- Intermediate Adjunction with A-Movement
It is widely assumed—often tacitly—that A-movement does not move through intermediate positions where it does not check morphological features (e.g., Baltin 2001). At the same time, Chomsky (2001), Fox (1999), Nissenbaum (2000), and others argue that Ā-movement must adjoin at least to every vP and CP on its path and leave an interpreted trace in each of these positions. In this squib, I present an empirical argument that A-movement as well must move through an intermediate position adjoined to vP. I then show that my discovery bears on a difference between Chomsky's (2001) theory of phases and Nissenbaum's (2000) version of it, corroborating Nissenbaum's proposal.
A well-studied case of A-movement is raising in English. With Chomsky (1995) and others, I assume that raising is movement of the subject from a position in the infinitival complement to the matrix Spec,TP position to satisfy the Extended Projection Principle (EPP). This is illustrated in (1).
Next, consider the interpretation of (2), where the universal quantifier is in the scope of negation. This interpretation can be paraphrased as Not every child is smart. It requires a special intonation with a rise on every and a fall on isn't, and is most natural if the sentence is followed by a clarifying continuation like In fact, half of them aren't smart (e.g., Jackendoff 1972, Büring 1997).
(2) [Every child]1 isn't t1 smart.
I assume with McCloskey (1997) that this interpretation is derived by total reconstruction of the subject to a position lower than negation. [End Page 308] One alternative to this assumption worth considering is that (2) allows covert raising of not to a position above the subject. However, two arguments speak against this analysis.
The first argument is that movement of negation actually would not alter the scope of negation on what seem to be the most straightforward assumptions about how movement is interpreted. This argument is based on the following consideration: Assume that movement of negation leaves behind an interpreted trace, as other movement processes have been argued to (e.g., Fox 1999, Sauerland 2000). Because negation is not quantificational, it could only leave behind a trace of the same semantic type as itself. But then, negation would actually be semantically reconstructed to its trace position in the interpretation process. Hence, movement of negation would not affect the scope of negation unless it was assumed not to leave behind any trace.
The second argument is based on the scope of negation relative to material occurring between the subject and negation. Both examples in (3) have only one interpretation, where negation takes scope below the quantificational adverb usually and the modal must, respectively. Unlike in (2), even with a special intonation and a clarifying continuation, the scope of negation is fixed, and in fact the clarifying continuations in (3) seem contradictory.
a. Jan mustn't get an A. (#In fact, he could get an A or a B.)
b. Tom usually doesn't follow. (#In fact, half the time he doesn't follow.)
It seems that in general, all adverbs and several modals (must, ought to, may) in English must take scope above negation when they occur between the subject and negation.1
Now consider examples where the subject is again a universal quantifier. (4) is an example with a modal.
(4) Every student mustn't get an A. At most a third of them can get one.
Here, the subject is able to take scope below negation. This reading is brought out by the following scenario: A junior teacher gave every student in his class an A. However, the school has a rule that only a third of all students may get an A, to prevent grade inflation. A senior teacher could then use (4) to reprimand his junior colleague. As in (2), this interpretation is dependent on a particular intonation contour with a fall on every, a rise on the negation, and destressing of the [End Page 309] material between the two. Furthermore, the clarifying continuations given in (3) helpin accessing this interpretation. In (4), even...