In the present paper I will discuss a theory of discourse processing on the basis of the results of an eye movement experiment. A three-stage human information-processing model is proposed. The differences in processing syntactical and lexical information and the claim that it is the lexical information which is important for the comprehension of the text, are put in the light of the theory of discourse processing. It is shown that word classes important for the construction of propositions of the text tend to be fixated longer and regressed to more often than words not included in these propositions. The different independent variables on the fixation time are related to the different stages in the model, which gives further evidence for the model as well as for the processes in the different stages.
To download this paper, please return to Proceedings of the 1997 Postgraduate Conference