h10
Abstract
The integration of generative artificial intelligence (AI) into academic writing practices has given rise to a new form of scholarly communication that demands critical investigation from the perspectives of applied linguistics and genre studies. This mixed-methods study examines the emergent genre of AI-mediated academic writing by analyzing three interconnected dimensions: 1) the instantiation of metadiscourse and stance markers in AI-generated texts compared to human-written counterparts, 2) the influence of multimodal design and perceived text origin on reader reading behavior, and 3) the dynamic, collaborative nature of the AImediated writing process. A parallel corpus of human and AI-generated texts was subjected to a Corpus-Assisted Discourse Study (CADS) using Swales' CARS model and a quantitative analysis of metadiscursive features. An eye-tracking experiment measured cognitive load and reading patterns across different text conditions. Finally, a longitudinal case study documented the evolution of prompting strategies and authorial voice. The findings revealed significant differences in the tonal profile of AI texts, increased cognitive load for readers, and the emergence of a distinct, iterative genre characterized by hybrid authorship. These results have profound implications for genre theory, academic literacy pedagogy, and the future of scholarly communication in the age of AI.