Contact Languages in the Bantu Area

CONTACT LANGUAGES IN THE BANTU AREA

Salikoko S. Mufwene

University of Chicago

1. INTRODUCTION

ÝÝÝÝÝ It is possible to interpret the phrase ìcontact languageî synonymously with ìlingua franca,î viz., as that variety that enables two or more (groups of) individuals speaking different vernaculars to communicate when they come in contact with each other. The fact that, consistent with its title, Status and use of African lingua francas, Heine (1970) includes several traditional African languages whose morphosyntax is not significantly restructured may encourage this interpretation. However, this essay is only on a subset of that long list, focusing on varieties that have been identified misguidedly as pidgins or creoles (see below). This essay is primarily on (Kikongo-)Kituba and on Lingala, both spoken in the two Congo Republics, as well as on Fanakalo (also spelled Fanagalo), spoken primarily in South Africa today but formerly also in the mining belt stretching all the way north to the Shaba province of the Democratic Republic of the Congo, and on Pidgin Ewondo, spoken in Cameroon. (See Map 1) These are the most commonly cited contact languages in the literature in relation to the Bantu area. The reason for this practice may be their highly simplified morphosyntaxes relative to their lexifiers (those languages from which they have inherited most of their vocabularies). Below, I refer to them as a group with the acronym KILIFAPE.

ÝÝÝÝÝ I also discuss Shaba Swahili (as a representative of inland varieties of Swahili) and Town Bemba (spoken in Zambia). They represent forms of traditional Bantu languages which, having been brought as exogenous varieties to colonial contact settings, have been influenced by the local and other languages they came in contact with, as well as conceptually by urban colonial and post-colonial, cosmopolitan life. Another reason is to clarify that they need not be lumped in the same category as KILIFAPE, although from the point of view of their developments under contact conditions, they represent outcomes of basically the same processes of language restructuring. That is, under specific conditions of language, dialect, or idiolect contact, some structural principles are lost, innovated, or modified, amounting to a new system for the language at the communal level. Sometimes the new, restructured system is identified as a new language or dialect, as in the case of KILIFAPE.

ÝÝÝÝÝ As fuzzy as the boundary proposed between the two subsets of language varieties is, the distinction appears useful to understanding the varying ecologies and consequences of language evolution in the Bantu area. I use the term ìecologyî as short for both the ethno‚graphic conditions of language contact andóby analogy to ëgene poolí in biologyóthe feature pool constituted by the languages in contact and from which new structural principles are selected into the restructured variety. I focus on how these Bantu contact varieties evolved and, from a genetic linguistics point of view, on how they bear on our understanding of the speciation of Bantu into its many subgroups and languages. I assume the languages evolve and diversify on more or less the same pattern as parasitic species in biology, as explained in Mufwene (to appear).

2. THE VARIABLE ROLE OF CONTACT:

ÝÝÝ EXOGENOUS VERSUS ENDOGENOUS LEXIFIERS

ÝÝÝÝÝ The tradition which in genetic linguistics has invoked internally-motivated change as the primary, more ordinary, more regular, or more natural reason for language change and speciation has led linguists to marginalize some languages as exceptional in having their origins in population movements and contacts. As they have been applied to KILIFAPE and to some extent to Shaba Swahili and Town Bemba, the terms ìpidginî and ìcreole,î reflect this particular approach to language evolution, on which I raise issues in (Mufwene 1997a, 1998). I refrain from using these labels in this essay for the following reasons among others:

1) The term ìcreoleî is not structurally-motivated, as creoles vary among themselves regarding almost any structural feature that is claimed to be typical of them.

2) Creoles of the New World and the Indian Ocean developed in specific conditions of settlement colonization that have not been replicated where KILIFAPE emerged. The fact that the new African language varieties are outcomes of restructuring under contact condi‚tions is not a sufficient justification, because it can be easily shown that all cases of language speciation have involved contact and restructuring to some extent. For instance, the Romance languages developed under contact conditions, and there is no evolutionary stage of English that can be dissociated from language contact. It is not obvious that in the case of Bantu languages the proto-language itself was structurally any more homogeneous than, say, the Kongo cluster of languages when Kituba developed. Nor is it evident that it would have speciated into so many subgroups and languages if the dispersing Bantu populations had not come in contact with other ethnolinguistic groups, such as the Sudanic, Pygmy, and Khoisan populations, and/or among themselves after earlier splits. Based on archaeological and, ironically also linguistic, evidence, Vansinaís (1990) and Newmanís (1995) studies of the Bantu dispersal suggests that contact must have been a central factor in the speciation of the relevant languages.

3) The common position that creoles have developed from erstwhile pidgins is not supported by the socioeconomic histories of the territories where these vernaculars developed, as pointed out by Chaudenson (1992).

ÝÝÝÝÝ In any case, aside from the fact that they have typically been identified as contact languages, KILIFAPE, Shaba Swahili, and Town Bemba have been singled out of Heineís list for the following reasons which make it interesting to discuss them as a group:

1) They are recent developments from the nineteenth century in the case of Kituba, Lingala, and Fanakalo, or later in the twentieth century in the case of Shaba Swahili, Pidgin Ewondo, and Town Bemba.

2) They are associated with the European political and economic colonization of Africa and with the mobilization of labor from and to different parts of Africa (see Map 2).

