Register and the features of translated language

Against the background of current corpus-based research on the features of translated language, this study investigates two research questions that emerge as "gaps" in existing research: (1) What are the occurrence patterns for the different hypothesised features of translated language, investigated together? (2) What is the relationship between register and the features of translated language? Utilising a comparable corpus of translated and original English produced in South Africa, the study tests two hypotheses based on the above questions. The first hypothesis is that the occurrence of linguistic realisations associated with particular features of translated language will demonstrate significant differences in a corpus of translated English texts and a comparable corpus of non-translated English texts, reflecting overall more explicit, more conservative, and simplified language use in the translation corpus than in the corpus of original writing. As a starting point for factoring in the variable of register, it was further hypothesised that the frequency of these features in the translation corpus will show no significant effect for the relationship between corpus and register-in other words, the translation-related features would not be strongly linked to register variation. This has the collateral effect of suggesting a broader hypothesis that in the translation corpus less register variation, or sensitivity to register, will occur, specifically as a consequence of translation-specific effects. The findings from the investigation provide limited support for the first hypothesis, with statistically significant differences between the two corpora for only two of the features investigated: the use of the optional that complementiser, and lexical variety. The second hypothesis, that the interference of the translation process will lead to a "levelling out" of registers, is not supported by the findings.