Age of L2 acquisition effects on Korean-English bilinguals' perception of phonetic contrasts in their two languages

Jessica Maye, Northwestern University
Yubin Lee, Northwestern University
Jenna Luque, Northwestern University
Midam Kim, Northwestern University
Thomas Farmer, Cornell University

Introduction: Korean speakers are known to have difficulty discriminating English /r/-/l/, while English speakers have trouble discriminating Korean plain vs. aspirated stop consonants. Studies of speech perception in a second language have reported mixed results, with some suggesting very strong effects of age of acquisition (AOA), others suggesting that speech perception remains malleable later in life. We tested Korean-English bilinguals' perception of these difficult phonetic contrasts to examine the effects of AOA and language dominance on bilinguals' perception in their two languages.

Participants: All bilingual participants were native Korean speakers, but varied in AOA for English. Early bilinguals (AOA < 8 yrs) were English-dominant, while Late bilinguals (AOA > 15 yrs) were Korean-dominant. Participants completed a 2AFC word identification task in each language. A set of English monolinguals completed the same task in English only.

Method: On each trial participants were presented with two pictures and were asked to click on one of the two pictures. On key trials the pictures formed a minimal pair (e.g. rock vs. lock). On control trials the two pictures did not form a minimal pair (e.g. doll vs. school). The same task was completed once in English (key items contained the /r/-/l/ contrast) and once in Korean (key items contained the plain vs. aspirated voicing contrast).

English Results: For English monolinguals, performance on the /r/-/l/ pairs did not differ from control pairs. However, for bilingual participants, performance on the /r/-/l/ pairs was modulated by age of exposure to English, with earlier exposure associated with greater accuracy. Early bilinguals (exposed to English before age 8) did not differ from English monolinguals in their accuracy on any of the test items. In contrast, although late bilinguals (exposed to English after the age of 16) performed as well as English monolinguals on control items, they performed significantly more poorly on the /r/-/l/ items compared to monolinguals (p < .05).

Korean Results: We were unable to test true Korean monolinguals, but have tested Koreans who have limited proficiency in English, which we will refer to as Korean monolinguals. For all groups of bilinguals, performance on the plain vs. aspirated pairs did not differ from control pairs. In addition, neither Early nor Late bilinguals performed more poorly than Korean monolinguals on the plain vs. aspirated pairs. However, Early bilinguals' performance on a set of practice items was significantly worse than Korean monolinguals (p < .05), confirming their lack of proficiency in their native language.

Conclusions: The English results suggest that early exposure to a second language can lead to native-like perception of speech sounds in the second language. However, the Korean results indicate that loss of dominance in a first language does not necessarily lead to impaired phonetic perception in that language. These results do not support the notion of a critical period for the acquisition of a perceptual system, but they do suggest that native exposure to a language may have a particularly lasting effect on perception.