Copyright © 1999 W3C® (MIT, INRIA, Keio), All Rights Reserved. W3C liability, trademark, document use and software licensing rules apply.
This document provides techniques for implementing the checkpoints defined in "User Agent Accessibility Guidelines 1.0". These techniques address the accessibility of user interfaces, content rendering, program interfaces, and languages such as HTML, CSS and SMIL.
This document is part of a series of accessibility documents published by the Web Accessibility Initiative.
This section describes the status of this document at the time of its publication. Other documents may supersede this document. The latest status of this document series is maintained at the W3C.
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.
While User Agent Accessibility Guidelines 1.0 strives to be a stable document (as a W3C Recommendation), the current document is expected to evolve as technologies change and content developers discover more effective techniques for designing accessible Web sites and pages.
This document has been produced as part of the Web Accessibility Initiative and intends to improve user agent accessibility for all users. The goals of the User Agent Guidelines Working Group are discussed in the Working Group charter. A list of the UA Working Group participants is available.
A list of current W3C Recommendations and other technical documents can be found at http://www.w3.org/TR.
Please send comments about this document to the public mailing list: w3c-wai-ua@w3.org.
Each checkpoint in this document is assigned a priority that indicates its importance for users.
The checkpoints in this document are numbered to match their numbering in User Agent Accessibility Guidelines 1.0.
This document is organized as follows:
A checkpoint map has been provided for navigation of the techniques. For each checkpoint, the map includes its definition (as it appears in the "User Agent Accessibility Guidelines 1.0") and links to applicable techniques for the checkpoint. In addition, the beginning of each section of this document lists the checkpoints that are addressed in that section.
This document contains a number of examples that illustrate accessible solutions in HTML, CSS, etc.
A list of assistive technologies and browsers designed for accessibility is available at the WAI Web site (refer to [USERAGENTS]).
Checkpoints in this section: 3.1, 3.2.
It is not sufficient to convert a rendering intended for one medium into a rendering for a different medium (e.g., a graphical rendering to speech) since structural information is lost. Examples: table cells, nested lists (want to know where list item ends).
Also, serial access to content not always convenient, so we need additional mechanisms to select and receive content.
Some combination of document source, style sheets (which may hide content or generate content), and user agent additions (which may add contextual information or dynamic information such as whether a link has been visited).
In addition, the user agent may want to provide "intelligent" access to content to simplify the view or to convey models more familiar to users than what is conveyed by the DTD alone.
In the Amaya browser ([AMAYA]), users may access attribute values as follows: Place the cursor at the element in question, open/swap to the structure view. You are shown list of attributes and values. Another technique: select the element (press escape in Linux), then the attributes are all available from the attributes menu. For alt, one can also look at the alternate view, which renders alt text instead of images - a lynx-like view. All the views are synchronized for navigation (and for editing).
Access means that the user agent renders selected content. Content includes text, video, audio, and alternative equivalents to them (which may be attribute values, element content, external resources, etc.).
In the simplest case, the user agent renders the document (e.g., two-dimensional graphical layout, audio stream, line-by-line braille stream) and the user has access to the entire rendering.
But this is not sufficient and so the user agent must provide navigation mechanisms that allow the user to set the selection/focus and then request the selected content (or information about the content - refer to another section...).
Checkpoints in this section: 3.3.
A user agent should treat content language as part of contextual information. When the language changes, the user agent should either render the content in the supported language or notify the user of the language change (if configured for notification). Rendering could involve speaking in the designated language in the case of an audio browser or screen reader. If the language was not supported, the language change notification could be spoken in the default language by a screen reader or audio browser.
Language information for HTML ("lang", "dir") and XML ("xml:lang") should be made available through the DOM ([DOM1]).
Checkpoints in this section: 3.4.
User agents should make available information about abbreviation and acronym expansions. For instance, in HTML, look for abbreviations specified by the ABBR and ACRONYM elements. The expansion may be specified with the "title" attribute.
To provide expansion information, user agents may:
Checkpoints in this section: 1.2, 1.1, 1.3, 1.4, 1.5, 1.6
For non-text or replaced content:
For client-side image maps:
To ensure accessibility, users must have final control over certain renderings.
For text and color:
Checkpoints in this section: 5.1, 5.2, 5.3, 5.4, 5.5, and 5.6.
[Ed. These may be rendered in a variety of ways. How do we specify rendering?]
Implement CSS ([CSS1], [CSS2]) including the CSS2 cascade order and user style sheets. The CSS2 cascade order ensures that user style sheets with "!important" take precedence over author style sheets, giving users final control. Style sheets give authors design flexibility while offering users final control over presentation (refer also to [WAI-WEBCONTENT], checkpoint 3.3). CSS should be implemented by user agents that implement CSS for text that it renders. CSS includes properties for audio, braille (fixed and refreshable), screen, and print rendering, and all relevant properties for supported output media should be implemented.
Note that in the CSS cascade order, markup is given less weight than style sheet rules. Thus, an author may use both presentation markup and style sheets, and user agents that support style sheets will prefer the latter.
A user style sheet can be implemented through a user interface, which means that the user may not have to understand how to write style sheets; they are generated or the user agent acts as though they were. For an example of this, refer to the style sheets implementation of Amaya ([AMAYA]), which provides a GUI-based interface to create and apply internal style sheets. The same technique could be used to control a user style sheet.
For images, applets, and animations:
Checkpoints in this section: 4.2 and 5.7.
