This document provides guidelines to user agent manufacturers for making their products more accessible to people with disabilities and for increasing usability for all users. This document emphasizes the accessibility of interoperability between two important classes of user agents - graphical desktop browsers and dependent assistive technologies (screen readers, screen magnifiers, and voice input software). However, it is meant to be applicable to user agents in general, including text and voice browsers, multimedia players, and plug-ins.
This document is part of a series of accessibility documents published by the W3C Web Accessibility Initiative.
This is a W3C Working Draft for review by W3C Members and other interested parties. It is a draft document and may be updated, replaced or obsoleted by other documents at any time. It is inappropriate to use W3C Working Drafts as reference material or to cite them as other than "work in progress". This is work in progress and does not imply endorsement by, or the consensus of, either W3C or Members of the WAI User Agent (UA) Working Group.
Please send comments about this document to the public mailing list: firstname.lastname@example.org.
This document has been produced as part of the Web Accessibility Initiative, and is intended as a draft of a Proposed Recommendation for how to improve user agent accessibility. The goals of the WAI UA Working Group are discussed in the WAI UA charter. A list of the current Working Group members is available.
A list of current W3C Recommendations and other technical documents can be found at http://www.w3.org/TR.
For those unfamiliar with accessibility issues pertaining to user agent design, consider that many users may be using documents in contexts very different from your own:
The guidelines in this document are designed to help developers understand and thereby reduce accessibility barriers that impede access to the Web. The associated Techniques Document provides practical solutions based on existing and upcoming technologies. Though developers may believe that implementing accessibility features in their products is difficult or of limited use, considering accessibility during the design phase of a product leads to more flexible, manageable, and extensible software.
These guidelines include information relevant to a wide class of user agents: graphical desktop browsers, screen readers, speech synthesizers, multimedia players, text browsers, voice browsers, plug-ins, etc., with a particular focus on two classes of user agents:
The guidelines emphasize interoperability between these two classes of user agents.
This document is organized according to several principles that will improve the design of any type of user agent:
This document addresses some aspects of user interface that directly impact accessibility, including device-independence, accessible product documentation, and configurability. However, the more general topic of accessible user interfaces for computer software exceeds the scope of this document. User interfaces must be intuitive, simple, and tested. Features that are known to promote accessibility should be made obvious to users and easy to find.
The user agent must allow users to control the style (colors, fonts, speech rate, speech volume, etc.) and format of a document. Users must also be able to access both primary and alternative representations of content.
The user agent must enable navigation among different parts of the document: links, form controls, tables, etc., in an input device-independent manner. Navigation should follow any tabbing order that has been specified by the author. How tabbing order is defined typically depends on markup language (e.g., the "tabindex" attribute in HTML).
The user agent should provide users with information about views of a document, document structure, document summary information, and events to help the user remain oriented. Perhaps more importantly, the user agent should minimize the chances the user will become disoriented by allowing the user to control when windows are spawned, warning the user when events occur, avoiding sudden movements of the viewport, etc.
Not all user agents satisfy the above principles natively, so they must make information available to assistive technologies. Even when a user agent implements a feature natively, it should make information available to other software. Information must be made available through interoperable and standard interfaces and generally according to operating system conventions (except where those conventions hinder accessibility).
Note. In general, software should be designed, installed, configured, and documented according to operating environment conventions.
This document refers to three concepts:
This document lists guidelines and the checkpoints that satisfy them.
The associated Techniques Document provides in-depth descriptions of the same checkpoints and examples of how to implement the checkpoints using HTML, CSS, and more. Please note that the Techniques Document will be updated more regularly than the current document in order to keep up with changes in technology.
Each checkpoint in this document is assigned a priority that indicates its importance for users.
The terms "must", "should", and "may" (and related terms) are used in this document in accordance with RFC 2119 ([RFC2119]).
This document defines two categories of conformance in order to promote a standard of accessibility within, and interoperability between, two important classes of user agents - graphical desktop browsers and dependent assistive technologies. The two categories of conformance are:
The conformance definitions use the following terms:
In order to conform as a desktop graphical user agent, the user agent must do all of the following:
Even for those checkpoints that must be satisfied natively, desktop graphical user agents should make information available to other software through standard interfaces (e.g., specialized dependent user agents may provide a better solution to a problem than a desktop user agent).
In order to conform as a dependent user agent, the user agent must do all of the following:
This section defines terms used in this document.
