SALIKOKO S. MUFWENE
The title of this chapter, which must be read as a frozen phrase or an idiom of some sort, is misleading. It suggests that creoles evolved from pidgins, but this genetic scenario is questioned by the colonial history of the territories where these varieties emerged, independently of each other, as I show in this chapter. Recently, some creolists have addressed the question of whether, as a group, creoles can be singled out as a structural type of languages. The answer is negative, even if one focused on creoles only, or on pidgins. Space limitations prevent me from developing this position, contra McWhorter (1998), which is discussed in Mufwene (2000) and DeGraff (2001). Creoles vary as much among themselves as indigenized Englishes do, taken as a group, and probably more than the “native Englishes” of the United Kingdom, North America, and Australia. They are not genetically related either, because the languages they have evolved from, misnamed lexifiers in creolistics, do not descend from the same parent language, although these are Indo‐European. It is plausible to argue that creoles, those that have evolved from European languages and that I discuss below, are new Indo‐European language varieties, but this position challenges the received doctrine in creolistics, which I show to be inconsistent. In order for this essay to be both informative and manageable within the space limits of this volume, I focus on what kinds of language varieties creoles and pidgins are, how they evolved, and some of what is entailed by the position I defend.
Strictly speaking, creoles and pidgins are new language varieties which developed out of contacts between colonial nonstandard varieties of a European language and several non‐European languages around the Atlantic and in the Indian and Pacific Oceans during the sixteenth through nineteenth centuries. Pidgins typically emerged in trade colonies which developed around trade forts or along trade routes, such as on the coast of west Africa. They are reduced in structures and specialized in functions (typically trade), and initially they served as nonnative lingua francas to users who preserved their native vernaculars for their day‐to‐day interactions. Some pidgins have expanded into regular vernaculars, especially in urban settings, and are called expanded pidgins. Examples include Bislama and Tok Pisin in Melanesia and Nigerian and Cameroon Pidgin Englishes, which are structurally as complex as creoles (based on, for instance, Féral 1989 and Jourdan 1991). One can certainly argue that the structural complexity of a language variety is ethnographically a function of the communicative functions it performs, even as a lingua franca, although from a typological perspective it is difficult to say whether a given language is structurally more complex than another, and especially whether a language that has complex morphosyntax is also more complex semantically or phonologically.
Creoles are vernaculars that developed in settlement colonies whose primary industry typically consisted of sugarcane plantations or rice fields and whose majority populations were non‐European slaves, in the cases of the Atlantic and Indian Ocean, or indentured laborers, in the case of Hawaii. The latter was colonized by Americans in the nineteenth century, when slavery was being abolished, and did not experience extensive ethnolinguistic mixing, which raises questions about using Hawaiian Creole English as an exemplar of how creoles developed everywhere (pace Bickerton 1981, 1984 in particular). Examples of other creoles include Cape Verdian Criolou (from Portuguese) and Papiamentu in the Netherlands Antilles (apparently Portuguese based but influenced by Spanish); Haitian, Mauritian, and Seychellois (from French); Gullah in the United States, and Jamaican and Guyanese (all from English); as well as Saramaccan and Sranan in Surinam (both from English, with the former heavily influenced by Portuguese and the latter by Dutch). Although Melanesian pidgins are associated with sugarcane plantations, they apparently originated in trade settings and were adopted on the plantations (Keesing 1988).
The terms creole and pidgin have also been extended to some other varieties that developed during the same period out of contacts among primarily non‐European languages. Examples include Delaware Pidgin, Chinook Jargon, and Mobilian in North America; Sango, (Kikongo‐)Kituba, and Lingala in central Africa; Kinubi in Southern Sudan and Uganda; and Hiri Motu in Papua New Guinea (Holm 1989; Smith 1995). Many of these varieties have historically been designated with the name jargon, which is much older in French and English, and simply means “a variety unintelligible to the speaker or writer.” The term pidgin did not arise until the early nineteenth century (Baker & Mühlhäusler 1990) or perhaps the late eighteenth century (Bolton 2002). Although it has usually been traced etymologically to the word business (as in business English), the Cantonese phrase bei chin (lit. ‘pay’ or ‘give money’) seems to be its more probable etymon (Comrie, Matthews, & Polinsky 1996: 146), because of the ecology of its emergence and also because it is phonologically more plausible to derive the word from the proposed Cantonese etymon than from the English alternative. Convergence need not be excluded as an explanation. In the original lay people’s naming practice, the term jargon was an alternate to pidgin.
