|
Abstract: We might profitably compare interpretation in discourse to
the solution of a simultaneous equation in multiple variables. In a
given utterance, there may be one or more anaphoric expressions or
ellipses whose interpretation must be resolved, and one or more
quantificational operators whose intended domain must be determined. I
sketch how a general Gricean principle of Retrievability guides the
resolution of all these types of variables and other kinds of
context-sensitivity as well, taking into account requirements of
relevance and salience which reflect the intentional structure of
discourse generally assumed by interlocutors. In particular, relevance
to an understood Question under Discussion constrains the search space
for anaphora resolution and domain restriction in a thorough-going
way, offering the foundation for a theory of salience. Moreover, the
interaction between conventional content and context can only be
captured in a dynamic model in which the formally modeled context
changes in the course of compositional interpretation. This
characterization of the role of context in interpretation stands in
contrast to a commonly assumed model which treats interpretation
(processing and comprehension) as a sequential affair, in which first
we parse, then resolve a few indexicals in the course of
compositionally determining the semantic content of the utterance, and
finally put Gricean icing on the propositional cake. But the proposed
model is also more constrained than some other models of
pragmatically-guided processing, including those of Sperber & Wilson
(1985) and those based principally on rhetorical relations. I review
experimental evidence from the psycholinguistic literature on both
processing and acquisition which suggests that the
intention-structured model of discourse is psychologically plausible
and explanatory, and that it reveals important features of the
interface between purely linguistic competence and more general
cognitive processes, including practical reasoning, and information
storage and retrieval.
|