Making IT Accessible

Accessibility: Making IT accessible

M Sasikumar, CDAC Mumbai,
ni.iabmumcadc|isas#ni.iabmumcadc|isas

Background

Digital divide refers to a social categorisation of people as those who are able to benefit from information technology and those who are unable to. It is an increasingly serious concern for many countries today, given the advent of a knowledge society where access to right information at the right time is a must for good quality of living. This access to information is critically dependent today on information and communication technology. Internet, for example, is by far the most common source one turns to for any kind of information. Applications running on the world wide web, caters to most aspects of life ranging from education and health to entertainment and shopping. Digital divide deprives a large section of our population from being able to benefit from these developments.

Digital divide is caused by a number of factors. The affordability aspect, primarily concerned with cost of an adequately powerful computing device is one major concern. Attempts to develop low cost PC, specialised devices such as PDAs and Simputer, etc address this aspect. The open source software movement contributes significantly to make available high quality software at affordable cost. The second and equally significant factor contributing to digital divide is accessibility. Accessibility – we use this term in a fairly broad sense in this document – has a number of dimensions:

  • linguistic accessibility: Use of most computing devices, from computers to cell-phones, still require knowledge of English. Nearly 95% of Indian population are not comfortable with English, and are thus deprived from any significant use of these technologies.
  • communication accessibility: More than knowing the language of interface, the user need to be able to read/write language. This is not the case for a large section of the population. The difficulty may be a literacy problem, a disability in vision/speaking/writing, etc.
  • user-interface accessibility: choice of poor metaphors for the concerned target audience. For example, the notion of a desktop may be difficult to comprehend for someone who has never seen a desktop. We need to identify different metaphors which may be easier for them to comprehend reducing the learning curve in use of these devices.

As information technology and associated devices are moving from the techno-savvy to the average citizen, these concerns will dominate usability and success of an IT solution. In this paper, we will be concerned with the second item — communication accessibility — primarily. Linguistic component is being addressed through other initiatives such as localisation, Indian language solutions for tasks like automatic translation, etc. However, these three are not totally independent aspects and hence some of our concerns will cover the other aspects also.

In the next section, we outline the relevance of this field for India, though the basic need should be visible from the references above to digital divide. We then look at characterising the various aspects of accessibility, and look at the scope of work possible in each. We end this report by looking where we are in terms of the various initiatives that exist, specifically some of the work we are doing at CDAC Mumbai.

Need and Relevance

As mentioned in the previous section, as ICT makes its impact on every walk of life from entertainment to education, there is a large section of society being left out of the picture. These are people with various constraints which disables them from using ICT. These include:

  1. Vision problems - blindness, low vision, colour blindness, blurred vision, etc.
  2. Hearing problems - Though, not a constraint for much of ICT today, as we move more applications to speech interface, this will also be a concern.
  3. Motor problems - Those suffering from Parkinson’s disease, musculine dystrophy, cerebral palsy, stroke, etc may have significant motor/mobility problems such as unable to type, unable to control fingers to adequately fine level for use of keyboard, too slow to type, etc. Similar concerns apply to use of mouse as well. Physical handicaps can also result in such difficulties.
  4. Speech problems - as in hearing, speech provides an alternative to typing in many cases. But if you have speech problems, then this is difficult.
  5. Cognitive problems - developmental disabilities, learning disabilities (dyslexia, etc), etc.

Categories of people such as handicapped, illiterate, senior citizens, etc all suffer from one or more of these hindrances and hence are deprived from the benefits of using ICT without the help of a third person. On an average, across the world, the fraction of population above 60 years of age is expected to double in the next few decades. Countries such as Italy and Japan are leading the way with over 40% of the population expected to be aged above 60 years by 2050. At that age, various disabilities are common and hence this segment is an important focus group for accessibility.

Accessibility is a problem, not restricted to one community or country - but truly an international concern. W3C has a set of guidelines for web accessibility under its web accessibility initiative (WAI) (http://www.w3.org/WAI/). Some countries are making legal options to ensure some degree of accessibility on applications; not providing adequate support becomes an offense. There are other initiatives such as User Agent Accessibility
Guidelines (UAAG) and Web Content Accessibility Guidelines (WCAG) from the World Wide Web consortium. While these are good first steps, there is a much larger range of problems to be addressed, as is clear from the list of difficulties given above. A number of major players in the information technology product and services space have significant initiatives in the area of accessibility. Microsoft, Adobe, Apple, Intel, etc all have sections devoted to accessibility on their website outlining initiatives to develop technologies to support accessibility and special provisions in their products with the same intention.

The work in this area needs to span a number of dimensions from awareness building to evolving effective frameworks for implementation. Many of the accessibility issues may not be of interest to commercial outfits, given the low market returns. And hence wide spread of awareness, community involvement, effective and easy mechanisms to make applications accessible in different ways, etc are all matters to be addressed.

