The effects of syntactic priming on the L2 production of relative clauses in English by Japanese-English bilinguals.

摘要

Recently, researchers have begun to use syntactic priming to investigate the mental representation of the two languages of bilinguals and their accessibility during comprehension and production (e.g. Bernolet, et al., 2007). In the syntactic priming technique, a priming sentence is presented to elicit a structurally similar target response, when an alternative structure is also available. Previous studies, which focused mainly on the L2 syntactic priming effects of dative structures, found priming effects when the target L2 structure is similar to the L1 but not when they are dissimilar (e.g. Branigan, 2007).;The present study investigated priming effects on the English L2 production of relative clauses by 18 Advanced adult Japanese-English bilingual speakers, using a modified version of Bock's (1986) sentence-repetition and picture-description task. It was predicted that the cross-linguistic differences between Japanese and English could potentially influence their L2 production. Experiment 1 comprised three prime-types (Full-relatives, Reduced-relatives, and simple actives) in English. Experiment 2 contained two Prime-types (Full-relatives and simple actives) in Japanese. In both experiments, the target responses were in English. Repeated-measures two-factor ANOVAs was used to test for Main and Interaction effects of Prime Type and Response Type on the mean proportion of responses in each Response Type category.;The results of both Experiment 1 (L2-to-L2) and Experiment 2 (L1-to-L2) indicated a significant main effect of Response Type; in both experiments, the participants preferred Simple Actives for picture-description regardless of Prime type. As for the interaction between Prime Type and Response Type, significant results were obtained only in the case of Experiment 1, where the directionality was from L2 prime to L2 response, but not in the case of Experiment 2 (Japanese-to-English). Specifically, in Experiment 1, the participants used Reduced-Relatives more often in the Reduced-Relative clause priming conditions than in the Full-Relatives and the Simple Active condition. The findings bolster the position that the L1 mental representation and the L2 mental representation of complex structures are separate, particularly when a bilingual speaker's two languages are typologically different, as in the case of Japanese and English. The findings also indicate that the L1 mental representation is not accessed during L2 production.