Using Kintsch's Computational Model to Improve Instructional Text: Effects of Repairing Inference Calls on Recall and Cognitive Structures

The goal of this study was to link a computational psychological model to instructional practice. Kintsch's ( Kintsch & van Dijk, 1978 ; Miller & Kintsch, 1980 ) reading comprehension model was used to identify locations where inferences were called for in a 1000-word expository text. Then each location was repaired to produce a principled revision. In an experiment with undergraduates ( N = 170) , free recall of the principled revision was much increased over that of the original version. Also, the author of the original text and 7 subject-matter experts provided measures of the shape of the original text's intended cognitive structure. The author's and experts' cognitive structure shapes correlated above .5 with the shapes provided by U.S. Air Force recruits who read the principled revision but only .1 with recruits who read the original version. Apparently, the principled revision carried out the author's intentions better than had the author's original text. It is concluded that Kintsch's model can be used to improve instructional text.