Usability Problems in Word Processing Applications
Michael S. Wogalter and Jennifer A. Cowley
North Carolina State University
Psychology Department, Human Factors and Ergonomics Program, Raleigh, NC
28695-7650 USA
Word processing is a major work activity for many occupations, including
professionals and students. Although word processing applications have been
available since the earliest personal computers, and some of these applications have
undergone many versions and revisions, it might be expected that usability problems
at this point in their evolution would be minimal. The present study suggests that
there is still room for improvement. Specific findings include: (a) a preference for
toolbars buttons with both icons and textual descriptors, (b) some automatic features
such as "assistants" are disliked, and (c) people wanting easier ways to turn off the
automatic features. Implications of these results are discussed with a focus on
encouraging developers to make use of user data to enhance their product's usability.
Such developments would benefit users’ work productivity and satisfaction.
INTRODUCTION
It has been approximately three decades since
computers, printers, and word processing software
replaced typewriters. The newer tools are vastly
better than the typewriter). Since then, people in
various jobs and learning environments use word
processing applications on a regular basis (e.g.,
Stainton-Skinn, 1982).
Rudimentary word processors have evolved
together with computers over time, including the
change from command based to graphical user
interfaces (GUI). Microsoft Word, developed
originally for the early GUI Apple Macintosh
computer, has become a dominant word processing
application across operating system (OS) platforms.
GUI platforms often utilize pull-down menus,
which are beneficial because users can make use of
people's excellent recognition capabilities as
opposed to making use of people's poorer ability to
recall information (e.g., Wiedenbeck & Davis, 1997).
However, pull down menus can also be
disadvantageous under some circumstances. In
GUI interfaces, making repetitive commands and
accessing deep hierarchical menus are generally
more difficult, awkward and slower. While novice
users generally prefer pull down menus, expert users
have a greater preference for keyboard commands to
accomplish tasks faster (e.g., Wogalter & Frei,
1990).
Another GUI method to deliver computer
commands is through selection from toolbar menus.
Toolbar menus are frequently comprised of set of
several icons on a rollup ribbon, toolbar or a ruler.
When turned on and displayed in an application, the
icons can be selected by a single pointer click. The
use of icons has sometimes been credited as one of
the reasons why GUI interfaces are easier than
command-line interfaces (e.g., Lodding, 1983).
This simple attribution is probably not exactly true
because there are numerous aspects of GUIs that aid
usability besides icons. Additionally, a large
growing body of research shows that icons (and
other kinds of symbolic graphical forms) vary
considerably in understandability (Wogalter, Silver,
Leonard & Zaikina 2006). Whereas, some icons
are comprehended with ease, others are not.
On-screen icons are usually small and of relatively
low resolution. They also may not represent some
concepts clearly (James, Lynk, Molinarie, & Caird,
1995). A symbol that directly represents a
concrete concept will generally be better understood
than one that indirectly represents a lesser-known or
abstract concept. As a result, people can fail to
comprehend the meaning of various kinds of
symbolic forms (e.g., Leonard, Otani & Wogalter,
1999; Mehlenbacher, Duffy, & Palmer, 1989;
Schwalm, Shaviv, & Goldschmidt, 2000). By
extension to the computer interface, if users cannot
identify or recognize the associated meaning of
icons in a toolbar, then the interface is not as useful
as it might otherwise be (Luximon, Ghambaryan, &
Mehrabyan, 2003).
A potential way to raise icon understanding is to
add textual definitions to the icons so that the
symbol and textual definition are shown together.
In other words, instead of the usual icons-only
toolbar, the toolbar would contain words or
abbreviations in addition to the icons. Indeed,
some computer programs give descriptor words for
icons in different ways. One method is to hover
over the icon (tool tip), which after a short delay,
textual descriptors are displayed in an adjacent area.
Other programs give icon definitions using the
hovering method but deliver the text descriptors at a
different designated (constant) location on the
screen. With both of these methods, each icon
must be hovered over individually before its
definition is shown and the process requires more
time and some effort compared to a method in
which the definitions always appear with the icons.
Thus a potentially better method of displaying
toolbar choices would be to have the descriptive text
always present with the icons on the tool bar. One
purpose of the present study was to evaluate
people's judgments regarding having icons and
textual descriptors in word processing toolbars.
