Information Technology Reference
In-Depth Information
whereas the high-knowledge students did not. Previewing training had no effects on
comprehension.
Discussion
Evaluations of both one-on-one SERT and group SERT found learning benefits for
students who received some form of self-explanation training. Unfortunately, both
of these SERT programs are limited to situations where an expert human teacher can
provide the training. Also, the trainers must be trained. The requirement of having
a trained human expert present to conduct the intervention means that SERT could
be expensive to implement and might also have a limited impact (as the largest
training group is 20-30 students at a time). It seemed unlikely at the time that this
intervention could be successfully carried out as a teacher-training program.
Additional problems emerge from the collaborative aspects of the group SERT
program. While the collaborative learning component to group SERT is an engag-
ing method of training, it can also lead to problems of poor feedback quality, no
feedback, and ineffective peer dyads. The group discussions heavily depend on the
dynamics of the group. This can be challenging for the best of high school teach-
ers. Moreover, creating an effective small group can be particularly difficult due to
a number of factors. Specifically, maturity, romantic interests, friendships, rivalries,
social survival, and many other social factors were observed to remove the focus
away from the learning task. These anecdotal observations in high school class-
rooms suggested that a one-on-one training may be more conducive to learning
within a high school environment, but this of course was entirely unrealistic if many
students were to receive training. Ideally, a hybrid version of these two SERT pro-
grams would appear to be the most effective. That is, the ideal implementation of
SERT would instruct students on an individual basis and also provide exposure to a
larger number of students.
iSTART
This technology seemed to provide the best solution to the dilemma of how to pro-
vide SERT to a wide audience, and still provide tailored training. At the time that
the initial one-on-one SERT evaluation projects were coming to a close, there was
a growing recognition of algorithms such as Latent Semantic Analysis (LSA) to
recognize the meaning of text (Hu, Graesser, & TRG, 1998; Landauer, Foltz, &
Laham, 1998). Such algorithms were necessary in order to build a tutoring sys-
tem that could interpret and respond to textual input such as self-explanations
(Graesser, Wiemer-Hastings, Wiemer-Hastings, Harter, Person, & TRG, 2000;
Wiemer-Hastings, Wiemer-Hastings, & Graesser, 1999).
Thus, iSTART was developed to instantiate SERT training within an ITS.
iSTART is a web-based version of the SERT program that can potentially be
