Rethinking discourse in secondary school English lessons through Translanguaging practices and AI-mediated generative facilitation

Abstract

The article explores new approaches to organizing verbal interaction in high school English classes under conditions of limited spontaneous oral practice and high cognitive load on students. The aim of the study is to assess the impact of a combined methodology that integrates meaningful use of the native language and generative support provided by an artificial intelligence system on the quantitative and qualitative parameters of classroom discourse. The study involved one hundred and sixty ninth-grade students with a B1 level of English proficiency, divided into one control and three experimental groups with different modes of work. The research methods included a pedagogical experiment with longitudinal observation, audio recording and transcription of lessons, discourse analysis, statistical processing of indicators such as the proportion of student and teacher speaking time, turn-taking frequency, lexical diversity index, and mean utterance length, as well as a questionnaire using a Likert scale. The results showed that allowing purposeful use of the native language at the stages of speech planning and idea discussion by itself leads to a noticeable increase in student speaking time and a reduction of perceived cognitive load, while the inclusion of a generative facilitator based on an artificial intelligence model further enhances the dynamics and polylogical nature of interaction by posing open-ended questions and modeling role situations. The strongest effects were observed in the group where both approaches were combined: here, student speaking time exceeded half of the lesson duration, lexical diversity and syntactic complexity indicators increased significantly, and students reported substantial growth in communication confidence and engagement. The novelty of the work lies in the integration of pedagogical practices and digital technologies within a single experimentally validated model of classroom interaction, which views each student utterance as a joint product of human and machine agents, thereby introducing new criteria for evaluating lesson success and communicative environment parameters. The article also discusses the potential scalability of the proposed model to other levels of foreign language education and subject areas, as well as the need for further research on long-term effects and ethical aspects of employing generative systems in school practice across diverse sociocultural contexts of contemporary education.