Language processing in bilingual children and adults : evidence from filler-gap dependencies and garden path sentences

Abstract

The present thesis examines morphosyntactic processing in bilingual children and adults. It bridges gaps in the existing literature in three ways. Firstly, previous work on bilingual children has focused on inflectional morphology but has not examined the timecourse of processing in terms of misinterpretation and real time reanalysis or use of information of different sources to aid disambiguation. Second, it extends the use of the visual world eye-tracking paradigm to research in morphosyntactic processing in bilingual children. Third, for the adults, it compares early/native bilinguals to monolinguals and late bilinguals / L2 learners. Two linguistic phenomena were investigated with adults and children; which-questions and garden-path sentences in English. Overall, both bilingual children and adults showed qualitatively similar patterns of processing to their monolingual counterparts. All groups experienced greater difficulty with structures where there was ambiguity and a need for syntactic reanalysis, i.e. object which-questions and garden-path sentences, suggesting incremental processing. The main difference between monolinguals and bilinguals is that of speed; bilinguals appeared to process sentences slower than monolinguals even when their comprehension accuracy was equally as high. This difference was found for both children and adults, as evidenced by the reaction times or changes in the gaze data, and was generally not more pronounced in sentences where reanalysis is required. With regards to the bilingual adults, the early/native bilinguals clustered with the L2 learners in terms of processing speed but more so with the monolingual adults in terms of accuracy. The bilingual groups showed a reduced utilisation of information from various sources to aid processing. Bilingual adults and children made use of number mismatch between the two noun phrases in the study on which-questions to facilitate disambiguation; however, they showed this effect just for off-line comprehension accuracy and not for real time processing, i.e. in the gaze data. The bilinguals also did not show consistent use of referential context to disambiguate in the study on garden path sentences, although this was also the case for the monolingual adults and children. In sum, the results from the studies in this thesis suggest the both bilingual adults and children were equally as able as their monolingual counterparts at an end stage but differed to the monolinguals on more fine-grained measures of real time processing. These measures point to qualitatively similar but more protracted over time processing for bilinguals and with more limited use of facilitatory information to disambiguate.