A Critical Evaluation of Google Translation: The Case of English-Sorani Kurdish Translation

Abstract

Google Translate is one of the most frequently and actively used machine translation services worldwide. Sorani Kurdish, otherwise known as Central Kurdish, was officially added to Google Translate in May 2022. At the onset of the launch of Sorani Kurdish translation services, it is crucial to evaluate the quality of the translation and pave way for improvements, if necessary. Taking into consideration both linguistic and cultural aspects, this paper aims to critically investigate the accuracy of English-Sorani Kurdish translations produced by Google Translate. For this purposes, the paper has adopted a textual analysis approach based on the notion of equivalence to critically analyze samples of translations generated by Google Translate. The results suggest that Google Translate is largely accurate at morphological and syntactic levels but improvements are needed to improve minor deficiencies detected in rendering ‘future perfect tense’, ‘third conditional sentences’ and passive tenses, especially in the case of ‘present progressive’ and ‘past progressive’ tenses. With regard to the semantic level, Google Translate has proved accurate in processing polysemous words, but it has assumed a total failure in translating idioms and proverbs. At the cultural level, in over third-fourths of instances Google Translate renders English cultural terms into Sorani Kurdish accurately, whereas in nearly one-fourth of instances it offers inappropriate translation. As far as English into Sorani Kurdish translation is concerned, Google Translate can be deemed effective as a supportive translation tool but it is far from being perfect, and continuous improvements will be needed to respond to the deficiencies unearthed in this study.