한국어 관용구의 기계 번역 양상 연구-{‘마음’+서술어} 형태를 중심으로-

Abstract

This study attempted to examine how items classified as idioms in Korean are translated through machine translation based on artificial neural networks. in the standard Korean dictionary The results of 14 items presented as idioms translated into English through Google Translator, Kakao Translator, and Papago Translator were compared focusing on the equivalence of the semantic and morphological aspects. This study looked at how the corresponding items were realized in the translation of the title words and examples, and whether the result of the title control translation was applied to the example translation, that is, the result of the title control translation and the result of the example translation were consistent in the same translator. In order to confirm the accuracy of the results translated into English, a BLEU evaluation was conducted to compare them with back translation and human-performed translation. Based on the results of this analysis, the results derived according to the translator were compared, and the types were classified based on form and semantic relevance. The significance of this study can be found in that it suggested the direction of the parts and utilization measures that machine translation should supplement to contribute to improving the quality of the translator through various aspects of the translation results.