Literature Review

This chapter reviews relevant literature and activities in order to locate AccessForAll work within the appropriate context.

The AccessforAll framework brings together two worlds of intellectual endeavour - accessibility work and metadata work. In both domains, there have been previous models and approaches on which the current work builds. Given the significance and complexity of both accessibility and metadata work, these concepts will be described and discussed in detail. In this chapter, there is a brief introduction only in order to make sense of the chapter. The focus will be an introduction to and critical discussion of the role of the previous models and frameworks and the context in which they are operating.

The Context for Web Resource Accessibility

Quantifying the accessibility context

For many years, Microsoft showed its skepticism for universal accessibility including by its lack of effort to make its Internet Explorer browser UAAG conformant. In 2003, however, Microsoft commissioned a study in the US to get some indication of who might be needing assistance with accessing information if they are to try using computers or other electronic devices. The overall population in the US in the age range 18 to 64 years was found to be divided into the following four groups: those with severe, mild, minimal and no difficulties, in the following proportions: 25% with severe, 37% with mild, and 37% with minimal or no difficulties resulting from disabilities.

Three more sets of figures provide the incentive to think carefully about accessibility in the general population:

paint a picture for the US that looks grim. There is clearly a worrying trend towards much higher proportions of the community being much older than at present, and therefore more likely at risk of disability. There is every reason to assume the figures will be similar in Australia.

This points to the fact that not all those who could benefit from computer use, in fact do use computers. There are many reasons for this, but as the trend to publish becomes electronic and the younger people adopt the technology, there is clearly going to be an increasing problem unless accessibility is also rapidly increased.

While Microsoft was working to convince, or otherwise, itself of the need to pay attention to accessibility issues, Texthelp Systems Inc. has a different slant because they have developed a solution at least for a high proportion of those with disabilities. They claim:

as justification for their product BrowseAloud. BrowseAloud is a service that can be offered by a web site to provide streamed reading aloud of the content of the site, assuming it is properly constructed.

In 2006, the US National Council on Disability released a policy paper that explores key trends in information and communication technology, highlights the potential opportunities and problems these trends present for people with disabilities, and suggests some strategies to maximize opportunities and avoid potential problems and barriers. In particular,

The report points out that technologies in common use change fast and unpredictably with the result that "assistive technology developers cannot keep pace". They cite convergence and competitive differences as having "a negative effect on interoperability between AT and mainstream technology where standards and requirements are often weak or nonexistent". The rapid increase in the number of aging people who have naturally increasing disabilities is, of course, always a concern.

On a more positive note, the NCD report summary lists a number of technological advances and says:

Accessibility as Economic Advantage

In 1998, the US Federal Government legislated in favour of accessibility of digital resources including applications when the US federal government is procuring content, systems or services [s508]. As the largest employer of people with disabilities in the US, the Federal Government is also responsible for social security (income replacement) including for people with disabilities. There may have been some connection between the two because it is clearly better in a number of ways for the US Federal Government to offer useful employment to their citizens with disabilities than to have to support them all on disability pensions.

Fairfax in Australia, however, has perhaps offered a similarly striking economic reason for being concerned about accessibility. In 2003, they redeveloped their Web site with accessibility in mind and the result is a saving of an estimated $1,000,000 per year in transmission costs.

In a 2004 presentation for the Web Standards Group [WSG], Brett Jackson, Creative Director of Fairfax Digital, reported that Fairfax achieved more than the following success with a major move to the XHTML/CSS platform.

In 2003, a surprisingly high proportion of the Webby award winners (organised by the International Academy of Digital Arts and Sciences) were found to have accessible sites despite their multimedia attraction. In the opinion of Bob Regan, the accessibility expert for Macromedia, the vendors of DreamWeaver and Authorware, the Webby winners did not have accessible sites so much because they were concerned about accessibility as because they were concerned to use the latest, smartest techniques, and these inevitably led to increased accessibility (Regan, 2004).

The Authoring Tools Accessibility Guidelines [ATAG] can be used as functional requirements for the accessibility of authoring tools of all kinds. The underlying belief is that if the tools are designed so as to promote accessible products, inadvertently, simply by using the tools, authors of resources will make their products accessible. The author, involved in the development of ATAG, asserts that, in fact, if those who are so concerned about training their authors about accessibility were to save the money and time involved and instead buy them better authoring tools, more might be achieved with the same amount of money.

