Exceptionalizing genetic creolistics: a rejoinder to Mikael Parkvall and Bart Jacobs on the emergence of Berbice Dutch

1 On the essence of the article

Parkvall and Jacobs (P&J) invite the reader to situate the timing of the unusual extensive grammatical and lexical contributions of Eastern Ijo to Berbice Dutch (BD) not in the (initial) process of “creolization” of Dutch in Berbice but in a second-stage post-formative process of the mixing of the Creole formed before 1713 with Ijo. Using the terminology of the science of complexity and emergence, the important “attractor” in this putatively unusual history was the arrival of the slave ship Sint Antony Galeij the same year, which “brought the right speakers to the right place at exactly the right time”. The ship “disembarked 403 slaves in Berbice” most of whom originated in Calabar. The newcomers apparently outnumbered the Creole and seasoned slaves who had preceded them, as the latter’s numbers were reduced to “give or take, 300” from 500 to 600 after a French raid of the colony in 1712 ended with 259 of them taken away. The colony had been founded in 1627. Although no information is given about the colonial population structure since the foundation to 1712, especially regarding the demographic sizes of the “plantations” and the general ratio of Europeans to Africans, I have no reason to doubt that a creole may have emerged already by then. Offhand, the hypothesis of the presence of a creolized Dutch spoken in Berbice prior to the 1713 arrival of the Ijos, assumed to be the majority of the Sint Antony Galeij’s human cargo, is plausible.

P&J use the above scenario to answer the question of whether BD, whose structures are detailed in Kouwenberg (1994), is a creole. Some creolists, including Mufwene (1993, overlooked by the authors), had asked it based on how the peopling of the colony was reported by Smith et al. (1987) as not involving extensive multilingualism. The alternative scenario in which Eastern Ijo speakers must still have been the overwhelming majority could, according to P&J, have involved primarily the contact of the already creolized Dutch with Eastern Ijo since 1713.

Our knowledge about the emergence of several creoles has evolved significantly since the 1980s. For instance, grammatical features of Palenquero have been claimed to reflect an almost exclusive substrate influence of Kikongo.^[1] Various indigenized Asian varieties of Portuguese claimed to be creoles are not the outcomes of extensive multilingualism either.^[2] I provide all this information because we must return to the question of what kinds of languages “creoles” are, whether there are structural features specific to them, and what the putative process of “creolization” really means.

As suggested above, ‘creolization’ can be defined as an evolutionary process by which a language, or a culture for that matter, becomes creole. This explanation defines a process by its outcome and presupposes that one already knows what the nature of the outcome is. It is like speaking of liquefying with the meaning of ‘turning into a liquid’. As I explain below, the notion of creole itself has become more and more elusive. I can say that the longer I work on the relevant language varieties, the harder it is to explain to my own satisfaction what they are without writing a whole paper. For language varieties identified with this name, is there a particular restructuring process, or a combination of specific structural changes, that can be characterized as creolization?

P&J capitalize on the loss of grammatical morphemes, or let’s call it ‘morphological impoverishment’. However, this is not unique to creoles. Compared to Vulgar Latin, the Romance languages have impoverished inflectional morphology, which actually prompted Schlieben-Lange (1977) to hypothesize that they may be characterized as creoles. Contradicting her, Posner (1985, 1996 and Trask (1996) argued instead that Romance creoles should be considered as the newest Romance languages. Chaudenson (1992, 2001 argued that French creoles have simply extended the morphological impoverishment that was already evident in the emergence of the Romance languages themselves and can still be observed in current français populaires, the actual lexifiers of the creoles. Based on some other studies, such as Klingler (2003), the Africans brought order out of chaos in the distribution of and irregularities in the remaining grammatical morphemes, by just omitting these. There is a kind of uniformitarian appeal to this anti-exceptionalist position, because the Romance languages are also outcomes of language contact.

