Journal Paper Published

Abstract

The increasing use of large language models like ChatGPT has sparked interest in their potential for translation tasks. However, little is known about what drives university students to adopt these tools or how disciplinary background and prior experience shape their decisions. Guided by the Unified Theory of Acceptance and Use of Technology (UTAUT), this study explores the adoption of ChatGPT for translation tasks among university students in Hong Kong. Survey responses from 308 students, including translation and non-translation majors, were analyzed using structural equation modeling. Results show that performance expectancy is the strongest determinant of adoption intention, followed by facilitating conditions, while effort expectancy and social influence were less significant. Experience level emerged as an important moderating factor: novice users relied on both social influence and performance expectations, whereas experienced users prioritized performance alone. Disciplinary differences were also pronounced. Translation students primarily valued performance benefits and used their technical expertise to evaluate ChatGPT independently. Non-translation students, however, were influenced by both performance expectations and facilitating conditions, suggesting a greater need for institutional support. These findings highlight the importance of tailored educational approaches that address the specific motivations of different student populations. For translation students, this means emphasizing advanced features and critical evaluation, while for non-translation students, it involves providing stronger support systems and guidance. The study also offers insights for LLM developers, underscoring the need for user-centered design that accommodates the diverse needs, experiences, and expectations of different student groups.

Keywords

generative AI, language learning, large language models (LLMs), technology acceptance, translation education