Gerunds (was Re: Filler-gap mismatches)

[Malouf R.]

Hi,

 > This
 > is, of course, a very different analysis from Malouf's "fine-grained
 > category" analysis. The analyses differ *empirically*, in terms of what
 > Malouf dubs lexical coherence (predicted by Malouf's analysis) vs. phrasal
 > coherence (predicted by Bresnan's analysis).

Just to clarify this point: the argument in my dissertation is that mixed
category constructions are universally characterized by lexical coherence.
Part of the evidence for this is the existence of mixed categories (such as
the Arabic masdar) which show lexical coherence but do not show phrasal
coherence. Now, I certainly wouldn't want to claim that no mixed category
construction shows phrasal coherence. On the contrary, the vast majority of
such constructs show both kinds of coherence, which is what makes it so
difficult to tease the two apart.  But, my claim is that the effects of
phrasal coherence can be explained through the interaction between lexical
coherence and other independently motivated constraints in the grammar.

What I do predict is that you won't find any mixed category constructions
which do not show lexical coherence, a generalization which, as far as I know,
seems to hold true. As I understand it, the constructions which Bresnan (2001)
discusses are all cases which show both lexical and phrasal coherence, and so
don't really constitute counterexamples to my analysis.

Since both approaches to mixed categories make largely the same predictions
for those constructions which show both kinds of coherence, the empirical
differences between the two analyses (if any) will be pretty subtle. However,
the existence of constructions like the Arabic masdar would seem to pose a
problem for the LFG approach. Or have I misunderstood something?

--
Rob Malouf
malouf at let.rug.nl