Wayne Dick, Ph.D.
Accessibility to information technology (IT) for individuals with disabilities is equally effective use of the same data model that the target audience uses to achieve the intended results of the system. This definition has three key points:
We understand the architecture of software well enough today to permit the universal use described in the definition of accessibility. This development is new and works as follows:
By encapsulating the basic system objects into three discrete modules: semantic declarations (the Model), presentational specifications (the View) and operational logic (the Controller), a system can achieve elastic access. This enables effective use of enterprise data for a wide range of individuals. The operational control has the flexibility to match many physical interfaces and the presentation of the view can be mapped to many perceptual modes. Universal access regardless of disability can exist in this framework.
Any enterprise can achieve accessibility in a reasonable time frame. The body must follow a careful program of repairing inaccessible applications to baseline standards determined by law and ethics; replacement of obsolete software with well structured systems when replacement is needed; building to best practice when new systems are built.
To achieve universal access for approved users of enterprise data, an IT System must provide four elements of behavior:
To use data it must be perceivable to the user. That means if a perceptual channel is unavailable or difficult for a user, communication must be conducted through a combination of available perceptual channels that are compatible with the user's needs.
A user must be able to operate the controls in order to access enterprise data. That means if a physical input device is unusable for an individual, then the same operational control can be achieved using physical input that meets the needs of the user.
The semantic model of the enterprise must be available to every individual who is qualified to use the model. When the view is made perceivable or control is made operable the user must not lose the meaning of the enterprise model.
To provide universal perception, operation and understanding the system must be robust enough to function on any reasonable platform. The IT should operate using assistive technology, older hardware and software, and modern alternative platforms.
Today a software design architecture exists that supports the requirements of accessibility. A design architecture describes the large components of a system and their interaction. The key to this technology is that each component has well defined behavior and data that are encapsulated within the component. The functions of components are strictly separated, and they only communicate with each other through well defined message protocols determined by the architecture. This encapsulation allows extreme flexibility regarding the internal behavior and data of components. That translates to extremely flexible and modifiable systems. The Model / View / Controller architecture is one such architecture. It is particularly useful for developing platform independent systems that interact directly with people.
The model records the meaning (semantics) of an enterprise. This is usually a database of some kind. The model determines the level of detail, approved access to material, and the conceptual link between the real enterprise and the system representation. The understanding element of accessibility must derive from the model. Any semantic rule asserted by the model must be perceivable and operable using any view or controller of the system.
The view determines how users perceive the model within the context determined by operational control. An independent specification of the view enables many data renderings of the same model. That flexibility serves the wide variety of perceptual needs presented by a typical user population. Thus the view component supports the perceivable output principle of accessibility.
The controller manages access to the model via operations that are determined by the context of use. At any time, there are a small number of responses available to a given state of the system. The controller abstracts this small collection of choices as set of numbers. Many physical devices can collect the user's physical responses and use them to indicate choice. These can be interpreted as numerical responses and passed to the controller's logic. In this way operability can vary with the physical abilities of the user. This guarantees physical operable control needed for accessibility.
The MVC Architecture encapsulates the functionality of components into three discrete objects: Model, View and Controller. These components communicate with each other via messages. So, any one can be extended, modified or replaced without breaking the other modules as long as the message interface is preserved. This enables the robustness required for accessibility.
| Model | View | Controller |
|---|---|---|
| The model encodes the semantics of the enterprise. Fidelity to the model semantics for every
view and control enables universal understanding. |
The view determines how the model is seen, heard or felt in a variety of control contexts.
This supports perceivability. |
The controller abstracts operations in each situation so that many input devices can be used to make the choices. This supports operability. |
| Model | View | Controller |
|---|---|---|
| Proper Web models are expressed in semantically rich markup languages like XML or Strict HTML that are neutral regarding presentation. | The Web view is specified by style languages like CSS or XSL. The style language is separated from the semantic markup. | The Web browser, media player, assistive technology - user agent is the controller. |
An accessible reading system permits completely flexible perceptual modes of reading and an instrument that can be manipulated by the reader. Books and standard print materials are examples of minimally accessible documents. Graphical images of print such as scanned GIF (Graphics Interchange Format) or PDF (Portable Document Format) encoding of GIFs are probably the most inaccessible electronic file formats. Since this form of PDF document was so easy to produce and distribute widely across the web, Adobe was the target of considerable ire from the disability rights community. In recent years Adobe has developed "tagged PDF" that is very usable by many if not most individuals with disabilities.
