"progress" in linguistics

[Tom Givon]

NOTE:
       THESE REFLECTIONS HAVE BEEN INSPIRED BY AN (ALAS BELATED)
       READING OF ESA ITKONEN'S "A UNIVERSAL HISTORY OF LINGUISTICS"
       (1991; AMSTERDAM: J. BENJAMINS). OBVIOUSLY, I COULD NOT RECOMMEND
       THE BOOK MORE ENTHUSIASTICALLY. IT IS A RARE TOUR DE FORCE. WHAT
       FOLLOWS BEGAN AS COMMENTS DURING CORRESPONDENCE WITH ESA, WHO OF
       COURSE BEARS NO RESPONSIBILITY FOR THE WAY I CHOSE TO INTERPRET
       HIS WORK. STILL, HE HAS MY THANKS.  TG



Dear Esa,

I have now made a brave if incomplete stab at both your Panini section &
the conclusions in ch. 6 . Panini has always been a rough going for me,
so I am thrilled at your determination & erudition in attempting to
decipher it for ignorami like myself. I wouldn't have been able to do it
on my own, not having enough patience (not to mention being burdened
with an atavistic recoil from obscurantism...). But I think I got the
point. Panini was just as sophisticated as any descriptive linguist of
the 19th Cent. tradition (yes, that includes us).

Having benefited so much from your tour-de-force of Panini, the Greeks &
Hermann Paul, I still wonder about your conclusions (ch. 6), in
particular the
discussion of progress in science and progress in linguistics.

Let me dispense with philosophy first. To my mind, Kuhn--and especially
a radical interpretation of Kuhn that foregrounds scientific revolutions
and downgrades 'normal science' (and makes a huge deal about the radical
difference between those two phases of science)--is univeresally
accepted neither in philosophy of science nor in many disciplines of
science itself. Most 'real' scientists I talk to (mostly chemists and
biologists) are much more conscious of the gradual nature of their
disciplines, the slow accretion of factual, methodological and
theoretical knowledge--even when great leaps occur, as they undeniably
do. It is not an accident that in Biology, Kuhn was embraced only by a
small tho media-savvy & articulate fringe of Marxist anti-graduality
ideologists--indeed, to my thinking, anti-evolutionary ideologists--like
Gould & Lewontin.

It is not an accident either, I think, that in Linguistics Kuhn was
embraced in by the Chomskians (and other self-perceived great
revolutionaries...). And it is not an accident either that most of the
overblown 'revolutionary
science' examples cited by Kuhn (or the Chomskians) come from Physics,
the most mathematics-dependent and the most deductively-driven of all
sciences. But I suspect that even in physics, Kuhn's sweeping taxonomy
of the two kinds of science represents a vastly over-blown
generalization (certainly Lakatos does not accept such a clean
taxonomy).  Tho admittedly, 'great revolutions' are easier to
demonstrate in physics, and thus help sustain the revoilutionary
philosophical biase.

In linguistics, it is precisely the people who view deductivism &
physics as the proper models for science that espouse the redical
Kuhn-type view.
Every time such a linguist farts, an inspired new theory is born. The
philosophy of science articulated by pragmatists such as Peirce & Hanson
does not make such a radical distinction between 'normal' and
'revolutionary' science. To begin with, it recognizes the fact-driven
nature of hypothesis formation (not Popper's 'miracle'...)--tho not via
induction but via abduction. What is more, such an approach also rejects
the strict Positivist (Carnap, Russell) division between factual and
theoretical statements. Rather, it concedes the theory-laden nature of
'facts' as well as their dynamic nature-- "Today's facts are yesterday's
theories". Such an approach is much less likely to be swept off its feet
by radical 'revolutions', not because it doesn't recognize the
profundity of new abductive insights (hypotheses), but because it sees
them as embedded in their historical & conteporary--often
cross-disciplinary--context. The best way I can remind all of us of this
is by quoting Peirce: "...Any philosophical doctrine that should be
completely new could hardly fail to prove completely false..." (1940, p.
269).

Now let us turn to your observation (ch. 6) that no 'real' progress
occurred in linguistics between Panini and 1960. Let me first tell you
why I am
sympathetic to this rather sweeping generalization--up to a point. One
thing that really marks the difference between a pre-scientific & a
scientific investigation of the very same domain is a commitment to
*methodology*. From such a commitment spring
both new data-bases and new hypotheses. Here Panini represents the best
of our
"competence" methodology, thus the best of arm-chair linguistics, thus
the best of
Plato-cum- Aristotle. The facts in such a method are crystal clear and
available to conscious reflection & analysis. Both intra-subject and
inter-subject variation is ignored as either methodologically or
theoretically irrelevant. The analytic method is heavily tipped toward
deductivism. This is very similar to Physics and inorganic chemistry
(you've seen one H2O atom, you've seen them all...).

Now, descriptive grammarians from Panini to the end of the 18th Century
(not 1960!) practiced this analytic "competence" methodology. They
applied it to
single languages, and the same *types* of facts were analyzed by the
same types of method. Should one be surprised that at the lack of
progress? Does one see much progress in philosophy over the same period?
(or, for that matter, up to the present?)

