法律翻譯語料庫建置及分析

Abstract

語料庫語言學及相關技術在翻譯領域中的應用日趨重要，專門語料庫作為特定專業領域的翻譯參考資源，也極具價值。為解決譯者在專門領域的翻譯工作中，可能面臨參考資源不足的問題，本研究嘗試應用現有之軟體資源輔助，建置中英平行結合英文單語之法律語料庫。建置過程採用軟體工具及半自動化方式，進行大批語料處理、自動斷詞、詞性標記、段落（句）對齊，以及詞組對應擷取之工作。建置完成的語料庫，以語料庫軟體輔助，進行關鍵詞、對應詞組、N連詞、雙語關鍵詞檢索，以及單語關鍵詞檢索之分析。研究結果顯示，本文中嘗試採用的語料庫分析方式，可有效幫助譯者取得多種翻譯過程中需要的參考範例。研究過程中取得的關鍵詞、詞組翻譯、常用表達方式和翻譯策略，也可經累積後應用於其他形式的翻譯資源建置。本研究採用的語料庫建置與分析方式，還可應用在其他專門領域之翻譯，以支援譯者工作需求；分析過程中觀察得的許多現象，也值得進一步分析探索，期能貢獻於未來的翻譯實務與研究工作。

Corpora, the well-organized bodies of “naturally occurring language data” sampled to represent a variety of language (McEnery, 2003, p. 449), have been making a growing impact in the field of translation (Bernardini, Stewart, & Zanettin, 2003). Scholars have asserted the immense value of corpora as reference tool for translation practice in specialized subject domains (Bowker & Pearson, 2002; Varantola, 2003), where intrinsic features of the language may cause difficulties for the translator. To address the potential lacking in reference tools for specialized translation assignments, this study explores a number of methods and computerized tools in compiling and analyzing a parallel and monolingual corpus of Chinese and English legislation. Incorporating semi-automated tools for text processing, part-of-speech tagging, sentence alignment, and phrasal alignment, this study utilizes keyword analysis, n-gram and n-gram part-of-speech sequence, as well as bilingual and monolingual concordance search to address identification of terminology equivalents, stylistic features, usage patterns, and translation strategies for legal contexts. Findings suggest that with the proposed methods, the corpus compiled in this study could effectively provide a number of information to aid the work of legal translators. The information identified can also be applied to compiling other forms of translation resources. It is hoped that in future research, the corpus tools and approaches employed in this study can be applied to facilitating other specialized fields of translation, and that preliminary findings observed here could be further explored to benefit future work in this discipline.