Layman's Summary

In a sentence like Harvey likes either jazz or polka, linguists have traditionally assumed that the function of either and or is to appear at the left edges of the two words or phrases being linked by or (in this sentence, jazz and polka). But either sometimes seems to appear even further to the left, resulting in sentences like Either Harvey likes jazz or polka. Current syntactic theory gives us at least two different ways of analyzing this second kind of sentence:

1. Either is still associated with the position immediately to the left of jazz, but has moved even further to the left:
   Either Harvey likes either jazz or polka.
2. The sentence has undergone ellipsis, similar to what happens in sentences like Harvey likes jazz and Joe does like jazz too.:
   Either Harvey likes jazz or Harvey likes polka.

I argue that, in fact, we need both of these kinds of analysis because different constructions with either behave in different ways. Specifically, in full sentences like Either Harvey likes jazz or polka, either has moved. But when only noun phrases are involved, as in Harvey plays either [ the saxophone or accordion ], ellipsis is a better way to describe what's going on (as in Harvey plays [ either the saxophone or the accordion ]). There are several reasons for believing that these two types of construction are different; some of them are described in the UCSC Graduate Research Symposium handout below.

Summary

My syntax qualifying paper examines the properties of "discontinuous coordination" constructions such as both...and..., either...or..., and neither...nor.... In particular, I am interested in cases in which the first conjunct (both, either, or neither, what I call "C1") does not appear at the left edge of the first conjunct, about which there already exists a substantial literature. My own conclusion is that displacement of C1 is best modeled as two distinct phenomena:

1. When C1 appears to be displaced out of a DP, an ellipsis process along the lines of Conjunction Reduction has applied in the second conjunt.
2. Elsewhere, C1 surfaces freely, subject to a locality constraint on how far away from the left edge of the first conjunct it can appear.

I have also examined some of the semantic properties of these constructions and concluded that the "scope" of coordination is independent of the position of C1 and is best modeled with a Hamblin semantics. Common to all three of these mechanisms — long-distance dependency, ellipsis, and scope — is a sensitivity to syntactic configurations such as islands and sentential negation.

Downloadable Stuff

(See CV for citation information. Listed from newest to oldest.)

Qualifying Paper
2007 Syntactic sensitivities of movement, ellipsis, and scope: Evidence from discontinuous coordination.
The most complete write-up of this project to date.

UCSC Graduate Research Symposium Handout
2007 An improved model of either...or... constructions
A condensed version of the WCCFL and LASC presentations, intended for non-linguists.

Data from the Treebank Corpus
2007 Discontinuous coordinations in the Treebank corpus: Unbalanced coordinations and high/low scope
A summary of examples of discontinuous coordination from the Treebank corpus. Of the 1,283 examples found, 446 are presented: those in which C1 is displaced and those in which the scope of coordination demonstrably does not coincide with the position of C1.

WCCFL Proceedings Paper
2008 The proper role of movement and ellipsis in discontinuous coordination
The proceedings paper for the WCCFL talk.

WCCFL Handout
2007 The proper role of movement and ellipsis in discontinuous coordination
Argues for a distinction between C1-displacement in clauses (instances of a long-distance dependency) and DPs (instances of ellipsis).

LASC Handout
2007 The proper role of movement and ellipsis in discontinuous coordination
A slightly earlier version of the WCCFL presentation.