back and forth

[Tom Givon]

Seems to me I hear another argument between the partly deaf. Everybody
concede that a correlation exists between grammatical structures and
semantic and/or pragmatic functions. But two extremist groups seem to
draw rather stark conclusions from the fact that the correlation is
less-than-fully-perfect. The "autonomy" people seem to reason:
     "If less than 100% perfect correlation,
      therefore no correlation (i.e. 'liberated' structure)"
The grammar-denial folks, all of them card-carrying functionalists,
seem to reason:
     "If less than 100% generative and autonomous,
      therefore 100% functionally motivated"
Now notice that both use a very similar Platonic reasoning:
     STRUCTURALISTS: "Something is functionally motivated
                      only if its is 100% so"
     ANTI-GRAMMFUNK: "Something cannot have independently-manifested
                      existence unless it is 100% so"
This is of course a silly argument, and that's why people keep repeating
the same reductionist extreme assertions again and again and again.

The facts suggests that neither extreme positions could be right. Grammar
is heavily motivated by semantic/communicative functions. But -- because
of grammaticalization and what Haiman calls 'ritualizatio' it never is
o100% so. It acquires a CERTAIN DEGREE of its own independent life.
This, however, does not mean 100% autonomy. And by the way, along the
diachronic cycle of grammaticalization, you can see a construction
changing its degree of iconicity. So that overall in the aggregate of
a synchronic grammar at any diachronic point, you can show constructions
that are much better correlated to their functions and some that are
much less so.

To those of you who know something about biology (and sorry Fritz and
Elen, I don't cont you among those...), this story of course looks rather
familiar. It also looks familiar in respect to another topic Ellen raised
(without calling it by its rightful name --  that of cross-language
typological vartiation. Yes, we do see the same communicative function
coded in different languages coded by different grammatical constructions.
That is because there is not only one way to perform a function, there
are alternative choices. Think about the function of AMBULASTION. There
are four major types bio-organisms seem to perform it: Walking, slithering,
flying and swimming. And among each of those there are minor sub-types.
And each one of those is associated wiuth its own -- highly adapted,
rather specific -- correlated structures. Now, does that mean that
structure is independent from function? Get real.

Bloomfield and his cohorts thought that cross-linguistic typological
variability suggested structures were 100% unconstrained by meaning. But
most serious typologists are keenly aware of how CONSTRAINED is the
range of syntactic types (major ones) that can perform the same
communicative functions. IN REL-clauses, for example, I have not
been able to find more that 5-6 major types. In passivization
(impersonal agent) maybe 4-5, etc. etc. This extreme paucity sould
be appreciated against the vast number of mathematically-possible types.

The idea that 'universality' means 100% universality is another version
of reductionist Platonic thinking. Species, biological populations, are
defined by evolutionary biologists (see Futuyma's text, eg) as A CURVE
OF DISTRIBUTION OF VARIANTS. These guys will tell you that VARIATION
GUARANTEES EVOLUTION. So the pernicious idea that somehow universality
demands 100% uniformity is really a bit primitive. As is the idea that
because there is more than one way of skinning a cat, ways of skinning
are not closely dependent upon the task at hand -- skinning a cat.
But of course, Aristotle in his brilliant rejection of structuralism
in Biology already said all that. One sometimes wonders why after
2300 years it seems nobody is listening.

Y'all be good y'hear. TG