Is OT a theory of syntax?

[Daniel L. Everett]

Joan and all,

I do not disagree with Nigel in the statement that OT is not a competitor 
theory. That is not the issue. Nor is it really an issue that OT's input 
and output are not really derivational. 

Nevertheless, once you are comparing candidate representations from a set 
provided by GEN you are able to manipulate the mix depending on your 
assumptions of what the 'input' is (understood nonderivationally).

This means that a certain way of translating theories into OT must be 
assumed. And it is what I said it was in the last message, modulo 
confusion I may have introduced about derivation.

The selection of inputs is a vital component for translating theories into
OT. Consider how OT might, for example, evaluate nonce forms (e.g. in
testing for native speaker intuitions of possible words - as in Everett
and Berent in progress on Hebrew [see an early, rough version of what we
are getting at in our posting to the OT list]).  If speakers reject a
nonce form, then one way to account for their behavior is to assume that
they first constructed a hypothetical input, reasoning backwards from the
output (along the lines Smolensky has argued for L1 acquisition in his
work on Richness of the Base, except without certain additional factors
involved in language acquisition, e.g.  where additional evidence can
override markedness constraints to rank faithfulness constraints over
markedness). The only way to account for their behavior is to get the
right 'input' to match the 'output' - and the particular set of
theoretical assumptions being thought of in this way, RG, LFG, Minimalism,
are irrelevant for this point - any theory will have to translate some of
its notions into this way of thinking (although, frankly, I think that an
OT version of Minimalism is less likely to work, since Minimalism
operates, at least in Chomsky 1995, with some OT-incompatible assumptions,
since it has violable local constraints, inviolable global constraints,
inviolable local constraints, and violable global constraints - see my
forthcoming review of Hornstein's Logical Form:From GB to Minimalism, in
Language. This criticisms may not apply to Chris Collins' version of
Minimalism, but the theory sketched in his MIT LI series book is too
schematic yet to tell for sure.)

Anyway, I don't think I disagree all that much with Nigel, if at all. 

-- Dan