A well formed markup language (HTML 4 or XHTML 1 Strict) is the best kind of model for storing reading documents. DAISY is a specialized XML dialect for storing digital talking books, both as text and recording. These are rich semantic specification languages and they are neutral regarding presentation. They are perfect for supporting highly effective reading models.
The reading view is expressed in a style sheet such as CSS (Cascading Style Sheet) or XSL (Extensible Stylesheet Language). These can be used to describe appropriate rendering of the document depending on the need of the reader. The view can be Braille, Aural Reading, large and otherwise modified print, cell phone display, and more traditional print documents or screen displays. The controller is a user agent like a browser, media player or assistive technology.
The following examples are intended for sighted individuals. They permit an experiential understanding of the kinds of transformations that are useful to one individual with low vision. The power of flexible view control is that this view can be modified to match the exact needs of other users with print disabilities.
The sample pages below will display well in either Mozilla or Internet Explorer. The plug-in CLiCk Speak only works with Mozilla.
The homepage of the Educational Outreach Working Group of the Web Accessibility Initiative is well written. The Model and View are completely separated. This first view is the normal view. The page has no view specification built into the HTML. All the presentation is controlled by style sheets. It looks good and is easy to use—if you have intact vision. The structure of the page permits an alternative view.
This view of the same page completely rearranges the typography:
All of the visual markers (headings, lists, paragraphs, etc.) are present in the alternative view. They convey the same meaning, but they waste little space doing this. That is very important when enlargement reduces the words per page to 100 or less. The result is a document that can be read by an individual with partial sight at the same level of understanding as a normal reader.
If you would like to experience listening to modified text use the CLiCk Speak add-on to Mozilla. Try it with either the Normal View or Alternative View. The tracking feature of CLiCk Speak allows sighted readers with print disabilities to follow the text that is being read. The combination of Alternative View and reading out loud with tracking highlights is very powerful for many people with partial sight or dyslexia.
As mentioned earlier, PDF has posed many problems for disabled users in the past. In recent years Adobe has developed a new variant called tagged PDF. This data model inserts semantic tags like XML tags into the PDF encoding to permit many perceptual modes for reading.
At present, tagged PDF does not permit the same style control as XHTML. The detailed textual access necessary to permit variable line spacing and leading does not exist. Also, when PDF is enlarged all page elements are enlarged by the same zoom factor. Thus, headings get huge and waste space.
Tagged PDF is exemplary because it is not perfect. It is really less courageous for a corporation to release a perfect product than it is to release a very useful product that is less than perfect. Adobe responded to the needs of many disabled users enough to permit reading when no alternative format is available. That is the action that is needed by organizations. When one cannot fix everything, fix what can be fixed. PDF permits transformation to many other model formats, so the combination of their native adaptation with this transferability, makes tagged PDF a reasonable file format for individuals with print disabilities.
This link to an application form in PDF shows many strengths and some shortcomings of tagged PDF. If one tries to zoom beyond 177% material is lost at the beginning and end of lines. However if the reflow feature from the view menu is set, then greater enlargement is possible. 400% enlargement looks quite good. The headings are too big and the form takes on some odd features, but the document is usable. For fun try the read out loud function from the view menu. It is not as good as CLiCk Speak, but it does the job. The inability to adjust line spacing and leading is missed, but even so, reading is possible. This particular document does not convert well to HTML with alternative style. The form is messy.
With considerable effort, Adobe has taken a format that was usually unreadable and turned it into a model that is fairly usable. When one considers that the original PDF was simply meant to be a platform independent storage of a single fixed view, this improvement is immense. Adobe Acrobat with tagged PDF is a reasonably flexible MVC system for reading documents.
No. Model / View / Controller architecture and its vital role in supporting platform interoperability is new technology. Most user interfaces were not designed with clear separation of these components. This software has accidental built-in barriers. It does not operate well on multiple platforms including assistive technologies.
Yes. Five years ago, PDF was the icon of inaccessibility. Today, tagged PDF is not perfect, but it is good enough for most people with print disabilities. Adobe faced a serious Model / View separation problem and fixed it well enough to make a huge difference. With commitment, the rest of the IT world can do the same. Adobe is now working with Bookshare, a provider of digital talking books, to create seamless conversion from tagged PDF to the DAISY digital talking book standard. They may hit perfection yet.
