A Critical Discourse Analysis (CDA) of ChatGPT’s responses to English grammar learning queries

Abstract

This study investigates how ChatGPT constructs knowledge, authority, and pedagogical discourse when responding to English grammar learning queries. While generative Artificial Intelligence (AI) tools are increasingly adopted in English as a Foreign Language (EFL) and English Education contexts, existing literature heavily favors technical efficacy and immediate pedagogical utility over critical examination of AI-generated content. To address this research gap, this study applies Norman Fairclough’s Three-Dimensional Framework of Critical Discourse Analysis (CDA) to scrutinize ten purposively sampled ChatGPT-4o responses to common English grammar learning queries (e.g., tense distinctions, passive voice mechanics, and article usage). Textual analysis reveals a pervasive reliance on high-affinity epistemic modality and imperative syntactical structures, positioning the AI as an absolute, objective authority rather than an assistive digital tool. Discursive practice analysis indicates that ChatGPT synthesizes standard prescriptive grammar textbooks while masking institutional authorship and algorithmic biases through naturalized, conversational interactions. At the level of social practice, this discourse reinforces traditional, Monolingual Anglo-centric linguistic norms and perpetuates deficit-based framing of EFL learners, undermining contemporary critical pedagogy that values linguistic diversity and translanguaging. The findings underscore the necessity for "Critical AI Literacy" among pre-service teachers and language learners to deconstruct the neutralized power dynamics and hidden ideologies embedded within algorithmic educational discourse.