Investigating The Structural and Lexical Accuracy of The Translations Done by Machines and Translation Centers
In this study, the structural and lexical accuracy of the translations done by machines and translation centers were analyzed and compared. The text under study included 40 Persian and 40 English sentences, which were translated by two randomly selected translation centers and the Google translate. The translated texts were analyzed and scored by two raters. The scores were then subjected a set of descriptive and inferential statistics. According to the results, the structural accuracy of the English to Persian translations were 87% and 76% for the translation Centers and the Google translate respectively, indicating that the centers outperformed machines by 11%. In lexical accuracy also the centers did better than Google Translate by 6%. The translation centers manifested a rather similar accuracy (i.e. 87% and 86% accuracy rates) on both target-to-source (English-to-Persian) and source-to-target (Persian-to-English) translations. The Google Translate, however, scored significantly lower (i.e. 76%) on target-to-source (English-to-Persian) translations, but manifested a rather similar accuracy (85%) on source-to-target translations, indicating that, as far as machines are concerned, accuracy on one area of translation may not be comparable to accuracy on other areas.
Index Terms - Machine versus professional translations, Translation accuracy, Translation agencies