Authorial Voice in AI-Assisted Academic Writing Among EFL University Students in Multilingual Contexts

Artificial intelligence (AI) writing tools have drastically changed academic writing in higher education, but little is known about how they affect authorial voice, a crucial aspect of academic identity, especially when it comes to English as a Foreign Language (EFL) learners from minority linguistic communities. This mixed-methods study examines how EFL university students use ChatGPT and Grammarly, two AI-assisted writing tools, to recognize, negotiate, and preserve their authorial voice in multilingual peripheral contexts. The study included a validated questionnaire, semi-structured interviews, and textual analysis of student writing samples from a sample of ۲۴۰ undergraduate and graduate students from three public universities. Descriptive statistics, independent samples t-tests, and one-way ANOVA were used to assess quantitative data, while theme analysis was applied to qualitative data. Significant conflicts between the maintenance of individual authorial identity and AI-mediated textual efficiency are revealed by the findings, with female students and postgraduate learners exhibiting higher metacognitive awareness of voice displacement. Stylistic homogenization, diminished epistemic confidence, and deliberate code-switching between AI and human expression emerged as the three main themes. In order to explain how EFL writers in peripheral multilingual environments manage AI reliance, the study offers a novel theoretical framework called the Authorial Voice Negotiation Model (AVNM). There is discussion of the pedagogical implications for teaching writing in EFL contexts.

کلیدواژه ها:

authorial voice ، AI-assisted writing ، ChatGPT ، EFL learners ، academic writing ، identity ، higher education ، multilingual contexts