This paper evaluates the usability of keyboard-only access to Windows 95. It establishes requirements for Windows-compatible expanded keyboards; determines whether keyboard shortcuts are usable in practice; highlights barriers faced by people who cannot easily use a mouse or other pointing device; and recommends ways to improve the keyboard-only interface.
Manipulating a mouse or other pointing device may be difficult or inefficient for people who operate computers with a single finger, toe, or stump, or use a head-stick, mouth-stick or similar appliance; have mobility impairments that affect upper-body coordination, such as cerebral palsy and dystonia; have mouse-induced repetitive strain injuries (RSIs); have low-vision; and are blind. For these populations, computer access is facilitated by keyboard-only techniques.
Three recent developments give rise to the hope that the Windows 95 environment is accessible without a mouse:
Notwithstanding these developments, questions remain about the viability of keyboard-only access to Windows 95. This paper explores three issues:
This research was conducted between 1996 and 1998 while I provided accommodation support services to people with disabilities. The clients included:
As potential buyers of expanded keyboard systems, the question of "mouseless" access to Windows is especially relevant to single-digit typists. Are commercially-available expanded keyboards compatible with Windows 95's keyboard-only interface?
While developing accommodations for the client who typed with one toe, we considered three expanded keyboard systems: the Intellitools IntelliKeys, the TASH WinKing; and the Unicorn Model II keyboard/Darci Too Computer Control Device.
Not all expanded keyboards are equal to the task of keyboard-only access. We determined that the keyboard interface is fully functional only if (a) the expanded keyboard has all 101 keys of a standard PC keyboard; (b) it supports latched and locked modifier keys, singly and in combination, with all keys, including mouse keys; and (c) its proprietary sticky key and mouse emulation software, if they exist, can be switched off. Access is compromised unless all three criteria are met.
Of the three keyboards, the IntelliKeys alone satisfied the criteria. However, other factors — relating to the size and spacing of the keys and the client's seating posture — compelled him to choose the Unicorn/ Darci Too system. He liked the bigness of the Unicorn, and was willing to trade a modicum of function for long-term physical comfort. With considerable effort, we were able to create workarounds for most of its incompatibilities with Windows 95.
There is no "perfect" expanded keyboard for single-digit computer work. When helping a single-digit typist select a new keyboard, tradeoffs must be made between productivity, technical features, health and safety factors, and personal preferences.
Keyboard-only access to Windows is always possible using MouseKeys. However, not everybody can use MouseKeys, and the technique is cumbersome at best. A simple drag-and-drop operation can use ten different keys and take 30 or more keystrokes. Choosing an item from a pull-down menu is easier, but still tedious.
Keyboard shortcuts, by contrast, are quick and positive. For example, to Exit from any Windows application, press the three-key sequence Alt, spacebar, C. Keyboard shortcuts save time, energy and frustration. Using MouseKeys, one of my clients needed 20 to 30 seconds to select an item from a menu. Using keyboard shortcuts, she completed the same task in under two seconds.
People who progress from MouseKeys to keyboard shortcuts see a dramatic increase in their productivity. However, mastery of the keyboard interface does not come easily, for reasons that will become clear.
Although Window 95's keyboard-only interface has many outstanding features, the goal of reliable access remains elusive. Some aspects of the keyboard interface appear to be retrofits, and consequently, are poorly integrated into the overall design of the operating system.
In addition, significant barriers stem from developers who ignore keyboard interface design guidelines, such as the standards set out by Microsoft , Vanderheiden & Lee , and others. Although keyboard-only access is almost always possible, it is not particularly usable.
Paciello  defines product usability in terms of five factors: (a) how easy it is to learn; (b) how easy it is to remember; (c) whether it promotes productivity; (d) whether it reduces the chances of error; and (e) user satisfaction. Access barriers associated with the Windows 95 operating system and many Windows applications render the keyboard interface less than usable.
The practical problems associated with using the keyboard-only interface are of three kinds: important information that is hard to see, navigation by keyboard that is overly complex, and operations that do not work properly.
Efficient use of the keyboard interface depends on the ability to immediately spot the focus and pick out important information. For keyboard-only users who can see, detecting information on the screen may not be easy:
Navigational complexity refers to difficulties moving around or performing tasks using keyboard equivalents. The threshold of navigational complexity is reached when an experienced user is compelled (or forced) to abandon shortcut keys for a pointing device to complete a task. Some specific examples:
Functional problems refer to access barriers that result from poorly designed or implemented software. In other words, things do not work as one might expect, which contributes to user frustration:
These problems (and others) create significant and needless obstacles for people who demand mouseless access to Windows. The barriers are maddening to the people I have taught keyboard-only techniques, and are completely unnecessary: the principles of accessible software design are widely known.
With recent software upgrades, there have been incremental improvements in keyboard-only access; but there have been setbacks as well. In Office 95, the menu bar and the selected menu title appear as contrasting colours. In Office 97, the menu bar is grey and the selected menu title appears as a raised grey button. The lack of contrast has created a new access barrier for keyboard-only users: now it is hard to tell when the menu bar has been activated.
The keyboard-only interface of Windows 95 is basically accessible, but not especially usable. Once mastered, the interface boosts productivity; but on other measures of usability, the design fails: it is hard to learn and remember, produces unnecessary errors, and does not promote user satisfaction. Of the six adults I have taught mouseless techniques, only one continues to use them.
The mouseless interface could be much better. These measures should go far to improving the keyboard-only interface of future versions of Windows:
The difficulties of the keyboard-only interface are symptomatic of a larger problem. Windows is a rich and complex computer environment. Its organizational and operating characteristics have not matured to the point where the environment is as "intuitive" as Microsoft's promotion materials would have us believe. Many aspects of the overall design could be improved by adhering to the cardinal Universal Design principle: consider all intended users. Significant improvements to the interface are achievable if developers and manufacturers consult with people who demand keyboard-only access. Such collaborations would result in applications that are easier to learn, easier to remember, promote productivity, reduce error, and increase user satisfaction.
 Cantor, Alan. An Evaluation of Keyboard-only Access to Windows for "Single-Digit" Typists. In RESNA `97 Proceedings. Rehabilitation Engineering and Assistive Devices Society of North America Annual Meeting, June 20-24 1997. Pittsburgh, PA. An updated version appears at http://www.cantoraccess.com.
 Snyder, Maryanne K. & Lowney, Gregory C. Microsoft Windows Keyboard Guide. October 17, 1996.
 No author. The Microsoft® Windows® Guidelines for Accessible Software Design: Creating Applications That Are Usable by People with Disabilities. Redmond, WA.: Microsoft Corporation. May 7, 1997 Edition.
 Vanderheiden, G. C. & Lee, C. C. Considerations in the Design of Computers and Operating Systems to Increase their Accessibility to Persons with Disabilities (Version 4.2). Madison, WI.: Trace R&D Center. 1988.
 Paciello, Michael. Accessibility By Any Other Name Is ...Usability. EIA/CEG "CE Network News." December 1993.