Aspects of Accessibility

As is clear from the above descriptions, accessibility involves a number of aspects ranging in complexity from trivial to highly challenging. Almost every aspect of interacting with a computer system – vision, use of hands, etc – has accessibility implications, since there are factors that affect that aspect and renders effective interaction difficult. In this section, we outline some of the outstanding issues. We will restrict our focus to the interface aspects, and will not discuss hardware issues such as improved power consumption, ability to withstand temperature variations, etc which also play a role in ensuring accessibility. We will discuss the various aspects in no particular order; the list is also not necessarily exhaustive.

Mode of communicating to a computer

As of today, keyboard and mouse dominates the modes of communicating to a computer. Keyboard and mouse are dependent on no disabilities involving hand, good hand-eye coordination, no significant vision problems, and so on. For those who do not meet one or more of these requirements, the keyboard/mouse of interaction is not appropriate. Touch screens are popular in kiosks, but not yet on common desktops. Touch screens to enter data such as text is very cumbersome and also uses a lot of screen space.

Keyboard usage has a high learning curve, for new users. The key layout in the Qwerty format continues for historic reasons, and can be a hindrance to mass usage of computers who have no typing practice. On screen keyboards provide some relief, since the layouts are not necessarily fixed.

Online handwritten character recognition is one approach to alleviate this problem. However, as of now, for Indian languages, handwriting recognition is in its early stages. The large number of characters (compared to English) and the complexity of character formation particularly for the compound letters makes this a challenging task.

Alternative mechanisms for interacting with a computer is a challenging direction of work. Use of head or retina tracking as an alternative to moving the mouse pointer with hand is a promising area. Effective devices built around this idea are yet to be commercially available. There is also the possibility of using the feet for mouse movement. None of these are again effective substitute for keyboard, which is required to enter text, for example, while creating documents, letters, etc.

Speech is potentially a very effective mode of interaction - as exemplified by its dominance in human-human communication. However, for use of speech, we need effective speech recognition capabilities on the computer. Speech recognition is still a research topic. Good systems are available commercially in restricted domains. These are normally speaker dependent, limited vocabulary, special ambience (low noise) systems. Relax one or more of these, the system’s performance degrades substantially. The situation is worse with Indian languages - with very little work even for restricted scenarios as above. From exploration of effective algorithms for the various aspects of speech recognition, to creation of rich enough corpora, there is a large range of work to be done.

Medium of Interaction

Currently, the language of human-computer interaction is almost completely English. A few Indian languages are beginning to make a mark in select software solutions such as office suite, browser, etc. Given the low fraction of Indian population (about 5 percentage) that is comfortable with English, this language divide prevents them from being able to use computers, and other ICT devices such as PDAs and cellphones.

Software localisation needs to happen at a large scale to address this problem. We are assuming that fonts, input methods, and rendering solutions for all languages are available. While this is certainly not true in general, one does not expect major technological challenges here - it is more a matter of community effort, awareness etc. Therefore, the major challenge of accessibility with respect to the medium of interaction lies with localisation. Much of the localisation process, in turn, is well understood and tools and frameworks are available to support the process.

However, localisation still offers many challenges. Given the large number of softwares with each having multiple versions over a period of time, and the large number of languages, automation has to play an important role for localisation to be effective in addressing the digital divide. The tremendous progress in the area of machine translation is being exploited, though partially, through ideas such as translation memory. These
systems still require active human involvement. Automated machine translation is still immature for many languages, more so, when moving from a language like English to Indian languages. Above all, the requirements from translation for localisation do differ from other domains. For example, consistent use of terminology across a full application and change of terminology across applications.

Another alternative is to disguise the actual language of interaction from the user. An effective speech-based interface, for example, can accept the commands from the user in Hindi, and convey to the application in English. This is the base for accessibility frameworks such as Gnome Accessibility framework. This kind of non-invasive approach to interface design is a tremendous advance in providing accessibility solutions. This is a scalable and source code independent approach to offer an interface convenient to the user, which may be quite distinct from what the software actually uses. This idea can be further extended to support alternate input devices for disabled people as well as supporting gesture recognition, eye-tracking, etc. While partial work exists for some of these, a number of open research issues are present requiring further work for building effective scalable solutions.

Receiving information from a computer

The most common mode of receiving information from a machine is visual - either on the screen or in print. This generally requires normal vision including the ability to perceive colour and font variations. This also requires the user to be literate in the language of interface to be able to read the text. There are circumstances requiring relaxation of all these assumptions - the visually impaired, blind, colour blind, illiterate, etc. Font magnifiers are today available on most browsers for the visually impaired, but not fully blind. Similarly, one can check with tools available on the net, how your site would look for a user who is colour blind in some way or other.

Using speech instead of printed/displayed text is one approach to making the output processable to the user. There are some solutions currently available such as screen readers - to convert the text on screen to speech. There are a number of challenges here.

