I'll defend the C/P distinction!

[Carson Schutze]

C = competence, P = performance below, because although I have the  
competence to type them, I know my performance system will make too  
many errors doing so :-)

But back to serious business. Anat Ninio said:

> I have a feeling -- following Joe Stemberger's excellent contributions
> to this strand -- that we're all a little scared stating the obvious,
> which is that the notion of competence as a separate entity from
> performance is a philosophical error, pure and simple.

I'm not scared to state what's obvious to me--that *rejecting* the C/P
distinction is a philosophical error. And the fact that I have no problem
with 98% of what's been said in this thread aside from denying C/P suggests
that there's some misunderstanding about what embracing it entails for one's
research program. So let me pick up from Joe's most recent posting and see
if I can pinpoint the problem.

[But let me also say: none of the 'problems' for C/P cited by Matthew Sexton
in his summary posting are problems for it at all, in fact they don't even
bear on it. This posting is already too long but we can get into that
separately if people think it would be productive.]

Joe says

> (1) the exact division between what is competence and what is
> performance, as well as the criteria that distinguish them, are largely
> unknown after more than 40 years.

That's also true for syntax vs. semantics, morphology vs. phonology,
semantics vs. pragmatics, etc. etc. That doesn't mean these aren't real and
useful distinctions. They might or might not be real, but no one's shown
that they aren't, and I defy anyone to argue that linguistics would have
made more progress by ignoring the distinction. (And some of these go back
hundreds if not thousands of years, so C/P has plenty of time to sort itself
out.)

> (2) There is no clear way to test competence except through performance.

A point I harped on in my 1996 book, too. Certainly true today. Arguably in
principle brain measures might someday falsify this, depending on a bunch of
assumptions we can't even guess at the veracity of right now. But not in any
of our lifetimes, so we can agree on this for all practical purposes. But
this also doesn't imply that C/P is useless, incoherent, etc. Until recently
there was no way to study genetics except by its effects on organisms, but
that doesn't mean it was wrong or useless to make the genotype/phenotype
distinction.

Joe then discusses overgeneralization of morphological irregularities, and
points out that these occur alongside overgeneralization of the "regular"
forms, and that therefore the idea that the latter should be handled by the
grammar while the former are handled by a processing theory is unmotivated,
and the thin edge of a wedge by which all linguistic phenomena could be
"dumped" into processing and therefore ignored by linguists. [I hope I'm not
oversimplifying too much, Joe--I believe that was the essential idea.]

For one thing, I agree that the division that Pinker, Clahsen et al. argue
for is not the right way to go--that, partly for the reason Joe gives and
partly for other reasons, all inflections should be treated in a single
system. Morris Halle agrees. So does Charles Yang. So do Albright and Hayes.
And we all want that system to be the grammar, not the processor. So in
practice, *for this case* I don't think the slippery slope worry is
justified.

But Joe wants a guarantee that in principle C/P can't be used to justify a
division of labor between grammar and processing that we would all consider
crazy. Is that a reasonable standard? Can anyone show that "syntax" and
"phonology" have been defined in a way that wouldn't allow everything we
commonly think of as syntax to be classified as phonology? Since all
sentences are built up out of sounds, clearly that cannot be excluded a
priori. But has that prevented linguists from making profitable use of the
distinction? Again, I think not.

So sure, C/P could be used in crazy ways. No scientific theory can prevent
itself from being used to do bad research. But don't throw the baby out with
the bathwater--even if you think that every piece of work ever done that has
explicitly embraced C/P has been wrong wrong wrong, that's no argument that
the C/P distinction itself is wrong, useless, nonsensical, etc. (You don't
even know until you look really hard whether research that explicitly
rejects C/P might not in fact be using it without realizing it.)

Joe summarizes thus:

> From a practical perspective, it seems to me that, since we have to work with
> performance data anyway, we want to work with as wide a range of  
> types of data
> as possible.

I agree, as does Chomsky--see quotes in my book.

> And we want to develop theories that account for all of those data,  
> in detail.
> If there's a distinction between competence and performance, that's fine, but
> it has to be explicit, and our theories need to explain exactly which
> phenomena are due to competence, which are due to performance, and why.

Unless we're holding ourselves to a standard beyond that of any other
science, I think a reasonable requirement would be that our theories should
include a *hypothesis* about which phenomena are due to C vs. P, and they
generally do, if only implicitly: people proposing theories of grammar are
hypothesizing that the data they try to explain are in C, those proposing
theories of processing are hypothesizing that the data they try to cover are
in P, and in some cases there are explicit arguments that things that some
theories treat as being in one should actually be treated as being in the
other (as in the theories of ir/regular inflection cited above). And of
course one may propose that a single theory should cover data that have
traditionally been treated with separate theories, as in Colin Phillips's
'Parser is Grammar [PIG]' model.

But to demand an explanation of why the division, if any, lies where it is
hypothesized to lie, while it would be nice, surely can't be a prerequisite
to employing C/P. Most theories in language don't have any "why" answers for
anything, really. I think parsimony is justification enough at this point.
(I.e., drawing the line here gives me more appealing theories of BOTH C AND
P than drawing it somewhere else.)

But of course Joe is correct that if you make a division somewhere and then
cannot point to any proposals, by yourself or anyone else, for how to handle
the facts that you've dumped in the other guy's backyard, you've taken a
step backwards, not forwards. I think typically when linguists say "I assume
X is due to performance," usually they have a vague sense that X resembles
some phenomenon for which performance accounts have actually been proposed.
The problem comes if they don't actually check whether their vague sense is
correct and whether those accounts could plausibly extend to X. In an ideal
world, they would go one step further than that: they would find a
collaborator who works on those performance theories and develop an account
that actually explains X. Nowadays, this is happening--not nearly enough,
but the trend is clearly in the right direction. 

      Carson
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://listserv.linguistlist.org/pipermail/info-childes/attachments/20071017/8097518b/attachment.htm>