Nik:
> Part of this lists' usual semantics debate comes about because people
> here have
> very different views about where complexity is to be found in grammar. For
> example, on the whole And thinks that complexity is found at the
> phonology-syntax (or phonology-syntacticosemantic) interface, whereas others
> find it at the syntax semnatics interface. Obviously a lot of this
> comes down to
> intellectual taste and before the event, philosophy of inquiry motivated,
> position taking. Part of the issue concerns what idealisations we're making.
> When And says that he wants to have lexical decomposition managed as relations
> between *words* this is partly because he is highly tolerant in his theorising
> of phonology-syntax mismatches, and higly intolerant of syntax-semantics
> mismatches (which is partly why he takes his reductive view of semantics.
> although this is also partly predicated on his anti-conceptualist
> stance). I am
> relatively intolerant of phonology - syntax mismatches. I accept a certain
> amount: you're forced to by phenomena like clitics and perhaps mixed
> categories
> (though perhaps not the latter). But I am highly suspicious of highly
> mismatched
> phonology - syntax relationships. You could, for example, quite
> logically end up
> arguing that there were not necessarily any syntactic relationship word-order
> rules (as the Burton-Roberts team do, I think) when the data make it look as
> though there is a clear relationship between morphosyntax and word order.

I don't understand how NBR et al get that move -- no linear order in syntax
-- to work, but if they could, I wouldn't find it objectionable, since
superficial order would be the phonological expression of some other
structural syntactic property.

I don't really understand the intellectual rationale for objecting to syntax--
phonology mismatches. We know they must occur in principle, and there's no real
learnability problem, since learners know both the sound and the meaning,
so are in a position to work out the mapping between them.

I'd be more sympathetic to an attempt to integrate syntax and phonology
(though I think it would be doomed to failure) or else (better) to insist
that non-phonological words don't exist and that syntactic categories
classify things (i.e. DOG doesn't exist, and Dog is a Noun). At least
those approaches would be striving for parsimony more.

As you recognize, the choice of where to locate complexity is more than
a mere matter of taste. If, like me and Noam, you think of language/grammar
as building structures to connect two interfaces, beyond which the
grammar cannot see, then the mapping between the two interfaces,
phonological and logical form is the only possible locus for complexity.
If, like Dick, grammar is no more than our memory of usage, and is in no
wise and to no degree whatever autonomous from the rest of cognition,
then at the very least there is no reason not to locate complexity at
the mapping from syntax to meaning, though I think that even in a
Dick/Joe/Jasp framework we can ask whether we really need a level of
syntax/logical form to mediate between phonology and meaning.

--And.