Background images may be controlled by the use of local style sheets, and more effectively if these can be dynamically updated. Animation rate depends on the players used. User agents that provide native rendering of animation (for example a movie player, a driver for animated GIF images, or a java machine) should enable the control of animation rates, or at least allow the user to stop, and to play frame-by-frame, as well as straight rendering. A user agent could provide control of the general timing of a presentation, combined with the ability to select from available tracks manually. An issue to bear in mind is that when animation is synchronized with audio, a user may need the ability to play the animation separately from the associated audio.
For time-based presentations:
Implement user controls to start, atop, rewind and pause presentations, and where multiple tracks are supported, to choose which tracks should be rendered. SMIL ([SMIL]) provides for a number of these features. A SMIL implementation should provide for direct user control, as well as activation of the controls through a published API, for developers of assistive technologies.
For user agents rendering audio:
Checkpoints in this section: 5.11, 5.13, 5.9, 3.9, 3.8, and 5.12.
On selecting from among available description tracks. SMIL ([SMIL]) allows users to specify captions in different languages. By setting language preferences in the SMIL player, users may access captions (or audio) in different languages.
The G2 player from Real Networks currently allows users to specify which language they prefer, which can be evaluated in a SMIL ([SMIL]) document to choose from among text or audio tracks. Currently only one language can be indicated which does not permit choosing, for example, English spoken audio with Spanish captions.
The Quicktime player currently permits turning on and off any number of tracks individually, which can include audio, video, and text.
For user agents rendering video:
Checkpoints in this section: 5.8, 5.10, 5.9, and 3.8.
Implement the CSS positioning and/or SMIL layout languages. Allow the user to freeze a presentation, manually move and resize component video tracks (including captions, subtitles and signed translations) and to apply CSS stylesheets to text-based presentation and SVG.
For user agents rendering speech:
Checkpoints in this section: 3.8, 5.14, 5.15, and 5.16.
CSS2 ([CSS2] properties for speech can allow users to control speech rate, volume, and pitch. These can be implemented by allowing the user to write and apply a local style sheet, or can be automatically generated by means of (accessible) user controls, which should also be controllable through an API.
User interface:
Checkpoints in this section: 5.17.
Checkpoints in this section: 4.1, 4.3, 4.4, 4.10, 4.9, and 4.5.
[Ed. Add note here that while useful to turn off support, say for all images, it's also useful to be able to view one particular image.]
User agents may:
For example, user agents may recognize the HTML construct target="_blank" and spawn the window according to the user's preference.
Checkpoints in this section: 4.14.
Page refresh according to an author-specified time interval can be achieved with the following markup in HTML:
<META http-equiv="refresh" content="60">
The user agent should allow the user to disable this type of page refresh.
Checkpoints in this section: 4.13.
Although no HTML specification defines this behavior formally, some user agents support the use of the META element to refresh the current page after a specified number of seconds, with the option of replacing it by a different URI. Instead of this markup, authors should use server-side redirects (with HTTP).
User agents can provide a link to another page rather than changing the page automatically.
Checkpoints in this section: 9.1.
Checkpoints in this section: 10.2 and 8.2.
Checkpoints in this section: 8.7
Configuration of navigation, navigating sets of items (but not always linear, e.g., tables).
Checkpoints in this section: 8.4.
Sequential navigation includes all active elements. User agents might provide other navigation mechanisms limited to a particular type of element. For example "Find the next table" or "Find the previous form". The following checkpoints suggest some types of navigation.
Direct navigation:
Excessive use of serial navigation can reduce the usability of software for both disabled and non-disabled users. As a developer, you need to determine the point at which tabbing gets in the way and provide a keyboard alternative. This is done through the use of keyboard shortcuts. Note that user agents must provide information about available shortcuts (the current keyboard configuration) to users.
Checkpoints in this section: 8.6
Checkpoints in this section: 8.3
All users should be able to quickly determine the nature and purpose of a table. Examining the table visually often conveys a sense of the table contents with a quick scan of the cells. Users with blindness or low vision, users who have difficulty translating printed material, or users in an eyes-busy or speech-based environment may not able to do this. Providing table summary information, when first moving the point-of-regard to a table allows the nature of a table to be easily determined. In HTML, summary information for tables comes from the "summary" attribute on the TABLE element as well as the CAPTION element.
An auditory rendering agent, when the point-of-regard moves to a table, might say, "Table: Tax tables for 1998," thus identifying the nature of the table. The user could then use keyboard commands to move the point of regard to the next logical block of information, or use a different command to "burrow" into the table.
The "burrow" command should have an opposite "up" command, which would move the point of regard from an individual cell to the table as a whole, so that the user can leave a table from any cell within it, rather than navigating to the end.
If the user moves the focus up to look over the summary information, it should be possible to burrow back to the same cell.
When navigating a table that contains another table, this strategy can avoid confusion. For example, if each row of a table contained five cells, but the second row contained a 4x4 table in the third cell, a user could be disoriented when the row did not end as expected. However, when the point of regard moved to the third cell of the table, a compliant browser would report that this was a table, and describe its contents. The user would have the option of navigating to the forth cell of the parent table, or burrowing into the table within this cell.
When rendering tabular information, the fact that it is tabular information should be apparent. For a graphical user agent, such information is commonly made obvious by the border attribute or by visually apparent aligned white space between columns. However, for a non-graphical agent, such information must also be made evident.
As the user agent shifts the point of regard to a table, it should first allow users to access summary information about the table (e.g., the CAPTION element or the "summary" attribute in HTML). Access to this information allows the user to determine whether or not to examine the contents of the table, or to move the point of regard to the next block of content. Users should be able to choose not to have the summary information presented, if, for example, they visit a table frequently and don't want to hear the summary information repeated each time.
