And Rosta wrote:
>
> > > #> (Well, maybe you
> > > #> (insanely!) do deny it, but if so, tell me what is the object of
> > > #> study when, say, the mathematics of chess is studied.)
> > > #
> > > #I assume that they are studying the possible moves. However, these
> > > #possible moves have no independent existence outside of the humans
> > > #that entertain such thoughts. Otherwise, we're getting into essences
> > > #and dualism. (In this case, at least, I think that my position is
> > > #pretty sane!)
> > >
> > > Obviously I disagree with you about criteria for "independent existence".
> >
> > And, you shouldn't disagree, unless you're a dualist.
>
> Well, yes, I am, though that doesn't mean I believe in disembodied minds.
> Knowers are embodied but what is known is not.

In other words, you're a dualist/monist. That sounds like a very
volatile combination!


> > > Anyway, I'll go along with you for argument's sake. It remains a fact
> > > that a system of possible moves does "fall out" from mental chess. That's
> > > not true of all games.
> >
> > What games are you thinking about?
>
> Football, say. You couldn't write a generative grammar of football that
> defines the set of all possible games of football. Or tennis. You
> could probably work out the rules of tennis inductively, but you still
> couldn't use them to create a generative grammar of tennis.

(Playing the Devil's advocate:) Why isn't language like soccer?


> > Certainly, but that's why taking development seriously is so
> > important. Besides trying to look at language as a formal system
> > that has no grounding in the human body (not to mention
> > interactions with the environment, etc.), you're also trying to
> > look at a steady state that has no developmental history. The
> > wager that I'm making is that both embodiment and development
> > are key factors for a full account of grammar. In fact, I'm
> > convinced that we won't understand syntax until we understand
> > neurology better. In any event, you learn grammar all your life.
> > It's true that, after about four, things start to slow down
> > somewhat, and after the onset of puberty the learning curve
> > flattens out quite a bit; but it doesn't flatten out entirely.
>
> I'd wager the same. But mathematical properties of language might
> be important too.

It's not necessarily the mathematical properties that matters, but
the TYPE of mathematical properties. Are they discrete, continuous,
stochastic, etc., etc., etc.?


> For example, it might be that any parsing algorithm
> is drastically simplified if the language is strictly continuous
> (non-Tangling); that would then explain why syntax is continuous.
> [I'm sure there is a known answer to that, but I happen not to
> know it.] We could then place further wagers on the relative
> importance of the biological and 'mathematical' bases of language.

Obviously, biology comes before the mathematical model (linguistic
engineering notwithstanding). Ultimately,  mathematical modeling
will ultimately have to be consistent with what our bodies (read:
brains, especially) can do. Take connectionism, for example. The
biological plausibility of the back-propagation algorithm that is
almost always employed is entirely dubious. If someone can come
up with an algorithm that is more in line with how neurons are
known to work, then that would be a big advance.


> > > #How do you account for the existence of graded grammaticality
> > > #judgments? The rules of chess are discrete; so, for example, you
> > > #can castle or do "en passant" only under certain specific conditions
> > > #but not others. There are no "kind of legal 'en passants'" but there
> > > #are many sentences that are only "kind of well-formed."
> > >
> > > Gradient acceptability judgements are due to:
> > >
> > > 1. pragmatic anomaly.
> >
> > Yes, but pragmatics(/semantics) can determine grammaticality, not just
> > acceptability. Take Brugman and Lakoff's argument:
> >
> > (1) a. * Under no circumstances I would allow behavior.
> >     b. Under no circumstances would I allow such behavior.
> >
> > (2) a. With no help I lifted the heavy box up onto the top shelf.
> >     b. * With no help did I lift the heavy box up onto the top shelf.
> >
> > In these examples, the difference IN GRAMMATICALITY is due the fact
> > that negative-polarity auxiliary inversion requires a negative
> > entailment. Example (1b) implies that I wouldn't allow the behavior
> > in question, and thus auxiliary inversion is okay here. On the other
> > hand, (2b) implies that I did in fact lift the box onto the shelf,
> > so inversion is no good.
>
> This is weird. It's normally me that insists on the grammaticality/
> acceptability distinction and you ('you lot') that deny it.

Actually, I'm saying that, insofar as the distinction is real,
acceptability is affecting grammaticality (i.e., pragmatics
is a condition on syntax).


> The grammaticality of given sentence is an axiom, not an empirical fact.

No!


> If you can explain the unacceptability of a sentence without invoking
> ungrammaticality then it's more economical to do so.

Actually, you can do that by fiat (we all know that this is done
quite often). But then you run into falsiability problems, in that
the scheme becomes inherently unfalsifiable.


> So, I see no reason to treat (1a) as ungrammatical.

You really think that "* Under no circumstances I would allow such
behavior" is grammatical?

