Shallow vs. deep : bilingual contrast processing
Katherine A. Yoshida · cIRcle (University of British Columbia) · 2009
Languages differ in the range of phones that cluster together to carry the same meaning, forming phonemic categories (e.g. all the possible manifestations of /l/ in English). These phonemes then contrast with one another to denote differences of meaning (e.g. /l/ vs. /r/, a valid contrast in English but not in Japanese). Infants are born with the ability to discriminate contrasts belonging to their native language, as well as those that are non-native contrasts. However, by 10-12 months of age they refine their repertoire, leaving only those utilized in the language input (Werker & Tees, 1984). The subsequent formation of non-native phonemes belonging to an ensuing second language (L2) is less clearly understood. This work uses Cantonese-native ESL speakers to investigate the distinction between "shallow" knowledge of non-native (Englishspecific) isolated phonemes and "deep" knowledge of the phonemes embedded within a linguistic context (Werker et al, 2001). Deep knowledge of the non-native contrasts is tested using a lexical decision task (Pallier, Colomé & Sebastián-Gallés, 2001) and compared to shallow knowledge, operationalized as ability to perceive differences between pairs of isolated phones. Results indicated that the bilinguals had difficulty perceiving the English-specific contrasts within the deep context, but not within the shallow context. Thus, support is found for a shallow-deep distinction within L2 speakers of a language. This helps explain previous divergent findings of L2 perception, and may potentially have implications for hypotheses of interference and a critical period.