Effects of translation memory software on English-to-Spanish translation: evidence of normalisation in translations by undergraduate trainees

This article reports on a translation quasi-experiment aimed at looking at how the textual and grammatical features of non-translated English and Spanish scientific texts are represented in translations from English into Spanish when they are mediated by translation memory software. Senior undergraduate translation trainees from three different Chilean universities translated abstracts for research articles in conservation biology from English into Spanish in a translation memory environment. The translations were then contrasted against a corpus of non-translated texts of the same genre written in English and Spanish, in terms of syntactic dependency relations. The results suggest that student translators, regardless of the use of TM systems, would tend to ‘normalise’ translation, i.e. they would be overly concerned about conforming to target language norms to the extent of overusing grammatical features.