3) While the immigrant laborers typically adopted a traditional local Bantu language, viz., Kikongo-Kimanyanga in the case of Kituba (Fehderau 1966), Bobangi in the case of Lingala (Heine 1970, Samarin 1982), Zulu (with some Xhosa vocabulary) in the case of Fanakalo (Heine 1970, Mesthrie 1989), and Ewondo in the case of Pidgin Ewondo, they also restructured it to where the ensuing system is so different from its lexifier that it is considered a separate language. To be sure there are varieties of Swahili in, for instance, Kenya, which are identified as Kisetla ësettlersí varietyí and Shengóblend of primarily Swahili and English ówhich will not be discussed here. My reasons are that Kisetla varieties are second-language approxi‚mations of the local Swahili spoken by the more indigenous African populations, and Sheng represents code-mixing, which, as discussed below, is understood as a process that may produce an autonomous variety but has not yet done so.

4) The varieties function as regional lingua francas, reflecting to some extent the linguistic regions where the colonial economic infrastructure provided the contact ecologies for their developments.

5) Their developments are partly reminiscent of those other cases also involving colonization in which European languages were adopted and restructured under contact conditions into varieties called ìpidginsî (e.g., Cameroon Pidgin English, Nigerian Pidgin English) and ìcreolesî (e.g., Cape Verdian Crioulo/Kriolu, So Tomense, Angolar, and Mauritian Creole on some of the offshore islands).

ÝÝÝÝÝ It is nonetheless useful to remember that while KILIFAPEís lexifiers were local vernaculars, such was not the case for Shaba Swahili and Town Bemba. Swahili was brought to the Congo from Tanzania. Fabian (1986:136) argues that it was imported into Shaba by a deliberate decision of the Belgian colonial administration in part in order to counter British colonial influence in the mining belt and to put an end to the use of Fanakalo. In addition, Kapanga (1991) observes that, unlike second-language varieties of Swahili discussed by PolomÈ (1968, 1971), regular Shaba Swahili did not develop by downright simplification of East Coast Swahiliís morphosyntax. Instead, this was adapted to patterns of the local Bantu languages. For instance, tense prefixes have been replaced by tense suffixes while aspect and mood are expressed periphrastically rather than affixally. The semantic distinctions havc remained basically the same where they have not increased.

ÝÝÝÝÝ Town Bemba was likewise imported from northern Zambia to its mining towns of the Copperbelt where speakers of ethnic Bemba dominated numerically and where it developed from contacts with other Bantu languages. It differs from the more traditional, rural varieties of its lexifier less by an impoverished morphosyntax than by more structural variation and by marks of influence from English especially in domains of popular culture and modernity (Spitulnik 1999).

ÝÝÝÝÝ It is the difference in the geographical origins of the lexifiers that justifies the distinction proposed above between endogenous and exogenous contact languages. The former had local lexifiers, whereas the latter had external ones. I argue below that this difference is correlated with variation in how the immigrant laborers were integrated among speakers of the lexifier and this difference in socialization bore on how the lexifier evolved structurally.

2.1. Whatís In a Name?

ÝÝÝÝÝ The motivation for grouping all the above contact varieties together may also lie in the popular names some of them bear. Town Bemba is known by other names too, such as Chikopabelti ëlanguage of the Copperbeltí, Citundu cukukalale ëtown languageí, Ichibemba ca bushimaini ëminersí languageí, and ChiLambwaza ëthe Lambasí way of speakingí (Spitulnik 1999:39; translations modified by SSM). All these names refer indirectly to the role of contact and the appro‚priation of Chibemba by non-native speakers as relevant factors in the development of its current peculiarities.

ÝÝÝÝÝ Kituba is also known under several names (see Heine 1970 and Mufwene 1997b), including the following: Kibula-matadi ëthe rock-breakerís way of speakingí in reference to the railroad construction between the Atlantic coast and Kinshasa, and Kileta ëthe administrationís way of speakingí in reference to the language variety heard from colonial administrators, especially their African auxiliaries. These too refer to the contact conditions of its development.

ÝÝÝÝÝ As made clear by Hulstaert (1974) and corroborated by Samarin (1989), the term ìBangala,î from which ìLingalaî was derived,Ý has been used rather elusively since Morton Stanleyís exploration of the Congo River in the nineteenth century. It designates not one single ethnic group but a cluster of populations speaking typolo‚gically similar languages in a geographic area along the Congo River bend, stretching from Irebu (south of Mbandaka), at the mouth of the Ubangi River, all the way to the Mongala River (May Mongala), past the town of Mankanza, also identified as Bangala. From west to east, the swampy area stretches from the Ubangi to the Congo Rivers (See Map 1). The Bangala played a key role in the riverine trade, apparently using the Bobangi language, then the most prestigious language downstream from the Congo River bend. Other populations who traded with them must have taken it to be theirs. Incidentally, Hulstaert (1989) reports an incident during a fire at a convent, when a nun overheard a helper speak to her coworkers in what turned out to be the local vernacular. Having until then mistaken Lingala to be the local vernacular, the nun asks the helper whose language the lingua franca was. The latter replied that the local population assumed it was the missionariesí language. In fact, the missionaries, who named the new language variety ìLingala,î also misapplied a Bantu class prefixation rule, using li- rather than ki- for ëlanguage of the Bangalaí or ëriverine languageí. Otherwise, Lingalaís lexifier is primarily Bobangi (Alexandre 1967, Samarin 1982, Hulstaert 1989).

ÝÝÝÝÝ Pidgin Ewondo is also known as Bulu bediliva ëmotoristsí Buluí, in association with its function as a lingua franca in trade centers and along trade routes in southern and central Cameroon, where it developed during the construction of the railway between YaoundÈ and Douala after World War I (Heine 1970)óa genetic history which is reminiscent of that of Kituba. (It has spread also to northern Gabon and to Rio Muni.

