Convergent developments in Dutch Turkish word order – A comparative study using ‘elicited production’ and ‘judgment’ data: Converging evidence?

Abstract

Language is a very dynamic entity, so language change is inevitable, especially in contact settings. Most contact studies are based on recordings of spontaneous speech. Useful as this has been, such data cannot tell us everything: for one thing, they do not tell us much about the degree to which the attested changes are entrenched in bilinguals' linguistic competence, and whether the changes are manifested mostly in bilingual speech, i.e. when the speakers are codeswitching at the same time. This paper aims to look for converging evidence, a fairly recent notion (Schönefeld, 2011:1), by employing two experimental techniques to explore a single issue: Dutch influence on the word order of immigrant Turkish in the Netherlands. More specifically, the focus is on the position of the matrix verb in complex clauses. Dutch and Turkish differ in their word orders, specifically in the position of the verb. Although Turkish can be claimed to have a relatively free word order, it is canonically verb-final. Dutch, on the other hand, is a verb-medial language, at least in main clauses. This study discusses findings as to whether immigrant Turkish verb positioning differs from TR-Turkish practice, and whether the differences are more pronounced when speakers are in a bilingual mode. It studies data from three groups of participants; 1) Bilinguals in bilingual mode, 2) Bilinguals in monolingual mode and 3) Monolinguals. One method employs elicited ‘production’ data reflecting bilingual ‘performance’ and the other one ‘judgments’ reflecting ‘comprehension’ or ‘linguistic knowledge’ (i.e. converging evidence). Unconventionalities, changes in preferences and frequency differences in these data are interpreted subtypes of a contact-induced change in this paper. Interestingly, the data from the two different sources did not converge completely: for the structures caught up in change, the ‘judgment’ data showed full acceptability of both the changed and the non/changed structures, while the ‘production’ data showed a clear preference for the former. The paper finally focuses on the question what kind of change mechanism may be responsible for the changes. The findings point to restructuring (Heine & Kuteva), which overlaps with Johanson's frequential copying and Matras' ‘other outcomes of pivot matching’. In conclusion, the results show that word order in Dutch Turkish is changing.