영어 ‘be like’ 비격식 인용 구문의 영한 기계번역

English-to-Korean Machine Translation ofBe Like Informal Quotative Constructions

This study investigates the quality of the machine translation (MT) of English be like quotative constructions into Korean. Quotative be like is an underexplored topic in translation studies although it has received considerable interest from linguists due to its innovative as well as colloquial and informal nature. The source texts used in this study were taken from the Corpus of Contemporary American English. A total of 179 be like quotative constructions (i.e., I’m like, he’s like, and she’s like) were examined. The target texts were generated by three different MT engines: Naver Papago, Google Translate, and DeepL. The findings revealed that translation errors were more frequent when quotation marks were absent. The most common mistranslations involved translating like as a preposition (‘similar to’) or a conjunction (‘as if’, ‘in the same way as’). The present study highlights the role of human translators as post-editors because translating informal expressions such as quotative be like remains challenging for MT despite recent advances in neural MT.