On Berbice Dutch VO status

Abstract

Berbice Dutch, once the vernacular of the Dutch-owned Berbice and Canje plantation areas of what is now Guyana (South America), is a VO language, even though both its substrate (Ijo, the Kalahari variety in particular) and its superstrate (16th and 17th century Dutch) are taken to be OV (and Verb Second (V2)) languages. Along the lines of Bickerton's Bioprogram hypothesis (Bickerton, 1984, et seq), some universalist analyses have taken Berbice Dutch to be a perfect illustration of OV markedness. Following Kayne's Lexical Correspondence Axiom that states that every language exhibits an underlying VO order and that OV orders are the result of additional object shifts (cf. Kayne, 1994), many scholars have taken OV languages to be syntactically more complex than VO languages. The emergence of Berbice Dutch VO would then be a reflex of a reduction of the complexity of a creole language in comparison to its substrate and superstrate languages (cf. Roberts, 1999). In this article we argue that Kalahari, unlike previous claims, is actually not a V2 language and that therefore the Berbice Dutch substrate language differs from its superstrate in not allowing surface VO structures. This, we subsequently argue, opens up the way to analyze the emergence of Berbice Dutch VO as a result of the interplay between first and second language acquisition. We conclude that VO word orders are actually likely to emerge in the Berbice Dutch contact situation and that, therefore, Berbice Dutch VO status does not constitute evidence in favor of Bickerton's universalism or similar approaches (in line with Mufwene's, 2001 and DeGraff's, 2001, 2003 arguments against creole exceptionalism), nor in favor of an alleged universal VO base order. In short, Berbice Dutch VO may directly result from the, fact that Kalahari speakers would not recognize Dutch V2, as their native language lacked it, and that therefore they analyzed SVO orders resulting from V2 as plain VO orders. Children growing up at the plantation would then take this input as evidence for a VO target language. We furthermore argue that such a reanalysis of OV + V2 structures as VO structures by children, both in main and embedded clauses, got facilitated by the existence of so-called VO leakages in 16th and 17th century Dutch (along the lines of Weerman, 1993). (C) 2016 Elsevier Ltd. All rights reserved.