The European perspective

Universal Accessibility - the W3C Approach

There is no sense in which one would want to 'fault' the work of WAI in the area of accessibility. Like others, they have struggled to deal with an enormous and growing problem and everyone has contributed all they can to help the cause. Nevertheless, it is clear that the work of WAI alone cannot make the Web accessible. Although there has been a lot written about the achievements of the universal access approach, that is not the topic but rather the context for the current work, and working to increase its effectiveness is a major goal.

The UK Disabilities Rights Commission Report

On 27/3/03, the UK Disabilities Rights Commission [DRC] issued a press release announcing its "First DRC Formal Investigation to focus on web access". They planned to investigate 1000 Web sites "for their ability to be accessed by Britain’s 8.5 million disabled people". They said that "A key aim of the investigation will be to identify recurrent barriers to web access and to help site owners and developers recognise and avoid them."

Significantly, this testing would not just be done by people evaluating the Web sites against a set of specifications, but they would also involve 50 disabled people in in-depth testing of a representative sample of the sites, testing in their case for practical usability. They claimed that "This work will help clarify the relationship between a site’s compliance with standards and its practical usability for disabled people." Bert Massie, Chairman of the DRC, said: “The DRC wants to see a society where all disabled people can participate fully as equal citizens and this formal investigation into web accessibility is an important step towards that goal.”

The DRC has legal power. As Mr Massie said: “Organisations which offer goods and services on the web already have a legal duty to make their sites accessible. The DRC is committed to enforcing these obligations but it is also determined to help site owners and developers tackle the barriers to inclusive web design.” (DRC, 2003)

On 30 April 2003, Accessify carried the following report on the briefing for the DRC project:

Beyond establishing the proposed methodology, the DRC project leader claimed that they would:

A year later, after the report was released, OUT-Law published an article about it (2004):

Out-law's commentary is interesting because it takes a critical position with respect to the report and its relationship and comments on the W3C WCAG Version 1 and 2. These comments will be considered in more detail in following chapters.

DRC Report findings

The report authors tend to use the term 'inclusive design' rather than universal design.

and further confirmed the lack of success in achieving accessibility of Web sites by the introduction of the guidelines and the local legislation. This time they were reporting on the state in the UK:

The overall finding includes the comment that compliance with the WAI guidelines does not ensure accessibility. Finding 2 contains the sub-point 2.2 :

The report goes on to describe many things that could be done by humans including training of web content providers and web users, proactive efforts by people with front-line responsibility such as librarians etc and more....

The level of compliance with the guidelines was amazingly low, even given the common perception that compliance levels are not high:

The report contains many statistics about the speed with which the users were able to complete tasks in what is generally to be understood as usability testing. It showed, in the end, that usable sites were usable and this, regardless of disability needs.

The final statement here is most important. It is the main thesis of the DRC Report that usability testing involving people with disabilities is essential to the accurate testing of content.

An important finding of the report was the extremely low level of accessibility of resources. It is explained:

What is significant here is that there is such a low rate of universal or, as Petrie says, inclusive accessibility. While this should not be taken as a sign of failure of those accessibility goals, it does suggest that there is a great need for more to be done, and that it is unlikely to be done by the original content creators. This means that third party support should be enabled, and that is dependent on protocols to enable it.

While this is a possible conclusion, it is asserted that the conclusion could equally have been that a better method of ensuring user satisfaction should be developed. There is a general emphasis on responsibility and training in many commentaries on accessibility. Many examples of calls for training of creators, for example, are similar to those within this report but perhaps this responsibility is misplaced. It is interesting to note also that the Report advocates more trust of users to select what they need and want (possibly represented by assistants).

If money is to be spent, the use of better authoring tools may prove cheaper than the training being advocated. And if users need to be served better, perhaps removing the need for them to translate their own needs into assistive technologies is somewhat more attractive.

But the question might be about what is the role these assistants should be trained to contribute. Perhaps training them to complete a simple questionnaire about their needs and preferences would be the easiest and most effective use of training time. Of course, this would only be possible if applications acted on those needs and preferences, and this does mean server improvements. (Note that this issue is considered in some detail in chapter ....)

But again this advice is based on a narrow definition of users that was the subject of the report, namely those with disabilities, as made clear in the following extract:

The Report's definition of users is implicitly limited by the scope of the Report. Accessibility, in general, is a far broader issue with a wider scope. There is no way that there could be user groups of the kind suggested by the Report that would cater for all the situations that account for inaccessibility. The various combinations of needs would not be different but identification of classes of needs would be too difficult and the individual differences in needs and preferences would be lost in the process.

