgchelewa. Tugachelewa. "We are late." B, as well as standard, uses the verbal prefix -ja- to signal negative pefect tense: STANDARD/B: Hajgfika "He hasn’t yet arrived" 231 As in standard, B has a completive, perfective aspect formed with huisha "to finish" as an auxiliary verb but the -li— marker is used rather than -me-: Aggsha jua Alisha Jua. "He already knows." 2.4.3 Conditionals3 The realis conditional of standard and B differ only phonologically. B uses only one of standards’s two irrealis conditional morphemes. -ngali-, the standard morpheme that is not used as an irreslis conditional has a different meaning in B (see 2.4.4). Standard B Aginiambia Aggniambia "If he tells me" Agggniambia Anggniambia "If he would tell me" Aggaliniambia Agggniambia "If he had told me" 2.4.4 The "still" tense‘ While standard has no "still tense", B does, signalling it with a that is phonetically equivalent to the standard’s past irrealis conditional’: B: Ungali kula! "You are still eating!" Ungali hapa. "You are still here." 2.4.5 Habitual In CS, the hu- verbal prefix signals habitual or customary action. In ZS as well as B, however, the suffix 232 -ka (or -§a) is added to conjugated verbs in the present tense: Standard: Watoto hgfika saa mbili ZS/B: Batoto banafikagg saa mbili. "The children usually arrive at 8 o’clock." 2.4.6 Distant time Hononymous with the ZS/B habitual marker is the suffix which, depending on the tense marker with which it cooccurs, marks distant future or distant past. Standard does not have a parallel morpheme. ZS/B: Utgendagg siku gani? "What day will you go?" Uliendagg siku gani? "What day did you go?" 2.5 Differences in verbal morphology Generally, the B singular imperatives are formed in the same way as standard imperatives, except for the three standard exceptions: Standard B Lete! Leta! "Bring!" Nenda! Kwenda! or "Go!" Uende! Njoo! Kuja!/Kuya! "Come!" The B plural imperative is the same as the second person plural subjunctive form. Thus, it differs from the 233 standard in that the -ni suffix is not used and the subject prefix is kept: Standard B Semeni! Museme! "Speak you all!" As in the singular, the B plural does not take the standard irregular imperative forms: Standard B Nendeni! Muende! "Go you all!" 'Njooni! Mukuye! "Come you all!" In B, the ku- infinitive marker is usually retained for the monosyllabic verbs--regardless of tense or mood. Thus. it differs from standard in wich the ku- marker is dropped in several verbal forms. Standard B hakula hakukula "He didn’t eat" hali hakule "He doesn’t eat" asile asikule "He shall not eat" ale akule "He shall eat" akila akikkula "If he eats" akala akakula "He ate" (narrative) aliyala alizikula "He ate them" Notes 1. I consider Nkiko et al. (1983) as the representative of .Zairean Standard Swahili. 2. To say that a noun has a 0 prefix is to speak in terms of concord rules more than the actual form of the noun in isolated context. Thus "chair" is kiti, "rock" is 'iwe, and "house" is nzumba in both 08/28 and B but in 08/28 different concord patterns are governed by the prefixes whereas B has the same pattern: 234 Standard B ki-ti ki-mepotea kiti i-napotea "the chair is lost" ji-we li-mepotea jiwe i-napotea "the rock is is lost" O-nyumba i-mepotea nyumba i-napotea "the house ' is lost" 3. Polome claims that specifically conditional morphemes are not used in Lubumbashi Swahili (1969:229) 4. About to use this label, I thought I better check with Comrie to see if he had a more esoteric sounding label. No; he also calls it the "still tense". He says that this tense "in the strcit sense of a single grammatical category.... seems[s] to be extremely rare cross-linguistically [but is] attested in a number of Bantu languages (1985:53). 5. Polome analyzes this as a "remarkable archaism" consisting of "the -nga- tense of -li- ’to be’ (with the meaning Bishop Steere illustrates for archaic kiUnggja with examples like niggali hai ’while I am still alive )" (1969:230). APPENDIX B Zairean Swahili Features Cited as Deriving from Local Language Influence In the following, the base form listed in the "Feature" column is the Standard Coastal Swahili form. The derived feature is the common Zairean form. 7 of the 10 Zairean Swahili features cited by LeCoste (1948) and/or Polome (1969;1983) as due to local language influence have also characterized coastal dialects; the dialects having these features are given in the right-most column. Coast- Cited al Di- Feature Exam le 1&1 alect2 Phonological patterns [j]~,[y] —jua~ -yua "know" P.129 Am VV «vVlV -kaa~'-kala "sit" P5224 Mg, Mr 3m~amu/ 1,3 class marker mtu~'mutu "person" Pu225 TD L294 m—vmu/ __ [-voca1ic] amkaovamuka "rise" P3225 TD flm’i/__N nchi~rinchi "country P5225 Class 9/10 mbwa~vimbwa "dog" ##uawbu/class 14 marker usiku~lbusiku "night" P5221 L296 Morphemes ka- diminutive sing., katoto "small child” P5226 DS not used in standard L297 4tu- diminutive plural, tutoto "small children" Pb226 L297 -ni lost and replaced nyumbaniA—ku nyumba "at P3226 Mrs by £2 "at, to, etc." the house" Mg,Ng -ka "habitual verb anaendaka "he often L298 Mr“ suffix". goes" 235 236 1. P.=Polome (1983); Pn=Polome (1969); L=LeCoste (1948). 2. Except for the first feature ([j] [y]), all information on the coastal dialects is taken from Sacleux (1909) who gives the following descriptions of the dialects listed: Abbgeviation Standgifor Description DS Dialectes Sud Mg Kngao dialecte de Mkao, parle du Fufidyi au Mozambique: centre principal Kilwa. (p.19) Mr KiMrima dialecte parle de vanga au Rufidyi (p.19); sur la cote faisant face a l’ile de zanzibar (p.7) . Ng KiNgozi dialects poetique, libre et visant a l’arcbaisme (p-ZO) TD Tous dialectes [j] [y] was cited in Stigand and Taylor (1915:34) as a feature of Kiamu, "the dialect of the town of Ann, or Lamu, on the island of that name." 3. Polome incorrectly says "the influence of the local languages is especially obvious in the shape of the class prefixes as shown by [the rule marked by this footnote)" (1969:225). 4. According to Sacleux (1909:311): "[ka- est un] prefixe nominal de qqs diminutifs plus emphatiques que les memes avec pref. ki-:...DS.... Dans les dialectes orientaux ces diminutifs sont des exceptions et n’ont pas de pluriel. Par contre le Dial. [Kingwana] emprunte volontiers aux langues voisines...le procede de former des diminutifs en -ka avec plur. tu-(tw-).... But it may be pointed out here that 52: is heard much more frequently than BB: in Bukavu, also. 5. "par emprunt aux langues voisines" (Sacleux 1909:620) 6. LeCoste realized that this was a feature of a coastal dialect but assumes that, rather than being brought from the coast, it "est reapparue" (1948:298). Polome, on the other hand, considers it as one of two "remarkable archaisms" (1969:230). Appendix C Texts Used in Translation Preference Test (1.3.4.2) MIDDLE V. LOW The follow pair of texts gives the Middle (Protestant Habari Njema) and Low versions of a passage adapted from St. Mark (see footnote 16, chapter 1 for explanation of why the actual Biblical passage was not used). The texts were numbered and presented side by side, on the same sheet of paper, to the readers (not as given below, of course) under the title "Mwalimu Kapanga". Beneath the texts, were the questions: Kwa maoni yako, namna gani ya kuandika ni mzuri zaidi kwa kusoma na kufahamu vizuri? Kwa sababu gani? "In your opinion, which way of writing is better for reading and understanding well? Why?" Below, the first line is the Protestant style and the second is the Low version (written by myself). a Wakati Mwalimu Kapanga alipokuwa akipita kando ya ziwa la Wakati Mwalimu Kapanga alikuwa anapita kando ya ziwa ya When Teacher Kapanga was passing side of lake of b Kivu, aliwaona Mirindi na ndugu yake Wasso. Walikuwa Kivu, aliona Mirindi na ndugu yake Wasso. Walikuwa Kivu he.saw.them Mirindi and relative his Wasso. They.were he.saw c wakivua samaki. Mwalimu Kapanga akawaambia: "Mukuyei; wanavua samaki. Mwalimu Kapanga akawaambia: "Mukuye; fishing fish. Teacher Kapanga he.told.them Come d masomo yataanza kesho." Mara moja, wakaenda kwao, masomo itaanza kesho." Mara moja, wakaenda kwao, school will.start tomorrow time one they.went to.theirs e wakakamata vitabu vyao, wakamufuata. wakakamata mabuku yao, wakamufuata. they.took books their they.followed.him f Wakafika Goma. Siku ya kwanza, wakaingia katika Wakafika Goma. Siku ya kwanza, wakaingia mu They.arrived Goma Day of first, they.entered in 237 238 8 masomo. Mwalimu Kapanga akafundisha. Wanafunzi masomo. Mwalimu Kapanga akafundisha. Wanafunzi school Teacher Kapanga he.taught. Students h waliokuwa wakimusikia kwa mars ya kwanza walishangaa wenyi walikuwa wanamusikia kwa mara ya kwanza walishangaa who.were hearing.him for time of first were.surprised who were i sana namna alivyofundisha, kwani hakuwa saws na sana juu ya namna alifundisha, sababu hakukuwa saws na very because of way which.he.taught because he.wasn’t same with he.taught J walimu wengine wasiotayarisha masomo vizuri. walimu wengine wenye hawatayarishake masomo muzuri. teachers others who.don’t.prepare lessons well. who don’t.prepare k Lakini, mwana wa mukubwa mumoja aliyekuwa na roho Lakini, mwana wa mukubwa moja mwenyi alikuwa na roho but son of big.shot one who.was with spirit who was 1 mbaya aliingia ndani ya masomo na kupaza sauti, akisema: mbaya aliingia mu masomo na kulalamika, aseme: bad he.entered in of school and raise voice saying: cry.out m . "Una nini nasi, wewe Mwalimu Kapanga? Ulikuja you.have what with.us you Teacher Kapanga You.came "We Kapanga, tuko na neno gani, sisi na weye? Ulikuja You Kapanga we.are with word which we and you You.came n kututesa? Ninakutambua wewe ni nani: wewe ni mwalimu to.bother.us I.reoognize.you you are who you are teacher kututesa? Minajua uko nani: uko mwalimu to.bother.us I.know you.are who you.are teacher 0 maskini tu!" maskini tu!" poor only 239 P Mwalimu Kapanga akamuhamakia yule mwanafunzi, Mwalimu Kapanga akamutombokea‘ ule mwanafunzi, . Teacher Kapanga spokecsternly.at that student q .- akamwamuru: “Nyamaza, toka somoni!" Mara moja, yule aseme: "Nyamanza, toka mu somo!" Mara moja, ule he.ordered Be.quiet leave.from in.class time one that in class r mwanafunzi akanyamaza. Akatoka. Alikuwa akisikia haya mwanafunzi akanyamanza. Akatoka. alikuwa anasikia haya student he.was.quiet he.left he.was feeling shame s sana. Wanafunzi wote wakashangaa sana, wakaulizana: sana. Wanafunzi wote wakashangaa sana, wakaulizana: very students all were.surprised very they.asked.one.another t "Neno gani hili? Mwalimu ana nguvu!" Hata mwana wa word which this teacher he.bas strength even son of "Hii ni neno gani? Mwalimu iko na ngufu! Hata mwana wa this is word which teacher he.be with strength even son of u mukubwa anamutii." Na kwa wepesi habari za Mwalimu mukubwa anamuheshimia." Na kwa upesi habari za Mwalimu big.shot he.obeys.him and by quickness news of teacher v . Kapanga zilienea po pote katika muji wa Goma. Kapanga zilifika fasi zote mu muji wa Goma. Kapanga spread place all in town of Goma (a)When Teacher Kapanga was passing by the shore of Lake (b) Kivu, he saw Mirindi and his relative Wasso. They were (0) fishing. Teacher Kapanga said to them: "Come; (d) school starts tomorrow." Right away, they went to their homes, (e) got their books, and followed him. (f) They arrived at Goma. Monday, they went to (3) school. Teacher Kapanga taught them. The students (h) who were hearing him for the first time were surpised (i) very much at the way he taught, because he was not like (j) other teachers who don’t prepare their lessons well. (k) However, the son of a bigshot who had a bad spirit (1) came into class and shouted (m) "What do you have to do with us, you Teacher Kapanga? Did you come (n) to bother us? I know who you are: you are only a poor teacher!" (p) Teacher Kapanga came down on that student and (q) ordered him: "Be quiet, leave the classroom!" At once, 240 (r) the student was quiet. He left. He was ashamed. (a) All the students were amazed, they asked themselves: (t) "What’s going on? This is a strong teacher! Even the son of (u) a bigshot obeys him." Quickly, news of Teacher (v) Kapanga spread throughout the town of Cons. I. This is the one instance in which the orthography of the Protestant version was changed from 1 to reflect the usual vernacular pronunciation. HIGH V. MIDDLE The following pairs the High Biblia Takatifu Catholic version (first) of excerpts from Isaiah 1:1-20 with the Middle Habari Njema Protestant version (second). The order of some lines in the Habari Njema selection have been altered to fadilitate comparison between the two'versions. l. Haya ndiyo aliyoyaona Isaya, mwana wa Amozi, these indeed which.he.saw Isaiah son of Amos Hays ni maneno Isaya, mwana wa Amozi, aliyofunuliwa these are matters Isaiah son of Amos which.he.was.revealed katika habari za Yuda na Yerusalema. in news of Judah and Jerusalem na Mungu juu ya inchi ya Yuda na muji Yerusalema. by God about of country of Judah and town Jerusalem. 