Procedural Text: Predictions of Importance Ratings and Recall by Models of TextComprehension

摘要

Two models of text comprehension, a process model proposed by Kintsch and vanDijk (1978) and a causal model proposed by Trabasso and Sperry (1985), were tested in two experiments with eight procedural texts. In Experiment 1, 24 female college students rated the importance of propositions, idea units, and sentences to the overall procedure described in the texts on a 7-point scale. In Experiment 2, 16 female college students recalled each of the eight texts immediately after reading it. Predictors derived from the models were used to predict the ratings and the recall in multiple regression analyses. The results showed that the amount of variance accounted for by the predictors varied from text to text. In general, for the ratings and recall, the causal model accounted for more variance than the process model did. For the process model, level within the hierarchy accounted for the most various and for the causal model, the causal chain factor accounted for the most variance. The models were not as powerful as in previous research with narrative texts perhaps because these models do not take into account the cognitive representation involved in doing the task described by a procedural text.