Traduction et Langues
Volume 25, Numéro 1, Pages 135-156
2026-06-30
Exploring The Role Of Ai In Diplomacy-related Translation Tasks: An Empirical Study On Chatgpt's Potential Use In Diplomatic Texts’ Translation
Authors : Akgün Yaşar . Yetkin-karakoç Nihal . İşi Mirza .
Abstract
Diplomatic language, with its culturally sensitive and formally stratified character, presents distinctive challenges for translation. Although artificial intelligence (AI) and large language models (LLMs), such as ChatGPT, have shown considerable potential in technical and pragmatic translation tasks, their applicability to the specialised domain of diplomatic translation remains underexamined. This study addresses this gap by empirically assessing ChatGPT’s performance in translating diplomatic texts in comparison with human translators. Three diplomatic texts—two translated from English into Turkish and one from Turkish into English—were assigned to five third- and fourth-year undergraduate students enrolled in the English Translation and Interpreting Department at Izmir University of Economics, all of whom had completed a course in political text translation. The same texts were translated by ChatGPT (GPT-4o) using the prompt: “Translate the following texts with a diplomatic tone.” To ensure blind evaluation, all translations were handwritten and anonymised before submission. A jury consisting of two scholars in international relations and one retired ambassador evaluated the translations using a five-criterion rubric. ChatGPT obtained the highest overall score among the six participants and outperformed the five human translators across all three texts. The margin was narrowest in Text 3, translated from Turkish into English, yet ChatGPT still achieved the highest score. This result may be attributed to the formulaic and template-based nature of diplomatic correspondence, which aligns with the pattern-recognition capacities of LLMs trained on large, English-dominant datasets. The study suggests that ChatGPT has substantial potential for written diplomatic translation, particularly in texts conforming to established diplomatic conventions. These findings have implications for translation pedagogy, diplomatic communication, and the integration of AI tools into professional translation workflows. Future research should extend the inquiry to interpreting contexts and to more rhetorically complex diplomatic texts.
Keywords
Diplomatic Translation; ChatGPT; Large Language Models (LLMs); AI-Assisted Translation; Diplomatic Discourse; Translation Quality Assessment; English–Turkish Translation
Les articles similaires
Birem Fatima
.
pages 284-301.
Birem Fatima
.
pages 44-67.
