Recurrent features of translation in Canada: A corpus-based study

Abstract

Based on the theory of translation universals, the general hypothesis that translated texts are distinguishable from non-translated texts by certain recurrent features of translation has been tested in recent contributions to Corpus-based Translation Studies. This hypothesis assumes that translation will leave similar traces in different languages. Major corpus-based studies have recently investigated three specific hypothetical recurrent features of translation (normalization, explicitation, and simplification). However, each of these research projects has hypothesized only one recurrent feature of translation at a time, using mainly literary, Anglo-European corpora, and using English as the sole target language of the translated texts. In the present study, all three of the above previously-studied recurrent features of translation are hypothesized and investigated, along with a fourth (levelling-out), which has not been the subject of previous study. Characteristics of translated and non-translated texts are compared in both English and French: appropriately for study of hypothetical "universal" features, the present research is carried out on target texts in more than one language. Our corpora consist of texts taken from Government of Canada Web sites; they constitute a broad sample of non-literary texts. Specific techniques of analysis are adapted from the literature, and where appropriate, new techniques are devised. WordSmith (versions 3 and 4) was the primary tool used for corpus analysis. The empirical evidence gathered in the present research supports the hypotheses of normalization and explicitation as recurrent features of translation into both English and French, but does not support the hypotheses of simplification and levelling-out. There is some indication that translated texts in both English and French tend to be more difficult to read (according to the standards of readability indices), an unexpected but interesting finding. All of these results must be interpreted in the light of future corpus-based study of recurrent features of translation, and it is recommended that a standardized protocol for recording the attributes of future comparable corpora should be adopted.