2. Enyi mbingu, sikilizeni; You heavens listen Bwana Mwenyezi anasema hivi: "Sikiliza Ee mbingu, Lord Ruler he.speakes thus Listen ee heaven nawe nchi, tega sikio, kwa maana Bwana amenena: and.you country set ear for reason Lord has.spoken nawe dunia utege sikio. and.you earth set ear Nimelea wana na kuwalisha, nao I.have.raised sons and feed.them and.they Nimewalea watoto, nimekomalisha, I.have.raised.them children I.have.made.them.grow 241 lakini wameniasi. but they.have.rebelled.against.me ingawa hivyo wameniasi mimi. even.though this they.have.rebelled.against.me me 3. Ngombe atambua mchunga wake, na punda cow recognizes shepherd his and donkey ngombe anajua mufugaji wake, na hata punda anajua cow knows shepherd his and even donkey knows hori la bwana wake, bali Israeli manger of master his however Israel nafasi bwana wake anapomulisha, lakini Waisraeli place master his where.feeds.him but Israelites hajui .kitu, taifa langu halifahamu cho chote. doesn’t.know thing nation my doesn’t.understand of all hawanijui mimi, hao watu wangu hawafahamu kitu. don’t.know me, these people my don’t.understand thing 5 Wapi niwapige, maana mnazidi kuasi? where I.shall.hit.you reason you.continue to.rebel Munataka kupigwa tena? Kwa nini munaendelea kuasi? You.want to.be.hit again for what you.continue to.rebel Kichwa chote kimeugua, moyo wote umezimia. head whole has.hurt heart whole has.fainted kichwa chenu kinajaa vidonda, nao moyo wenu unaumiza. head your is.full sores and.it heart your hurts 8 Binti Siyoni amebaki daughter Zion has.remained Ni muji Yerusalema tu ndio umebaki, It.is town Jerusalem only indeed has.remained kama msonge katika mizabibu. like shed in vines umebaki kama vile kibanda katika shamba la mizabibu. it.remains like that shed in garden of vines. kama kilindo kati ya shamba la matikiti, like watch.tower midst of garden of melons Unaonekana kama vile kibanda katika shamba la maboga, It.resembles like that shed in garden of vegetables 242 kama mji uliozingiwa. like town which.is.surrounded kama vile muji unaozungukuwa na waadui. like that town which.is.surrounded with enemies 10 Sikilizeni neno la Bwana, Hear word of Lord Musikilize maneno ya Bwana Mwenyezi, Listen words of Lord Supreme enyi watawala you rulers ninyi wakubwa you leaders Sodoma; Sodom wa Yerusalema, ninyi mulio waovo kama vile wale wa Sodomo of Jerusalem, you you.who.are rotten like that those of Sodom pokeeni amri ya Mungu wetu, receive command of God our Mutege masikio, musikilize mafundisho ya Mungu wetu set ears listen.to teachings of God our ninyi wakaaji wa Gomora. you dwellers of Gomorrah. ninyi mulio waasi sawa vile watu wa Gomora. you you.who.are rebels like that people of Gomorrah. 16 Jiosheni, jitakaseni: wash.yourselves make.yourselves.holy Mujisafishe, mujitakase. clean.yourselves make.yourselves.holy ondoeni uovu wa matendo yenu, usiwe tena mbele ya macho yan take.away rot of acts your don’t.be again front of eyes my Mutupe mbali na macho yangu matendo yenu maovu, throw far with eyes my acts your rotten Acheni kufanya mabaya. Quit to.do bad mwache kufanya mabaya. Quit to.do bad 243 18 Njoni, tukabishane, asema Bwana. Come, let’s.discuss says Lord Bwana Mwenyezi anasema hivi: "Mufike tukate shauri. Lord Sovereign says thus come let’s.cut counsel zambi zenu zikiwa kama nguo nyekundu, sins your being like cloth red Hata zambi zenu zikionekana wazi sawa vile rangi nyekundu sana, even sins your appearing clear like that color red very zitageuka kuwa nyeupe kama seluji; they.will.change to.be white like snow nitazisafisha ziwe nyeupe kama pamba. I.will.olean.them they.ahall.be white like cotton. zikiwa nyekundu kama damu they.heing red like blood Hata zikionekana sawa vile rangi nyekundu sana, even if.they.appear like that color red very zitageuka kuwa kama sufu. they.will.change to.be like wool nitazigeuza nyeupe kama vile ohokaa. I.will.ohange.them white like that whitewash 19 Mukikubali kutii, If.you.agree to.cbey Ikiwa munakubali na kunitii If.it.be you.agree with to.obey.me mtakula mema ya nchi. you.will.eat good of country mutakula mavuno mazuri ya inchi. you.will.eat harvests good of country 20 Bali, mkikataa, upanga utawameza . ninyi However if.you.refuse sword will.swallow.you you Lakini ikiwa munakataa, mutauawa kwa upanga. but if.it.be you.refuse you.will.be.killed by sword maana kinywa cha Bwana kimenena haya. reason mouth of Lord has.spoken these 244 Ni mimi Bwana Mwenyezi ninayesema maneno hayo. It.is me Lord Sovereign I.who.speak words these 1. The vision of Isaiah the son of Amoz, which he saw concerning Judah and Jerusalem... 2. Hear, O heavens, and give ear, 0 earth; for the Lord has spoken: "Sons have I reared and brought up, but they have rebelled against me. 3. The ox knows its owner, and the ass its master’s crib; but Israel does not know, my people does not understand." 5. Why will you be smitten, that you continue to rebel? The whole head is sick, and the whole heart faint. 8. And the daughter of Zion is left like a booth in a vineyard, like a lodge in a cucumber field, like a besieged city. 10. Hear the word of the Lord, you rulers of Sodom! Give ear to the teaching of your God, you people of Gomorrah! 16. Wash yourselves; make yourselves clean; remove the evil of your doings from before my eyes; cease to do evil. 18. "Come now, let us reason together, says the Lord: though your sins are like scarlet, they shall be as white as snow; though they are red like crimson, they shall become like wool. 19. If you are willing and obedient, you shall eat the good of the land; 20. But if you refuse and rebel, you shall be devoured by the sword; for the mouth of the Lord has spoken. (Revised Standard Version Translation) APPENDIX D INTERVIEW SCHEDULE: Language Use And Attitudes In this appendix, the interview schedule as used by my research assistant is given first. The translation follows. Italicized print indicates instructions to the interviewer. Regular print indicates main parts of the interview and questions asked to the subjects. Interviewers did practice runs before administering the interview schedules on their own and were encouraged to use their own words rather than simply read the instructions as I had written them. The recorded texts on which the first part of the interview schedule were based are given at the end of this appendix. Swahili Text Maulizo juu ya cassette .Mbele ya kuuliza maulizo 1-5, utaeleza: Nitakusikilizisha cassette kwenye watu watatu wanasema kwa kiswahili. Watasema juu ya photo walionyeshwa. Mukisha kusikia namna yao ya kusema, ninataka uniambie bitu juu ya kila mtu: mawazo yako juu ya kiswahili ' yake, anatoka wapi, kazi yake, na tabia yake. Hakuna jibu nzuri ao mbaya. Ninataka tu kujua mawazo yako juu ya hawa watu. Chasesha cassette kwa musemaji moja, kisba uliza maulizo #1-5 juu ya ule musemaji. Endelea vile kwa wasemaji wote watatu. 1. Kiswahili yake ni namna gani? Musemaji #1 #2 #3 mbaya sana mbaya kadiri mzuri mzuri sana MRNNH 2. Uyu mtu 1. anajivuna 2. katikati 3. anajinyenyekeza 245 3. 4. 5. 246 Anapokea mushahara gani (zaires kwa mwezi moja)? lo 0-19992 2. 2-3999Z 3. 4-69992 4. 7-11,999Z 5. 12,000+ Alifanya masomo ngapi? 1. Ecole primaire - 1 4- 3. Ecole secondaire 1- 4. " " 5- 5. Universite 3 6 4 6 Anatoka wapi? 1. Inje ya jimbo la Kivu 2. Kivu (lakini si Bukavu) 3. Bukavu Maulizo juu za kutumia na kusikia kiswahili 6. Huku Bukavu walimu wanatumia kiswahili mingi kwa kufundisha miaka mbili ya kwanza ya ecole primaire. Kisha wanafundisha sarufi ya kiswahili kwa miaka ine ya kufuata. Unawaza ni mzuri vile? Au unawaza ingekuwa mzuri zaidi kutumia na kufundisha kifransa tu tangu mwaka wa kwanza? Au unawaza ni mzuri kutumia na kufundisha kiswahili zaidi? 1. Endelea kama sasa 2. Kifransa tangu mwanzo 3. Kiswahili zaidi Unawaza ni lazima kwa mtu kujua kiswahili bora? 1. Hapana, hata kidogo 2. Si lazima, lakini ni mzuri 3. Lazima kabisa Unaweza kusikia kiswahili bila shida-- Kidogo Badiri Ming; Yote a. pamoja na marafiki b. ku kazi c. ku radio d. mu kanisa 247 9. Unaweza kutumia kiswahili kwa gisi unafurahia--kwa kusema kitu chochote na wakati yoyote munapenda kukitumia? 1. mara mingi, hapana 2. mara mingi, ndiyo 3. kila mara, ndiyo Maulizo juu za maisha 10. 11. Uko na ndugu mu Bukavu? a. James ngapi? b. Wanaishi wapi? Unafanya kazi na watu wengine (wawili au zaidi) wa mtaa/zone.yako? 12. 13. moja? 14. a. Biko banaume au banawake (ao bote bawili)? b. Unaona hawa watu kisha kazi kwa kuzungumza na kujifurahisha nao? 1. Hapana 2. 1-2X/mwezi 3. 1-2X/posho 4. 3X au zaidi/posho Unaenda ku kanisa? 1. chini ya 1X/mwezi 2. 1-2X/mwezi 3. kila Siku ya Mungu 4. zaidi a. Wanatumia luga gani? b. Unafanya kazi kwa kanisa? Kazi gani? Unasafiri inje ya Bukavu mars ngapi kwa mwaka/mwezi a. Wapi? b. Kufanya nini? 1. Kuona wandugu 2. Commerce 3. Unasema luga gani na: a. wazazi b. watoto c. rafiki d. watu wenye unatumika nao 248 English Translation of Interview Schedule Questions concerning the cassette Before asking questions #1-5, you will explain: I will play you a cassette on which three people speak Swahili. They will speak about a photo that they are shown. After hearing their way of speaking, I want you to tell me things about each person: your opinions about his Swahili, his education, where he comes from, his work, and his personality. There is not a goor or bad answer. I just want to know your opinions concerning these people. Play the cassette far one speaker, then ask questions #1-5 concerning that speaker. Continue like that for the three speakers. 1. How is his Swahili? Speaker #1 #2 #3 1. very bad 2. bad 3. in between 4. good 5. very good 2. This person: 1. is arrogant 2. in between 3. is humble, pleasing 3. He receives what salary? 1. SO-20/month 2. $21-40/month 3. $41-70/month 4. 971-120/month 5. $121+Imonth . 4. How much schooling does he have? 1. 1-3 years of primary school 2. 4-6 " 3. 1-4 years of secondary school 4 . 5-6 N 0' 9' N 5. University 249 5. Where is he from? 1. Outside of the Kivu province 2. Kivu (but not Bukavu) 3. Bukavu Questions about use and understanding of Swahili 6. Here in Bukavu, teachers use much Swahili to teach in the first two years of primary school. Then they teach Swahili grammar for the next four years. Do you think this is good? Or do you think it would be better to use and to teach only French from the first year on? Or do you think that it is good to use and to teach Swahili more? 1. Continue as is done now 2. French from the beginning 3. More Swahili 7. Do you think it is necessary to know Standard Swahili? 1. No, not at all 2. It is not necessary but it is good 3. It is very important 8. Can you understand Swahili without problem-- > 1...: y... A little Fair amount Much a. with friends b. at work c. on the radio d. in church 9. Can you use Swahili in the way you please--to say anything, whenever you want to use it? 1. Many times, no 2. Many times, yes 3. Always, yes Questions about living situation 10. Do you have relatives in Bukavu? How many families? Where do they live? 250 11. Do you work with other people (2 or more) from your neighborhood? Are they men or women (or both)? Do you see them after work for talk and amusement? 1. No 2. 1-2X/month 3. 1-2X/week 4. 3X or more/week 12. Do you go to church? 1. Less than once a month 2. 