The Creole Origins of African American Vernacular English:
Evidence from copula absence
John R. Rickford
(In African American English, ed. by Salikoko S. Mufwene, John R. Rickford,
Guy Bailey and John Baugh. London: Routledge, 1998)
(Revised final version: 9/17/97)
Two issues loom large in discussions of the development of African American Vernacular English (AAVE).1 The first is the "creole origins issue"--the question of whether AAVE's predecessors, two or three hundred years ago, included creole languages similar to Gullah (spoken on the islands off the coast of South Carolina and Georgia) or the English-based creoles of Jamaica, Trinidad, Guyana, Hawaii or Sierra Leone. The second is the "divergence issue"--the question of whether AAVE is currently diverging or becoming more different from white vernacular dialects in the US.
The creole origins issue is the older issue. The earliest linguists to suggest the possibility that AAVE had pidgin or creole roots were Schuchardt (1914), Bloomfield (1933:474), Wise (1933) and Pardoe (1937).2 The case was articulated in more detail by B. Bailey (1965) and repeated in Hall (1966:15). It was vigorously championed by Stewart (1967, 1968, 1969) and Dillard (1972, 1992), and it was subsequently endorsed by Baugh (1979, 1980, 1983), Holm (1976, 1984), Rickford (1974, 1977), Fasold (1976, 1981), Smitherman (1977), Edwards (1980, 1991), Labov (1982), Mufwene (1983), Singler (1989, 1991a, 1991b, to appear), Traugott (1976), and Winford (1992a, 1992b, 1997), among others. Arguing against the creole hypothesis, and asserting instead that the speech of African Americans derives primarily from the dialects spoken by British and other white immigrants in earlier times (hence the label "dialectologist") were Krapp (1924, 1925), Kurath (1928), Johnson (1930), Brooks (1935, 1985:9-13), McDavid and McDavid (1951), McDavid (1965), Davis (1969, 1970), D'Eloia (1973), Schneider (1982, 1983, 1989, 1993b), Poplack and Sankoff (1987), Poplack and Tagliamonte (1989, 1991, 1994), Montgomery (1991), Tagliamonte and Poplack (1988, 1993), Montgomery et al (1993), and Ewers (1996), among others. It should be added that positions are not always as polarized as these lists of creole proponents and opponents might suggest. For instance, while McDavid and McDavid (1951) felt that most AAVE features came from White speech, they recognized creole influence in the case of Gullah, and urged careful study of African and creole languages to see whether AAVE features in other areas might be traced to these. Similarly, Winford's (1997) paper is self-described as written from "a creolist perspective"--but it is one which allows for considerably more influence from British and other white dialects than creolists like Stewart and Dillard would concede. And Mufwene (1992:158) argues that "neither the dialectologist nor the creolist positions accounts adequately for all the facts of AAE" and that new intermediate positions are necessary.
The divergence issue is more recent, first advanced in a 1983 conference paper by Labov and Harris (published as Labov and Harris 1986) on the basis of data from Philadelphia, and supported by other researchers from the University of Pennsylvania--Ash and Myhill (1986), Graff, Labov and Harris (1986), Myhill and Harris (1986)--with data from the same city. Data from the Brazos Valley, Texas, and from elsewhere in the South were also introduced in support of this claim by Bailey and Maynor (1985, 1987, 1989). The issue was debated by Ralph Fasold, William Labov, Fay Boyd Vaughn-Cooke, Guy Bailey, Walt Wolfram, Arthur Spears and myself in a panel discussion at the fourteenth annual conference on New Ways of Analyzing Variation (NWAV14), held at Georgetown University in 1985 (Fasold et al 1987). Butters (1989) is a critical book-length review of the divergence literature. Other contributions to this issue, several recognizing convergence as well as divergence in the recent history of AAVE, include Bailey (1993), Denning (1989), Butters (1987, 1988, 1991), Myhill (1988), Rickford (1991b) and Edwards (1992).
I will concentrate now on the creole origins issue since it is the older and better investigated one and the one which continues to inspire more controversy and new research.
