Google Translate and DeepL: Breaking taboos in translator training. Observational study and analysis

Abstract

In recent decades, when we read about the use of machine translation, we were told that it was not suitable for professional translation. The reason was that machine translation used to produce poor results as long as it was not applied to controlled language of pre-edited texts within a specific field of specialization, and with post-editing of the target text, all of it using the appropriate specialized tools.

However, three considerations led us to want to assess to what extent machine translation could be used in a German-Spanish/Spanish-German Specialized Translation course, since we observed that

1) despite warnings, students made use of machine translation,

2) Google Translate and the newcomer DeepL were improving their machine translation results, and

3) in short, computer-assisted translation uses machine translation as a secondary help tool, although always controlled by the human translator.

As translator trainers, we planned to use post-editing as an opportunity to teach translation and focus on how to diagnose machine translation errors as a means of improving proficiency in both the mother tongue and the foreign language. In this work we focus on Spanish-German translation of specialized texts, which poses many challenges to students and usually requires reviewing and deepening the grammar of the target language in class. Our aim is to contribute to the didactics of translation, teaching the students how to prevent errors analyzing machine translation and handling post-editing with a critical mind, in order to improve linguistic competence.