Upper limit effects in DPs and the interdependence of worlds and times

Abstract

The distribution of upper limit effects suggests that an expression’s evaluation time cannot be bound independently from its evaluation world. Specifically, even in cases where worlds and times can be non-locally bound (which implies the availability of mechanisms with object-language quantification or equivalent power), both coordinates must be bound by the same (non-local) antecedent. From this, I conclude that the mechanism manipulating worlds and times in natural language is less expressive than a mechanism that manipulates independent object-language variables that range over worlds and times. At most, the mechanism that binds worlds and times can express binding of evaluation indices (i.e. world-time pairs).