Machine Translation Problems in English-Arabic Collocations and Post-Editing: Google Translate as a Case Study

Abstract

Every language has its own collocations and fixed expressions that represent its related culture and ideological, social, and religious differences. In the field of translation technology, although there have been developments in Machine Translation (MT), problems still exist, particularly when translating English collocations into Arabic. This paper aims to investigate the stylistic problems in (Google Translate) outputs from English into Arabic when translating collocations in a scientific(semantic) context and stand on the limits and strategies of Machine Translation from the perspective of language and thought production, also to reveal the extent to which one of the neural Machine Translation programs (Google Translate) can translate scientific texts from English into Arabic, which include many collocations extracted from Ogden and Richards’ (1946) linguistics text, The Meaning of Meaning: A Study of the Influence of Language Upon Thought and the Science of Symbolism, with Supplementary Essays. Furthermore, it highlighted the stylistic problems in collocations when translating. In addition, it aims to Post-Editing output of Machine Translation according to the rules of translation quality and acceptability. The study adopts the comparative analytical approach for the data collected from the Google Translate (2021) output, thus comparing it to human translation, then applying Post-Editing to these outputs. The results indicate that Google Translate fails to utilize the appropriate technique(s) when it renders metaphors from English into Arabic because it lacks the required profound knowledge of the linguistic and cultural legacies of both the source and target languages. Also, Google Translate output of collocations within a scientific text need a fully Post-Editing with a revision.