Contextual Scaffolding and Self-Efficacy: Supporting Computer Skill Development among Blind Learners in India
Akshay Kolgar Nayak, Yash Prakash, Sampath Jayarathna, Hae-Na Lee, Vikas Ashok · 2026 · Proceedings of the 2026 CHI Conference on Human Factors in Computing Systems (CHI '26) · doi:10.1145/3772318.3791509
Summary
This paper presents a four-month contextual inquiry at two Indian computer training centers serving 94 blind or visually impaired (BVI) students — one urban center in Bengaluru focused on job-oriented ICT training (KEONICS certification) and one rural residential blind school in Ranebennur offering K-12 education with basic computer instruction. The authors argue that existing HCI literature on inclusive computer literacy education draws almost exclusively from affluent Global North settings where accessibility resources, legal frameworks, and pedagogical training are comparatively mature, and that very little is known about in-situ teaching and learning in resource-constrained, multicultural Global South contexts like India. Framed by Bandura's social cognitive theory (self-efficacy), Vygotsky's zone of proximal development, and Lave and Wenger's situated learning, the team observed classes twice weekly (four-five hours per visit), conducted pre- and post-study semi-structured interviews with newly enrolled students, recent graduates, and past trainees, and translated conversations from Hindi and Kannada. They used hybrid inductive/deductive thematic analysis across 617 pages of transcripts, mapping emergent codes to theoretical constructs. Two research questions structure the work: how do trainers teach computer skills to BVIs in India, and how do BVI learners' lived experiences shape their learning behaviors and self-efficacy? The paper contributes empirical findings on misaligned curricula and unstructured pedagogy, a characterization of the compounded cognitive load BVI learners face, and design proposals for culturally responsive, AI-assisted EdTech that scaffolds both trainers and trainees.
Key findings
The curriculum at both centers is ocularcentric and certification-oriented, treating screen readers and accessibility as peripheral 'add-ons' to a visually-centric syllabus imported wholesale from sighted learners. Trainers were hired for technical proficiency as SR users (JAWS or NVDA), not pedagogy or computer science — several had literature or data-entry backgrounds — and defaulted to rote demonstration via earbud-sharing, with one-on-one demos proving unscalable. Exams tested generic shortcuts (Ctrl+S), not SR-specific navigation (Insert+F7), producing brittle, context-bound skills that did not transfer across applications. BVI learners bore a compounded cognitive load: simultaneously learning an English QWERTY keyboard, Kannada Inscript mappings, SR commands, UI semantics, and rapid monotone English SR output, often with only basic English proficiency. Late-blind and low-vision learners received no tailored transitional support and were grouped with congenitally blind peers. Infrastructure was thin — pirated Office 2007, Windows XP/7, 6-9 working computers per center. Subjective norms (family, peers, employer stigma) further eroded domain-specific self-efficacy even when general self-efficacy was high. Urban graduates could apply for bank clerk jobs but were rejected when asked about Word collaboration or HTML; many returned home, lost computer access, and saw fragile skills collapse. Peer learning filled trainer gaps but introduced inconsistent shortcut sequences and deepened confusion.
Relevance
For practitioners working outside North America and Europe, this paper is a direct challenge to the assumption that Global North accessibility curricula, certifications, and train-the-trainer models transplant cleanly. It reframes the 'inaccessible curriculum' narrative as a deeper question of how assistive technologies get positioned inside pedagogy — and argues for locally-adaptable contextual scaffolds, multilingual learning, and interdependence rather than imported independence models. The authors propose concrete EdTech directions (LLM-based personalized tutors, multilingual bilingual coaches that shadow-interpret SR output, trainer-facing pedagogical mirrors) while warning that AI systems trained on Global North data risk reproducing ableist, ocularcentric, and infantilizing patterns. Limitations: single region of India, participants mostly totally blind or with very little residual vision, desktop-focused, and sighted teachers in mainstream schools were not directly studied. The work is especially relevant to NGOs, rehabilitation agencies, funders, and AT vendors designing for the majority of the world's BVI population, which lives in the Global South.
Tags: screen readers · blind and low vision · computer literacy · Global South accessibility · India · self-efficacy · scaffolding · contextual inquiry · culturally responsive pedagogy · vocational training · BVI learners · JAWS · NVDA · digital divide
Standards referenced: Rights of Persons with Disabilities Act (India, 2016) · UN CRPD · ADA · Section 504 · European Accessibility Act · IDEA