인공지능 기반 번역기 활용 교육: Pre-editing

English education with AI-based translators: Pre-editing

The purpose of this study is to propose simplified pre-editing rules of Korean-English machine translation for ordinary users of AI-based translators. Pre-editing is the process by which an author edits text before applying machine translation in order to improve the quality of the output of machine translation. While some pre-editing rules are universal such as avoiding syntactically complex structures or lengthy sentences which cause processing difficulties, some pre-editing rules are language specific and thus unidirectional, such as null subjects and objects. Homophonic words, especially Sino-Korean words, would be mistranslated if the contexts are not clear in a sentence. Mistakes in punctuation would trigger mistranslation as well. The simplified pre-editing rules include the aforementioned 4 rules. A 4-hour online course of the simplified pre-editing rules showed that they were easily acquired and applied to source texts for enhanced machine translation results. In addition to the satisfactory translation results, the ordinary users recognized the similarities and differences between the two languages in the process of pre-editing. In particular, English learners were able to understand and utilize the syntactic characteristics of a pro-drop language such as Korean. Thus, AI-based translators, with the simplified pre-editing rules, can be used in English education for learners to notice similarities and differences between Korean and English.