Responses to DRD Report

Kelly et al (2005) argue that the DRD report and other evidence show that there is not yet a good solution to the accessibility problem but that it clearly does not rest in a set of technical authoring guidelines. In fact, they list factors that need to be taken into account in the determination of accessibility:

This framework offers one way of thinking about the problems. But a year later many of the same authors offered what they call the 'tangram' approach (see below). It should be noted that the proposed AccessForAll approach provides an operational framework that can include any and all of these contextual issues.

From accessible code to accessible services

The authors of "Developing A Holistic Approach For E-Learning Accessibility" (Kelly, Phipps & Swift, 2004), point to surveys of accessibility of higher educational sites undertaken in the UK before the DRC Report and comment that the findings are similarly not good,

But the thrust of their paper is that there is more to accessibility than a technical analysis of conformance with WAI guidelines and that such things as blended learning may provide better solutions in the UK. Blended learning is learning that is not only technology based but includes physical objects and the role of people such as assistants, maybe family members. Jutta Treviranus, on the other hand, in her Keynote address at the 2004 OZeWAI Conference [OZeWAI 2004], emphasised that there is an effort in Canada to use the technology, to exploit the artificiality of it and let it provide for people according to their needs and preferences in ways that humans in the physical world have and often can not (Treviranus & Roberts, 2006). This position does not deny the possibility of human and physical help, but it does make strong demands on the technology for those situations in which it is involved.

There is no reason to follow one approach or the other but rather it is important to be aware of both. Within educational contexts in the UK, the 'SENDA' legislation requires reasonable accommodations to be made to promote inclusive learning. Kelly et al (2005) argue this is done by adopting a holistic approach to accessibility. Where learning is being undertaken in an online environment, the technology should be operating to its highest level of support for accessibility, as required in Canada.

Kelly et al (2005) are raising the expectations in terms of responsibility for teachers, parents, institutions and their performance; the DRC expect the support communities to take a greater role, and Treviranus claims the AccessForAll approach wants more from the technology: while Kelly et al argue for standing back from the online life and including other aspects of life, Treviranus argues that standing back from the original resource and providing what it contains in a form the user can access is what is needed. Kelly et al do this offline and AccessForAll requires the server to do it. In essence, they share the holistic model although they differ in their dependence on computers because they are working in different contexts. Another point of view on their perspectives, and those of the DRC, W3C, and others, asks what burdens are they placing on the humans, and how well can they respond?

By 2006, Kelly and colleagues (Kelly et al, 2006) are moving away from what they describe as their earlier absolute solution to what they refer to as their tangram metaphor with multiple possibilities for satisfaction. They argue that the W3C tests provide a good base for accessibility but do not solve the problems and cannot - there are too many other factors involved.

In a more recent exercise., Kelly and Brown (2007) proposed Accessibility 2.0 and called for greater variety being incorporated into the provision of accessibility.

Responsible for accessibility

In referring to the Australian legislative context for discrimination, Michael Bourk says:

Additionally, Bourk (1998) makes the point that early on, in the case of Scott and DPI (A) v Telstra (HREOC, 1995),

In other words, says Bourk, the case establishes it is the service not the objects that must be accessible. He says:

Bourk was writing as a student of Tom Worthington, a recognised Australian expert in accessibility, and an expert witness in the Maguire v SOCOG accessibility case (HREOC, 1999). Bourk makes two points of interest: accessibility is a quality of service and the need for attention is not merely that some people have medical disabilities. Both ideas are fundamental to the work being reported and of particular relevance in Australia.

In fact, the guidance notes for Australian regulations that extend the Australian Disabilities Discrimination Act say:

While Australian legislation, for example, follows others in using WCAG as the standard specifications for Web content, it is clear that the test of accessibility is not just conformance to the guidelines.

Kelly at al (2005) point out that the W3C Guidelines do not claim to be the arbiters of accessibility but it is clear from most work in the field that they are often used this way. With respect to the W3C position, Kelly et al (2005) argue:

The summary of the US National Council on Disability's "Over the Horizon: Potential Impact of Emerging Trends in Information and Communication Technology on Disability Policy and Practice" concludes with the comment that:

The AccessForAll framework developed for descriptions of accessibility needs and preferences and of resource characteristics enables the development of tests (of descriptions of resources) that are far more objective and testable than the WCAG criteria that have been shown to be both frequently misjudged and abused when negative results are likely to have adverse ramifications, and that are not able to guarantee what they aim to achieve even if they are correctly evaluated. The AccessForAll framework does no more than identify the objective characteristics of resources.

Correctable accessibility errors

Given the widespread faith in universal design, any resource that is repaired is likely to be done so 'retrospectively'. This is not a well-structured technical term but rather one that has simply become part of the vernacular of those working in accessibility.

