Exploring Translation Strategies of ChatGPT and an English Native Speaker for Korean Onomatopoeia and Mimetic Words

This study aims to identify key characteristics and factors in the translation methods of ChatGPT and an English native speaker when translating into English 100 Korean sentences comprising 50 onomatopoeia and 50 mimetic words. The analysis revealed several key findings: Firstly, both ChatGPT and the English native speaker preferred translating Korean onomatopoeia into verbs, whereas they favored translating mimetic words into adjectives or adverbs. The native speaker did this 40% of the time, compared to 32% for ChatGPT for onomatopoeia, and 72% versus 52% for mimetic words. Secondly, ChatGPT replaced Korean onomatopoeia with English equivalents at a rate of 22%, nearly double that of the native speaker's 12%. This suggests ChatGPT prioritized faithfulness to the original text, while the native speaker considered the socio-cultural context of onomatopoeia use in English. Thirdly, explicative paraphrase was the most consistently used technique for both types of word. Fourthly, omission was frequently employed in translation, often correlating with the choice of adjectives or adverbs by both ChatGPT and the native speaker. Finally, ChatGPT used transliteration for onomatopoeia, while neither translator used mimetic word techniques.