DP Is Not a Scope Island

1 The Issue

This squib is concerned with inverse linking constructions. May (1977) introduced this term for cases where a quantificational DP occurs inside another DP and takes wider scope than the containing DP. The construction is illustrated in (1), where every linguist is an argument of book, but takes wider scope than one book by t on the most salient interpretation.

(1) Tom read [_cQP one book by [_iQP every linguist]].

In the following, I use the terms iQP and cQP for the inversely linked, contained QP and the containing QP, respectively. My use of the two terms is indicated in (1). For concreteness, I furthermore assume that syntactic movement, specifically Quantifier Raising (QR), is the only scope-changing mechanism.

May (1977) proposed that in inverse linking, the iQP (every linguist in (1)) undergoes QR to take clausal scope at VP (in the framework May was assuming, the iQP actually took scope at S). This LF structure is shown in (2).

(2)

Today, however, a different analysis of inverse linking is almost universally accepted. May himself proposes, in his 1985 book, that DP is an island for QR. The inversely linked interpretation of (1) is, on this proposal, derived from the structure in (3).

(3)

Rooth (1985:117-118) and Larson (1985b) propose the same analysis of inverse linking, though with a different scope-taking mechanism. The DP-adjunction analysis is also adopted by and Kratzer (1998:232-233) and, with a different scope-taking mechanism, by Barker (2002). May and Bale (2002) provide an accessible discussion of the analyses by May (1985) and Larson (1985b). Representation (3) requires some version of type-shifting to be interpretable since the sister of the iQP is of type 〈e, 〈〈e, t〉, t〉〉 while iQP itself is of type 〈〈e, t〉, t〉. (Rooth and Heim and Kratzer provide concrete implementations of the type shift.)

In this squib, I argue that May's (1977) TP-adjunction analysis of inverse linking is required, which entails that DP cannot be an island for QR. Specifically, I show that the iQP can take scope separate from the cQP when scope relative to a scope-taking verb or scope relative to negation is considered. I look at these two phenomena in sections 2 and 3. In section 4, I discuss Larson's (1985b) finding concerning the scope of iQP and cQP relative to a third QP that is the main argument for the adjunction-to-DP analysis.

2 Intensional Verbs

The study of examples containing three quantificational elements requires great care. I first introduce the tests for wide and narrow scope of a QP relative to the verb want and then apply them to inverse linking. It is well known that indefinites provide a good test for narrow scope relative to an intensional verb. Consider (4).

(4) John wants to marry someone from Spain.

(4) has two distinct scopal construals. On one reading (sometimes called the de dicto reading), marrying any person from Spain would satisfy John's desires. This construal allows (4) to be true in a situation where John does not know anybody from Spain. For this reading, the indefinite someone from Spain must take scope below want, since there is no single person such that John holds the desire to meet this person specifically. The second reading holds in a situation where there's a Spanish person such that John holds the desire to marry him or her. This reading (sometimes called the de re reading) arises if the indefinite someone from Spain takes scope above want.

Work on indefinites has proposed that indefinites do not need to undergo QR to acquire wide scope (e.g., Fodor and Sag 1982, Ruys 1993, Reinhart 1997). Therefore, the second reading of (4) could also arise from an LF structure where someone from Spain remains in the c-domain of want. Examples like (4), however, successfully argue for narrow syntactic scope of the indefinite: for the first reading to be available, the indefinite someone must be syntactically represented in the c-domain of want. In what follows, I use indefinites to test for...