A Comparative Study of the Quality of Persian Translations by Google Translate and SDL Free Translation

Abstract

As the aim of machine translation is to replicate human translation, it is of much significance to incorporate human judgments of translation quality into system development. While there have been a large number of studies on the quality of machine translation outputs in many language pairs, there are limited studies on the quality of machine translation outputs in the context of English into Persian translation as assessed by human evaluators. With this in mind, the present research investigated into the quality of translations by Google Translate and SDL Free Translation in terms of clarity, accuracy, style and informativeness as assessed by professional as well as student translators in the context of translation from English into Persian. The texts chosen for evaluation by two machine translation systems consisted of four extracts adapted from two different genres. The first two passages were travel guides while the other two passages were manuals. Our data indicated that while both systems provided translations which were of poor quality in terms of clarity, accuracy, style and informativeness as assessed by our evaluators, the quality of translations by Google Translate was higher than that of SDL.