In many data tables, the meaning of the contents of a cell are related to the contents of adjacent cells. For example, in a table of sales figures, the sales for the current quarter might be best understood in relation to the sales for the previous quarter, located in the adjacent cell.
In order to provide access to contextual information for individuals using non-graphical browsers, or for individuals with certain types of learning disabilities, it is necessary for the user agent to allow the point of regard to be moved from cell to cell, both right/left and up/down via keyboard commands. The UA should inform the user when navigation has led to a table edge.
The most direct method of performing such navigation would be via the cursor keys, though other navigation strategies might be used.
Users of graphical browsers can easily locate cells within a table that are at the intersection of a row and column of interest. To provide equivalent access to users of non-graphical browsers, equivalent means of navigation should be provided. The search function of a browser will allow the user to locate key terms within a table, but will not allow the user to find cells that are at the intersection of rows and columns of interest.
Techniques:
The header information visible in a TH cell may be abbreviated, in which case it should be user preference to see the "abbr" value if any or the full contents.
Axis information may also help the user search into confined portions of the table.
Column groups and row groups are other confining partitions of a table in which a search may be limited.
Software:
Checkpoints in this section: 8.5.
Checkpoints in this section: 8.1, 9.3.
[Ed. @@CMN: For example, Opera does allow the user to navigate between views - in it's case various windows, as does emacs-w3. Other systems rely on the Operating System to do it - in MSWindows and the X window system there are keyboard and mouse driven methods for moving among open windows.]
Author-supplied navigation mechanisms such as navigation bars at the top of each page may force users with screen readers or some physical disabilities to wade through numerous links on each page of a site. User agents may facilitate browsing for these users by allowing them to skip recognized navigation bars (e.g., through a configuration option). Some techniques for doing so include:
Checkpoints in this section: 10.4.
Status information - on resource loading - should be provided in a device-independent manner. Techniques include text and non-text status indicators. Users should be able to request status information or have it rendered automatically. User agents may allow users to configure when status information should be rendered (e.g., by hiding or showing the status bar).
Screen readers may provide access on demand (e.g., through the keyboard) to the most recent status information, or to announce the new information whenever it changes.
Useful status information:
User agents may allow users to configure what status information they want rendered.
Checkpoints in this section: 9.4.
Checkpoints in this section: 9.5, 9.6, and 9.7.
Lynx ([LYNX] numbers each link and other element and provides information about the relative position of the section of the document. Position is relative to the current page and the number of the current page out of all pages. Each page usually has 24 lines.
Information about link status and other properties can be provided in an information view such as that provided by Netscape Navigator about how many and what types of elements are in a document.
User agents should not consider that all local links (to anchors in the same page) have been visited when the page has been visited.
User agents may use graphical or aural icons to indicate visited links or broken links.
Users should be able to:
Checkpoints in this section: 9.10 and 9.9.
Users of screen readers or other serial access devices cannot easily glean information about a page "at a glance". This is particularly difficult when accessing two-dimensional tables and trying to determine their content. Therefore, contextual information about tables (available from author-supplied markup or generated by the user agent) is very important to making them accessible.
Text metadata about tables can come from a number of elements, attributes, the structure of the table itself, or other sources. Useful information to make available to users includes:
Checkpoints in this section: 9.11, 10.6, 9.8.
Statement of form submission problems from Gregory Rosmaita:
Point A: As a user, I do not want to be prompted time I submit a form, provided that I submitted the form by activating its submit button. If, however, I simply hit the ENTER or the RETURN key from within a FORM control (i.e., rather than explicitly activating the SUBMIT mechanism), I would like the UA to request confirmation before submitting the form content.
Point B: As a user, I do NOT want the form content automatically submitted if I inadvertently press the ENTER or RETURN key.
PROBLEM STATEMENT FOR POINT B:
Inadvertently pressing the RETURN or ENTER key is quite a prevalent phenomenon amongst users of every level of expertise - especially those who often find it necessary to switch between user agents. Lynx, for example, uses the ENTER key within FORMs as a means of exposing drop-down (or pop-up, depending upon your point of view) SELECT menus. Thus, when one encounters a SELECT menu using Lynx, one: exposes the content of the menu by pressing the ENTER key, and then is able to navigate between OPTIONs using the up and down arrows or via Lynx's text-search feature. When one finds the appropriate OPTION, it is selected by pressing ENTER, which causes the selected item to be displayed in the SELECT menu listbox.
The problems posed by the default "submit on enter" feature of most GUI browsers, is not limited to the SELECT menu problem outlined above. Lynx (as well as several other text-based browsers) uses the ENTER/RETURN key as a means of toggling several FORM controls, such as the selection of checkboxes and radio buttons.
Moreover, I would like to stress that the "Auto-Submit-On- Enter" feature is not only quite problematic for one operating in an eyes-free environment, but for those unaccustomed to using online forms, and for those unfamiliar with a particular user agent's default key- bindings for forms, as well as those (like myself and countless others) who surf the web using a variety of browsers, often switching from browser to browser -- ALT- TAB-ing from Lynx32 to MSIE to Opera, for example -- in order to better comprehend the contents of a page or while attempting to navigate an poorly structured site or a poorly marked-up form
Point C: As a speech user, I am constantly frustrated and misdirected by the use of javascript and event handler controlled pseudo-forms, wherein the user is presented with a menu (in the form of a listbox in GUI browsers), and is redirected to a different viewport upon selection of an OPTION.
