Against ambiguity and underspecification: evidence from presupposition as anaphora

Abstract

I propose a critical assessment of both the underspecification (sense-generality) and ambiguity positions. I argue that within the frameworks that espouse the dynamic approach to meaning construction, a non-ambiguous interpretation ensues. My Default Semantics that merges the logical form with the output of the intentionality of the corresponding mental states [Default semantics, pragmatics and intentions in Turner, ed. (1999); Disourse, Beliefs and Intentions (1999); The default-based context-dependence of belief reports in Jaszczolt, ed. (2000)] demonstrates that the concept of underspecification is redundant in the case of definite descriptions. In the case of apparently semantically ambiguous referring expressions, the speaker's intentions interact with the logical form of the sentence and produce a unique propositional representation. The problem of ambiguity does not arise because intentions ‘intrude’ into the semantic representation. The view is compatible with pragmatic intrusionism of dynamic-semantic approaches such as Discourse Representation Theory. In this paper I present an argument against underspecification that comes from the approach to presupposition as anaphora by van der Sandt [J. Semantics (1992)] as supplemented by Krahmer and van Deemter [J. Semantics (1998)]. Van der Sandt suggests that presuppositional expressions are anaphoric expressions. When accounted for in terms of a dynamic semantic theory, they exhibit the capacity to bind to an antecedent or to accommodate in the relevant context, and these processes are aided by pragmatic factors. I propose an amendment to van der Sandt's and Krahmer and van Deemter's approach by demonstrating that where multiple representations are viable, they are ordered on the scale of preference that does not allow for ambiguities. I replace van der Sandt's admittance of some degree of ambiguity of presuppositional anaphors with a scale of salience of possible anaphors, ranging from the default to the most unlikely one. Underspecification is not denied its theoretical status: if the theoretical discussion focuses on the theoretical question as to to what extent sentences have to be disambiguated for logical reasoning to proceed [van Deemter, J. Semantics (1998)], distinguishing the stage of underspecification is justified. Instead, I question the need for distinguishing underspecified representation as a separate stage in utterance interpretation, at least in the fragments of English that I have tested so far.