AI-Mediated English Language and Literature Education: Digital Literacy, Cultural Identity, Student Engagement and the Moderating Role of Architectural Learning Environments

Abstract

Generative AI (GAI) has come into English classrooms before the evidence base to regulate it, and nearly all the evidence surrounding GAI treats the classroom as a neutral receptacle. This study examined whether the benefits of AI-mediated instruction are related to the architecture of the space in which the instruction is delivered, and what the instruction does to learners' cultural identity. The study employed a 16-week quasi-experimental design with convergent qualitative strand with four universities of public sector in Punjab, Pakistan. Twenty-four intact undergraduate sections (N = 597 students) were assigned to AI mediated or conventional instruction, in either an Adaptive Learning Studio (reconfigurable furniture, distributed display surfaces, high illuminance, low reverberation) or a Conventional Lecture Room. Outcomes were literary interpretive competence (blind double-rated analytic essays), four-dimensional engagement, digital/AI literacy and three facets of cultural identity negotiation; indoor environmental quality was measured objectively in every room. AI-mediated instruction produced large gains in digital/AI literacy (η²p = .241, g = 1.10 [0.93, 1.27]), engagement (η²p = .182) and interpretive competence (η²p = .216), but the instruction × environment interaction was significant for attainment, F(1, 592) = 37.22, p < .001, η²p = .059: the AI effect was more than twice as large in the studio (g = 0.95 [0.71, 1.20]) as in the conventional room (g = 0.40 [0.17, 0.63]). The indirect effect of serial mediation (AI → digital literacy → engagement → attainment) was 0.88 points [0.67, 1.14] and the index of moderated mediation was 0.52 [0.10, 0.99]. While there was a slight increase in perceived cultural dilution (g = 0.21) and a slight decrease in heritage cultural affirmation (g = −0.27), there was a medium increase in awareness of AI and Global Englishes (g = 0.45), with both of the latter results mitigated by a stronger ethical AI-literacy (r = −.30, p < .001). The findings suggest that AI integration and learning space design should be considered as a single intervention and that the mediational and cultural pathways need to be experimented and demonstrate a correlational relationship.