摘要：The aim of this study is to examine the effect of high variability training on “-n” and “-ng” in Mandarin Chinese. 46 Japanese native speakers who study Mandarin Chinese as a second foreign language at a university in Japan were divided into one experimental group and one control group. Participants assigned to the experimental group trained themselves using software, which was developed in our early research. Participants in the control group took Chinese lessons normally. Identification tasks (ITs) were conducted before and after the training. After 12 sessions, the average identification accuracy of nasal codas increased 7.1% (p .05) in the control group. Results of the training process prove that the training was effective for some participants, while had little effect on other participants. From the questionnaires, we found that there is a disagreement between the participants’ subjective ability to identify nasal codas and the actual identification accuracy in ITs. Participants underestimate their ability to identify “-n” and “-ng” before training and overestimate their ability to identify “-n” and “-ng” after training. We made a formula to predict the accuracy of ITs (y) in the last 4 training sessions (x): y = 2.13x2 - 2.35x 1.23. This formula will help participants to grasp the training process and predict their identification ability.
关键词：Japanese speaker; nasal coda; Mandarin Chinese; training
