Pickup Usability Dominates - Human-Computer Interaction and Innovation in Handheld, Mobile and Wearable Technologies - page 49

keys for touch-screens. Mackenzie and his team
have conducted a series of experiments on alter-
native layouts that are optimised for entry using
a stylus (single touch entry). They investigated
both unambiguous keyboards and an optimised
12-key ambiguous keypad, inspired by the success
of T9 and the fundamental rule of interaction that
large targets are faster to tap than small ones (Fitts,
1954). Their results estimate that an expert user
could achieve 40+ wpm on a soft QWERTY
keyboard with novice soft-keyboard users achiev-
ing around 20 wpm (MacKenzie, Zhang, &
Soukoreff, 1999). The alternative layouts were
predicted to give higher entry rates for expert use:
the unambiguous Fitaly layout was predicted to
reach up to 56wpm and ambiguous JustType
44wpm (Figure 7). However, novice users
achieved only around 8wpm using these alterna-
tive keyboard layouts—highlighting the carry-
over effect of desktop QWERTY layout.
While simple and fast, the on-screen keyboard
approach can be tiring for users as they are required
to repeatedly hit very small areas of the screen.
The patented technology underlying the XT9™
Mobile Interface from Tegic Communications
attempts to address this problem by including a
level of disambiguation in an otherwise unam-
biguous keyboard (Robinson & Longe, 2000).
For example, if the user taps letters adjacent to
the letters in the intended word, then the more
likely letters are used instead of the letters actu-
ally tapped. Their approach defaults to the most
likely full word given the approximate letters

entered, while offering alternative corrections and
word completions as well as the letters actually
typed (Figure 8). XT9 technologies have been
developed by Tegic for multiple platforms, includ-
ing hand-printing and small physical keyboards.

Handwriting

To many the obvious solution to text entry on
handheld devices is handwriting recognition.
Modern hand-writing recognition systems, for
example on Windows™ Vista™ tablets, are ex-
tremely good at recognising in-dictionary words
but struggle on words that are not previously
known and are inherently limited by writing
speeds (about 15 wpm (S. Card, Moran, & New-
ell, 1983)). Furthermore, handwriting recognition
needs a reasonably large physical space people
to write in and processing power that is more in
line with modern laptops/tablets than phones. To
target mobiles better, the unistroke (Goldberg &
Richardson, 1993) approach introduced a sim-
plified alphabet to reduce both the processing
complexity and the space, and for experienced
users the time, needed for writing while increas-
ing accuracy. Here each letter is represented as a
single stroke with letters typically drawn on top
of each other in a one letter wide slot and requires
users to learn a new alphabet (Figure 9). Palm
popularised a more intuitive version, Graffiti™,
on their palmtops—a mostly unistroke alphabet,
Graffiti™ was composed mainly of strokes with
high similarity to standard capital letters. CIC's

Figure 7. Fitaly and JustType keyboard layouts