Problems with Chomsky

[Tom Givon]

                         1-7-2000 (wow)


Dear FUNK people,

I had the best intention of staying away from the wonderful recent
exchange y'all ran in December, about Noam Chomsky & what not. For one
thing, I thought I had nothing to add to it. For another, it was a
crowded field. So, while you can count the following missive as
certainly related, it was actually triggered by an article by Malcolm
Gladwell in the most recent issue of The New Yorker. In his article,
Gladwell reviews three recent books, by a neurologist, three
developmental psychologists, and another psychologist, resp.:
  (i) John Bruer "The myth of the first three years" (Free Press)
  (ii) Alison Gopnik *et al* "The scientist in the crib" (Morrow)
  (iii) Jerome Kagan "Three seductive ideas" (Oxford)
But the thing that caught my attention was a quote from our national
expert oin children issues, Hillary Rodham Clinton, taken out of her
welcoming address to a White House Conference entitled: "What new
research on the brain tells us about our youngest children" (April
1997). To wit:

     "...Fifteen years ago, we thought that a baby's brain structure
      was virtually complete at birth... Now we understand that it is
      a work in progress, and that everything that we do with a child
      has some kind of potential physical influence on the rapidly
      forming brain. A child's earliest experiences--their relation-
      ships with parents and caregivers, the sights and sounds and
      smells and feelings they encounter, the challenges they meet--
      determine how their brains are wired..."

What struck me about this quote is the hopeless reductionism about child
learning/development, the same reductionism that struck me on first
reading Chomsky's (1959) review of Skinner's "Verbal Behavior": The
absolute extremism -- or, if I may be forgiven, the *intellectual
Stalinism* -- of it all: "You are either with me (and Descartes and
Plato) in the Innate Ideas camp, or you are with Skinner (and Bloomfield
and Hume and Aristotle) in the S-R camp. No room in the middle".

The three books (and the more minor studies) that Gladwell reviews
actually, all suggest that the action is right in the middle. That the
child and its brain are *interactive* from the word 'go' and to a ripe
old age; that innate biases & pre-wired structure interact with input,
are pre-wired to seek input, to form hypotheses about the input, and to
evaluate the empirical evidence that is or isn't compatible with the
hypothese -- and in the latter case re-formulate and come up with new
hypotheses (and then test them).

So, maybe Chomsky's intemperedness is only alive and well as Hillary's
Choice (oops, couldn't resist that 'n). But then I recalled the tenor of
many of the contributions to the December FUNK-exchange, how
reductionist they were, how they followed Chomsky's line of either/or
reductionism but not -- God forbid -- the more complex, more realistic,
more sophisticated *middle*. You are either a rabid functionalist
("Grammars are not really really REALLY real"... Well, by the way,
Chomsky also things that grammatical *constructions* are not real, as of
1992, in case you are looking for company...), or you are a
died-in-the-wool structuralist ("Functions are mushy speculations"). No
room in the middle...

And likewise with methodology: Either you do only "competence"
sentences, (as Chomsky clearely insisted in 1965), or you do only "live"
communication data (as many functionalists insist). But God forbid that
you should try to be be multi-methodological (as most complex sciences
are).

So I wanted to ask a simple minded question: How can you be a
functionalist without automatically also being a structuralist? The
function of WHAT are you going to study? Because in all
biologically-based systems (sorry guys, we're included, kicking or
screaming...), functions are carried out by paired *structures*. Have
you ever met a pulmonary physiologist who is NOT interested in the
anatomy of the heart & lungs & circulatory system? Or a brain
physiologist/cognitivist NOT interested in cerebral anatomy?

In the rather disorienting context of some of the discussion, it also
struck me that two people I have respected for many years in spite of
severe local disagreements on occasion -- Fritz Newmeyer and Joan
Bresnan -- have been making honest attempts to be just that,
functionalist cum structuralists. So you may quibble with some of their
specific conclusions, as I sometimes do; but you've got to respect their
honest attemp to escape the stranglehold of *reductionism* that Chomsky
has saddled us with.

The correlate, of course, is that just because Chomsky was (and still
is, alas) an extremist, we need not ignore the *many* things that we did
learn from him. So I wanted to enumerate just a few, some of which I
shared previously (and privately) with Wally Chafe & Brian MacWhinney.

This business of all of us trying to find as many ways of saying either
I love you, NC or I hate you, NC, sure reminds my of the funeral oration
in Julius Caesar ("We have not come to praise you, NC, but to bury...").
Sure, we all know the many ways in which NC made linguistics a rather
miserable morass. But at least for my generation, he also saved us from
the Bloomfieldians, who were in some ways just as bad, in others even
deadlier and even more dogmatic about the irrelevance of meaning,
function & mind.

Perhaps it would also help to mention that NC, rather paradoxically,
engineered our generation's return to functionalism: Aspects (1965)
fairly reeked with semantics, both propositional & lexical. Sure, it is
all couched in obfuscatory structuralist jargon, but it's still there.
Sure, the 'format' was licenced by Fillmore (1962) and Katz & Postal
(1964); nothing really original (but the jargon...).