Over the years, the number of features provided
in word processing programs has increased, and
consequently, the number of potential icons has
increased. While the available space for icons has
increased over the years due to the availability of
larger screens, the amount of screen space allocated
for toolbar icons will to some extent limit the area
for the main document being worked on. The
point here is that large areas used for toolbars and
icons could adversely affect word processing
performance because parts of the document are
partially obscured (or a reduced, less legible
document is used). Thus it is possible that
combining icons with textual descriptions would not
be desirable due usurpation of valuable screen space
for the document itself.
The present study also examined people's beliefs
about other word processor features beyond the
toolbar. The aspects examined were based partly
on the items used in the research of Levine and
Wogalter (2000). They found that people strongly
disliked certain features of word processing
applications. In the present study, some of the
aspects identified by Levine and Wogalter (2000)
were examined. The categories of problem
features examined concerned formatting, on-line
help, and automation. These aspects were focused
upon because participants in the earlier study
indicated that some functions were needlessly
complex, not obvious to use, and that turning off
functions was difficult and that default automation
was frequently disliked. Thus, as a follow-up to
earlier Levine and Wogalter (2000) study,
participants were asked which features gave them
difficulty in using word processors. These data
were collected to determine if there were
consistently voiced problems.
Thus one purpose of this research was to
examine whether participants would prefer textual
descriptions added to tool bars. Another main
purpose was to examine whether people had any
other issues with the word processor(s) that they use.
We believed that there are substantial usability
problems to current word processor programs
despite numerous iterations of revisions and
versions over the past 30 years. On the other hand,
if we were to find that participants provide ratings
indicating that they experience only limited
problems with their current word processing
programs then this would suggest that the software
has reached a level of maturity in terms of usability.
METHOD
Participants
A total of 187 individuals, with ages ranging
from 18 to 79 years (M = 32.2, SD = 15.1)
participated (47% male and 53% female).
Ninety-eight were non-student adults and 89 were
undergraduates. Overall, race/ethnic composition
was 83.3% Caucasian, 8.1% Asian, and 4.3%
African/African American.
Materials and Procedure
Participants were given a multi-topic
questionnaire. One section contained questions
about computers including which word processors
they have used, and the amount of time they use a
computer and word processing software per week.
Participants provided a self-report of their word
processing expertise on a Likert-type 9-point scale
with the following text and numeric anchors: (0)
beginner, (4) intermediate and (8) advanced. They
were then asked to evaluate a set of features of word
processing software.
One set of questions had three items that
described toolbars with respect to having icons and
text descriptors. They were then asked to rate the
importance of each item on a 9-point rating scale
with the following anchors: (0) not at all important,
(2) slightly important, (4) important, (6) very
important, and (8) extremely important. The three
items regarding the use of icons and text descriptors
in toolbars were:
a. The toolbar should have only text
descriptions for the different functions.
Icons are not necessary.
b. The toolbar should only have icons for
the different functions. Text descriptions
are not necessary.
c. The toolbar should have both text
descriptions and icons for the different
functions.
In addition, several other features concerning
formatting, on-line help, and automation were
evaluated. Table 2 shows the specific list of items
evaluated. The same 9-point Likert-type rating
scale measuring judged importance, as described
above, was used.
Participants were presented with one of two or
orders of features to rate. One was randomized
and the other was the reverse order of that.
Lastly, participants were asked to list difficulties
that they have experienced with their current word
processing program(s) that have caused difficulty,
reduced their productivity, and that they would
recommend being improved by software designers.
RESULTS
Participants reported spending an average of
26.57 (SD = 18.5) hours per week using the
computer and 8.47 (SD = 10.3) hours per week
using word processing software.
The data indicated that 88.2% of the participants
reported using Microsoft Word, 2.1% Word Perfect
and 1.6% Microsoft Works. Several other word
processing applications were mentioned less
frequently. On average, participants rated their
level of expertise between intermediate and
advanced according to the rating scale anchors (M =
5.27, SD = 1.8).
Table 1 shows the means and standard
deviations for the ratings of toolbars with icons only,
text only, or both icons and text. A one-way
repeated measures ANOVA identified a significant
effect, F(2, 372) = 82.70, MSE = 4.73, p < 0.0001.
Comparisons among the means using Tukey’s HSD
test revealed that the “text/icon combination” (M =
5.54) was rated significantly higher than both the
icon-only (M = 3.26) and the text-only (M = 2.85)
mean ratings. A planned comparison indicated the
icon-only toolbar was rated significant higher than
the text-only condition, F(1, 186) = 4.29, MSE =
3.69, p < 0.05.
