And Rosta wrote:
> 
> > Or take an even more relevant example: I recall reading about an
> > experiment that induced tip-of-the-tounge (TOT) states in speakers
> > of Italian. The task was to name words on the basis of definitions.
> > The experimenters gave definitions of uncommon words, which led
> > to a TOT state. Subjects could often access grammatical gender
> > even if they couldn't access the phonology.
> 
> That shows either that there's a direct route from the sense concept to
> gender or that there's a route from the sense concept to the lexeme
> and further links from the lexeme to the gender and to the phonology.
> The former scenario sounds to be the one you're envisaging. The latter
> would be the WG scenario.

There's a third logical scenario: both routes are activated. Using a
little ASCII art, the diagram would look something like this, I think:

                 CONCEPT
                /   |   \
               /    |    \
          LEXEME----|---GRAM. GENDER
               \    |    /
                \   |   /
                PHONOLOGY

 
> > > > An opposite type of case is gapping:
> > > >
> > > > (3) Noam has married Carol, and me, Carmina.
> > > >
> > > > It's obvious that the second conjunct has the sense of "has
> > > > married," but it can't have its morphosyntactic properties.
> > >
> > > It can't have its phonological properties. The issue of whether it can
> > > have its morphosyntactic properties is part of what we're debating.
> >
> > I don't think that what I'm saying is controversial if the proof
> > for morphosyntax is morphonology:
> >
> > (3') * Noam has married Carol, and me (has married) Carmina.
> >
> > (3'') * Noam has married Carol, and me (have married) Carmina.
[snip!]

Sorry, And, for the lack of explicitness. You've completely mis-
understood me (my fault). Maybe it would be easier to understand
like this:

(3') Noam has married Carol, and me (* has married) Carmina.

(3'') Noam has married Carol, and me (* have married) Carmina.

There are two things wrong with (3'): (a) person agreement;
and (b) finite verbs don't take accusative subjects (excluding
coordination from consideration, that is--e.g., "me and you").

(3'') is ungrammatical only because of reason (b).


> In fact,
> given the actual data judgements I was hitherto aware of, the conclusion
> would seem to be either that the parenthesized material is not syntactically
> present at all

This is my point. This shows that semantics and phonology are
not completely isomorphic.

> or that it is syntactically
> present but the has/have distinction is a case of purely phonological
> inflection rather than morphosyntactic concord.

This is cheating. Lasnik says something similar: deletion is not
sensitive to agreement constraints. If true, then falsification
in linguistics is essentially impossible.

Joe 
__________________________________________________________
Home page: http://lingua.fil.ub.es/~hilferty/homepage.html




%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%%