PROBLEM STATEMENT FOR POINT C:
The markup behind such pseudo-forms is a mix of javascript (in particular the "function switchpage(select)" command) and HTML FORM controls, which utilize HTML4's event handler script attributes (in particular the "onchange" event handler attribute has been defined. An example (gleaned from the document source for one Web site follows:
<SELECT NAME="condition" onchange="switchpage(this)">
When such a menu is encountered by a Web surfer who is using speech synthesis in conjunction with a javascript enabled user agent, his or her instinctual reaction will be to use the UA's navigation mechanism (usually the up and down arrows) to review the available OPTIONs. However, each time a new OPTION is displayed, the user is abruptly taken to a new viewport. Conversely, if one is using a user agent that does not support javascript (or has javascript support disabled), then the menu is displayed, but since there is no SUBMIT mechanism associated with it, there is no mechanism by which one can use the menu to quickly switch viewports - the apparent purpose of this type of pseudo-form. And while one can avoid having the viewport abruptly changed when encountering the menu (at least in the Windows environment) by using the ALT-LEFT-ARROW keystroke to display the menu in a drop-down list, (a) very few users know this keystroke, and (b) when one encounters a listbox on a page in an aural environment, one usually assumes that he or she is navigating a valid FORM, in which there are no unexpected side effects to perusing the contents of a SELECT menu using the arrow keys
Note. I have chosen to address the issue of pseudo-forms in this context, for, although they straddle the boundary between "Form Controls" and Checkpoint 5.8, pseudo-forms rely on FORM elements for activation This issue has been raised in the past, particularly by Chris Kreussling.
Techniques:
Refer also to [WAI-WEBCONTENT], checkpoint 6.3: Content developers must ensure that pages are accessible with scripts turned off or in browsers that don't support scripts.
Checkpoints in this section: 9.2.
User agents should provide information about:
Checkpoints in this section: 10.1, 10.3.
Metadata of all sorts: titles, dimensions, dates, relationships, etc. promotes accessibility by providing additional context to users. Text metadata is particularly useful since it can be rendered graphically, as braille, and as speech.
Checkpoints in this section: 2.1.
Checkpoints in this section: 2.2, 2.3, 2.6.
Some suggestions:
Some reserved keyboard shortcuts are listed in the appendix on accessibility features of some operating systems.
Checkpoints in this section: 9.12
Checkpoints in this section: 11.1
Configuration profiles allow individual users to save their user agent settings and re-apply them easily. This is particularly valuable in an environment where several people may use the same machine.
The user should be able to easily transfer profiles between installations of the same user agent. One way to facilitate this is to follow applicable operating system conventions for profiles.
Users should be able to switch rapidly between profiles (or the default settings). This is helpful when:
User agents may apply a profile when the user logs in. They may also allow users to apply settings interactively, for example by allowing them to choose from a list of named profiles in a menu.
Sample profiles (based on common usage scenarios) can assist users in the initial set up of the user agent. These profiles can serve as models and may be copied and fine-tuned to meet an individual's particular needs.
Checkpoints in this section: 2.4, 2.8, 2.7, and 2.5
[Ed. Discuss keyboard access here. How access is specified in HTML: "accesskey".]
[Ed. New section. Users must be allowed to control keyboard configuration based on specific needs. For poor motor control, keys far apart on a regular keyboard. For poor mobility, keys close together. General principle: fewest keystrokes, short distance to move.]
Checkpoints in this section: 11.2.
Checkpoints in this section: 12.2.
Universal design means that access to features that help accessibility should be integrated into normal menus. User agents should avoid regrouping access to accessibility features into specialized menus. Documentation includes anything that explains how to install, get help for, use, or configure the product. Users must have access to installation information, either in electronic form (CD-ROM, diskette, over the Web), by fax, or by telephone.
Checkpoints in this section: 12.3.
Include references to accessibility features in these parts of the documentation:
Checkpoints in this section: 12.1.
Documentation created in HTML should follow the Web Content Accessibility Guidelines.
Electronic documentation created in open standard formats such as HTML and ASCII can often be accessed in the user's choice of application such as a word processor or browser. Accessing documentation in familiar applications is particularly important to users with disabilities who must learn the functionalities of their tools and be able to configure them for their needs. Commonly used applications are also more likely to be compatible with assistive technology. Electronic documentation should not be provided in proprietary formats.
Users with print impairments may need or desire documentation in alternative formats such as Braille, large print, or audio tape. User agent manufacturers may provide user manuals in alternative formats. Documents in alternative formats can be created by agencies such as Recording for the Blind and Dyslexic and the National Braille Press.
User instructions should be expressed in an input device-independent manner. Provide instructions for using or configuring the user agent in a manner that can be understood by a user of any input device including a mouse or keyboard. For example, "Select the Home button on the toolbar" or "Select Home from the Go menu to return to the Home page."
Checkpoints in this section: 6.6
Develop the UA User Interface (UI) with standard interface components per the target platform(s). Most major operating system platforms provide a series of design and usability guidelines, these should be followed when possible (see platforms below).
These checklists, style guides, and human interface guidelines provide very valuable information for developing applications (e.g., UAs) for any platform/operating system/GUI. If your custom interface cannot provide information or operation as defined above, then you may need to design your UA using any additional options provided by that platform.
For instance, software should use the standard interface for keyboard events rather than working around it.
Evaluate your standard interface components on the target platform against any built in operating system accessibility functions (see Appendix 8) and be sure your UA operates properly with all these functions.
For example, take caution with the following:
User agents should support operating system or application environment (e.g., Java) conventions for loading assistive technologies. In the case of Java applets, the browser's Java Virtual Machine should support the Sun convention for loading an assistive technology. Writing an application that follows the Java system conventions for accessible software does not allow the applet to be accessible if an assistive technology designed for that environment cannot be run to make the applet accessible. Appendix 9 explains how an assistive technology developer can load its software into a Java Virtual Machine.
