Mapping the Terrain: A Scoping Review of Artificial Intelligence Tool Classification Frameworks in English Language Teaching

Abstract

While the number of artificial intelligence (AI) tools has been growing at an extraordinary pace in the field of English Language Teaching (ELT), it has been difficult to remain organised and clear in this field, where the tools come in all shapes and sizes. The need to capture a shared perspective on what a given AI tool is, how it is described, and how it is used has become clearer for researchers and educators alike. This is a scoping review based on the framework recommended by Arksey and O'Malley and guided by the PRISMA-ScR guidelines, covering AI tools in ELT and related language-education fields between 2016 and 2026. After systematically searching the open-access literature from Scopus, Wiley, Springer, and DOAJ, 22 studies were included. The review identifies five recurring classificatory logics across the studies: function-based - classifications based on the language skill supported; technique-based - classifications based on the primary AI technique, such as automatic speech recognition or natural language processing; educational-role-based - classifications based on the interaction between the learner and the AI, such as tutor or tool; ethical-spectrum-based - classifications based on the degree of learner agency preserved; and literacy-oriented - classifications based on the capabilities that learners and educators must build. The review shows that most frameworks are single-study proposals derived from a small number of tools, with limited testing across contexts. Few frameworks are ELT-specific, and there is little cross-testing between frameworks. It concludes with a synthesis that maps a shared spectrum of classificatory dimensions and outlines an agenda for a more empirically grounded ELT classification.