Letter To The Editor: Pie Menus and Soft Keyboards

Just got Vol. 1, Number 5 and read your comments on the use of pie menus. A couple points that may be of interest. First, the idea of pie menus was first suggested in: Wiseman, N.E., Lemke, H.U. & Hiles, J.O. (1969), PIXE: a new approach to graphical man-machine communication. Proceedings of the 1969 CAD Conference, Southampton, IEE Conference Publication 51, 463.

They were then described in: Newman, W.M. & Sproull, R. (1973), Principles of Interactive Computer Graphics, 1st ed., New York: McGraw Hill. 2nd ed., 1979. As you correctly state, the first formal study was: Callahan J., Hopkins, D., Weiser, M. & Shneiderman, B. (1988), An empirical comparison of pie vs. linear menus. Proceedings of CHI’88, pp. 95-100.

You commented that you would like to see an evaluation using a stylus rather than a mouse, and to also see a comparison. That study has been run and is currently under review by the journal, Human Computer Interaction. The following is the proper citation and abstract: Kurtenbach, G., Sellen, A. & Buxton, W. (1991), Markings and pie menus: making stylus-driven interfaces self-revealing. [submitted to Human Computer Interaction]

We argue that pie menus can provide a self-revealing way of showing users the association between markings and commands they can invoke in stylus-based interfaces. By adding a virtual ink trail to the path of the cursor, markings are the natural by-products of the strokes necessary to select a sector from a subdivided circular menu (or “pie”). Thus, novice users can “pop-up” a pie menu and make a selection, whereas experts can simply make the mark without waiting for the menu to appear.

With this interface concept in mind, we carried out an experiment to elucidate the articulatory and cognitive aspects involved in combining pie menus with marks. One group of subjects used fully visible or “exposed” menus and two other groups of subjects used “hidden” menus: half with an ink trail, and half without. Exposed menus were designed to reveal the articulatory aspects of menu selection. Hidden menus involve recalling or mentally reconstructing menu layout, and thus were designed to inform about cognitive aspects of performance. Both menu size (number of slices per menu) and input device were systematically varied in order to assess their effects on both aspects of performance.

Among the findings was the fact that performance on hidden menus was as good as performance on exposed menus for small menu sizes. Performance on hidden menus quickly improved with practice, whereas it stayed relatively constant for exposed menus. Certain menu sizes (4, 8, and 12) facilitated performance when menus were hidden because these particular menus allowed subjects to more easily infer the position of the target. We also found the trackball to be the worst of the three devices tested. Mouse and stylus yielded similar performance, but the stylus outperformed the mouse given that an ink trail was present.

Another paper describing the use of pie menus in pen-centric applications is the following paper which is forthcoming in the UIST proceedings: Kurtenbach, G. & Buxton, W. (1991), Integrating mark-up and direct manipulation techniques. To appear in the Proceedings of the Fourth ACM SIGGRAPH Symposium on User Interface Technology (UIST), Nov. 11-13, Hilton Head, S.C.

The direct manipulation paradigm has been effective in helping designers create easy to use mouse and keyboard based interfaces. The development of flat display surfaces and transparent tablets are now making interfaces possible where a user can write directly on the screen using a special stylus. Part of the intention of these interfaces is to exploit user’s existing handwriting, mark-up and drawing skills. This paper reports on a test-bed program which we are using for exploring hand-marking types of interactions and their integration with direct manipulation interactions.

Pie menus are more than a gimmick. They are especially well suited for pencentric applications. The Momenta implementation is a good start, and an example of appropriate use of technology from a human-factors perspective. That is not to say that there is not more research to be done. For example, there are interesting questions about embedded, or hierarchic pie menus, something which we are currently studying.

What is important is that the use of pie menus in this class of system is not proprietary, and the community at large could benefit from their use. (The citations above establish clearly that they are in the public domain.)

Soft Keyboards Faster than Printing

One other brief point. In your article on the Momenta system, you are correct in stating that “This is slow and inefficient when extensive typing is necessary” when discussing the stylus-driven “soft keyboard” on the Momenta. However, what about when there is not extensive text entry? And how does text entry with the “soft keyboard” compare to stylus entered printed text? (The first of these relates directly to your piece on “Getting By Until Real Recognition is Ready.”

Let’s take the second point first. Typing on a soft numerical keyboard (such as a calculator or telephone keypad) with a stylus is significantly faster and more accurate than printing the same information with a stylus. This can be verified empirically, or by modelling using Fitts’ Law. (This is not true if the soft keys are too small.) Second, typing on a soft QWERTY keyboard using a stylus is faster for someone familiar with the keyboard layout than printing the same text with a stylus is. It is also more accurate.

Your point about the soft keyboard, therefore, is equally true for printed character entry. Secondly, the “Getting By…” article could well have mentioned the soft keyboard entry technique.

My comments do not suggest that typing is the same as printing. Printing obviously lets us enter location, size and face (e.g., bold by pressure), for example, all at once — something that typing will never accomplish. However, virtually no current recognizers deliver on this potential. As long as we are constrained to print in little boxes in special locations, typing on soft keyboards is a viable design alternative which may well result in faster and more accurate input for the user (especially with numerics), and significantly lower development cost for the applications programmer.

William Buxton, Adjunct Professor, Dept. of Computer Science, University of Toronto