抄録
The present paper reports on auditory and acoustic differences observed in learners' utterances before and after a series of pronunciation tests. A group of Japanese junior high school students (N=66) participated in a 24-session pronunciation course, in which the main activity was one-on-one performance coaching/testing. Audio recordings were made of a short sentence read aloud by a subgroup (n=34) of the participants both at the beginning of Session 1 and at the end of Session 24. The pre- and post-course recordings were compared in terms of (a) perceived degree of holistic foreign accentedness, (b) perceived phonological accuracy of segments, (c) the range and SD of vocal pitch (F0), and (d) acoustic characteristics of /r/ as reflected in F3. Holistic accentedness was judged by L1 English speakers, segments were rated by trained L1 Japanese teachers of English, and F0 and F3 analyses were conducted using Praat speech analysis software. The results indicated that in terms of every variable measured the post-course recordings were closer to the targeted, model than the pre-course recordings were. Through the one-year training, participants' utterances were segmentally more accurate, lower in the degree of foreign accent, came to use a wider pitch band, and F3 frequencies for /r/s became lower. Significance of these results is discussed and an argument is presented that the observed effects should be considered beneficial backwash of performance testing of pronunciation.