8.1. Some definitions
To understand the "creole origins issue," we need to have some idea of what pidgins and creoles are, and for this, I will draw on Rickford and McWhorter (1997:238):
Pidgins and creoles are new varieties of language generated in situations of language contact. A pidgin is sharply restricted in social role, used for limited communication between speakers or two or more languages who have repeated or extended contacts with each other, for instance through trade, enslavement, or migration. A pidgin usually combines elements of the native languages of its users and is typically simpler than those native languages insofar as it has fewer words, less morphology, and a more restricted range of phonological and syntactic options (Rickford 1992a:224). A creole, in the classical sense of Hall (1966), is a pidgin that has acquired native speakers, usually, the descendants of pidgin speakers who grow up using the pidgin as their first language. In keeping with their extended social role, creoles typically have a larger vocabulary and more complicated grammatical resources than pidgins. However, some extended pidgins which serve as the primary language of their speakers (e.g. Tok Pisin in New Guinea, Sango in the Central African Republic) are already quite complex, and seem relatively unaffected by the acquisition of native speakers . . .
Although it was assumed for a long time that creoles evolved from pidgins, Thomason and Kaufman (1988:147-166) and others have argued that many creoles, particularly those in the Caribbean and in the Indian Ocean, represent "abrupt creolization," having come into use as primary or native contact languages before a fully-crystallized pidgin had had time to establish itself.
We also need to take into account creole continuum situations, like those in Guyana, Jamaica, and Hawaii, where, in between the deepest Creole (the basilect) and the most standard variety of English (the acrolect), there exists a spectrum of intermediate varieties (the mesolects). In the pioneering work of DeCamp (1971) and many of his successors, it was assumed that such continua developed from earlier bilingual creole/standard situations through a process of decreolization in which the creole variety was gradually levelled in the direction of the standard. However, Alleyne (1971) suggested that in Jamaica, a continuum-like situation may have existed from the very beginnings of Black/White contact, depending on the degree and nature of the contacts which house slaves, field slaves, and other segments of the slave community (e.g. old hands vs. the newly arrived) had with metropolitan English speakers. Subsequently, Baker (1982, 1991:277), Bickerton (1986) and Mufwene (1996) suggested that, given the lower proportions of Blacks to Whites in the founding phase of most colonies, creole continua may actually have "backwards," with the first generations of Africans acquiring something closer to metropolitan English, and later generations acquiring successively "restructured" or creolized varieties as they had less access to White norms and learned increasingly from each other.
The reason this issue is relevant to us is that early creolists like Dillard and Stewart tended to assume that the earliest variety of AAVE was a relatively uniform and basilectal creole which subsequently decreolized into mesolectal forms increasingly closer to English. However, more recent discussions of the creole issue, for instance by Rickford (1997) and Winford (1997), provide more explicitly for variation across a continuum of varieties from very early on, although I (for one) contend that creole varieties were a significant mix of the early contact situation, particularly in the South, and that a gradual process of quantitative decreolization must have been taking place in the USA over time, with fewer speakers using Creole varieties, and more speakers using varieties closer to standard English.
8. 2 Relevant questions and evidence in relation to AAVE. From the point of view of the creolist/dialectologist debate, the fundamental question is whether a significant number of the Africans who came to the United States between the seventeenth and nineteenth centuries went through processes of pidginization, creolization and (maybe) decreolization in acquiring English (the creolists' position), or whether they learned the English of British and other immigrants fairly rapidly and directly, without an intervening pidgin or creole stage (the dialectologists' position).
Although linguists who address the creole issue typically concentrate on one kind of evidence, or at most two, there are at least seven different kinds of evidence which could be brought to bear on the primary question of whether AAVE was once a creole, each of them involving secondary questions of their own.
