Web-Content's Syndication in Sign Language
Oussama El Ghoul, Nour Ben Yahia, Mohamed Jemni · 2012 · Proceedings of the International Cross-Disciplinary Conference on Web Accessibility (W4A) · doi:10.1145/2207016.2207050
Summary
This paper addresses the inaccessibility of web content syndication (RSS feeds) for deaf users, particularly those in developing countries where over 80% of deaf people are illiterate. While RSS and Atom feeds are efficient mechanisms for pushing updated content to users, all syndicated metadata — titles, descriptions, author names, dates — is delivered in text, which is inaccessible to deaf individuals whose primary language is a sign language. The authors note that even websites like WebSourd (a French deaf services company) that present main content in sign language video still syndicate their RSS feed metadata in written text. The paper proposes extending the RSS specification to support sign language content by leveraging two existing but underused standards: the ISO 639-2 language codes for sign languages (e.g., sgn-US for ASL, sgn-GB for BSL) and avatar technology for rendering sign language from textual descriptions. Rather than embedding bandwidth-heavy video in lightweight RSS files, the authors propose using SML (Sign Markup Language), an XML-based language they developed for annotating signer gestures, which can be rendered by a 3D avatar in real time. The paper also discusses sign language transcription systems including HamNoSys (Hamburg Notation System, 1984) and SignWriting (Valerie Sutton, 1974), noting their limitations for deaf readers.
Key findings
The authors' approach embeds SML code directly into RSS XML elements, replacing textual content with sign language animation descriptions that a 3D avatar can perform. An example shows an RSS channel where the <title> and <description> elements contain <sml> blocks instead of plain text, with the <language> element set to "sgn-US" (American Sign Language) per ISO 639-2. Two prototype tools were developed: a deaf-accessible RSS editor that allows content creators to animate a virtual character using a WYSIWYG interface with key-framing (drag-and-position joint rotation along Euler axes), and a deaf-accessible RSS reader that interprets SML code and renders 3D avatar animations when users click on feed items. The avatar's armature follows the H-Anim specification, and the animation approach groups joint rotations by time intervals. The SML approach is lightweight compared to video — critical for RSS's design goal of small, fast-downloading files — while preserving the visual-spatial nature of sign language that text-based transcription systems fail to capture.
Relevance
This paper highlights a frequently overlooked aspect of deaf accessibility: the assumption that text-based content is accessible to deaf users. Research consistently shows that many deaf individuals, particularly those who are prelingually deaf, have significantly lower reading levels in written languages — the reading capability of deaf high school graduates is comparable to that of 8-9 year old hearing children. This challenges the common misconception that providing text alternatives (as WCAG emphasizes) fully addresses deaf accessibility. The avatar-based approach to sign language content avoids the cost and bandwidth of video while being more expressive than written transcription systems. However, the paper acknowledges significant limitations: avatar technology in 2012 produced signing that lacked the naturalness and facial expressivity of human signers, no user evaluation with deaf participants was conducted, and the text-to-sign-language translation problem remained fundamentally unsolved. The broader argument — that web standards should formally accommodate sign languages as first-class content languages — remains relevant and largely unaddressed.
Tags: sign language · deaf accessibility · avatar technology · RSS · content syndication · web accessibility · virtual signing
Standards referenced: ISO 639-2 · H-Anim · MPEG-4