Component processes in the comprehension of simple prose

Abstract

The comprehension process is generally viewed as consisting of several component subprocesses. These processes were presented in the form of a conventional process model, and then characterized more explicitly by a simulation model. The model was used to account for the reading times of individual sentences in simple passages obtained in a free reading task, a topic choice task, and a recall task. Multiple regression was used to obtain predicted reading times from the number of processing steps performed in each process of the simulation model. Additional variables were defined to account for task differences. The model accounted for about 96% of the variance among mean reading times. The fit between the model and the data was excellent except for incoherent passages in the recall task. The results of the regression analysis indicate how much time is contributed by each component process to the total reading time, and suggest that the reading tasks differ in the presence or absence of task-specific component processes.