La traduction automatique comme ressource pour la traduction humaniste-littéraire : un instrument ‘artificiel’ peut-il transférer un texte essentiellement ‘humain’ et ‘social’ ?

Abstract

The usefulness of machine translators (MTs) is virtually indisputable. Although, until recently, they could be considered almost the ‘enemy’ of translators, the rise of technology in our profession has changed from a threat to an opportunity to broaden skills and take on new roles (Moorkens, 2018). MT has evolved since its inception and, in particular, Statistical MT (SMT) and Neural Network-based MT (NMT) systems have become very popular. One question that deserves particular attention is whether these automatic resources are just as effective for a humanistic-literary translation as they are for a specialised translation in the scientific-technical field. In this article, we have compared the degree of appropriateness of a corpus of EN>ES translations produced by around one hundred 4th year degree Translation and Interpreting students at the University of Córdoba. The analysis shows, as could be predicted, that, although it is always necessary to revise the automatic target text, in the case of humanistic-literary texts it is essential to use MT only as a basis for a ‘human’ translation. Furthermore, and very pertinently, the students’ translation results were better when the texts were related to their own lives or had social repercussions with which they could sympathise.