1-2X/month 3. Each Sunday 4. More often What language do they use? Do you do church work? What kind? 13. How often do you travel outside of Bukavu per . month/year? Where? To do what? a. to see relatives b. for business I 14. What language do you speak with: a. your parents b. children c. rafiki d. people you work with Texts That Subjects Evaluated 1. Vernacular with little French Hii picha ya kwanza unaona hapa ni tribinali. Iko ku La Botte. Kama batu biko na mashitaki banaweza kukuya pale. Picha ya pili ni fasi kwenyi bakubwa ba kanisa ya Katoliki banafanyiaka kazi yao. Nyuma ya pale kuko Catedral. Pale kunakuwaka misa. Kila Siku ya Mungu kuko misa tatu. Siku zingine kunakuwaka misa moya ya asubui. Picha ya tatu ni ya balobaji benyi biko banaloba samaki mu bahari ya Kivu. Biko mu mitumbu ya miti. Siyui vizuri hii fasi kwenyi kuwa picha ya ine. Nazani ni Lycee, pale yulu ngambo ya Kadutu. 251’ Picha ya tano ni kiwanja ya kabumbu. Batu banachezaka kabumbu pale kila Siku ya Mungu. "The first picture you see here is the Court House. It’s at La Botte. If people have legal problems they can go there. . "The second picture is the place where Catholic church leaders do their work. Behind there is the cathedral. There are masses there. Each Sunday, there are three masses. 0n the other days, there is a mass in the morning. "The third picture is of fishermen fishing on Lake Kivu. They are in wooden boats. "I don’t know very well the place where this fourth picture is. I think it’s the Lycee, there in the high part of Kadutu. "The fifth picture is a soccer field. People play soccer there every Sunday." 2. Near-Standard Hizi ni picha za Bukavu. Ya kwanza ni baraza ya sheria. Watu wenyi kuwa na ugomvi kati yao watakuja pale kwa kutafuta haki. Ya pili ni ya nyumba ya wakuu wa kanisa la Katoliki. Jumba hili lililo nyuma yake ni kanisa la Wakatoliki. Watu wengi huja hapa Siku ya Mungu. Ya tatu ni ya wavuvi wanaovua samaki kwa ziwa la Kivu. Wamo katika vyombo vyao vilivyojengwa kwa miti. Picha ya ine in ya Lycee Wima. Ni shule kwa vijana wanawake. Ipo kwa kilima kimoja juu ya barabara ya kuenda Kadutu. Kuna wabikira wanaofundisha pale. Picha ya tano ni ya nyumba za Kadutu tu. "These are pictures of Bukavu. The first is of the Court House. People with disputes between them go there to look for justice. "The second is of the house for leaders of the Catholic Church. The large building behind it is the church of the Catholics. Many people come here on Sundays. "The third is of fishermen fishing on Lake Kivu. They are in their boats which are made of wood. "The fourth picture is of Lycee Wima. It’s a school for girls. It's on a hill above the road going to Kadutu. There are nuns who teach there. "The fifth picture is just of houses in Kadutu. 252 Vernacular speaker using much French Kwanza, hapa ni photo ya tribunale. Pale kunakuwaka baavocats na bajuges. Kama mutu fulani anakutesa, utaenda kwa avocatna kupima kupata justice. Photo ya pili ni mabureau ya archeveche na batu bingine ba kanisa. Kisha pale yulu unaona Catedrale. Photo ya tatu ni ya bapecheurs mu mapirogues yao. Biko banatengeneza mafillets yao. Photo ya ine ni ya Lycee Wima--ecole secondaire pour les filles. Photo ya tano ni ya Stade, fasi kwa kucheza futbol. Kila dimanche, des milliers des personnes banafika pale kwa kuangalia match. "First, this is a picture of the Court House. There are lawyers and judges there. If someone bothers you, you go to a lawyer and try to get justice. "The second picture is the offices of the archbishOp and other people of the church. Then there above you see the cathedral. "The third picture is of fishermen in their canoes. They are fixing their nets. "The fourth picture is of Lycee Wima--secondary school for girls. "The fifth picture is of the stadium, a place for playing soccer. Each Sunday, thousands of people go there to watch a game." APPENDIX E Interview Schedule Scores VARIABLE LABELS: 8 'sex' E 'educational level' ed 'education in years' a 'question #1 of inter- view schedule b '#2' c '#3' d '#4' e '#5' f '#6' g '#7' h '#8a' i '#8b' j '#8c'. R '8d" 1 '#9 m 'network strength based on answers to ##10-11 n '#12' o '#13' p 'l4a' q '#14b' r '#l4c' s '#l4d' VALUE LABELS: g: f 'female' m 'male'; B: 0 '0 years of school' 1 '1-3 years' 2 '4-6 years' 3 '7-9 years' 4 '10-12 years' 5 '13-15 years' 6 '16+ years'; 22: number represents maximum number of years achieved; 2;}: values correspond with numbers of response choices as listed in interview schedule; 2: 0 'no affirmatives to questions in 10-11 of interview schedule' 1 'affirmative to #10' 2 'affirmative to ##10,11 3 'affirmative to;##10,11, co-workers are of same sex (11a)' 4 'affirmative to ##10,11, co-workers are of same sex, and they meet together at least once a week'; 2: values correspond with numbers of response choices as listed in interview schedule; 0: e 'English' f 'French' 1 'Lingala' m 'ethnic language' 3 'Swahili' 253 NIHNNIIINIII III lull alt-ell» IIIIIIIIHI achansonsneonates-Ntea-sconea“waaauwuwwwwouwwuwnwwwnwwwwumnwuwuwnu I“ 344 342 543 352 243 453 243 342 513 342 342 352 345 253 453 254 244 344 235 343 352 354 4?? 354 353 243 452 352 351 344 453 453 455 352 142 454 453 253 443 344 453 b 331 231 232 321 321 321 321 321 321 321 331 331 321 232 331 321 231 221 313 332 321 321 321 321 331 321 231 321 321 321 132 312 331 331 331 231 331 311 322 231 211 231 221 231 322 332 332 322 321 331 321 211 312 321 321 321 112 223 123 223 223 122 122 234 123 223 234 232 122 132' 122 122 231 522 125 123 223 142 133 120 123 124 123 122 122 222 224 124 1 3 223 121 ??? 112 452 212 122 243 214 123 123 335 234 224 143 123 334 322 344 345 441 314 334 234 245 244 235 235 234 234 134 345 243 234 045 234 245 454 024 2?? 233 245 445 234 344 355 334 2.45 245 345 345 143 234 225 234 445 331 445 “NNWHNWNNNNNPNWI‘NN NQNNNNNNHNNNN I NNNNNNNHNNNNNNNNNIQNNNNNIf K 254 NNNNWNNNNNNQNNNQIIW NQNNNNQQWNN “WNNNNNNCOF‘IJJNNNNQNNNN wnwaaaawwhawaaacoaooaanewabhwaacopanacea-.5ohawanbwwawaaoowwaoowwhII 5' WNNQNN on I wwanaaaco l auwahwwacowwaauwwwwaooawwu~— fiNNwPNMNNNNNNNwl-‘NN I hpuwwnanwwuwmwwwuwwnwwnwwwnnnwwwupwpnus— hub““WebbbhwwwwhéhhwaQOWNthflhéhbbwbw-bNubvbWabwhHNhhfiWNWNNQUbllW $.5th WIWN oawcoacoaaoo wwnwwwwnwnwunwwwwnwmucous:wwwwooconcow”mwuwwwuwwwwwwwwwwwwwwconI— bananas-aslacshahahauabahwhaasahahhwabIII IbubthubH N NW Awwawaroaooaowprbwwwcowbnwmwwawwaoowwcocwawwpb “NQNOQWQQWQQWII 5 IbeubH I ah “N09 ab3 3a a1 0 b52 mmmgmmmmmmmumm'mm I! mm;- (on SH .8 SI I8 .3 mmummmmmmmummmmmmmmmammum'ummmmmmmmmmmmmmmmmmmmmummuuhnu«.0 In N I! mmmmmmmmmmmmmmmmmmmns mmum p—a fse sfl MHNHMMMMMM h I F5 ‘3! is Is is is 80 III-NIIIIIIIWINIIIIIIIIIIIIIIIIIIIIIINWIIIIIIlla OOOOOOOONOOPOOOhNNN“ONObwGNwhbl-‘abwl‘lbahhfl(”0016.501 an 321 231 332 333 312 222 321 321 331 321 312 231 332 321 332 332 322 223 223 134 122 212 242 223 223 324 223 212 233 112 234 321 435 113 234 245 135 245 355 '35 233 334 345 235 354 234 234 235 33 325 244 223 354 345 254 245 344 245 335 345 344 245 345 123 234 443 234 234 455 345 334 345 234 345 234 435 345 "O Mlll N l-‘ll 231 233 322 233 233 332 333 323 331 321 321 233 334 223 333 232 331 331 321 231 231 321 323 321 321 321 321 321 331 233 321 321 321 321 332 321 321 312 312 322 311 321 “(0090060009909NNNNNPQHNQNQHNQNNWQWHPNwwwwNQNQNNWNN II” 255 wwwwwwwwwnwwwuwnuwwwwwwawwwwuwwwuwwwuwwwnunw ll“ wwwwwwwwuoaaapwwurchwwawbwaawapawwwbwaabuaaaw "5' O) l newwwwwaawawwwoowwoanasuaauabwhaa l ““450“ II.- N“ l‘HNlO-bNNNQNQNNHH“WNQNwwthNbbNQWQWIbD-fi“NubNh IIG-n- NM NNWWAOQQWQIDQWQWOIDthfi“bhhfibhwibthabl-‘bhahbwb "3' coco P“MN“NNNNNNNNNH“WM“NNNGQNHNNNGRNNQNQ‘QN“NH ll!" NH bhubthubtbubhnbh oom-cooooa-ooo-bawounwaaahaaauanaaaan ll- Nab oouuwwwhaoouuancowwahw ll: nuawwaoauwwwuaaaohaun “N a1 a1 a2 1a is al 3a 2a 3a a2 a2 a1 3a a3 3a a1 83. 1a 33 a1 . gummmangmulwun he a «mawazo. n an N at fsm at at Appendix F Individual (j) Scores for Each Environment Arranged Accordng to Age Groups' Implicational Scales For Young and Middle-Aged groups' scores: Column # Environment 1 -ja1a 'be full' 2 -yala 3 -ja- 'not yet' 4 -ya- 5 . -kuja 'come' 6 -kuya 7 moja 'one' 8 moya 9 mos 10 ngoja 'wait' 11 ngoya 12 ngoa 13 majani 'grass' 14 mayani 15 maji 'water' 16 mayi 17 ‘ -jua 'know' 18 -yua 19 -jenga 'build 20 -yenga 21 kijana 'youth' 22 kiyana 23 fujo 'noise; probem' 24 fuyo 25 stressed -ja- 26 stressed -ya- 27 -ja- before stress 28 -ya- before stress 29 -ji- after stress (excluding ##15, 16) 30 -yi- after stress 31 stressed -ji- 32 stressed -yi- 33 (-)ji- before stress 34 (-)yi- before stress 256 257 lOllG All IIIDLI'AGE GROUPS' 1'! IIPLICAIIOILL 8611! N 0‘ H 0. U ‘0 N) n a— J p h at Al - l; I“? "1 H l 1 ’5‘ 1' .. ‘3 3‘ Ill! 11 21 l 1 1a ya 1| to so I1 I1 I a a 1 y aa ea 1| 1| 1: ye 1 ll 13) s s l y 11 y 11 y 11 y 11111 11 11 11 11 11 11 11 21 11 1 1 1 1 1 11 11 11 11 2 1 1 1 13 1 11 1 bet 1 11 11 11 11 11 11 11 11 11 11 11 1 11 1 1 '11 1 11 1 salt 12 11 11 13 11 11 11 11 11 1 1 11 11 11 11 2 1 11 1 11 1 c111 12 11 1 1 11 12 11 31 11 11 11 1 2 2 11 11 11 11 1 3 1 1 12 1 11 1 12 1 111 a 13 11 1 2 11 11 11 11 11 21 11 1 2 11 11 11 11 1 1 1 1 11 1 11 1 saga! 11 11 1 1 11 11 oo 11 11 23 11 1 1 1 1 1 11 11 11 11 1 11 1 1 1 1 11 1 12 1 11 1 I'lll 11 11 1 1 11 11 11 21 11 12 11 1 1 1 11 11 11 13 11 11 1 2 1 1 11 1 22 1 11 1 slla 11 11 11 12 11 31 11 11 11 1 2 1 11 11 11 11 1 11 1 1 2 1 11 1 11 1 lea b 11 11 1 1 11 11 11 31 13 22 11 1 1 1 11 11 11 21 1 3 1 1 1 1 1 1 13 1 12 1 pap: 11 11 11 11 11 21 11 21 11 1 2 1 11 11 11 11 1 11 1 1 11 1 21 1 1010 11 11 1 1 11 11 11 11 11 11 11 1 1 1 11 11 11 11 3 2 1 1 11 1 13 1 11 1 asloa 11 11 1 1 11 12 11 11 2f 13 1 2 1 1 1 3 11 11 11 21 12 11 1 11 1 11 1 11 1 last: 11 11 11 11 11 21 11 32 13 11 21 1 1 1 1 13 1 11 1 les 3 11 11 1 1 11 11 11 31 11 21 21 1 1 1 11 11 21 11 1 1 3 1 11 1 11 1 Iasla 11 11 1 2 11 12 11 31 11 11 11 1 1 2 11 11 11 11 l 1 1 1 13 1 12 1 103:. 11 11 11 11 11 13 11 1 2 11 11 12 11 11 11 1 11 12 1 11 1 litat 11 11~ 11 11 11 22 11 11 11 11 11 11 13 11 11 2 11 1 1 3 1 11 1 11 1 22 1 vlclo 11 11 11 12 11 12 11 11 11 12 11 11 11 1 11 12 1 11 1 loyol 11 11 11 12 11 11 11 11 11 11 11 1 11 1 1 1 1 2 1 11 1 11 1 claal 11 11 1 1 11 12 11 21 11 31 11 1 1 1 1 1 13 11 13 11 11 11 2 12 1 1 2 1 1 1 13 1 iaerl 11 11 1 2 11 11 11 23 11 11 21 1 1 1 11 11 11 11 12 11 1 2 1 1 2 1 12 1 11 1 I1: I 11 11 11 12 11 12 11 21 21 1 2 1 11 11 11 11 11 11 2 1 1 1 11 1 11 1 0111 11 11 1 1 11 12 11 11 11 11 11 1 3 11 11 2 1 1 1 13 1 13 1 expo: 11 11 11 . 11 11 11 1 3 1 11 11 11 13 2 11 1 1 11 1 1:11: 11 11 11 11 11 11 11 11 12 ’11 11 3 11 1 1 11 1 ea: 1 11 12 11 13 11 11 11 21 11 1 1 1 1 1 12 11 11 11 12 11 1 1 3 1 11 1 21 1 layll 11 12 1 1 11 11 11 13 11 13 1 1 1 1 11 31 1 1 12 1 11 1 °1anll 11 11 11 11 11 11 12 11 11 1 2 1 11 11 11 11 3 11 2 1 1 1 11 1 ' late 11 11 11 12 11 12 11 23 11 1 1 1 1 2 11 11 11 11 1 13 3 1 13 1 11 1 ea: 1 11 11 1 1 11 12 11 11 11 11 12 1 1 1 1 1 11 11 11 11 11 11 2 1 1 1 11 1 22 1 11 1 10:1: 11 13 11 11 11 21 11 11 11 1 1 11 11 11 11 1 1 2 1 12 1 11 1 3:131 21 11 11 11 11 11 11 11 11 1 1 1 11 11 11 21 1 11 1 2 11 1 11 1 31111 21 12 11 11 11 33 11 11 32 11 21 2 11 3 1 13 1 alaia 21 11 11 11 11 21 11 11 11 1 2 1 1 1 11 11 13 11 11 11 1 1 2 1 2 1 13 1 11 1 op! 21 12 1 3 12 12 11 21 13 21 11 1 1 1 1 1 13 1 11 11 11 13 2 11 1 1 3 1 13 1 12 1 31 1 11:11 21 11 11 11 11 32 11 31 21 11 1 1 11 1 23 1 allal 21 11 11 13 11 11 11 11 11 1 1 1 11 11 1 1 13 1 13 1 131:. 