OK, in response to what Matthew suggests “is wrong with the C/P distinction”:<br
/><br />· “In everyday usage, the notions of knowing and using a language
seem inextricably bound” (Kaufer, 1979, p.257). To illustrate, Kaufer points out
that one must strain hard to find cases of people using a language they do not
know (reciting Latin prayers, perhaps) or, on the other hand, knowing a language
they cannot use (e.g., some forms of aphasia). A similar point is also
attributed by Stemberger to Fromkin: “assuming that performance is based on
competence, we expect performance to reflect many aspects of competence”
(InfoCHILDES, October 14th 2007)<br /><br />This seems to be an argument against
the idea that C and P are totally unconnected, do not interact, do not jointly
influence behavior, or something like that. But to claim that they are logically
distinct does not imply any of those other things, so this does not bear on the
issue.<br /><br /> “performance factors” can always be invoked to explain away
awkward-shaped pearls that fall from children’s mouths. If they don’t fit the
hypothesis under scrutiny they are rejected, seemingly on good grounds, but, of
course, invoking performance in this way is not at all well motivated.<br /><br
/>I agree, as discussed in my previous posting. But this isn’t the fault of
drawing the C/P distinction, this is the fault of insufficient research.<br
/><br /><br />· Another view: the competence-performance distinction relies
on a particular account of adult grammar (Chomsky’s) that is not to everyone’s
taste. If one rejects a Chomskyan version of grammar as the endpoint of
development, then the competence-performance distinction is rendered redundant
in the process (Ambridge, Rowland & Pine, in press).<br /><br />This is a
total nonsequitur—see Gary’s posting. C/P has nothing to do with any particular
account of grammar, even if both were written about by the same guy (who has
repeatedly noted that he did not “propose” C/P since it’s always been around as
an assumption, except perhaps for the Skinnerians).<br /><br />· One
possible alternative is offered by McClelland & Bybee (in press), based on
the notion of gradience, which they take to be “an inherent feature of language
representation, processing, and learning” (McClelland & Bybee, in press,
p.1). This view clashes with the traditional notion of grammatical competence in
which a given utterance is either grammatical or ungrammatical.<br /><br
/>Another nonsequitur. In fact, in the earliest work on generative syntax
(Logical Structures..., 1955) already incorporated degrees of “grammaticalness”,
as Chomsky called it then, as do many versions of OT today (while still being
models of grammar, not models of processing).<br /><br />· Connectionist
approaches also often clash with idealizations like competence, since the latter
“excludes aspects of linguistic performance that are .... central to the
structure of utterances” (Seidenberg & MacDonald, 1999, p.572). This point
is exemplified by Hoff (under review) with her observation that social factors
can affect the linguistic form of observed child speech output (for example, the
contrast witnessed in two-year-olds’ conversation with their own mothers versus
a researcher).<br /> <br />Competence doesn’t exclude that or anything else, if
it is systematic behavior (it does exclude blunders, as Gary says). I sense here
some allusion to the passage in Aspects that talks about an “ideal
speaker-hearer...”, but even on the ridiculous assumption that everyone working
in this paradigm is bound by every word uttered by its de facto leader 40+ years
ago, no such exclusion would follow.<br /><br /><br />> Perhaps we should
talk instead about two kinds of competence: one concerned with linguistic
>competence, the other with speech production competence. At some level,
therefore, it may be proper to >acknowledge a split between competence and
performance, or competence in one domain from >competence in another. <br
/><br />Right. Certainly Chomsky and everyone else I know would be content with
positing grammatical competence and sociolinguistic competence, for example
(which does not preclude interactions between them, just like positing syntactic
competence and phonological competence doesn’t preclude such interactions). It’s
all about carving up the empirical pie (which is all behavior, hence
performance, as I think we all agree) in a way that gives us the most
scientifically appealing theories of each of the domains. <br /><br />
Carson<br /><br />