OT, functionalism, etc.

[bbergen at socrates.berkeley.edu]

Since the cents seem to be flying, I'd like to bring up a couple critiques
of functionalist OT. I apologize if this commentary is on the long side,
but I think that this position has been inadequately expressed in the
recent discussion on this list.

First, like other "natural" phonological models, it is very difficult for
functional OT  to explain " unnatural" phonological behavior. Juliette
Blevins and Andrew Garret have recently shown for consonant harmony and
metathesis that is is essentially impossible to explain why these
processes have variant behavior on the basis of synchronic natural models.
Instead, synchronic idiosyncracies result from regular, natural diachronic
developments.

The problem is not that there are no "natural" behaviors in synchronic
phonology - the problem is that trying to explain all synchronic
phonological properties by appealing to "natural" effects obscures an
essential question: what properties of phonology are a direct result of
(caused by, explained by) the synchronic grammar (including, perhaps
functional constraints), and which are historically contingent (although
historically functionally motivated)?

Another questionable property of OT relates to issues brought up by Brian
MacWhinney. OT has become an extremely powerful device. Such notions as
Sympathy, Overlapping Constraints (Hayes), Output-Output correspondences
(e.g. Benua), integration of phonological and extraphonological
constraints, and levels of OT derivation have rendered OT essentially
omnipotent. This makes it difficult to call it an explanatory theory. Of
course, there are various attempts to reduce its power, but the real
problem, I think, lies in the very notion that this OT grammar is actually
explanatory of the language it models, independent of the "semantics" of
the constraints. (Although I do recognize that Joan Bresnan disagrees
with me here, I simply mean that there is no way in which the
extra-constraint properties of functional constraints enter into their
behavior within an OT model.).

I would bet that most functionalists agree that whatever capacity for
language individual speaker-hearers have, it is extremely powerful,
and that certain restrictions are levied on what form a phonology can
have by factors external to it. Some of these restrictions will come from
the typical sources - processing constraints, neural architecture
constraints, information processing constraints, physiological
constraints, cognitive representation constraints, and so on. It might be
a problematic step to reverse these roles as functional OT seems to - the
grammar becomes explanatory, and the constraints have very little
explanatory role.

On the other hand, I couldn't be happier that the interface between
phonetics and phonology is being closely scrutinized. Functional OT may
very well provide the kind of theoretical framework required for good
empirical work to get done.

Ben Bergen

-------------------------------------------
Benjamin K. Bergen
Graduate Student
Department of Linguistics
U.C. Berkeley
www.icsi.berkeley.edu/~bbergen
-------------------------------------------