22 12 11 11 11 11 33 11 11 11 11 11 11 1 1 1 11 1 11 1 11 1 lelea 23 11 1 1 11 11 11 21 11 11 11 1 3 3 11 22 1 11 2 1 1 1 11 1 12 1 11 1 laait 23 11 1 1 11 11 11 11 11 11 11 1 3 1 11 11 11 32 11 12 1 11 2 1 11 1 11 1 11 1 tilaa 23 11 11 11 11 11 11 11 11 1 1 1 1 1 11 13 13 11 1 11 1 1 1 1 11 1 11 1 hallo 21 11 1 1 11 11 11 23 11 21 12 1 2 1 1 1 13 11 12 11 1 12 1 1 2 1 11 1 11 1 11 1 111a 21 11 1 1 11 11 11 21 11 11 11 1 1 11 31 1 2 1 1 1 1 2 1 1 1 13 1 33 1 11111 21 11 11 11 11 11 11 13 13 2 1 1 11 11 11 11 1 11 2 1 3 1 11 1 11 1 11 1 1111: 21 11 11 11 11 11 11 11 11 11 11 11 11 12 11 1 11 2 1 1 1 1 1 21 1 11 1 11 1 even: 21 11 1 1 11 11 11 21 11 13 11 1 1 1 11 13 12 23 11 11 1 11 2 1 11 1 11 1 11 1 halal 21 11 11 11 21 11 11 21 11 1 2 1 11 11 11 11 2 11 1 1 3 1 1 1 12 1 11 1 nasal 21 11 1 2 11 11 11 23 11 11 11 1 1 1 11 12 11 11 11 11 2 11 1 1 1 1 1 1 11 1 13 1 11 1 258 1112 11 £1 1 y la ya 1| 1| I1 I1 I1 I a a 1 y la a: 1| 1| 1: y: t 11 1 y s s 1 y 11 y 11 y 11 y laaba 21 11 1 1 11 11 11 21 11 21 11 1 2 1 1 1 11 11 12 21 11 12 1 11 2 1 11 1 11 1 1f 1 salon 21 12 1 1 11 11 11 11 11 21 12 1 1 23 11 1 2 1 1 1 12 1 11 1 aatll 21 11 11 12 11 11 11 1 1 1 33 12 11 11 1 11 1 2 1 1 1 13 1 21 1 11a 1 21 11 11 11 11 21 11 11 11 1 1 1 11 11 11 21 11 11 3 2 2 1 1 1 11 1 11 1 1:1:- 21 11 11 11 11 21 11 22 11 11 11 13 33 11 11 1 11 1 1 2 1 11 1 21 1 ISSII 21 11 1 1 11 21 11 22 11 11 11 1 1 1 11 11 11 11 11 11 2 1 2 1 13 1 12 1 11 1 raaa: 21 11 11 12 11 13 12 11 12 11 11 11 11 1 1 2 1 12 1 11 1 21:11 21 12 1 1 11 11 11 12 11 11 12 1 2 1 11 11 13 11 1 1 1 1 21 1 21111 21 11 1 1 11 11 11 11 11 11 11 2 1 1 2 2 11 11 11 21 1 2 1 1 1 1 13 1 11 1 11 1 tarot 31 11 11 11 11 11 11 11 11 1 1 1 1 1 11 11 11 21 1 11 1 1 1 1 11 1 13 1 22 1 11:11 11 11 11 11 11 13 11 22 11 1 2 1 11 13 11 11 11 11 1 11 2 1 1 1 12 1 13 1 11 1 like: 31 11 1 2 11 11 11 11 11 21 11 2 1 11 31 1 1 1 1 1 1 12 1 11 1 1 111 31 11 11 11 11 31 11 11 11 1 3 1 11 12 2 1 12 1 22 1 aaall 31 11 1 1 11 11 11 21 12 12 11 1 1 1 11 11 11 1 1 1 1 1 2 1 13 1 11 1 11 1 11:10 11 11 11 11 12 11 12 11 11 1 2 11 11 23 11 11 12 1 11 2 1 13 1 11 1 11 1 aarce 31 11 11 11 11 11 11 13 11 1 2 1 11 11 11 11 1 3 2 1 12 1 21 1 last: 12 13 12 11 11 11 11 11 11 2 1 1 11 11 11 11 1 11 1 1 11 1 11 1 11 1 Intel 11 11 12 11 11 12 11 11 11 1 1 1 11 11 11 11 1 11 1 2 1 1 11 1 11 1 11 1 12 33 11 1 2 11 12 11 12 11 11 12 1 2 1 11 11 11 11 12 1 12 1 21 1 trail 13 11 11 11 11 11 11 11 11 1 2 1 11 13 12 11 1 1 13 1 11 1 reglu 33 12 11 11 11 31 11 11 13 1 1 1 11 11 11 21 1 1 3 1 12 1 11 1 aalaa 31 11 11 13 11 11 21 12 11 1 1 1 13 11 23 11 1 11 1 1 1 1 12 1 21 1 21 1 pill! 31 11 1 1 11 11 11 11 11 31 11 1 2 1 1 1 11 12 12 11 13 11 1 11 1 1 1 1 11 1 12 1 lllOl 11 11 11 11 11 21 12 21 11 11 12 21 11 11 11 1 11 1 2 1 1 11 1 11 1 11 1 259 For Old-Age group's scores: Column # Scores are the question. Environment -ja- 'not yet’ -kuja 'come' -kuya maji 'water' mayi ngoja 'wait' ngoya ngoa majani 'grass' mayani moja 'one' moya moa -jenga 'build' -yenga -jua 'know' -yua, kijana 'youth' kiyana fujo 'noise; problem' fuyo stressed -ja- stressed -ya- (-)ja- before stress (-)ya- before stress -ji after stress -yi after stress stressed -ji- stressed -y1- (-)ji- before stress (-)yi- before stress total occurrences of the variant in 260 1/2 IIVIIONIEIIS ABILIGED 20 011-16! GROUP IIPLICATIOIIL SCALE 1. 5. ‘ 391101113131Iguo 13552129142722. «)0 1G 11 1I 11 1| 1| II II I I I 1 11 II II II 11 yo 1I15 1 $12 11 y 11 y 11 , 11 11 12 12 13 11 11 11 11 11 11 11 11 11 11 11 1 1 1 1 11 1 11 1 21 1 11 11 11 11 11 21 11 11 1 1 1 1 11 11 21 12 11 13 11 21 2 1 12 1 11 1 11 1 11 11 11 11 11 11 11 11 1 1 1 2 1 11 21 11 11 11 12 11 1 1 11 1 11 1 11 12 11 12 11 11 1 11 11 11 11 1 1 11 1 11 11 11 11 11 11 12 11 1 11 11 11 11 12 1 11 11 1 1 1 11 1 11 1 13 1 12 11 11 11 11 11 11 11 1 11 11 12 11 11 11 11 21 2 1 12 1 11 1 11 1 11 11 11 11 11 11 11 11 11 1 1 11 1 11 1 11 11 11 12 12 11 1 1 1 1 1 11 11 11 11 11 11 11 1 1 11 1 11 1 12 11 11 11 11 11 1 1 1 1 11 11 13 11 11 11 11 11 1 1 2 1 11 1 11 1 11 1 12 11 11 11 11 11 12 11 1 1 1 21 11 12 12 12 11 11 1 1 1 1 1 1 11 1 11 1 21 1 11 11 11 11 11 21 11 11 1 13 11 11 11 11 11 12 11 2 1 12 1 11 1 11 1 11 11 11 11 11 11 11 11 3 12 11 31 11 11 11 12 23 2 1 11 1 11 1 11 1 11 11 11 11 11 21 11 13 1 12 11 11 11 13 12 12 33 2 1 1 1 1 1 13 1 12 1 11 1 11 11 11 21 11 11 1 2 1 11 11 21 11 11 11 11 11 2 1 2 1 3 1 11 1 11 1 11 1 11 11 11 11 11 11 1 1 1 11 11 11 11 11 11 11 1 1' 11 1 12 11 11 12 11 11 1 1 1 11 11 11 11 11 2 1 1 1 21 1 11 1 12 1 12 11 11 11 11 11 1 11 21 11 11 11 23 13 3 1 1 1 11 1 11 1 11 11 11 11 11 11 11 11 1 1 1 2 11 13 11 11 31 11 12 11 1 11 1 11 11 11 12 11 21 11 11 2 11 11 11 11 11 12 11 1 1 11 1 11 1 11 1 11 11 11 11 11 11 11 11 1 3 1 1 11 31 11 11 11 11 11 1 1 1 1 11 1 11 1 11 12 11 11 11 11 11 11 1 2 1 2 11 11 11 11 11 11 11 1 1 11 1 11 11 1 3 12 11 1 2 1 1 11 11 11 11 11 21 11 1 1 11 1 11 1 11 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 11 11 11 11 11 15 11 12 1 15 11 25 10 11 11 02 2! 11 11 11 15 11 15 12 11 2 12 11 22 11 01 11 11 35 1 1 11 1 13 12 11 11 12 11 20 11 11 2 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 1 3 11 1 1 11 12 10 11 13 01 11 1 1 1 1 11 1 21 1 15 11 11 12 11 21 11 12 11 1 1 11 21 11 01 11 11 11 1 1 —~ 1 11 11 13 0 1 11 22 2 1 12 12 11 11 11 11 11 1 1 11 11 10 11 12 11 15 21 1 11 11 11 21 11 11 1 2 1 1 1 11 11 11 11 11 11 0 1 2 1 11 21 10 11 20 11 12 1 5 15 1 11 11 11 15 01 21 11 11 1 1 1 1 10 12 11 02 11 11 15 l 1 1 2 12 1 11 1 21 11 11 11 1 11 11 11 0 1 1 2 11 11 11 1 1 01 11 3 1 1 1 12 1 12 1 11 1 12 11 11 11 11 11 10 11 1 5 1 2 11 11 13 11 11 11 11 2 1 3 1 1 2 11 1 11 1 11 2 15 11 11 18 1 1 1 2 11 11 02 11 11 11 11 12 1 11 1 1 1 0 12 1 11 1 261 UN CO I CO 0 GN NG NG NN N» Nm Nm Nm No NG NG >uvanNx o NsnpwN>urmm. N .mccuoon. N .waun eons. u . » . mutu + zw\0N. m .llrcxac. m . awcmc. u . maana: soawmwmn. savanna. m e . draw. NG . .ooacvwnwos. NN .lluotnuutw**.llNN HoNNmNcs. N» .III> + 2w. le .IlwtnNN. naNonNosawpv. Na .Ilnaaoaannnnwcoxoc> N H I I O 0 1'0 0'8 '0 02 cu '0 02 W9 '0 69 '1 '0 69 W0 '0 69 9“ '0 69 99 '0 69 “Q 1'0 1 8 2'0 1 8'0 1'0 1'0 2'1 0'2 0'11 0'9 1'0 2 8'0 8'1 9'0 era yao I-I 2'1 1'1 1 8'0 1'0 9'0 2'0 2'1 1'0 1'0 2'1 0'2 0'1 0'1 0'6 1 8 1'1 9'0 1'0 1'0 0'1 0'1 0'2 00 OH O 3?? 3?? 335 578 5?? N+V an S1 53fi mog emu] man man 8133 12 an; “In 01 11 21 91 61 81 21 02 22 82 Appendix H Individual (ni) Scores for Combined Environments Variable i "interviewer" b ed 1a 1b 2a 2b 3a 3b sex 'born’ 'education' 'occupation' 'ethnic group' 'Necessity of Swahili 'network strength' 'non-nominal 232 'non-nominal 353' 'Simple NP 3;; 'Simple NP 252' 'Complex NP Bi 'Complex NP :52' Values 0 'Zairean only' 1 'Zairean + me' 2 'me only' 0 'male' 1 'female' 'before 1950' 2 '1950-1968' '1969-1976' wH 'none' 1 '1-3 years' 1 '4—6 years' '7-9 years' 4 '10-12 years' '13-15 years' 6 '16+ years' UIUDO 'never full-time paid employment' 'manual laborer' 2 'teacher' 'skilled laborer' 4 'administrator' 'preacher' UIUJI-‘O 1 'Shi' 2 'Rega' 3 'Other' 1 'not necessary at all' 2 'not necessary but good' 3 'very necessary' 0-5 - weakest to strongest Numbers . total occurrences in text II I! I! II II II 266 2W COIBINED ENVIRONMENTS NI/'K0: 0 IA I) lailh 3a 3b 1 s b 040 e S N k no no no no co co 12 2 15 01 10 2 81 51 3 423332333 ‘3 34 2 453.31.55.05 3 233333323 3331331131332112113 312311 1 1110 2234 111 060 22 1135 0102 111111 1 12 030141400000 4 091 3 o 8 09 1 111 o 2 107288392512 143 181 171 1 3‘. 1110100 31011-01300 0 30 0111013 ‘010012143 2 1. 1 33823711923071931141353 1 00331014921403312355111 5 55 3 33 01 19 1 32 00 3 444 0‘. 3 I. 1.2 3 41.5 4‘. 5 51.55 3 233 33 2 3323 00300532001100102 0 o 20 55 66 996 11381241100217.3112 21 101 131 3211 00 230 012 6036 1 13 8.3910917691130191 1 32181831242501023 333 I. 34.1. 3 552 5 355 5 253 33 3 114 232 2 232 35 52 111221132312233112111221332 1101111 11141111111112221131311313144111110111311 20000000000000001151122222223433344424403188004122222333 11111111111111111111111111111111111122222222222222222222 0000001001000010000000000000000000°00001101100001101010,0 00000100000101000120000001011200011010100002010000010000 268 0 no no no CO CO 8 a: a: == == 3: kn N: S: e: 03 08 b8 83 18 .0119n£4f4Qu2n2971Qu1n00:6060a1 Dalnvonvfl:0110n0110 izunvonvflflnvonUO-OAUI:UnVO¢1 2‘10 2 957417. 1. 60662§=v1 48a39- 0. 8 1 .8417119o1068a14:3Q:2Q03.150I3I1121118.1929062¢11:2nv2 4:19.4.59:8Ru5n05:3114901000 .2050112u2oLOLU 4 1:2002‘1 .0 92241 0 .1 onvouznuiflnvoaLO. 110 1nunv 2 nv1.1110050139- 11 .DQuZAVO. 9a 900‘“ 4 AU 1.1nv3.1150.19-O;U 041 nv1ni 0 4 3 44 5 3 11 2 27152811377811227251811775210419421153371513051835526 4 10 1 3 34850261513112714012558090311002101000400202261230001 434 3 3 23444343 4 33 323 4333 33 3 3 4342 32 15 5 2 554255 5 25 434 5555555 5 4 5553 4 23 2 2 2333233 2 32 232 2332222 1 2232 32 3902412Qa19¢1&01413902901901413902903411 2062062901¢1141341290141241390141 a3 1111 2412 1nv1.12062060n01:2993060:§Avonv0nv0n0010120nvoAUOAVOAUO;UnvonVOAUOAU Qv3nJQu4‘Ilrinllf‘nilr4151:4115‘9415:0Rv312nv1n69-2069-2n6923n6922ndfig3oafiu3.1%0313‘1403 952n£932062;29o2n2922069f2n29u206922n6922oJQu3n3Qu313Qu3n32u3139u313223n3Qu3n3Qu3n3Qu3oJ 111111nvo410nv1nv0nv0nVOAv0410zu110nvonun. 111Av0:0110¢11:0110411410410ZU£11110nv1 nvoAVOAUOAVOAVOAVOnYo060411nvonVo;10;Unv0110nv0nv0:09:005OAU0nV0nV0nU0:1nv0nV0411 Appendix I “n- 0.— 2-6-6 yra; 3-7-9 ’o-old-ece'group; OCCUPATION “2151. . g?! :coree according to occupation. c-above averege church ettendance: c urch attendance not known; E-Educatlonal level (0‘0 lel 1'1'3 7‘83.“ 1": 4'10-12 7:03 5-13+ yrs); Ia-Intervtewad by Zairean + nyaelfs Ib-Intervleued by ayaelf only; fl-nlddle-age IIOUPI Y-young-aae group. Plot of £323; Scores : --._”ruah-lm-m-“_ “ .u __u- . ‘A - - PLOT OF V7 81TH VS _.- ...... ‘0‘. . ' ' Z ' ' ' ° ‘ :' : ° ' .f?-.. x ‘ 0'3 ......._.E in,“ .3 ~ 1 1n.t; '— >13133° 1 o. 21.... - ._-._._.1_I-I. e. lb..- , ‘ I - ._._.___ \ x ’_-' t 1 m.. t — 0. .—~ - _ a. - ‘— _vlfi 73o ' I O t”“‘”“’ _ _ Y I “""—’T“’ I ! “ 6205; I"m.1°;tj~ ~- .9 -‘ u x ' O ..-“1...-- -urw --1 __. xH.1a I v ' 0.13 1 ' E 509 1 I - . -- .. __--. -, H C? I_.- .-.. I ...__... --- ..-_....._...- ...—.. .--- . a : 1u.c7 “' _ I ’3! c 1 1 10.;2 c - -_ .. .. . __ -.-..__.. _...__._-._i_ ‘ U D 310-; 15.0 . '- -——- mg- -.- __-o— n. t . . - _.,J-- i 10.22 I _ . -- ...-_ -.._ -- u.c:.__-._ -.__.--.._ - Y 2:; gs. u.c-. ' ' '° " I 0.22; n.c I -1 _. -__-_____ 1m“; _ w” 1 .1 I1 1 1'13. '1. 0"“ 1 12.5. 1 3' ' ..I ,1. _ 10.25.11: ‘ ' 1 t -- I“. I . 0.30 3 I --m‘.--_.__-.-._...,__ ..--. .3 -- o. 12 1 a t 00“ ‘ 3 A ‘ ‘ ‘ ‘ . O -- .v v v v 7 _¢ ¢ -¢-.----.— 3 3 ¢ 0--.. - Haaua1 Teacher ' Skilled Adnlniatrator Prbacher " Laborer Laborer ' 269 Appendix J Independent Variables and Phonological and Lexical Scores VARIABLE LABELS: l "interviewer" l "sex" 2 "born" 4 "education" 2 "Occupation" g "ethnic group" "necessity of Swahili" 8 "network strength" "church attendance l2 "wa + verb" ll wa + noun 2 "moja" lg "jua" lg "French frequency" 5 "lakini v. mais" Ihflwhfl H VALUE LABELS: l: O 'Zairean only 1 'Zairean + me' 2 'Me only'; l: 0 'male'. 1 'female'; 2: 1 'Before 1950' 2 '1950-1968' 3 '1969—1976': 3: 0 'no education' 1 '1-3 years' 2 '4-6 years' 3 ‘7-9 years' 4 '10-12 years' 5 '13-15 years' 6 '16+ years': 2: 1 'manual laborer 2 'teacher' 3 'skilled laborer 4 'administrator' 5 'preacher'; g: 1 'Shi' 2 'Rega' 3 'Other'; 1 1 'not at all necessary' 2 'not necessary but good' 3 'very necessary'; 8: 1-5 = lowest to highest; 2: 0 'does not—attend' 1 'less than lX/mo. 2 'l- 2X/mo. 3 '1X/week' a 'more than lX/week' lQ-lg: percentage of standard variant occurrences: l3: 6-99 =-lowest to highest frequencies of Swahili words relative to French: l2: percentage of lakini occurrences. 1 2 3'4 5 6 7 8 9 10 11 12 13 15 15 1 s b e o e S N k an waN noJ Jua Ire lak 0 0 1 2 1 2 3 4 4 003 006 000 000 047 100 0 0 1 0 1 3 2 4 4 000 000 000 007 052 100 0 0 1 0 1 1 3 5 2 000 000 000 004 099 0 0 1 0 1 3 3 3 3 001 000 000 000 063 1 0 1 0 1 3 3 5 3 000 011 048 100 0 0 1 0 1 3 008 002 000 006 069 100 0 0 1 0 1 1 3 5 2 000 003 000 008 074 100 0 0 1 0 1 3 3 5 3 000 006 053 0 1 1 0 0 1 2 5 3 000 000 000 000 093 100 0 0 1 0 1 3 3 5 3 000 000 005 001 054 100 0 0 1 1 1 1 3 5 4 000 003 005 053 066 100 1 0 1 0 1 3 006 004 000 001 065 100 0 0 1 2 1 3 3 2 3 030 075 009 062 025 100 1 0 1 0 1 1 3 5 3 000 000 019 007 086 100 1 0 1 1 1 1 000 000 000 017 040 100 0 1 1 0 1 3 4 3 000 000 008 098 099 100 270 I... OOOOOOOOOO OONHOOOHPOOHNOHOOOOOOCFflN°NOHHOOO°OOOHOO OOHOOI‘OOOO OOOOHOMHOHFOF‘HOOOOF‘OOOOOOOOQOOOOGCOOOOOH' N JONNNNNNNNN NNNNNNNNNNNNNNNNNNNNHHflh‘flflhfiHl-‘HFHFF‘HHFOHt-DH w apwpmwaoanAhacmwmwmhwpmanuoowawnmbaamunanwwopwnwoo 5 hut-s» ovum—.5» ”kuwuauwwwwphhn—sNHwHNflpflnprwwHHh-Huupnp NH 9‘ p FNGNI‘NQ nnnwwwwpw» P“NfluwhNNHNMW“MNNHwNHHHHHI-‘w “09106303 “N” was» com 09 03 03 N 63030601 “N60 “NM 0' (henchmen $.50! 