華語語音向來是日籍學習者難點所在，歷來文獻雖討論不少，然多用一般實驗語料進行研究，語料量少且多非自然情境下的語言輸出。本文採用「臺師大華語學習者中介語口語語料庫」，從中抽樣進行中介語音分析，希望透過量多且較接近自然情境下輸出的語料，進一步檢驗日籍學習者常見之發音偏誤。此外，本研究依研究所得將偏誤語音加以排序，呈現更具科學性的日籍學習者之華語語音偏誤頻率。研究結果發現，語料庫中日籍學習者的發音偏誤現象大致與文獻所述相同。其中，偏誤現象最顯著者分別是：在聲母中為舌尖後音/zh/、/ch/、/sh/、/r/；在韻母中則為後高圓唇音/u/。最後根據分析結果提出針對日籍學習者語音教學之建議，期望能藉此提升對華語語音教學之成效。

Instead of using general experimental data collected as research stimuli, this study aims to analyze the phonological errors that Japanese learners of Chinese frequently make by using Chinese Learners’ Oral Corpus. After obtaining and analyzing the corpus data, this study ranks the error type frequency on different Chinese consonants and vowels to suggest a broader view of the interlanguage phonology of Japanese learners. The error type ranking suggests that the post-alveolar retroflex consonants /zh/, /ch/, /sh/, /r/ in initials and the rounded vowels /u/ in finals are the most common errors for Japanese learners. Based on this error type analysis, several pedagogical suggestions aimed at Japanese learners are further proposed to enhance the effectiveness of Chinese phonetics teaching.