Evaluating the Arabic–Indonesian Translation Quality of ChatGPT on Quranic and Hadith Texts

Abstract

This study examines the reliability of ChatGPT in translating Arabic–Indonesian religious texts, addressing the concern that AI often struggles with culturally embedded, legal, and theological expressions as highlighted in recent translation studies. The issue is significant because Qur’anic and hadith texts carry normative meanings whose alterations may lead to doctrinal misinterpretation. Data were collected from ten Qur’anic legal verses and ten hadiths from al-Arbaʿīn al-Nawawiyyah, translated by ChatGPT using a standardized prompt. The translations were evaluated through Nababan et al.’s accuracy–acceptability–readability model by expert reviewers and general readers. Findings show that ChatGPT performs well in accuracy and readability but displays limitations in rendering legal terminology, cultural nuance, and doctrinally sensitive expressions. This research contributes empirical evidence on the strengths and risks of AI-assisted translation for Islamic texts and underscores the necessity of expert post-editing to ensure theological precision and culturally appropriate interpretation.

Ähnliche Arbeiten