ChatGPT's Translation of Medical Terms: Techniques, Shifts and Acceptability

Health articles contain various medical terms that require appropriate terminology choices. Several local media cite information from English health articles, which necessitates accurate translations. Translating medical terms has become easier with the help of artificial intelligence (AI) such as ChatGPT. However, machine translation is only used in the initial step of translation process as meaning distortion and ambiguity may occur. This research focuses on the medical terms translation by ChatGPT in the form of nouns and noun phrases from English to Indonesian, with data obtained from two health articles from Mayo Clinic website. The translation was analyzed based on the theories of 18 translation techniques by Molina & Albir, translation shift by Catford and translation acceptability by Nababan et al. A descriptive-qualitative approach is applied to obtain deeper insight of medical terms translation using AI. The results show that out of 120 data, most medical term translations apply established equivalent (47 data), followed by literal translation (41), calque (21), borrowing (18), linguistic amplification (7), adaptation (6), description (4), and amplification (1). From the overall data, there are 107 shift-indicated translations and 13 no-shift-indicated translations. Through this study, ChatGPT can be considered as reliable machine translation for translating medical terms. However, it is important to note that AI-tool only serves to assist in enhancing time effeciency, and translators must review the output thoroughly since errors may occur.

Keywords: medical terms translation, ChatGPT translation, health articles, translation techniques, translation shifts, translation acceptability