Although the term must have been taken from China to Melanesia by sailors and traders in that part of the world (with Tok Pisin perhaps suggesting ‘business talk/language’), linguists are the ones who have generalized usage of the term, unfortunately without providing operational criteria for the extension to other colonial trade lingua francas. Hall (1966) and Mühlhäusler (1986) argue that pidgins are more stable and jargons are an earlier stage in the “life cycle” that putatively goes from jargon, to pidgin, to creole, to postcreole by progressive structural expansion, stabilization, and closer approximations of the base language (i.e. “lexifier”) from which the variety evolved. As explained by Mufwene (2001), the fact that the term pidgin emerged in Canton, thousands of miles away from the American Iberian colonies where the term creole originated in the stxteenth century, should have cast doubt on the scenario that derives creoles from pidgins by a putative process of nativization interpreted as structural expansion through the acquisition of native speakers. So should the fact that expanded pidgins have equally complex structures developed largely through the agency of adult L2‐speakers using it increasingly as a vernacular. The socioeconomic histories of the territories where creoles developed speak against the Hall‐Mühlhäusler position, to which I return below.
Chaudenson (1992) and Mufwene (1997) argue that creoles developed by basilectalizing away from the base language, that is, by developing a basilect – the variety the most different from the acrolect, the variety of the upper class. Mufwene (2001) emphasizes that creoles and pidgins developed in separate places, in which Europeans and non‐Europeans interacted differently – sporadically in trade colonies but regularly in the initial stages of settlement colonies. The main justification for this position is that plantation settlement colonies typically developed from homestead societies, in which the non‐Europeans were well‐integrated minorities, and their children spoke the same colonial koinés as the children of European descent. It is only during the later stage of the plantation phase that the basilects, typically identified as creoles, developed by the regular process of gradual divergence from earlier forms of the colonial language.
The term creole was originally coined in Iberian colonies, apparently in the sixteenth century, in reference to nonindigenous people born in the American colonies. (See Mufwene 1997 for references.) It was adopted in metropolitan Spanish, then in French, and by the early seventeenth century in English. By the second half of the same century, it was generalized to descendants of Africans or Europeans born in Romance‐language colonies. Usage varied from one colony to another. The term was also used as an adjective to characterize plants, animals, and customs typical of the same colonies (Valkhoff 1966).
Creole may not have applied widely to language varieties until the late eighteenth century, though Arveiller (1963) cites La Courbe’s Premier voyage (1913: 192), in which it is used for “corrupted Portuguese spoken in Senegal.” Such usage may have been initiated by metropolitan Europeans to disfranchise particular colonial varieties of their languages. It is not clear how the term became associated only with vernaculars spoken primarily by descendants of non‐Europeans. Nonetheless, some speakers of creoles (or of pidgins) actually believe they speak dialects of their lexifiers (Mühlhäusler 1985; Mufwene 1988).
Among the earliest claims that creoles developed from pidgins is the following statement by Bloomfield (1933: 474): “when the jargon [i.e. pidgin] has become the only language of the subject group, it is a creolized language.” Hall (1962, 1966) reinterpreted this, associating the vernacular function of creoles with nativization. Since then, creoles have been defined inaccurately as “nativized pidgins,” that is, pidgins that have acquired native speakers and have therefore expanded both their structures and functions, and have stabilized. Hall then also introduced the pidgin‐creole life cycle, to which DeCamp (1971) added a “postcreole” stage (see below).
