Loading...
Artificial Intelligence for Kenyan Sign Language to Support Gender-Inclusive Education for Deaf Learners
Abstract
In Kenya, approximately 1 million people are deaf, facing limited access to education due to the scarcity of trained sign language interpreters. To address this challenge, the AI4KSL project has developed an innovative assistive technology that interprets spoken English into Kenyan Sign Language (KSL) in real-time using virtual signing characters. This technology aims to bridge the communication gap between deaf and hearing learners of all genders, thereby enhancing inclusion and improving educational outcomes. The study employed a mixed methods approach and was guided by design thinking theory, as developed by the Hasso-Plattner Institute of Design at Stanford D School (2010). The project involved creating a comprehensive dataset comprising 14,000 English sentences with equivalent KSL glosses and approximately 20,000 signed words from KSL videos collected from 610 learners, 16 teachers and 6 non-teaching staff from 1 girls’, 1 boys’, 3 co-educational schools and 1 technical and vocational training institution using a stratified sampling technique. Notably, the dataset revealed gender variations in terms of fluency and nonmanual signals among signers across age groups. Female learners articulated more fluent signs compared to male learners. Further, non-manual signals such as facial expressions were more evident among female signers as compared to male ones. The AI4KSL avatar incorporates both male and female gender representations to avoid bias and to provide mentorship for the male learners. By leveraging generative Artificial Intelligence, the AI4KSL technology supports Sustainable Development Goals 4 and 5, promoting lifelong learning and ensuring that no one is left behind, regardless of gender. It has significant potential for global scalability in enhancing accessibility and promoting inclusive education.
PCF11 Sub-Theme: Gender, Technology and Innovation in Open Education
Paper ID: 0832
Country
Kenya
Region
Africa
Collections
Files
Loading...
PDF
Adobe PDF, 656.85 KB
Link
Date
2025-09
Author
ORCID
Corporate Author
Editor
Publisher
Commonwealth of Learning (COL)