By Amelia Bellamy-Royds
A First Public Working Draft of the WAI-ARIA Graphics Module was published today. This new vocabulary for describing graphical documents allows improved representation to and interaction with people with disabilities, and will be of interest to those working in any graphics format on the web or in XML documents. It is being developed by the SVG Accessibility Task Force, a joint effort of the Accessible Rich Internet Applications Working Group and the Scalable Vector Graphics (SVG) Working Group.
Graphics can be particularly problematic from an accessibility perspective, because they are oriented so strongly towards a holistic, visual perception of the content. Graphics are used to convey complex patterns, relationships, and details of shape and structure that cannot be easily conveyed in text—not even in the 1000 words a picture is allegedly worth. Numerous possible differences in abilities, from color blindness to complete blindness to alternative cognitive processing, can affect a person’s potential to extract information from an image. So can technological limitations, from black-and-white printing to tiny screens.
Nonetheless, graphics formats such as SVG allow authors to define an image, not as the final visual representation, but as a structured document with meaningful parts and embedded text descriptions. Unfortunately, current software tools make limited use of this information to assist end-users. Furthermore, there are no standards for encoding many important aspects of graphical structure and data, limiting the development of specialized accessibility tools. A truly accessible graphical document could be explored regardless of the medium used to access it. The person accessing it could identify meaningful parts, follow the relationships between them, and understand the concepts or data they represent.
In combination with other standards in development, the WAI-ARIA Graphics Module aims to establish a language with which graphics creators can communicate the structure of their document to assistive technologies, so those technologies can effectively present the content to users. It does so by introducing graphics-specific content roles to the WAI-ARIA (Accessible Rich Internet Applications) model.
The WAI-ARIA roles model is a standard vocabulary through which content creators can describe the structure and function of elements within a document. Web browsers, computer operating systems, and assistive technologies use this information to help users perceive, understand, and interact with the document—even if they do so using interfaces quite different from those used by the author. However, the established taxonomy of ARIA roles only address a subset of document features. In particular, they emphasize landmarks within standard web page layouts (such as headers, footers, navigation menus, and articles) and user-input widgets (buttons, sliders, text boxes, and so on).
The WAI-ARIA Graphics Module proposes three new ARIA roles which will provide the foundation for a structured, semantic approach to graphical content. The graphics roles are being developed as part of a modular approach to extending ARIA with domain-specific vocabularies. For example, another new ARIA module in development is the Digital Publishing Module for describing parts of a book.
The new roles would allow authors to:
- Declare a document, or section of a document, to be a complex graphic using the
graphics-docrole. This would warn web browsers and assistive technologies that the visual presentation and layout of the content may convey information not contained in its plain text content. Users may wish to explore connections between content in a two-dimensional manner that does not follow the linear structure of the document source code. Alternative visual presentations—such as simplified “reader” modes—should respect the graphical nature of the content.
- Describe a section of a document as a meaningful object, using the
graphics-objectrole. The vocabulary used to describe the structure of text documents—sections, lists, and groups—can be confusing when applied to a graphic. The new graphical object role will allow authors to apply labels and descriptions to composite objects, while clearly distinguishing these from a grouping of distinct elements.
- Identify an image or graphic as a standard symbolic representation of a concept or category, using the
graphics-symbolrole. Graphical symbols are used in many contexts: on maps, in charts, in weather reports. In each case, the specific appearance of the symbol is less important than the category or value it represents. With graphics clearly identified as symbols when appropriate, people experiencing the content through a screen reader can know whether a full description is required. Alternative visual or tactile representations of the graphic might even substitute all instances of the symbol with a simplified equivalent.
These roles would complement the existing
img role, which tells assistive technologies to treat an element or section of a document as a single, indivisible complex image with an alternative text description.
Future extensions of this module will focus on information-heavy graphics such as data charts and maps, defining more specific roles for common graphical objects such as legends and axes, and providing standard ways for authors to annotate their graphical elements with the data and relationships they represent. With this additional information, advanced assistive technologies could be developed to convert and adapt the data into other formats more suitable for a particular user. Nonetheless, the core graphics roles should provide a framework for assistive technology developers to start addressing the particularly needs of graphical comment, for layout and for navigation.
In the near term, work on the WAI-ARIA Graphics Module will focus on developing a suitable mapping in the SVG Accessibility API Mappings of the new ARIA roles to roles and properties used by operating system accessibility APIs (which are, in turn, used by many assistive technologies). This may require new extensions to some of those application programming interfaces. The editors would also like feedback on the current draft specification, in particular:
- Are proposed roles clear and appropriate to the needs of interactive graphics?
- Is the use of the “graphics-” prefix in role names to avoid potential collision with other ARIA roles acceptable?
- What mechanism would be suitable for addition of new roles?
- Is the relationship of this specification to WAI-ARIA 1.1 clear?
Feedback may be submitted as an issue to the W3C’s ARIA specifications GitHub repository or via feedback to the SVG Accessibility Task Force’s e-mail list, email@example.com.