For a similar reason Bailey and Maroldt (1977) also argued that Middle English was a creole, as it was likewise the outcome of language contact. More recently, Emonds and Faarlund (2014) argued that (Middle) English was the outcome of language contact as it owes a substantial part of its grammar to Old Norse. These include: the use of auxiliary modals such as will and shall to express future tense; double modal constructions such as should may (antecedents of might could and the like in, for instance, American Southern English); the use of have + past participle for past tense in nonfinite clauses, as in will have served; and stranded prepositions in relative clauses and questions, as in the topic that Mary wrote about. If there is such a thing as Germanic structural peculiarities, Modern English is also inflectionally impoverished compared to Old English and even Modern German. It has also lost the V2 pattern of other Germanic languages, having adopted a strict SVO structure in affirmative sentences. In Mufwene (2005), I argue that if we follow this line of reasoning, virtually every modern language must be stipulated as a creole, since they all owe some of their current structures to contact. The term creole would thus become synonymous with language.

If P&J want to be convincing about their two-phase scenario of the emergence of BD, they must be more explicit about what ‘creole structural features’ or particular restructuring processes must be identified as creole. McWhorter (1998) proposed that creoles can be defined on the prototypical model based on the absence of tones and loss or impoverishment of inflectional and derivational morphology. This hypothesis was a stillborn brainchild, because the vast majority of language varieties identified as creoles would be more or less creoles; they fail to match the stipulated stereotype in one respect or another. Chaudenson (2001) argued that the absence of tones was not an impressive feature, because the European lexifiers themselves are toneless. DeGraff (2001) showed that Haitian Creole not only has retained numerous derivational morphemes from French but has also been quite creative in producing new ones. Clements (2022) shows that Asian Portuguese creoles have some inflectional and derivational morphemes too. In other words, P&J appear to be toying with a vacuous speculation, similar to a mathematical ‘vacuous truth’ that cannot be satisfied. These observations support my remark that it is becoming more elusive to me to explain what creoles are, especially based on structural features. So, the process of ‘creolization’ itself becomes vacuous if it is defined by an outcome that is both elusive and illusive.

I also wonder whether the two-phase scenario conjectured for the emergence of BD is fundamentally different from that of gradual creole formation hypothesized primarily by Chaudenson (1979, 1992, 2001 and Arends (1989), based on their analyses of documentary evidence and sociohistorical information. In the case of French creoles, the information included the emergence of plantation societies from the homestead communities and the pattern of population growth. Chaudenson then hypothesized that creoles emerged by gradual basilectalization, i.e., structural divergence farther and farther away from the lexifier. An issue brought to my attention recently by Enoch Aboh (p.c., Sept. 2022) is that the term basilect, which is implicit in this hypothesis, is defined relative to acrolectal varieties of the relevant lexifiers, while the actual lexifiers of creoles were nonstandard dialects of European languages. In fact, reality tells us that they consist of continua that are not the outcomes of what has been characterized since Schuchardt (1914), in modern terminology, as ‘decreolization’. Contrary to what P&J claim in their conclusions, the conjecture cannot be “falsified” not only because there are no data but also because their assumptions about what creoles are remain question-begging.

To be sure, P&J’s purpose is to show that BD straddles between being identified as a creole and being claimed as a mixed language, because, according to them, in the latter situation only two principal languages were involved: BD and Eastern Ijo. One must assume that access to Dutch ended in 1713 or soon thereafter and also that the putative minority of non-Ijos that came on the Sint Antony Galeij had no role, or just a negligible one, to play in the (re)shaping of BD through the putative mixing. This is like saying that the Dutch had no role to play in the emergence of Sranan, while it should be evident that the Dutch in Surinam were not native speakers of English, which they used to communicate with the Africans. Baker (1993) had exposed the inadequacy of this kind of approach regarding the exclusive role that, according to relexificationists, Ewe speakers must have played in shaping the structures of Haitian Creole. It’s surprising that an issue that seemed resolved is resurrected with the apparent motive of simplifying explanations of evolutions that are otherwise naturally complex.