Ensure your UA can be operated using the standard interfaces on the target platform(s). Some example tests include:
Checkpoints in this section: 6.1, 6.2, 6.3, 6.5.
The operating system application programming interfaces (APIs) that support accessibility are designed to provide a bridge between the standard user interface supported by the operating system and alternative user interfaces developed by third-party assistive technology vendors to provide access to persons with disabilities. Applications supporting these APIs are therefore generally more compatible with third-party assistive technology.
The User Agent Accessibility Guidelines Working Group strongly recommends using and supporting APIs that improve accessibility and compatibility with third-party assistive technology. Third-party assistive technology can use the accessibility information provided by the APIs to provide an alternative user interface for various disabilities.
The following is an informative list of currently public APIs that support accessibility:
Many operating systems have built-in accessibility features for improving the usability of the standard operating system by persons with disabilities. When designing software that runs above an underlying operating system, developers should ensure that the application:
Write output to and take input from standard system APIs rather than direct from hardware controls where possible. This will enable the I/O to be redirected from or to assistive technology devices - for example, screen readers and braille devices often redirect output (or copy it) to a serial port, while many devices provide character input, or mimic mouse functionality. The use of generic APIs makes this feasible in a way that allows for interoperability of the assistive technology with a range of applications.
User agents should use standard rather than custom controls when designing user agents. Third-party assistive technology developers are more likely able to access standard controls than custom controls. If you must use custom controls, review them for accessibility and compatibility with third-party assistive technology.
For information about rapid access to Microsoft Internet Explorer's DOM through COM, refer to [BHO].
Checkpoints in this section: 6.4.
[Ed. Need to emphasize more of what DOM can do.]
A Document Object Model (DOM) is an interface to a standardized tree structure representation of a document. This interface allows authors to access and modify the document with client-side scripting language (e.g., JavaScript) in a consistent manner across scripting languages. As a standard interface, a DOM makes it easier not just for authors but for assistive technology developers to extract information and render it in ways most suited to the needs of particular users. Information of particular importance to accessibility that must be available through the DOM includes:
User agents should support W3C DOM Recommendations, including [DOM1] and [DOM2]. The W3C DOM Level 1 ([DOM1]) provides access to HTML and XML document information. The DOM Level 2 ([DOM2]) is made of a set of core interfaces to create and manipulate the structure and contents of a document and a set of optional modules. These modules contain specialized interfaces dedicated to XML, HTML, an abstract view, generic stylesheets, Cascading Style Sheets, Events, traversing the document structure, and a Range object.
It is important to note that DOM is designed to be used on a server as well as a client and therefore many user interface-specific information such as screen coordinate information is not relevant and not supported by the DOM specification.
Assistive technologies also require information about browser highlight mechanisms (e.g., the selection and focus) that may not be available through the W3C DOM.
Note. The WAI Protocols and Formats Working Group is focusing its efforts on the DOM as the conduit from which to extract accessibility information from and to enhance the accessibility of a rendered document through a user agent. It is this are should concentrate on for providing access to user agent documents.
[Ed. Link to W3C Note on HTML Accessibility Features instead of this section]
Checkpoints in this section: 7.2, 7.1
Following is a list of accessibility features in HTML:
Checkpoints in this section: 3.1.
User agents must be able to recognize sources of alternative representations of content.
Checkpoints in this section: 4.6 and 4.7.
User agents that recognize the following sources of blinking text and animations must enable users to freeze that text.
Checkpoints in this section: 3.5, 3.6.
Speech-based user agents providing accessible solutions for images should, by default, provide no information about images for which the author has provided no alternative text. The reason for this is that the image will clutter the user's view with unusable information adding to the confusion. In the case of an speech rendering, nothing should be spoken for the image element. This user should be able to turn off this option to find out what images were inaccessible so that the page author could be contacted to correct the problem.
In the case of videos, an assistive technology should, by default, notify the user know that a video exists as this will likely result in the launch of a plug-in. In the case of a video, user agents should indicate what type of video it is, accompanied by any associated alternative equivalent. User agents should prefer plug-ins that support system-specific accessibility features over those that don't. <>In the case of applets, an assistive technology should, by default, notify that an applet exists, as this will likely result in the launch of an associated plug-in or browser specific Java Virtual Machine. In the case of an Applet, the notification should include any associated alternative equivalent. This especially important for applet as applets typically do provide an application frame that would provide application title information.
When an applet is loaded, it should support the Java system conventions for loading an assistive technology (refer to Appendix 9). When the applet receives focus, the browser user agent should first notify the user about the applet as described in the previous paragraph and turn control over to the assistive technology that provides access to the Java applet.
Other techniques:
Checkpoints in this section: 3.7 and 5.9.
Scenario-video showing professor writing complex equations and graphs on the overhead and discussing them but not describing what he/she actually wrote on the overhead. Without description this would be inaccessible to people with visual impairments. This could be generalize to any video presentation of visually rich or complex information where the visually presented information is critical to the understanding of the presentation.
Checkpoints in this section: 3.7.
Checkpoints in this section: 4.12 and 4.11.
[Ed. See Scott's suggestions]
In HTML, "summary" attribute. Also, the "abbr" attribute for headers (see the section on table cells and headers).
Checkpoints in this section: 9.8.
[Ed. Talk about CSS pseudo-classes for :hover]
[Ed. Talk about using :before to clearly indicate that something is a link (e.g., 'A:before { content : "LINK:" }')]
Tables were designed to structure relationships among data. In graphical 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. The content of any table cell that visually wraps onto more than one line can be a problem. If only one cell has content that wraps, there is less problem if it is in the last column. 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.
