A Corpus-based Study of Translation Universals in Thesis/Dissertation Abstracts

This article argues that the translated English texts hold particular linguistic properties distinctive from the non-translated English texts on a substantial level. By compiling the purpose-end comparable corpora of translated and non-translated English abstracts of theses and dissertations written by both Korean and American graduates, particular translational instances were examined through the translational predictors pertaining to the four dimensions of the translation universals hypotheses: the simplification, explicitation, normalization, and leveling-out shifts. First, the translated English texts were lexically less diverse and dense and syntactically simplified than the non-translated texts when gauged using the values of STTR, function words, frequency words, and mean sentence length. Second, a high preference for adverbial connectives was indicative of the translational explicitation of the translated texts. Third, the use of high-frequency lexical bundles was overtly recurring and detectable across the translated texts, which signifies the instance of translational normalization. Finally, using the standard deviations of STTR and mean sentence length, the translated texts were found to be less homogeneous lexically and syntactically when compared to translated texts, thus being incompatible with the leveling-out hypothesis. All in all, the results indicate that a variety of translational shifts were salient even though some predictors were not sufficiently feasible to verify the universal features of the translated English texts.