It is also good to remind ourselves that there was relatively  modest
*theoretical* progress in Biology between Aristotle and (ca.) the 16th
Century--and for very similar reasons. Yes, people did very useful
*taxonomy* work, vastly enriching Aristotle's initial *scala natura*.
But the general thrust of the Linnaean taxonomy--the apex of that
protracted effort--was really not all that different from Aristotle's.
No insightful new explanations were forthcoming (except for the
tried-and-true God, or His Romantic stand-in, *force vitale*...).

Eventually, the conflation of new methods cum new data-bases did occur,
and that's what (I think) stimulated the eventual theoretical
'revolution' (Darwin): The fossil record and the mothods of Geology;
Microbiology. But still, it was the methodological/factual expansion to
detailed micro-variation within species (Galapagos!), on the one hand,
and the link to adaptive *behavior* (the birth of what eventually became
ethology), borrowed from another discipline (Political Economy; Darwin
was reading Malthus & Adam Smith) that eventually prompted the new
theoretical perspective(adaptive selection; competition for limited
resources). And it was the later merger of chemistry and biology into
molecular genetics that finally clinched the rest of the mechanism (the
site of mutations, thus the source of spontaneous variation). So from my
perspective, there was a profound thoretical stasis in biology
bewtween 300BC and 1600AD.

Now, why do I think scientific progress in linguistics started in the
19th century (rather than in the 1960s)? Two reasons, both of them well
known to you.

(i) UNIVERSALS: Till the 19th century, all Panini-type descriptive
grammarians may have *assumed* universals implicitly, but they did not
systematically study cross-language (typological) variation. They
described single language but did not systematically compare them. The
advent of the expanded data-base in the 19th Century, first within IE
and soon across-families, provided a much more realistic factual
benchmark for raising the question of universals anew--beyond Plato &
Aristotle. Whether the early answers by Schleicher, von Humboldt,
Bloomfield, Sapir/Whorf etc. were successful theoretical manoeuvers is
almost beside the point. We did not had enough cumulation of data-base-
cum-methodology to raise this issue seriously before the 19th century.

(ii) DIACHRONY & FUNCTIONALISM: Until the 19th century, change was not
the focus of the study. But it is precisely the expansion of the
data-base-cum-methodology in the 19th century to *diachrony* that
allowed Hermann Paul to make his inspired  generalizations about the
real locus of the "causal nexus" in an explanatory theory of Language
(rather than descriptions of languages)--the
cognitively/adaptively-driven *process* of on-line communicative
*behavior*. This is, surprisingly, a synchronic perspective, but a novel
one--not of competence, but of *performance*. It is thus not only a
shift from stasis to process thinking that made the difference, but also
from competence-based to performance-based methodology. (In saying this,
I merely interpreting your own description of H. Paul's work).

Now, have we progressed far beyond H. Paul since 1960? First, I think
Jespersen was already right there with Paul, I see no serious retreat
(tho no advance
either). Was Saussure progess? I don't see how. How about Bloomfield &
Chomsky (or,
for that matter, the European structuralist)? I am tempted to respond
with
*ditto* again. While one may have a lot of respect for Bloomfield the
descriptivist (the Panin-type Bloomfield), as a theoretician he
represents the same general retreat from H. Paul (his teacher?) as
Saussure.

So what actually started in the 1960s, really? Maybe two or three things
that represent a more systematic--methodological--return to the
prophetic vision of H. Paul. First, a much more broad-gaged expansion of
the cross-linguistic (typological) data-base; albeit with an implicitly
Bloomfieldian biase toward "inductive generalizations".

Second, incompletely and haltingly, the realization that typology *is*
diachrony, so that the locus of explanatory universals (Paul's "causal
nexus") is in the process of grammaticalization. But here again, not
many of the grammaticalization people understand the implications of
this. Most of them are still looking for Bloomfieldian "inductive
generalizations"--i.e. a taxonomy of grammaticalization types. And while
"emergence" is a wonderful battle cry, it is not yet a clear research
programme. And one of thesae days we'll need to moved on from
methodology to theory. Taxonomy is very useful (hooray for Linnaeus!),
but it is not exactly an ambitious theoretical agenda.

Third, and again only dimly & haltingly, the beginning of some
*cognitive* explanatory theory, at our interface with experimental
psycholinguistics
and neuroscience. This was, in essence, the third leg of Paul's research
programme--the psychological "causal-nexus" hinging on processing
*behavior* (performance). And in the 1960s it begand to--slowly,
bashfully--rear its delicate head. But again, the majority of linguists
who descroibe themselves as "cognitive" or "functionalist" are still
dyed-in-the-wool *competence* philosophers. They expresse their
Panini-like generalizations in cognitive-sounding terms. These terms
have been either invented by inspired but still arm-chair linguist, or
they are 100% isomorphic to structural categories and thus have no
independent mtrhodological validity (except for our endearing faith in
100% iconicity...).

So yes, the 1960s did represent a return to H. Paul's
potentially-scientific theoretical agenda. But we are still at the very
threshold of this process. Most
of us are still--hopelessly, incurably--methodological humanists
('competence Platonists'?). Science is still an alien planet to us.

So, has there been any progress? Will linguistics become, finally, a
responsible science? Three moves could help us accelerate the shift from
Panini/Chomsky:
   (a) Interaction with neighboring adaptively-based disciplines
       (like Darwin).
   (b) Commitment to a performance-based empirical methodology
       (like Labov; like psycholinguistics)
   (c) The conflation of diachrony, universals & performance
       (H. Paul's agenda).


Best & many thanks,  TG