Uso de pós-edição de tradução de máquina (MTPE) na tradução de histórias em quadrinhos: um estudo de caso

Abstract

This undergraduate thesis taps into case studies collected to analyse the possibility of using two machine translation engines, DeepL and Google Translate, for the Machine Translation Post-Editing (MTPE) of comics from English to Portuguese by taking into account the technical and temporal efforts involved. A corpus was compiled that consisted of: a page from Maus by Art Spiegelman, and a page from Calvin and Hobbes by Bill Watterson, both in English; the same pages published in Portuguese in Brazil; and the same pages translated from English into Brazilian Portuguese by two web-based MT engines, Google Translate and DeepL. The corpus was analysed as to the number of characters, excerpts with grammatical and translation errors, and BLEU-based translation quality scores. A human analysis of the machine translations was also carried out by considering the perceived difficulty in correcting grammatical and translation errors through MTPE. The results were partially satisfactory for post-editing and seemed to correlate with decreased temporal, cognitive and technical efforts.