Among the creolists who dispute the pidgin ancestry of creoles is Alleyne (1971), who argues that fossilized inflectional morphology in Haitian Creole and the like proves that Europeans did not communicate with the Africans in foreigner or baby talk (see below). As noted above, Chaudenson (1979, 1992, 2001, 2003) argues that plantation communities were preceded by homesteads, on which mesolectal approximations of European koinés, rather than pidgins, were spoken by earlier slaves. Like some economic historians, Berlin (1998) observes that in North American colonies, creole Blacks spoke the European language fluently. In advertisements about runaway slaves in British North American colonies, “bad English” is typically associated with slaves imported as adults from Africa. Diachronic textual evidence also suggests that the basilects developed during the peak growth of plantations (in the eighteenth century for most colonies) when infant mortality was high and life expectancy short, so the plantation populations increased primarily by massive importation of labor, and the proportion of fluent speakers of the earlier colonial varieties kept decreasing (Baker & Corne 1986; Chaudenson 1992, 2001; Mufwene 2001).
According to the life‐cycle model, as a creole continues to coexist with its base language, the latter exerted pressure on it to shed some of its “creole features.” This developmental hypothesis may be traced back to Schuchardt’s (1914) explanation of why African‐American English (AAE) is structurally closer to North American English than Saramaccan is to English in the Caribbean, namely, coexistence with the base language in North America and absence of such continued contact in Suriname. Jespersen (1921) and Bloomfield (1933) anticipated DeCamp (1971), Bickerton (1973), and Rickford (1987) in invoking decreolization as “loss of ‘creole’ features” to account for speech continua in creole communities.
It is in this context that DeCamp (1971) coined the term postcreole continuum, which must be interpreted charitably. If a variety is a creole because of the particular sociohistorical ecology of its development (see below), rather than because of its structural peculiarities, it cannot stop being a creole even after some of the features have changed. Besides, basilectal and mesolectal features continue to coexist in these communities, suggesting that the local creole variety has not died (yet). Lalla and D’Costa (1990) present copious data against decreolization in Caribbean English creoles, just as Mufwene (1994) adduces linguistic and nonlinguistic arguments against the same process in Gullah. On the other hand, Rickford and Handler (1994) show that in the late eighteenth century, Barbados had a basilect similar to those of other Caribbean islands. It is not evident that it was spoken by all the Barbadian slaves. It now seems to have vanished. How and why it was lost here but not elsewhere in the Caribbean calls for an explanation.
Closely related to the above issue is the common assumption that creoles are separate from both their base languages, whereas related noncreole colonial offspring of the same European languages are considered to be their dialects. Such is the case for the nonstandard French varieties spoken in Quebec and Louisiana, as well as on the Caribbean islands of St. Barths and St. Thomas. Likewise, except for Palenquero, New World nonstandard varieties of Spanish and Portuguese are not considered creoles, despite structural similarities with Portuguese creoles which they display. Has the fact that similar varieties are spoken by descendants of both Europeans and Africans in territories where there has been more race hybridization influenced the naming practice? Although not officially acknowledged by creolists, the one obvious criterion behind the naming practice has been to identify as creoles those varieties of European languages which have been appropriated as vernaculars by non‐European majorities (Mufwene 2001). This is well captured by what DeGraff (2003, 2005) called “creole exceptionalism.” There is otherwise no yardstick for measuring structural divergence from the base language, especially since feature composition of the latter was not the same in every relevant contact setting. Besides, contact was a factor in all colonial settings, including those not associated with creoles (Mufwene 2008).
It has also been claimed that creoles have more or less the same structural design (Bickerton 1981, 1984; Markey 1982). This position is as disputable as the other, more recent claim that there are creole prototypes from which others deviate in various ways (Thomason 1997; McWhorter 1998). The very fact of resorting to a handful of prototypes for the essentialist creole structural category suggests that the vast majority of them do not share the putative set of defining features, hence that the combination of features proposed by McWhorter (1998) cannot be used to single them out as a unique type of language. On the other hand, structural variation among creoles that have evolved from the same base language can be correlated with the sociohistorical ecologies of their developments (Mufwene 1997, 2001). The notion of “ecology” includes, among other things, the structural features of the base and substrate languages, the ethnolinguistic makeups of the populations that came in contact, the segregated population structures that fostered the new vernaculars, how regularly the residents interacted across class and ethnic boundaries, and the rates and modes of population growth.
