Reconciling Austinian and Russellian accounts of the Liar paradox

Introducing The Liar Barwise and Etchemendy's The Liar presents two accounts of the semantics of the liar sentence \this sentence is not true", named for Russell and Austin, and informed by situation semantics and non-well-founded set theory. The accounts diier over how a speaker uses a sentence and the world to express a proposition. \By a statement we will understand certain sorts of datable events, those where a speaker asserts or attempts to assert something using a declarative sentence. In contrast, we take a proposition to be a claim about the world, the kind of thing that is asserted by a successful statement." (The Liar, page 11) Under the Russellian account, \sentences are used to express propositions, claims about the world, and these claims are true just in case the world is as it is claimed to be." (The Liar, page 26) Thus, a Russellian statement is an utterance of a sentence, the proposition expressed by that statement is a claim about the world, and that proposition is true ii the claim holds. Under the Austinian account, \a legitimate statement A provides two things: a historical (or actual) situation s A , and a type of situation T A. The former is just some limited portion of the real world; the speaker refers to it with what Austin in his paper \Truth"] calls \demonstrative conventions." The latter is, roughly speaking, a property of situations determined from the statement by means of \descriptive California under a postdoctoral fellowship provided by the Systems Development Foundation. I am deeply indebted to Peter Aczel, John Barwise and an anonymous referee for invaluable comments on earlier drafts of this paper.