Answers are grounded in this paper and cite their sources.
English for Academic Purposes (EAP) has traditionally focused on helping students enter disciplinary discourse communities yet has largely overlooked explanatory competence. This gap is particularly pronounced at cross-disciplinary universities where students must articulate specialized concepts across diverse audiences. This study investigated the role of AI-mediated dialogue in helping EAP students enhance their explanatory competence, as manifested in their metaphor refinement across three dimensions: clarity, accuracy, and audience adaptation. A three-week intervention was conducted with 41 first-year undergraduate EAP students at a cross-disciplinary university in Guangzhou, China. Students refined their metaphorical explanations of the term ‘recursion’ under two conditions: Group A (n = 26) engaged in AI-mediated dialogue via ParallelX.cn, an AI-powered educational platform, featuring branching conversation prompts and peer feedback, while Group B (n = 15) refined independently. Results showed that both groups reported positive gains in perceived explanatory competence. In comparison, however, Group A scored slightly higher than Group B in perceived clarity, accuracy, and audience adaptation, while also rating higher AI’s capacity to balance simplicity with accuracy though this difference did not reach statistical significance. Group B valued independent reflection but also expressed a desire for additional support. By contrast, Group A registered slightly higher scores in their belief that AI-mediated dialogue facilitated their deep thinking and metaphorical clarity. Qualitative analysis of AI dialogue logs further suggested iterative refinement and increased technical precision, consistent with the survey findings. These findings suggest that AI-mediated refinement can support the development of metaphor-mediated explanatory competence, with implications for EAP pedagogy and AI-enhanced language learning.
No takes yet. Share an insight, caveat, or question.
Xing et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: