Web Browsing through Adaptive Technology: A Consumer Information Resource (April 1996)Kevin Nguyen and Linda Petty, University of Toronto
Kevin K. Nguyen and Linda S. Petty
Adaptive Technology Resource Centre (ATRC), University of Toronto, Ontario, Canada
The Adaptive Technology Resource Centre (ATRC), University of Toronto coordinated efforts with the Diversity Management Directorate (DMD), Public Services Commission of Canada to systematically evaluate how commercial World Wide Web (WWW) browsers functioned with various types of adaptive technology under different operating systems. The types of adaptive technology examined include screen magnifiers, scanning / switch access systems, alternative keyboards, screen readers, Morse code input devices and voice recognition systems. The objectives of the project were to: document the strengths and weaknesses of controlling different browsers with a representative set of adaptive technologies and operating systems; note customizations of the assistive devices or browsers that were needed for effective control; incorporate feedback from manufacturers of the adaptive technology products into the final document; and summarize the research findings into a set of bilingual, on-line guidelines targeted at users with disabilities. Important information was obtained that provided clues for what needs to be done on both the browser and adaptive technology sides to make web browsing accessible.
The popularity and pervasiveness of the World Wide Web has increased exponentially over the last few years; however, the mouse-based point-and-click interface that makes the Web appealing as a method of serving information also creates barriers for people with disabilities. There are several commercial assistive technology devices currently available that allow people with disabilities to access their computer systems (11,12). These include voice recognition systems, screen readers, screen magnifiers, Morse code input devices and alternative keyboards. However, how well do these assistive computer devices, often an add-on to most computer systems, work with standard interfaces such as Internet Web browsers? Information on how access technology and browsers work together is relatively uncommon. Users with disabilities have often relied on their own trial and error methods to determine how their adaptive technology devices can access the Internet. A joint venture formed between the Diversity Management Directorate (DMD) of the Public Service Commission and the Adaptive Technology Resource Centre (ATRC) of the Information Commons, University of Toronto to help eliminate the lack of information on how various adaptive technologies and Web browsers work together.
Objectives of the Project
- To document the strengths and weaknesses in controlling browser functions of a full range of access technologies.
- To document any customization required for effective use of the adaptive technology products with the identified web browsers.
- To inform the manufacturers of each adaptive technology product tested of the results of the research on their product and to incorporate their feedback into the final "How To" guidelines.
- To address the direction needed to overcome future access barriers.
- To prepare a final report explaining the tools and methodologies were used, the results found, and the customization which will be of most help.
- To summarize the research results in a set of "How To" guidelines targeted at users with disabilities to be converted into accessible HyperText Markup Language (HTML) format for posting on the DMD and ATRC web sites.
The Diversity Management Directorate provided funding to develop guidelines and practical tools to help persons with disabilities access the WWW using identified adaptive technologies. The project began March 1, 1996 and the majority of testing and report writing was completed on May 17, 1996. Further support was provided by the DMD to translate the information into French and to convert all information into accessible HTML format for publication on the ATRC and the DMD web sites. A simultaneous public launch of the web project took place on January 20, 1997.
Adaptive technology products were selected for the project based on availability and consumer interest. In particular, screen reading software has received considerable interest and controversy over its ability to access web browser features, so leading products in this group were targeted for testing. A representative sampling of technology for alternative keyboard and mouse control was also tested, as very little attention has been paid to this area in the literature or in adaptive technology conferences. Browsers were selected based on market share and popularity (1st Quarter 1996) and reviewed mainly in the Windows 95/Windows 3.1 and Macintosh OS environments. Netscape Navigator was the browser of choice for most testing, with some products also tested on Microsoft's Internet Explorer as the strongest contender. IBM's Web Explorer for OS/2Warp and the commonly used text-based web browser, Lynx, were reviewed in the DOS and UNIX platforms, to accommodate individuals using these operating systems. Testing included accessing the features of each browser's menus, buttons, status bar, etc.-- and navigation of web pages, hypertext links, forms, frames, etc. Each combination of browser and assistive technology was tested with the known content of the ATRC web site and with a mixture of other web sites, including Netscape's home page and form-based search engine pages such as Infoseek and Lycos.
Before testing the adaptive technology it was necessary to carefully review each browser and list its features and requirements for use. Comprehensive documents for each browser version were produced describing features and providing other important information such as keyboard equivalents for menu items, etc. It is anticipated that the browser information will be helpful to users who want to learn about available browsers prior to selecting or using one. The documentation is also useful for learning about available configurations and customizations for each product. Knowledge of browser keystroke equivalents can be invaluable for automating and optimizing commands for users of adaptive technology. Also, many browsers only have help files available on-line; thus, the collected browser information can be a useful off-line reference tool.
Five ATRC student-staff were involved in the testing and writing process. The students were finishing the third or fourth year of their university studies and were familiar with both assistive technology and browsing the World Wide Web. Three of the students have disabilities and so were able to test the products from the perspective of consumers. Four other people with disabilities were paid an honorarium to test the browser/assistive technology combinations and provide us with their feedback. These four individuals used a screen reader, Morse code, single switch scanning and alternative keyboard input respectively to access computer systems. One user tested Lynx at home using his own screen reader; the remaining users came to the ATRC for testing. The ATRC assistant manager reviewed several products and assisted in the user testing to ensure optimal access setups. The assistant manager also solicited the feedback from the software developers/product manufacturers and arranged for equipment purchases or loans for the project.
Browsers and Assistive Technology Products Reviewed:
World Wide Web Browser Software
Documents identifying functions and features, keyboard equivalents, and other useful access information were collected for the following browsers:
- Netscape 2.0 for the Macintosh
- Netscape 2.0 for Windows 95
- Netscape 2.0 for Windows 3.1
- Internet Explorer for Windows 95
- Internet Explorer for Windows 3.1
- NCSA Mosaic for Windows 3.1
- NCSA Mosaic for Windows 95
- Lynx 2.6 for Unix
- IBM Web Explorer for OS/2 Warp
Assistive Technology Products
The following products were reviewed, as listed, in conjunction with web browser and operating system:
Screen Enlargement programs
- ZoomText for Windows 3.1 with Netscape 1.1
- ZoomText for Windows 95 with Netscape 2.0
- InLarge for the Macintosh with Netscape 2.0
Scanning access methods
- Ke:nx 3.0 for the Macintosh and Netscape 2.0
- WiViK 2.2 for Windows 3.1 and 95 and Netscape 2.0
Alternative keyboards with mouse emulation
- Glide point ALPs keyboard for the PC
- Intellikeys expanded keyboard for Mac and PC
- TASH Mini keyboard for the Mac
- ASAP and Lynx in the UNIX/DOS environment
- IBM ScreenReader/2 with IBM Web Explorer
- Outspoken with Netscape and Internet Explorer
- Jaws for Windows with Netscape 2.0
- Screen Power for Windows with Netscape 2.0
- Dragon Dictate 2.0 with Internet Explorer and Netscape in Windows '95
- Dragon Dictate 1.1 with Netscape and Internet Explorer in Windows 3.1
- Kurzweil LawTalk for Windows 3.1 with Netscape 2.0
- Power Secretary for the Macintosh with Netscape 2.0
Morse Code access
- Mini-Morse with Macintosh with Netscape 2.0
- Mini-Morse with Windows 3.1 and Netscape 2.0
The results were compiled into an integrated set of public guidelines posted on the Adaptive Technology Resource Centre and the Diversity Management Directorate Internet sites in both French and English. The Public Services Commission of Canada provided French translation and both versions of the documents were converted from various word processor format into HTML 2.0. The complete set of guidelines was entitled "Universal Internet Access Project" and "Projet d'accès universel à l'Internet" and is linked to from the following URLs: http://idrc.ocad.ca and http://www.psc-cfp.gc.ca/. Several features of the site were added during the process of HTML conversion to increase the overall experience and resource value. All pages were rendered accessible for the users of screen readers (4,6,9). ALT-text was provided for all images and all graphical links had text equivalents. Also, each page contained a set of standardized "page navigators" that allowed easy movement through the resource. Contact information for all products reviewed was provided as well as direct hypertext links to company web pages and electronic mail addresses (2). In addition, a glossary of adaptive technology and Web browser terminology was integrated into the site (8). This is an invaluable resource for anyone new to adaptive technology or/and the Internet. The guidelines also invited electronic mail, fax, and telephone feedback from readers about their own findings or suggestions for future product reviews (1).