8.2.1. One could ask, first of all, whether the sociohistorical conditions under which Africans came to and settled in the United States might have facilitated the importation or development of pidgins or creoles. With respect to importation, Stewart (1967), Dillard (1972), and Hancock (1986) favor the hypothesis that many slaves arrived in the American colonies and the Caribbean already speaking some variety of West African Pidgin English (WAPE) or Guinea Coast Creole English (GCCE). Rickford (1987a:46-55) and Schneider (1991:30-33), among others, feel that such slaves were probably not very numerous. However, the case for significant creole importation from the Caribbean in the founding period has been bolstered by recent evidence that "slaves brought in from Caribbean colonies where creole English is spoken were the predominant segments of the early Black population in so many American colonies, including Massachussetts, New York, South Carolina, Georgia, Virginia and Maryland in particular." (Rickford 1997:331).
With respect to conditions for the creation or development of contact varieties on American soil, low proportions of target language (English) speakers relative to those learning it as a second language favor pidginization and creolization. The frequency of small US slave holdings and the relatively high proportion of whites to blacks in the US--in contrast with Jamaica and other British colonies in the Caribbean (Parish 1979:9, Rickford 1986:254)--are thought by some to make it less likely that these processes took place in the US, particularly in the founding period (Schneider 1989:35, Mufwene 1996:96-99, Winford 1997). However, as Schneider (ibid.) points out, "just because a majority of plantations was small does not necessarily imply that a majority of the slaves lived on small plantations"; he cites Parish's (1979:13) observation that "the large-scale ownership of a small minority meant that more than half the slaves [in the mid 19th century US] lived on plantations with more than twenty slaves."
Moreover, there were striking differences from one region to another. A creole is much more likely to have developed in South Carolina, where "blacks constituted over 60% of the total population within fifty years of initial settlement by the British" (Rickford 1986:255) than in New York, where blacks constituted "only 16% of the population as late as the 1750's, one hundred years after British settlement" (ibid). When one considers that from 1750 to 1900, 85% to 90% of the Black population lived in the South, and that African Americans in other parts of the country are primarily the descendants of people who emigrated from the South in waves beginning with World War I (Bailey and Maynor 1987:466), it is clearly the demographics of the South rather than the North or Middle colonies which are relevant in assessing the chances of prior creolization (Rickford 1997).
To variation by region must be added considerations of variation by time period. For instance, both Mufwene (1996) and Winford (1997) are more sanguine about the possibilities of creole-like restructuring in Southern colonies in the eighteenth and early nineteenth century than in the seventeenth century, as the proportions of Blacks to Whites increased. Finally, as Rickford (1977:193) has noted, "Questions of motivation and attitude must also be added to data on numbers and apparent opporunities for black/white contact." We have striking contemporary examples of White individuals in overwhelmingly Black communities (Rickford 1985) and Black individuals in overwhelmingly White communities (Wolfram, Hazen and Tamburro 1997) who have not assimilated to the majority pattern because of powerful cultural and social constraints. This is likely to have been equally if not even more the case two or three hundred years ago, when the constraints against assimilation were more powerful. _Constraints like these might have been sufficient to provide the "distance from a norm" which Hymes (1971:66-67) associates with the emergence of pidgin/creole varieties.
Although sociolinguists have recently begun to do substantive research on the sociohistorical conditions under which Africans came to and settled in the American colonies, and the possibility that they imported or developed pidgin-creole speech in the process, there is still need for more research at the levels of individual colonies or states, counties and districts, and plantations or households.
8.2.2. The second kind of evidence one might consider is textual attestations of AAVE from earlier times, or "historical attestations" for short. The known evidence of this type can be divided into two broad categories: (a) Literary texts, including examples from fiction, drama and poetry as well as those from travellers' accounts, records of court trials and other non-fictional works (Brasch 1981); and (b) Interviews with former slaves and other African Americans--many born in the mid nineteenth century--from the 1930s onward, including the two subcategories distinguished by Schneider (1993b:2): "the so-called ex-slave narratives" published by Rawick (1972-1979) , and the tape recordings made for the Archive of Folk Songs (AFS), published and analyzed by Bailey et al. (1991)." A third source of early twentieth century data are the interviews with 1605 African Americans concerning "hoodoo" which were recorded by Harry Hyatt between 1936 and 1942 on Ediphone and Telediphone cylinders and subsequently published (Hyatt 1970-1978) and analyzed (Viereck 1988, Ewers 1996).
