& 
.
[skip to section contents list]
A search of the digital libraries at ACM and IEEE, as well as Citeseer and the HCI Bibliography Project found very few recent published work about human factors of electronic glossaries to support the use of technical or academic texts, or experiments with such tools. The most notable resources found were studies of how and why people annotate texts (e.g. works by O'Hara et al. [1998], Marshall [1998], and Jackson [2001]). We did however find several excellent articles about the development, and occasional testing, of such systems (e.g., [Kaindl et al., 1999] and [Brown & Brown, 2004]). We agree with the view that prototyping can be a good way to elucidate user needs however in this work we take a more basic approach: we are trying to determine what users need in a hypertextual glossary tool before we commit to build a full-scale one.
In the rest of this review section we summarize what is known about the use of hypertext specifically for glossaries (not annotation in general) in discursive works.
Cognitive overhead can be defined as the added concentration and exertion required in order to maintain multiple tasks or trains of thought at a single time [Wright, 1991].
It is believed that readers form a coherent mental
representation of a text as they read, and that they use that
so-called map to help them to comprehend and use the
text [Charney,
1994; Dillon, 2000].
Since reading hypertext requires a higher cognitive overhead on
readers reading than regular text it is especially important
that nothing in the interface imposes more load without providing
significant benefit [Blustein, 2000].
In her survey of applications of hypertext
applications Charney [1994] notes that if relevant textual passages
are not physically close then readers will be unlikely to make
mental connections between them.
Since glossaries can provide a way to bridge between two parts
of a text they have the possibility to improve readers
understanding of hypertextual documents.
One would expect such bridging to be best achieved by having a
glossary tool that was physically close to a current use of a term
and include sufficient context to remind the user of the other
locations where the term is used, based on Charney's observation.
(For more speculation about glossaries as bridges and their
relationship to open linking services you can jump to the hypertextual significance of this work
section.)
Black et al. [1992] discuss several different styles of glossary tools. Their research involved thorough experimentation into the factors surrounding the usability and effectiveness of different styles of glossary tools. Varying styles of glossary entry presentation were also studied.
Most of the tools studied by Black et al. [1992] employ the use of links embedded in the main text that the user is reading. The links are anchored to terms in the text that have corresponding entries in the glossary. Activating a link presents the corresponding definition. They call them integrated glossary links.
Some form of visual marking was also typically used to identify which words in the text had associated entries [Black et al., 1992]. Wright [1991] states that simply making functionality available to a user is not enough unless the user is made explicitly aware of the presence of this functionality. This view is echoed by Norman [1998] who pontificates that the perceived functionality of a system is what is important to users, even though it may be very different from the system's actual functionality. For example when a user encounters a completely or partially unknown term that is defined in the glossary, then the user must be made aware, perhaps by highlighting the term in the text, that clicking on the term will provide the corresponding definition.
According to Wright [1993] providing the user with the ability to access glossary entries is not always enough to be entirely useful. She expects that the functionality would be much more useful if it had visual support from within the display itself. Black et al. [1992] also mention the importance somehow marking the presence of terms that have entries in the glossary.
The visual presentation of information affects users' consultation strategies when using a glossary tool. Readers' willingness to consult definitions was influenced by whether the definitions occluded the main text or left it visible [Black et al., 1992]. Users were more likely to consult the glossary tool when the text remained visible.
Black et al. [1992] also discuss varying presentation styles for glossary entries. Once accessed, a glossary entry may occlude the main text or it may be presented in the margins alongside the text. Alternatively, the entry may be presented to the user through auditory means. If the auditory presentation style is used, attempting to access a glossary entry would result in a sound file reciting the entry as opposed to the visual presentation of the same entry.
An alternative to integrating glossary links into the main text is presenting a list of glossary terms separately from the main body of text. Terms in the text that is currently being viewed on the screen may be highlighted in the list to make their presence in the glossary more prominent to the user. The glossary entries can then be accessed using the list. This style of glossary is directed more at readers who wish to supplement their vocabulary either before or after reading, but do not want to interrupt their reading.
The list style of glossary has certain advantages and disadvantages when contrasted against glossaries that employ the use of integrated glossary links. Bailey [2002] concluded that varying line lengths in online texts produce different reading speeds in users. While leaving the original text unaltered is an advantage of not having integrated glossary links, Wright [1991] found that the lack of these links produced a drop in users' willingness to access glossary entries. They speculated that this decrease in willingness was due to the increase in cognitive overhead involved .
Despite this disadvantage, this style of glossary is preferable over other glossary styles in certain respects, including ease of implementation, as well as allowing users free access to glossary entries at any time as opposed to only when they encounter terms in the course of their reading the main text. Combining integrated glossary links with this style of glossary allows for individual differences in reading styles. As Wright [1993] states, any generalizations about users' styles of reading should be made with care.
Individual user styles can vary depending on several factors.
A user's level of expertise with a system usually has a
major impact on how the user will interact with the system.
Other simpler factors, such as level of fatigue and mood can
also cause differences in user behavior. Because of differences in
individual users, care should be taken to ensure that systems are
not designed for a specific kind of user behaviour [Nielsen, 1995].
Dillon & Watson [1996, p. 633] reached a similar
conclusion but note that even though there are differences amongst
users that predict performance with interactive systems, appropriate
design of the interface and/or training can reduce these
differences.
While Black et al. [1992] state the willingness of user to access glossary entries when they encounter completely unknown terms increases when the glossary terms are visibly marked, they also note that users' willingness to look up partially known terms depends on whether the method of marking the glossary terms in the main text disrupts the users' flow of reading. It has already been established that signaling the presence of glossary terms is necessary in order to allow the glossary tool's potential to be employed more fully. Underlining these terms does not disrupt the flow of the main text as well as signaling to the users that a corresponding glossary entry exists for the underlined terms. Many readers may also be familiar with the convention of underlined hypertext links on the World Wide Web, and will therefore, be more comfortable with clicking on the glossary term to access its entry.
Wright [1993] states that separating users' access to glossary terms from the main text (e.g. having the terms listed in another division of the window or in a separate window entirely) decreases their willingness to access the glossary entries. This drop in willingness was found to be the case even if the list of glossary terms and the main text were presented on the screen at the same time without occluding each other. This research supports the decision to integrate glossary links into the main text. In this way, users' willingness to access glossary entries will rely more heavily on their perceived level of familiarity with the terms they encounter and less heavily on the glossary tool's interface.
Some readers may prefer to enhance their vocabulary with completely or partially unknown terms in the text either before or after they are done reading, but not during [Black et al., 1992]. If users' access to glossary entries is restricted to clicking on the integrated glossary links, users who prefer this style of reading will have to adapt to a different style of reading or will be discouraged from using the glossary tool at all. This adaptation process will cause an increase in cognitive overhead. For this reason, a list of glossary terms is provided in another division of the interface. Having a separate list of terms allows users to access entries without having to locate them in the main text. Integrating both styles of glossary term presentation caters to both styles of reading, so that neither type of reader is at a disadvantage.
References for works cited in this text chunk appear below. References for all works cited are available in a separate chunk.
Article No. 237, 2004-02-03, 2004.
& .