Summary of Results/Findings
The research yielded some surprising results and some important details on known problems and issues. The full 120 page listing of results can be linked to from http://idrc.ocad.caor http://www.psc-cfp.gc.ca/.The following general conclusions were drawn from the collected, in-depth individual reviews of assistive technology and browsers:
- Although Netscape Navigator was the acknowledged leader among web browsers at the time, Internet Explorer 2.0 offered some useful access features, such as commands to universally increase or decrease the font size of the text on the web page. Internet Explorer appeared to have been created primarily as a Windows 95 tool and was less intuitive to use in its Windows 3.1 version.
- Screen enlargement packages generally worked well with web browsers in the Macintosh, Windows 3.1 and 95 environments, although color palette conflicts existed in some Windows video modes. For proficient use of screen enlarger with browser, the user required that the browser window be resized to the viewable area. Otherwise, following text in the browser window became confusing.
- Four of the screen reading packages for graphical user interfaces were tested, with very mixed results. The best function found during the testing period required macro files to configure the screen reader for the browser, which limited the user's choice of software. No screen reader tested was able to clearly present information found on a web page containing groupings of graphics, text and information in columns. If text information was broken by columns, frames and graphics, then it was very difficult for the user to gain a comprehensive and cohesive understanding of the page contents in a short time. Frames and columns or mixed text and graphics are challenges still to be fully conquered by many screen readers (5,9,13).
- Voice recognition software works best when using built-in macros for accessing standard menu features and these vary in availability with off the shelf packages. Mouse control of web browser features is somewhat difficult and much more time consuming using voice recognition than the use of equivalent spoken menu commands.
- Alternative keyboards with built in mouse emulation worked adequately for browser access, as was expected. The ability to move quickly from one hypertext link to the next would have benefited testers with motor skill deficits.
- Scanning for single or dual switch users can be configured to permit functional control of a browser. Scanning browser functions can provide a more direct access than simply using a switch to operate mouse emulation. For example, scanning to and selecting a "Scroll Down" scan menu function can be more efficient than selecting "Mouse Right" to move the pointer to the scroll bar and then subsequently trying to manipulate the scroll bar with the mouse pointer. Further development is needed to produce efficient scanning interfaces for browsers that suit a variety of platforms, switch access skills and software interfaces.
- Morse code access provided adequate mouse control and text input for a browser. The smoothest interface is one that provides the greatest ease of mouse control, as most surfing can be done through mouse emulation.
- Frames are a challenge to many assistive technology users (5,7). First, the user must know that frames are present. A frame will not be read by a screen reader unless it is selected and given focus. For the graphical browsers reviewed, switching focus between frames was only possible through mouse control. Only after the frame is selected can a user navigate the frame using a screen reader, voice recognition system or a scanning array.
- A keyboard equivalent for moving forward and backward through hyperlinks would be useful for most alternative access tools. Screen Reader/2 and Jaws for Windows provided this in Web Explorer 2.0 and Netscape 2.0 respectively. This feature would be useful to alternative keyboard users and other screen reading packages if built into the browser itself.
This project was valuable in confirming the effectiveness, and identifying the inadequacies, of various assistive technology devices for World Wide Web access. Detailed reviews of each product with at least one web browser provided much useful information on the product's functionality. As the products approached screen reading or mouse emulation differently, their effectiveness in controlling web browsers varied considerably. The most useful configurations of both the assistive technology and the browsers were also identified, allowing a reader of resulting guidelines to start with a customized configuration of the software in advance. A number of technical obstacles were encountered in setting up both the browsers and the assistive technology, demonstrating that substantial time and technical support is required to support not only a project like this but also individuals wishing to access the Web through adaptive technology.
