Human-AI Academic Writing, Integrity, and Language-Communicative Culture in Teacher Education: Evidence from Ukrainian Student Texts

Abstract

Generative artificial intelligence has turned academic writing into a hybrid communicative process in which human authors interact with language models, automated editing systems and digital source environments. For teacher education, this shift is not only a question of academic misconduct; it is a question of language-communicative culture, authorship, source accountability and professional responsibility. The article examines how Ukrainian student academic texts reveal linguistic, compositional, terminological and integrity-related risk zones that become especially significant in AI-mediated academic writing. The empirical material consists of 26 anonymised student texts: 12 course papers, 8 master’s theses and 6 research articles. The study applies qualitative content analysis with descriptive quantification across four textual categories: lexico-stylistic normativity, logical-compositional cohesion, citation and source use, and terminological control. The findings show that only 19% of texts demonstrated a high level of language-communicative culture, while 54% were sufficient and 27% were low. Lexical and stylistic violations were identified in 65% of texts, citation and source-use problems in 38%, cohesion problems in 35%, and terminological imprecision in 19%. The article argues that these weaknesses should not be treated as isolated language errors but as indicators of preparedness for responsible human-AI academic writing. The study’s contribution is a text-centred framework that connects academic literacies, language-communicative culture, and AI-use transparency in teacher education.