top of page

Non-locality in clausal structure: Extended Projections

Thursdays 11.15-12.45, Seminarraum H1 5.16, GWZ


An extended projection (Grimshaw 2000, among others) is a chunk of structure that consists of a sequence of functional categories projected above a lexical category. E.g. the extended projection of an N would be the DP or perhaps the PP, and the extended projection of the V would be the CP, involving functional projections like I and C, and potentially others. Extended projections are a way to deal with patterns that are nearly local, but involve a certain inherent kind of non-locality, and thus cannot straightforwardly be modeled in terms of selection. For example there is a rigid sequence of possible auxiliary verbs in English: modal above perfect 'have' above progressive 'be' above passive 'be'. We might try to capture this by saying that the modal selects the perfect, which selects the progressive, which selects the passive, which selects the main verb. But this won't work, because any one of these auxiliaries can be omitted without disturbing the rest of the sequence. Unless we want to say that e.g. even non-perfect clauses involve a perfect head, we need a mechanism which can state the ordering of these heads, but which is not directly stated on the heads themselves. What is the basic spine of a clause or a nominal phrase, what parts are obligatory, and what order to they come in? This is the sort of thing that devices like extended projections, Adger's Hierarchy of Projections (Adger 2003 et seq.), or the functional sequence/"fseq" (Starke 2011) of Nanosyntax have been proposed for.

Some of the questions we will look at are:


  •  What are the structural wellformedness conditions on an extended projection? (how big can they be, how small can they be, where can there be "gaps" in the sequence, and how do we decide)?

  • Relatedly, what are the different formal ways to implement this basic idea: e.g. Grimshaw's original extended projection, Adger's Hierarchy of projections, a cartographic functional sequence/"fseq"? -- and how do we empirically decide amongst them?

  •  What is the connection between an extended projection and a phase? (e.g. a CP is considered to constitute a single extended projection but consists of two phases -- vP and CP).

  • How much crosslinguistic uniformity/variation do we see in the structure and inventory of extended projections? 



Course Syllabus


































bottom of page