AI-ASSISTED URDU-MEDIUM PHONETICS INSTRUCTION IN RESOURCECONSTRAINED HIGHER EDUCATION  

Abstract

Generative artificial intelligence research in education has focused largely on student-facing applications, leaving teacherled, mother-tongue content creation in resource-constrained higher education underexamined. This exploratory mixedmethods case study examined a zero-cost workflow for creating Urdu-medium phonetics videos in a Government Graduate College for Women in Punjab, Pakistan. Conventional classroom teaching was followed by six short articulation videos scripted with Claude AI, generated as narrated multimedia presentations through Google NotebookLM, reviewed by the teacher, and distributed through YouTube links shared in a class WhatsApp group. Thirty female BS English students completed a post-intervention Google Form containing Likert-scale items and open-ended responses; available video comments and platform information were also considered. In the post-test, 93.3% of students agreed that visual elements enhanced comprehension, 83.3% reported increased confidence, and 76.7% reported mentally translating English phonetics terminology into Urdu. Eighty percent trusted AI-generated content when teacher-verified; 46.7% accessed the videos through shared family devices, 13.3% reported offline viewing or downloading, and 60% reported independent family access. These results are descriptive and exploratory. They are consistent with Cognitive Theory of Multimedia Learning, Cummins’ Interdependence Hypothesis, and microlearning as a design approach, but they do not establish causal effects or population level generalisation.