Track: Machine Learning

Abstract

Machine translations are commonly used for translation, but the result may not be correct, and the user might not know if the mistranslation happened. The research aims to identify the translation result of Japanese onomatopoeia that was translated using google translate to Indonesian, the difference between the google translate result and the translated version, and the cause of mistranslation. The qualitative method through library research was applied. The data source was from the comic entitled Komi Can't Communicate by Tomohito Oda. Translation results would be analyzed using the descriptive analysis method. It can be concluded that the research found that google translates onomatopoeia translation had differences from the Indonesian version from the difference in word usage to mistranslation due to language ambiguity and missing data. The author also finds the possibility of the lack of Japanese onomatopoeic data on google translate or the lack of data for translating Japanese onomatopoeic to Indonesian, which causes translation errors. In addition, the authors also found that the translation technique used by Google Translate in translating onomatopoeia mainly uses the word-for-word translation method so that it produces one word from each translation process. However, there is also translation using the free translation method, which uses two words to translate one onomatopoeia so that the translation results can be better understood in the target language.