To date, the best‐known creoles have evolved from English and French. Those of the Atlantic and Indian Ocean are, along with Hawaiian Creole, those that have informed most theorizing on the development of creoles. While the terms creole and creolization have been applied, often uncritically, to various contact‐induced language varieties, several distinctions which have not been clearly articulated have also been proposed, for instance, koiné, semicreole, intertwined varieties, foreign workers’ varieties of European languages (e.g. Gastarbeiter Deutsch), and indigenized varieties of European languages (e.g. Nigerian and Singaporean Englishes). The denotations and importance of these terms deserve reexamination (Arends, Muysken, & Smith 1995; Mufwene 1997, 2001).
The central question here is: how did creoles develop? The following hypotheses are the major ones competing today: the substrate, the superstrate, and the universalist hypotheses.
Substratist positions are historically related to the baby talk hypothesis, which I have traced back to nineteenth‐century French creolists: Bertrand‐Bocandé (1849), Baissac (1880), Adam (1883), and Vinson (1882). Putatively, the languages previously spoken by the Africans enslaved on New World and Indian Ocean plantations were the primary reason why the European languages they appropriated were restructured into creoles. These French philologists assumed African languages to be “primitive,” “instinctive,” in a “natural” state, and simpler than the relevant “cultivated” European languages. Creoles’ systems were considered to be reflections of those non‐European languages. The baby‐talk connection is that, in order to be understood, the Europeans supposedly had to speak to the Africans as if to babies, their interpretation of “foreigner talk.”
The revival of the substrate hypothesis (without its racist component) has been attributed to Sylvain (1936). Although she recognizes influence from French dialects, she argues that African linguistic influence, especially from the Ewe group of languages, is very significant in Haitian Creole. Unfortunately, she states in the last sentence of her conclusions that this creole is Ewe spoken with a French vocabulary. Over two decades later, Turner (1949) disputed American dialectologists’ claim that there was virtually no trace of African languages in AAE and showed phonological and morphosyntactic similarities between Gullah and some West African (especially Kwa) languages. He concluded that “Gullah is indebted to African sources” (Turner 1949: 54).
Mufwene (1990) identifies three main schools of the substrate hypothesis. The first, led by Alleyne (1980, 1996) and Holm (1989), is closer to Turner’s approach and is marked by what is its main weakness: invocation of influence from diverse African languages without explaining what kinds of selection principles account for this seemingly random invocation of sources. This criticism is not ipso facto an invalidation of substrate influence: it is both a call for a more principled account and a reminder that the nature of such influence must be reassessed (Mufwene 2001, 2008).
The second school has been identified as the relexification hypothesis. The proponents of its latest version, Lefebvre (1998) and Lumsden (1999), argue that Haitian is a French relexification of languages of the Ewe‐Fon (or Fongbe) group. This account of the development of creoles has been criticized for several basic shortcomings, including the following: (a) its “comparative” approach has not taken into account several features that Haitian shares with nonstandard varieties of French; (b) it downplays features which Haitian shares with several other African languages which were represented in Haiti during the critical stages of its development; (c) it has not shown that the language appropriation strategies associated with relexification are typically used in naturalistic second language acquisition; and (d) it does not account for those cases in which structural options not consistent with those of Ewe‐Fon have been selected into Haitian. Moreover, relexificationists assume, disputably, that languages of the Ewe‐Fon group are structurally identical and that no competition of influence among them was involved. For the most detailed critiques of the relexification hypothesis, see DeGraff (2002) and Aboh (2015).
The least disputed version of the substrate hypothesis is Keesing (1988), which shows that substrate languages may impose their structural features on the new, contact‐induced varieties if they are typologically homogeneous, with most of them sharing the relevant features. Thus Melanesian pidgins are like (most of) their substrates in having DUAL/PLURAL and INCLUSIVE/EXCLUSIVE distinctions and in having a transitive marker on the verb. Sankoff and Brown (1976) had shown similar influence with the bracketing of relative clauses with ia. However, the pidgins have not inherited all the peculiarities of Melanesian languages. For instance, they do not have their VSO major constituent order, nor do they have much of a numeral classifying system in the combination of pela with quantifiers. For an extensive discussion of substrate influence in Atlantic and Indian Ocean creoles, see Muysken and Smith (1986) and Mufwene (1993).
