Ezeizabarrena, M. J. & Munarriz, A. .
University of the Basque Country
Different proposals have been made with regard to the syntactic constraints that may regulate a) language choice and b) placement of the linguistic components in mixed utterances. Some models distinguish between the languages available to the bilingual in terms of strong(est)/ weak(est) language, matrix/embedded language or the language that provides /receives functional elements as a possible explanation (prediction) of the source of items contained in mixed utterances. Evidence for that hierarchy has been found in early and late unbalanced bilingualism, speech pathologies etc. In contrast, models which consider mixed utterances as the result of one computational system working with two lexicons predict that it is the head of the phrase what will decide the placement of its complement.
The present study analyses mixed utterances containing elements of B(asque) and Sp(anish), two languages which differ with regard to the head directionality parameter —head final (B)/ initial (Sp)— and to the morphology of some functional heads—bound (B) vs. free (Sp) . The analysis of longitudinal and cross-sectional corpora of spontaneous or elicited production of mixing by early bilingual children (2L1 and L2) and by language impaired adults revealed the existence of a similar pattern of language mixing across corpora: mixed utterances mostly contain complex full XPs in the other language (B or Sp) as complement or adjunct of Sp or B heads respectively. In addition, head-complement word order in mixed DPs corresponds to the language providing the head, head initial in DetSpNPB phrases and head final in NPSpDetB mixed DPs though no clear preference is observed for any of the combinations, what contrasts with predictions based in the feature specificity of the head. Finally, the overall scarcity of mixed phrases suggest that the morphosyntactic properties of the languages in contact play an important role in the typology of mixed utterances.