Towards a Universal Accessibility for Textual Information
Vasile Topac · 2011 · Proceedings of the International Cross-Disciplinary Conference on Web Accessibility (W4A) · doi:10.1145/1969289.1969309
Summary
This student award paper from Politehnica University Timisoara proposes a universal framework for understanding and addressing textual information accessibility across all media formats and contexts — not just web accessibility, but also document accessibility, desktop applications, printed text, and spoken text. The author introduces a layered model where any textual information instance is characterized by three "access layers": media format (digital text, printed text, audio text), language (English, Romanian, etc.), and language specificity (lay, medical, technical). Each unique combination of these layers represents a distinct accessibility challenge. The key insight is that accessibility barriers are not limited to disability — everyone faces "temporary disabilities" when encountering text in a foreign language or specialized terminology (e.g., a lay person reading medical diagnostics). The research originated from a diploma project tool that used a webcam with OCR, Google Translate, and TTS to read printed text aloud in different languages, demonstrating the power of orchestrating existing text technologies.
Key findings
The proposed cloud-based platform orchestrates existing web services (OCR, TTS, machine translation, speech recognition, NLP, text simplification) into composable pipelines addressable through the browser, eliminating local installation and hardware requirements. Pre-defined templates handle common scenarios — e.g., speech-to-speech translation chains speech recognition, machine translation, and TTS services together. Users can also customize pipelines by including or excluding any available text transformation component. A specialized terminology service was developed for medical language (English and Romanian) and integrated into a teleassistance system, addressing the overlooked problem of technical jargon being inaccessible to lay users. The universal model also enables a novel approach to accessibility evaluation: applied to YouTube, it revealed that while the platform offered content in 1 audio and 59 digital text formats, it lacked multilingual audio output and multilingual printable captions — gaps that the framework made easy to identify systematically.
Relevance
This paper offers a genuinely expansive view of text accessibility that transcends the typical web-centric focus. The three-layer model (format, language, specialization) provides a useful analytical framework for identifying accessibility gaps that traditional WCAG-focused evaluation would miss entirely — such as the inaccessibility of medical terminology to patients, or the barriers created by monolingual content for non-English speakers. The web services orchestration approach anticipated the API-first accessibility services that have since become common (Google Cloud Speech-to-Text, Amazon Translate, Azure Cognitive Services). The YouTube evaluation example demonstrates the model's practical analytical value. As a student research direction paper, it presents the framework concept and early prototypes rather than a complete implementation or evaluation. The idea that "we all have disabilities" in certain information access contexts — language barriers, jargon, format mismatches — broadens accessibility thinking beyond the traditional disability focus toward truly universal design.
Tags: content adaptation · cloud computing · text-to-speech · speech recognition · multilingual accessibility · natural language processing · plain language · accessibility frameworks