Abstract
Using corpora in translator education has increased greatly in importance over the past decades (Zanettin, 1998, Bowker, 2000, Zanettin et al., 2014). This research demonstrates how the use of small specialized corpora - often termed virtual corpora or ad hoc corpora (Sánchez-Gijón, 2009, Hernández, 2010) can improve translator trainees’ understanding of the source texts and production of high-quality target texts. This study provides an experiment comparing translations produced by a group of Chinese translator trainees. One translation was completed by using traditional resources; the other with the use of small bilingual specialised comparable corpora (English - Chinese) compiled with WebBootCaT toolkit and concordance provided by Sketch Engine. The results reveal that translations achieved by using corpus were of higher quality. Benefits brought by corpus-derived insights include: firstly, the source language corpus provides explanations of the terms that are not contained in conventional lexical resources. For example, the term flat spot cannot be found in conventional resources and was literally rendered as 平点 (ping dian, flat spot), a translation that does not exist in Chinese expert discourse. In contrast, by investigating concordances in the English corpus, trainees acquired the meaning of flat spot as the momentary hesitation of engine during acceleration. Based on this, feasible translations were produced. Secondly, the target language corpora can be used as a testbed for checking the conventionality of translation candidates. For example, tuning box can be translated as 调试盒 (tiaoshi he, tuning box), 外挂盒 (waigua he, plug-in box), or 外挂电脑 (waigua diannao, plug-in computer), etc. By checking the frequencies of such translations in the Chinese corpus, trainees can make better decisions. Thirdly, the comparable corpus provides more idiomatic expressions in both languages, enabling trainees to improve their level of linguistic and domain expertise.