Footnotes

 

(1)

I am grateful to Joan Bresnan, Peter Sells and Andrew Spencer for discussion of the issues in this paper - all errors are my responsibility. I began working on these issues while a Visiting Scholar in Cognitive Science at CSLI during Autumn 1996, am opportunity for which I am very grateful.  

(2)

These aspects set the Welsh possessive apart from the Semitic Construct State construction.  

(3)

A host of papers show how head movement analyses can be recast in LFG: see [Kroeger 1995], [King 1995] and [Bresnan 1997a], among others.  

(4)

It may be that the possessor is in fact structurally the specifier of DP: we briefly consider this alternative analysis below, and for present purposes nothing hangs on this distinction.  

(5)

I assume that POSS is a lexically selected function, introduced in semantic forms, but alternatively it may be the case that POSS is a specialisation of the ADJ function itself.  

(6)

Interestingly, [Alexiadou and Stavrou 1998] apparently argue, within the Minimalist/PPT framework and independently of the present paper, that noun movement accounts based solely on word order considerations are undermotivated and therefore problematic. I am grateful to Andrew Spencer for bringing this to my attention.  

(7)

The relationship bewteen semantic role and syntactic complement for nominal dependents has been much less well studies than for verbal dependents: nonetheless the claim that we are making is not without precedent in the literature: see [Zucchi 1993][Grimshaw 1990] for some discussion.  

(8)

[Rouveret 1994], citing grammatical judgements from Gwen Awbery) provides the following examples:

 

(9)

[Rouveret 1994] is forced to assume the existence of a rightward specifier position as well as the leftward specifier position in which NP possessives appear. Accounting for the NP/PP_gan alternation in the leftward specifier position requires the postulation of an invisible Genitive case on the possessive NP.  

(10)

Notice that the constraint is required, since (exist ^DEF) can be satified by a definite possessor, as we will see shortly.  

(11)

For expository purposes, I use the label Num as a categorial shorthand, without commitment to categorial status (as Adj or N) of numerals.