Contextual Language Learning with Capti ESL Assistant
Yevgen Borodin, Yury Puzis, Andrii Soviak, Vikas Ashok, Andrii Melnyk, I. V. Ramakrishnan · 2016 · Proceedings of the 13th International Web for All Conference (W4A) · doi:10.1145/2899475.2899508
Summary
This paper presents Capti ESL Assistant, a web application designed to help English language learners develop reading and listening skills simultaneously while remaining accessible to users with print disabilities including dyslexia and vision loss. Built on the Capti Narrator platform, the Assistant addresses a critical gap identified by ESL instructors: existing language learning tools like Rosetta Stone and Duolingo rely on predefined, canned materials that only cover beginner levels, cannot be customized by teachers, and are inaccessible to learners with disabilities. The authors frame language barriers as an accessibility issue in the broadest sense, noting that over 1.5 billion people worldwide are learning English. The system combines text-to-speech narration with synchronized visual highlighting, built-in translation, and a gamified vocabulary exercise called Word Challenge. Its core technical components are a User Knowledge Model that tracks which vocabulary a learner knows or does not know, and a Natural Language Processing module that performs linguistic analysis including part-of-speech tagging, sentence parsing, and named entity recognition to generate contextually appropriate exercises from any user-supplied text.
Key findings
A three-month pilot deployment at East Hampton School District High School with 20 students (10 beginners and 10 advanced) produced qualitative observations of strong student engagement. Teachers reported that students — typically difficult to motivate — played the Word Challenge game competitively for sustained periods, comparing scores and competing for fastest answers. Students were able to choose their own reading materials from sources like NewsELA, listen to articles narrated multiple times, and then reinforce vocabulary through the game. The NLP-driven exercise generation created contextually relevant questions where incorrect answer choices were specifically selected to be plausible distractors rather than obviously wrong. Interviews with over 100 ESL instructors confirmed the value of allowing custom materials and identified requested features including additional game types (pronunciation-focused, fill-in-the-blank), and student progress tracking. The cross-platform web-based design meant the tool worked across devices without installation barriers.
Relevance
This work sits at the intersection of language accessibility and disability accessibility, demonstrating how universally designed educational tools can serve multiple populations simultaneously. The Capti ESL Assistant shows that text-to-speech technology originally developed for users with print disabilities — people with dyslexia, low vision, or blindness — can be repurposed to support language learners who similarly benefit from synchronized reading and listening. For accessibility practitioners, the paper reinforces that accessibility features are not niche accommodations but universal design assets that expand a tool's audience. The approach of allowing any user-supplied content rather than relying on pre-built materials is particularly relevant for inclusive education, where learners bring diverse backgrounds, interests, and skill levels. The pilot's small scale and qualitative-only evaluation are limitations, but the concept of accessible language learning tools remains underexplored and important.
Tags: language accessibility · print disabilities · text-to-speech · dyslexia · education accessibility · universal design · natural language processing