On the role of semantics in a theory of adverb syntax

Abstract

The recent literature on adverb positioning is roughly divided between two theoretical camps, one of which claims that adverbs are licensed in spec positions by empty functional heads (Cinque, 1999), the other taking adverbs as adjoined and licensed largely by semantically based principles (Ernst, 2002). This paper defends the latter view, presenting two sets of evidence. First, the syntactic approach invoking functional heads holds that certain aspects of an adverb's meaning are encoded in the functional heads, such as the different scopes of frequency adverbs, e.g. in Texans often drink beer versus Texans drink beer often. It is shown that this approach to adverb meaning leads to significant redundancy and great difficulties in representing adverb semantics in a coherent way. Second, while the functional-head theory does account for the normal order of subject-oriented adverbs and negation ( She probably didn’t leave versus *She didn’t probably leave), the existence of exceptional cases where adverbs may follow negation requires this theory to adopt a semantically based explanation (invoking the adverbs’ status as positive polarity items) in any case. As a result, its original account of subject-oriented adverbs is redundant. For both sets of evidence, the semantically based theory handles the data easily: scope and related patterns are predicted by its basic semantic principles, and the polarity explanation for the second set also fits naturally into this framework without redundancy. Thus, the two phenomena provide evidence for the semantically based theory.