Bound Variables and C-Command

Abstract

It has long been assumed in linguistics that bound variable interpretations of pronouns are possible (only) when a quantified expression c-commands the pronoun. In two studies in which readers’ eye movements were recorded, we examined the processing of pronouns bound by universal quantifiers. Experiment 1 compared examples where the quantifier c-commands the pronoun (‘Every British soldier thought he killed an enemy soldier’) with examples where it doesn’t (‘Every British soldier aimed and then he killed an enemy soldier’). Although there were no first pass differences, re-reading time showed that both quantifier examples took longer to read than nonquantified controls (‘The old British soldier. . . ’), but there was no special penalty in examples when the quantifier failed to c-command the pronoun. Experiment 2 investigated intersentential binding (telescoping): ‘John Frederick/Each executive/Every executive went home. He broiled a steak. He ate dinner. Then he watched television.’ Second pass and total reading times in the region containing the first pronoun were longer for quantified examples than name examples. But there was no indication that telescoping is tightly restricted, for example, to contexts with ‘each’ or to discourses describing stereotypical events composed of predictable subevents. The results suggest that bound variable interpretations are more generally and more readily available than is often assumed. They fit well with Bosch’s (1983) attempt to limit pronoun occurrences to just two types: anaphoric referential pronouns and syntactic agreement pronouns. On this view, ‘bound variable’ interpretations without c-command are really anaphoric pronouns with inferred antecedents. This view is discussed along with the challenges it faces.