A document may be seen as a hierarchy of elements. Elements are defined by a language specification (e.g., HTML 4.0 or an XML application). Each element may have content, which generally contributes to the document's content. Elements may also have attributes that take values. Some attributes are integral to document accessibility (e.g., the "alt", "title", and "longdesc" attributes in HTML).
An element's rendered content is that which a user agent renders for the element. Often, this is what lies between the element's start and end tags. However, some elements cause external data to be rendered (e.g., the IMG element in HTML). At times, a browser may render the value of an attribute (e.g., "alt", "title" in HTML) in place of or in addition to the element's content. Rendering is not limited to only visual displays, but can also include audio (speech and sound) and tactile displays (braille and haptic displays).
A user agent renders a document by applying formatting algorithms and style information to the document's elements. Formatting depends on a number of factors, including where the document is rendered: on screen, paper, through speakers, a braille device, a mobile device, etc. Style information (e.g., fonts, colors, voice inflection, etc.) may come from the elements themselves (e.g., certain style attributes in HTML), from style sheets, or from user agent settings. For the purposes of these guidelines, each formatting or style option is governed by a property and each property may take one value from a set of legal values. (The term "property" in this document has the meaning ascribed in the CSS2 Recommendation.) A reference to "styles" in this document means a set of style-related properties.
The value given to a property by a user agent when it is started up is called the property's default value. User agents may allow users to change default values through a variety of mechanisms (e.g., the user interface, style sheets, initialization files, etc.).
Once the user agent is running, the value of a property for a given document or part of a document may be changed from the default value. The value of the property at a given moment is called its current value. Note that changes in the current value of a property do not change its default value.
Current values may come from documents, style sheets, scripts, or the user interface. Values that come from documents, their associated style sheets, or via a server are called author styles. Values that come from user interface settings, user style sheets, or other user interactions are called user styles.
User agents may handle different types of source information: documents, sound objects, video objects, etc. The user perceives the information through a viewport, which may be a window, frame, a piece of paper, a speaker, a virtual magnifying glass, etc. A viewport may contain another viewport (e.g., nested frames, plug-ins, etc.).
User agents may render the same source information in a variety of ways; each rendering is called a view. For instance, a user agent may allow users to view a document in one window and a generated list of headers for the document in another.
The view is how source information is rendered and the viewport is where it is rendered.
Generally, viewports give users access to all rendered information, though not always at once. For example, a video player shows a certain number of frames per second, but allows the user to rewind and fast forward. A visual browser viewport generally features scrollbars or some other paging mechanism that allows the user to bring the rendered content into the viewport.
The content currently available in the viewport is called the user's point of regard. The point of regard may be a two dimensional area (e.g., for graphical rendering) or a single point (e.g., for aural rendering or voice browsing). User agents should not change the point of regard unexpectedly as this can disorient users.
User agents generally offer several mechanisms for referring to parts of content in a view: focus, selection, and insertion point.
The user focus designates an active element in a document. Which elements are active depends on the document language. In HTML documents, for example, active elements include links, form controls, elements with a value for the "longdesc" attribute, and elements with associated scripts. An element with the focus may be activated through any number of mechanisms, including the mouse, keyboard, an API, etc.
The effect of activation depends on the element. For instance, when a link is activated, the user agent generally retrieves the linked resource, which may be another Web page, program, etc. When a form control is activated, it may change state (e.g., check boxes) or may take user input (e.g., a text field). Activating an element with a script assigned for that particular activation mechanism (e.g., mouse down event, key press event, etc.) causes the script to be executed.
A view has at most one focus. When several views co-exist, each may have a focus, but only one is active, called the current focus. The current focus is generally presented (e.g., highlighted) in a way that makes it stand out.
The user selection generally specifies a range of content (text, images, etc.) in a document. The range may be restricted to the content of a single element or may span several elements. The selection may be used for a variety of purposes: for cut and paste operations, to designate a specific element in a document, to identify what a screen reader should read, etc.
The user selection may be set by the user (e.g., by a pointing device or the keyboard) or through an application programming interface (API). A view has at most one user selection. When several views co-exist, each may have a user selection, but only one is active, called the current user selection.
The user selection is usually presented in a way the stands out (e.g., highlighted). On the screen, the selection may be highlighted using colors, fonts, graphics, or other mechanisms. Highlighted text is often used by dependent user agents to indicate through speech or braille output what the user wants to read. Most screen readers are sensitive to highlight colors. Dependent user agents may provide alternative presentation of the selection through speech, enlargement, or refreshable braille display.
