베한 기계번역 오류 양상 및 포스트에디팅 – 베트남 외교부 입장문을 중심으로 –

Abstract

This study evaluates errors in Vietnamese-Korean machine translation by referencing post-edited texts, focusing on the Vietnamese Ministry of Foreign Affairs' statements from 2018 to 2023. The research utilized Naver Papago and Google Translate to analyze a total of 51 sentences. By comparing the machine-translated texts with their post-edited versions, the study identified the predominant types of errors. The errors in the 102 sentences translated by the two engines were categorized into seven types: semantic, grammatical, word order, lexical, addition, omission, and numerical errors. The analysis revealed that errors occurred in approximately 77% of the texts (39 texts) translated by Google Translate and in about 75% of the texts (38 texts) translated by Papago. The most frequent errors in Google Translate were 'omission errors' and 'semantic errors', while Papago predominantly showed 'omission errors'. This pattern indicates that 'omission errors' were a common issue across both translation engines. The significance of this paper lies in its contribution to understanding the accuracy of Vietnamese-Korean machine translation in the context of Vietnamese diplomatic communications, a relatively under-researched area in Vietnam.