FROM LANGUAGE SUPPORT TO ACADEMIC INTEGRATION: GENERATIVE AI IN ACADEMIC ENGLISH EDUCATION FOR CHINESE PhD STUDENTS IN UKRAINE

Abstract

The article provides a theoretical and methodological substantiation of the pedagogical potential of generative artificial intelligence (GenAI) in Academic English education for Chinese PhD students studying at Ukrainian higher education institutions. The problem is shaped by the convergence of three developments: the internationalisation of doctoral education, increasingly demanding expectations for English-medium research communication, and the rapid normalisation of generative language models i n academic writing practices. Drawing on a critical synthesis of recent scholarship in English for Academic and Research Purposes, international student adaptation, Chinese students’ multilingual practices, doctoral writing pedagogy and GenAI -assisted writ ing, the study argues that language support should not constitute the final pedagogical purpose of AI use. A four -level model, four stages: Language Support, Academic Literacy, Critical AI Literacy, and Academic Integration, is proposed. Within the model, GenAI gradually changes its role from a linguistic mediator and local editing assistant to a resource for metacognitive scaffolding, critical comparison of alternatives and preparation for autonomous participation in international academic communication. F ive pedagogical zones of GenAI use are specified: engagement with research literature; development of academic vocabulary and rhetorical patterns; planning, drafting and revising academic texts; preparation for oral research communication; and reflection on authorship, evidence, transparency and academic integrity. Particular attention is given to the situation of Chinese doctoral researchers, including movement between Chinese and English, differences in academic rhetorical conventions, unequal access to g lobal and domestic AI platforms, and the risk of technological dependency. The paper introduces the principle of pedagogical fading of AI assistance: as the doctoral researcher’s competence grows, direct text generation should decrease while verification, reasoned rejection of AI suggestions and independent authorial decision -making should increase. The model offers a framework for designing Academic English for Researchers, English for Research Publication Purposes and academic adaptation courses for international doctoral students in Ukrainian universities. The contribution is conceptual and methodological and is intended for subsequent empirical validation in authentic doctoral education settings.