Sistemi di traduzione assistita e testualità: un’analisi di testi istituzionali della Confederazione Svizzera

Computer-Aided Translation tools, or CAT tools, have been widely used by professional translators for at least thirty years. While production companies praise their help in boosting productivity and improving the quality of translations, several scholars have argued that their architecture, based on the use of translation memories which pre-segment the text into sentences, may produce negative effects on text coherence and cohesion. This paper sets out to investigate the implications of the use of CAT tools on text structure, paying attention to textually relevant shifts from source to target texts. The analysis is conducted against the backdrop of a model specifically designed for translated texts belonging to the institutional domain (Ferrari, Pecorari, 2022). The model is applied to the contrastive study of a parallel corpus of press releases translated from German into Italian by the Swiss Confederation, partly with the help of the CAT tool Transit, partly through a manual translation. The results show that textually relevant features are translated with more freedom, as expected, when the CAT tool is not employed; however, the difference between the two translation environments is slighter than expected and highly variable depending on the feature.