In general, the literary texts--the primary data sources for Stewart (1967) and Dillard (1972)--take us back much further in time, to the early eighteenth century, at least; but they tend to be relatively brief and open to serious questions of authenticity (Viereck 1988:301, fn 1, Schneider 1993b:1-2). Of the early twentieth century interviews, the AFS materials--the data source for the analyses by various researchers in Bailey at al (1991)-- are generally considered the most reliable, but the audible recordings consist of only a few hours of speech from a dozen former slaves, and like the other nineteenth century materials, these represent a relatively late or recent period in African American history (cf Rickford 1991a:192, Wald 1995). Moreover, as Bailey et al note, in their introduction (p. 18-19), "the recordings and transcripts often lend themselves to a variety of interpretations" and their representativeness is limited both in terms of speaker type and time-period (cf. also Rickford 1991). The reliability of the ex-slave narrative materials--the primary data sources for the studies by Brewer (1974) and Schneider (1989), among others--has recently been questioned by Maynor (1988), Wolfram (1990) and Montgomery (1991) on the grounds that errors were introduced by field-workers who set down the texts by hand and by editors who subsequently over-represented certain stereotypical dialect features. However, Schneider (1993b) has made a spirited defense of these materials, arguing that their errors and distortions are detectable from comparisons with the AFS materials and by other means. The reliability of the Hyatt recordings--especially the early Ediphone recordings which required the interviewer to "repeat into a speaking-tube every word or phrase spoken by the informant" (Hyatt 1:xx)--is open to question. But the later Telediphone recordings (made with a microphone) and tape-recordings are better, and Ewers (1996:27) assumes that despite drawbacks, the Hoodoo material :is in principle a sufficiently reliable basis for carrying out morphological and syntactic studies." 3
8.2.3. The third source of evidence is modern-day recordings from the African American diaspora or "diaspora recordings" for short. These consist of audio recordings with descendants of African Americans who left the United States for other countries in the late eighteenth or early nineteenth century, and who, because of their relative isolation in their new countries, are thought to represent an approximation to the African American speech of their emigrating foreparents. The first diaspora data to be examined in relation to the creole issue came from the Samaná region in the Dominican Republic, where the descendants of African Americans who emigrated there in the 1820s constitute an English-speaking enclave in a Spanish-speaking nation (Poplack and Sankoff 1987, Poplack and Tagliamonte 1989, Tagliamonte and Poplack 1988, DeBose 1988, 1994). The second source of diaspora data was Liberian Settler English, the variety spoken by the descendants of African Americans who were transported to Liberia by the American Colonization Society between 1822 and 1910 (Singler 1991a:249-50). The third and most recent source of diaspora data is African Nova Scoatian English, the English spoken by the descendants of African Americans who migrated to Nova Scotia, Canada in the late eighteenth and early nineteenth centuries (Poplack and Tagliamonte 1991). Attractive though these diaspora varieties are as sources of extensive tape-recorded data on which quantitative analysis of selected variables can be performed, the significant question which they leave unanswered is whether they can indeed be taken as reflecting late eighteenth or early nineteenth century English, unaffected or only minimally affected by internally or externally motivated change (e.g. from contact with neighboring varieties of English or Spanish), and also unaffected by the Observer's Paradox (Labov 1972b:209).
8.2.4. The fourth type of evidence is similarities between AAVE and established creoles, or "creole similarities" for short. The theoretical justification for considering this type of evidence, which has been widely applied to other cases, is provided in Rickford (1977:198): "If a certain set of clear cases are agreed upon by everyone to constitute pidgins and creoles in terms of the standard theoretical parameters, and these cases display certain characteristic linguistic features, then other cases that also display these characteristics can be assumed to belong to the same type or class, unless evidence to the contrary is shown." The primary creole varieties to which AAVE has been compared are the English-based varieties spoken in Barbados (Rickford and Blake 1990, Rickford 1993), Guyana (Bickerton 1975, Rickford 1974, Edwards 1991), Jamaica (B. Bailey 1965, Baugh 1980, Holm 1984, Rickford 1991c), Trinidad (Winford 1992a, 1992b), and the South Carolina Sea Islands ("Gullah"--Stewart 1967, Dillard 1972, Rickford 1980, Mufwene 1983) and Liberian Settler English (LSE, Singler 1991a, 1993). The importance of attending to intermediate or mesolectal creole varieties rather than basilectal ones has been stressed by several researchers (Rickford 1974, Bickerton 1975, Winford 1992a), and quantitative analysis of selected features has, for the last two decades at least, become the standard comparative method. Mufwene (personal communication) has suggested that connections of AAVE to Caribbean mesolectal varieties might be informative typologically, but not historically, since "there has been no historical connection established between those varieties and AAVE." But recent sociohistorical evidence indicating the importance of Caribbean slaves in the early settlement of many American colonies (Rickford 1997) helps to provide precisely this connection.