Liaisons with the manufacturers/software developers were made by phone or in person at the CSUN Technology and Persons with Disabilities conference in March 1996. Although there was general support for our project, not all companies were consistent or forthcoming in our request for feedback on the product reviews. On-going discoveries of how to enhance the utility of the adaptive technologies and web browsers were relayed to the manufacturers. Several manufactures gave constructive feedback on the project, sometimes requesting re-phrasing of product descriptions or operational information. Generally, they concurred with our findings and sometimes requested that we document information on future versions of their product which they claimed would improve function with a web browser. On our recommendation, Hentre-Joyce created new macros enabling their Jaws screen reader to handle column text on web pages. Connections made with consumers for user testing broadened our pool of contacts for future user testing.
Columns and Frames: All tested screen readers read across the page from left to right with disregard for column boundaries. However, after this information was presented to Hentre Joyce, their staff developed four custom Netscape Navigator macros for the reading of columns. This information was passed onto the public through our published guidelines. To the detriment of accessibility, the use of multi-column text may be increasing. A popular web publishing technique is to use invisible tables to gain fine control of web page design layout (10). Usually the entire page is embedded in a complex table resulting in text presented in multiple columns interspersed with images and links.
Multiple invisible Frames are also being used the same way tables are to gain control of the positioning of web page elements (10); however, the problem is further compounded with a focusing issue. Frames are generally inaccessible because the user that is unable to use a mouse cannot control which frame has focus. Only one frame can have focus at a time so the user needs a way to easily alter focus between multiple frames. Additionally, an action in one frame may result in a change in another frame, of which a person using a screen reader would be unaware. In response to the focus issue, the latest version of Microsoft Internet Explorer (version 3.0) provides a keystroke equivalent for this feature.
The World Wide Web Consortium (W3C) is considering a recommendation for the addition of Cascading Style Sheets (CSS) into HTML 3.2 that will allow absolute positioning of elements within an HTML document (14). This will provide an alternative for invisible tables and frames and hopefully should increase the accessibility of web pages with complex layout design if web page creators adopt these style sheets.
Hypertext Links: This project has supported the fact that keyboard equivalents or hotkeys are fundamental to the accessibility of a program. Automation of a frequently used task (which would usually take several mouse movements) into a single keystroke command can increase the efficiency of web browser control at any level. The research demonstrated a direct correlation between the number of tasks automated and the ease of use of the program for web surfing.
Tasks that could not be performed directly with a keystroke equivalent (i.e. choosing a hypertext link in Netscape Navigator 2.0) were longer and more difficult to complete. A simulated mouse mode usually needed to be employed to choose a hypertext link. Controlling the mouse pointer within a graphical user interface can a very difficult and unintuitive task for the user of adaptive technology. Innovative mouse movements systems such as sector arc scanning (WiVik) and grid scanning (Dragon Dictate) can make mouse navigation easier for some, but any function that can be performed using a direct keystroke equivalent as opposed to mouse manipulation should be favored by adaptive technology users. Fortunately, industry has acknowledged the access community and has recently incorporated keystroke equivalents into their web browsers.
At the start of the project, Netscape Navigator 2.0 and Internet Explorer 2.0 were the latest versions available for testing. Although these programs had built-in keystroke equivalents to execute menu / toolbar functions and navigate dialog boxes, they did not have keystroke equivalents for the principal functionality of a web browser: navigating hyperlinks. Soon after the completion of our project, Microsoft Corporation's Windows Accessibility Group added major navigation keystrokes to the release of Internet Explorer 3.0. Tab, Shift-Tab allows the user to quickly move between hyperlinks. Ctrl-Tab and Ctrl-Shift-Tab allows the user to change focus between Frames. Combined, these new features provide momentous functionality for both user and non-users of adaptive technology. A foundation of native keystroke equivalents can greatly increase accessibility to any computer application. Macros for voice recognition and screen reading are easier to create and custom keyboard layouts are easier to make. It was found during the user testing trials that Tab navigation for hyperlinks circumvented the need for mouse control thus significantly decreasing the effort needed for persons using Ke:nx and single-switch scanning to select a hypertext links. Time needed for choosing hypertexts link decreased from approximately 2 minutes/link to approximately 15 seconds/link. Also, with Tab navigation of hyperlinks there is no longer a problem with small targets for persons using alternative pointing devices and/or lacking fine motor control. Netscape Navigator 4.0 Beta also added keystroke navigation features in response to an access committee which included the ATRC.
