Abstract/Summary

The two studies in this thesis examine the activation of the L1 during the processing of L2 collocations. Models of bilingual lexical representation and access such as the Bilingual Activation Model (Dijkstra & Van Heuven, 2002) and the Multilink Model Dijkstra and Rekke (2010) posit that non-selective, cross-lingual activation is a feature of the bilingual lexicon. Following from increased research in formulaic language, this study uses online processing measures to investigate whether cross-lingual activation can be extended beyond single lexical items to collocations in bilingual children. A self-paced reading (Study 1) and an eye tracking (Study 2) experiment were conducted with Tamil-English bilingual children (age 8-11). In both studies, reading times on English collocations embedded in sentences were measured. All collocations were congruent or incongruent with collocations in Tamil. Study 1 with 58 participants was conducted in India and Study 2 with 80 participants was conducted in the UK and participants across the two studies varied substantially in their English and Tamil proficiency. All children also completed vocabulary tests (English & Tamil) and children in Study 2 also completed a questionnaire on their language background. Results clearly show that children rely on their vocabulary knowledge in L1 to aid their processing of collocations in L2 and that this L1 activation is immediate and can be captured in real time. An important finding from these studies is that the extent of L1 activation is dependent on L1 vocabulary as well as language exposure and dominance: the children in Study 1 showed a much bigger congruency effect than the children in Study 2. These findings indicate that L1 activation in bilingual children is not just for single words but goes beyond word level to collocations as well.