010109 “'0‘ :5 but“ “:5 O 0| NON“ OIO'IO'I CI vb bub“ acne- 35““ (a) ”hub 60 nib“. H0903 60 #0035 2” 10 M0 0M 0M 0” M0 MW M4 0U 0” M3 M8 M3 0M M4 0M M3 0M 1M M4 M0 0M M3 M0 0M M0 M0 0M M0 0M 0M 0M 0M on M0 0M M8 M0 0“ 0M M0 M3 M4 0” 0M M0 0% 0M M6 11 M0 M7 M0 M4 M0 0M M1 M0 M3 0” 1M M0 M0 M9 0% M1 0M 1M M3 0M 0M M9 M0 M3 0M 0M M5 0“ 0M 0M 0M 0M M4 M0 0M M4 0M M4 0M 0M 0M 0M 0% M0 M0 0M M3 M0 12 0M 0M 0M 0M 0M 0M M7 M5 0M 0M M4 0M 0M M4 0M M1 0% M5 M0 0M 0M 0M 0M 0M 0M M0 M4 0M 0M M0 0M 0M M0 0M M5 M2 0M 0M 0M 0% M3 0% M3 0M M0 0” M0 0M 13 M0 0M 0M M8 0M 0“ NO 0M M9 0% N0 0M M0 1M M5 M4 M1 M0 M0 M2 0” 0M M4 on M0 M0 1M 0” M4 M6 M0 0M M1 0M 0M 0M M3 0M 0M M3 M3 M3 M0 M4 M5 M6 M6 M0 M9 M9 M2 0“ M9 M9 0M M4 M9 0” 0M 0M M3 M4 M4 M3 M0 M9 M9 0” 0“ M3 M6 M9 0M 0M 0M 0” M7 0M M7 M1 0M M0 0M M3 0M 0M M1 0M M2 M0 M3 M5 M9 0M M2 M0 M N0 N0 1M mo mo mo M9 mo 0M 1M M2 0M M3 0M M3 M0 M0 1M M2 1M M5 0M M0 M0 0M 0M M3 0M 0M 0M M3 1M M2 M9 M4 M9 0M mo 0M 0“ M0 M0 mo 0M M6 F‘OOOOOOOOOOOQOOOOONNOHOO¢HOO°3°OOOHOOOO°°O ll!- OHHOHOOOP‘OOF‘OOFI‘HHOHOI‘HHOOOHOOHHOOHOHHOOO ll“ woowwwwo:uwwwwwwwwawwwwwwwwwuwwnwnunnwwwuwnw IIU' l-‘NNNNNNONNNNOO“”MwwwflwbwwathIbhOwhéNNwhNQQ IIG COOOOOOOOOOOOOOOOOOPOOOOOOOWOONP NOMF‘HOHHH IIO ”HWMFHQHHMMQFHN 110 NNHHNNPNMHF‘MPHGNI‘HH”WHO-.0300 “N ”0360”” N N um N“ NNNN “N “H” we: "2 w 0' :h-Nnb I!” O‘NHOIC'UIN wwwwapuu 010' 000%“ 030' mono: Nab but») 03“” N“ woo 0903 03:5 an 043 000 022 000 000 004 000 003 092 018 000 003 012 000 049 006 000 000 000 006 016 027 016 001 003 003 000 000 000 000 092 000 000 000 001 000 020 002 000 000 000 000 272 MIN 075 000 040 000 005 007 000 006 100 040 000 008 000 004 074 038 000 000 009 007 009 021 033 003 063 008 009 000 000 000 085 014 000 000 000 003 019 010 012 010 000 000 no.1 042 000 000 000 000 007 002 004 100 003 000 000 019 000 000 003 016 000 014 000 000 021 020 000 021 002 000 000 003 005 068 003 000 014 009 000 029 002 000 000 000 002 Jua 095 008 020 083 019 016 021 098 099 050 022 004 036 000 100 075 075 083 033 010 006 100 000 097 100 048 000 075 070 055 025 100 059 042 003 000 100 094 100 100 000 064 tre 032 018 035 011 028 031 045 014 051 050 055 026 062 046 017 012 019 026 027 030 023 020 099 018 022 035 062 036 021 015 037 021 046 041 040 037 075 027 017 034 018 lak 089 080 095 9.000 100 031 091 007 093 100 100 000 100 010 000 041 086 100 100 000 094 019 100 100 092 081 100 100 100 093 095 097 090 100 094 086 053 Appendix K Samples of Recorded Interview Texts In this appendix, I give excerpts from several texts to provide extended examples of subjects’ speech. The first is from the speech of one of the oldest subjects. His father was a powerful chief of the Bukavu area when it was ethnically homogeneous, before the incursion of Europeans. He had no formal education. The second text excerpt is of a subject from the middle age group with formal education whose linguistic scores used for testing the hypotheses are fairly close to the median scores of her age group. This excerpt is given as an example of fairly typical Bukavu Swahili. The third and fourth excerpts are from two whose educational and.socioeconomic statuses are high; both use more French than most other subjects (the fourth excerpt is from one who had the highest rate of French borrowings/code switches). The fifth excerpt is from a preacher who uses a relatively high amount of Standard forms. The final excerpt is from a third-grade teacher who uses Swahili as a medium of instruction in the class-room. I have chosen excerpts in which the subject speaks with a minimum of interruptions from the interviewer. Capital letters signify ellipsed sounds. Italicized words are borrowed from French. "_____2 signifies that the recorded speech is unclear. 273 .274 Excerpt #1 Birth: 01894 Formal education: none Occupation: farmer Ethnic group: Shi Sex: male In this text, two features distinguishing this subject’s speech from founger speakers are: the use of 2 instead of g as the second person, singular prefix (e.g., sentence 8) and the formation of negative verb phrases with the isolated word hapana "no" rather than the negative verb prefix ha- (e.g., sentences 9 and 20). Text 1. Nakuambia kama iko zamani. 2. kwa ile siku ya I.tell.you that it.be long.ago at that day of mutoto wangu, mutoto wangu mukqua, iko saws child my child my big he.be he.be like Huyu, Mirindi. 3. Na nani, kama, kama siku minayenga, this.one Mirindi. and what* if if day I.build minalunga ndoa, minakamata mwanamuke wa mbere, I.join marriage I.take woman of before tunasinda miaka saba naye bado kuzala. we.conquer years seven and.her not.yet to.bear 4. nakamata ungine, anazala mutoto, mutoto wa mwanamuke. I.tske another she.bears child, child of woman 5. na ule mutoto mwanamuke anagufa, namufukuza. and that child woman she.dies I.send.her.away 6. nakamata ungine, ye njo iko, ye njo miye niko I.take another she whom she.be she whom I I.be naye, ye njo anazala Miri--mutoto wangu mukqua Mirindi. with.her she who she.bears child my big Mirindi 7. iko sawa Huyu Tabura Huyu iko Hapa.... she.be like this Tabura this she.be here 27S 8.0kakamata mwanamuke, okaona iko Hapana sikia vire if.you.take woman if.you.see she.be no hear that onamwambia onamufukuza unakamsta ungine. 9. kama you.tell.her you.aend.her.away you.take another if Hapana sikia vire onamuambia, onamufukuza onakamata no hear that you.tell.her you.aend.her.away you.take ungine. 10. kama anasikia onamuba--onabakia naye. another if she.bears you.atay with.her 11. njo vire kwa zamani. 12. hakulungage ndoa so that at long.ago one.didn’t.Join marriage saws Hivi banalunga ndoa. l3. kama banalunga ndoa, like this they.join marriage if they.Join marriage BAnaweka rupeta. 14. ule, ukafika mu nyumba na ule they.put ring that if.you.srrive in house and that munalunga ndoa naye, akakamata fimbo Hii nyama... you.Join marriage with.her if.she.takes stick this animal 15. zamani ukamua-- ukamwambia: lets chakula we, long.age if.you.tell.her bring food you we mwanmuke, alete; leta, fanya Hivi na Hivi you woman, she.shall.bring bring do this and this afanye; kwenda kulima, akwende kulima. she.shall.do go to.garden she.shall.go to.garden 16. njo vile ilikuwa zamani. 17. kwa sasa, Hapana so that it.was long.ago at now, no 18. kwa sasa akwambie, lets franga nikwende ku at now she.shall.tell.you bring money I.shall.go to soko we. 19. ba sasa, leta franga kwenda ku soko. market you of now bring money go to market 20. kama we Hapana na franga, we toka, uko if you no with money you get.out you.be saws maskini we. like poor you *The primary meaning of nani is "who?;whom?" but it is commonly used as a filler phrase, as in this instance, to mean "what?"; it signals that the person is searching his memory for the right expression. 276 Free translation 1. I tell you it was long ago. 2. At the time of my oldest child M, he was like this one [indicating child in room]. 3. And what, on the day I build a house, I marry, I take my first wife, we live together seven years without her giving birth. 4. I take another wife, she bears a baby girl. 5. That girl dies; I send the woman away. 6. I take another wife, she’s the one that I am with now--the one who bore M, my oldest child. 7. She is Tabura, here. 8. If you take a woman and see that she doesn’t obey you, you send her away and take another. 9. If that one doesn’t obey you, you send her away and take another. 10. If she obeys, you stay with her. 11. That’s how it was long-ago. 12. One didn’t marry like one does now. 13. If they marry, they put on a ring. 14. If you come home and the one whom you married takes a stick--animal. 15. Long ago, you tell her "You, bring food, you woman" and she brings it. "Bring it, do this and that," and she does it. "Go do the gardening" and she goes to do the gardening. 16. That’s how it was long ago. 17. Now, no. 18. Now, she says to you "You, give me money so that I can go to the market." 19. Those now: "Give me money for going to the market." 20. If you don’t have any money: "You, get out!" You’re like a poor person, you!" Excerpt #2 Birth: 1953 Formal education: Three years of secondary school Occupation: Mid-wife ' Ethnic group: Rega Sex: Female Text Interviewer’s Question: [ugonjwa] ilikuwa inakukamata aye? sickness was taking.you how "How did you come down with the sickness?" 1. ginsi ilikuwa inanikamata? 2. siyue minAWezA manner it.was taking.me I.don’t.know I.can eleza aye juu wakati inanikamata ni ma-deux heures du explain how since when it.takes.me be two hours of matin, trois heures. 3. ile saa mi sitakuwa busingisi. morning,three hours that hour me I.won’t.be sleep. 277 4. saws saa Hii mtu analalaka, unasikia saWa like hour this person sleeps, you.feel like kabusingisi kanataKa kukukamata--lakini tenA little.sleep wants to.take.you--but again unashituka unaseMa: kumbe minataKa lala. you.are.startled you.aay: really I.want sleep. 5. njoo sasa malali yangu ilikuWA inanikamata ile. so now sickness my it.was taking.me that. 6. lakiNi mbele wakati ilianzaka, nilikuwa nasikia but before when it.started,I.was feeling saWa fbie, fbie na rate. 7. inaniluma tiii, naenda like liver liver and spleen It.hurts.me then, going kupimisha ni damu, ni mukoyo, ni wapi, Habaone to.test be blood, be urine be where, they.don’t.see ataa jibu moya. 8. apres, nikasikia tu minaanza even answer one. After, I.fe1t only I.begin kuwa na-- nashitukaka vile. 9. saa moya be with-- being.start1ed like that. hour one minaona batu banaingia mu-- mule mu cha-bra, I.see people they.enter in.there in room, tunaanza pigana pigana pigana pigana pigana, spree we.begin hit.each.other after minatas-- nItasikia tu niko na-- nslalamika. I.will.feel only I.sm crying.out. 10. kama mwenzangu minalala naye, kuna if my.companion I.sleep with.her, there.is saa nasikia ananitingiza, aseMe: uko unasemasema hour I.hear she.shakes.me saying you you.talk.talk nini? 11. minasema: Hapana. 12. wala ananiambia what I.say no or she.says.to.me aseMe: uko unalota nini? 13. minaseMa: saying you.be you.are.dreaming what I.say: siko nalota, niko macho mais niliona Hivi, I.not.be dreaming I.be eyes but I.saw this, nikaona Hivi, nikaona Hivi. 14. ere anasema: Hapana, I.saw this, I.saw this. she says: no M 278 hauonake vile. 15. minasema: ndio, minaonaKa you.aren’t.seeing those.things I.say yes I.sm.seeing vile. 16. tukaikalaka. 17. apres kuikala tena those.things. we.aat after to.sit again 18. nikaonaka minalala tu Hivi mu chambre juu I.saw I.lie.down only like.this in room since alikuwaKa zamu ku Hopitali, njoo kule kule she.was night.shift at hospital, so there there masisi Rule, 18. iko saWa balikuwa bananijaribia Masisi there, it.be like they.were they.are.bothering.me bale benzangu juu hatukukuwa tunasikilizana those companions since we.weren’t we.get.along.with nabo. l9. minalala mu chambre nasikia mulango ya with.them. I.sleep in room I.hear door of chambre ina-- inafunguka, inalia. 20. kipande room it.cpens it.creaks part minaona kipande ya migulu inakuya kushimama Hivi pale I.see part of legs it.comes to.stand like.this there ku kitanda. 21. minakuwa NaseMa: sitaki yangu, at bed I.sm saying I.don’t.want mine, sitaki yangu, sitaki yangu, sitaki yangu. I.don’t.want.mine I.don't.want.mine I.don’t.want mine 22. sasa mains Yangu nilikuwa now sister-in-law my I.was naye ku chambre akasikia. with.her at room she.beard. 23. anakuya aseMe: nini? 24. uko unasema na nani? she.comes says what? you.be you.apeak with who uko unalota? 25. minasema: Hapana, siko nalota, you.be you.dream I.say no I.not.be dreaming niko macho aLakini nilikuwa napigana, niliona kitu fulani, I.be eyes but I.was fighting I.saw thing some na kitu fulani. 26 na njo tuliikalaKa tiii... and thing some. and so we.aat then... 279 Free translation 1. How I came down with it? 2. I don’t know if I can explain how since when I did it was two or three o’clock in the morning. 3. At that time, I wasn’t asleep. 4. It was a time when people sleep; you feel drowsy enough to sleep but then you are startled. You say "Really, I want to sleep." 5. That’s how I was coming down with my sickness. 6. But before it started, I was feeling my liver, liver and spleen. 7. It hurt. Later, I go to test whether it’s in the blood, urine, or where but they didn’t have any answer. 8. Then, I just feel that I’m beginning to be startled like that. 9. One time, I see people entering the room, we begin to fight, fight, fight, fight, fight, then I just hear myself crying out. 10. If I’m sleeping with a co-worker, there are times I feel her shaking me: "what are you jabbering about?" 11. I say: "Nothing." 12. Or she says to me: "What are you dreaming about?" 13. I say: "I’m not dreaming; I’m awake but I saw this, I saw that, I saw this. 14. She says: "No, you aren’t seeing those things." 