Metacognitive Awareness versus Linguistic Politeness: Expressions of Confusion in Tutorial Dialogues

Metacognitive Awareness versus Linguistic Politeness: Expressions of Confusion in Tutorial Dialogues Gwendolyn E. Campbell (Gwendolyn.Campbell@navy.mil) Naval Air Warfare Center TSD, Code 4651 12350 Research Parkway, Orlando, FL 32826-3275 Natalie B. Steinhauser (Natalie.Steinhauser@navy.mil) Naval Air Warfare Center TSD, Code 4651 12350 Research Parkway, Orlando, FL 32826-3275 Myroslava Dzikovska (M.Dzikovska@ed.ac.uk), Johanna D. Moore (J.Moore@ed.ac.uk), Charles B. Callaway (Charles.Callaway@ed.ac.uk) & Elaine Farrow (Elaine.Farrow@ed.ac.uk) Human Communication Research Centre, University of Edinburgh 2 Buccleuch Place, Edinburgh EH8 9LW, United Kingdom Abstract Research suggests that students who are aware of their own confusions and take steps to resolve those confusions are most likely to benefit from a learning experience. At the same time, there are conversational maxims, such as Leech’s politeness maxims, that may inhibit a student from expressing and pursuing confusions within a tutorial dialogue. We investigated students’ expressions of confusion while working through a series of learning activities with a tutor. We found that, during the times when students were working independently on an activity, their expressions of confusion were reliable indicators of their (lack of) understanding; however, when they were conversing with their tutors, these same students did not express confusion and, in fact, the more often the expressed comprehension, the worse they performed on the post-test. This suggests that student metacognitive statements should not be interpreted without taking into consideration the context in which they were expressed. We briefly consider implications for human tutors and the development of computer tutoring systems. Introduction Twenty years ago, researchers were somewhat surprised to discover that students who expressed confusion while studying worked examples were more likely to learn from that activity than students who appeared to easily understand those examples (Chi & Bassok, 1989). The key is in the qualifier “appeared” to understand. Further analyses and additional research confirmed that the students who expressed confusion often followed up by taking steps to resolve that confusion, while many of the students who breezed through a worked example were simply not aware of the fact that they didn’t understand why each step was taken and how that process might be applied to related but different problems (e.g., Ferguson-Hessler & de Jong, 1990; Pirolli & Recker, 1994; Renkl, 1997). In other words, accurate metacognition was a necessary precursor to taking remedial action. This work has led to a number of efforts, both in the classroom and within computer- based tutoring systems, to understand and train metacognitive skills such as self-explanation (e.g., Bielaczyc, Pirolli & Brown, 1995; Chi, de Leeuw, Chiu & LaVancher, 1994; Conati & Van Lehn, 2000; Renkl, Stark, Gruber & Mandl, 1998). Robust findings include the fact that many students do not spontaneously self- explain, but that they can learn this skill and will benefit from this type of training. Of course, studying a worked example independently is not the same task as working through lesson materials with a tutor. In the second environment, it is possible that