Browser Commands: Functions of a browser or any other program can be accessed through either direct or indirect keyboard equivalents. A direct keyboard equivalent is defined as a combination of keys when pressed simultaneously will execute an explicit command. An indirect keyboard equivalent means a command must be navigated to through a series of keyboard equivalents. There are advantages and disadvantages to each.
Direct keystroke equivalents can be quickly executed but require memorizing the association between function and key combination. However, there are several keystroke equivalents reserved by the operating system (i.e. Ctrl+C to copy), so a program can eventually run out of reasonable keystroke combinations. "Ctrl+Alt+Shift+F7+Ins" is unreasonable.
Indirect keystroke equivalents require more time to find commands, but easy to follow hierarchies can exist. Most adaptive technology programs adapt to such a hierarchy. For example, if you have a program containing 50 commands, it is easier to break them down into 5 groupings of 10 commands as opposed to a single grouping of 50 commands. A well designed program will use both methods. A frequently used command will be quickly executed (via keystrokes, toolbar buttons, right mouse button menu, etc.) as well as having the command embedded in the menu bar hierarchy. The most commonly used commands have direct keystroke equivalents while the less commonly used commands reside in the menus hierarchy that can be accessed through indirect keystrokes.
New Access Challenges
Future access tools may need to become HTML-aware in order to deal with the increasing complexity of web page formats (7). These access tools would do more than simply relay the information presented in the browser window. They would parse the HTML and present it in a browser-independent accessible format best suited for the user. pwWebspeak is one example of such a tool.
However simply being HTML aware may not be enough. HyperText Markup Language, the foundation of the World Wide Web, is slowly evolving into an application programming interface. It is no longer just the HTML document itself that needs to be accessible, but also the embedded in-line, interactive components (12). Before graphical web browsers such as Netscape and Mosaic, when the WWW was mostly text based, people with disabilities could functionally access the Internet using screen readers. Any obstacles that were encountered were usually not fundamental to the experience of the web site. However, today the World Wide Web is largely image based and is becoming increasingly more dynamic. The Web browser is a GUI (Graphical User Interface) application within a GUI operating system with access obstacles such as tables, columnar text, graphical icons and bitmapped text. Increasing bandwidth will lead to increases in multimedia content and interactivity.
The functionality of browsers such as Netscape Navigator and Microsoft Internet Explorer can be greatly expanded through either third party plug-ins, such as Macromedia's Shockwave player; or embedded components/objects, such as Java applets. Accessing/interpreting and controlling the visual information from embedded applets will become increasingly important for screen readers and alternate access tools. However, because applets and plug-ins have the ability to add functionality to a web page they may provide a means to increase web page accessibility; for example, they may present the same information through multiple modalities.
Scope of Project and Future Direction
The project undertaken was ambitious, attempting to document the features and functions of the major web browsers and provide in-depth reviews of a representative sampling of assistive technology products in 10 weeks or less. The list of web browsers analyzed is substantial and this work will not need to be repeated, only updated, as new versions are released. Previously, the information was collected in a word processor and manually converted by a single HTML editor. In the future, the researchers will directly code their projects in HTML. An HTML template is now provided for future reviewers wishing to update the project. The Diversity Management Directorate will provide French translations of future results.
Subsequent phases of the project will be to address a broader range of assistive technology products on the market, or being released in the near future. The potential for expansion of this project is based on product needs in each category.