A linear view of tables -- cells presented row by row or column by column -- can be useful, but generally only for simple tables. Where more complex structures are designed, allowing for the reading of a whole column from header downward is important as is carrying the ability to perceive which header belongs to which column or group of columns if more than one is spanned by that header. It is important for whole cells to be made available as chunks of data in a logical form. It might be that a header spans several cells so the header associated with that cell is part of the document chunk for that and each of the other cells spanned by that header. Inside the cell, order is important. It must be possible to understand what the relationships of the items in a cell are to each other.
Properly constructed data tables generally have distinct TH head cells and TD data cells. The TD cell content gains implicit identification from TH cells in the same column and/or row.
For layout tables, a user agent can assist the reader by indicating that no relationships among cells should be expected. Authors should not use TH cells just for their formatting purpose in layout tables, as those TH cells imply that some TD cells should gain meaning from the TH cell content.
When a table is "read" from the screen, the contents of multiline cells may become intermingled. For example, consider the following table:
This is the top left cell This is the top right cell of the table. of the table. This is the bottom left This is the bottom right cell of the table. cell of the table.
If read directly from the screen, this table might be rendered as "This is the top left cell This is the top right cell", which would be confusing to the user.
A user agent should provide a means of determining the contents of cells as discrete from neighboring cells, regardless of the size and formatting of the cells. This information is made available through the DOM ([DOM1]).
Non-graphical rendering of information by a browser or an assistive technology working through a browser will generally not render more than a single cell, or a few adjacent cells at a time. Because of this, the location of a cell of interest within a large table may be difficult to determine for the users of non-graphical rendering.
In order to provide equivalent access to these users, compliant browsers should provide a means of determining the row and column coordinates of the cell having the point of regard via keyboard commands. Additionally, to allow the user of a non-graphical rendering technology to return to a cell, the browser should allow a means of moving the point of regard to a cell based on its row and column coordinates.
At the time the user enters a table, or while the point of regard is located within a table, the user agent should allow an assistive technology to provide information to the user regarding the dimensions (in rows and columns) of the table. This information, in combination with the summary, title, and caption, can allow the user with a disability to quickly decide whether to explore the table of skip over it.
Dimensions is an appropriate term, though dimensions needn't be constants. For example a table description could read: "4 columns for 4 rows with 2 header rows. In those 2 header rows the first two columns have "colspan=2". The last two columns have a common header and two subheads. The first column, after the first two rows, contains the row headers.
Some parts of a table may have 2 dimensions, others three, others four, etc. Dimensionality higher than 2 are projected onto 2 in a table presentation.
The contents of a cell in a data table are generally only comprehensible in context (i.e., with associated header information, row/column position, neighboring cell information etc.). User agents provide users with header information and other contextual information. Techniques for rendering cells include:
User agents should use the algorithm to calculate header information provided in the HTML 4.0 specification ([HTML40], section 11.4.3).
Since not all tables are designed with the header information, user agents should provide, as an option, a "best guess" of the header information for a cell. Note that data tables may be organized top-to-bottom, bottom-to-top, right-to-left, and left-to-right, so user agents should consider all edge rows when seeking header information.
Some repair strategies for finding header information include:
The user may choose the form and amount of this information, possibly announcing the row heads only once and then the column head or its abbreviation ("abbr") to announce the cell content.
[Ed. The following issues were raised by Harvey Bingham.]
Checkpoints in this section: 5.17
Checkpoints in this section: 4.12.
Possible solutions:
Checkpoints in this section: 4.8.
Certain elements of the document language may have associated event handlers that are triggered when certain events occur. User agents must be able to identify those elements with event handlers statically associated (i.e., associated in the document source, not in a script).
[Ed. Talk about EMBED? Charles recommends that it be put inside an OBJECT if used.]
Checkpoints in this section: 7.2, 7.1
The accessibility features of Cascading Style Sheets (refer to CSS, level 1 and CSS, level 2) are described in [CSS-ACCESS].
Cascading Style Sheets may be part of a source document or linked externally. Stand-alone style sheets are useful for implementing user profiles in public access computer environments where several people use the same computer. User profiles allow for convenient customization and may be shared by a group.
Checkpoints in this section: 7.2, 7.1
The accessibility features of SMIL 1.0 (refer to [SMIL] are described in [SMIL-ACCESS].
Accessible presentation means here that the information can be easily accessed by using different media and that the user can easily control whether to show the information or not.
Several of the more popular mainstream operating systems now include a common suite of built-in accessibility features that are designed to assist individuals with varying abilities. Despite operating systems differences, the built-in accessibility features use a similar naming convention and offer similar functionalities, within the limits imposed by each operating system (or particular hardware platform).
The following is a list of built-in accessibility features from several platforms:
The next three built in accessibility features are not as commonly available as the above group of features, but are included here for definition, completeness, and future compatibility.
The next accessibility feature listed here is not considered to be a built in accessibility feature (since it only provides an alternate input channel) and is presented here only for definition, completeness, and future compatibility.
For information about Microsoft keyboard configurations (Internet Explorer, Windows 95, Windows 98, and more), refer to [MS-KEYBOARD].
