Background and Objectives: Children with Down syndrome (DS) commonly present with hypotonia of the orofacial musculature, delayed phonological development, and limited working memory, all of which constrain the acquisition of spoken second-language (L2) skills under conventional classroom methods. In Iran, English is introduced as a foreign language from the early years of schooling, yet learners with intellectual and developmental disabilities are rarely accommodated in mainstream English Language Teaching (ELT) curricula, and inclusive practice in provincial cities such as Sari, Mazandaran Province, remains underdeveloped. Artificial intelligence (AI)-mediated tools, including speech-recognition applications, conversational agents, and adaptive feedback systems, have been proposed as a means of individualising pace, repetition, and reinforcement for learners with developmental disabilities, but empirical accounts specific to Persian-speaking children with DS are scarce.

Methodology: A qualitative, multiple-case study design was proposed, drawing on 12 children with DS (ages 8-14) recruited from special-education and inclusive schools in Sari during the 2025-2026 academic year. The proposed intervention combines a speech-recognition and pronunciation-feedback application, a generative-AI conversational partner for guided dialogic practice, and visually supported AI-generated prompts, delivered across 10 weeks of twice-weekly 30-minute sessions. Data sources comprise structured classroom observations, semi-structured interviews with English teachers and parents, and reflective field notes. To demonstrate the analytic procedure, illustrative (simulated, non-empirical) case vignettes and thematic findings are presented following Braun and Clarke's (2006) thematic analysis framework.

Findings:The illustrative analysis suggests four candidate themes: (a) increased willingness to vocalize in low-pressure AI interactions compared with peer settings, (b) variable benefit of automatic speech recognition depending on the severity of dysarthric speech features, (c) parental and teacher perceptions of AI as a motivational scaffold rather than a replacement for human instruction, and (d) the importance of slowed speech rate, visual cues, and predictable AI feedback loops in sustaining engagement.

Conclusion: AI-assisted tools appear conceptually promising as a supplementary, individualised resource for developing English speaking skills among children with Down syndrome, provided they are embedded within teacher-mediated, multimodal instruction rather than used as a stand-alone substitute. The proposed design offers a replicable framework for future empirical research in Mazandaran Province and comparable under-studied EFL contexts.