Search Results (3)

The Perceptual Assimilation Model (PAM)-L2 framework posits that the discriminability of L2 speech contrasts can be predicted from how L2 phones are assimilated to L1 categories. This study examined how such assimilation types relate to variability in L2 vowel discrimination within the PAM-L2 framework, with particular attention to assimilation overlap. Chinese listeners were tested with six English vowel contrasts (/i-ɪ/, /e-ɛ/, /æ-ɛ/, /ɑ-ɔ/, /ɔ-ʌ/, /u-ʊ/) using an assimilation task with goodness ratings and an AXB discrimination task. The vowel contrasts formed three Uncategorized-Categorized and two Uncategorized-Uncategorized contrasts, with both types exhibiting either partial or complete overlap, along with one Category-Goodness contrast. Discrimination results showed that partial versus complete overlap accounted for some differences in discrimination accuracy and absence of overlap between dominant L1 response categories likely facilitated discrimination even when secondary categories overlapped. Large differences in perceived goodness appeared to facilitate discrimination for a vowel contrast with complete overlap. The results are discussed in particular relation to the PAM-L2 account of the assimilation overlap-discrimination relationship, and additionally how the influence of overlap may be modulated by category-goodness differences in contributing to variability in L2 vowel discrimination. Full article

The present study examined the predictions the Perceptual Assimilation Model in the context of naïve bilingual speakers while also considering whether the predictions of third language (L3) models of morphosyntax could be extended to L3 phonology. In particular, it was asked whether several groups of Spanish-English bilinguals (constituting a fully combined design) would categorize sounds in two unknown languages, French and German, using their first language (L1) categories, second language (L2) categories, or a combination of both. 199 participants took part in the study, who made up 4 total groups: L1 English–L2 Spanish (n = 55), L1 Spanish–L2 English (n = 59), English monolingual (n = 59) and Spanish monolingual (n = 29). The participants completed a vowel categorization task, where they were asked to match four vowel sounds in French and German to their existing English and Spanish categories. The results of a series of Bayesian Multinomial regression models suggested that bilinguals categorize L3 vowels using both L1 and L2 categories according to the acoustics of the input. There was no evidence of a clear bias for either the L1 or L2 when an L3 vowel exists in both the L1 and L2. Additionally, the bilingual English participants differed from English monolinguals in the their categorization of new language sounds. These results have implications for both the PAM-L2 and L3 models, by showing that the language learners are not solely guided by their native language, and have access to both L1 and L2 categories when accounting for novel language sounds. Full article

Japanese learners of English can acquire /r/ and /l/, but discrimination accuracy rarely reaches native speaker levels. How do L2 learners develop phonological categories to acquire a vocabulary when they cannot reliably tell them apart? This study aimed to test the possibility that learners establish new L2 categories but perceive phonological overlap between them when they perceive an L2 phone. That is, they perceive it to be an instance of more than one of their L2 phonological categories. If so, improvements in discrimination accuracy with L2 experience should correspond to a reduction in overlap. Japanese native speakers differing in English L2 immersion, and native English speakers, completed a forced category goodness rating task, where they rated the goodness of fit of an auditory stimulus to an English phonological category label. The auditory stimuli were 10 steps of a synthetic /r/–/l/ continuum, plus /w/ and /j/, and the category labels were L, R, W, and Y. Less experienced Japanese participants rated steps at the /l/-end of the continuum as equally good versions of /l/ and /r/, but steps at the /r/-end were rated as better versions of /r/ than /l/. For those with more than 2 years of immersion, there was a separation of goodness ratings at both ends of the continuum, but the separation was smaller than it was for the native English speakers. Thus, L2 listeners appear to perceive a phonological overlap between /r/ and /l/. Their performance on the task also accounted for their responses on /r/–/l/ identification and AXB discrimination tasks. As perceived phonological overlap appears to improve with immersion experience, assessing category overlap may be useful for tracking L2 phonological development. Full article