On Fri, Jan 18, 2013 at 12:15 PM, Gary Shannon <[log in to unmask]> wrote:
> On Fri, Jan 18, 2013 at 11:23 AM, Garth Wallace <[log in to unmask]> wrote:
> ---snip---
>> "Part of speech". Or "word class", or "lexical category", etc. At
>> least if I'm understanding you correctly.
> I'm referring more to the method of collecting sets. Besides, each of
> those terms is already loaded down with specific meanings, so if I use
> them to describe something different there will be no end of
> confusion.
> And apparently I'm not doing too good a job of getting my idea across.
> Things like "part of speech" are similar, but the criteria for placing
> a word in one part of speech or another is different. If I said that a
> possessive pronoun and a definite article were the same part of speech
> people would think I was incredibly stupid.

But you can say that they're both specifiers and be entirely correct.

If you treat classes as a flat, mutually exclusive categorization
scheme, of course it doesn't work. But I don't think there's any
reason to do so. Langauges have all sorts of subclasses that restrict
the possible syntactic contexts they can appear in beyond those of the
larger categories, or allow exceptional syntactic contexts:
intrasitive vs. transitive verbs, verbs requiring certain oblique
arguments, verbs requirinf "quirky subjects", deponents, mass vs.
count nouns, pluralia tantum, prepositions with different valency (if
you accept that theory), and so on. You can even allow for some
overlap, for example the "specifier" class in English contains
articles, demonstratives, and possessive NPs.

> But I CAN say that with
> respect a particular specified template possessive pronouns and
> definite articles CAN BE isotactic. So "part of speech" doesn't cut
> it.
> Also, I'm interested in sets of words that are BOTH grammatically AND
> semantically _sensible_ in the specified context.

"Semantically sensible" is pretty nebulous. It can depend on any
number of other words in the sentence, and even on information outside
of the sentence. Even the audience is part of the context. I'm not
sure that, even given all of that, you could always strictly say that
a word is or is not semantically sensible (possibly more or less
semantically sensible). It'd be tough to come up with useful
generalizations if that's a criterion.