P&J make a few allusions to “intertwined languages” as representative of “mixed languages”. But they also point out that the structures of these languages have not evolved in identical ways. They have no common structural template; they are identified as such because only two ethnolinguistic groups were involved in the contacts, which were intimate, if we do not count BD. For example, Media Lengua, (Ki)Ma’a, and Michif do not have a common morphosyntactic pattern exclusive to them, no more than creoles would have shown to some extent if their lexifiers had not been Indo-European. Moreover, it’s not evident that the intimacy of contacts was the same in these cases. If P&J insist on grouping them together, perhaps they should consider including, say, Irish English in the same waste basket, because it is the outcome of the contact of two languages.

And it is not obvious why intertwined languages are claimed to fit in the category of mixed languages while creoles are not. To wit, Thomason (2001) includes creoles in this category whereas Meakins and Stewart (2022) exclude them, although, historically, it is the latter that prompted linguists in the late 19th and early 20th centuries to posit the category of mixed languages. The purpose was to justify by fiat the disenfranchisement of creoles from the club of “normal languages”. In note 13, P&J claim that “creoles can be said to be ‘mixed’ insofar as the lexifier vocabulary is combined with a grammar of other origin, but this ‘other’ does not equal any previously existing language(s)” (this issue, p. 184). This is actually close to the way Thomason and Kaufman (1988) define creoles and conclude that they cannot be assigned to a particular genetic/genealogical language family. (Note that this conclusion raises issues about how the authors can explain the actuation of the speciation of Indo-European languages, which was a consequence of population movements and language contacts.) My concern about P&J’s statement regards how the reader must interpret “but this ‘other’ does not equal any previously existing language(s)”. What about all the polemics creolists have had about the significance of substrate influence on the structures of creoles? The fact is that, for instance, Michif is no more identical with Cree than French is.

To buttress their position, P&J identify Eastern Ijo as an adstrate, as opposed to other African languages (whichever they were) that were brought to Berbice before 1713 and are presumably more legitimate substrates. Literally, since Africans are not indigenous to the Americas, the term substrate has been applied to their languages from a social perspective, which makes irrelevant the distinction between adstrate and superstrate, because all Africans were subordinated to European colonists (other than the indentured servants). Noteworthy is also the fact that the peopling of the colonies where creoles emerged was incremental. The different ethnolinguistic groups did not all arrive at the same time, and any one of them, under favorable demographic conditions, could have introduced new changes to the current system. Chaudenson’s basilectalization hypothesis captured this evolution.

By the way, would the P&J scenario not apply to, say, French too? Vulgar Latin started to change because of its contact with the Gaulish languages and then underwent the influence of Frankish. This was the point of Schuchardt’s (1882, 1884 publications. Almost the same can perhaps apply to Asian Portuguese “creoles”, which emerged first out of the close interactions of the Portuguese merchant colonists with specific indigenous populations speaking the same respective languages and diverged farther away from heritage Portuguese after Portugal lost its colonies to other European colonizers. An important difference of course is that the earlier stage would, I suppose, involve language mixing according to P&J’s explanations and the post-Portuguese-rule period involves perhaps another layer of mixing with some other indigenous language.

What does all this make of the Thomason and Kaufman (1988) typology of language contact that P&J claim remains unaffected by their conjecture? Or are they just trying to save distinctions made in the 19th century when European philologists (with the exception of Hugo Schuchard and Adolfo Coelho) assumed that non-Europeans were evolutionarily too inferior and spoke languages too simple to be able to learn the sophistications of European languages? The difference is that there are no more explicit invocations of such explanations today. Only the categories have been maintained. I wonder what the reader can learn from historical language evolution (as opposed to the protracted phylogenetic emergence of languages) from P&J’s article in the absence of structural data from earlier stages of BD and of more information about the growth of the Berbice colony. To be charitable, the answer may come from more careful studies of creoles whose evolutions are better documented (notwithstanding the usual colonial bias in the documentation).

