失語症病患(aphasic patients)的研究對於語言的運作機制提供了一些訊息，讓我們能從不同的角度認識大腦的運作。本篇研究的目的在於探討失語症病患的自我修補表現(self-repair performance)，我們利用一套具有五大自我修補項目的分類系統分析病患的表現並將參與研究的四位失語症病人區分為流利型(fluent)以及非流利型(non-fluent)病患，之後再分別將四位病人與其治療師一對一的各別談話錄音、譯寫。在流利型的組別之中，我們收集了207個威尼克病患(Wernicke’s aphasics)的自我修補語料，而在非流利型組別當中，則收集了347個布洛卡病患(Broca’s aphasics)的自我修補語料。 結果顯示，威尼克病患與布洛卡病患在自我修補的整體表現基本上是相似的，也就是說，在這五大項目的自我修補模式中，他們使用每一項的頻率是沒有顯著差異的。因此，我們認為自我修補是一種溝通能力的表現，而這樣的能力並不會受到病人各自的語言缺陷所影響。然而，我們得知兩種類型的病患在自我修補的表現上，依然具有一些細微的不同。威尼克病患使用了較多的填補模式(the completion pattern)，這或許與他們較流暢的表達能力有關係，因為他們能較不費力的把尚未完整說出的詞彙補述完成。布洛卡病患則是使用了較多的取代模式(the substitution pattern)，這或許是因為他們的理解能力較好，幫助他們監控了自己說話裡頭的錯誤並進而修正。 此外，比較中文正常說話者與失語症病患的表現，我們發現每一種自我修補項目被使用的比率亦是相似的。這種情況或許提供了一種證明，表示我們在使用自我修補的模式上，都有一種共同的傾向(universal tendency)。 總結來說，在自我修補的表現上，影響失語症病患的三個可能因素是1、在語言處理過程中的複雜度。2、不同類型病人各自具有的語言缺陷。3、每一種自我修補模式所具有的實用性。 / Studies of aphasic patients provide more information for understanding mechanisms of language processing, since we can know more details of the human brain from a different perspective. The purpose of this study is to investigate self-repair performance of aphasics. A classification system made up of five major categories was adopted to analyze their performance. We divided four patients into fluent and non-fluent groups and collected conversation recordings of these four therapist—patient dyads. In fluent group, 207 self-repairs were recognized in the conversation between therapists and Wernicke’s (fluent) patients. In the non-fluent group, 347 self-repairs were found in the conversation between therapists and Broca’s (non-fluent) patients. Results indicated that the overall self-repair performance of Wernicke’s (fluent) and Broca’s (non-fluent) aphasics was similar. That is, the distribution frequency in the five categories was not significantly different. We therefore suggest that the self-repair performance is a communication-oriented ability, and it is not interfered by specific language deficits. However, there were still subtle differences between the two patient groups. We speculate that the relative higher percentage of the completion pattern in Wernicke’s (fluent) group is due to their rather fluent production ability, since they can fulfill an unfinished word without too much difficulty. On the other hand, Broca’s (non-fluent) patients used more of the substitution pattern. That is, they can detect the problems within their own speech and further correct them. This could be attributed to the self-monitoring ability brought by their better comprehension. In addition, comparing with the studies of Mandarin normal speakers (Chui, 1996; Tseng, 2006), we found that the distribution frequency of each self-repair pattern in aphasic patients was similar with the distribution in normal speakers. This similarity might indicate a universal tendency of the self-repair use. In sum, there are three possible factors affecting the self-repair performance of aphasics, they are the processing difficulty, the linguistic deficits of different patients, and the value of the repair pattern.