Jasp:
> > That said, there are pros and cons to nonterminal nodes. The cons, Dick
> > has pointed out in some old articles in _Linguistics_. The pros include
> > adjunct/complement ordering and handling what WG has to handle by
> > invoking semantic phrasing.
>
> We'll have to go through all of this (again?) at some point. What
> ordering properties are you referring to?
Complements are nearer to the head than adjuncts are (though with verbs
this generalization gets a bit compromised). This is as predicted by
standard PSG analyses of adjuncts. In WG it either must be stipulated,
or it must be made to somehow follow from semantic phrasing, which
would entail WG's semantic level constraining the syntactic level
with a potential clunkiness that I shudder to contemplate...
--And.
|