A Study on the Usability of Korean-Arabic Machine Translation for Speeches: Comparison of Google, ChatGPT, and Papago

This study compares the usability of machine translation tools—Google Translate, ChatGPT, and Papago—for Korean-Arabic translation and examines their potential applications. Focusing on the Korean-Arabic language pair, the study analyzes speeches commonly used in translation and interpretation practice. The dataset consists of five Korean speeches delivered by President Yoon Suk-yeol during his state visits to Saudi Arabia and Qatar in October 2023, totaling 2,639 words (11,269 characters) translated into Arabic. The translations were evaluated based on fidelity, fluency, formality, and pragmatics. The analysis revealed that Google Translate produced the highest quality translations with the fewest errors in fidelity and fluency. ChatGPT exhibited frequent omissions from the source text but performed well in fluency, formality, and pragmatics, showing potential for simultaneous interpretation, albeit with necessary source-text verification. In contrast, Papago displayed significant distortions and the lowest quality in fluency and formality, highlighting its limitations for interpretation purposes.