The Linguistic ‘Fingerprint’ of Academic Texts
Yue Shi Yang
by This research compares an AI-simulated academic interview (ChatGPT-4) with a genuine human interview (Chinese) using qualitative discourse analysis. The goal is to identify linguistic and discursive "fingerprints" of authorship. The AI transcript is highly cohesive and refined, featuring a formal rhetorical style with polished paraphrasing. However, it critically included false information (made-up anecdotes), raising serious issues about academic honesty. In contrast, the human transcript is spontaneous and disjointed, marked by disfluencies, pauses, and overlapping speech. It exposes a developing authorial voice and a greater degree of critical involvement and emotional depth. The study questions the belief that polished rhetoric indicates true authorship or intellectual validity. It demands a critical examination of AI-produced academic writing and calls for institutional policies that address both the stylistic fluency and factual integrity of authorship in the generative AI era.