While I'm at it, here's another new paper of mine, this one in collaboration with Tanja Temmerman. It starts out from a seemingly crazy idea originally proposed by Kyle Johnson in a paper in Lingua, namely that two elements that for all intents and purposes look to be non-adjacent, can nonetheless be considered adjacent under a multidominant analysis. I was quite skeptical of the idea at first—witness footnote 22 in Johnson's paper—but as Tanja and I discovered, it allows you to have your cake and eat it too when it comes to negative indefinites such as no car. As many peope have pointed out, negative indefinites seem to be semantically—and in some cases morphologically—composite in that they contain both negation and an existential. Accordingly, some of the earliest analyses of the phenomenon assumed a kind of fusion or amalgamation approach, whereby clausal negation and the indefinite determiner of the direct object fuse into a single negative determiner no. However, while that works fine for an OV-language like Dutch, where negation linearly precedes the indefinite determiner of the object, it goes awry in English, where the verb intervenes between clausal negation and the direct object (John did not eat a cookie.).
Enter Johnson's idea, and all of a sudden, negation and direct object can be adjacent even in a language like English. What's interesting about this approach—and this is actually the main topic of our paper—is that it correctly predicts negative indefinites to interact with ellipsis in ways that are unexpected under a movement- or Agree-based approach. The paper is relatively technical—and some of the trees rather funky-looking—but the general message I think is clear and quite thought-provoking: being close to one another does not necessarily mean the same thing for morphosyntax as it does for Spell-Out. In other words, Metallica had it right all along.