GenAI-Mediated English Text-Work Breadth and Localized English Acceptance in Teaching: The Role of Linguistic-Norm Awareness

Abstract

This study investigated whether pre-service English teachers’ breadth of generative artificial intelligence (GenAI)-mediated English text work is associated with their awareness of linguistic norms embedded in GenAI suggestions and with their subsequent acceptance of localized English in teaching. By focusing on the micro-level mechanism linking everyday GenAI-assisted revision to pedagogical dispositions toward linguistic plurality, the study addresses a tension extending beyond China to teacher education worldwide, as GenAI-driven standardization conflicts with the value of local Englishes. Its novelty lies in testing linguistic-norm awareness as a mediator through a temporally separated design, rather than assuming exposure alone transforms attitudes. The study surveyed 316 Chinese pre-service teachers across three waves, controlling for baseline acceptance, and estimated associations via confirmatory factor analysis and a baseline-adjusted path model. Results showed that broader GenAI text work was reliably associated with linguistic-norm awareness. This awareness, in turn, had a statistically reliable but substantively small indirect association with later acceptance. The direct path from text-work breadth to acceptance was negligible, and baseline acceptance accounted for most variance, which indicates high attitudinal stability. These findings suggest that broad GenAI text work may make normative bias noticeable. However, noticing alone is insufficient, and teacher education must transform it into principled comparison of standard, idiomatic, and localized forms. Teacher education programs should therefore use GenAI-mediated text work as a starting point for such comparison, because exposure alone cannot change entrenched attitudes and principled comparison is needed for inclusive teaching.