Competing with those genetic views has been the superstrate, or dialectologist, hypothesis, according to which the primary, if not the exclusive, sources of creoles’ structural features are the nonstandard varieties of their base languages. Speaking of AAE, Krapp (1924) and Kurath (1928), for example, claimed that this variety was an archaic retention of the nonstandard speech of low‐class European Americans with whom the African slaves had been in contact. According to them, African substrate influence was limited to some isolated lexical items such as goober ‘peanut’, gumbo, and okra. It would take until McDavid (1950) and McDavid and McDavid (1951) before allowance was made for limited African grammatical contributions to AAE. D’Eloia (1973) and Schneider (1989) invoke several dialectal English models to rebut Dillard’s (1972) thesis that AAE developed from an erstwhile West African Pidgin English brought over by slaves. Since the late 1980s, Shana Poplack and her associates have shown that AAE shares many features with other nonstandard vernaculars in North America and England; thus it has not developed from an erstwhile creole. (See Poplack & Tagliamonte 2001 and Poplack 1999 for a synthesis.) Because some of the same features are also attested in creoles (Rickford 1998), we come back to the question of whether most features of creoles did not after all originate in their base languages.
Regarding French creoles, the dialectologist position was first defended by Faine (1937), according to whom Haitian Creole was essentially Norman French. This position was espoused later by Hall (1958: 372), who argues that “the ‘basic’ relationship of Creole is with seventeenth‐century French, with heavy carryovers or survivals of African linguistic structure (on a more superficial structural level) from the previous language(s) of the earliest speakers of Negro Pidgin French; its ‘lexical’ relationship is with nineteenth‐ and twentieth‐century French.” Chaudenson (1989, 1992, 2001) is more accommodating to substrate influence as a factor that accounts for the more extensive structural divergence of creoles from their base languages compared to their noncreole colonial kin.
The universalist hypotheses, which stood as strong contenders in 1980s and 1990s, have forerunners in the nineteenth century. For instance, Adolfo Coelho (1967 [1880–1886]) partly anticipated Bickerton’s (1981) language bioprogram hypothesis in stating that creoles “owe their origin to the operation of psychological or physiological laws that are everywhere the same, and not to the influence of the former languages of the people among whom these dialects are found” (p. 105). Bickerton pushed things further in claiming that children made creoles by fixing the parameters of these new language varieties to their unmarked, or default, settings as specified in Universal Grammar. To account for cross‐creole structural differences, Bickerton (1984: 176–177) invokes a “Pidginization Index” that includes the following factors: the proportion of the native to nonnative speakers during the initial stages of colonization, the duration of the early stage, the rate of increase of the slave population after that initial stage, the kind of social contacts between the native speakers of the base language and the learners, and whether or not the contact between the two groups continued after the formation of the new language variety. To be sure, a number of these factors are the same as those invoked by superstratists to account for the basilectalization of creoles after the first, homestead phase, during which mostly closer approximations of the European languages were spoken by the slaves (Chaudenson 1979, 1992, 2001; Mufwene 2001).
Some nagging questions with Bickerton’s position include the following: Is his intuitively sound Pidginization Index consistent with his creolization qua abrupt pidgin‐nativization hypothesis? Is the abrupt creolization hypothesis consistent with the social histories of the territories where classic creoles developed (Mufwene 1999, 2001)? How can we explain similarities of structures and in complexity between abrupt creoles and expanded pidgins when the stabilization and structural expansion of the latter is not necessarily associated with restructuring by children? Is there convincing evidence for assuming that adult speech is less controlled by Universal Grammar than child language is? How can we account for similarities between abrupt creolization and naturalistic second‐language acquisition? Not all creolists who have invoked universalist explanations have made children critical to the emergence of creoles. For instance, Sankoff (1979) and Mühlhäusler (1981) make allowance for Universal Grammar to operate in adults, too.
Few creolists subscribe nowadays to one exclusive genetic account, as evidenced by the contributions to Mufwene (1993). The complementary hypothesis (Baker & Corne 1986; Hancock 1986; and Mufwene 1986, 2001) seems to be an adequate alternative, provided we can articulate the ecological conditions under which the competing influences (between the substrate and superstrate languages, and within each group) may converge or prevail upon each other. This position was well anticipated by Schuchardt (1909, 1914) in his accounts of the geneses of lingua franca and of Saramaccan. More and more research is now underway uncovering the sociohistorical conditions under which different creoles have developed, for instance, Arends (1989, 1995), Baker (1982), Chaudenson (1979), Corne (1999), and Mufwene (2001).
