مجلة طبنـــة للدراسات العلمية الأكاديمية

Ai-enabled Accessibility In Digital Learning For Deaf And Hard-of-hearing Learners

Auteurs : Boudjemai Djouhra .

Résumé

This article develops an evidence-calibrated framework for evaluating artificial-intelligence-enabled accessibility in digital learning for deaf and hard-of-hearing (DHH) learners in Arabic-speaking contexts. A critical integrative review examined six focal Arabic-region works spanning audiovisual translation, DHH media preferences, teacher readiness, neural text-to-sign translation, and AI-based sign recognition. International evidence and current accessibility standards were used for targeted verification. The analysis distinguishes five validation levels: technical performance, linguistic validity, usability and accessibility, educational effectiveness, and real-world impact. The selected literature shows technical progress but an incomplete evidential chain. Four focal works addressed AI, only one directly recruited DHH participants, and none reported a controlled classroom test of learning outcomes. Direct learner evidence indicates that captions, sign-language interpretation, and combined presentation are conditionally useful rather than universally interchangeable. Their value varies with knowledge type, reading proficiency, sign-language proficiency, visual density, and learner control. Recurring constraints include limited continuous-sign datasets, weak localization, insufficient training, infrastructure gaps, privacy risks, and restricted Deaf-led participation. The article proposes a six-layer implementation model, a concise evaluation checklist, and a field-validation protocol. Its central contribution is methodological: algorithmic accuracy must not be treated as proof of natural sign-language quality, learner usability, or improved learning.

Mots clés

accessible digital learning ; deaf and hard-of-hearing learners ; captions ; sign languages ; audiovisual translation ; artificial intelligence