Abstract
The purpose of the study was to examine the efficacy of self-explanation for helping elementary school students solve mathematical word problems through computer-based support over one year. Seventy-one fifth graders participated in the study as the training condition. The students solved worked-out examples for thirty minutes once a week in four three-week training sessions. Then they completed a word problem test. Sixty-two fifth-graders were used as the control condition who did not self-explain the solution steps. The results showed that all of the students gradually solved more word problems correctly than before. We classified students in the training condition into three groups according to the patterns of their test scores. Students in the upper group, who had consistently higher scores, self-explained their solution processes using inferences. Some students in the rising group, who were gradually increasing their scores, self-explained their solution processes using inferences. Most students in the lower group, who had consistently low scores, only took notes on the results of their past records about solution steps. Self-explanation as a metacognitive strategy is discussed.