But still, NC embraced it, to his great eventual sorrow -- since Ross &
Lakoff's paper (1967) "Is [syntactic!] deep structure necessary?"  was a
direct consequence of the 'semanticism' of Aspects, simply drawing it to
its ultimate conclusions & exposing its incompatibility with the rest of
the structuralist machinery. I know that was the point that licensed me
to bolt. And if I am not mistaken, Generative Semantics was directly
licenced by it. And maybe even Wally's 1970 book "Meaning & the
Structure of Language"? Though Wally was probably old enough by then to
have harboured those ideas earlier... Sapir? Whorf? Common sense?

Even the much-maligned notion of "deep structure" had its salutary
consequences -- given its historical context. It focused attention on
the semantic correlates of syntactic constructions. Those stock
sentences -- "Sally is easy/eager to please" and "Flying planes can be
dangerous", etc. etc. -- played an important role in demonstrating that
syntax had semantic correlates. And in fact, many of NC's (and Pstal's)
arguments agains the IC analysis prevalent at the time actually hinged
on semantics (even if he didn't say so), as did R.B. Lees' arguments in
"The grammar of English nominalizations", even if he didn't say so
himself.

And, for that matter, even Harris's original 1956 paper ("Co-occurrence
and transformations..."), where semantics was hiding under the
forbidding, empiricist/structuralist moniker "co-occurrences".

Which brings me to *transformations*: One of Noel Rude's contributions
misrepresented what I think/said/thought about this issue, so here is
the real thing: True, transformations obfuscated a lot of issues. But if
you read Aspects carefully (which I do, with my grad students, once a
year), a curious thing may strike you: As a "process" T-rules surely
*are* a mess. But as RELATIONS between structures they are most
revealing: The very same propositional-semantic contents persists, as
leitmotif, through multiple syntactic structures ("transforms"). With NC
reminding us (following Fillmore's 1962 paper in WORD...) that
"transformations don't change meanings". But -- we asked ourselves in
the late 1960s, if they don't change meaning, what do we use them for?
What is their FUNCTION? And the obvious answer was (with some help from
Joe Emond's dissertation, another structuralist classic that contributed
to my development as a functionalist... and with some help from a
WONDERFUL paper by Joan Hooper/Bybee and Sandy Thompson with supplied a
functionalist interpretation of Emonds) -- they must be there to code
discourse-pragmatic (communicative, interactional, etc.) function.

And so, I am almost tempted to say, NC licensed semantics quite in spite
of
himself, to his own eventual sorrow. Which brings to (the wandering)
mind the Pythia's warning to Xenophon (when he tried to cheat on going
to join the rebellion of Cyrus against the Persian empire):

      "Invoked ot uninvoked, the God will be there".

Finally, there is one more area that NC should get (grudgin) credit for
-- the more sophisticated notion of *syntactic constructions*.
Bloomfieldian
syntax was a veritable mess, mostly morphology and non-hierarchic IC.
The notion of constructions, with constituency AND hierarchy AND
embedding -- clearly isomorphic to what I see as grammar-coded
*functional domains* (communicative pragmatics) -- was really not easy
to derive from from Bloomfieldian IC analysis. It is, in my  humble
estimate, NC's pairing of deep structures with propositional-semantic
interpretations that licensed the next move by *functionalists*; or at
least by functionalists who grew up in NC's incubator...

It is of course ironic that by 1992 NC denounced this very notion, for
which SS (1957) and Aspects (1965) were largely responsible, in effect
calling constructions figments of our methodological imagination. To
wit:

     "...The notion of grammatical construction is eliminated, and with
it,
      construction-particular rules. Constructions such as verb-phrase,
      relative clause, passive, etc. are taken to be *taxonomic
artifacts*,
      collections of phenomena explained through the interaction of
      of principles of UG, with the values of parameters fixed..."

      ("A minimalist program for linguistic theory" (1992), p. 3;
       emphases added)

This is truly bizare, granted, me coming to the rescue of  REAL
*structure* from NC, who has decided to finally dump them, 'cause they
kept sticking in his craw, couldn't swallow them, couldn't spit them
out. Major bummer dude.

But still, have a heart, y'guys: For those of us who believe that
grammar is cognitively real, (and neurologically real, if you don't
mind, Liz & George...); and that it involves not only morphology; for
us, Aspects was never a total waste of time. At the very least, it gave
us something to work from, build on, transcend, bounce off (as George &
Haj did in 1967), eventually escape from (boy, those strictures...).

Well, actually, there was one giant around, and we could have learned
from him about REAL grammar -- Otto Jespersen. But nobody told us about
him, and Bloomfield dismissed him together with his (and B.'s) teacher
Herman Paul, as speculative philosophers, not *real* scientists. And
yes, Dwight Bolinger was around, but we didn't know he existed (most of
us found him in the 1970s). So who else was round (in the US) to teach
us about the semantic/pragmatic correlates of grammar? Perversely, we
got it from Chomsky, tho we had to escape his strictures to get it
straight. But I don't see how we could have gotten it without Aspects. I
do go back every so often to the high shelf, in the left corner, and
pull out & re-read the Joost collection (1962), as a refresher, just to
remind myself how truly deadly, indeed suffocating, and above all smug,
the atmosphere was in US linguistics in 1956...

I apologize for taking so much of your time.

Happy New Millenium, y'all, Brave New World! TG