Predicting discrimination difficulty of Californian English vowel contrasts from L2-to-L1 categorization

Abstract

L1-Spanish-Catalan speakers often face difficulties when learning the English vowel system. PAM/PAM-L2 posits that native language experience shapes perception of L2 sounds, so discrimination of L2 contrasts can be predicted from L2-to-L1 categorization. The current study focused on whether discrimination of the nine Californian English vowel contrasts /i/-/ɪ/, /ɛ/-/ɪ/, /ɛ/-/ə/, /æ/-/ʌ/, /æ/-/ɑ/, /ɑ/-/ʌ/, /ə/-/ʌ/, /ə/-/ʊ/, and /u/-/ʊ/ can be accounted for by the perceptual assimilation of these vowels to L1-Catalan categories. Experienced L1-Spanish-Catalan learners of English performed categorization with goodness rating and AXB discrimination tasks. The vowels /i/, /u/, /æ/, and /ɑ/ were categorized, while /ɪ/, /ɛ/, /ʌ/, /ə/, /ɝ/, and /ʊ/ were not consistently mapped onto a single Majorcan Catalan category. Discrimination of the vowel contrasts /i/-/ɪ/, /u/-/ʊ/, /ɛ/-/ə/, and /ʊ/-/ə/ was very high (>90%). The contrasts /ɪ/-/ɛ/, /æ/-/ɑ/, and /æ/-/ʌ/ were also well discriminated (>80%). In contrast, /ʌ/-/ɑ/ and /ʌ/-/ə/ were discriminated less accurately. Percent overlap scores predicted discrimination accuracy at the group level. At the individual level, discrimination could only be predicted for the contrasts /i/-/ɪ/, /ɪ/-/ɛ/, /u/-/ʊ/, and /ɛ/-/ə/. The results are discussed in terms of three L2 speech learning models, PAM-L2, SLM-r and L2LP.