Unlocking the Potential of Web Localizers as Contributors to Image Accessibility: What Do Evaluation Tools Have to Offer?
Silvia Rodríguez Vázquez · 2015 · Proceedings of the 12th International Web for All Conference (W4A) · doi:10.1145/2745555.2746662
Summary
This paper investigates the role of web localization professionals (translators) in maintaining image accessibility when websites are adapted for different language audiences. The author identifies a significant gap: when websites are localized from one language to another, text alternatives for images are frequently overlooked as translatable content, degrading accessibility in the target language version. The study recruited 14 professional translators with English-to-French web localization experience but no web accessibility training to form a control group. Participants were asked to localize a test website containing 130 images with various alt text conditions (missing, empty, appropriate, and inappropriate across 10 rule categories). The experiment used a within-group design where participants first translated without accessibility tools, then used two evaluation tools — Acrolinx (a controlled-language checker adapted with 40 accessibility-oriented rules for verifying text alternative quality) and aDesigner (an Eclipse-based visual impairment simulator with WCAG 2.0 checking capabilities). The research addresses two key questions: how often translators identify alt text as translatable content without accessibility knowledge, and whether evaluation tools help them recognize and improve text alternatives during localization.
Key findings
A striking 43% of participants (6 out of 14) failed to identify text alternatives as translatable content during initial translation without tools. Among those who did find alt text, most were either guided by computer-assisted translation tools that surfaced the text, or they spotted well-formed sentences in the HTML source — isolated nouns or short phrases were frequently missed. When evaluation tools were introduced, Acrolinx triggered significantly more text edits, with participants modifying 70% of text alternatives (811 out of 1,152). This led to higher translation rates even among participants who had initially missed alt text entirely. Acrolinx was rated higher for correctness and documentation clarity, with participants praising its explanations, examples, and improvement suggestions. aDesigner was preferred for specificity and its ability to check errors in context and detect missing alt attributes. The combination of both specialized and general evaluation tools produced the best results, suggesting that no single tool adequately addresses all aspects of image accessibility verification during localization.
Relevance
This research highlights a frequently overlooked accessibility gap in multilingual web development: the loss of image accessibility during localization. For organizations operating multilingual websites, the findings underscore the need to explicitly include alt text translation in localization workflows and to equip translators with appropriate evaluation tools. The study demonstrates that even without accessibility training, translators can significantly improve text alternative quality when given tools that provide clear explanations and contextual examples — a principle applicable to any content authoring workflow. The research also reinforces the broader finding that combining specialized and general accessibility evaluation tools yields better outcomes than relying on any single tool, a practical lesson for accessibility testing strategies.
Tags: image accessibility · text alternatives · web localization · accessibility evaluation tools · quality assurance · multilingual accessibility · alt text
Standards referenced: WCAG 2.0