Pickup Usability Dominates - Human-Computer Interaction and Innovation in Handheld, Mobile and Wearable Technologies - page 48

approximately 40% reduction in text entry
speed—however, still considerably faster than
2D date-stamp approaches for very small devices.
An interesting alternative input method for small
devices is to use a touch-wheel interfaces, such
as those on iPods ™ . (Proschowsky, Schultz, &
Jacobsen, 2006) developed a method where users
are presented with the alphabet in a circle with
a predictive algorithm increasing the target area
for letters based on the probability of them being
selected next, so that users are more likely to hit
the correct target when tapping on the touch-wheel.
User trials showed around 6-7 words per minute
entry rates for novices, about 30% faster than the
same users using a 1D date-stamp approach on
a touch-wheel.
The letters on an ambiguous phone keypad do
not, of course, need to be laid out alphabetically.
Here the disambiguation method introduces an
additional aspect to designing an optimal layout:
the letters can be rearranged to minimise the level
of ambiguity for a given language in addition to
looking at minimising finger movement. How-
ever, experiments predict that a fully-optimised
phone layout would improve text entry rates by
only around 3% for English (Gong & Tarase-
wich, 2005). We found a larger but still small
improvement of around 8% in keystrokes for a
pseudo-optimised 4-key letter layout (Dunlop,
2004). Gong & Tarasewich do, however, show
that stretching the standard phone pad from
eight to twelve keys for text entry (see Figure 5)
is likely to reduce prediction errors by around
65% for optimised keyboard layouts (Gong &
Tarasewich, 2005).

for the screen and leads to natural mouse-like
interaction with applications. Lack of a physical
keyboard has led to many different approaches
for text entry on touch-sensitive screens that will
be discussed in this section: on-screen (or soft)
keyboards, hand-writing recognition and more
dynamic gesture-based approaches.

On-Screen Keyboards

A simple solution to text entry on touch-screens
is to present the user with an on-screen keyboard
that the user can tap on with a stylus, or on larger
touch-screens with their fingers. The most com-
mon implementation is to copy the QWERTY
layout onto a small touch sensitive area at the bot-
tom of the screen (Figure 6). Following a similar
experimental protocol to (James & Reischel, 2001)
we conducted an initial experiment on three expert
iPhone users. James & Reischel measured expert
performance on chat style messages at 26wpm;
using the same phrases our initial study showed
iPhone entry around 50wpm (mean 51.6, stdev 1.5
with similar error rates, though the iPhone spell
checker corrected about half of these). Although
a very small independent sample study, this does
indicate that practiced iPhone users may be up to
twice as fast as T9 users.
As with physical keyboards and keypads, there
has been research into better arrangements of the

Figure 6. iPhone™ on-screen QWERTY keyboard

TOUCH-SCREEN TEXT ENTRY

Compared to mobile phones, personal organisers
(PDAs) have made more use of touch screens and
stylus interaction as the basis of interaction and
this is now emerging on high-end phones such as
Apple's iPhone. This frees up most of the device