AI and Academic Discourse: A Sociolinguistic Analysis of Language and Interaction

Abstract

The integration of artificial intelligence (AI), particularly large language models, into academic writing has reshaped scholarly communication by influencing language use, rhetorical practices, and authorial identity. This study employs Mediated Discourse Analysis (Scollon, 2001) to investigate AI as a mediational tool that co-constructs academic texts alongside human authors, redistributing agency and shaping interactional processes. Using purposively selected academic writing samples from postgraduate students and early-career researchers, both AI-assisted and human-authored, alongside reflective narratives, the study examines lexical choices, syntactic complexity, and Meta discursive markers. Findings reveal that AI-assisted texts demonstrate formal vocabulary, simplified syntax, and assertive stance markers, whereas human-authored texts exhibit nuanced hedging, complex syntactic structures, and interactive engagement. Participants negotiated AI suggestions to preserve disciplinary authenticity, epistemic stance, and authorial voice, highlighting AI’s dual role as both a facilitator and constraint in academic discourse. The study underscores the sociolinguistic, pedagogical, and institutional implications of AI integration, demonstrating its influence not only on textual outcomes but also on interactional practices and scholarly identity within academic communities.