Language, sort of; rubber bands

[From Bill Powers (930128.1230)]

One thing I forgot to say to the linguists: I just barely know
what I am talking about, and sometimes I don't even know that
much. If my remarks seem silly and uninformed, just think of them
as a cry of frustration.

Bruce Nevin (930128.0830) --

The segue from your reply to Rick's (930125.1500) on output and
perception "bushes" is actually a non-sequitur. A phrase-
structure tree with NPs, VPs, etc. is a structural description
that applies to an indefinitely (but not infinitely) large set
of sentences.

I guess this is my main problem -- not understanding what a
phrase-structure tree is supposed to be. It doesn't seem to be a
model. It doesn't seem to correlate with any presumed operations
that the brain does in constructing or recognizing a sentence.
It's clearly something that some linguists do with sentences --

I guess I just can't express the nature of the gap I sense
between manipulations of phrase structure trees and what would
look to me like a model of a system capable of doing language.
Maybe I'm just expressing what it feels like to be ignorant.

The bottom-up control-theoretic approach that you counterpose
to this is I think what Avery has been proposing. (Correct me
if I'm wrong, Avery, or elaborate if you will.)

Yes, please do.

Where I keep falling off the bandwagon is always at the same
bump. We start out talking about the relationship between words
and nonwords, and then suddenly all the direct experiences of
nonwords drop out of the conversation and we're back to talking
about the properties of words all by themselves. This feels to me
like talking about chess by discussing only the way the players
move their hands, without ever referring to the chess pieces and
their properties, or the point of the game. Yet it's ONLY because
of the chess pieces and their properties that the players move
their hands at all. No doubt you could find a lot of structure in
the hand-movements, and there would be provable preferred
associations of one hand movement with another. You could build
up some very complicated theories about the way chess players
move their hands without ever referring to the chess pieces or
the board or the underlying rules of the game.

I don't see how there can be any rational account of language
that doesn't include an account of the properties of the
experienced world. We don't say "John hit Mary" just because
that's what other people say. In the experienced world, there are
agents doing actions on objects of action. If we want to convey
such processes to someone else, we have to settle on some sort of
convention for designating the agent, what was done, and the
object. We could say "The act was hitting, and the doer of the
act was John, and the victim of the act was Mary," listing those
conjoined descriptions in any order. After a while we would tire
of that and find shortcuts; word order, prefixes and suffixes
that are understood to carry the same information, or whatever an
inventive mind can think up. But whatever convention we arrived
at, it would have to provide ways of indicating the essential
elements of the experience. There would be no way to explain
these conventions without knowing what those essential elements
are. Yet it seems to me that linguists continually try to analyze
language without any (or enough) reference to what it is that
language is supposed to enable us to talk about.

Well, don't take me too seriously. I know we're all on the same
side.

ยทยทยท

---------------------------------------------------------------
Ed Ford (930128) --

What's nice about the rubber band experiment is that you can
demonstrate just about all the features of PCT with it. Your
latest is a fine example. Sounds like something Dag Forssell
could use in his seminars.
----------------------------------------------------------------
Best,

Bill P.