  • Current screen readers are too crude. This is partly due to the kind of speech synthesis system used. Most TTS engines (text to speech) are unable to express prosody - emotions expressed in what is being read. This requires work on the speech synthesis component and also the ability to guess the relevant emotion from the text being read.
  • Another aspect of difficulty with screen readers is the challenges in converting a relatively stationary two-dimensional display into a transient single-dimensional audio stream. Navigation among the various items on a page is a major concern in this transformation. Current solutions are quite primitive.
  • Colour blindness is a task currently to be addressed by using style sheets for HTML pages. For outputs produced by other programs, even this is not applicable.
  • Processing multimedia outputs such as video and audio is another challenging aspect. How about a display device which produces Braille scripts as output, which a blind can feel and understand?

Alternate models of computing devices

As cellphones and PDAs are approaching the capabilities of a minimal desktop, the user interface challenges would need to cover these devices as well. The restricted screen area and device size is a major constraint even for normal users. Therefore, the challenge of making them usable by the disadvantaged is more complex. This may lead to further innovations in the HCI aspects of these devices.

The growth of ubiquitous and pervasive computing and ambient intelligence are also likely to further increase the need for dealing with devices and information technology on a day to day basis. Our discussion has so far been assuming the interaction with a normal desktop or laptop kind of computers, whereas these developments introduce other types of computing devices such as mobile phones, embedded devices, etc. Accessibility concerns for these will have additional dimensions of concern.

Accessibility Today

As mentioned earlier, many countries have legal provisions to ensure accessibility to some degree. Some are amending their disabilities discrimination laws to extend its reach to software products and services, particularly those over the web. A number of companies are active in this space too. IBM has a large initiative on accessibility. There is a Human Ability and Accessibility Centre as part of the India Research Lab of IBM in India. IBM is focusing on development of products specifically aimed at various types of accessibility such as visual impairment, low vision, etc. aDesigner is a product to simulate disabilities to help software developers in supporting various accessibility issues in their product. Voiceweb is another initiative visualising a web over voice, instead of vision, which can be used by visually impaired people.

A number of guidelines already exist to support accessibility - though at a very primitive level. A fair amount of attention has been so far on the visually impaired. For example:

  1. To indicate hyperlinks on a page, use underline AND colour (for the colourblind persons)
  2. Use meaningful names and alternate text for links and images so that the screen readers can convey some sense to the listener when these are encountered.
  3. For links and other clickable entities, provide a large area for those who are unable to control mouse precisely.
  4. Reduce flash effects or provide to switch it off, for those prone to seizures.

Similarly, a number of software and hardware tools are available (most of them to be purchased) to help improve accessibility. These include

  • Screen readers which can read out text on screen. Though far from perfect (as discussed earlier), they are a tremendous help at present.
  • braille terminals which ’display’ pages through raising pegs on a flat surface.
  • screen magnifiers available on most operating systems.
  • speech recognition systems (though limited in capability).
  • keyboard overlays for those with motor difficulties.
  • sip-and-puff switches which can be operated by mouth.
  • Mouse smoothing software to control hand-tremor.

Mostly these are offering superficial solutions without changing the basic mechanism of interaction in any serious way. In the long run, addressing the disabilities would require a lot more deeper attention and new ways of interacting with computing devices.

CDAC Mumbai has been involved in a number of activities relevant to accessibility in different ways.

  • Open source software - particularly relevant for development of alternative interfaces, software localisation, etc. There is a full-fledged division working in this area with expertise in a number of open source development environments.
  • Software localisation. The activities include localisation of Moodle learning management system in Hindi, compilation of a how-to guide for localisation, and project on incremental localisation to support quicker localisation of new versions of an existing localised software.
  • Accessibility framework. We have attempted to build speech interfaces to e-mail client Pine and browser Mozilla, using available open source speech recognition engines. The Mozilla work was done using the Gnome accessibility framework.
  • Predictive writing. This is important to enhance the typing capabilities of those having mobility problems (for example, due to old age). A working prototype is available for Hindi. It predicts text at a word level, as opposed to predictive writing supported by common word processors at a character level.
  • Natural language processing. As part of the AI group, there is fair amount of expertise on various aspects of natural language processing. NLP techniques will be important in addressing many of the aspects relating to language of interaction including automation of translation in localisation.
  • Handwriting Recognition - A framework for online handwriting recognition for Indian languages is being explored as a PhD research work, using Kannada as the base.

Conclusion

Accessibility is becoming a major concern since it is preventing a large section of the society from benefitting from ICT. It has a number of dimensions including handwriting recognition, speech recognition and synthesis, alternate modes of input and output, etc. Some early systems and results are available in most of these areas. Most of these areas are active topics of research while addressing a completely practical concern. Accessibility ought to be a subject in all information technology curriculum, and we require more researchers to consider this area for their research work. It is technologically challenging and at the same time, a highly socially relevant concern.

Unless otherwise stated, the content of this page is licensed under Creative Commons Attribution-ShareAlike 3.0 License