No. Word processors, spread sheets and presentation software do not support clear separation of model and view. Today most office automation software makes it much easier to produce documents with the model and view enmeshed. The concept that model elements like heading, list and table are distinct from their view properties like font size, line spacing and font style was not prevalent at the time when most of this software of today was developed. As of March 2007, commercial office automation software does not embrace this concept, and the ordinary user must understand far to much about the technology to produce accessible documents.
Yes. Style templates and plug-ins already exist for leading word processors to help with formatting theses and professional publications. Many professional societies provide templates for electronic submission of publications. The same technologies (templates and plug-ins) could be used to bridge the gap between existing office automation products and the next generation of MVC oriented software. Within five years office automation software developers could first build bridge tools for the short term, and redesign their products to a pure MVC architecture to support accessibility and other forms of reading platform interoperability.
Yes. Repair of flawed systems is expensive, and repair to best practice may be impossible. The rule of thumb is this: Repair to law. Build to best practice. There is no way around it, the IT industry has a backlog of deferred maintenance, and it will take time and money to address the problem. The payoff is worth the cost. Each repair to increase accessibility will also increase general interoperability. IT architecture can follow the path taken by construction architecture. Identify critical bottlenecks and remove them. Build a priority list of repairs needed and set aside an annual fund to address barriers. Finally, when new systems are built use a Model / View / Controller architecture and ensure clean encapsulation.
No. Well designed webs and applications are more robust. Modifiability and interoperability are built-in. We have a challenge to pass through, but on the other side we will have functional software that offers access and flexibility for everyone.
Early law focused on reasonable accommodation as the solution for individuals with disabilities in the workplace. While some accommodation through direct service like human readers or deaf translators may be necessary at times, accessible media and IT can eliminate the need for much of this costly accommodation and increase the independence of the person receiving the service. It is much easier to download a talking book from Bookshare than it is to arrange a recording or set up appointments with human readers. It also feels better from the self respect perspective.
Most organizations can afford some reasonable level of accommodation. The problem is that in an inaccessible environment, accommodation resources can be overwhelmed quickly. A high level of IT accessibility, say above 75% of products being compliant, will reduce the need for individual accommodation to a manageable level. That accomplished, complete access to the workplace will be possible for everyone.
There is no checklist or rule book to ensure complete accessibility. Standards give lists of known barriers and suggested solutions for removing them. As time progresses and standards get more complete and accurate, we will get closer to guaranteed tests for accessibility. For now, the standards certainly take us to the point where the only accommodations we need are for truly unique circumstances. That is good enough.
In 1954 (Brown vs. Board of Education) the US Supreme Court struck down the principle of "Separate but Equal" that had been used to justify segregated education in the United States. Today the notion that separate cannot be equal is generally accepted, but with information technology achieving universal access to data models has not been possible until recently.
In the world of reading documents, books had to be copied to analog talking books. Braille translation always followed normal printings. The result was a separate and inferior library of written information. Inaccessible media on the web or in other electronic formats often causes the need for alternative copies of media such as text only versions or otherwise modified documents. Again this process creates a separated maintenance process, and inferior access for users of the alternative media.
The Model / View / Controller architecture enables everyone to use the same semantic model and consume it using the view and controller that meets their needs. Two libraries are no longer necessary. The example of an accessible Web that uses standard model and view languages to elicit different behavior from the same controller shows how this architecture can produce efficient solutions to the problem. Segregation of people with disabilities into one group and the rest of society into another is no longer required in the world of information. With modern IT we can all share access to the same library of models with equal effectiveness.
The idea of using Model / View / Controller as the base architecture comes from the article Accessibility requirements for systems design to accommodate users with vision impairments by P. Brunet, B. A. Feigenbaum, K. Harris, C. Laws, R. Schwerdtfeger, and L. Weiss that appears in the IBM Systems Journal, November 3, 2005. The POUR Principles of Accessibility (Perceivable, Operable, Understandable and Robust) were developed by the Web Accessibility Initiative of W3C and is articulated best in the Web Content Accessibility Guidelines 2.0 (Final Draft).
The overall scope and context of this article has developed over the past year with my colleagues on the California State University System Accessible Technology Initiative. I must give special thanks to Tom Jewett, Mary Cheng and Mark Turner for creating the conversation that helped me understand the exact relationship between MVC and POUR and how to administer its implementation in a practical workplace, the CSU System.
Copyright © 2007, by Wayne Dick, wed@csulb.edu. Links to this site are welcome and encouraged. Individual copies may be printed for non-commercial classroom or personal use; however, this material may not be reposted to other web sites or newsgroups, or included in any printed or electronic publication, whether modified or not, without specific permission from the author.