8.2.5. The fifth type of evidence is similarities between AAVE and West African languages or "African language similarities" for short. Although the existence of lexical Africanisms might be considered of little significance, no matter how extensive, the demonstration that contemporary AAVE parallels West African languages in key aspects of its grammar might be taken as evidence of the kind of admixture or substrate influence which is fundamental to pidginization and creolization (Rickford 1977:196). Alleyne (1980), Holm (1984) and DeBose and Faraclas (1993) have provided such evidence for copula absence in AAVE, a variable to which we return in more detail below.
8.2.6. The sixth type of evidence is differences from other English dialects, especially those spoken by whites, which we might refer to as "English dialect differences" for short. As Rickford (1977:197) notes, "The question of prior creolization [of AAVE] has been frequently defined in terms of how different it now is from other English dialects and how different we can presume it to have been in the past . . . " The theoretical assumption for this is that dialects involve linguistic continuity with earlier stages or other varieties of the language, while pidgins and creoles involve "a sharp break in transmission and the creation of a new code" (Southworth 1971:255). The principal dialects to which AAVE has been compared with respect to this criterion is white vernacular dialects in the US (Davis 1969, Labov 1972a, Wolfram 1974, Bailey and Maynor 1985), although British varieties thought to have influenced AAVE through contact in the US (Schneider 1983) have also received some attention. As we will see below, this type of evidence has been more fundamental in discussions of the divergence issue than in discussions of the creole issue, with Fasold (1981) and others warning that contemporary difference might mask earlier similarities, or vice versa. Nevertheless it is still of relevance to the creole issue.
8.2.7. The seventh and final type of evidence is that which is potentially available from comparisons across different age groups of African American speakers, or "age group comparisons" for short. Such evidence could provide fundamental indications of decreolizing change in apparent time (Labov 1972b:275), but it has virtually never been invoked in relation to the creole issue. Indeed, Stewart (1970) and Dillard (1972), the principal proponents of the creole hypothesis, have argued that because of age-graded avoidance of creole forms by adults , African American children in fact use the significant creole forms more often, the exact opposite of what a theory of prior creolization and ongoing decreolization would predict. Age group data have, however, been considered more often in relation to the divergence hypothesis.
Table 1 provides a summary of the different kinds of evidence which bear on the creole hypothesis. In order to review this hypothesis further, I will now go on to survey one linguistic feature using all but the first and the last kinds of evidence (the ones which are least frequently used). Several different features have been examined in relation to the creole issue--including third person present tense and plural s-marking, perfect and past tense marking, habitual be, and completive done--but the one that has been considered most often, using the widest variety of evidence, is the absence of present tense forms of the copula be (e.g., "He Ø tall," "They Ø going") and I will accordingly survey the data on this feature.
1. Sociohistorical conditions (suitable for pidginization and/or creolization)
2. Historical attestations (literary texts; ex-slave narratives and recordings)
3. Diaspora recordings (Samaná, Liberian Settler, African Nova Scotian English)
4. Creole similarities (between AAVE and Caribbean creoles, Gullah, Hawaiian, etc.)
5. African language similarities (between AAVE and West African varieties)
6. English dialect differences (between AAVE and British/White American dialects)
7. Age group comparisons (across different generations of AAVE speakers)
Table 1: Possible types of evidence bearing on issue of creole origins of AAVE