ABSTRACT
The study of translators’ styles is an important research topic that has attracted increasing attention in translation studies (cf. Baker, 2000; Huang, 2015; Li and Shao, 2016; Marco, 2004, among others). However, to date, virtually all these studies have relied on manual identification of linguistic features characteristic of a translator. To bridge this gap, this study introduces state-of-the-art tools in computational linguistics that can facilitate semi-automatic identification of translators’ styles. We demonstrate how to draw on natural language processing (NLP) tools to construct a paragraph-aligned parallel corpus based on The Old Man and the Sea and its three Chinese translations by Eileen Chang, Shaowei Zhao, and Kwang-Chung Yu. Through a series of corpus processing procedures including Chinese tokenization and bilingual alignment at paragraph and word level, lexical divergences among different translators can be automatically derived. The statistical information from bilingual word alignment and Keyword List empowers researchers to explore the systematic stylistic differences of word choices among the three translators in a straightforward manner, facilitating the identification of translators’ styles in a way that was difficult to obtain through traditional methods. Compared with previous approaches, our proposed corpus-based computational approach can address translators’ styles more efficiently and effectively by presenting both quantitative and qualitative evidence.
