ChatGPT in English-Arabic Translation: Learner Perceptions and the Treatment of Humour and Empathy

Abstract

Although English is truly the global lingua franca today, machine translation (MT) of paralinguistic elements like humour and empathy is not totally seamless in the English­Arabic language pair. This study examined the effectiveness of ChatGPT, a generative artificial intelligence (AI) model, in translating these elements, and how Saudi translation students who use AI tools for translation perceive its performance. The study applied a mixed­methods approach by gathering quantitative data through a Likert­scale questionnaire from 46 undergraduate translation students at Qassim University, Saudi Arabia, and qualitative data was gathered through an analysis of texts translated by ChatGPT. The questionnaire was validated and tested, demonstrating high reliability with a Cronbach’s alpha of 0.670. The findings showed that, overall, translation output in ChatGPT was strong, and users have high confidence in the tool for formal and academic translation, with a high average score of 4.00. However, its performance in conveying humour (mean = 3.33) and empathy (mean = 3.37) is moderate, which can lead to neutral or culturally mismatched results. The study concludes that while ChatGPT is a useful educational tool for improving accessibility and motivating translation practice, it needs significant improvements in cultural and emotional intelligence. Additionally, the language dataset can be enhanced with rich Arabic content. Despite the mixed results, it can be recommended that educators critically incorporate ChatGPT into curricula, while simultaneously focusing on post­editing and cross­cultural comparison exercises for best practices.

Ähnliche Arbeiten