Both the current focus and the current user selection must be in the same view, called the current view. The current view is generally highlighted when several views co-exist.
The insertion point is the location where document editing takes place. The insertion point may be set by the user (e.g., by a pointing device or the keyboard editing keys) or through an application programming interface (API). A view has at most one insertion point. When several views co-exist, each may have an insertion point, but only one is active, called the current insertion point
The insertion point is generally rendered specially (e.g., on the screen, by a vertical bar or similar cursor).
When certain events occur (document loading or unloading events, mouse press or hover events, keyboard events, etc.), user agents often perform some task (e.g., execute a script). For instance, in most user agents, when a mouse button is released over a link, the link is activated and the linked resource retrieved. User agents may also execute author-defined scripts when certain events occur. The script bound to a particular event is called an event handler. Note. The interaction of HTML, style sheets, the Document Object Model [DOM1] and scripting is commonly referred to as "Dynamic HTML" or DHTML. However, as there is no W3C specification that formally defines DHTML, this document will only refer to event handlers and scripts.
Certain types of content (e.g., images, audio, video, applets, etc.) may not be accessible to all users, so user agents must ensure that users have access to author-supplied alternative representations of content. The Techniques Document describes the different mechanisms authors may use to supply alternative representations of content. Please also consult the Web Content Accessibility Guidelines ([WAI-PAGEAUTH]).
User agent accessibility requires that the user be able to install the software, learn about its features, configure it, and find help when necessary.
Documentation must be accessible so that users may learn about software features, notably those that relate to accessibility. Users who may not be able to access print material, including individuals with visual impairments, learning disabilities, or movement impairments, may be able to use accessible electronic documentation or documents in alternative hardcopy formats.
Documentation includes all product documentation, notably installation instructions, the help system, and all product manuals.
See also checkpoint 7.2.4.
One of the keys to an accessible user interface is device-independence. Users must be able to interact with a user agent (and the document it renders) using the supported input and output devices of their choice. Input devices may include pointing devices, keyboards, braille devices, head wands, microphones, and others. Output devices may include monitors, speech synthesizers, braille devices, and printers.
Please note that "device-independent support" does not mean that user agents must support every input or output device. User agents should offer redundant input and output mechanisms for those devices that are supported. For example, if a user agent supports keyboard and mouse input, users should be able to interact with all features using either the keyboard or the mouse.
For those user agents that support keyboard input, the input methods must be accessible. Some users require single-key access, others require that keys activated in combination be physically close together, while others require that they be spaced physically far apart. Configurability is vital to accessible keyboard input.
The more apparent the keyboard commands are to all users, the more likely it is that new users with disabilities will find them and use them.
Users must be able to turn on and off support for features that may interfere with accessibility. User agents are only expected to provide this control for content that it recognizes. See also guideline 5.1.
In order to access a document, some users may require that it be rendered in a manner other than what the author intended. Users who are color-deficient may not be able to perceive certain color combinations. Users with low vision may require large fonts. Users who are blind may require audio or tactile rendering. Users who are deaf may require captions for audio files.
User agents must therefore allow the user to control:
The user must be able to control the colors, fonts and other stylistic aspects of a document and to override (have final control over) author styles and user agent default styles. Otherwise, users who are blind, have visual impairments, some types of learning disabilities, or any user who cannot or has chosen not to view the primary representation of information will not know content of the information on the page.
The following checkpoints state which styles the user must be able to control. The checkpoints also require user agents to allow users to override author styles and user agent defaults. The checkpoints apply to alternative representations of content as well as primary content.
Checkpoints for fonts and colors:
See also checkpoints 5.3.6 and 5.3.7.
Checkpoints for applets and animations:
Checkpoints for video.
Checkpoints for audio:
Checkpoints for speech:
Checkpoints for changes to the user interface:
User agents must give users access to author-supplied alternative representations of content (descriptions of images, captions for video or audio, etc.) since some users cannot perceive the primary content due to a disability or a technological limitation (e.g., browser configured not to display images). See also guideline 5.3.
Checkpoints for audio:
Checkpoints for video:
For all link navigation or searching, the focus is moved to the target link. See also guideline 4.3.
The following checkpoints refer to links to long descriptions of content. See also guideline 5.2.