The following accessibility features can be adjusted from the Accessibility Options Control Panel:
Other keyboard shortcuts:
The Keyboard Response Group (KRG) contains three functions: RepeatKeys, SlowKeys, and BounceKeys. The KRG can be turned on from the keyboard with the pre-stored user default settings. There should also be an emergency activation scheme to turn the KRG on in some minimal configuration for those times or for those users who cannot operate the computer keyboard without a particular KRG function (e.g., SlowKeys). Note. SlowKeys and BounceKeys are mutually exclusive. In other words, if the acceptance delay for SlowKeys is some value other than "0", then the delay value for BounceKeys must be "0". SlowKeys and BounceKeys can both be "0", or in effect off, while RepeatKeys is on, or either SlowKeys or BounceKeys can be on with RepeatKeys. Therefore the following KRG combinations can be set by the user:
The common modifier for activation of the KRG is to press and hold the right VK_SHIFT key for 8 seconds (note, emergency activation when the right VK_SHIFT key is held for 12 or 16 seconds.
Additional accessibility features available in Windows 98:
The following accessibility features can be adjusted from the Easy Access Control panel (Note: Apple convention uses a space within the accessibility feature names.)
The following accessibility features can be adjusted from the Keyboard Control Panel.
The following accessibility feature can be adjusted from the Sound or Monitors and Sound Control Panel (depends upon which version of the OS).
Additional accessibility features available for the Macintosh OS:
(Note: AccessX became a supported part of the X Window System X Server with the release of the X Keyboard Extension in version X11R6.1)
The following accessibility features can be adjusted from the AccessX graphical user interface X client on some DEC, SUN, and SGI operating systems. Other systems supporting XKB may require the user to manipulate the features via a command line parameter(s).
The following accessibility features are available from a freeware program called AccessDOS, which is available from several Internet Web sites including IBM, Microsoft, and the Trace Center, for either PC-DOS or MS-DOS versions 3.3 or higher.
There are several methods for developers to accomplish this. Most of these methods fall into four categories:
These methods are ordered as developments within a rapidly changing technology with the most recent advances/methods listed first.
Note. This method and the method described in the next section are very similar. What differs is the amount of, or capability of, the AT that actually gets loaded in the same process or address space as the User Agent.)
Access to application specific data across process boundaries might be costly in terms of performance. Therefore, user agents may wish to provide a mechanism to load the entire assistive technology (AT), into the process space of the application as a separate thread with direct access to the DOM.
One technique is to store a reference to an assistive technology in a system registry file or, in the case of Jave, a properties file. Registry files are common among many operating system platforms.
In Windows you have the system registry file. On OS/2 you have the system.ini file and on distributed network client networks you often have a system registry server that an application running on the network client computer can query.
In Java 2, the existence of an "accessibility.properties" file causes the system event queue to examine the file for assistive technologies required for loading. If the file contains a property called "assistive_technologies", it will load all registered assistive technologies and start them on their own thread in the Java Virtual Machine that is a single process. An example entry for Java is as follows:
assistive_technologies=com.ibm.sns.svk.AccessEngine
In Windows, a similar technique could be followed by storing the name of a Dynamic Link Library (DLL) for an assistive technology in a designated assistive technology key name, AT pair. An example entry for Windows could be as follows:
HKEY_LOCAL_MACHINE\Software\Accessibility\DOM "ScreenReader, VoiceNavigation"
Once the assistive technology is determined from the registry, any user agent on the given operating system can now determine if an assistive technology needs to be loaded with their application and load it.
On a non-Java platform, a technique to do this would be to create a separate thread with a reference to the User Agent's DOM using a Dynamic Link Library (DLL). This new thread will load the DLL and call a specified DLL entry name with a pointer to the DOM interface. The assistive technology's task will then run until such time as is necessary to end communication with the DOM.
Once loaded, the assistive technology can monitor the DOM as needed. The assistive technology has the option of communicating with a main assistive technology of its own and process the DOM as a caching mechanism for the main AT application or be used as a bridge to the DOM for the main assistive technology.
In the future, it will be necessary to provide a more comprehensive reference to the application that not only provides direct access to it's client area DOM, but also multiple DOM's that it is processing and an event model for monitoring them.
Java is a working example where the direct access to application components is performed in a timely manner. Here, an assistive technology running on a separate thread monitors GUI events such as focus changes. Focus changes give the AT notification of which component object has focus. The AT can communicate directly with all components in the application by walking the parent/child hierarchy and connecting to each component's methods and monitor events directly. In this case an AT has direct access to component specific methods as well as those provided for by the Java Accessibility API. There is no reason that a DOM interface to UA components could not be provided
In Java 1.1.x, Sun's Java access utilities load an assistive by monitoring the Java awt.properties file for the presence of assistive technologies and loads them as shown in the following code example:
import java.awt.*; import java.util.*; String atNames = Toolkit.getProperty("AWT.assistive_technologies",null); if (atNames != null) { StringTokenizer parser = new StringTokenizer(atNames," ,"); String atName; while (parser.hasMoreTokens()) { atName = parser.nextToken(); try { Class.forName(atName).newInstance(); } catch (ClassNotFoundException e) { throw new AWTError("Assistive Technology not found: " + atName); } catch (InstantiationException e) { throw new AWTError("Could not instantiate Assistive" + " Technology: " + atName); } catch (IllegalAccessException e) { throw new AWTError("Could not access Assistive" + " Technology: " + atName); } catch (Exception e) { throw new AWTError("Error trying to install Assistive" + " Technology: " + atName + " " + e); } } }
In the above code example, the function Class.forName(atName).newInstance() creates a new instance of the assistive technology. The constructor for the assistive technology will then be responsible for monitoring application component objects by monitoring system events.
