The following are summaries of the presentations based on meeting notes, recordings, and edits of presenters.
TV Raman, Google
In this presentation, T.V. Raman asserts that new technologies should be seen as new opportunities, rather than as obstacles and complications. Through coordinated development of content, the user interface, and assistive technology, the reach of people using assistive technology can be increased. Newer light-weight components even make it possible to quickly create custom interfaces, delivering useful information in a wide range of contexts and usable by a variety of technologies.
Hosted Web applications offer the promise of easy deployment, light-weight user interaction, ubiquitous access to data, and easy upgrades, but a shift in approach will be needed before they work with assistive technologies (AT). To an extent, the blind community has put itself in a bind by saying something is accessible if it works with a screen reader. A better approach is to view accessibility as usability done properly, with graceful degradation for less capable technologies.
Access goals with Web applications are to:
The building blocks of access are the following:
To build speech access, we need to be able to identify what to speak, determine how to speak it, and decide when to speak.
The Web application model provides an opportunity for AT to extend and embrace the Web model.
The Web browser becomes a container for the Web application, functioning as a universal client.
Custom interfaces are doable in the Web world, providing new opportunities for access.
New adaptive technologies can be built from the growing arsenal of light-weight components.
Separation of content from interaction is laying the foundation for mashups. What is the accessible equivalent of a mashup?
Inspiration for innovative AT could come from audio games. Games often lead to UI innovation.
In conclusion, it is important to build on what we have, but thinking only in terms of current AT is too limiting.
TV Raman, Google; Doug Geoffray, GW Micro
Early AT software simply captured text, but as pages and interfaces became more complex, accessibility aids such as Microsoft Active Accessibility were needed. Now, with the appearance of dynamic HTML content, new approaches must be developed. IBM is working with W3C, FireFox, and others to address this need. A key challenge is the lack of coordination among AT developers.
With early AT software, content could be read by capturing the text output, but as pages became more complex, that approach could yield gibberish. Microsoft Active Accessibility (MSAA) gave AT a way to get at content within, but still pages could have much more complexity that was indicated by MSAA. GW Micro's Window-Eyes screen reading software is designed to use MSAA and the Windows Document Object Model (DOM) to move through content, but is still basically designed for reading static pages. Users can move back and forth between a virtual mode, which navigates a copy of the page, and a browse mode which interacts directly with the page, such as when doing forms. GW Micro is working with IBM to find ways to handle dynamic HTML elements.
As Web applications move into the browser, one challenge is that virtual mode is becoming useless because it does not know about dynamic changes in page content between page loads. A standard method is needed for AT to communicate with the Web applications. How can menu items be presented to the blind user? How can Web applications have the consistency of method AT needs when there are so many different authors and applets?
Early text browsers, such as Emacspeak, demonstrated the advantages of formatting text in terms of its type (from, to, subject). How can we bring the content out of the prison of the screen reader? Why can't the user community build its own set of tools using tools such as GreaseMonkey scripts?
Among assistive technologies, a key problem is that there are too many control key combinations. Every possible combination has been taken and getting AT software developers to coordinate is very difficult. Also, people writing self-voicing applications do not necessarily know what people who are blind need. Not every blind person wants audio. Some just want Braille. Other problems are that the infrastructure in Windows is inconsistent and there are a huge number of APIs AT must be designed to deal with. The result of this confusion is that innovation with respect to the blind has come to a standstill.
Publishing and sharing knowledge about these challenges is the way to get things moving. IBM has focused on open standards:
With the growth of cell phones and MP3 players, are we seeing a shift to aural delivery? Not necessarily. Quality is the key. What a blind user needs to hear is different from what a sighted person needs. You need to have a fairly high threshold of pain to deal with current voice software. The general user has a lower pain threshold and expects higher sound quality. An interesting aspect of the aural delivery idea is how the interaction and aural content might be managed to produce a distinct sound and feel, such as of a Nike shoes commercial site.
The wide variety of ways data in pages are organized is a major obstacle in writing AT for rich Internet applications. A standard interoperativity contract between the Web application and AT is needed. The Roadmap for Accessible Rich Internet Applications identifies gaps that need to be filled. A review and update of guidelines, standards, and laws relating to accessibility will be needed as these new methods are developed and implemented.
We are moving from the static document into the rich desktop experience. The Roadmap for Accessible Rich Internet Applications (ARIA) is a gap analysis aimed at providing plans and guidelines we can use to addressing the holes.
Accessibility of a graphical user interface (GUI) is about getting to the data, but everyone's data is organized differently. We need an accessibility application programming interface (API) that defines a standard contract between the Web application and the assistive technology which should include the following:
Keyboard accessibility is an example of the problems we face:
Work is underway to develop new standards for XHTML:
The goal is to provide information to map the content to the accessibility API, even as the content changes. As we see more distributed development of mashups, these standards could be built into the tools used to build them. The Dojo AJAX library is already including many of these ideas.
Part of the effort is to clarify best methods, such as in menus utilizing tabs only for the highest level and using arrow keys within the menu tree.
We need to address legislation relating to accessibility. Most of the legislation (WCAG1, 508) is based on 1998 browser technology and is rapidly becoming an obstacle to finding solutions. We need to focus on interoperability and usability, rather than try to exclude technologies. The IMS Global Learning Consortium accessibility specifications are an example of what could done.
Resources on this topic include the following:
Recognizing the importance of accessibility in online education, process and procedures addressing it were built into course design, quality assurance, and instructor training. Instructors want to use more rich media. Evaluating sites is challenging as content is built with more technologies. Developers, excited about new technologies such as AJAX, can easily forget about accessibility.
