Modelling Reading Times in Different Reading Tasks with a Simulation Model of Comprehension.

摘要

Reading times are a very detailed source of data on comprehension processes, especially in the case of recording reading times on individual sentences in a passage. But conventional verbal or mathematical theories of comprehension fail to fully exploit such data because the comprehension processes are not represented in enough detail. Computer simulation models of language processing are adequately detailed, but there has been difficulty in applying such models to experimental data. This paper presents the results obtained by using a multiple-regression method to compare an ordinary language processing simulation model to single-sentence reading times by fitting a linear equation to the reading times. The candidate predictor variables in the regression analysis are variables representing the number of steps in each process performed by the simulation in understanding each sentence.