- Screen Readers: Other GUI screen readers to be tested include Arctic WinVision, Syntha-Voice Window Bridge, GW Micro Window Bridge and Vocal Eyes. PW Webspeak is a dedicated Screen-Reading Browser that will be tested as well.
- Screen Enlargers: MAGIC, from Handiware, will likely be reviewed.
- Voice Recognition: Voice recognition for the Macintosh and Windows has been covered comprehensively. Future directions include developing Apple scripts for more effective control of Power Secretary with web browsers and testing the Kolvox configuration for web browsers.
- Switch Access: Switch access for the Macintosh needs to be expanded with Click-It from Intellitools and Trakker/Magic Cursor from Madenta. Ke:nx version 4.0 could provide important improvements in scan line access for the Macintosh. Further switch and Morse code access to Windows can be investigated with Madenta's Windows products, SofType and EZKeys.
This research was supported through funding from the Public Services Commission of Canada, Industry Canada. Acknowledgment to everyone at the ATRC who contributed to the project: Chris Serflek, Dena Shumila, Jon Kameoka, Jan Richards, Lake Porter and ATRC Manager, Jutta Treviranus. Recognition goes out to our DMD partners including Peter Field, Peter Stockdale, Miriam Maas and Yves Seguin.
- Adaptive Technology Resource Centre and Diversity Management Directorate. "Universal Internet Access Project"; http://www.utoronto.ca/atrc/rd/uiap/; 1996.
- Adaptive Technology Resource Centre and Diversity Management Directorate. "Contact Information (Universal Internet Access Project"; http://www.utoronto.ca/atrc/uiap/en/contact.htm; 1996.
- Diversity Management Directorate. "PSC Web Site Self Evaluation Test." http://www.psc-cfp.gc.ca/dmd/access/welcome1.htm.
- Gunderson, J. University of Illinois at Urbana; "World Wide Web Browser Access Recommendations"; http://www.staff.uiuc.edu/~jongund/access-browsers.html.
- Letourneau, C. "Accessible Web Page Design" http://www.starlingweb.com/webac.htm; 1997.
- National Centre for Accessible Media. "NCAM/Web Access Symbol"; http://ncam.wgbh.org/webaccess/symbolwinner.html.
- Nguyen, K. "The Accessible Web". Presented at: The National Educational Association of Disabled Students Conference: Futures: Tomorrow Is What We Make It; 1996 Nov 8-10; Toronto, ON.
- Nguyen, K. Adaptive Technology Resource Centre. "Glossary of Terms (Universal Internet Access Project)"; http://www.utoronto.ca/atrc/uiap/en/glossary.htm; 1996.
- Shumila, D.; Richards, J. Adaptive Technology Resource Centre; "Increasing Access to World Wide Web Sites for Blind and Visually Impaired Computer Users"; http://www.utoronto.ca/atrc/rd/library/papers/accessWWW.html; 1995.
- Siegel, D. Creating Killer Web Sites: The Art of Third-Generation Site Design. Indianapolis: Hayden Books; 1996. 270 p.
- Treviranus, J.; Serflek, C. Adaptive Technology Resource Centre; "Alternative Access to the World Wide Web"; http://www.utoronto.ca/atrc/rd/library/papers/WWW.html; 1996.
- Trace Research and Development Centre. "Designing an Accessible Web" http://www.trace.wisc.edu/world/web/.
- Vanderheiden, G.; Chisholm, W.; Ewers, N. Trace Research and Development Centre; "Making Screen Readers Work More Effectively on the Web: A List of Critical Needs"; http://trace.wisc.edu/archive/screen_readers/screen.htm.
- World Wide Web Consortium. "Web Style Sheets"; http://www.w3.org/pub/WWW/Style/; 1996.
Author Contact Information:
Kevin K. Nguyen
Adaptive Technology Resource Centre
University of Toronto
130 St. George Street
M5S 3H1 CANADA
Phone: (416) 946-3001
Fax: (416) 971-2896
Linda S. Petty
Adaptive Technology Resource Centre
University of Toronto
130 St. George Street
M5S 3H1 CANADA
Fax: (416) 971-2629