Mapping AI-mediated pragmatic competence in EFL: A systematic review of Human–AI interaction and communicative development

Abstract

The study of AI-mediated pragmatic competence in English as a Foreign Language (EFL) is theoretically divided and methodologically disproportionate models, and there is no pre-existing synthesis of the intellectual organization of the field. The current systematic review fills this gap by considering 361 articles in the Web of Science Core Collection (2021–2025). The study based on the improved bibliometric and science mapping methods (co-occurrence networks, thematic mapping, Lovaka Law and factorial analysis) through Bibloshyny shows the explosive annual growth (74.56%), most significant in China (168 records) and with the terms ChatGPT, artificial intelligence, and discourse. Results indicate that productivity of authors does not follow the Lotka law and thematic maps put Generative AI and Pragmatics as thematic motors. Findings suggest that the existing AI interaction can perhaps socialize learners into machine compatible hybrid pragmatic competences in lieu of entirely genuine competence. The review offers practical solutions to teachers and curriculum creators to intelligently incorporate AI without losing the sociocultural richness. It sets out a methodological model of applied linguistics that is replicable. The originality of the study lies in the fact that it is the first systematic multi-dimensional synthesis of AI-mediated pragmatic competence in EFL, as the individual studies have been replaced by a unifying knowledge structure.