Parasitic participles: Evidence for how to Agree

Monday, 4 October 2010, Colloquium

Susi Wurmbrand, University of Connecticut

Abstract

In this talk, I investigate a series of Germanic verb constructions, which appear to involve the ‘wrong’ morphology on one or more of the verbs involved: Norwegian parasitic participles, Frisian upward and downward parasitic participles, and the German ‘Skandal’ construction, a construction where every verbal element appears to occur with the ‘wrong’ morphology (see (1)).

(1)

  1. Jeg hadde villet lest / lese boka. (Norwegian)
    I had want.PART read.PART / read.INF book.DEF
    ‘I would have liked to read the book.’ [Wiklund 2001 :201]
     

  2. hy soe it dien / dwaan wollen ha. (Frisian)
    he would it do. PART / do.INF want.PART have.INF
    ‘He would have liked to do it.’ [den Dikken and Hoekstra 1997: 1058]
     

  3. hy soe it dien ha wollen / wolle. (Frisian)
    he would it do.PART have.INF want.PART / want.INF
    ‘he would like to have it done.’ [den Dikken and Hoekstra 1997]
     

  4. ohne es verhindert / verhindern haben zu können (German)
    without it prevent.PART / prevent.INF have.INF to can.IPP
    ‘without having been able to prevent it’ [Vogel 2009]

I provide an explicit syntactic account unifying the phenomena and deriving the differences from independent differences among the languages (specifically word order and morphology). In doing so, I show that the distributional variation found among these constructions is fully in line with standard grammatical principles of the languages under consideration. The account I propose is based on a top-down definition of Agree, namely the claim that an unvalued feature is valued by the closest c-commanding element with the appropriate valued feature. I demonstrate that this view, in contrast to standard Agree and licensing under specifier/head agreement, allows for a uniform treatment of the morphological and syntactic properties of these constructions, which, so far, have been assumed to be unrelated. Lastly, this view of Agree will be shown to extend directly to other syntactic dependencies, such as Case assignment and multiple Agree(ment) relations in general.