AI technologies in teaching foreign languages to children with speech disorders

Abstract

Teaching a foreign language to learners with speech disorders faces a contradiction: adapted programmes assume acquisition of a second language, while the teacher's toolkit is designed for a typically developing child, and AI tools are applied piecemeal, without regard to the type of speech defect. The paper aims to build a didactic classification of AI technologies for foreign-language teaching of children with speech disorders and to propose a model for embedding them into an adapted programme. Using conceptual-categorial analysis, structural-functional modelling and content analysis of regulatory documents, five functional classes are identified – diagnostic-adaptive, pronunciation-training, meaning-supporting, dialogue-generating and assistive-compensatory, each matched to a barrier produced by a particular group of disorders. A five-circuit model of application is proposed, aligned with the Federal State Educational Standard for learners with disabilities, the federal adapted programme and the standard GOST R 71657-2024. Distinguishing tools by their leading didactic function shows that the effect of AI is achieved by an accurate match between the instrument, the child's speech profile and the planned result; the correctional load is carried by multimodal duplication of content.