기계번역을 활용한 한영번역에서의 번역품질 비교연구

Abstract

This paper aims to suggest how to produce a better translation method in Korean-English machine translation. For this purpose, it compares translation error types and frequencies according to three machine translation methods; the pre-editing, post-editing and the combination of pre- and post-editing. In the experiment, 31 subjects are divided according to the TOEIC score and a selected translation method. The subjects are required to translate a 450-word report text through Papago, one of the machine translation tools and participate in the survey related to the reason for a selected translation method. The analysis shows subjects with the lower TOEIC scores below 600 select the pre-editing method while those with the higher TOEIC scores above 600 choose the post-editing and the combination of pre- and post-editing methods. It also indicates depending on a selected translation method, translation error types and frequencies are different. The better machine translations are produced by subjects who have higher TOEIC scores and choose the combination of pre- and post-editing method. This paper is meaningful in that the machine translation materials are analyzed and compared according to the three machine translation methods and the findings are utilized in translation and writing classes.