15. I say: "Yes, I-am seeing those things." 16. We sat there. 17. After sitting, again I saw. I was just lying like this in a room since she had the night shift at the hospital-~the one there in Masisi. 18. It was as if my co-workers were testing me since we weren’t getting along with them. 19. I sleep in my room, I hear the door open. It creaks. 20. I see part of someone’s legs coming to stand like this, there at the bed. 21. I am saying: "Stay away from me! I don’t want to be hurt!" 22. Now, my sister-in-law with whom I was staying in the room heard. 23. She comes to me: "What is it? 24. "Who are you speaking with? Are you dreaming?" 25. I say: "No, I’m not dreaming, I’m awake but I was fighting. I saw such and such. 26. And so we sat, then... Excerpt #3 Birth: 195 Education: University graduate (3 years) Occupation: Executive secretary in prosperous private enterprise Ethnic group: Shi Sex:.female Text 1. ile siwezi yua parce que-- hapendi bibi that I.can’t know because he.doesn’t.like wife Hatumike juu ya nini? 2. mi sionAKe¥ ma-- she.doesn’t.work because of what I I.don’t.see sionAKe¥ probleme. 3. anajiimposer kwangu? I.don’t.see problem he.imposes.himself on.me e5 280 4. bien sur anapashwa jiimposer sinon mi well sure he.must impose.himself if.not I ndamuonea 5. anapashwa kujionesha I.will.look.down.on.him he.must to.show.himself kama ye njo chef. 6. bon, kaMA anasema: that he the.one.who boss well if he.says ”usitoke saa Hiyi". 7. tunatoka par example seize don’t.leave hour this we.leave for example sixteen beures mains le quart. 8. ambia "Hapana usItoke hours minus the quarter says no don’t.lesve na mi siyakuya. 9. mi nikala. 10. ban, kama sasa and me I.haven’t.yet.come me I.sit well, if now tuseme ningekuwa na bwana ku nyumba mwenye let’s.say I.would.be with husband at house who ananingoya seize heures. ll.seize beures mi he.waits.for.me sixteen hours sixteen hours me sikuye, niko na minachunga patron mpaka saa I.don’t.come I.be with I.sttend boss until time atakuya. 12. ona pale anaweza sirika. 13. comment he.will.come see there he.oan be.angry how anachelewa kule na mi niko namungoYs. l4. mais he.is.late there and me I.be be.waiting.him but mi sione, sione pourquoi mtu snAWeza kukataza me I.don’t.see I.don’t.see why person he.oan to.refuse bibi yake kUtumika.... 15. en tout cas' minazani wife his to.work in all case I.think Habsnionesge. 16. siItikage they.don’t.look.down.on.me I.don’t.sccept banionee en tout ass. 17. mais minazani they.ahall.look.down.on.me in all case but I.think inadependre, inafatana na -- gisi mutu ye peke it.depends it.follows.with with way person she alone ansjikamata. 18. kama minajikamata-— she.tskes.herself if I.take.myself sijipatie ile nguvu, mi peke I.don’t.get.for.myself that strength I alone 281 sijipatie ile bucbef, njo vile na mi I.don’t.get.for.myself that leadership so that and I bataniksmata. 19. kama saa ya kukamata decision they.will.take.me if time of to.take decision minasnza "ee tutasema-- niseme nini?" batsyua I.begin "ee we.will.say I.shall.say what they.will.know kama mi si-— sins ile uwezo ya kusema. that I I.don’t.have that ability of to.speak 20. inafaa mwanaume anisisaidie njo it.is.necesssry man he.shall.help.me so.that ntasema. I.will.speak tTranscribers agreed that a "raised tone" on last syllable indicates that the -ak- "habitual" or "emphatic" marker is ellipsed. Free translation 1. That I can’t know because--why doesn’t he want his wife - to work? 2. I don’t see any problem.... 3. Does [the boss] impose himself on me? 4. Of course! He must impose himself; if not, I won’t respect him. 5. He must show that he’s the boss. 6. If he says: "Don’t leave at this time--" 7. We leave [work], for example, at quarter of four. 8. He says: ”No, don’t leave if I haven’t yet come. 9. I stay. 10. If now-- let’s say I had a husband at the house who was waiting for me at four o’clock. 11. Four o’clock, I don’t come; I’m looking after the boss until he comes. 12. You see, he can be angry in that situation: 13. " How can she be late there while I’m waiting for her?" 14. But I don’t see, I don’t see why a person can keep his wife from working.... 15. Really, I don’t think they look down on me. 16. I don’t accept that they look down on me, really. 17. But, I think it depends; it’s according to how the individual person carries herself. 18. If I carry myself-- I don’t get for myself that strength, I alone don’t get the leadership, and like that they will take advantage of me. 19. If, when it’s time to make a decision, I begin "Hmmm, what should we-- what should I say?", they will know that I don’t have the ability to speak. 20. It’s necessary for a man to help me so that I can speak. 282 Excerpt #4 Born: 1957 Education: Post-secondary degree (three years) Occupation: Assistant director of a financial institution Origin: Kasongo area (he arrived in Bukavu in early teens) Sex: male (The use of French in the following excerpt is typical of the whole text) Text Interviewer: labda kesho au kesho kutwa perhaps tomorrow or tomorrow is.met unaweza pita directeur kabisa kule you.can pass director completely there ku [name of institution]. unawaza nini at you.think what suivant gisi uko unatumika? following way you.be you.work 1.miye personnellement, j’ai mes principes. 2. Principe mi personally, I’ve my principles Principle yangu, je n’ambitionne jamais un paste. 3. Depuis que my I not.seek never a position. since that nilianzaKA tu--si¥ unaishs ons gisi nilikuelezea. I.began not.be you.already see way I.sxplained.to.you 4. Nilimaliza, on me dit: "vous etes engage telle part; I.finished one me tells you are employed certain place je suis parti. 5. compte tenu ya msmatieres, peut-etre I am parted account held of materials maybe njo batu banajuger abo banyewe. 6. nilifika ku so people they.judge them themselves I.srrived at [name of institution] comme inspecteur. 7.quelques temps as inspector some time apres, j’etais promu chef de service; quelques temps after I was promoted head of service some time apres j’etais promu assistant du directeur. 8. si after I was promoted assistant of director if 283 banye biko na-- naapprecier bananielever ku rang ya those.who they.he evaluating they.raise.me to level of directeur, je ne trouve pas-- je ne trouverai rien de director, I not find not I not find nothing of special parce que minaassumaka mafbnctions du directeur. special because I.sssume functions of director. 9. a chaque Ibis qu’il est absent, c’est moi qui at each time that.he is absent, it’s me that dirige l’enterprise. 10. donc pratiquement, il n’y directs the.enterprise so practically, it not.there a rien de nouveau ya kuseMA bataniapprendre mule has nothing of new of to.say they.will.teach.me there mu maison. 11. parce qu’il y a meme des fbis Je fais in house because there are even some times I do trois mais d’interim. 12. je connais bien l'enterprise. three months of.interim I know well the.business parce que wakati nilipita assistant du directeur, because when I.passed assistant of.the director, nilisubir periods moYa ya formation yenye ilinipermettre I.underwent period one of training which it.permitted.me kujua enterprise, comment ca se passe. 14. c’est a dire to.know business how it occurs that’s to say le-- l’ancien directeur akaNIpatia un mais, rien que the.former director he.gave.me one month nothing that lire et comprendre tous les documents qui existent dans read and understand all the documents which are in la maison. 15. entant que assistant, si vous me poser the house. as.much as assistant if you me pose une question sur la [name of organization], je sais que as a question on the I know that it existe telle part, peut-etre sitakuws tu njo na exists such place maybe I.won’t.be only so with mareferences, mais ninajua kama inaexister mu [name of references but I.know that it.exists in institution]. 16. tout un mais, tu ne fais que lire do all a month you not do but read of ‘= 284 sept heures trente a quinze heures trente. 17. alors seven hours thirty to fifteen hours thirty so ilinipermettrags en tout cas kuniformer mwa ile cadre. it.permitted.me in all case to.train.me in that respect tThe negative copula g1 is commonly used to introduce a rhetorical question to which the expected answer is "yes". It is much less commonly used in the standard sense for signalling the negative mode. Rather, the :59 copula is generally used with a negative prefix (21: in first person, singular and ha: in all other persons). Free translation Interviewer: Maybe you’ll have the highest position at [institution] one of these days. What do you think about that in terms of how you are working? 1. Me, personally, I have my principles. 2. My principle is never to seek after a position. 3. Since I began-- you’ve already seen what I explained to you, right? 4. When I finished [school], one said to me: "You are employed in a certain place." I left for there. 5. With regard to content, maybe people judge for themselves. 6. I arrived at [name of institution] as an inspector. 7. A while later, I was promoted to be the head of service. A while later, I was promoted to be the assistant director. 8. If those who are evaluating me raise me to the director level, I won’t find that to be anything special since I have already taken on the responsibilities of director. 9. Whenever he is absent, it’s me who directs the business. 10. So, practically, there’s nothing new in-house that they will teach me 11. because there are even times when I will do three months as interrim director. 12. I know the business well because when I became assistant director, I underwent one month of training which permitted me to know it, how it operates. 14. In other words, the former director gave me one month to do nothing but read and understand all the documents which are in the house. 15. In the position of assistant, if you ask me a question about [name of organization], I know that [the answer] exists in a certain place; maybe I won’t have the references but I know if it exists in [name of institution]. 16. A whole month, you don’t do anything but read from 7:30 AM to 3:30 PM. 17. So, it really permitted me to train myself in that respect. 285 Excerpt #5 Birth: 1949 Education: Secondary school Occupation: Pastor Ethnic group: Shi Note: In this text, I underline high phonological, morphosyntactic, and lexical forms that are rarely used in Bukavu Swahili. This text is in answer to the interviewer’s question: na watoto munawasomesha "and you pay for your children’s studies?" Text 1. ndio. Mungu ni wa ajabu kama vile nimgtoka1 kusema. yes. God be of wonder like that I.have.left to.say 2. kweli hiygz elfu moja na mia tano, so elfu truly this thousand one and hundred five or thousand moja, so zaires elfu mbili kama vile nilivzosemaa mbele, one or zaires thousand two like that which.I.said before wakati unapgzipata', zinakuwa na mibaraka. time when.you.get.them they.hecome with blessings 3. tunazikamata zile pesa ambazo’ kanisa inatupatia, we.take.them those money which church it.gives.to.us tunakamata zaires elfu moja tunapatia mwanamuke. we.take zaires thousand one we.get.to woman 5. mwanamuke anapita mu Kadutu, anachuruza ndaggg‘, woman she.passes in Kadutu, she.sells smelts anachuruza visamunyu, viazi so bunga, so anachuruza she.sells bananas, potatoes or flour or she.sells vitunguu, ao pilipili. 6. hivi faida anazgpataa onions or pepper like.this profit which.she.gets kwa ile, mars nyingi‘ ananunus‘ vyakula vya siku ambszoa by that times many she.buys food of day which tutakula ns watoto7 ganapata' kukula. 7. hivi we.will.eat and children they.will.get to.eat like.this kama kunaombwa malipo kwa masomo ya watoto, mama if there.is.asked fee for school of children mother huyg2 anasaidia kwa njis ile ile kusudi tunapata namuna this she.belps by road that that so.that we.get way 286 ya kusomesha watoto wetu. 8. Hivi Mungu of to.make.study children our like.this God anaibariki na tunapata kuishi. 