A Computational Modelling Approach to the Development of L2 Sound Acquisition

A computational framework for quantitative studies of sound acquisition in a second language is presented. The framework supports the exploration of issues such as the effect of the amount and type of exposure to L2 categories, interactions between L1 and L2 sound systems as well as L1 attrition. The model is illustrated by simulating a Chinese listener’s responses to English vowel-consonant-vowel exemplars. The simulations suggest that even very small quantities of L2 material can lead to rapid improvements in recognition of L2 target consonants and that with balanced amounts of exposure to the two languages, L2 models can contribute to the recognition of L1 tokens.