> It's possible to
> construct contexts in which it becomes acceptable, and its unacceptability
> is easily explained without invoking grammar.

Such as?


> As for (2b), the sentence literally means I didn't lift the box, so if
> I use that sentence in an utterance that tries to communicate that I
> did lift the box then, obviously, unacceptability results.
>
> > > 2. (un)familiarity
> >
> > This is a HUGE problem for the notion of disembodied grammar, because
> > it shows that people are a necessary condition for language. How can
> > you (by which I mean "anyone in general") say that a given structure
> > is okay in abstract, but speakers may vary in their judgments? That's
> > begging the question.
>
> Yes, it is.

Hmmmm...


> The statement that a given structure is okay in abstract
> is a fact only to the extent that it plays a role in a full model
> of acceptability judgements.
>
> The grammar is based on the set of valid sentences, and these in
> turn are found out by a kind of triangulation or algebraic equation:
>
>   grammatical sentences + pragmatics-etc. = acceptability judgements
>
> You know the acceptability judgements and you know pragmatics-etc.,
> and so you can deduce what the set of grammatical sentences is.
>
> To me, all this is so obviously a part of the methodology of
> formal armchair linguistics that it baffles me how so many
> people (including most of my colleagues over the years) who
> inveigh against armchair linguistics can fail to grasp it.

Did you know that there is quite a controversy concerning event-
related potentials and the grammaticality-acceptability distinction?
At the very least, the distinction is at present unclear (at least
in terms of electrical activity in the brain). If the distinction
(taken across the board) strikes people anti-intuitive, *maybe*
there's a physiological basis for it.


> > > 3. the salience of alternative ways of communicating the same
> > > information (-- the lack of alternative ways improves the
> > > acceptability)
> >
> > Again, this implies people, not abstractions. Clearly, people
> > know that grammar and discourse go hand-in-hand to some (a great?)
> > extent: in fact, some constructions exist for discursive reasons!
> > So, I don't see how an abstract analysis is going to be very
> > revealing in such cases.
>
> In case it's not clear from what I've already said, acceptability
> judgements indubitably involve people.

And grammaticality doesn't? Isn't this a double standard?


> If a sentence is unacceptable whatever the context, and if we can
> explain the unacceptability without invoking grammar, then it
> doesn't really matter whether we treat it as grammatical or not;
> it's one where we can let the rest of the grammar decide, as per
> Chomsky 1957.

Chomsky has also admitted that people are not capable of making a
clean distinction between grammaticality and acceptability. Which
Chomsky do we accept? And which Chomsky do we accept for ourselves?


> > Hmmm... So, if we happened to find a new animal, say, deep at the
> > bottom of the ocean, then it was abstract before we found it and
> > concrete after we found it. This situation fits your description:
> > the animal was unknown to humans so it had no corresponding human
> > brain state. This would make it abstract. However, once we found
> > the animal, then it wouldn't, by your definition, be abstract any
> > more. Being a realist myself, I'm not willing to accept the logic.
>
> No, you misunderstood me. If it's abstract then it can't be reduced
> to patterns in material spacetime. Clearly a fish does exist in
> material spacetime. But some stuff that we know may indeed not
> be reducible to patterns in those bits of material spacetime that
> our behaviour and brains constitute.

I disagree; but anyway...


> > But say that we trained a bonobo monkey (a very smart one) to make
> > only legal moves, but nothing else. Would the bonobo be playing chess,
> > or merely be moving the pieces in the manner that he was trained. I
> > don't think that you can really say that the bonobo is playing
> > chess--that would be the equivalent of saying that the man in the
> > Chinese room knows Chinese.
>
> There is a distinction between the activity of playing chess, which
> is a prototype category based on human activity, and, on the other
> hand, the formal definition of a valid chessgame.

Searle wouldn't let you get away with this so easily!


> > Of course, by the same token, I probably shouldn't have brought in
> > computers if I'm going to get into the intentionality thing. But
> > there is something called extended cognition: acts of cognition
> > that require external cues, such as an abacus, or instruments that
> > are designed to perform certain actions instead of humans. So, by
> > virtue of extended cognition Clueless might be playing chess.
> > However, this is not the case of the bonobo.
>
> Aren't we just arguing over definitions of terms here? Call
> situated chessplaying 'Chess1' and abstract chess 'Chess2'. Then
> when I talked about 'Chess', replace that by 'Chess2' and we no
> longer need to argue this particular point!

Yes, but Chess2 (=Clueless) only exists because we realize that was
the programmer's intention. It doesn't exist in the abstract.

Joe
_________________________________________________________________
Home page:        http://lingua.fil.ub.es/~hilferty/homepage.html
__________________ http://www.ub.es/filoan/hilferty/homepage.html