本研究旨在探討台灣英文作文教師校正學生錯誤的能力並探究教師常誤判的錯誤類別及其原因。研究中，共有四名大學英文系三年級英文作文教師參與。主要研究工具包括四名教師批改後的學生英文作文，共計二十四份，現場批改短文測驗、同步出聲思考紀錄（Think-aloud protocol）以及兩次的面談紀錄。研究結果發現，教師修正學生錯誤的表現在不同的情況下有其差異。從批改後學生作文中看來，教師校正學生錯誤的準確度達百分之七十八，但在批改短文測驗中，準確度則降至百分之四十八。另外，研究也發現教師常將他們的建議以校改錯誤的模式呈現。在分辨出教師的建議以及校正後，教師校正錯誤的準確度可提高至百分之八十五及百分之五十五。研究指出，教師最經常誤判的錯誤類別為用字以及句子結構。誤判錯誤以導致校正過多的原因包含教師對於英文用法的認知有錯或是不完整、對於型式上的過度重視、對於作文主題熟悉度和批改作文角度。而將錯誤誤解為其他類別的錯誤則多源於教師批改作文的習慣問題。最後，研究者針對教師校正錯誤表現差異的原因，以及改善教師校正錯誤的方法上提出見解。 The present investigation seeks to examine the quality of non-native English speaking teachers’ error correction and explore the causes of teachers’ unnecessary and inaccurate corrections. Four college professors participated in the study. Data were gathered from four sources: (1) twenty-four corrected student essays, (2) error correction task, (3) think-aloud protocol, and (4) interviews. The results revealed that the quality of teacher error correction differed in contextualized and de-contextualized contexts. While about 78% of the corrections on the student essays were accurate in location and correction, the accuracy rate of the corrections on the error correction task was as low as 48%. It was also found that teachers had the tendency to blur the distinction between suggestions and corrections and offer them in the same form. The accuracy rate of the correction in both contexts could increase by 8% after the suggestions were identified and withdrawn from the corrections. Causes of teachers’ unnecessary corrections were rather versatile, including false or partial understanding of English grammar and usage, focusing on style, content and correcting from readers’ perspective. On the other hand, teachers’ inaccurate corrections mainly resulted from the correction strategy undertaken. Possible factors contributing to the difference in the accuracy rate and implications to improve teachers’ error correction practice were discussed.