In "Evaluation and Enhancement of Web Content Accessibility for Persons with Disabilities", Xiaoming Zeng (2004) considered a number of surveys of accessibility of Web sites, showing that in those studies, the same sort of results were obtained as in the DRC example. He pointed out that in the case of the study by Flowers, Bray and Algozzine (1999), "Their findings indicated that 73% of the universities’ special education homepages had accessibility errors, yet, with minimal revisions, 83% of those errors [were] correctable" (Xiaoming, 2004, p25).

Overwhelmingly, in most of the cases cited by Zeng, the problems were associated with failure to give a text label to images or giving one that was not appropriate. He points, however, to an exception:

The difference between a site that contains images that lack proper description and sites that cannot be used at all is, of course, huge. Zeng's contention is that good reporting on evaluations would make it easy for the site owners to correct the defects (Xiaoming, 2004, p36).

He goes on to work on numerical representations of accessibility, and develops his own, and later argues that with suitable software, often the major flaws in the pages can be corrected 'on the fly' to make the sites accessible in the broad sense, even if the images may lack a description. Zeng's contribution is to provide a way of moving from the accessible/inaccessible dichotomy which, as he argues, can cause a huge site to fail the test of accessibility when only one tag is missing while a smaller site can pass and be quite unusable. He limits the scope of his numerical evaluation to those features of accessibility that can be reliably tested automatically (Xiaoming, 2004, p38).

Zeng argues for a numerical value for accessibility mainly for the convenience and machine properties it would have but he states:

He cites a number of other benefits such a number might have but he fails to convince the reader that a number would provide useful information for making a site more accessible, although if one is to judge a site, perhaps his system would give a fairer evaluation of a site than the existing and generally used checklist provided by WCAG, which is what people usually refer to for the dichotomous evaluation. He calls his metric the 'Web Accessibility Barrier' score (Xiaoming, 2004, p45).

This approach is in line with that taken by the EuroAccessibility group leading to a smaller group's work for a quality mark. Again, the quality mark, however constructed, approach does not seem to contribute to accessibility for the user, or acccess to resources that would be accessible to the user if not to everyone. It might act as a motivation for content developers to be more careful about the accessibility of their resource, but it is also a source of revenue for those few organisations certified to evaluate sites (in some cases those who proposed the certification scheme) and so there has been deep suspicion about it.

EuroAccessibility

In April 2004, the EuroAccessibility Workshop was held in Copenhagen and came up with an annotated draft of the original WCAG that attempted to make WCAG testable (EuroAccessibility, 2004).

Earlier, in Paris in 2003, the following press release was issued by the EuroAccessibility group:

One of the things they did was try to make WCAG testable: the plan from the April 30 2004 meeting was as follows:

There is no evidence the group managed to go much further than to develop the statements before the group was disbanded due to lack of funding. in the current context, it is interesting to note that the group were trying to find ways to insist that within a resource, any necessary alternatives should be identified. This is also considered important in relation to AccessForAll and shares the use of what might be called metadata. In the former case, metadata would be embedded in the resource and in the latter it can be independent of it. The practical difference is that the EuroAccessibility approach would not support third party, distributed or asynchronous annotation as easily as does the AccessForAll approach. Nor would it support the continuous improvement of the resource by the addition of accessible components, which is a major aspect of the AccessForAll approach.

The two approaches differ fundamentally in that the EuroAccessibility approach was intended to make a judgmental statement about the resource whereas the AccessForAll approach strictly avoids that.

Explanation	Example
Take an original WCAG guideline	Guideline 1. Provide equivalent alternatives to auditory and visual content.
and the original WCAG checkpoints	1.1 Provide a text equivalent for every non-text element (e.g., via "alt", "longdesc", or in element content). This includes: images, graphical representations of text (including symbols), image map regions, animations (e.g., animated GIFs), applets and programmatic objects, ascii art, frames, scripts, images used as list bullets, spacers, graphical buttons, sounds (played with or without user interaction), stand-alone audio files, audio tracks of video, and video.
and provide Clarification Points	• A text equivalent (or reference to a text equivalent) must be directly associated with the element being described (via "alt", "longdesc", or from within the content of the element itself). It is not unacceptable for a text equivalent to be provided in any other manner i.e. an image being described from an adjacent paragraph
and testable statements	Statement 1.1.1: All IMG elements must be given an 'alt' attribute. Text: Related Technique:
	Statement 1.1.2: The appropriate value for the text alternative given to each IMG element depends on the use of the image. etc
and provide a list of terms used for a glossary.

