Generative AI and linguistic creativity in digitally multilingual higher education

Introduction Generative AI is reshaping language education by turning text production, revision, genre transformation, and stylistic choice into forms of human–AI interaction. In digitally multilingual higher education, this opens possibilities for developing linguistic creativity, while raising concerns about standardization, algorithmic bias, overreliance on AI texts, and the weakening of learners’ authorial voice. Methods The study conceptualizes linguistic creativity as a four-component construct: text-production, figurative-expressive, transformational, and motivational. The empirical part was a quasi-experiment involving 88 first- and second-year students from three Ukrainian higher education institutions. The control group completed traditional tasks, whereas the experimental group worked with structured ChatGPT-mediated tasks. The data were analyzed using descriptive statistics, t-tests, and Cohen's d. Results Both groups showed positive dynamics, but the experimental group improved more strongly. The groups were comparable at the initial stage. At the second and final stages, the differences became statistically significant, with a strong practical effect by the end of the experiment. Qualitative observations suggested that changes were most visible in text production and genre-stylistic transformation, whereas changes in the figurative-expressive component were more moderate. Discussion The findings show that ChatGPT can support linguistic creativity when it is embedded in a structured pedagogical sequence: prompt formulation, generation of alternatives, comparison, selection, revision, and creation of a final learner-authored text. The study links linguistic creativity to specific AI-mediated task types and emphasizes AI literacy, academic integrity, multilingual sensitivity, and learner agency.