話輪修補在語言教學中的地位至關重要。基於會話分析理論，對話輪修補進行深入探討，有助於第二語言習得與教學法的進一步發展。儘管相關語言學研究通過分析北京或台灣普通話會話材料以解決話輪修補的問題，但甚少側重漢語作為第二語言學習者在漢語課堂會話中的自我修補機制。同時相關研究僅注重區分話輪修補的具體方法，並未詳細討論每種話輪修補機制的發啟與結果。因此本文旨在研究漢語課堂會話中二語學習者自我修補機制的實際現象，以期激勵漢語教師在實際教學中對二語學習者提供更為有效的幫助。本文採用會話分析的傳統方法，調查了兩種自我修補機制（即自我發啟自我修補與他人發啟自我修補）並驗證了相應的運用策略。本研究數據來自武漢大學公共語料庫，錄音時長共計147分鐘，涉及四名漢語教師與兩組班級間的課堂會話。本研究結果表明，除他人發啟修補機制外，上述兩種自我修補機制均出現在會話數據中，此外教師通過重複之前的提問來強調或間接指出學生話輪中的癥結所在。該結果證明，會話分析的方法適用於漢語作為第二語言課堂師生會話的研究。漢語課堂會話中表現出的自我修補的不同方式與對英語日常會話分析的以往研究結果亦大致相同，但仍存在自我特徵。從教學法意義言之，會話分析可以幫助漢語教師透過觀察學習者自我修補機制的運用，瞭解學習者的漢語掌握及運用能力。

The organization of repair plays the most significant role in language instruction. Insight into repair, informed by Conversation Analysis (CA), contributes to the development of second-language acquisition (SLA) and pedagogy. While a number of Chinese linguists have worked on the issue of repair with the Beijing and Taiwan Mandarin Chinese conversation data (e.g., Chui 1996; Tao 1995; Yang 2009), attention is rarely focused on Chinese as a second language (CSL) learners' self-repair and the repair devices applied in Chinese institutional talk. Also, the initiation and the outcome of each repair were not properly discussed, as these studies only analyzed the classifications of detailed repair methods. Therefore, this study aims to look into the details of L2 learners' self-repair devices in Mandarin Chinese classrooms, in order to inspire CSL teachers to reach out to L2 learners more effectively and successfully in an authentic teaching context. To do this, I investigated the two types of self-repair (i.e., self-initiated self-repair and other-initiated self-repair) and identified the appropriate repair practices in Chinese conversation. The analysis was based on 147-minutes of audio involving institutional conversation between four teachers and two CSL classes, drawn from a corpus recorded at Wuhan University, China. The results indicated two types of self-repair while the absence of other-initiation methods were noted in the data. Instead, the repetition of the original question was utilized by CSL teachers in order to emphasize and indirectly target the trouble source of the students' turns. Therefore, I concluded that although the CA-informed repair mechanism is highly applicable to Chinese conversations, some particular features of self-repair in the CSL classroom cannot be ignored.