TABLE 1. Mean importance ratings and standard
deviations of word processing features on a 0 (low)
to 8 (high) point scale.
Toolbar Content Mean SD
_____________________________________
Text only 2.85 2.3
Icons only 3.26 2.4
Both text and icons 5.53 2.2
________________________________________
Table 2 shows the mean ratings (and standard
deviations) for several word processor features.
The Tukey's HSD critical difference at p < .05 is .68,
meaning that any difference between means bigger
than this number in Table 2 is significant. Although
the list of features vary considerably, the mean
ratings suggest several trends. In general,
participants gave higher importance ratings to
features that allow users to do specific tasks such as
file insertion and formatting. Lower ratings were
given to features involving greater automation.
TABLE 2. Mean ratings and standard deviations
of importance levels of other features in word
processors on a 0 to 8 point scale.
_____________________________________________
Feature Mean SD
_____________________________________________
Inserting graphs, pictures, etc. from
other files 6.34 1.8
Being able to copy formats from one section
to another 6.04 1.9
Being able to apply many different types
of tabs 6.01 1.8
Auto highlighting of improper grammar 5.88 2.1
Being able to apply different headers and
footers within sections of the document 5.75 1.9
Auto spelling corrections 5.72 2.3
Being able to apply different numbers of
columns in sections of a document 5.31 2.0
Auto capitalization of words 4.55 2.6
Automated outline creation 3.73 2.3
Having a “wizard” character pop-up to
give direction (for example,
Mr. Paperclip) 2.72 2.3
_____________________________________________
The final question asked participants to list
aspects of word processors that have most decreased
their productivity and need improvement. The
most common response category (mentioned by
50.4% participants) concerned auto-formatting.
Representative responses were: (1) “many auto
features are more of a nuisance than a help,” (2)
“does things for you without you wanting it at
times,” (3) “auto-formatting takes over too much
and is sometimes hard to overcome,” and (4) "there
ought to be an obvious way to undo the automatic
functions." Other frequent responses were: (1)
spelling or grammar checker gives wrong
suggestions, (2) setting of the margins and tabs is
difficult, (3) more control wanted over
auto-capitalization, and (4) annoyance with
assistants/wizards.
DISCUSSION
While word processing applications have
grown in complexity and sophistication since the
introduction of microcomputers about 30 years ago,
usability problems have not disappeared. One
fundamental feature of contemporary word
processors (and many other applications) is tool bars.
Most toolbars contain only icons, and no text. The
present results show a preference for icon-only
toolbars over text-only toolbars (Garcia, Badre, and
Stasko, 1994; Lodding, 1983). However, most
preferred was having both icons and textual
descriptors in toolbars. This result makes sense
given that research suggests that people do not
understand the meaning of some of the icons (even
some commonly used ones). The added text
definition or description provides an apparent way
to learn what those less understandable icons mean.
Although some word processors provide text
that relate to the icons, text descriptors are provided
in awkward manner that is time consuming by
requiring the user to hover the pointer over
individual icons to deliver text in the form of bubble
help or presented at another location on the screen.
Some applications (but not many word processing
applications) have toolbars with both icons and text
(full words or abbreviations).
A disadvantage of including text in the
toolbars is that it would consume potentially
valuable screen space. For example, it could limit
the size of the document window and adversely
affect word-processing performance. It could add
visual clutter. However some of the problems might
be overcome by allowing the user to easily switch
off the text portion of the icon bar when not needed
or after the icons have been learned. Of course, the
text would need to be quite small when
accompanying the icons. It should be displayed
with a font that is maximally legible. Also, some of
the text descriptors may need to be abbreviated due
to space (letter count) restrictions. Research has
shown that small text size and abbreviations can
adversely affect usability (Sanders and McCormick,
1993). However, if needed, short abbreviations
could be combined with a “hover over” maneuver to
enlarge the space for a larger, longer description.
Alternatively, users could be allowed to enlarge the
size of the default text if it is too small for them.
Despite the disadvantages of using small text and
losing some screen space, the present study’s data
showed a preference for having the text visible and
available as opposed to it being absent. Usability
testing could help determine the most important
parameters here and what tradeoffs and
compensatory aspects should be incorporated.
The ratings of selected word processor
features showed an interesting general pattern.
The highest rated features suggested a desire to have
some level of manual control in executing certain
word processor tasks. Although all of the features
participants rated involve some form of automation,
the lowest rated items in the list (but not all of them)
tended to be ones that involve more ongoing
assistance. Also they tended to be features that
make "assumptions" about what the user wants.
