Research Group: Attitudes, Concepts and Objects

Prag: A Model for Processing Conditional Utterances Lázló Nagy Issue. Although the terms reasoning and pragmatics cover two vast areas of research in psychology and linguistics, their interdependence has only been the subject of occasional explorations. More specifically, current theories of reasoning often incorporate some "independently motivated pragmatic comprehension processes" (Braine et al., 1995). One of the dominant concepts of utterance comprehension is Grice's (1975, 1989) conversational implicature. Grice argues that hearers frequently infer more than the literal meaning of an utterance, and that these inferences show distinctive properties.

In GrKK since 10/94 as a Ph.D. student

Goal: The project has two main purposes. First, to test experimentally whether these 'Gricean properties' distinguish implicatures from other inferences. Second, to specify a model of pragmatic comprehension that defines pragmatics in a cognitive framework.

Method: Conditional phrases like promises (e.g. "If you allocate the resources, our sales will double") and threats ("If you drive fast, you will pay a fine") were chosen as the domain of investigation. Conditionals appear to be appropriate as a starting point, because their conversational implicature is well-documented and predictable. The basic material, including 12 context stories, has been evaluated in a questionnaire study using ranking scales. Then, the predicted properties were addressed in in three computer based experiments, considering both frequencies and reaction times as relevant data.

Results: As to the first aim, the data supply only very limited evidence for the Gricean predictions. However, the cognitive model named Prag can be summarized in this way: Pragmatic interpretation starts immediately after the utterance recognition phase is completed. The addressee builds up a speaker intention representation. This involves conversion from linguistic surface structure to a more abstract, propositional format. Then the individual utterance will be mapped on a speech act category. This pre-processing speeds up a contingency evaluation, where the addressee has to check which possible events or actions are compatible with the utterance, and then prepare for action. This model may be seen as an addendum to established theories of reasoning, as a component that tackles the early phase of discourse comprehension rather than subsequent, more complex inferences.