Henrik Theiling wrote:
> > What I meant is that you are assuming that "case" can exist
> > apart from some actual construction or form in the language. ...
> > This is precisely what Ray and I have been arguing against: that
> > such a thing as case is "abstract" rather than overt.
> Ok, I think I understand. 'case' is the morphological way of marking
> which nouns belongs into which argument slot of a verb. Word order is
> a different way, postpositions yet another. Is that right? This
> operation as such has no standard name, but maybe 'argument assignment'
> may be appropriate. I, obviously mistakenly, used 'case' for both things:
> the implementation, and the assignment operation as such.


> > It is generally agreed, whether you take Trask's first or
> > second definition, that the category of "case" is a mapping
> > from verbal arguments to thematic roles (whether by derivation
> > or not).
> WHAT?? Confusion alert! But now you are mixing up levels, no? This
> is not in line with the rest of your argument.

Actually, it is. What I was stating here was what the two definitions
(both the case-as-morphology and case-as-abstraction) share.  They
both map from thematic roles to arguments in syntax, but the morphological
analysis says this is purely by virtue of the overt marker, not because
there is some covert ghost property floating around in the syntax.

> You say in the very next paragraph:
> > ...
> > How those morphological forms map onto thematic roles is an
> > entirely separate question.
> 'case' is merely a way of assigning arguments to their argument slot
> of the verb. That's an implementation not of a semantical operation,
> but of a syntactical one. Which thematic role is represented in the
> XY argument of a verb is not expressed by case, but typically
> lexicalized with the verb. As you say later, case names are merely
> arbitrary labels.
> ASCII-graphically:
> syntactical operation: noun phrase --is-assigned-to--> verbal argument slot
> Possible implementations: - word order - adpositions - case ...
> semantical operation: verbal argument slot --is-assigned-to--> thematic role
> Implementation: - lexicon lookup for the verb ...?
> Now, would we agree up to here?

No, we wouldn't.  I favor an architecture of grammar in which the
separate modules are quite autonomous (though not entirely so).
What you have suggested here is that the syntactic and morphological
modules are not distinct: case and adpositions are lumped together
as "implementations" of syntactic processes.  To use LFG terminology,
you are conflating lexicocentricity with endocentricity: properties
that are inherent properties of words and word-structure, and those
that arise as a result of phrase structure rules.

Thomas Wier	       "I find it useful to meet my subjects personally,
Dept. of Linguistics    because our secret police don't get it right
University of Chicago   half the time." -- octogenarian Sheikh Zayed of
1010 E. 59th Street     Abu Dhabi, to a French reporter.
Chicago, IL 60637