9. watu gengi he.blesses.it and we.get to.live people many wanashsngsa vilevile kwa namuna gani tunaishi, they.marvel also at way which we.live 10. tunasomesha watoto, ganagonjws, tunawavalisha, we.make.study children they.become.sick we.dress.them 11. lakini Mungu anaibariki kwa ile njis nimgtoka1 but God he.blesses.it by that road I.have.left kukwambia, kwamba wasaidizi wetu yani wake wetu nao to.tell.you that helpers our that.is wives our and.they ganajipa mayo, 12. ganakazana kusudi na they.give.themselves heart they.persist so.that and wapate kusaidiana ns sisi, na Mungu they.ahall.get to.help.one.another with us, and God anawabariki mahali ganapgpita'. l3. kwa njia hiyg‘v He.blesses.them place where.they.pass by road this watumishi gs Mungu ganaishi. workers of God they.live 1. The present perfect marker -me- is usually replaced by the present tense -na- or past tense -li-. (See Appendix __, section_____. 2. Standard Swahili has three demonstrative base forms. The counterparts of the hip "this, already referred to" form used in sentence 2 are hi; "this" and ile "that". Only the last two are generally used in the Bukavu vernacular. 3. Generally, only juxtaposition or the isolated morphemes gig or -enyi (the latter prefixed by a class marker) are used to indicate relative clauses in the Bukavu vernacular. 4. Generally, an [1] would be inserted between the final two vowels: ndagsla. 5. The class 10 concord prefix 91: accords with the plural noun mars "time; occurrence” (class 9/10). However, in the Bukavu vernacular, mdngi is generally used to qualify nouns in all but the second ([+humanl) class. 287 6. Generally, -uza is used rather than :gpnua for the verb "buy" (in standard, uza means just the opposite: "sell"). 7. Generally, the class 2 marker of plural, human referents is pronounced as an obstruent (stop or fricative). . Free translation 1. Yes. God is wonderful, as I’ve just said. 2. Truly, this 1,500 or 1,000 or 2,000 zaires, as I said before, when you get them, they become blessed. 3. We take the money which the church gives us and give 1,000 zaires to my wife. 5. My wife goes to Kadutu and sells smelts, sells cooking bananas, potatoes, or flour, or sells onions or hot peppers. 6. In this way, the profit which she gets by that she often uses to by the day’s food which we will eat and the children get to eat. 7. Thus, if the children’s school fees are demanded, the wife in that very way so that we will have a way to educated our children. 8. Thus, God blesses the money and we are able to live. 9. Many people are amazed also at the way we live. 10. We educate our children, they become sick-- we clothe them. 11. But God blesses the salary in the way I have just told you: our helpers, that is, our wives are courageous. 12. They persist and are able to help us, and God blesses them whereever they go. 13. In this way, God’s workers live. Excerpt #5 Birth: 1962 Education: Secondary School Occupation: Third-grade teacher Ethnic group: Shi Sex: Female Text Interviewer asks her to tell about her work, the way she teaches, and advantages and drawbacks of teaching. 1. miye . 2. kwa Hiyo kusikilizana na batoto, me for this to.get.along.with with children ya kwanza haifae kuwa mkali. 3. juu of first it.isn’t.suitable to.be mean. since 288 ukisha kuwa mkali, kila neno utafundisha, if.you.slready be ~mean each word you.will.teach mwanafunzi ataitika aseme slisika et pourtant student he.will.agree saying he.understood and yet hakusikia. 4. juu ya nini? juu ya he.didn’t.understand because of what because of bukali yako. 5. we mwalimu inafaa unakuwa meanness yours you teacher it.is.suitable you.be simple mbele ya batoto. 6. Inafaa simple before of children It.is.suitable munazoeana, we na batoto. you.become.accustomed.to.each.other you and children 7. Mukisha zoeana, utayua If.you.already are.used.to.each.other you.will.know aseme HuYu mwanafunzi iko na akili fulani, Huyu saying this student he.be with intelligence some, this mwanafunzi iko na akili ya chini, Huyu iko na student he.be with intellegence of low, this he.be with akili ya yulu. 8. banafunzi bote habakuage saws. intelligence of high students all they.aren’t same. 9. kila mutu iko na yake akili yenye Mungu each person he.be with his intelligence which God alimuumbaga nayo. 10. Usiforcer nature yake he.created.him with.it Don’t.force nature his kabisa juu spate akili ya kutambuka greatly since he.shall.obtain intelligence of to.surpass ile yenye Mungu alimuumbaga nayo. ll. Hapana. that which God he.created.him with.it no 12. Ban, minafurai minaonags ku balimu, . Well, I.smohappy I.see at teachers 13. Nilifundisha mutoto, mwisho wa mwaka anapata, I.taught child, end of year he.gets, l4. ni furaha yangu mi mwalimu juu k—- sifa be happiness my me teacher since praise itaenda asema mwalimu fulani anayUa kufundisha. it.will.go saying teacher certain she.knows to.teach. 289 15. na hangekuwa mwalimu fulani, and she.badn’t.been teacher certain, singepataka diplome yangu ou I.wouldn’t.have.gotten diploma my or singepataga Hivi na Hivi mu maisha. 16. njo I.wouldn’t.have.gotten this and this in life so furaha minaonaga ku bwalimu ile. 17. maalaka happiness I.see for teaching that ziko mingi sana, alakini na we walimu, unayUa kaMA they.be many very, but ' and you teacher, you.know that kazi ya mwalimu ni kusema. 18. na kama uko nafundisha work of teacher be to.speak and if you.be teaching inafaa kupana mamifano mingi juu bansfunzi it.is.suitab1e to.give examples many so.that students basikie. 19. banasikiliaga s travers ile they.may.understand. they.understand through that mifsno. 20. Bon, na kisha fika ku nyumba, ile saa example well and after arriving at house, that time mapreparation, manini, inafaa kwanza we preparations, what.things it.is.suitab1e first you mwalimu mbele ukule, ukuwe na nguvu. 21. kama teacher before you.eat you.be with strength if haukule muzuri, unalala mubsya, kusiriks you.don’t.eat well you.lie.down badly to.be.mad.about nini. 22. ile yote iko napunguza ma-- bitu mu kichwa. what that all it.be lowering things in head 23. byote benye ungeenda fundisha kule ku masomo, all.things which you.would.go teach there at school haiko tens. 24. so ukisha wa na bisirani ku it.isn’t again or if.you.slready be with anger at masomo, "ah, leo bibi yangu alinikosea, leo school ah, today wife my she.did.me.wrong today bwana ’angu alinifanya Hivi na Hivi." unAenda lumizs husband my he.did.me this and this you.go hurt mutoto wa batu. 26. so byote benye ungifundisha child of people or all.things which you.would.teach 290 batoto, hauna tens idee ku kichwa yako unsanza children you.don’t.have again idea in head your you.begin kuwa nssema bingine bingine. 27. unaona pale to.be saying other.things other.things you.aee there banafunzi habswezi sikis. 28. mwalimu students they.csn’t learn teacher inafaa unakuwa simple, unszoes ns it.is.suitab1e you.be simple, you.be.accustomed with batoto nA unsz——, unayus namuna ya kubaencadrer. children and you.know way of to.care.for.them 29. bale banafunzi utaona munafikisha those students you.will.eee you.enable.them.to.arrive mwisho wa mwaka muzuri. 30. na ule mweNYE alikuwa wa end of year well and that.one who he.was of kupata, ws akili, atspata. 31. ns mweNYE alikuwa to.get of intelligence he.will.get and the.one.who he.was wa kukosa, juu mu somo kunaikalsgs wa kupata ns ws of to.miss since in class there.ait of to.get and of kukosa, anskosa. to.miss he.misses Free translation 1. I ___, 2. So, to get along with children, first, one musn’t be mean. 3. Since if you’ve already been mean, the student will accept each word you teach, saying that he understood even though he did not. 4. Why? Because of your meanness. 5. If you are a teacher, you need to be simple in front of the students. You need to get used to each other, you and the children. 6. Once you’re used to each other, you will know this student has a certain intelligence level, this one has a low level, this one has a high one. 8. Students aren’t all the same. 9. Each person has his intelligence level which God created him with. 10. Don’t force his nature to obtain a level of intelligence surpassing that which God created him with. 11. No. 12. Well, I’m happy I see teachers . 13. I taught a child and at the end of the year he gets to go on. 14. It’s happiness for me since my reputation will go out. People will say: "This one knows how to teach. 15. And if it wasn’t for her, I wouldn’t have gotten my diploma or I wouldn’t have gotten this and that in life." 16. So, that’s the happiness I see in teaching. 17. There are many ’ 291 responsibilities but, as teacher, you know that a teacher’s work is to speak. 18. And if you are teaching, it’s necessary to give many examples so that the students will understand. 19. They understand through those examples. 20. Well, after coming home, the time for preparations, what all, it’s necessary first that you eat, so that you’ll be strong. 21. If you don’t eat well, you sleep poorly, and get mad about anything. 22. All that hampers your ability to think. 23. All that you were planning to teach there at school no longer exists. 24. Or you are already fuming at school: "oh, today my wife did me wrong," ”today my husband did this and that." You go and hurt the children of others. 26. Or, all that you were going to teach the children, you don’t have an idea in your heard and you begin to talk about all kinds of other things. 27. You see there that the students can’t understand. 28. If you’re a teacher, you need to be simple, get used to your students and -- know how to take care of them. 29. You’ll see that you’ll enable them to get to the end of the year well. 30. And the one who had enough skill to pass will pass. 31. And the one who fails--since in class there are always those who pass and those who fail, fails. References Annicq, Camille. 1967. Le swahili véhiculaire. Lubumbashi: Editions IMBELCO. Anon. n.d. Eléments de kiswahili véhiculaire: a l’usage des militaires de la base de Kamina. Ministere de la Défense Nationale. Anon. 1938. Langage kiswahili. Bruxelles: Imprimerie Industrielle et Financiere. Anon. 1985. English-Swahili grammar. Nyankunde, Zaire: Editions Evangélique. Bailey, Charles-James N. 1973. Variation and linguistic theory. Arlington, VA: Center for Applied Linguistics. Barrett—Reach, Elizabeth. 1986. Word-internal evidence from Swahili for Aux/Infl. Linguistic Inquiry 17.2. 559-64. Barton, H. D. 1980. Language use among Ilsla residents. Language in Tanzania, ed. by Edgar C. Polome and C. P. Hill, 176-205. Oxford: University Press. Bickerton, Derek. 1971. Inherent variability and variable rules. Foundations of Language 7. 457-92. ------ . 1973. Quantitative vs. dynamic paradigms: the case of Montreal ’que’. New ways of analyzing variation in English, ed. by C.—J. Bailey and R. Shuy, 23-43. Washington, D. C.: Georgetown University. ------ . 1975. Dynamics of a creole system. Cambridge: University Press. ------ . 1981. Roots of language. Ann Arbor: Karoma. Bizuru Burasa. 1973. Nsisssnce de Bukavu. Institut Supérieur Pédagogique-Bukavu mémoire. Burssens, Amaat. 1954. Introduction a l’étude des langues bantoues du Congo Belge. Anvers: De Sikkel. Cedergren, H. J., and D. Sankoff. 1974. Variable rules: performance as a statistical reflection of competence. Language 50. 333-55. Comrie, Bernard. 1985. Tense. Cambridge: University Press. Cooper, Robert L. 1982. A framework for the study of language spread. Language spread, ed. by R. L. Cooper, 5-36. Bloomington: Indiana University Press. 292 293 Craig, Dennis R. 1980. Models for educational policy in creole-speaking communities. Theoretical orientations in creole studies, ed. by Albert Valdman and A. Highfield, 245-266. New York: Academic Press. Deans, William A. 1953. Congo-Swahili, a lingua franca of Central Africa. The Bible Translator 4. 77-82. DeCamp, David. 1971a. Introduction: the study of pidgin and creole languages. Pidginization and creolization of languages, ed. by Dell Hymes, 13—39. Cambridge: University Press. ------ . 1971b. Towards a generative analysis of a post- creole speech continuum. Pidginization and creolization of languages, ed. by Dell Hymes, 349-70. Cambridge: University Press. Duran, James J. 1979. Non-standard forms of Swahili in West- Central Kenya. Readings in creole studies, ed. by Ian F. Hancock, 129-51. Ghent: E. Story-Scientis. Fabian, Johannes. 