Pickup Usability Dominates - Human-Computer Interaction and Innovation in Handheld, Mobile and Wearable Technologies - page 43

Unambiguous Keyboards on their handhelds. Researchers in academia and

industry have been working since the emergence of
handheld technologies for new text entry methods
that are small and fast but easy-to-use, particularly
for novice users. This article will look at different
approaches to keyboards, different approaches
to stylus-based entry, and how these approaches
have been evaluated to establish which techniques
are actually faster or less error-prone. The focus
of the article is both to give a perspective on the
breadth of research in text entry and also to look
at how researchers have evaluated their work.
Finally, we will look at perceived future direc-
tions attempting to learn from the successes and
failures of text entry research. Throughout this
article we will cite words-per-minute (wpm) as
a fairly standard measure of typing speed, for
reference highly skilled office QWERTY touch
typists achieve speeds of around 135wpm while
hand-writing with pen and article achieves only
about 15wpm.

Small physical keyboards have been used in
mobile devices from their very early days on
devices such as the Psion Organiser in 1984 and
the Sharp Wizard in 1989 and have seen a recent
resurgence in devices targeting email users, such
as most of RIM's Blackberry range. While early
devices tended to have an alphabetic layout, the
standard desktop QWERTY family of layouts, e.g.
QWERTY, AZERTY, QWERTZ and QZERTY,
was soon adopted as there is strong evidence
that alphabetic layouts give no benefits even for
novice users (Norman, 2002; Norman & Fisher,
1982). When well-designed, small QWERTY
keyboards can make text entry fast by giving the
users good physical targets and feedback with
speeds measured in excess of 60wpm (Clarkson et
al 2005). However, there is a strong design trade-
off between keys being large enough for fast, easy
typing and overall device size with large-fingered
users often finding the keys simply too small to
tap individually at speed. Physical keyboards also
interact poorly with touch-screens, where one
hand often needs to hold a stylus, and they reduce
the space available on the device for the screen.
The QWERTY keyboard layout was designed
as a compromise between speed and physical
characteristics of traditional manual typewriters:
the layout separates commonly occurring pairs of
letters to avoid head clashes on manual typewriters
and is imbalanced between left and right hands.
Faster touch-typing office keyboards such as the
Dvorak keyboard (Figure 1) are significantly faster
but have not been widely adopted—primarily
because of the learning time and invested skill-
set in QWERTY keyboards. This investment has
been shown to carry over into smaller devices,
where the sub-optimality issue is even stronger as
users tend to type with one or two thumbs—not
nine fingers envisaged of touch-typists. While
optimal mobile layouts could be designed around
two-thumb entry, these are likely to be so differ-
ent from users' experiences that initial use would

KEYBOARDS

The simplest and most common form of text entry
on small devices, as with large devices, is a key-
board. Several small keyboard layouts have been
researched that try to balance small size against
usability and text entry speed. Keyboards can be
categorized as unambiguous, where one key-press
unambiguously relates to one character, or am-
biguous, where each key is related to many letters
(e.g. the standard 12-key phone pad layout where,
say, 2 is mapped to ABC ). Ambiguous keyboards
rely on a disambiguation method, which can be
manually driven by the user or semi-automatic
with software support and user correction. This
section looks first at unambiguous mobile key-
board designs, then at ambiguous designs and,
finally, discusses approaches to disambiguation
for ambiguous keyboards.