When AI Speaks the Student's Language: Toward a Theory of Linguistically Responsive Generative AI in African Higher Education
TL;DR
The paper proposes the Linguistic Mediation Proposition, which posits that the educational value of GenAI is partly determined by the alignment between learners’ linguistic repertoires and the linguistic responsiveness of AI-mediated learning environments.
Abstract
Generative artificial intelligence (GenAI) is rapidly reshaping higher education, yet dominant approaches to AI-enabled learning often treat language as a neutral medium of interaction rather than a constitutive dimension of learning, identity, knowledge construction, and educational participation. This assumption is particularly problematic in African higher education, where multilingualism, code-switching, linguistic hierarchies, and unequal recognition of indigenous languages shape students’ educational experiences. This conceptual paper advances Linguistically Responsive Generative AI (LR-GAI) as a theoretical construct and develops a theory of how linguistic responsiveness can shape the educational value of GenAI in multilingual contexts. Drawing on Sociocultural Theory, Translanguaging Theory, and the Capability Approach, the paper argues that access to GenAI does not necessarily translate into equitable educational capability when learners’ linguistic repertoires are poorly accommodated by AI systems. It distinguishes linguistic responsiveness from simple multilingual processing or translation and conceptualises LR-GAI through five interrelated dimensions: linguistic accessibility, multilingual responsiveness, semantic and cultural contextualisation, pedagogical responsiveness, and linguistic agency. The paper proposes the Linguistic Mediation Proposition, which posits that the educational value of GenAI is partly determined by the alignment between learners’ linguistic repertoires and the linguistic responsiveness of AI-mediated learning environments. From this proposition, the paper develops theoretically grounded propositions linking linguistic responsiveness to learner–AI interaction, knowledge construction, epistemic and linguistic agency, and educational participation. The resulting framework challenges language-neutral models of AI in education and provides a theoretical foundation for designing, evaluating, and governing GenAI systems that support linguistically diverse learners in African higher education.