ÝÝÝÝÝ Fanakalo has long been associated with Indian South Africans and identified as Isikula ëcoolie languageí (based on a pejorative colonial designation of Indian labor). It is also known by the names Silunguboi ëEuropean-servant languageí, Kitchen Kafir (in more or less the same sense as the preceding name, with more explicit and derogative reference to Black domestics), and Basic Bantu, in reference to its oversimplified structure, marked especially by lack of agglutination on the verb and of prefixes on the noun (Mesthrie 1992:306).

ÝÝÝÝÝ It is such awareness of the role of interethnic and therefore interlinguistic contact in the development of these new varieties that led linguists to extend the terms ìpidginî and ìcreoleî as technical labels to them (e.g., PolomÈ 1968, 1971; Samarin 1982, 1990; Ngalasso 1984, 1993; Mufwene 1988, 1989; Mesthrie 1989). However, Mufwene (1997a) argues that these names actually prevent us from realizing that the mechanisms by which KILIFAPE and the like developed are the same as those which account for the more traditional cases of language speciation in the first place. NicolaÔ (1990:183f) may be right in arguing that Songhay must have developed pidginization and creolization, from the contact of Tuareg with Mande languages (see below).

2.2. The Significance of a Simplified Morphosyntax

ÝÝÝÝÝ KILIFAPE have been associated with an impoverished morphosyntax compared to the Bantu canon, and with absence of tones in the case of Kituba, Fanakalo, and Shaba Swahili. Such restructuring is another reason why all the above varieties have been singled out as contact languages. As a matter of fact, Mufwene (1988) misguidedly used structural arguments alone to conclude that Kituba is a creole. He then disregarded the fact that Lingala, which preserves a reflexive prefix and both the lexical and grammatical tone systems, has not been restructured to the same extent under similar contact conditions (Mufwene 1989).

ÝÝÝÝÝ The apparent morphosyntactic impoverishment of KILIFAPE has been associated with loss of Subject-Verb and Head-Modifier agreement systems (as in Kituba and Fanakalo), or its reduction (as in Lingala). Thus Kituba has b·na mÈneÝ dia ëthe children have eatení, dÌnu lÈnda buk·na ëthe tooth may breakí, in which the verb stem carries no agreement prefix, and b·na na mÛno ëmy childrení and dÌnu na mÛno ëmy toothí, in which possession is expressed by word order and the invariant connective na. However, while Urban Lingala exhibits much of the same Head Noun + Modifier behavior attested in Kituba, it has preserved a variable subject prefix on the verb and has also innovated a Subject-Verb agreement based in part on animacy, as is moto a-kwÈÌ ëa man has fallení and elÛkÛ e-kwÈÌ ësomething has fallení, with a- and e- as the alternating singular agreement prefixes of the third person. It is still not clear what accounts for the varying ways morphosyntactic simplification has affected the devel‚opment of KILIFAPE.

ÝÝÝÝÝ Arguments invoking loss of tones, or perhaps their replacement by a pitch-accent contrast, have likewise ignored the fact that East Coast Swahili and related languages of the area are not tonal. Besides, as observed by Ngalasso (1991) and corroborated by Mufwene (1997b), the pitch-accent system, with the high pitch born predictably by the penultimate syllable, is only partial in Kituba, which shares some lexical tone patterns with Lingala. The latter has preserved both lexical and grammatical tones, such as between m tÛ ëheadí and mto ëfireí and between ·mna (SUBJUNCTIVE) ëlet him/her seeí and amn· (REMOTE PERFECT) ëhe/she saw [A LONG TIME AGO]í. To be sure, Kituba resembles Lingala only in notÝ always having the high pitch on the penultimate syllable. It also has polysyllabic words with low tones only, as in mun k Ýëmouth, openingí, or with more than one high tone (pitch?), as in mal·l· ëorangesí, and other combinations. Together such examples call into question the position that there is a pitch-accent in the more common pattern attested in words such as dÌsu ëeyeí and kut™ba ëspeakí. The latter word counts among the evidence that the apparent hetero‚geneity of the tonal/pitch-accent system in Kituba is not necessarily due to Lingala influence, in which the word is k l ba, with low tones only. The above structural evidence questions the significance of impoverished morphosyntax as justification for identifying a restructured variety as a pidgin or creole, unless one could demonstrate that such structural complexity is later development.

ÝÝÝÝÝ Structure-based arguments for identifying particular languages as pidgins or creoles would work if there were any combination of structural features that singled such languages out developmentally apart from others. McWhorterís (1998) effort to define a creole prototype have proved unrewarding, chiefly because of exceptions within the small set of the proposed prototypes. The very attempt to identify a handful of creoles as prototypical of the category is an implicit recognition of the fact that the vast majority of them diverge from this ideal‚ization and have features that reveal how similar they are to other languages. Mufwene (1998) also shows that, the specific sociohistorical, exogenous colonial conditions of their emergence set aside, creoles have evolved by the same kinds of contacts that have influenced the develop‚ment of, for instance, the Romance languages. These involved populations shifting from a particular language to the lexifier in naturalistic second-language acquisition settings, except that the contact ecology for every language restructuring was different. It is not helpful to treat every other case as one involving creolization when there is no specific combination of diachronic structural processes that ineluctably produce creoles. We will learn more by comparing more systematically cases of language speciation accepted in genetic linguistics with those of the development of creoles, paying attention to the ecologies of their evolutions (Mufwene 1998).

