AWE and LLMs in L2 Writing Feedback: An Exploration of EFL Learners’ Experiences

In second language (L2) writing instruction, providing real-time personalized feedback at scale within traditional classroom settings poses significant challenges for teachers. Technology tools such as the established Automated Writing Evaluation (AWE) and the emerging Large Language Models (LLMs), with their respective strengths, have emerged as viable solutions. To investigate leaners’ experiences with these two distinct computer-generated feedback types, this study examined the cognitive load and perceptions of English as a Foreign Language (EFL) students when using AWE and LLMs for writing feedback. A between-group experiment with 76 Chinese university students revealed that both tools induced low to moderate cognitive load, indicating favourable technology acceptance of both tools. Students’ perception ratings for LLMs were slightly higher than those for AWE across all three dimensions. Qualitative data also indicated a strong student preference for LLMs due to their detailed, adaptive, and interactive feedback, though some still favoured AWE for its simplicity. The findings supporting the previous research results that LLMs exhibit significant potential in providing multi-dimensional feedback, AWE remains relevant for certain learners. The study highlights the need for tool selection based on learner proficiency and the potential benefits of integrating Generative AI with traditional AWE methods in L2 writing instruction.