Discursive Constructions of AI in Research Article Abstracts: A Corpus‑Based Faircloughian CDA of Enhancement, Disruption, and Tensions in Applied Linguistics and Beyond

With the advent of ChatGPT, scholarly articles on artificial intelligence (AI) have increased dramatically, yet there remains a paucity of studies exploring how researchers, especially in the field of applied linguistics and L2-related fields, conceptualize the opportunities and threats of AI in the compact genre of research article abstracts. This study drew on a corpus of 360 research article abstracts from six disciplines, such as Applied Linguistics and other hard and soft sciences, to reveal how AI is discursively represented in the field of L2/applied linguistics and, in comparison, to other disciplinary contexts. Enhancement framings, which focused on efficiency, innovation, and accessibility, were adopted by more than 80 percent of the abstracts, and disruption framings, which were associated with ethics, integrity, and equity, were less common and were often in reactive relation to enhancement narratives. The two cross-cutting tensions that were apparent in the disciplines were: Innovation vs. Integrity and Equity vs. Inequality, which were articulated through specific lexical and rhetorical tools. Drawing on a corpus-based Faircloughian Critical Discourse Analysis (CDA) combined with a reflexive thematic analysis (RTA), this study revealed how fine-grained (micro-level) linguistic choices evolved into overarching (macro-level) ideological orientations and reproduced specific power dynamics, which manifested a form of sedimented techno-optimism that constrained the possibilities of critical reflection. The findings carry implications for L2 teacher education, multilingual equity, and AI literacy in applied linguistics, as well as for policy, pedagogy, and AI tool design across the broader academy.