2.3. The Conspicuous Absence of KILIFAPE-like Varieties in Precolonial Africa

ÝÝÝÝÝ The trend to treat KILIFAPE, and to some extent Shaba Swahili and Town Bemba, as so special from a genetic linguistics perspective has been encouraged also by something pointed out by Samarin (1986), viz., no precolonial KILIFAPE-like varieties have been reported in the early colonial literature to have been in usage even in the continentís largest precolonial kingdoms. Apparently only those who knew the neighboring lan‚guages, or were accompanied by such speakers, would venture outside their homelands, communicating with the people that they visited in their own local languages or in a lingua franca that was not significantly restructured. Lack of evidence to the contrary suggests that precolonial African kingdoms must have been regionally multilingual, in ways that must have enhanced the political status of interpreters in African royal courts, as well as during colonial expeditions. The current state of affairs in which lingua francas are associated with particular administrative regions are a legacy of the European colonial administration, which has been interpreted since the early post-colonial days as a means of fostering national or regional unity.

ÝÝÝÝÝ The long list of African lingua francas presented by Heine (1970) which do not seem to have evolved in the same way as KILIFAPE suggests that perhaps throughout African history several languages must have experienced evolutions similar to those that have produced Shaba Swahili and Town Bemba. Perhaps because their speakers had more prestige, some ethnic languages were adopted as trade languages. In some instances, they also became vernaculars of their later speakers. Missionaries used such lingua francas (for instance Bobangi, which they identified as Lingala) for the purposes of evangelization and formal education. European colonization appears to have depended on them in building its lower-level economic and administrative infrastructures. NicolaÔís (1990) hypothesis on the develop‚ment of Songhay may not be so far-fetched if it is true that while it functioned as a lingua franca in trans-Saharan trade routes, Tuareg was eventually adopted as a vernacular among the people with whom its native speakers traded and was restructured in the process. As a matter of fact, it may be informative to re-examine the spread and speciation of, for instance, Bantu on the model of such a scenario, interpreting the dispersal of its speakers eastwards and southwards in sub-Saharan Africa as a form of colonization. During this sequence of migrations and domi‚nations, they gradually assimilated the more indigenous populations, viz., the Pygmy and the Khoisan people, linguistically, but in turn the latter must have influenced the languages that they adopted.

2.4. The Contrary Fates of Endogenous and Exogenous Models

ÝÝÝÝÝ There is something that distinguishes the emergence of KILIFAPE from the minor restruc‚‚turing that has affected the other lingua francas. They are endogenous in the sense that they developed from a local language which was appropriated by outsiders who were brought from other parts of Africa as colonial auxiliaries and laborers but were apparently not absorbed by the indigenous populations. According to colonial history, these outsiders lived in special labor camps built by the companies that recruited them. As some of the names cited above suggest, the outsidersí varieties of the indigenous languages became associated with the new world order instituted by the colonial administration and accommodations to its exogenous speakers nurtured the development of these new varieties, especially after they were exported outside their birthplaces. In a way, the emergence of these varieties is also reminiscent of the development of foreign workersí varieties of European languages in Germany and France especially, except that speakers of the latter varieties carry no prestige. Nonetheless, in both cases an indigenous language was adopted by exogenous populations who were kept on the margins and precluded by the circumstances of their immigration from participating in the regular lives of the natives.

ÝÝÝÝÝ The developments of Shaba Swahili and Town Bemba represent the more traditional African trend, according to which an exogenous population succeeds in having their language prevail in the host territory. Its adoption by the local population and other immigrants causes it to be restructured but apparently not under the same conditions of linguistic heterogeneity that is typical of the contact ecologies in which KILIFAPE developed, nor to the same extent of divergence from the base language. An interesting ecological peculiarity of the varieties of Swahili discussed by PolomÈ (1968, 1971) is that they more or less reproduce the scenario of the development of KILIFAPE, because that allegedly pidgin variety of Swahili is spoken by people who have just migrated to the Swahili-speaking area and do not use it as a vernacular. In Lubumbashi and similar African cities, such speakers live with people who speak their ethnic languages and resort to the urban vernaculars (Swahili in this case) only to communicate with city-born children who often do not speak those ancestral languages or with people from other ethnic groups. The incipient pidgin-like varieties are considered transitional interlanguages and are often derided by native or more fluent speakers of the targeted varieties. Kapanga (1991) shows that the established position of Shaba Swahili as a vernacular and the continuous absorption of the newcomers into the local urban population prevents the PolomÈ varieties from gelling into a separate variety. Similar interlanguage varieties must obtain in all urban centers where rural populations have been migrating in search of jobs, trying to communicate in the local vernacular. Differences between mother-tongue and second-language varieties are to be expected under such contact conditions. Second-language varieties reflect the extent to which the speaker has become proficient in the variety.

2.5. The Impact of European Official Languages

ÝÝÝÝÝ Spitulnikís (1999) discussion of the situation of Town Bemba reveals yet another aspect of the fate of African languages that function as lingua francas and/or as vernaculars in especially the urban settings in which they are ethnographically ranked second to the official languages, typically those of former colonizers. These lingua francas have constantly been challenged to adapt to communicative demands of non-traditional aspects of the new, cosmopolitanÝ African culture. Many of their speakers are educated and experience the normal challenges of language contact in diglossic situations in which it is tolerated to mix the lower language with elements from the higher language, viz., English, French, or another European language, depending on the country. Although the literature has discussed discourse in mixed codes as if it were a deviation from putatively normal, monolingual discourse, its recognition as typical of some varieties of Town Bemba and other lingua francas suggests that the coexistence of European official languages and African lingua francas may be leading to new forms of African languages that admit contributions from European languages not only in the lexicon but perhaps also in grammar and discourse strategies.

