AUTHENTICITY AT STAKE: COMPARING SPEECH ACTS IN AI-GENERATED VS. HUMAN-AUTHORED ELT DIALOGUES

Abstract

This study examines the authenticity of English Language Teaching (ELT) dialogues by comparing speech acts in AI-generated and human-authored textbook materials. Pragmatic competence in ELT requires learners not only to master grammatical accuracy but also to use language appropriately in social contexts. However, textbook dialogues are often considered overly idealized and insufficient in representing real-life communication. This study employed a qualitative comparative content analysis design using six dialogues derived from two sources: the Indonesian senior high school English textbook Life Today and AI-generated dialogues produced through controlled prompts. The analysis focused on speech acts, pragmatic markers, and contextual authenticity. The findings indicate that AI-generated dialogues demonstrate greater pragmatic fluency, more varied interactional features, and higher contextual naturalness compared to textbook dialogues, which tend to be more rigid, didactic, and dominated by assertive functions. AI-generated dialogues also incorporate pragmatic markers and conversational features that more closely resemble authentic communication. These findings suggest that artificial intelligence has significant potential to support the development of more authentic and contextually relevant ELT materials in the digital era.