In the following code example, the constructor for the assistive technology "Access Engine," adds a focus change listener using Java accessibility utilities. When the assistive technology is notified of an objects gaining focus it has direct access to that object. If the Object, o, implemented a DOM interface the assistive technology would now have direct access to the DOM in the same process space as the application.
import java.awt.*; import javax.accessibility.*; import com.sun.java.accessibility.util.*; import java.awt.event.FocusListener; class AccessEngine implements FocusListener { public AccessEngine() { //Add the AccessEngine as a focus change listener SwingEventMonitor.addFocusListener((FocusListener)this); } public void focusGained(FocusEvent theEvent) { // get the component object source Object o = theEvent.getSource(); // check to see if this is a DOM component if (o instanceof DOM) { ... } } public void focusLost(FocusEvent theEvent) { // Do Nothing } }
In this example, the assistive technology has the option of running standalone or acting as a cache for a bridge that communicates with a main assistive technology running outside the Java virtual machine.
Access to application specific data across process boundaries might be costly in terms of performance. Therefore, user agents may wish to provide a mechanism to load part of the assistive technology (AT) into the process space of the application as a separate thread, with direct access to the DOM, to provide the specific functionality they require. This could consist of a piece of stub code, a DLL, a Browser Helper Object, etc. An example of how to do this follows.
In order to attach to a running instance of Internet Explorer 4.0, you can use a "Browser Helper Object." A "Browser Helper Object" is a DLL that will attach itself to every new instance of Internet Explorer 4.0 (only if you explicitly run iexplore.exe). You can use this feature to gain access to the object model of a particular running instance of Internet Explorer. You can also use this feature to get events from an instance of Internet Explorer 4.0. This can be tremendously helpful when many method calls need to be made to IE, as each call will be performed much more quickly than the out of process case.
There are some requirements when creating a Browser Helper Object
For more information, please check out:
http://support.microsoft.com/support/kb/articles/Q179/2/30.asp.In order for native Windows ATs to gain access to Java applications without the creating a Java native solution Sun Microsystems provides the "Java Access Bridge." This bridge is loaded as an AT as described in section 6.1.3. The bridge uses a Java Native Invocation (JNI) to Dynamic Link Library) (DLL) communication and caching mechanism that allows a native assistive technology to gather and monitor accessibility information in the Java environment. In this environment, the AT determines that a Java application or applet is running and communicates with the Java Access Bridge DLL to process accessibility information about the application/applet running in the Java Virtual Machine.
Specialized user agents might also include the necessary assistive technology as part of their interface, and thus provide possibly the best of both worlds. An example would be pwWebSpeak, from The Productivity Works (refer to [Prodworks]).
[Ed. Have The Productivity Works provide short description.]
Access to application specific data across process boundaries or address space might be costly in terms of performance. However, there are other reasons to consider when accessing the User Agent DOM that might lead a developer to wish to access the DOM from their own process or memory address space. One obvious protection this method provides, is that if the User Agent application fails, it doesn't disable the user's AT as well. Another consideration would be legacy systems, where the user relies on their AT for access to other applications as well as the User Agent, and thus would have their AT loaded all the time, not just for accessing the User Agent.
There are several ways to gain access to the User Agent's DOM. Most User Agents support some kind of external interface, or act as a mini-server to other applications running on the desktop. Internet Explorer is a good example of this, as IE can behave as a component object model (COM) server to other applications. Mozilla, the open source release of Navigator also supports cross platform COM (XPCOM).
An example of using COM to access the IE Object Model can be seen in the code snippet below. This is an example of how to use COM to get a pointer to the WebBrowser2 module, which in turn allows you to get a interface/pointer to the document object, or IE DOM for the web page in view.
/* first, get a pointer to the WebBrowser2 control */ if (m_pIE == NULL) { hr = CoCreateInstance(CLSID_InternetExplorer, NULL, CLSCTX_LOCAL_SERVER, IID_IWebBrowser2, (void**)&m_pIE); /* next, get a interface/pointer to the document in view, this is an interface to the document object model (DOM)*/ void CHelpdbDlg::Digest_Document() { HRESULT hr; if (m_pIE != NULL) { IDispatch* pDisp; hr = m_pIE->QueryInterface(IID_IDispatch, (void**) &pDisp); if (SUCCEEDED(hr)) { IDispatch* lDisp; hr = m_pIE->get_Document(&lDisp); if (SUCCEEDED(hr)) { IHTMLDocument2* pHTMLDocument2; hr = lDisp->QueryInterface(IID_IHTMLDocument2, (void**) &pHTMLDocument2); if (SUCCEEDED(hr)) { /* with this interface/pointer, IHTMLDocument2*, you can then work on the document */ IHTMLElementCollection* pColl; hr = pHTMLDocument2->get_all(&pColl); if (SUCCEEDED(hr)) { LONG c_elem; hr = pColl->get_length(&c_elem); if (SUCCEEDED(hr)) { FindElements(c_elem, pColl); } pColl->Release(); } pHTMLDocument2->Release(); } lDisp->Release(); } pDisp->Release(); } } } }
For more information on using COM with IE, please visit the Microsoft web site:
http://www.microsoft.com/com/default.aspFor more information on using XPCOM with Mozilla, please visit the Mozilla web site:
http://www.mozilla.org/For a working example of the method described in 6.1.4, please visit the following web site and review HelpDB, developed as a testing tool for web table navigation.
http://trace.wisc.edu/world/web/document_access/This index lists each checkpoint and the sections in this document where it is discussed. Furthermore, each guideline number links to its definition in the guidelines document. Each checkpoint also links to its definition in the guidelines document.
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, Glen Gordon, John Grotting, Markku Hakkinen, Earle Harrison, Chris Hasser, Kathy Hewitt, Philipp Hoschka, Masayasu Ishikawa, Phill Jenkins, Jan Kärrman (for help with html2ps), 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, Gregory 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,
Note. W3C cannot maintain stability for any of the following references outside of its control. These references are included for convenience.