Large language models (LLMs) as co-constructors of linguistic knowledge: a socio-cognitive analysis of AI-mediated English learning
Year: 2026
Type: article
Abstract: The emergence of Large Language Models (LLMs) such as GPT-4 has triggered a fundamental shift in computer-assisted language learning, moving from rule-based or narrowly interactive systems to generative, dialogic agents that can engage learners in open-ended linguistic exploration.This paper offers a sociocognitive analysis of LLMs as potential co-constructors of linguistic knowledge, drawing on Vygotskian concepts of mediation, the Zone of Proximal Development, and the More Knowledgeable Other (MKO).Synthesising recent empirical research on LLM-mediated English learning, we examine how learners interact with generative AI for vocabulary acquisition, emotional regulation, and motivation, and how these interactions are shaped by perceived usefulness, social influence, and learner autonomy.The analysis reveals that while LLMs can provide contingent scaffolding and lower the affective filter, their algorithmic nature lacks genuine intersubjectivity and shared intentionality.We further explore emergent cognitive capacities of LLMs-particularly analogical reasoning-as a resource for deeper conceptual understanding.The paper concludes that effective integration of LLMs requires rethinking task design and AI literacy, positioning the model not as a replacement for human teachers but as a mediational tool within a pedagogically structured sociocognitive environment.
Source: Гуманитарный научный журнал
Language: English
Subfield: Language and Linguistics
Field: Arts and Humanities
Domain: Social Sciences
SDG: Quality education
Open Access status: green
OpenAlex ID: W7223482886