ÝÝÝÝÝ In this vein, Knappert (1979:162) observes that ìpresent day Lingala in Kinshasa displays a complex spectrum of lexical and phonemic variationî with new sounds introduced from French. He concludes that ìLingala is changing from a simple Bantu language of the middle Congo banks, and becoming a mixed language with a considerably extended vocabulary and phonemic inventory, a development that is comparable to that of Swahili, and in many ways to that of Englishî (163). The future will tell what the structures of these lingua francas will be like and how many of the present Bantu structural features they will change or preserve. What is treated today as code-mixing or code-switching may very well contain micro-evolutionary processes that will shape up the evolutions of Lingala and other urban vernac‚ulars into varieties more different from their lexifiers, the more traditional Bantu languages, than they are now. We must, however, note that not all speakers of these new Bantu vernac‚ulars and lingua francas code-mix (as freely), although the phenomena are so common as to have stimulated the production of a large body of scholarship on code-switching (see Myers-Scotton 1993).

2.6. The Impact of Standardization

ÝÝÝÝÝ Observing that these new Bantu languages, especially Kituba and Lingala, were too impov‚erished for the purposes of evangelization and formal education, missionaries zealously developed standard varieties that would correspond to an idealized Bantu canon in which the Bible was translated and other literary texts were produced. The standard for Kituba was Kikongo-Kisantu. Ironically the modifier Kisantu means ëof saintsí, while it stands for the name of the mission-town in the Lower Congo region where Catholic missionaries developed this variety, in a senior seminary. To native speakers of Kituba like myself, it was as incom‚prehensible as any foreign language and rare are individuals who learned to speak it, although the most successful of us who went on to earn a degree and proceeded with our formal education later in French passed our examinations on the texts we read. The missionaries themselves did not speak it either in preaching to, or in com‚municating with, the indigenous populations.

ÝÝÝÝÝ More successful in this respect was the development of Mankanza Lingala, developed at the Congo River bend, which continues to be used in the education system and in much of the media (Dzokanga 1979, Meeuwis 1998). It will probably not replace the regular urban Lingala that has developed naturally without conscious engineering, nor the ìmixedî variety described by Knappert (1979). The evidence suggests that these languages may be speciating in quite interesting ways that reflect both colonial and post-colonial trends.

ÝÝÝÝÝ Interestingly, attempts to standardize Swahili and Chibemba have not been so extreme nor artificial. In the case of Chibemba, Spitulnik (1999:34) observes that ìWhile Town Bemba does exhibit minor morphological simplification in comparison to rural Bemba, there is no evidence that it was ever a minimal communication code. Town Bemba is very complex linguistically, even if there is some morphological reduction.î It ìis better understood as a variety of Bemba or a hybrid language based on Bemba.î She also observes that ìthe notion of standard Bemba derives from rural Bemba, and in particular the Bemba language from the villages at the center of the Bemba royal and ritual life, Chitimukulu and Chinsali, and from the neighboring town of Kasanta.î In the case of Swahili the standard is an approximation of coastal Tanzanian Swahili.

ÝÝÝÝÝ In any case, one learns from the above observation that these exogenous languages adopted by the local and other populations as lingua francas and/or vernaculars in the contact settings were influenced by these other languages. Kapanga (1991) suggests that the fact that a large proportion of the labor population was Luba and Lulua from Kasai, thus presenting a certain amount of substrate homogeneity, must have prevented extensive restructuring or simplification of the kind observed elsewhere. I personally suspect that integration with the native-speaking population in the industryís residential areas must have favored appropria‚tions of the target with the least restructuring, past the interlanguage stage discussed by PolomÈ (1968, 1971). All the immigrant laborers were housed in the same labor camps, in specific quarters of the mining towns; and the new language varieties spread from there.

2.7. Examining Things Genetically

ÝÝÝÝÝ From a genetic linguistics point of view, one cannot dodge a question that arises from NicolaÔís (1990) conclusion that Songhay must have developed from Tuareg by pidginizing and creolizing from its contact with Mande (and presumably other West African languages), viz., is it possible that, as surmised above, several languages that are not identified as pidgins or creoles developed through contact-induced restructuring?

ÝÝÝÝÝ Although, given the history of population movements and contacts in Africa, it is difficult not to answer this question affirmatively, I do not subscribe to the view that there are restructuring processes specific to pidgins and creoles (Mufwene 1998). Nor should it be necessary to account for the development of Bantu contact languages by assuming the same conditions that obtained on plantations of the Atlantic and Indian Oceans, where creoles lexified by European languages developed. Nor is it even necessary to assume the same sporadic contact conditions that obtained in settings where pidgins lexified by the same languages developed before the political colonization of Africa (Fabian 1986:93, 100, 108), after the Berlin Treaty in 1885. Differences in the kinds and extent of restructuring that occurred reflect variation in the specific ecologies in which the relevant lexifiers came in contact with other languages (Mufwene 1998). It is this correlation which suggests that the speciation of Bantu languages, due to the dispersal of Bantu populations southwards and eastwards, must have been on the model of the development of Shaba Swahili and Town Bemba. According to this, an exogenous language was appropriated by the local populations and others as a lingua franca and/or a vernacular and was restructured in the process. Contact is certainly a factor that need not be ruled out by fiat, because internally-motivated language change does not answer the actuation question, even if one admits that Proto-Bantu was internally variable already before the dispersal.