There was a division of labour among the various members of the EuroAccessibility group and a small group received funding to pursue their ideas as the CEN/ISSS WS/WAC developing a "CWA on Specifications for a Complete European Web Accessibility Certification Scheme and a Quality Mark" (CEN/ISSS WS/WAC, 2006).

There was, at the time, significant concern with respect to the EuroAccessibility work after the group had split. It seemed to the author, and others, that the motivation for the work was not simply improving the accessibility of the Web, but also the creation of an industry, in citrcumstances when there was doubt about the value of such an industry and fear it might actually stifle better work. Such concerns were notified formally to the CEN process and discussed informally in many contexts (see corres about this with Makx etc ... early Jan 2006...).

Liddy - July 2003 - for AGLS????? note the text is in the code - fix!!!

I have been working on what might be useful in DC metadata (see http://dublincore.org/groups/access/). DC metadata is arguably the most common, international metadata anywhere. In the DCMES, we have an architecture and practices that ensure that DC metadata will be transportable, interoperable, easily developed, langauge independent and internationally deployed. It is also easily and consistently expanded to include locally specific information.

DC metadata is formally expanded, i.e. the number of elements and the qualifications for them, when it is clear that this is going to work well for a large community and not break or damage any existing systems or metadata schemas. The DCMI Usage Board works hard on this and has worked to provide suitable specifications for implementations of DC metadata in HTML, XML and RDF - so it is fairly versatile.

Well, what does it make sense for DC metadata users to do about accessibility?

After a lot of work, I have come to think the answer is fairly simple and should work for a lot of people. I think that metadata about accessibility will be used for compliance purposes and for discovery purposes but also for repair purposes.

With existing DC structures:

Using the dc:format element we can indicate the format of something as it is.
Using dc:relation:conformsTo we can point to standards, specifications or guidelines to which the resource conforms, e.g. WCAG 1.0. For those who want to know not just to what a resource complies but who said so, and when, we can use dc:relation:conformsTo:certified for this information.
Using dc:relation, we can indicate that there is an alternative modal rendering of content by using the dc:relation:isAlternative qualified term to point to the URI of some alternative object or material.
Using the principle that you can have a multiplicity of any DC element, we can also provide (free text) information that complies with a local service's needs, such as that this object will be rendered as a sound object if you use a particular style sheet, or whatever.

Using some of the more 'obvious' DC elements such as dc:audience is not advisable, in my opinion. That element, e.g., is about the content or purpose of the resource, not its accessibility and it is not appropriate or useful, in my understanding, to refer to audiences by their accessibility needs or capacities. My explanatons for this are already on the DCMI site at http://dublincore.org/groups/access/.

All of this information is useful and human readable, as well as machine readable, but it falls short of the very precise and rich information that can be conveyed within an EARL statement. The foregoing information is also fairly easily made available by people.

So my proposal is likely to be for one more recommended DC element - dc:access.

For this element, I am thinking that it should do exactly what most of us want, and not pay attention to what disability a user may have, but be an open element about the accessibility of some resource (and BTW, I am including services etc when I say resource). This element would, if I were Queen, be used to draw attention to the user in the Web world, to encourage Web providers to think about their audience and how they are providing for them, to encourage governments and organisations to think about compliance, etc etc. And the value for this element would be a comprehensive EARL statement, pointed to by a URI.

Following this line, I am working on what should be in that EARL statement so that application developers can make the very most of it.

I am also keen to raise the profile of the user and to that end am arguing that there should be more specific DC metadata profiles for people. We change our access needs according to where we are, what we are doing, etc, and for this, we need a set of profiles, I think. We are also affected by what devices we are using to make the access and what it is that we are trying to do - setting the timer on a microwave oven, getting money from an ATM, reading a newspaper. For this we need profiles of contexts, devices, etc. I believe that if we have more specific metadata profiles, we can mix and match them better. See http://ausweb.scu.edu.au/aw03/papers/nevile/index.html for our first pass at this.

At the Dublin Core conference in Seattle in late September, we will be working on these issues. I urge you to send me comments or come if you can (see http://www.ischool.washington.edu/dc2003/).

Liddy

The value of this correspondence at this time is that it establishes the desire, on the part of the author, to shift from a focus on conformance assertions for universal accessibility to qualitative statements about the resource that can be evaluated by the user.

Brown and Gerrard (2006) argue that broadband makes it easier to make accessible content. This is in line with other expectations for the future; as the technology improves, the opportunities should, and one believes do, improve.