Information Technology Reference
In-Depth Information
one perspective in the discourse. Whether the discourse was presented as monolog
or dialog had no significant effects. This finding indicates that Fox Tree's (1999) ear-
lier evidence of better comprehension in her dialog condition was probably because
multiple perspectives are more likely to be presented in natural dialog than monolog.
It should be noted, though, that Schober and Clark (1989) obtained superior perfor-
mance among those who participated in dialog than those who only overheard the
dialog, a discrepancy that remains to be resolved. Fox Tree and Mayer (2008) also
explored whether accuracy correlated with seven stimulus variables. Three correla-
tion coefficients were significant (pp. 172-173). Accuracy was positively correlated
with the liveliness of the description, but there were two significant negative correla-
tions, one between accuracy and number of discourse markers (cf. Fox Tree, 1999)
and another between accuracy and the number of disfluencies. It seems quite possi-
ble that the enhanced comprehension that resulted from multiple perspectives was
due to an increased likelihood that listeners overheard a perspective that (partially)
mapped onto an existing mental model (or schema) and allowed deeper processing
(Chi, 2000; Chandler & Sweller, 1992).
McKendree et al. (2001) explored the role of questions in another referential
communication task, one involving the use of the Map Task Corpus (see Map Task
Corpus, URL, in references; also Brown, Anderson, Shillcock, & Yule, 1984). In
the Map Task each of the two participants received schematic maps that were not
totally consistent, in that some landmarks were represented differently. An instruc-
tion giver's map contained a route drawn from a start point to an end point. The
giver's task was to describe the route on the map to an instruction follower , whose
task was to draw in the route.
McKendree et al. (2001) noted that when Anderson (1995; Anderson, Clark,
& Mullin, 1994) examined a large corpus of Map Task data she found that com-
prehension was enhanced when new map features were introduced into the dialog
by interrogatives rather than declaratives. Inspired by these findings, McKendree
et al. (2001) selected two audiotapes of dialogs between givers and followers from
the original Map Task Corpus. Followers in the McKendree et al. (2001) study
attempted to draw routes on maps identical to those used to collect data in the
original Map Task Corpus. In one taped dialog 12 new map features were intro-
duced using questions (High-Intro), while in the other only five new map features
were introduced using questions (Low-Intro). Listeners heard either the tape of the
High-Intro or Low-Intro dialog. Those in the Low-Intro condition made deviation
errors nearly three times as large as those in the High-Intro condition. McKendree
et al. pointed out that comprehension of their overhearers in the High-Intro group
was also slightly superior to the total group of active participants (the followers
conversing with the givers) in the original Map Task research (Brown et al., 1984),
who used the same map and route but were not selected on the basis of the number
of questions included in the dialog. It seems reasonable to conclude that a second
feature of discourse that facilitates comprehension, in addition to overhearing mul-
tiple perspectives (Fox Tree & Mayer, 2008), is increasing the number of embedded
questions used to introduce new information (Anderson, 1995; Mckendree et al.,
2001).
Search WWH ::




Custom Search