Tom Wier wrote:

> We're not talking about semantic relations here, but syntax.
> Nouns and verbs are abstract grammatical categories that only
> exist if there is reason to posit they exist in the syntax of a language.
> No one (except maybe solipsists) denies that there really are things
> out there that fit into semantic equivalents of those abstract
> categories -- but those are semantic categories, not grammatical
> categories.

OK, let's play that game.

> So, when we're talking about syntax, we can't talk about "disguising"
> the existence of a category -- the category itself either exists or it doesn't.
> (whether or not it's reasonable to have that category is a different
> question).   If a language (like Yupik Eskimo) likes to incorporate
> stems whose semantic meanings represent physical objects into
> an abstract category called "verbs", there's nothing stopping it from
> doing that.  It's not disguised;  that's a confusion of semantics and
> syntactics.  For example, in Oneida, you can incorporate the noun
> itself directly into the verb:
> lanV,stayV,'sthos  "he [or she] plants corn"
> (where my <V,> here is [V], I think)
> la-         3rd person singular affix
> yV,-      to plant
> -s          present tense
> nV,st-    incorporated form of o-nV,ste?, 'corn'

Well, that's cool, and English stuffs nouns into verbs too,
anyway even here it is difficult *not* to see "corn" as
essentially nominal and "implant" as very verb-like.
I don't think either can be analyzed away. My own analysis
would try to break down "implant" to "make be planted", but
I can't because the Yupiks apparently made it a basic concept.
So, I must admit "implant" as a true verb-type-thing.

> > In that sense, Spanish may not "warrant". But it's nonsense,
> > e.g. in "I'm telling you this" I count 1 verb and 3 nominals.
> Why are you assuming I agree with Comrie?

Not. I just respond to the ideas as stated. You explain well.

> All I meant to
> do was to show you that the world doesn't usually fit into our
> neat little categorizations as often as we'd like it to.

Granted. But I still don't think any natlang dispenses with
either nouns or verbs, however fuzzily defined. Would you
propose Yupik as a good counterexample?

> When you
> make a claim about linguistic typology, one needs to be
> able to back it up with facts, closely analyzed.

Well, I do not necessarily agree with Comrie's (etc) methods
of analysis. It is a good survey, but not the last word.

> And please -- don't dismiss Comrie unless you've read what
> he has to say. :)

Not dismissing, and did read through. I thank you for your
attempts to enlighten me on this, but I am such a poor student.
Once any field is named, there must be experts, and dogmas,
which change frequently; and this is a good thing, overall.
But I have little faith in these things.