When does ChatGPT help or hinder? A scoping review of influencing factors in college EFL learners' affective and...

Existing reviews have established the overall direction of ChatGPT's effects on college English as a Foreign Language (EFL) learners' affective and cognitive experiences but have not systematically addressed under what conditions these effects arise. This scoping review maps the influencing factors reported in 26 empirical studies examining ChatGPT and closely related LLM-based generative AI tools as shaping these experiences. A systematic search of Web of Science, Scopus, and ERIC following PRISMA-ScR guidelines and the (Arksey and O'malley, 2005) framework yielded 829 records, with 26 studies meeting inclusion criteria after deduplication and two-stage screening. Influencing factors were extracted from each study and organized into learner-level, instructional-level, and technology-level categories based on the Cognitive-Affective Theory of Learning with Media. Ten types of influencing factors were identified. Evidence density varied markedly across the three levels, with learner-level factors supported by the largest body of research and technology-level factors by the smallest. Self-efficacy showed consistent mediating pathways across eight studies, while metacognitive awareness produced the most clearly documented conditional differentiation between strategic and surface-level ChatGPT use. Structured instructional integration was the factor with the most consistently positive effects across the three studies that examined it. The influencing factor mapping table reveals that the effects of ChatGPT on EFL learners are not uniform but contingent on the interaction among learner characteristics, instructional design, and technology use patterns, supporting differentiated rather than uniform approaches to ChatGPT integration in college EFL instruction.