Grammar with a G

[david_tuggy at SIL.ORG]

     Rob Freeman wrote:

     ****
If you really accept 'shifting schemas' then do 'rule-based and analogy-based
accounts differ only in degree, not in kind' as you say in your first message?
To fit my model this would mean a 'rule-based account' which had a new 'rule'
for every concept - essentially infinite.

Most of the time when people think of a rule-based account they think of a
finite number of rules. Analogical models which seek to extract rules or label
structure can only get a finite amount of it. I think that the relevant
structure is essentially infinite. That's the difference I'm trying to
highlight.

Apart from that I agree totally with Langacker's model. An idea of '_shifting_
schemas' would seem to fit the bill precisely.
     ****

     Again, I'm not sure what you mean. If "shifting schemas" means
     the generalizations people extract are constantly being adjusted
     (in their current computational form) by context, sometimes
     drastically, and (in their stored form, usually more
     incrementally) by such current usages, yes, the model has that
     built in. Or if you mean that different systematizations of
     low-level generalizations may coexist and rise to or fade from
     prominence over time, yes that is also built-in. And yes, this
     does imply that the grammar of the language is not something that
     you can ever describe exhaustively or get all of (it) down on
     paper.

     A new "rule" for every concept? I don't know about that. Not
     every concept by any means is part of any language, but only
     those that are conventionalized: shared and known to be shared by
     the users of that language. And conventionalization is of course
     a matter of degree. It does seem probable that all
     conventionalized concepts are schemas, generalizations over
     more-specific particular concepts that occur in the individual
     users' minds (which concepts of course may themselves be
     schematic in some degree). So in that sense, sure, there is a new
     "rule" for every new concept in the language. And of course any
     new usage can prompt a new systematization or higher-level
     connecting "rule" generalizing over such low-level rules; either
     the user tumbles to the new generalization (which wouldn't make
     it part of the language), or he comes to believe that he shares
     it with his interlocutor (which does make it conventional, and
     thus part of at least his version of the language--their version
     if he is right.) But an awful lot of what we use in communicating
     has already undergone this conventionalization process, and those
     concepts constitute grammar (and lexicon and everything else).

     Hey, if this discussion is boring to the rest of you, tell us so,
     and Rob and I can talk in private! (Or maybe we're done?)

     --David