2 Other concerns

P&J also claim that “the rather strict svo word order of the language, despite both Dutch and Ijo being (at least underlyingly) sov, has been claimed to show not only that svo is a creole universal, but that it is a universally more unmarked order altogether” (this issue, p. 181). To begin with, according to Kouwenberg (1994), BD is not strictly SVO; it is partly SOV too. Keesing (1988) showed convincingly that the reason why Melanesian pidgins have SVO word order is that Melanesian indigenous languages are agglutinating. In specific contexts in which the sentence arguments are known and need not be repeated, utterances consist only of the verb complex, which includes affixes specifying the syntactic roles of the arguments in an SVO order, although the order of major constituents in the full sentence is VSO. Dutch has an SVO order in the basic nuclear sentence. Since some of its sentences can also begin with a non-subject argument, it is a language with a mixed word order.^[3] Creoles have a dominant SVO order because most of their European lexifiers have this basic word order. Claims of the universality and unmarked status of SVO are biased if not unwarranted when the facts from some of the languages in contact can explain what is going on. Markedness values are specified in specific interactional ecologies, which involve complex dynamics at the population level. Since there are alternative solutions to the same problem, different language communities wind up producing different patterns; some of them may even be competing ones. They are not universal, especially because there is no universal major constituent order for sentences.

I was not aware that “both [creoles and intertwined languages] are often thought to emerge rather abruptly, in the space of only one or two decades” (this issue, p. 182; my emphasis). In fact, the literature had discussed their evolutions in terms of generations (e.g., Bickerton 1981; Chaudenson 1992; Holm 1988). Also, as I pointed out in Mufwene (2001), a rapid evolution is not synonymous with an abrupt one. There is room for gradual evolution even within the two-decade window of time that P&J claim, especially since financial factors bore significantly on how fast the labor force grew either in the initial homesteads or the emergent plantations. Even in settings such as Hawaii, where there may have been no homestead phase worthwhile discussing, the emergence of Creole may not have been that abrupt.

Besides, as can be inferred from the colonization of Jamaica and South Carolina, for instance, shortage of financial capital delayed the emergence of large plantations for 40–50 years, thus about two generations (Dunn 1972; Wood 1974). Although they both were second-generation colonies, settled from Barbados, their agricultural industries did not start as fast as one may expect, contrary to Chaudenson’s (1979, 2001 claim that such colonies benefited linguistically and infrastructure-wise from the experiences of their mother colonies. Until the end of the 17th century, soon after a 1692 earthquake destroyed Port Royal, the headquarter of pirates and employers of a large proportion of settlers in Jamaica, the dominant European colonial population resided on small farms dispersed all over the Island. During the first 50 years most of the English colonists in South Carolina survived on especially fur trade, which did not need much labor force, with Native Americans.

P&J do not mention when the sugarcane plantation economy started in Berbice and when the expected residential segregation between the Europeans and Africans, which is critical to the emergence of creoles, began. This is relevant to their conjecture, because absence of a segregated population structure is part of the explanation for why Brazil has not produced a creole although it engaged in sugarcane cultivation a century earlier than the Caribbean and imported many more slaves than all the latter colonies. Another reason is that, although Brazil produced more sugar than all of the Caribbean colonies in the 17th century, its sugar production system was organized differently, based on farms (rather than plantations). Farmers teamed up for harvesting and the distillation of sugar (Schwartz 1985). Moreover, the demographic disproportion between Europeans and Africans was not as pronounced in Brazil as in the Caribbean.

The latter factor also explains why in the United States Gullah emerged only in coastal South Carolina and Georgia, where rice was cultivated, but not in the hinterlands, where the cultivation of cotton and tobacco relied on mostly European indentured servants and smaller populations of Africans. The lesson is that all evolution is constrained by the local socioeconomic ecology, including its history. Today, we should refrain from generalizing too hastily on the putative universality of a contact-induced restructuring process just to maintain language categories that were ill-defined in the late 19th century owing to incomplete information and some colonial prejudices toward the non-European speakers of European languages.

