Nonrestrictive modifiers in non-parenthetical positions Introduction

The systematic but often subtle semantic differences between prenominal and postnominal adjectives first noted by Bolinger (1967) in many respects remain poorly understood. There remains a similar murkiness surrounding some of the systematic but often subtle semantic differences between preverbal and postverbal adverbs, of the sort noted by Jackendoff (1972), Bellert (1977), Ernst (1984, 2002), and Cinque (1994), among many others. This chapter focuses on one difference of this sort that occurs in both these murky domains: for both adjectives and adverbs, nonrestrictive interpretations are possible without resort to parenthetical intonation only in pre-head positions.¹

The proposal is to derive this striking parallel from a broader principle governing how nonrestrictive interpretations are built up. More precisely, I will suggest that the semantic mechanism that gives rise to these interpretations – understood here more or less in the framework of Potts (2005) – is characterized by a fundamental structural asymmetry that prevents it from assigning such interpretations to constituents on right branches. This is in one respect a surprising proposal: linear precedence normally has no effect on semantic interpretation, and it’s not altogether clear that information about linear precedence should be present at LF at all. What this may reveal is that such nonrestrictive, non-truth-conditional meaning is fundamentally quite different from ordinary meaning. One of the broader questions that underlie the proposal here, then, is how and where truth-conditional and nonrestrictive meaning interact. The other broader question that will frame the discussion is the general empirical one of how modifier position and interpretation relate.

1 The term ‘nonrestrictive’ does not have a precise self-evident definition. I use it here essentially as a convenient label for the problem, in part because it is widespread in the literature