Exposing ChatGPT-Assisted Plagiarism in Student Assessment in Higher Education: Linguistic and Non-linguistic Clues

ChatGPT offers second language writers’ limitless opportunities for engagement with this technology. This exploratory study focuses on Gen-AI academic plagiarism in the context of unsupervised written assessment and pursues the concept of opportunity in traditional academic fraud theory, in an attempt to evaluate its applicability to Gen-AI academic plagiarism. The research questions aimed to gather perceptual and textual observations by departmental faculty regarding their students’ unpermitted use of ChatGPT in written assessments. Thirteen experienced faculty members in the English and Translation Department at a publicly funded university in the Sultanate of Oman completed a questionnaire asking them to identify clues of plagiarism evident in their students’ written work. Additionally, assessment artefacts in the form of 15 student-authored literature reviews were examined in search of these clues. Using an inductive, mixed-methods approach, the analysis drew on faculty members’ growing understanding of the affordances of large language models, coupled with their situated knowledge of their students’ writing abilities in terms of the lexico-grammatical and discoursal features characterising their submitted texts. The findings were summarized in a model which highlighted the interrelationships amongst the various factors leading to writer disengagement principally manifested through language, subject-matter, and behavioural clues. The paper concludes by adopting a utilitarian, pragmatic perspective on academic plagiarism, with a view to transforming these limitations into opportunities for writer engagement and, ultimately, learning.