The Meaning of 'Life': Capturing Intent from Web Authors
Rhys Lewis · 2006 · Proceedings of the 2006 International Cross-Disciplinary Workshop on Web Accessibility (W4A): Building the Mobile Web: Rediscovering Accessibility? · doi:10.1145/1133219.1133231
Summary
This paper explores the fundamental problem that web pages lack explicit semantic information, relying instead on visual cues — layout, color, juxtaposition — to convey meaning. Lewis argues that this semantic deficit creates parallel challenges for two distinct user groups: people with disabilities who cannot perceive visual context, and mobile device users viewing content on small screens that destroy spatial relationships between page elements. The paper examines how traditional approaches to both problems — assistive technologies for disabled users and server-side transcoding for mobile devices — struggle because they must guess at author intent from markup that was designed for sighted desktop users. Using a detailed hypothetical movie website example, Lewis demonstrates how even moderately complex pages embed critical semantic information only in visual arrangement: navigation bars identified solely by horizontal positioning, form label-to-field relationships conveyed only through adjacency, and multiple form actions hidden within shared processing URLs. The paper then surveys emerging W3C standards work addressing this deficit through two complementary strategies: semantic enrichment (adding metadata like the WAI-PF role taxonomy to annotate existing markup) and semantically rich markup languages (new languages like XForms and the Device Independent Authoring Language that encode meaning directly). Lewis also examines the concept of "delivery context" — device characteristics and user preferences that drive content adaptation — and the W3C Mobile Web Initiative's relationship to the Web Accessibility Initiative.
Key findings
The paper identifies a core insight that became foundational for later web standards work: the web's reliance on implicit, visually-conveyed semantics is the root cause of accessibility barriers for both disabled users and mobile users. Lewis demonstrates that transcoding and assistive technologies are fundamentally limited because they must reverse-engineer author intent from markup that never explicitly encoded it — what he terms "intelligent guesswork" or "heuristics." The WAI Protocols and Formats Working Group's role taxonomy, which would later evolve into WAI-ARIA, is presented as a key solution, allowing authors to label page sections with explicit semantic roles like "navigation," "main," and "spreadsheet." Lewis shows how the same role-based semantic annotations needed for assistive technologies are equally useful for mobile content adaptation. The paper also introduces the concept of "authored units" and "delivery units" from the W3C Device Independence Working Group — the idea that a page is not monolithic but composed of selectable content units that can be assembled differently for different contexts. Privacy concerns around transmitting user preference data (including disability-related preferences) through adaptation chains are flagged as an important unresolved challenge.
Relevance
This paper is historically significant as an early articulation of the synergy between mobile web access and disability access — an insight that proved prophetic as responsive design and WAI-ARIA became mainstream. The role taxonomy discussed here directly evolved into the ARIA specification that is now fundamental to web accessibility. Lewis's argument that authors must encode intent explicitly rather than relying on visual presentation anticipated the semantic HTML movement and HTML5's introduction of structural elements like nav, main, article, and section. For practitioners today, the paper remains a useful conceptual framework: the distinction between application-level and user-interface-level semantics helps clarify why semantic HTML matters and what ARIA roles are actually for. The privacy concerns raised about transmitting user preferences through adaptation chains foreshadowed ongoing debates about disability disclosure in personalization systems.
Tags: mobile accessibility · web semantics · content adaptation · device independence · transcoding · semantic markup · ARIA roles · authoring practices
Standards referenced: XHTML 2.0 · XForms · WAI-ARIA · WCAG