Investigating the Accuracy of Google Translate in Translating Urdu Linguistic Elements

Abstract

This study aims to investigate the accuracy of Google Translate regarding the translation of idioms and proverbs from and to English language. The shortcomings of Google Translate and the limitations of Neural Machine Translation have also been pointed out after in-depth analysis of carefully chosen examples from authentic lexicons. An experiment has been conducted by running Urdu and English idioms, proverbs and other linguistic expressions through Google Translate. The theoretical framework of Antoine Berman's twelve deforming tendencies has been employed to analyse these examples and Mixed Methods design has been applied for the evaluation in the study. The error analysis highlighted all the linguistic mistakes and errors committed by the software during Google machine translation. Major contribution of this study is the provision of the accurate alternatives for the wrong translation. This study plays a diagnostic role in identifying and classifying those errors. Furthermore, it contributes in improving the quality of Machine Translation from Urdu to English and English to Urdu. In the end, suggestions for the effective advancements in the software have also been provided.