逐句口譯之錯誤分析研究

An Error Analysis of Sentence by Sentence Interpretation

葉舒白

指導教授 ： 林世華

國立臺灣師範大學/文學院/翻譯研究所/博士(2012年)

摘要

本研究旨在以Corder（1974）的學說為基礎，探索適用於口譯研究的錯誤分析法，並以該法分析英譯中逐句口譯。
本研究設計一結合單字辨識、句子理解與句子口譯任務的實驗，以30位大學主修英文者為對象蒐集語料。所有原文句和譯文句均在抄錄後分解成命題。為規範錯誤的辨認與分類，本研究提供一參考譯文，並根據命題結構確立比對及對應中英文命題的原則。為解釋並評估錯誤，本研究從任務及表現兩方面分析錯誤類別、句長及句速之間的關連。本研究的主要發現如下：
1. 參考譯句、比對原則及對應原則，似乎有助於降低主觀，使辨認與分類的結果趨向一致，讓單語人也可擔任評定者。
2. 除了遺漏、替換以及扭曲原意的增加之外，涉及蘊含（entailment）及預設（presupposition）概念的語意錯誤，似乎也可做為錯誤類別。譯文句內最大的錯誤類別為扭曲原意的增加，原文句內最大的錯誤類別則是遺漏，顯示邊加邊漏似乎是逐句口譯的特性。此外，譯句的長度及速度似乎是可靠的品質及表現指標，尤其是譯句秒數。
3. 相較於理解度，辨識度與原文句譯出與否之間的關係似乎較為密切，且長度較能反映辨識與理解的難度，顯示原文句越長時：（a）參與者似乎越在已完成辨識但未充分理解的狀況下，以零星的意義單位建構譯文句，使扭曲原意的增加處比例上升；且（b）譯文句就越長，原意似乎也越可能因記憶的衰減及干擾而遭到遺漏。
4. 不論從譯文句或原文句觀之，錯誤的比例均過半，顯示整體品質欠佳。遺漏與扭曲原意的增加之間具共發性，屬較嚴重的出錯型態。

關鍵字

錯誤分析 ； 命題分析 ； 難度指標 ； 表現指標

並列摘要

In line with steps suggested by Corder (1974a), this study aimed to develop procedures tailored to the analysis of interpretation errors, and conducted an Error Analysis of English to Chinese sentence-by-sentence interpretation.
An experiment consisting of word recognition, sentence comprehension and sentence interpretation tasks was carried out to collect target outputs from 30 bilingual college graduates with dominance in Mandarin Chinese. Both the source inputs and target outputs were transcribed and deconstructed into propositions. To facilitate the identification and classification of errors, reference interpretations and guidelines were established to govern the comparison and alignment of source and target propositions. To explain and evaluate errors, correlation analysis by task and by performance was applied to investigate the connections within and between error type, sentence length and sentence speed. The major findings of this study are listed as follows:
1. The reference interpretations, and comparison and alignment guidelines seemed to effectively limit subjectivity, ensuring more consistency in error recognition and classification, and allowing monolinguals to participate as raters.
2. In addition to omissions, substitutions and meaning-altering additions, entailing and presupposing propositions were also classified as errors. Meaning-altering additions were the largest error type of the target sentence, while omissions the largest of the source sentence. The speed and length, especially duration, of a target sentence seemed to be reliable indicators of quality and performance.
3. Compared with comprehension rate, recognition rate seemed to correlate better with how much a source sentence was correctly interpreted. Length was a more reliable difficulty indicator of recognition and comprehension than speed. This seemed to imply that the longer the source sentence, the more likely that: (a) it was recognized but only partially comprehended, thus its interpretation was characterized by meaning-altering additions composed of fragmented meaning units; and (b) its target sentence was lengthier, thus omissions occurred due to the decay and interference in memory
4. On average, more than 50% of a source and target sentence were errors, indicating a generally poor quality. Meaning-altering additions and omissions seemed to co-occur, thus were considered grave errors.

並列關鍵字

Error Analysis ； Propositional Analysis ； difficulty indicator ； performance indicator

參考文獻

賴則中 (民99)。從文本難度與特色看視譯之困難 (未出版碩士論文)。國立臺灣師範大學。臺灣台北。取自http://ir.lib.ntnu.edu.tw/handle/309250000Q/76382

Abadia-Barrero, C., Rowinsky, P., Hausmann, J., Battisti-Ashe, A., Peters, G., Flores, G., Laws, M. B., et al. (2003). Errors in medical interpretation: Our concerns for public health and a call for caution. Pediatrics, 111(6), 1495–1497.

Agrifoglio, M. (2004). Sight translation and interpreting: A comparative analysis of constraints and failures. Interpreting, 6(1), 43–67.

Al-Khanji, R., El-Shiyab, S., & Hussein, R. (2002). On the use of compensatory strategies in simultaneous interpretation. Meta, 45(3), 548–557.

Al-Salman, S., & Al-Khanji, R. (2004). The native language factor in simultaneous interpretation in an Arabic/English context. Meta, 47(4), 607–626.