[Corpora-List] ad-hoc generalization and meaning

[John F. Sowa]

Rob,

You can define patterns by means of a grammar, but logic
takes the next step of specifying criteria for what it
means for a pattern to be true or false as a description
of something else.

 > OK, then doesn't any _grammar_ define a logic of its own?

Grammars only define patterns.  They do not say anything about
whether the patterns are true or false about something outside
the set of patterns.

 > Remember the ad-hoc grammatical patterns found by syntax in
 > the model I'm suggesting all reflect meaningful regularities
 > in a corpus (e.g. "strong" = "black" or "strong" != "black").
 > They are only ad-hoc in the sense they select between
 > contradictory "truths".

That extra assertion brings in the notion of truth or falsity:

    "These patterns represent regularities in that corpus."

But this sentence is not part of the grammar itself.  It
makes another assertion, which could be true or false.
That is sufficient to make it a rudimentary logic.  To make
it a more complete system of logic, you could add operators
for 'and', 'or', 'not', 'if-then', 'some', and 'every' with
rules of inference for reasoning about those operators.

 > Isn't cognitive grammar essentially an attempt to enumerate
 > "a logic covering the whole gamut of grammatical constructs
 > found in a Corpus"? They try to equate grammar with meaning,
 > anyway.

I would never try to characterize anybody else's work in
a one-sentence summary.  In any case, I don't believe anyone
working on cognitive linguistics equates grammar with meaning.

More generally,

  1. A grammar specifies patterns.

  2. Semantics (which includes logic) brings in the question of
     whether some pattern is true or false about something else.

  3. Pragmatics brings in the questions of how and why anyone
     would be interested in what some pattern says about
     something.  (In short, pragmatics involves the purpose
     or intention of the speaker and/or hearer.)

The word 'meaning' is an informal term that involves issues
related to all three of these points, but it would be a
mistake to "equate" it with any one of them.

 > Perhaps I have misunderstood your article. Were you not
 > presenting Wittgenstein's games as an ad-hoc form of logic?

In that article, I merely mentioned Wittgenstein's work as
a important way of characterizing pragmatic issues about
how and why language is used.  I didn't even attempt to work
out the details of how W's ideas relate to all the issues
of syntax, semantics, pragmatics, etc.

That's an important question, but not one I would try to
answer in a short article.

John


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