| This study was designed to evaluate predictions made by various models of L1 transfer, namely, the Contrastive Analysis Hypothesis, the Perceptual Assimilation Model and the Speech Learning Model. It also investigated the effect of speaker gender and learner gender on perception and production of contrasts involving the AE phonemes /i, I, eI, epsilon ae/ among adult Mandarin speakers. Thirty Murray State University Chinese students volunteer were asked to perform a discrimination task, an imitation task and a word reading task. In all three tasks, performance was best for /i/ (beat), which is the only American English phoneme to have exactly the same phoneme in Mandarin. In the discrimination task, all the contrasts that fit the Two Categories pattern (TC) and the Category Goodness pattern (CG), had better discrimination than the Single Category pattern (SC) contrast of /epsilon/-/ae/ [bet - bat]. However, contrary to what had been predicted, the /eI/- /i/ [bait-beat] contrast, which was expected to fit the TC pattern, showed lower discrimination accuracy than the CG contrast /i/-/I/ [beat -- bit]. In general, better perception was also associated with better production in both imitation and word reading. Speaker gender and participant gender did not affect participants' production in imitation task and word reading task. In the discrimination task, speaker gender and participant gender had no effect for the easiest and hardest contrasts, but had some influence on the moderately difficult contrasts. |
