Interplay of native and non-native vowels in Japanese late learners of English

This study investigates perception of native (/i/-/e/) and non-native (/i/-/ɪ/) contrasts by Japanese late learners of English (N = 40). We hypothesized that speakers with greater proficiency in L2 might show more effects of L2 learning on L1 perception than speakers with lower L2 proficiency. We found first that self-rated proficiency correlated relatively well with L2 vowel contrast categorization (r = 0.622) indicating that late bilinguals can achieve more nativelike categorization of L2 vowels (/i/-/ɪ/) as their overall proficiency improves. Turning to the effect of L2 proficiency on L1 perception, we found that the bilinguals who exhibited more nativelike English /i/-/ɪ/ categorization in their L2 were more likely to have a broader /i/ category when identifying /i/vs./e/ in Japanese (compared to monolingual Japanese controls). We argue that while L2 English learners may be improving in identifying English /i/vs./ɪ/, the non-native /ɪ/ is not fully differentiated from their Japanese /i/ category, possibly providing atypical exemplars that influence the perceptual boundary of /i/ in their L1. The complex interplay of L1 and L2 vowels will be further discussed by reference to production data for the learners and native speaker controls for Japanese and English.