Abstract for Daniel Altshuler's talk

A non-Sequence of Tense approach to temporality in indirect reports: Evidence from Russian
Daniel Altshuler (Rutgers)
October 9, 2007 (Tuesday)
6:30 PM - 8:00 PM; Room 7102, The CUNY Graduate Center

It is generally held that an indirect report like "Abelard said that Heloise looked pale" has two readings: one corresponding to the direct report "Abelard said: 'Heloise looks pale'" (overlapping reading) and the other to "Abelard said: 'Heloise looked pale'" (past-shifted reading). In order to predict this ambiguity, many researchers have posited an optional syntactic Sequence of Tense ('SOT') rule, which renders the embedded past tense semantically vacuous (Abusch 1988, 1997; Stowell 1993, 1996; Ogihara 1995, 1996; von Stechow 1995). The SOT hypothesis is difficult to reconcile with current assumptions about syntactic theory: it is not clear how the SOT rule instantiates any independently motivated types of syntactic rules (though see Kratzer 1998; Schlenker 1999, 2003, 2004; von Stechow 2003). In this talk, I offer an alternative, semantic/pragmatic account, which does not require such a rule. Based on data from Russian I argue that an overlapping reading is possible when the following two conditions are met: (i) the reference time in the matrix and the embedded clause is the same and (ii) the embedded predicate describes a state which is not a consequent state (Webber 1988) of an event and which persists throughout the reference time (Hinrichs 1981, Partee 1984 and related work). When one of these conditions is not met, a past shifted reading is the only one available. The analysis put forward is that temporality in indirect reports is a context-sensitive phenomenon, governed by rules independently needed to account for narrative progression in discourse. On this semantic/pragmatic account, there is no need to stipulate any SOT rule in the syntax.