Abstract
This chapter gives a detailed survey of the semantics of free relatives. After providing a brief characterization and typology of free relatives, the chapter reviews the arguments that have played a role in the debate about the semantic nature of free relatives.
The arguments clearly support the traditional view that free relatives correspond to definite descriptions. At the same time, however, there is a class of ever free relatives, called here universal-like ever free relatives, which match the behavior of universal quantifiers.
The chapter then turns to a discussion of so-called ever free relatives and particularly the modal inferences they convey - such as indifference or ignorance. A brief crosslinguistic survey is included, suggesting that ever free relatives are primarily non-modal, which in turn lends support to the approaches that attribute the potential modal inferences to a source external to the ever free relative itself.
The final part of the chapter is devoted to how the semantics of free relatives can be derived compositionally.