I’m equally surprised by the claim that in the case of creoles “the grammar (with the occasional exception of more or less isolated features), is not clearly traceable to any of the (alleged) input languages” (this issue, p. 183). In the case of Caribbean English creoles and Gullah, for instance, the major constituent order in a sentence and the positions of the article, quantifiers, and adjectives have been selected from the lexifier. Moreover, the notions of adjective and article can definitely be traced back more to the lexifier than to many of the substrate languages which lack these grammatical categories and use other strategies to modify the noun and to express definiteness. Much of the structures of relative clauses and interrogative sentences (minus Subject-Auxiliary Inversion) are from the lexifier, taking into account the stranded preposition at the end of the clause. Serial verb constructions, which once figured prominently in discussions of the origins of grammatical features, have been attributed to substrate languages. What are the “more or less isolated features” of creoles that P&J are referring to? Aboh (2006, 2015 has illustrated clearly the extent to which some features of the lexifier and some substrate languages have “hybridized” in the case of Haitian Creole. Corne (1999) underscored the role of (partial) structural congruence in the selection of grammatical patterns in the emergence of creoles.

I remain unconvinced that there are “typical features of both creoles and intertwined languages” to the exclusion of other languages. I argued in Mufwene (1986) that some of the features claimed to be creole are more consistently attested in the grammars of some languages such as Chinese (e.g., number delimitation and serial verb constructions) than in creoles, whose structural features are not monolithic. Things are more complicated when creoles are not reduced to idealized basilects and their mesolects, which are the more common in practice (Rickford 1990), are included. And there are indeed structural differences among those lexified by the same language. For instance, Gullah and Guyanese have a dedicated habitual marker, respectively, /dəz/ and /d⊃z/, while most, if not all, others do not. In Jamaican Creole, the plural marker dem follows the noun while the definite article precedes the latter, whereas in Gullah both are conflated into a prenominal dem. Serial verb constructions vary among themselves; several creoles do not have one that expresses a dative function. More detailed comparative analyses may reveal more inter-creole differences, such as whether a variety uses an indefinite article or the quantifier ‘one’ instead. To be sure, even French itself can be confusing in this respect, except when the delimited noun is feminine.

Such facts make the notion of ‘creoloid feature’ problematic, even if one assumed that there are some ‘prototypical creoles’ somewhere. And what is a “process of contact-induced change”? I didn’t know that the phrase “contact-induced change” (which, by the way, can be found in any modern language) denoted a specific kind of restructuring process, as suggested in the article. P&J allege that “Indo-Portuguese […] was once less mixed, resembling more ‘traditional’ creoles” (this issue, p. 184). It is not clear what quantification method or yardstick they have used for this determination.

The article contains several other inaccurate or questionable statements that constitute a major distraction from P&J’s main thesis discussed in the first part of this rejoinder. All the above comments concern various claims in sections 1–2.2. My conclusion is that while the authors’ speculations have prompted me to discuss some issues about what should make creoles and intertwined languages more special than any other language in historical evolutionary linguistics, the authors have not made a convincing case for the position they set out to defend, namely that BD emerged in two phases: ‘creolization’ followed by a mixing of two languages. The claim is complicated by the fact that, if they think structural considerations prevail over sociohistorical ones in identifying a language as a creole or a mixed one, the latter category does not have structural features shared universally by its members, no more than creoles do. No one would ignore the significance of structural features in discussing language change and speciation, which is true of creoles. Sociohistorical information (overlooked in traditional genetic linguistics) helps us determine what specific socioeconomic and demographic ecological factors influenced the emergence of specific features out of the contacts of the relevant languages. It is incautious to underestimate the complexity of the dynamics that produced modern language varieties, regardless of whether they are identified as creoles, pidgins, intertwined languages, or “normal” languages.