earticle

영어

The purpose of this study is to investigate punctuation translation strategy of the Korean-Russian machine translation and categorize the types of its errors. This study is also aims to seek practical implications for improving product quality of machine translation by establishment of pre-editing rules and its application. For this goals, corpus of articles in KBS World Radio were established, and quantitative and qualitative analyses were carried out. The analysis results showed that both the punctuation translation strategy of machine and of human were almost same. However, the punctuation error rate by machine translation was 24.19% while the error rate by human translation was only 2.4%. To determine the cause of this, only the errors about punctuation by machine translation were classified, and the translation strategy was addition > shift > literal translation > omission. The type of punctuation marks causing errors in the source texts by machine translation was revealed. Then, the errors by machine were categorized by punctuation marks, and Pre-editing rules were created based on this. In addition, to verify the rules’ utility, the source texts and the texts translated by machine, the pre-edited source texts and the newly-produced texts by machine were comparatively analyzed. The analysis result showed that only the pre-editing rules applied to punctuation marks considerably improved sentence structure, grammar, omission, mistranslation and non-sentence.