Web Posted on: August 4, 1998

An observation method of Internet electronic documents accessibility

Bernard Oriola
Nadine Vigouroux
Philippe Truillet
Franck Lacan
oriola, vigourou, truillet, lacan{@irit.fr}
IRIT, UMR CNRS 5505
118, Route de Narbonne
F-31062 Toulouse Cedex

Summary

The goal of this paper is to report on the "usability" of browsers by visually impaired persons (VIPs). Firstly, accessibility problems concerning the use of Internet by VIPs will be addressed. Secondly some considerations about the evaluation methods of interactive systems will be discussed. Then the evaluation goals and the experimental protocol applied to evaluate three browsers will be described. The protocol as well as the analysis results in terms of satisfaction and efficiency degrees will be also commented.

1 Introduction

The World Wide Web represents one of the challenges in the field of electronic services. It allows:

• information exchange by message services such as electronic mail between several users, and discussion lists about various topics (newsgroups),
• access and navigation through electronic Hyper Text Markup Language document available on an infinity of sites through the World-Wide-Web.

For the visually disabled persons, it is an extraordinary integration tool in terms of education, information access and teleservices, etc. But the human-machine interfaces for accessing the WWW are mainly based on direct interaction: this means highly visual presentation and mouse-based interaction. The aim of this paper is to demonstrate for VIPs, both the possibilities and the difficulties to access the WWW with existing tools on the market. The studies undertaken will report on two important points concerning the non-visual interaction:

• the efficiency of notification about visual items (texts, frames, graphics, motion pictures, etc.),
• the efficiency of the navigation strategy. These two points will determine the degree of the WWW usability by VIPs.

| Top |

2. Internet accessibility problems for VIPs

Like already above mentioned, human-machine interfaces of telematics tools as well as of programming software and so on, are becoming more and more based on visual presentation and this prejudices the non-visual communication. More this fact, the visual elements of HTML documents can also raise some troubles or miss-understanding during their reading by VIPs. Among of these we can notice as crucial problems of accessibility:

• the frames which share the screen in several independent windows; therefore screen readers read line by line crossing eventually these frames;
• the image maps containing interactive areas which allow to activate an hyperlink by means of the mouse pointer; the problem is how the VIPs have notification of this area ?
• applets; these are little programs whose execution can produce images, sounds and graphics. The VIPs don't know often when and where there is an applet running;
• etc.

| Top |

3. Technical aids to access WWW by VIPs

There are different technical solutions to allow access to the WWW. Some solutions consist in using browsers specially designed for the VIPs which do not need to be coupled to a screen reader. Among these approaches, we can mention pwWebspeak [1] (Productivity Works), Smart-Net [2] or DraculaNet (euroBRAILLE) [3], etc. Another solution is to use a screen reader coupled with a standard Web browser. There are the textual browser Lynx [4] and some graphical browsers like Internet Explorer (Microsoft Corporation), Netscape Navigation (Netscape Corporation), etc. In addition, the HTML documents can be filtered by means of a proxy ; it is the case of WAB [5] which rewrites the document in a form supposed being more accessible for the VIPs. Independently of the approach used, these technical methods need to:

• allow the review of visual screen contents (i.e. visual objets in the HTML document),
• implement non visual interaction methods for each interaction object (for example, applets, image maps, etc.),
• assist the blind in navigating through the structure of the HTML document.

These solutions may be designed only on speech output, or on a braille output, or integrating the alternative or combined use of braille, speech and sounds [2]. The question that we will try to answer is thus: What is accessible during a browsing task through the WWW or a HTML document? Which man-machine interface elements are (or could) be used to improve the non-visual interaction on the WWW?

| Top |

4. The observation goals

Firstly, we will place our observation method. What we want to know is the World Wide Web usability by VIPs. As mentioned by [6] the usability is "the effectiveness, efficiency and satisfaction with which specified users can achieve specified goals in a particular environment". [7] defines this point of view as "the ease of use and acceptability of a system or product for a particular class of users carrying out specific tasks in a specific environment; where ‘ease of use’ affects user performance and satisfaction, and ‘acceptability’ affects whether or not the product is used".

This precision made, the attributes considered for the evaluation steps will be described. This list is based on the Nielsen’ proposition [8]: the ease of learn and of use, the efficiency of use,

the ease of memorisation, the ease without interaction errors and the satisfaction.

The "ease of use" determines if a software can be used or not by VIPs and how it can. It can be determined by both the user satisfaction degree and his behaviour and by the user interface. These criteria must be analysed in several interaction contexts —several user classes, tasks, etc.— The "efficiency of use" represents the degree of use by an user. This means that the interface can allow to realise some defined tasks; in our case, for instance, to reach a site.

| Top |

5. The evaluation protocol

Evaluation should form an integral part of the design process of any new technical aid for disabled persons. In this framework, we intend to study man-machine interaction acts for blind users during several Internet accessing tasks. A set of exercises to realise with different browsers was defined. Three browsers were selected according their design and working features:

• Lynx v. 2.7, a textual browser running with the screen reader Jaws (Henter-Joyce),
• pwWebspeak v. 1.2, a vocal browser
• and the Internet Explorer (IE) V 3.01 running also with the screen reader Jaws.

