Varden 和Sato（1996）藉由日語發音實驗，首先發現台灣人學習者對於無聲子音條件下的高母音[i, ɯ]，不易無聲化。本研究以2名台灣學習者為對象，藉由較自然之音讀文章情形，針對此課題做了一次更整體性、系統性的探討。研究結果得知（1）其中1名學習者，在34個高母音當中出現母音無聲化現象的音有2個，另1名學習者則出現6個。（2）其音節上的子音持續時間，日語音讀範例者和2名台灣人學習者之間未出現顯著性差異，惟音節上高母音部分，學習者的持續時間長度將近於日語母語話者兩倍之長，可見母音的持續時間是重要因素。本研究將針對以上兩點，簡潔地探討教學方面的建議事項及今後的研究內容。

Varden and Sato (1996) were the first to find that, on a pronunciation task, Taiwanese learners of Japanese tend to fail to devoice the high vowels [i, ɯ]in the voiceless consonant environment. The present study, using two Taiwanese learners and an oral reading task, verified Varden and Sato's findings, and thoroughly investigated the issue in a systematic manner. The main findings were that (1) one learner devoiced 2 out of 34 high vowels, and the other, 6, and (2) while the voiceless consonant durations are comparable between a model Japanese speaker and two Taiwanese learners, the durations of the high vowels were almost twice longer for the Taiwanese learners than for the model reader, thereby suggesting that vowel duration is a key factor. Educational implications and further research are briefly discussed.

臺灣人學習者 ； 音讀 ； 日語高母音 ； 無聲化 ； 音響分析

Taiwanese learners ； oral reading ； Japanese high vowels ； devoicing ； acoustic analysis