Using lexical bundle analysis as discovery tool for corpus-based translation research

摘要

The aim of this paper is to propose a method of analysing lexical bundles from the perspective of translation research oriented toward revealing characteristic features of translated texts vis-a-vis non-translated texts and demonstrate its usefulness by analysing 3-ejel lexical bundles in translation and non-translation comparable corpora of Korean journalistic texts. The method proposed consists of four steps: extracting lexical bundles from the corpora; filtering out patterns tied to source texts; selecting individual patterns likely to be reflective of inherent properties of translated texts; and carrying out in-depth analysis and explaining the results. The two case studies reported in this paper showcase this process, leading to discovery of what could be important distinctive features of translated Korean journalistic texts as compared to those of non-translated texts, while offering clues about possible variation in stylistic choices across (groups of) translators.