ABSTRACT
Understanding one's native language is something that comes so naturally to people that it can be difficult to stand back from this process and characterize it properly, in order to study it scientifically. Similarly, the mechanism underlying our ability to recognize everyday objects is something we do not usually worry about in our ordinary lives. Partly for this reason, Marr (1982) suggested that the study of a cognitive system should start with a task analysis: What does the system do and why? As we have already suggested, in comprehending discourse, readers and listeners are typically aiming to construct a representation of a situation, or a set of ideas, that the discourse is about. This claim leads directly to the notion of a mental model, a mental representation of part of the real or imaginary world (see, e.g., Garnham, 1987; Johnson-Laird, 1983). The structure of such a model should parallel the structure of the part of the world it represents, and not the structure of the text in which it is described or the structure of the sentences of that text. Thus, a mental model is different from the kind of propositional representation of text meaning that was postulated in the early 1970s (e.g., Kintsch, 1974). Johnson-Laird (1983) distinguished between a propositional representation of a text as being "close to the surface form of the sentence" and "a mental model whose structure is analogous to the state of affairs described by the discourse" p. 244).