1985. Scratching the surface: observations on the poetics of lexical borrowing in Shaba Swahili. Anthropological Linguistics 24:1. 14—50. Ferguson, Charles A. 1959. Diglossia. Word. 325-40. Ferguson, Charles A. and Charles E. DeBose. 1977. Simplified Registers, broken language, and pidginization. Pidgin and creole linguistics, ed. by Albert Valdms, 99-125. Bloomington: Indiana University Press. Ferrez, Luiz. 1983. The origin and development of four creoles in the Gulf of Guinea. The social context of creolization, ed. by Ellen Woolford and William Washabaugh, 120-25. Ann Arbor: Karoma. Fishman, Joshua A. 1967. Bilingualism with and without diglossia; diglossia with and without bilingualism. Journal of Social Issues 23:2. 29-37. Gilman, Charles. 1976. Lexical reinterpretation in Zairean Swahili. Kiswshili 46:2. 13-16. ----- . 1979. Convergence in Lingala and Zairian Swahili. Anthropological Linguistics 21.2. 99-109. ----- . 1985. Proto-creole /r/. Diversity and development in English-related creoles, ed. by Ian F. Hancock, 33-43. Ann Arbor: Karoma. 294 Glass, Gene V. and Kenneth D. Hopkins. 1970. Statistical methods in education and psychology. Englewood Cliffs, NJ: Prentice-Hall. Goyvaerts, D. L. and A. Kabemba. 1986. Le contexte historique de la naissance et de la diffusion du kingwana. Antwerp Papers in Linguistics 44. 199-244. Goyvaerts, D. L., M. W. K. Semikenke, and D. Naeyaert. 1983. Language and education policy in the multilingual city of Bukavu. Journal of Multilingual and Multicultural Development 4:1. 47~62. Guy, Gregory R. 1980. Variation in the group and the individual: the case of final stop deletion. Locating language in space and time, ed. by William Labov, 1-36. New York: Academic Press. Haddad, Adnan. 1983. L’arabe et 1e swahili dans la république du Zaire. Paris: Société d’Edition d’Enseignement Supérieur. Hancock, Ian F. 1971. A survey of the pidgins and creoles of the world. Pidginization and creolization of languages, ed. by Dell Hymes, 509-23. Cambridge: University Press. Harries, Lyndon. 1955. Swahili in the Belgian Congo. Tanganyika Notes and Records 39. 12-15. Harries, Lyndon. 1956. Congo Swahili. Tanganyika Notes and Records 44. 50-53. Hasher, Lynn and Rose T. Zacks. 1984. Automatic processing of information: the case of frequency of occurrence. American Psychologist 39.12. 1372-88. Hemphill, R. J. 1974. Language use and language teaching in the Primary Schools. Language in Kenya, ed. by W. H. Whiteley, 455-80. Oxford: University Press. Heylen, Walter. 1977. Initiation pratique au swahili. Bukavu, Zaire: Centre d’Etudes de Langues Africaines. Hinnebusch, Thomas J. and Sarah M. Mirza. 1979. Kiswahili: msingi wa kusema, kusoma, na kusndika. New York: University Press of America. Hudson, R. A. 1980. Sociolinguistics. Cambridge: University Press. Hunter, William F. 1956. A guide to the study of Congo Swahili. Bunia, Congo Belge: Editions Evangeliques. 295 Inkelas, Sharon and Liobs Moshi. 1988. The internal structure of adjectives in Kivunjo Chaga. 19th Conference on African Linguistics, Boston University. Abstract. Kabsnga Mpanda, Mudimbs Lsmata, Muke Zihisire, Mupendawatu Safi, and Ksnga. 1983. La situation de l’enseignement au Kivu. La problématique du développement au Kivu: actes du troisieme colloque du CERUKI Bukavu, ed. anon., 341-55. Bukavu: CERUKI. Kajiga, G. 1967. Grammaire Swahili: initiation a la culture ntu. Goma, Zaire: Eveque de Goma. Kasongo Lubinda Nzuzui. 1985. Bukavu: site socio-historique de la dualité des villes sfricaines (1960-1973). Institut Supérieur Pédagogique—Buksvu. Kasongo Mukslay Mu Luba Wa Sevelino. 1985. The phenomenon of Indoubil: a case study from Bukavu. Insitut Supérieur Pédagogique-Buksvu memoire. Labeye, 1e Lieutenant-Colonel. 1928. Cours de Langue Indigéne. Bruxelles: Imprimerie R. Louis. Labov, William. 1966. The social stratification of English in New York City. Washington, D.C.: Center for Applied Linguistics. ----- . 1972. Sociolinguistic patterns. Philadelphia: University of Pennsylvania Press. ----- . 1980. The social origins of sound change. Locating language in time and space, ed. by William Labov, 251-65. Philadelphia: University of Pennsylvania Press. LeCoste, Beaudoin. 1948. De l’influence des langues congolaises bantu sur le ki-swahili. Bulletin des Juridictions Indigenes et du Croit Coutumier Congolais 10. 293-301. ------ . 1954. Le ngwana: variété congolaise du swahili. Kongo—Oversee 20:4/5. 391-408. Le Page, R. B. 1980. Theoretical aspects of sociolinguistic studies in pidgin and creole languages. Theoretical orientations in creole studies, ed. by Albert Valdman and A. Highfield, 331-51. New York: Academic Press. Manessy, Gabriel. 1977. Processes of pidginization in African languages. Pidgin and creole linguistics, ed. by Albert Valdman, 129-54. Bloomington: Indiana University Press. 296 Masumbuko Lusamaki. 1987. The use of Swahili in Bukavu primary schools. Institut Supérieur Pédagogique— Bukavu mémoire. Milroy, Lesley and Sue Margrain. 1980. Vernacular language loyalty and social network. Language in Society 9. 43-70. Milroy, James and Lesley Milroy. 1985. Linguistic change, social network and speaker innovation. Journal of Linguistics 21. 339-384. Mougeon, Raymond and Edouard Beniak. Forthcoming. The extra-linguistic correlations of core lexical borrowing. Variation in language: NWAV-XV, ed. by K. Denning. Stanford University: Dept. of Linguistics. Mpenda Gatana. 1983. L’analyse de la structure économique dans la sous-region de Bukavu. La problématique du développement au Kivu: actes du troisieme colloque du CERUKI Bukavu, ed. anon., 117-30. Bukavu: CERUKI. Muhlhausler, Peter. 1980. Structural expansion and the process of creolization. Theoretical orientations in Creole studies, ed. by Albert Valdman and Arnold Highfield, 19-56. New York: Academic Press. ----- . 1986. Pidgin and Creole linguistics. New York: Basil Blackwell. Mugsnza Lwamba, Ntamayariro Uw’Imana, and Ngendanganys Jean. 1983. Quelques reflexions sur l’école et le développement. La problémstique du développement au Kivu: actes du troisieme colloque du CERUKI Bukavu, ed. anon., 357-74. Bukavu: CERUKI. Mutamba Mskombo and Kabemba Assan. 1979. Les classes moyennss et le développement du Kivu. La problématique du développement au Kivu: actes du troisieme colloque du CERUKI Bukavu, ed. anon., 449—64. Bukavu: CERUKI. Mwamba Kapanga. 1985. Lubumbashi Swahili revisited. Presentation at the 16th Annual African Linguistics Conference. MS. Natalia, Ernest. 1965. La langue swahilie. Université de Liege: Fondation de 1’Université de Liege pou les Recherches Scientifiques en Afrique Centrale. Nida, Eugene A. and Harold W. Fehderau. 1970. Indigenous pidgins and koines. International Journal of American Linguistics 32:2. 148-56. 297 Nkiko, Busane, Mudekereza, and Mujinya. 1983. Eléments de grammaire swahili. Lubumbashi, Zaire: Editions Impala. Nurse, Derek and Thomas Spear. 1985. The Swahili: reconstructing the history and language of an African society. Philadelphia: University of Pennsylvania Press. Nzongola Mukoka. 1983. Croissance démographique et emploi a Bukavu. La problematique du developpement au Kivu: actes du troisieme colloque du CERUKI Bukavu, 287-312. Bukavu: CERUKI. d’Offay de Rieux, Danielle. 1980. Creole and educational policies in the Seychelles. Theoretical orientations in Creole studies, ed. by Albert Valdman and Arnold Highfield, 267-72. New York: Academic Press. Parkin, D. J. 1974. Language switching in Nairobi. Language in Kenya, ed. by W. H. Whiteley, 189-216. London: Oxford University Press. Platt, John T. 1977. A model for polyglossia and multilingualism with reference to Singapore and Malaysia. Language in Society 6. 361-78. Polomé, Edgar. 1969a. Lubumbashi Swahili. Journal of African Languages 7:14-25. ------ . 1969b. The position of Swahili and other Bantu languages in Katsnga. Texas Studies in Language and Literature 11:2. 905-13. ------ . 1983. Creolization and language change. The social context of creolization, ed. by Ellen Woolford and William Washabaugh, 126-36. Polome, Edgar C. and C.P. Hill, eds. 1980. Language in Tanzania. Oxford: University Press. Rickford, John R. 1987. The haves and have nots: socio- linguistic surveys and the assessment of speaker competence. Language in Society 16. 149-78. ‘Russell, Joan. 1980. Networks and sociolinguistic variation in an African urban setting. Locating language in time and space, ed. by William Labov, 125-40. Philadelphia: University of Pennsylvania Press. Rwanika Mwisha, Bukasa Kalonji, and Mwantuali Epoks. 1986. Swahili: livre du formateur. Bukavu: Centre de formation du Corps de la Paix. 298 Sacleux, Ch. 1909. Grammaire des dialectes swahilis. Paris: Procure des PP. du Saint-Esprit. - Samarin, William J..1980. Standardization and instrumentalization of creole languages. Theoretical orientations in Creole studies, ed. by Albert Valdman and Arnold Highfield, 213-36. New York: Academic Press. Sampson, Geoffrey. 1987. Review article of "Rumelhart, McClelland, et al.: Parallel Distributed Processing." Language 63.4. 871-86. Sankoff, G. 1980. The social life of language. Philadelphia: University of Pennsylvania Press. Scotton, Carol Myers. 1972. Choosing a lingua franca in an African capital. Edmonton: Linguistics Research, Inc. ------- 1977. Review Article: Language in Kenya, ed. by W. H. Whiteley. Language 53.1. 174-189. ------ . 1982. The possibility of code-switching: motivation for maintaining multilingualism. Anthropological LinKUiSticB 24e4a 432-44e ------ . 1986. Diglossia and code switching. Sociolinguistics and the sociology of language, 2, ed. by Joshua A. Fishman et al., 403-15. Berlin: Mouton de Gruyter. ------ . Roots in Swahili? n.d. A locative copula becomes a stative marker. MS. - Siegel, Jeff. 1985. Koines and koineization. Language and Society 14. 357-78. Singler, John. 1983. Variation in tense-sspect-modality in Liberian English. UCLA Ph.D. dissertation. Stapleton, Walter Henry. 1903. Comparative handbook of Congo Languages. Bologo, Congo Independent State: Baptist Missionary Society. Stigand, Captain C. H. 1915. A grammar of dialectic changes in the Kiswahili language. Cambridge: University Press. Taabu Sabiti. 1975. Cours de swahili. Bukavu: Libreza. Thomason, Sarah G. and Terrence Kaufman. n.d. Language contact, creolization, and genetic linguistics. MS. Published in 1987 by U of Cal Press, Berkeley. 299 Timm, Lenora A. 1981. Diglossia old and new--a critique. Anthropological Linguistics 23. 356-67. Trudgill, Peter. 1974. Sociolinguistics. New York: Penguin. ------- -. 1983. On dialect. Oxford: Basil Blackwell. Turk, Laurel H. and Aurelio M. Espinosa. 1983. Mastering Spanish. Lexintion, MS: D.C. Heath and Company. Whiteley, W. H., trans. 1959. Maisha ya Hamed bin Muhammed el Murjebi yasni Tippu Tip: kwa maneno yske mwenyewe. East African Swahili Committe Journals Supplement. 28:2 and 29:1. van de Weyer, P. 1929. Vocabulsire francais-kiswahili (et bemba) et éléments de conversation. Bruxelles: Imprimerie Industrielle et Financiere. van den Eynde, F. 1944. Grammaire swahili. Bruxelles: Editorial office. van Spsandonck, Marcel. 1965. Practical and Systematical Swahili Bibliography. Leiden: E. J. Brill. Verbeken, A. 1944. Petit cours de Kiswahili pratique suivi d’un vocabulaire Francais-Kiswshili, Kiswahili-Francais et de phrases usuelles. Elisabethville: IBELCO. Wardhsugh, Ronald. 1986. An introduction to sociolinguistics. Oxford: Basil Blackwell. Whitehead, J. 1926. The possibility of a standardized Swahili. Congo Mission News 55. 20-23. Whitehead, John and L. F. Whitehead. 1928. Manuel de Kingwana. Wayiko, 1e Lualaba, Congo Belge: Imprimerie de la Mission. Whiteley, W.H., ed. 1974. Language in Kenya. Oxford: University Press. Winford, Donald. 1985. The concept of "diglossia" in Caribbean creole situations. Language in Society 14. 345-56. Woolford, Ellen. 1983. Introduction. The social context of creolization, ed. by Ellen Woolford and William Washabaugh, 1-9. Ann Arbor: Karoma. 300 Wurm, Stephen A. 1977. Pidgins, creoles, lingue franche, and national development. Pidgin and creole linguistics, ed. by Albert Valdman, 333-57. Bloomington: Indiana University Press. Young, Crawford and Thomas Turner. 1985. Rise and Decline of the Zairean State. Madison: University of Wisconsin Press.