Pickup Usability Dominates - Human-Computer Interaction and Innovation in Handheld, Mobile and Wearable Technologies - page 52

tap individual letters, improving pick-up-and-use
usability, or to enter words in one gesture by fol-
lowing the path of the letters on the touch keyboard
(experts can then enter the gestures anywhere on
screen) (Zhai & Kristensson, 2003).
Dasher is a drastically different approach to
text entry that attempts to exploit interactive dis-
plays more than traditional text entry approaches.
In Dasher (Ward, Blackwell, & MacKay, 2000)
(Figure 13), letters scroll towards the user and (s)
he picks them by moving the stylus up and down
as the letters pass the stylus. The speed of scrolling
is controlled by the user moving the stylus left and
right with predictive text entry approaches dynami-
cally changing the space allocated to each letter
(so that likely next letters are given more space
than less likely ones, but all letters are available
at each stage). Experiments show that users can
enter at over 30 words per minute.

discussed above, in this section we focus on the
evaluation methods themselves.

Technical Evaluation

The literature commonly uses three methods for
reporting the performance of text entry: average
ranked list position (ARP), disambiguation accu-
racy (DA) and keystrokes per character (KSPC).
The average ranked-list position (e.g. (Dunlop
& Crossan, 2000)) for evaluating ambiguous text
entry methods is calculated in two phases. First
language models, e.g. in the simplest case word
frequencies, are learned from a corpus appropriate
to the target language. Once trained, the second
phase involves processing the same corpus one
word at a time. Each word taken from the corpus
is encoded using the ambiguous key-coding for
the target keypad (e.g. home is encoded as 4663 )
and a ranked list of suggested words produced
for that encoding based on the learned language

EVALUATION

Figure 13. Dasher

Unlike many areas of mobile technology where
market forces and commercial ingenuity dominate,
the field of text entry has benefited from consider-
able scientific study to establish the benefits of
one method over another. These studies have been
conducted by academic and industrial research
groups, often in collaboration, and are used both
to compare techniques and to tune their usage to
how users actually enter text. Much of the related
evaluation work and results have already been

Figure 12. The ATOMIK keyboard with SHARK shortcuts