I'm not sure the above subject heading is the best anymore, since we seem
to be off on a sort of theoretical discussion.

Bob Yates brings up the examples that show how performance underpredicts
competence, such as the 'grandstanded' example.

Current mainstream theories don't believe that analogy is enough to
account for acquisition of language; cognitive linguistics** believes that
analogy _is_ enough. But it does not claim that analogy is _all_ that is
at work in language acquisition. I think that anyone looking at the
discoveries of linguistics in the last three centuries would be foolish
not to conclude that there is some innate predisposition interacting with
language learning. The big debate is: how much innate stuff is devoted
_exclusively_ to language, viz., is a module autonomous from other
cognitive functions. A subthread of this debate is whether subsystems of
language, such as syntax, semantics, phonology are also autonomous from
one another, connected by 'interfaces', or whether they all share the same
cognitive space, as it were.

**The Cogling list is not connected with anthropology; it was started by a
UCSD graduate student in Cognitive Linguistics to serve, primarily, people
working in cognitive theories of linguistics, but of course the list is
open to anyone who is interested in joining.

It is now a well-known and oft-discussed fact about languages that new
additions to the word stock that _could_ follow an irregular rule tend to
follow the regular rule instead, in spite of the lack of a model.
'Grandstanded' instead of 'grandstood' is one example, and 'the Toronto
Maple Leafs' (not 'Leaves') is another. 'Sweet tooths' would be another.
Would you say your family abounds with 'sweet teeth'? I couldn't. I wish I
had the reference (from within generative morphology) that deals with this
at hand, but it is at home. A. Spencer's 'Morphological Theory', 1991,
Blackwell, would have something on it. It's an intro to generative
morphology, and a good one.

Such forms have _two_ models which could be invoked for marking past tense
or plural: the regular and the irregular. There may be more than one
factor at work. (1) Irregular patterns of marking tend to be much less
productive than regular ones. Hence new words in the language or category
_tend_ to inflect regularly. (2) 'Grandstand' and 'Maple Leaf' share
phonological content with 'stand' (take up a vertical position) and 'leaf'
(plant part), but they have little _semantic_ content in common.

One extracts a regular past-tense
marker from hundreds of experiences of that marker on a variety of verbs
that are also heard without the marker. The learner thus infers that any
word with something that looks like the suffix is analyzable (as in a
clever Family Circle cartoon in my collection in which Dolly, entering
church with her family, asks her father 'Are you going to sit with us,
Daddy, or are you going to ush?') The learner then infers that the marker
can be applied to any new word that seems to be in the proper category,
and applies it, as in 'grandstanded' and 'Maple Leafs'.

What does it prove when someone applies a rule and produces a form that
has never been uttered? It proves they have learned a rule and are
generalizing it to new data. This doesn't prove that you can't acquire
language by being exposed to performance only. In fact, it argues for
the opposite: you need only to be able to extract generalizations, which
is an extremely general trait of human cognition in no way limited to
language. This is, I'm sure, an innate ability, and therefore does not
derive from exposure to language performance -- or any other kind, for
that matter -- but it doesn't prove that we need to posit a kind of
competence special to language only, that is used in language acquisition
exclusively.

I believe there are other facts about language that point to there being
something innate that, if not devoted exclusively to language, definitely
helps shape grammars. One example is the 'me first' principle, according
to which greater similarity to human beings or to the speaker enable
grammatical markings reserved for humans or animates. Another is
agent-focus: the tendency to attend more to agents, to prioritize them in
your attention, than to patients or goals. The tendency to foreground
moving objects and background stationary ones also shows up in language.
A cognitive linguist friend recently told me that an exploration of the
brain functions required to reach and grasp objects and use the arms in
other ways translates directly into the semantics of aspect in verbs
across languages!! This strikes me as farfetched, but who knows??

This connects up to Jim Kenkel's point that at least some writers on
generative grammar don't exclude pragmatics or performance entirely from
the scope of linguistics. Yes, there are linguists who practice generative
theories of semantics and pragmatics. They tend to be strict autonomists,
in my experience -- for instance, they do not want to hear that syntactic
form may be motivated _mostly_ by semantics or discourse function. They
are developing a theory of a given module. The interface development is
viewed as further theoretical work. Cog. linguistics is trying the direct
opposite, of taking as the working hypothesis that everything comes from
semantics (and functionalism that everything comes from discourse, with
points in between) and pushing that to its limit. It would be nice if the
camps would treat each other with mutual respect. They don't. Generativism
is currently dominant, and generativists have not felt obliged, until
recently, to pay attention to or cite the work of cognitivists and
functionalists. I attended a conference in Chicago a number of years ago
about the interaction of pragmatics with grammar, and a noted linguist
stood up and said, in so many words, 'what y'all are doing here is very
interesting, but it's not linguistics'. Linguists are individuals; some
take a broader perspective than others. I have seen generativists deal out
very shabby treatment indeed to those in the other camp. Yeah, I've seen
it go the other way, too. On the other hand, I respect the positive
contributions of generativism and I cite generative work in my own.

~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~
Johanna Rubba   Assistant Professor, Linguistics              ~
English Department, California Polytechnic State University   ~
San Luis Obispo, CA 93407                                     ~
Tel. (805)-756-2184     Fax: (805)-756-6374                   ~
E-mail: [log in to unmask]                           ~
Office hours Winter 1999: Mon/Wed 10:10-11am Thurs 2:10-3pm   ~
Home page: http://www.calpoly.edu/~jrubba                     ~
~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~