Human vs AI-Generated Texts in Language Learning: A Linguistic Comparison

Abstract

The rapid integration of artificial intelligence (AI) into education has significantly changed how we approach language learning and teaching. Even though AI-generated materials are becoming more common in English language instruction, there is still not enough linguistic research comparing these texts with those designed for textbooks. This study aims to fill that gap by comparing human-authored and AI-generated texts used in language learning. The goal is to identify the main linguistic differences between these two types of texts and assess their effectiveness as language-learning models. The study uses texts from High Note and Language Leader alongside those generated by ChatGPT on similar topics, all within the B1–B2 proficiency range. The analysis examines lexical richness, syntactic complexity, cohesion, and discourse organization using both quantitative and qualitative methods. The findings show that AI-generated texts exhibit greater lexical diversity and syntactic complexity; however, they often exhibit structural uniformity, overuse of cohesive devices, and limited pragmatic depth. In contrast, textbook-based texts demonstrate balanced grammar, controlled vocabulary selection, greater communicative authenticity, and a stronger fit with teaching goals. The study emphasizes the need to balance technological innovation with effective teaching in language education. The results suggest that while AI-generated texts can be helpful as supplementary resources, they should not replace professionally designed educational materials. This research contributes to ongoing discussions on artificial intelligence in English language teaching and offers insights into the linguistic and educational impacts of AI-assisted learning resources