인공지능(AI) 번역기의 한일 번역 양상에 관한 고찰―파파고(papago)에서의 명사를 중심으로―

Abstract

In this study, translations by an artificial intelligence translator of newspaper editorials from Korean to Japanese were analyzed to examine their accuracy and the patterns of errors. Naver's Papago was used as an AI translator targeting Korean newspaper editorials to determine the accuracy and quality of Papago translations into Japanese, focusing on nouns. The review divided the nouns into those that matched in the Papago translation and the Japanese editorial translation and those that did not. On average, more than 80% of the words were consistent. Among the words in which the Papago translation and the Japanese translation were inconsistent, we confirmed that authentic Japanese expressions were not reflected in the Papago translation, and most of these translations were close to the literal translation of the Korean text. Moreover, a few mistranslations were confirmed. Although this study was limited to nouns, it confirmed that Papago translations from Korean to Japanese have improved significantly in terms of accuracy and quality.