When developing independent study programs, the question of accessibility came up. If someone identified a specific problem we would fix it. Then Sheryl Burgstahler gave us a presentation on accessibility, making it clear it was a lot easier to build Web sites that are accessible from the start, rather than go back and retrofit sites. As a result, Educational Outreach set up processes and procedures including improved Web skills, incorporating checking for accessibility features into the Quality Assurance process, and incorporating accessibility into the design process so that everyone would be talking about it, including the instructors.
A system of indicators was developed for evaluating distance learning education sites.
University of Wisconsin-Madison, which has a resource center for students with special needs, has had a Web accessibility policy since 2001.
As a lead technical analyst for electronic research administration systems at the University of Washington, Cheryl has become passionate about accessibility.
Developers, enthusiastic about trying new technologies like AJAX, are frequently an obstacle to accessible design. Developers want to provide more interactive interfaces and quicker response time. What roadblocks are in the way between us and using new technologies?
Developers who work on internal and faculty applications can develop a false sense of security of only dealing with a limited audience. As we make ourselves more dependent on electronic technology, we are going to need to insure that our technology works for everyone now and for any people we may hire in the future.
How can you evaluate accessibility?
When there are legal requirements, it is worth doing more work to fully understand why the requirements are the way they are. When federal money is involved, you should be 508 compliant.
How accessible is AJAX? .Net developers will be using Microsoft libraries, including ATLAS. Open source libraries are way ahead in terms of accessibility. The Dojo Toolkit is available and includes many accessibility features. Saying technology is not available is a punt.
Bob Regan, Adobe
Current accessibility standards tend to push us toward text oriented content, but for many people, interactivity addresses how they learn. In rich media, label, role, state, and structure need to be provided for for good interaction between AT and content. Evaluating usability of rich media presents some puzzles, including control the time axis and notifying the user of content changes.
The techniques for making rich Adobe applications accessible are easy. The hard part is getting developers to use the techniques.
In Web applications, single screen updates, diverse controls (buttons, sliders), and live data updates (constantly streaming data to the browser) can be difficult for someone using AT to follow.
When developing Web applications with technologies like AJAX and Flash (including Flex, a structured language for creating Flash content), we have a number of standards we can work with.
We be able to evaluate designs, we need a base set of assistive technology to test against. The disabled are not the only community we have to address, but designs must work for the blind community - that is a baseline requirement. Needs of the blind can be addressed with both audio and tactile interfaces.
Usability is an important question. A design feature can work in AT, but will anyone use it? Associating usability and accessibility is tricky. One site with 37 frames, each fed by a different data stream, followed standards, but will it make sense and be meaningful to the user?
One approach to accessible design is disability use cases:
Needs can be conflicting. For someone who is blind, the most accessible form of content is text. For a person with a cognitive disability, the least accessible form of content may be text.
If we prohibit interactive media such as Flash, we will be leaving a large group of people behind. Interactivity addresses how many of us learn.
Training developers about accessibility is challenging:
Successful development of accessible Web applications must address four characteristics of elements and objects:
Rich applications present some puzzles:
Preserving opportunity and availability can be approached with three techniques; (1) standards based development, (2) redundant interfaces offering multiple approaches to content, and (3) designs that mimic functionality the user is already comfortable with. Menus can be built from unordered lists with display controlled by CSS or scripts, for example. AT that does not use the CSS or scripts can still access the information because its of simple, semantic structure.
Todd has been working on the menu bar in the YUI library. Menus are a basic way people navigate a site. The menu mechanism needs to be accessible for the site to work.
Doug is developing AT software and often gets calls from developers who want to do the right thing.
Libraries of tools provide a means to provide and support well thought-out methods of design. Yahoo! wanted to move to CSS layout but did not always have people with CSS layout skills. The solution was to develop a grid kit, available in the Yahoo! UI Library.
Web 2.0 is about getting it right the second time. Now we have browsers that can support a range of standardized technologies (CSS, XHTML, XML, JavaScript). If we use the technology as designed (such as avoiding non-semantic class names) we can create evolvable platforms that preserve opportunity and availability.
One definition for accessibility is "a general term used to describe the degree to which a system is usable by as many people as possible without modification."
We can approach that kind of accessibility by using three techniques:
A bad practice is to think that Web 2.0 means you can throw out all we have learned about accessibility.
With these techniques in mind we can look at how menus should be made in Web technology.
The list approach goes "with the grain" of web technologies and results in pages that are truly available to all. What we create is likely to be out there a long time. It is important to do things right from the beginning.
Redundant interfaces offering multiple means of input make possible flexible interactions. A user can have the choice of GUI or command line input, text fields can have the option of auto complete, and support can be provided for tab and arrow keys. This an example of progressive enhancement:
Semantic markup makes content portable. Progressive enhancement allows for the development of redundant interfaces that give users a choice.
Redundant interfaces are better for everybody. The keyboard is as important as the mouse. Users can choose among multiple task flows. The approach is limited by insufficient communication with accessibility APIs on the desktop at present. While it can require development of two experiences, it does not necessarily require twice the effort and can benefit the development process.
Design that is faithful and predictable to the desktop experience has greater learnability and discoverability. Completeness is critical to preserve the illusion of consistency between the desktop and the Web application.
The WAI-ARIA roles and states utilize a powerful and well-understood desktop accessibility API, providing a standard and predictable enrichment of markup.
The benefits of this approach are more options, better discoverability, better usability, and support of many working styles. Drawbacks are that it is not easy, seems heavier and more complex, and is not the path of least resistance.