Tables were designed to structure relationships among data. In visual media, tables are often rendered on a two-dimensional grid, but this is just one possible interpretation of the data. On the Web, the HTML TABLE element has been used more often than not to achieve a formatting effect ("layout tables") rather than as a way to structure true tabular data ("data tables")
Layout tables cause problems for some screen readers and when rendered, confuse users. Even data tables can be difficult to understand for users that browse in essentially one dimension, i.e. for whom tables are rendered serially. Large tables pose particular problems since remembering cell position and header information becomes more difficult as the table grows.
User agents facilitate browsing by providing access to specific table cells and their associated header information. How headers are associated with table cells is markup language-dependent.
Tabular navigation is required by people with visual impairments and some types of learning disabilities to determine the content of a particular cell and spatial relationships between cells (which may convey information). If table navigation is not available users with some types of visual impairments and learning disabilities may not be able to understand the purpose of a table or table cell.
For all form control navigation or searching, the focus is moved to the target form control. See also guideline 4.3.
Orientation - the sense of where one is within a document or series of documents - is fundamental to a successful Web experience. Visual clues that help orientation often abound:
Not all users can use visual orientation clues, however, so user agents must complement them with other mechanisms, including:
See also guidelines 5.3, 5.4, and 5.5.
Users that are viewing documents through linear channels of perception like speech (since speech is temporal in nature) and tactile displays (current tactile technology is limited in the amount of information that can be displayed) have difficulty maintaining a sense of their relative position in a document. The meaning of "relative position" depends on the situation. It may mean the distance from the beginning of the document to the point of regard (how much of the document has been read), it may mean the cell currently being examined in a table, or the position of the current document in a set of documents.
For people with visual impairments, blindness, or certain types of learning disabilities, it is important that the point of regard remain as stable as possible. The user agent should not disturb the user's point of regard by shifting focus to a different frame or window when an event occurs without notifying the user of the change.
Hierarchical navigation (through the document tree) is useful for efficiently navigating the major topics and sub-topics of a document.
It is important to alert users, in an output device-independent fashion, when important events occur during a browsing session. To avoid confusion that the effects of scripts may cause, users should be notified when scripts are executed (or be able to disable scripts entirely). This is also important for security reasons; users should be able to decide whether to allow scripts to execute on their machines.
Many types of software: assistive technologies, scripting tools, automated test engines, etc., benefit from having access to user agent information.
A user agent that supports a language should implement the accessibility features for the language. The techniques document lists the accessibility features of the following languages, defined by W3C specifications:
When a user agent communicates with other software (dependent user agents, the operating system, plug-ins), must do so through applicable interfaces. To promote interoperability, open standards should be used wherever possible.
Some operating systems have operating system-level flags and settings that are pertinent to accessibility, such as high-contrast colors and "show" sounds for people with hearing impairments. User agents should take these global settings into account for their own settings.
Many thanks to the following people who have contributed through review and comment: Paul Adelson, James Allan, Denis Anson, Kitch Barnicle, Harvey Bingham, Olivier Borius, Judy Brewer, Bryan Campbell, Kevin Carey, Wendy Chisholm, David Clark, Chetz Colwell, Wilson Craig, Nir Dagan, Daniel Dardailler, B. K. Delong, Neal Ewers, Geoff Freed, John Gardner, Al Gilman, Larry Goldberg, John Grotting, Markku Hakkinen, Earle Harrison, Chris Hasser, Kathy Hewitt, Philipp Hoschka, Masayasu Ishikawa, Phill Jenkins, Leonard Kasday, George Kerscher, Marja-Riitta Koivunen, Josh Krieger, Catherine Laws, Greg Lowney, Scott Luebking, William Loughborough, Napoleon Maou, Charles McCathieNevile, Masafumi Nakane, Mark Novak, Charles Oppermann, Mike Paciello, David Pawson, Michael Pederson, Helen Petrie, David Poehlman, Michael Pieper, Jan Richards, Hans Riesebos, Joe Roeder, Lakespur L. Roca, Greg Rosmaita, Lloyd Rutledge, Liam Quinn, T.V. Raman, Robert Savellis, Rich Schwerdtfeger, Constantine Stephanidis, Jim Thatcher, Jutta Treviranus, Claus Thogersen, Steve Tyler, Gregg Vanderheiden, Jaap van Lelieveld, Jon S. von Tetzchner, Willie Walker, Ben Weiss, Evan Wies, Chris Wilson, Henk Wittingen, and Tom Wlodkowski,