A Corpus-based Study of Human-translated vs. Machine-translated Texts: The Case of Ellipsis in English-Persian Translation

Abstract

The present study was intended to investigate the translation of different types of ellipsis in Persian and English using a descriptive method. Given the importance of the study of human translation (HT) and machine translation (MT), the present study focused on the ellipsis based on Halliday and Hasan’s model (1976). In so doing, a bilingual parallel corpus called "Mizan corpus", including more than one million translated sentences, was compared with Google Translate, as an example of machine translation. The results of the study indicated that there were some mismatches in verbal and nominal ellipsis between English and Persian, but almost no mismatches were found for clausal ellipsis. There was a significant difference in the quality of human translation and machine translation (i.e. Google Translate) in favor of human translation. The findings also depicted that HT’s quality was still higher than MT for all kinds of ellipsis. So, it was concluded that human translation is more dependable than machine translation.