Comparing Post-Editing Translations by Google NMT and Yandex NMT

Abstract

This study is aimed at examining the naturalness of post-editing translations using Google NMT and Yandex NMT by English Department students and to determine which of the two NMT tools came closest to the naturalness of a short story's translation. The subjects of this study were English Education students from the University of Mataram who come from the native area. Meanwhile, the object of this study was a short story entitled “Jack and The Beanstalk,” in English version. In this study, the researcher used Larson’s theory as the study’s reference to analyze the naturalness of translation in the short story “Jack and The Beanstalk” from English to Indonesia. The data were obtained by two methods of data collection: observation and documentation. The total data in this study was 1248 sentences, which were analyzed descriptively. The result of this study showed the percentage of text quality in naturalness translation that students produced in conducting post-editing. In GNMT most of the post-editing quality is “highly natural” with 88%, followed by “natural” with 5%, “less natural” with 6%, and “unnatural” with 1%. On the other hand, little few differences were found in YNMT, which shows that most of the post-editing quality is “highly natural” with 81%, followed by “natural” with 6%, “less natural” with 8%, and “unnatural” with 4%. According to that percentage, it can conclude that the quality of naturalness translation in post-editing from GNMT is easier to edit and produce better-translated text than YNMT.