Evaluating Artificial Intelligence (AI) Translation Performance in Diglossic Iraqi Courtroom Discourse: A Comparative Study of ChatGPT and Microsoft Translator (MST)

Mkhlaf, Asst. Lect. Fatimah Awad; Abdullah, Asst. Lect. Osamah Mohammed

doi:10.25130/Lang.10.2.P2.19

Evaluating Artificial Intelligence (AI) Translation Performance in Diglossic Iraqi Courtroom Discourse: A Comparative Study of ChatGPT and Microsoft Translator (MST)

Document Type : Original Article

Authors

Asst. Lect. Fatimah Awad Mkhlaf ¹

Asst. Lect. Osamah Mohammed Abdullah ²

¹ Department of English Language, College of Arts, Tikrit University, Tikrit, Iraq

² Department of Translation, College of Arts, Tikrit University, Tikrit, Iraq

https://doi.org/10.25130/Lang.10.2.P2.19

Abstract

This research examines how artificial intelligence translation systems deal with diglossic language in Iraqi courtroom discourse. Iraqi courtrooms commonly involve a mixture of Modern Standard Arabic and Iraqi Colloquial Arabic, which creates serious challenges for machine translation systems. The study compares the performance of ChatGPT and Microsoft Translator (MST) in translating 30 instances of authentic Iraqi courtroom dialogues collected from publicly available trial videos on YouTube. The data were manually transcribed and analyzed using both qualitative and quantitative methods, with special focus on diglossic shifts and pragmatic meaning. The results show that ChatGPT achieved higher accuracy in interpreting diglossic shifts, correctly identifying about seventy three percent, while MST achieved about sixty seven percent. However, both systems failed to provide fully reliable sentence level translations suitable for legal use. The findings confirm that current AI translation tools cannot be safely used independently in courtroom settings and must be supported by human translators to avoid serious legal misunderstandings.

Keywords

Arabic diglossia

Iraqi courtroom discourse

Machine translation

ChatGPT

Microsoft Translator (MST)