ÝÝÝÝÝ To understand more of the issues involved, it should help to provide a genetic synopsis of the development of Swahili on the East African coast before it penetrated its hinterland. Nurse and Spear (1985) show that it did not start as a pidgin or creole out of the contact of Arabic with Bantu languages spoken on the coast of East Africa. Although it has been penetrated lexically by Arabic in much the same way as English has by French (Nurse and Hinnebusch 1993), its essential morphosyntactic structures are very similar to those of related Bantu languages of the east coast. Nurse (1997) groups it with languages of the Sabaki group and observes that even its use of pitch-accent, instead of the more common Bantu tone system, is a peculiarity it shares with some ìother languages in East Africa (e.g. Tumbuka and Nyakyusa), spoken far from the coast and not known to have been lingua francas, [which] have also replaced tone by stressî (279). Among his conclusions about the genetic status of coastal Swahili are the following:

ÝÝÝÝÝ Little by little, Swahili was adopted as the lingua franca of East Africa, all the way to Eastern part of the Democratic Republic of the Congo, except in Uganda and in the mono‚lingual Rwanda and Burundi. Its spread gained momentum when European missionaries, colonial administrators, and industrialists also adopted it to serve their missions of evangelization, formal education, colonization of the natives, and the development of colonial industry. In the urban centers which developed subsequently, Swahili soon became the vernacular, especially for the locally born children. Such a spread and contact with the local Bantu languages lead to its speciation into several varieties distinct from the coastal varieties.

ÝÝÝÝÝ However, second-language and xenolectal varieties set aside, none of these new dialects has features that can be characterized as non-Bantu, unlike the extensive loss or reduction of Bantu morphosyntactic characteristics observed in, for instance, Kituba (Mufwene 1989) and Fanakalo (Mesthrie 1989, 1992). Kapanga (1991) interprets the restructuring that produced this speciation of Swahili in the eastern part of Central Africa as adaptation to the morpho‚syntax of the local Bantu languages, in fact in some cases as a complexification, instead of simplification. Thus he confirms the conclusion by Nurse and Spear (1985), Nurse (1997), and Fabian (1986) that Swahili has not developed by a process comparable to what is called pidginization or creolization. The commitment of missionaries and colonial admin‚istrators to teach Standard Swahili must have also prevented more extensive divergence from East Coast Swahili, although one wonders why such an effort was not successful in the development of Kituba. Nor did a similar effort prevent the entrenchment of a simplified urban Lingala alongside Mankanza Lingala used in literature and the media.

ÝÝÝÝÝ The development of morphosyntactically impoverished varieties called Pidgin Swahili , Kisetla, and Kihindi in Kenya, or theÝ Lubumbashi non-native subvariety discussed in PolomÈ (1968, 1971) correspond to the kinds and extents of the contacts in which the speakers are involved with those who speak Swahili as a vernacular. It reflects transitional stages in the way Swahili became the most important language of East Africa. More regular interactions with fluent speakers lead to fewer and fewer deviations from the base language, despite influence from the local languages.

ÝÝÝÝÝ Town Bemba seems to have evolved according to a similar process as Shaba Swahili, as noted above, by importation and dominance of an exogenous language of northern Zambia to the mining towns of its Copperbelt and by its appropriation as a lingua franca and/or vernacular by the ous populations and speakers of other languages in these contact settings. However, it has remained essen‚tially Bantu in its structures, according to Spitulnik (1999). This Bantu-preserving develop‚ment of exogenous lingua francas is so different from what happened in the development of KILIFAPE, which has led linguists to associate them misguidedly with contact languages lexified by European languages and identified earlier as pidgins and creoles.

2.8. Back to KILIFAPE

ÝÝÝÝÝ Although everything seems to have to do with the importation of exogenous labor to the contact settings in which these lingua francas have developed since the 19th century, the social ecologies of interaction that influenced language evolution must have varied. In the cases of both Shaba Swahili and Town Bemba, on the one hand, and KILIFAPE, on the other, there were indeed workersí camps in the emerging factory or industrial towns. However, the fact that KILIFAPE turned out to be significantly restructured and simplified morpho‚syntactically in comparison with their lexifiers suggests that in this case the exogenous laborers who targeted the indigenous vernaculars were not absorbed nor integrated by the indigenous populations.

ÝÝÝÝÝ In fact, the natives did not always want to participate in the colonial administrationís work projects and were only forcefully drafted to them. Samarin (1989) observes that the Bakongo people did not participate much in building the railroad which crosses their territory from the Atlantic Ocean to Kinshasa. Neither did they volunteer to accompany colonial administrators outside their homeland, fearing to be enslaved. Thus Kimanyanga speakers played a limited role as model speakers to the exogenous laborers. They participated even less in the spread of the emerging Kituba outside its birthplace and we may surmise that this new language variety was appropriated in the Bakongo area as a colonial language perhaps remotely related to languages other than Kimanyanga in the Kikongo cluster.

ÝÝÝÝÝ Perhaps such ethnographic factors explain why it wound up so simplified, although the majority of the exogenous laborers who appropriated Kimanyanga as their lingua franca were Bantu-speakers. Samarin (1982, 1990) may have been correct in emphasizing the role of West African colonial auxiliaries as the originators of Kituba. They identified Kikongo-Kimanyanga as the lingua franca of precolonial trade in the region, targeted it and shared it with the exogenous Bantu laborers. Lack of documentation makes it difficult for us to answer myriads of relevant questions conclusively. The scenario is thus partly reminiscent both of those plantations settings where creoles lexified by European languages developed and of European settings where foreign workersí varieties have developed recently. I still maintain Mufweneís (1994) position that the appropriation of Kituba or Kikongo-Kimanyanga by speakers of other Bantu languages prevented it from diverging much further from its lexifier, for instance in maintaining time reference distinctions which are almost the same in number as in the lexifier (Mufwene 1990), although the morphosyntactic system is now largely periphrastic. One must also bear in mind Owensí (1998) observation that the extent of restructuring toward an isolating morphosyntax is relative to the starting point. Ceteris paribus, lexifiers that have a very rich agglutinating morphosyntax have typically not been simplified to the same extent as those with a poor one, the case of western European lexifiers of classic pidgins and creoles.