Some of those assumptions may be incorrect, which
could add to the higher workload and potentially
add errors to the worked-on document. Together,
this pattern of data suggests that people are having
negative experiences in the ways the automatic
features have been implemented.
The lowest ratings were for the
wizard/assistant tool. While this tool has extensive
and complex capabilities, the resulting actions may
not be what the user wants or intends.
A somewhat similar account is given by the
responses to an open-ended question asking to list
the word processor features have caused them
difficulty and need to be improved. Like the
ratings, the open-ended responses tended to point to
automatic features as being problematic. The
responses suggested a dislike of automation because
they often generated undesired responses that were
difficult to change or turn off.
The intent of this article is not criticize
particular word processing applications or the
companies that manufacture and distribute the
software, but rather to point out that current word
processors still have substantial usability problems.
Thus, despite many years and multiple versions and
revisions of word processing software, the software
has not approached a high level of maturity in terms
of usability. Word processors have not evolved to
the point where usability problems have been
minimized.
This is unfortunate, because word processors
are a main productivity tool for many occupations
and production tasks. It is used in a wide range of
tasks, as for example producing intellectual products.
Poor usability can translate into wasted time,
disturbance, and potentially the production of lower
quality work documents. The goal of word
processing software should be to enable writers to
record their verbal ideas without being encumbered
by the interface. Ideally the application should not
get in the way of idea flow. In other words, the
tool should be nearly "invisible" so as not to distract
attention from the writing task.
REFERENCES
Garcia, M., Badre, A. N., and Stasko, J. (1994).
Development and validation of icons varying gin
their abstractness. Interacting-with-Computers, 6,
191-211.
James, K., Lynk, D., Molinarie, and Caird, J.K., (1995).
The comprehension and use of word processing
icons. Proceedings of the Human Factors and
Ergonomics Society, 39, 941.
Leonard, S. D., Otani, H., and Wogalter, M. S. (1999).
Comprehension and memory. In M. S. Wogalter, D.
M. DeJoy, and K. R. Laughery (Eds.). Warnings
and Risk Communication. (pp. 149-188). London:
Taylor and Francis.
Levine, J. D., and Wogalter, M. S. (August, 2000). On
determining the relative importance of word
processor functions and deficiencies. Proceedings
of the International Ergonomics Association & the
Human Factors and Ergonomics Society Congress,
44 (1), 362.
Lodding, N. K. (1983). Iconic Interface. IEEE Software,
March-April, 11-20.
Luximon, A., Ghambaryan, A., and Mehrabyan, A.
(2003). Usability of operating Systems' (OS icons.
Proceedings of the International Ergonomics
Association, Seoul, South Korea.
Mehlenbacher, B., Duffy, T.M., and Palmer, J. (1989).
Finding information on a menu: Linking menu
organization to the user’s goals. Human-Computer
Interaction, 4, 231-251.
Mehlenbacher, Wogalter and Laughery (1999). On the
reading of product owner's manuals: Perceptions
and product complexity. Proceedings of the
Human Factors and Ergonomics Society, 46,
730-734.
Sanders, M. S., & McCormick, E. J. (1993). Human
factors in engineering and design (7 Ed.). New
York: McGraw-Hill.
Schwalm, N. D., Shaviv, V., and Goldschmidt, G. (2000).
Can icon animation enhance human performance. or
is it just another gimmick? Proceedings of the
International Ergonomics Association & the Human
Factors and Ergonomics Society Congress, 44 (1),
323-326.
Stainton-Skinn, J. (1982). Business and secretarial
studies: The impact of word processors. Journal of
Vocational Education and Training, 34, 7-9.
Wiedenbeck, S., and Davis, S. (1997). The influence of
interaction style and experience on user perceptions
of software packages. International Journal of
Human-Computer Studies, 46, 563-588.
Wogalter, M. S., & Frei, R. L. (1990). Social influence
and preference of direct-manipulation and
keyboard-command computer interfaces.
Proceedings of the Human Factors Society, 34,
907-911.
Wogalter, M.S., Silver, N.C., Leonard, S.D. and Zaikina,
H. (2006). Warning symbols (Chap. 12). In M.S.
Wogalter (Ed.) Handbook of Warnings (pp.
159-176). Mahwah, NJ: Lawrence Erlbaum
Associates.
