Syntax Word order and restrictive vs. nonrestrictive modification

When one begins to look in detail at adverb-ordering facts , a further issue arises concerning the interaction of adverb syntax and semantics. Even if both examples in (1) have a manner interpretation, (1a) has a reading that (1b) lacks: (1b) can only be true if the manner of dodging the question is part of what caused the annoyance, while that is not a requirement for the truth of (at least one reading of) (1a).

Marcin Morzycki’s contribution to this volume accounts for the contrast in (1) by proposing that the adverb in (1a) can function as a nonrestrictive modifier and thus does not contribute to the “at issue” content of the clause (in this he builds on Potts 2005; see below), while the adverb in (1b) serves as a restrictive modifier.

The restrictive/nonrestrictive distinction, more familiar from the adjective domain, is most easily perceived when the modified expression is known to have a unique referent (for example, when it is a proper name); however, such modification is very clearly marked syntactically in some languages, such as Spanish, where non focused, prenominal modification by intersective adjectives is always interpreted non restrictively (see Demonte, this volume). Contrast the unacceptable postnominal adjectival in cases where, as in (2b), the modification is intended to be nonrestrictive.¹

The semantics and pragmatics of nonrestrictive adjectival modification remains a topic of debate. Potts (2005) argues for a multidimensional semantics which distinguishes “at issue” content from conventionally implicated content, with nonrestrictive modifiers falling into the latter category. In contrast, Macia (2002) and Schlenker (2007) defend the position that the kind of nonrestrictive modification illustrated in (2a) (so-called “expressives”) is simply a special type of presuppositional phenomenon. Either way, however, the analysis involves positing a special relationship between syntactic structure and semantic interpretation.

1 (2b) is acceptable in a context where we distinguish between two Juana Garcias via restrictive modification by the adjective.