CAR-TR-553 October 1992
CS-TR-2662
Investigating Touchscreen Typing:
The effect of keyboard size on typing speed
Andrew Sears
Doreen Revis
Janet Swatski
Rob Crittenden
Ben Shneiderman
Department of Computer Science &
Human-Computer Interaction Lab
University of Maryland
College Park, MD 20742
Abstract: Two studies investigated the effect keyboard size has on typing speed
and error rates for touchscreen keyboards using the lift-off strategy. A cursor
appeared when users touched the screen and a key was selected when they lifted
their finger from the screen. Four keyboard sizes were investigated ranging from
24.6 cm to 6.8 cm wide. Results indicate that novices can type approximately 10
words per minute (WPM) on the smallest keyboard and 20 WPM on the largest.
Experienced users improved to 21 WPM on the smallest keyboard and 32 WPM
on the largest. These results indicate that, although slower, small touchscreen
keyboards can be used for limited data entry when the presence of a regular
keyboard is not practical. Applications include portable pocket-sized or palmtop
computers, messaging systems, and personal information resources. Results also
suggest the increased importance of experience on these smaller keyboards.
Research directions are suggested.
1 Introduction
Touchscreens are being used in a growing number of situations: information kiosks,
banking machines, office directories, financial systems, and even by news analysts for the national
news (Sears, Plaisant, & Shneiderman, 1992). One of the motivations for this research is to
continue exploring the possibility of using touchscreens in even more situations, such as palmtop
and pocket-sized computers, portable message systems, and personal information resources.
Currently, the primary method for data entry is the standard QWERTY keyboard. This
works well for computers that are stationary; however, as computers continue to become smaller
and more portable, alternative methods for data entry must be investigated. Touchscreens may
provide an interesting alternative. Touchscreens can be used with pocket-sized or palmtop
computers, something that is difficult or impossible with a mouse. Touchscreens also allow the
interface to be customized to fit the user and the task. For instance, users could specify the
keyboard layout, language, and size that they wish to work with, something that is impossible with
a traditional keyboard. By specifying a smaller touchscreen keyboard users can make additional
screen space available for other uses which could be important on a portable computer with a small
screen. In addition, the keyboard can be displayed only when data entry is required, allowing
additional space to be used for displaying data.
Since there are many situations where the portable computer may be smaller than a
traditional keyboard, alternative input devices must be investigated. The flexibility of touchscreen
keyboards suggests that they may be appropriate for such situations. These studies investigated
the effect keyboard size has on typing performance and user preference for touchscreen keyboards.
Four keyboard sizes were investigated, ranging from 24.6 to 6.8 cm from the Q to P keys. Based
on several previous studies that are discussed in the following section, we expected that typing
would become slower as the keyboard became smaller (Sears, 1991; Sears & Shneiderman,
1991). Time, errors, and preference ratings were collected to assess user performance and
satisfaction.
2
2 Method
2.1 Subjects
Twenty-four seniors and graduate students from the Computer Science Department of the
University of Maryland volunteered to participate in the first experiment as novice subjects. All
were familiar with the QWERTY keyboard, averaging 8.6 years of typing experience.
Four seniors and graduate students from the Computer Science Department of the
University of Maryland participated in the second experiment as experienced users. All subjects
had moderate experience using the touchscreen keyboards. Since these subjects had extensive
knowledge of the purpose of the experiment subjective ratings were not gathered for this second
experiment.
2.2 Equipment
2.2.1 Hardware: An NEC PowerMate 386/25 PC with a Sony Multi-scan HG monitor and
MicroTouch touchscreen was used for both experiments. A special desk allowed the monitor to be
mounted below the surface of the desk at various angles (Figure 1). The keyboard fit into a drawer
which slid into the desk when not in use. The display area of the monitor measured 24.6x18.3 cm
and was used in VGA mode (640x480 pixels). The MicroTouch touchscreen is a capacitive
touchscreen that provides continuous information about the location of a touch on a 1024x1024
grid. It requires only a light touch to be activated and averages the location of all simultaneous
touches, returning a centroid location. The touchscreen was cleaned once before the first subject
began each experiment and was not cleaned again.
