How well do screen readers support web accessibility guidelines?

DO-IT Factsheet #1166
http://www.washington.edu/accessit/articles?1166

For the World Wide Web to be accessible, several parties must support web accessibility guidelines and standards. First, web designers must develop content that follows accessibility guidelines. Second, both web browsers and assistive technology products must render this content appropriately. If any of these parties falls short in their support for accessibility guidelines and standards, the end result may be exclusion of people with certain disabilities.

For example, if a web designer carefully follows all W3C Web Content Accessibility Guidelines [1] but a user's assistive technology product doesn't support these same guidelines, the user may have difficulty accessing the web content despite the designer's good intentions.

A classic example in HTML is the "longdesc" attribute. This attribute was first defined in HTML 4.0, which became a W3C Recommendation in December 1997. It allows web authors to specify a link to a long description of an image, for those images that are too complicated to be sufficiently described using only a few words in an "alt" attribute. Despite its importance for making complicated images accessible to users who couldn't see them, until recently no Windows-based screen reader utilized "longdesc" in any way. Freedom Scientific's JAWS for Windows was the first product to do so, but not until version 4.01 was released in February 2002 (over four years after "longdesc" became an official HTML attribute).

In March 2002, the organization CAMO pour personnes handicapées [2] released a report How Assistive Software Supports Web Accessibility [3]. The publication reports on the evaluation of several popular screen reader applications regarding their support for the WCAG. Their assessment focused specifically on GW Micro Window-Eyes 4.2, IBM Home Page Reader 3.02, and three versions of Freedom Scientific JAWS (3.5, 3.71, and 4.02). They tested each of these screen readers on forty-one guideline-related variables. Results show that no screen reader fully supports the established guidelines. Comparative findings for all screen readers across all variables are presented in the full report.

References