Jeff Runner

My talk will focus on a series of studies investigating the range of interpretations available to reflexives in "picture" noun phrases (PNPs) such as 'a picture of himself'. A standard analysis of PNP reflexives is to treat them as "exempt" from the Binding Theory (Pollard & Sag, 1992; Reinhart & Reuland, 1993), due to the nature of the picture noun's predicate argument structure. A second well-known analysis posits a null pronominal possessor in the PNP, which binds the post-nominal reflexive (Chomsky, 1986; Davies & Dubinsky, 2001). Both approaches account for examples in which no structurally more prominent antecedent is apparently available as a binder. The shared claim of these approaches is that it is the structure of PNPs themselves -- the predicate argument structure of the head picture noun, or the availability of the null pronominal possessor -- that is relevant for explaining the unusual behavior of PNP reflexives.

Another line of research has investigated what are sometimes called "statue" reflexive interpretations (Jackendoff, 1992; Lidz, 2001; Reuland, 2002), which arise in cases where the reflexive is a representation (e.g., a statue) of the antecedent (as in 'Ringo touched himself' in the context of a wax museum in which Ringo Starr touches a statue of Ringo Starr). Statue reflexives restrict in various ways what sort of linguistic antecedent they can accept; and cross-linguistically a number of languages have different reflexive elements that may or may not receive statue interpretations.

PNP reflexives and statue reflexives have certain similarities. For one, they both can be interpreted coreferentially, while true reflexives are typically interpreted only as bound variables. Secondly, a more obvious similarity is that both involve the notion of representation. This leads to the question of the possibility that they are the same thing. If reflexives in PNPs are actually a type of statue reflexive then approaches that blame their unusual behavior on particular structural characteristics (e.g., having a certain head noun like 'picture', or having a null pronominal possessor antecedent) will have to be reconsidered. I will present results from a series of experiments done in collaboration with Micah Goldwater that test this question directly.