Figure 1 - Desk used in experiment
3
2.2.2 Monitor Position: A recent study by Sears (1991) investigated touchscreen keyboards. This
study suggested that the standard monitor position is sub-optimal, at least when using a
touchscreen. The results indicated that mounting the touchscreen so the surface of the screen
makes a 30 degree angle with the surface of the desk may result in reduced fatigue and increased
user preference ratings. These results are also supported by work conducted by Ahlström and
Lenman (1991). For this reason, the monitor used in these studies was mounted at 30 degrees.
2.2.3 Key Size and Selection Strategy: Sears (1991) also investigated the minimum size keys
necessary to yield over 99% accuracy when using the land-on strategy. The results indicate that
square keys must be approximately 2.27 cm per side to capture over 99% of a user’s touches.
These results are similar to those reported by Pfauth and Priest (1981), and by Hall, Cunningham,
Roache, and Cox (1988) who recommended 2.2 cm and 2.6 cm per side respectively.
Using this information, we can conclude that reducing the size of keys below
approximately 2 cm per side will result in unacceptable error rates if the land-on strategy were
used. Since small keyboards are of particular interest, we resorted to an alternate selection strategy,
lift-off. When using the lift-off selection strategy, a cursor appears when a user touches the screen.
The user can then drag the cursor to the desired location and a selection is made where they lift
their finger from the screen (Potter, Weldon, & Shneiderman, 1988). The lift-off strategy has been
shown to be fast and accurate for targets approximately 0.2 cm per side (Sears & Shneiderman,
1991).
2.3 Design and Procedure
Four keyboard sizes were used, with alphabetic keys measuring 2.27, 1.14, 0.76, and 0.57
cm per side. The largest size, 2.27 cm, was chosen to result in the largest keyboard that could be
displayed on the available monitor. The remaining three sizes are 50%, 33%, and 25% of the
largest key size. The space bar, backspace and done keys were all proportional in size. Number
and punctuation keys were not included on the current keyboards to allow comparisons with the
previous study by Sears (1991). Future studies will investigate complete keyboards (Plaisant &
Sears, 1992). The keyboards measured 24.6, 13.2, 9.0, and 6.8 cm from the Q to P keys. Keys
on traditional keyboards are approximately 1.35 cm per side and measure approximately 19.0 cm
from the Q to P key. Keyboards appeared as shown in Figure 2 and will be referred to as Large,
Medium, Small, or Extra Small (L, M, S, XS). The string that users were to type appeared 3.2 cm
above the keyboard, and the string the users actually typed appeared 1.9 cm below the string to be
4
typed (Figure 2). The touchscreen used the lift-off selection strategy, making selection of the
smallest keys possible. The cursor was placed directly below the user’s finger when using the
three larger keyboards, and slightly above the user’s finger when using the Extra Small keyboard.
This difference was necessary to allow users to see the key they were selecting when using the
Extra Small keyboard. Both visual and audible feedback were provided to users. When a key was
touched, it highlighted indicating that the key would be selected if the user lifted their finger. When
subjects did select a key, the key returned to normal colors and a clicking sound was made.
Q
W
E
R
T
Y
U
I
O
P
L
K
J
H
G
F
D
S
A
Z
X
C
V
B
N
M
DONE
THE QUICK BROWN FOX JUMPED OVER THE LAZY DOG
STRING: THE QUICK _
BACKSPACE
SPACE
Figure 2 - Keyboard layout and string positions
Keyboard size and stings entered were within subject variables. Every subject entered one
practice string and three additional strings for which data were collected with each of the keyboard
sizes (Table 1 contains the three strings for which data were collected). Keyboard size was
randomized to prevent any biases. The time from when the first character of each string was typed
until the done key was touched was automatically recorded. In addition, the number of corrected
and uncorrected errors were also recorded. A sequence of backspaces was considered one
corrected error. Any error in the final string was considered an uncorrected error. The time,
corrected errors, and uncorrected errors for each set of three strings using each keyboard size were
recorded for each subject. In addition, subjects in the novice user experiment answered several
questions after completing the experiment.
MONDAY
5
