Nik:
> I am in the middle of moving, so have only read the current spate of emails
> hastily - if I am repeating something already said, forgive me.
>
> Isn't there a difference between a CP/IP view of syntactic structure and the
> kind of constructional fact that Dick drew our attention to? I.e.
>
> 1. A top down view of syntax: S-->NP,VP
> 2. A bottom up view: 'a word defines the company it keeps'
> 3. A problem to do with the status of constructions.
>
> Both theories of grammar in (1) and (2) can accommodate constructions
> - early TG
> certainly did; and both can eschew constructions, as GB did (and the WG of my
> thesis was anti-constructional, too).
>
> And one of the interesting things about the CP/IP view of syntax that
> it claims
> that syntax has the structure indicated in (2), but it still gives clauses/
> sentences a definition independent of the words in them.
>
> Contemporary minimalism seems to be pretty constructional, in fact. E.g. the
> exploitation of ForceP, TopicP and FocusP in Rizzi's (1997) left
> periphery paper
> gives a characterisation of certain syntax/meaning facts which looks like the
> standard definition of a construction.
I don't know about CP/IP, but ForceP, TopicP etc. can't be projected
from the lexicon, so must be part of a prefabricationist grammar.
But if (2) is revised to 'a node defines the company it keeps', so
as to cover nodes that don't correspond to anything from the lexicon,
then I think (2) is sustainable.
On a slight tangent, it seems to me that nodes like FocusP -- i.e.
head nodes for which there is no associated word -- are tantamount
to exocentricity. I'm curious how it can be alleged that the
principle of endocentricity is preserved.
(One of these years, I hope to write up a summary of arguments from
English in favour of exocentricity. Basically the idea is that once
you have discarded X-bar structure in favour of DG, the burning
need for a principle of endocentricity evaporates, and nonterminal
nodes in general and exocentricity in particular can be (re)admitted
but put to much better use than in X-bar and pre-X-bar days.)
--And.
|