ÝÝÝÝÝ We are led to more or less the same kind of conclusion with the development of Lingala, as with the emergence of Kituba. Samarin (1982) and Hulstaert (1989) agree that Bobangi was the riverine trade language between Stanley Pool and Irebu, south of the Congo River bend, before the late nineteenth century. Samarin (1989) observes that, like the Bakongo, the Babangi did not want to go upstream with the colonists and missionaries, but the latter took their language in the Bangala area as identified above. That is, the model which was presented in the new contact settings was non-native and it was restructured during its appropriation under exogenous conditions. We may also surmise that the predominance of the local populations speaking languages typologically close to Bobangi (among those who accom‚modated the colonists and missionaries in what they assumed to be the latterís language) kept Lingala close to the Bantu canon. Lingala has a somewhat simplified Subject-Agreement system, a reflexive pronominal prefix, and a normal Bantu tonal system (Mufwene 1989), unlike Kituba, which has no Subject-Agreement system, no reflexive prefix, and only a partial tonal system in coexistence with a pitch-accent system (Ngalasso 1991, Mufwene 1997a). Perhaps, the large involvement of populations speaking languages structurally close to Bobangi in the development of Lingala just prevented it from diverging from its lexifier as extensively as Kituba did. As a matter of fact, Owens (1998) argues that if creoles were defined by their structural features, a count of the morphosyntactic features that distinguish Kituba from Kimanyanga would make it one of the creole prototypes par excellence.

ÝÝÝÝÝ Pidgin Ewondo developed in yet a different way, on a trade route and apparently without a stable population of speakers who would develop it into a vernacular. Details of the ecology of the development of Fanakalo still remain unknown (Mesthrie 1992), it appears to be one of the two contact language varieties in the Bantu area to which the term ìpidginî could be extended, because it does not function as a vernacular. It is also the only one to the develop‚ment of which European settlers seem to have actively contributed. Colonial administrators and European missionaries typically provided the contact settings in the development of the other contact languages, except when they engaged themselves zealously in the production of varieties more consistent with the Bantu canon, viz., KiKongo-Kisantu and Lingala-Mankanza, which, ironically, were developed at extreme geographical opposites of where Bobangi is spoken. Apparently close to 30 % of Fanakaloís vocabulary is Germanic, from English and Afrikaans; 70% of it is from the Nguni languages, especially Zulu and to some extent Xhosa. It developed in the heart of the Afrikaner territory and the patterns of the handbooks designed to teach it (see further readings) reflect many of the hierarchical contexts in which it was developed, in interactions between the white rulers and the non-European workers (including the Indian indentured laborers) on the plantations, in the mines, and in the Europeansí kitchens. Thus variation in the social and linguistic ecologies alone can account for differences in the kinds and extents of restructuring in the developments of KILIFAPE.

3. CONCLUSIONS

ÝÝÝÝÝ Overall, African contact languages are colonial phenomena. They do not seem to have developed in identical ways and fall in two or three categories from the point of view of their restructuring. Shaba Swahili, Town Bemba, and the like have preserved their traditional Bantu canon, despite marks of contact with other Bantu languages, whereas KILIFAPE reveal simplified morphosyntax compared to their lexifiers. On the other hand, Lingala seems less simplified than Kituba, and Kituba apparently less so than Fanakalo, and it appears that the specific ecologies of their developments, having to do with the exogenous identities of their developers should shed light on the extent of restructuring. There is no strong argument against assuming that contact must have played a role in the speciation of Bantu languages over time and the model seems to be that of the development of Shaba Swahili and Town Bemba as exogenous varieties appropriated also by indigenous populations in the contact settings.

References

Alexandre, Pierre (1967) Langues et langage en Afrique noire. Paris: Payot.

Brumfit, A. (1980) The rise and development of a language policy in German East Africa. Sprache und Geschichte in Afrika 2.219-332.

Chaudenson, Robert (1992) Des Óles, des hommes, des langues. Paris: LíHarmattan.

Dzokanga, Adolphe (1979) Dictionnaire lingala-franÁais, suivi díune grammaire lingala. Leipzig: VEB Verlag Enzyklop”die.

Fabian, Johannes (1986) Language and colonial power: The appropriation of Swahili in the former Belgian Congo 1880-1938. Cambridge University Press.

Fehderau, Harold (1966) The origin and development of Kituba. Ph.D. dissertation, Cornell University.

Heine, Bernd (1970) Status and use of African lingua francas. Munich: Weltforum Verlag.

Hulstaert, Gustaaf (1974) A propos des Bangala. ZaÔre-Afrique 83.173-185.

Hulstaert, Gustaaf (1989) L'origine du lingala. Afrikanistische Arbeitspapiere. Institut f¸r Afrikanistik, Universit”t zu K–ln.

Kapanga, Mwamba Tshishiku (1991) Language variation and change: A case study of Shaba Swahili. Ph.D. dissertation, University of Illinois at Urbana.