Still, the future of research on the development of creoles has some problems to overcome. So far, knowledge of the colonial nonstandard varieties of the European languages remains limited. There are few comprehensive descriptions of creoles’ structures – which makes it difficult to determine globally how the competing influences interacted among them and how the features selected from diverse sources became integrated into new systems. Few structural facts have been correlated with the conclusions suggested by the sociohistorical backgrounds of individual creoles. Other issues remain up in the air. For instance, what are the most adequate principles that should help us account for the selection of features into creoles’ systems? For developmental issues on pidgins and creoles, the following edited collections are good starting points: Hymes (1971), Valdman (1977), Hill (1979), Muysken and Smith (1986), Mufwene (1993), and Arends, Muysken, and Smith (1995). More specific issues may be checked in volumes of the Creole Language Library (published by John Benjamins) and of Amsterdam Creole Studies, in the Journal of Pidgin and Creole Languages, and in Etudes Créoles. Several issues of Pacific Linguistics also include publications on Melanesian creoles.
There is much more literature on the genesis, sociology, and morphosyntax of pidgins and creoles than on their phonologies, semantics, and pragmatics. With the exception of time reference (Singler 1990; Michaelis 1993; Schlupp 1997) and nominal number (Tagliamonte & Poplack 1993), studies in semantics and pragmatics are scant. On the other hand, the development of quantitative sociolinguistics owes a lot to research on AAE since the mid‐1960s (Labov 1972) and on Caribbean English creoles (Rickford 1987). Numerous publications in American Speech, Language in Society, and Language Variation and Change reflect this. There are also several surveys of creolistics today, including Romaine (1988), Holm (1989), Manessy (1994), Arends et al. (1995), and Mühlhäusler (1986). They vary in geographical areas of focus and adequacy. Kouwenberg and Singler (2009) has become a standard reference now, with which Chaudenson (2003) and Mufwene (2005, 2008) will have to compete in regard to their divergence from the received doctrine. DeGraff (2003, 2005) has emerged as a forceful deterrent from treating creoles as having exceptional evolutions and a good wakeup call for uniformitarianism. Efforts to bridge research on the development of creoles with that on other contact‐based varieties and phenomena (Thomason & Kaufman 1988; Mufwene 2001; Myers‐Scotton 2002; Thomason 2001; Winford 2003) are noteworthy.
Studies of structural aspects of creoles have yet to inform general linguistics beyond the subject matters of time reference and serial verb constructions. For instance, studies of lectal continua (Escure 1997) have had this potential, but little has been done by creolists to show how their findings may apply to other languages. The mixed nature of mesolects, those intermediate varieties combining features associated both with the acrolect and the basilect, should have informed general linguistics against the fallacy of assuming monolithic grammatical systems (Mufwene 1992; Labov 1998). The notion of “acrolect” deserves rethinking (Irvine 2004), especially since in places such as Surinam and the Netherlands Antilles the current acrolect is not the same language as the lexifier (Dutch as opposed to English in the former polity and Portuguese in the latter). In any case, creoles evolved from nonstandard varieties, not the standard ones. Creolistics has been bridging with research on grammaticalization, an area that promises to be productive, as evidenced by Kriegel (2003). Andersen (1983) was an important step to consolidate common interests between second‐language acquisition and the development of creoles. DeGraff (1999) bridges research on the latter topic with research on (child) language development and on the emergence of sign language. Lefebvre,White, & Jourdan (2006) and Kouwenberg and Patrick (2003) bridge in interesting ways with the scholarship on second language acquisition, as does Mufwene (2010). Aboh (2015) does this extensively with generative syntax. A monumental contribution to research on language typology has been made by Michaelis, Maurer, Haspelmath, and Huber (2013). Creolistics can also contribute fruitfully to research on language vitality, including language loss (Mufwene 2002, 2004, 2008).