TRÍ TUỆ NHÂN TẠO (AI) CÓ GIÚP TRẠNG QUỲNH “CHƠI CHỮ” ĐƯỢC BẰNG TIẾNG ĐỨC?

Abstract

This paper examines the limitations of Artificial Intelligence (AI) in translating Vietnamese humorous narratives, particularly those deeply rooted in Vietnamese cultural contexts. This literary genre requires a high level of sensitivity to culture, context, and linguistic nuance. The study analyzes a selection of Trang Quynh stories translated from Vietnamese into German by AI systems (ChatGPT and Gemini). These translations were subsequently evaluated by a group of German native readers. Based on an analysis of the AI-generated translations and a comparison with the readers’ feedback, the paper discusses cases in which AI encounters significant difficulties or where translation proves effectively “untranslatable”. Through the evaluation of AI-generated translations based on Reiß’s framework, as well as on assessments by native readers, this study highlights the major challenges that AI faces in handling the distinctive features of Vietnamese humorous narratives, particularly in dealing with wordplay. The results of this initial small-scale dataset indicate that the ability to convey the humorous function and express deeper, multi-layered meanings through the characteristic use of wordplay in Trang Quynh narratives remains limited. Furthermore, the study suggests potential directions for the use of AI in teaching and learning German–Vietnamese translation.