Gemini-Assisted Task-Based Academic Writing: Exploring EFL Students’ Reasoning Processes During Argumentative Writing
DOI:
https://doi.org/10.36312/jllet.v1i2.77Keywords:
Argumentative writing, Generative artificial intelligence, Google gemini, Reasoning processesAbstract
The increasing integration of generative artificial intelligence into higher education has created new opportunities and concerns regarding academic writing, particularly for English as a Foreign Language (EFL) learners who must coordinate linguistic accuracy with complex argumentative reasoning. This qualitative study investigated how EFL university students used Google Gemini during task-based argumentative writing and examined the reasoning, decision-making, and challenges that emerged when learners interacted with AI-generated suggestions. Seventeen EFL learners at Tshwane University of Technology, South Africa, were purposively selected based on their experience with online learning and academic writing. Data were collected through classroom observation and semi-structured interviews and analyzed iteratively through data condensation, data display, and conclusion drawing. Findings showed that Gemini served different functions across writing stages: exploratory idea generation during planning, linguistic and rhetorical scaffolding during drafting, critical scrutiny during evaluation, and selective integration during revision. Students demonstrated evaluative reasoning, source verification, rhetorical judgment, comparison, and metacognitive monitoring when accepting, rejecting, modifying, or checking AI-generated suggestions. However, challenges included ineffective prompting, factual uncertainty, possible overreliance, and ambiguity concerning authorship and academic integrity. The findings suggest that Gemini supports higher-order reasoning only when learners actively interrogate its outputs rather than accept them uncritically. The study concludes that effective Gemini-assisted writing requires explicit instruction in prompt literacy, evidence verification, reflective revision, and ethical AI use. Its contribution lies in shifting attention from final writing performance toward the interactional processes through which EFL learners negotiate AI-generated input and maintain epistemic responsibility.