Knappert, Jan (1979) The origin and development of Lingala. In Readings in creole studies, ed. by Ian Hancock, 153-64. Ghent: E. Story-Scientia P.V.B.A.

Lukas, J. (1942) Das afrikanische Verkehrssprachenproblem. Beitr”ge zur Kolonialforschung 2.15-24.

McWhorter, John H. (1998) Identifying the creole prototype: Vindicating a typological class. Language 74.788-818.

Meeuwis, Michael (1998) Lingala. Munich: Lincom Europa.

Mesthrie, Rajend (1989) The origins of Fanakalo. Journal of Pidgin and Creole Languages 4.211-240.

Mesthrie, Rajend (1992) Fanakalo in colonial Natal. In Language and society in South Africa, ed. Robert K. Herbert, 305-324. Johannesburg: Witwatersrand University Press.

Meyer, E. (1944) Das Problem der Verkehrssprachen von Tropisch-Afrika, insb. von Kamerun. Mitt. geogr. Ges. hamburg 48.253-288.

Mufwene, Salikoko S. (1988) Formal evidence of pidginization/creolization in Kituba.Ý Journal of African Languages and Linguistics 10.33-51.

Mufwene, Salikoko S. (1989) La crÈolisation en bantou: Les cas du kituba, du lingala urbain, et du swahili du Shaba. Etudes CrÈoles 12.74-106.

Mufwene, Salikoko S. (1990) Time reference in Kituba. In Pidgin and creole tense-mood-aspect systems, ed. by John V. Singler, 97-117. Amsterdam: John Benjamins.

Mufwene, Salikoko S. (1994) Restructuring, feature selection, and markedness: From Kimanyanga to Ki‚tu‚ba. In Historical issues in African linguistics, ed. by Kevin E. Moore, David A. Peterson, and Comfort Wentum, 67-90. Berkeley Linguis‚tics Society.

Mufwene, Salikoko S. (1997a) Kituba. In Contact languages: A wider perspective, ed. by Sarah G. Thomason, 173-208. John Benjamins.

Mufwene, Salikoko S. (1997b) Jargons, pidgins, creoles, and koinÈs: What are they?î In The structure end status of pidgins and cre‚oles, ed. by Arthur K. Spears and Donald Winford, 35-70. Amsterdam: John Benjamins.

Mufwene, Salikoko S. (1998) What research on creole genesis can contribute to historical linguistics. In Historical Linguistics 1997, ed. by Monika Schmid, Jennifer Austin, and Dieter Stein, 315-338. Amsterdam: John Benjamins.

Mufwene, Salikoko S. (to appear) The ecology of language evolution. Cambridge: Cam‚bridge University Press.

Myers-Scotton, Carol (1993) Social motivations for codeswitching: Evidence from Africa. Oxford: Oxford University Press.

Ngalasso, Mwatha M. (1984) Pidgins, creoles ou koinËs?: A propos de quelques langues vehiculaires africaines. Cahiers de l'Institut de Linguistique de Louvain 9.135-61.

Ngalasso, Mwatha M. (1991) Tons ou accents? Analyse des schËmes intonatifs du kikongo vÈhiculaire parlÈ dans la rÈgion de Bandundu (Zaire). Etudes CrÈoles 14.147-162.

Ngalasso, Mwatha M. (1993) Les procÈdÈs rÈpÈtitifs en kikongo: le redoublement et la rÈduplication. In Topics in African linguistics, ed. by Salikoko Mufwene and Liona Moshi, 45-66. Amsterdam: John Benjamins.

Newman, James L. (1995) The peopling of Africa: A geographic interpretation. New Haven, CN: Yale University Press.

NicolaÔ, Robert (1990) ParentÈs linguistiques: ý propos du songhay. Paris: CNRS.

Nurse, Derek (1997) Prior pidginization and creolization in Swahili? In Contact languages: A wider perspective, ed. by Sarah G. Thomason, 271-294. Amsterdam: John Benjamins.

Nurse, Derek and Thomas Spear (1985) The Swahili: Reconstructing the history and language of an African society, 800-1500. Philadel‚phia: University of Pennsyl‚vania Press.

Nurse, Derek and Hinnebusch (1993)Ý Swahili and Sabaki: A linguistic history. Berkeley, Los Angeles, & London: University of California Press.

Owens, Jonathan (1998) Representativeness in the data base: Polemical update for the twenty-first century. Language Sciences 20.113-135.

PolomÈ, Edgar (1968) Lubumbashi Swahili. Journal of African Languages 7.14-25.

PolomÈ, Edgar (1971) The Katanga (Lubumbashi) Swahili Creole. In Pidginization and creolization of languages, ed. by Dell Hymes, 57-60. London: Cambridge University Press.

Samarin, William J. (1982) Colonization and pidginization on the Ubangi River. Journal of African Languages and Linguistics 4.1-42.

Samarin, William J. (1986) Protestant missions and the history of Lingala. Journal of Religion in Africa 16.2.138-163.

Samarin, William J. (1989) The black man's burden: African colonial labor on the Congo and Ubangi Rivers, 1880-1900. Boulder, Colorado: Westview Press.

Samarin, William J. (1990) The origins of Kituba and Lingala. Journal of African Languages and Linguistics 23.47-77.

Spitulnik, Debra (1999) The language of the city: Town Bemba as urban hybridity. Journal of Linguistic Anthropology 8.30-59.

Vansina, Jan. (1990) Paths in the rainforest: Toward a history of political tradition in Equatorial Africa. Madison: University of Wisconsin Press.