Pickup Usability Dominates - Human-Computer Interaction and Innovation in Handheld, Mobile and Wearable Technologies - page 46

Figure 5. 12-key phone pad (Nokia N73) and stretched phone pad (Blackberry™ 7100)

through 9 plus numeric digits (with the 1, *, #
and 0 keys typically acting as space, shift and
other control keys depending on handset). These
can, however, represent many characters once
accented characters are included (e.g. 2 maps to
ABC2ÄÆÅÀÁÂÃÇ). The method of disambigu-
ating the multiple letters per key is discussed later.
Recently some phones have been released with
a slightly stretched mobile phone pad, typically
with two extra columns, reducing the number of
characters per key and considerably improving
disambiguation accuracy (Figure 5).
While the 12-key mobile phone pad is the
smallest commonly found keyboard layout, there
has been some work on very small keyboards/
devices with as few as three keys. Text entry on
these usually involves cursor movement through
the alphabet using the 3-key date-stamp method
widely used in video games (left and right scroll
through an alphabetic strip of letters with fire
entering the current letter) and 5-key variant using
a joystick with a 2D keyboard display. Experi-
ments have shown users of 3-key date-stamp
entry can achieve around 9wpm while with a
5-key QWERTY layout users can reach around
10-15wpm (Bellman & MacKenzie, 1998; MacK-
enzie, 2002b). These experiments also tried dy-
namically adjusting the layout based on proba-
bilities of next letter but these didn't have the
expected speed-up, probably due to the extra at-
tention load of users cancelling out the reduced
time to select a letter. An alternative approach for

non-ambiguous very small keyboard entry is to
use short-codes representing the letters, for ex-
ample short sequences of cursor keys. (Evre-
inova, Evreino, & Raisamo, 2004) showed that
users could achieve good entry speeds with 3-key
combinations of cursor keys, e.g. left-up-left for
A , and that, despite high initial error rates, users
could learn the codes quickly.

Disambiguation

The traditional approach to disambiguating text
entry on a mobile phone keypad is the manual
multi-tap approach: users press keys repeatedly
to achieve the letter they want, e.g. on a standard
phone keypad 2 translates to A with 22 translating
to B etc. This approach has also been adopted in
many other domestic devices such as video remote
controls. Multi-tap leads to more key-strokes
than an unambiguous keyboard, as users have to
repeatedly click for most letters, and to a problem
with disambiguating a sequence of letters on the
same key, e.g. CAB is 222222 . Users typically
manually disambiguate this by either waiting for
a timeout between subsequent letters on the same
key or hitting a time-out kill button (e.g. right cur-
sor key)—clearly an error-prone process and one
that slows users down. (Wigdor & Balakrishnan,
2004) refer to multitap as an example of con-
secutive disambiguation—the user effectively
enters a key then disambiguates it. An alternative
manual disambiguation approach is concurrent