Information Technology Reference
In-Depth Information
interactive
condition were recorded and presented to a participant in a
yoked-
vicarious
condition. In a second vicarious condition a virtual tutor presented a
monolog
on each of the 12 topics that contained all (102) content statements in the
ideal answer and expectations from AutoTutor's script. In a
half questions
vicarious
condition, half (51) of the sentence in the ideal answers and expectations were each
preceded by a deep question spoken by a second voice engine, with the remainder
presented as monolog. In a
full-questions
vicarious condition each sentence in both
the ideal answers and the expectations was preceded by a deep question (total
=
102
questions). Total time in the latter three conditions was about 30 min.
Results yielded a significant effect of instruction condition. Learners in the full-
questions vicarious condition significantly outperformed those in each of the other
four conditions. The pretest-to-posttest learning gains for the other four conditions
were reasonably comparable (see Craig et al., 2006, Table 1). A major finding
of the study, of course, was that learners in the full-questions vicarious condition
outperformed those who were engaged in the interactive tutorial dialog. This find-
ing highlights the impact on vicarious learning of embedding deep questions into
educational content as part of dialog when presenting course content. It should be
pointed out, though, that the 51 deep questions presented in the half questions con-
dition were not sufficient to significantly enhance learning compared to tutoring in
the interactive condition, the yoked vicarious, or the monolog condition. Because
the impact of embedding deep questions in educational content may have impor-
tant implications for curriculum design in distance learning and in computerized
learning environments in general, it was deemed necessary to replicate the finding,
first among college students (Craig et al., 2006) and then among younger students
(Gholson et al., 2009).
In Exp. 2., Craig et al. (2006) included four instruction conditions: interactive
tutoring by AutoTutor, yoked-vicarious, full-questions vicarious with (102) deep
questions and content statements presented as dialog, and full-questions vicarious
with deep questions presented as part of monolog. In the full-questions monolog
condition, the same agent and voice engine used in the interactive tutoring and the
yoked-vicarious conditions spoke both the deep questions and the content state-
ments. In the full-questions dialog condition, the questions were asked by a second,
distinct voice engine.
Analyses revealed that the two deep questions conditions significantly outper-
formed both the interactive and the yoked-vicarious conditions. Neither the two
former nor the two latter conditions differed from each other. The Craig et al. (2006)
research, then, replicated the facilitating effects of deep questions as a feature of
discourse among college students during vicarious learning, even when compared to
interactive tutoring (Craig et al., 2000; Gholson & Craig, 2006; Graesser et al., 1996,
2004; King, 1989, 1994; Kintsch, Welsch, Schmalhofer, & Zimny, 1990; Otero &
Kintsch, 1992; Rosenshine et al., 1996). Moreover, what appears to be important is
the number of deep questions that are presented as part of the discourse, not whether
the questions are embedded as part of dialog or as part of monolog.
The next study Gholson et al. (in press) in the series explored the generality
of the deep questions effect in research with younger students and included a new
