[Corpora-List] ad-hoc generalization and meaning

[Michael Maxwell]

Rob Freeman wrote:
> It is of course John who claims the syntax of his examples is captured by:
>
> "A person supported NP1 with NP2"

...if that's what he's saying, then I'm agreeing with him.

> I think the syntax of "supported" in each context will be different.

What is your definition of "syntax"?

> Failure to treat the syntax of "supported" on an ad-hoc basis in this way
> means you have no way of capturing the information that, in a grammar of
> English, "supported" = "accompanied" but also "supported" !=
> "accompanied".

The question is whether it's appropriate to capture that meaning
distinction in the syntax, rather than in the semantics.  It seems to me
that there's a class of facts that can be captured quite well by
distinguishing syntax from semantics, but which is obscured when you try
to collapse the two.

> That means you will be unable to capture detailed syntactic restrictions
> which prevent you from saying "slightly odd" things like "Tom accompanied
> his tomato plant to the garden (where he planted it.)"

I would put it differently; I would say that collapsing syntax and
semantics makes it difficult (maybe impossible :-)) to capture the fact
that the above sentence is perfectly comprehensible, if (as you say) odd;
whereas the following sentence is only comprehensible to English speakers
with great difficulty, although the intended meaning is quite normal:
   garden-the-to accompanied tomato-plant-his Tom

> John would claim such restrictions are purely semantic, but in point
> of fact you can capture them with an ad-hoc search for syntactic
> regularities along the lines I recommend.

I'm sure you can find such regularities; the question is whether treating
them as syntactic doesn't obscure the generalizations that I would
consider truly syntactic.

> I don't know why his is resisting this.

I'm with John here.

   Mike Maxwell
   CASL/ U MD


_______________________________________________
Corpora mailing list
Corpora at uib.no
http://mailman.uib.no/listinfo/corpora