Blind users have to carry out exercises and to answer to a questionnaire. The interview schedule used was elaborated in an iterative process, with several draft versions of exercises which were driven on a number of blind interviewees. Users were divided in two categories:

• inexperienced users: this class of users knows Internet principles and the use of Windows 95 but is inexperienced to use the tested browsers,
• expert users: these users know the principles of Windows 95, Internet tools and the browsers tested.

Research studies in the predictive evaluation field of interactive systems have shown that users must have a basic knowledge —principles and rules of interaction, what is possible to do with the interactive system, which notifications are available?, etc.— of the analysed browser. So a short presentation of the browsers tested will be made to inexperienced users. The same goes for the screen reader if the user does not use it usually.

After a preliminary questionnaire concerning personal information and general use of the telematic tools, the users have to realise seven exercises with each browser. Each exercise has a growing difficult level: loading a given HTML document, navigating through HTML documents including only textual items, using a search engine, consulting HTML documents including frames, maps and accessing a form (read and fill up).

To avoid tiredness and lassitude effects, several structurally equivalent HTML documents (e.g. documents with the same kind of visual items —frames, image maps, applets, forms— but concerning a different subject) were designed. Textual descriptions can be joined to images.

| Top |

6. The observation sessions

The observation has taken place from June to November in 1997 with seven blind users in the IRIT laboratory. Three engineers in computer science (experts) and four beginners (inexperienced users) participate in these. Each user had to undergo through an half day to perform the set of exercises with each browser. The screen reader Jaws is coupled to the braille display ABT3-series from Alva BV and to the TTS ProVerbe Speech Engine of Elan Informatique Company.

| Top |

7. First results

For the two class of users —expert and inexperienced users— the three browsers tested can be used to load a HTML document, to run a search engine, to navigate through the HTML document: the value of the attribute "ease of use" is evaluated at "high level possibility" for these three browsers analysed independently of the user class. The degree of satisfaction is more important with the textual browser than with the graphic browser IE for the inexperienced user for these three activities. These results allow us to formulate the following hypothesis: it seems that the users who have experience in accessing graphical user interfaces increase their performance and their satisfaction degree in accessing the WWW with graphical browser. Another important fact is the low satisfaction degree of the pwWebSpeak use for both user classes for the frames and image maps. The users observed find the pwWebSpeak not well appropriate: a potential explanation could be that this browser is based only on an auditory mode (we must notice that the version 1.2 of pwWebspeak could not use frame, map or sound).

Table 1 – Some results about the observation sessions.

The table 1 illustrates the efficiency notification for the images and the frames according the browser type. For instance, the notification is particularly important when the loading time is high.

It also shows that the VIPs can navigate through hyperlinks (links by links) and frames of a given HTML document.

| Top |

Acknowledgements

The authors are grateful to VIPs who participate in these observations.

| Top |

8. Conclusion

These observation sessions show that the WWW is accessible even if some new visual items in HTML documents raise problems. This study points out that experimented users and half of the beginners prefer to use a standard Web browser coupled to a screen reader rather than as dedicated tool. It seems also that efforts must be carried on notification and interaction strategies as well for VIPs than sighted persons. It would be utopian to hope a WWW without its presentation elements, nevertheless we can wish that Webmasters make an effort to associate a textual description to an image and to respect the accessibility recommendations of HTML 4.0.

| Top |

9. References

[1] pwWebspeak, http://www.prodworks.com/pwwebspk.htm

[2] Truillet Ph., Oriola B. & Vigouroux N., Multimodal Presentation as Solution To Access A Structured Document, in 6th World-Wide-Web Conference (Santa-Clara, 07-11 April 1997, USA).

[3] DraculaNet, euroBRAILLE, http://www.ccr.jussieu.fr/braillenet/eurob.html

[4] Lynx, http://lynx.browser.org

[5] WAB: W3-Access for Blind, http://inf.ethz.ch/publications/ea.html

[6] Brooke J., Bevan, N., Brigham, F., Harker, S. & Youmans, D, Usability and Standardisation – Work in Progress in ISO, in D. Diaper, D. Gilmore, G. Cockton, & B. Shackel (Eds.), Proceedings of IFIP TC 13 Third International Conference on Human-Computer Interaction, INTERACT’90, pp. 357-361, Cambridge, UK, Elsevier Science.

[7] Bevan N., Kirakowski J., Maissel J., What is usability? In H-J. Bullinger (ed.) Human Aspects in Computing: Design and Use of Interactive Systems and Work with Terminals (pp. 651-655), Amsterdam, The Netherlands. Elsevier Science Publishers.

[8] Nielsen, J., Usability engineering, San Diego, CA, Academic Press.

| Top | |TIDE 98 Papers |