good news from generative grammarians

[Robert Levine]

I think Carl's note captures and corrects very concisely the rather
large array of category errors in KPM's observations about theory
comparison in linguistics. I think it's worth stressing the erroneous
nature one of the points in particular: the idea that transformations
represent some kind of procedural implementation, that they are an
algorithm of some kind. Derivations are not aspects of implementation,
which is what the procedural/declarative distinction refers to. A
theory with transformations pairs with each sentence a partially
ordered set of representations with successive pairs of
representations in the ordering satisfying constraints on structural
descriptions defined in terms of Boolean conditions on
analyzability. You can implement a such a grammar purely
declaratively. You can implement it procedurally. You can implement a
grammar specifying for each sentence a single structural object
associated with multiple attributes at each node declaratively or
procedurally. Derivations and procedures are fundamentally different
things and, as Carl stresses, the implementation style is irrelevant
to the issue of properties of the grammar itself.

Another point which merits some discussion is the idea that there is a
legitimate distinction between generative and nongenerative theories
depending on the ontological interpretation of the theory. In KPM's
view, apparently, two totally separate notions---the issue of
psychobiological realism on the one hand and of explicit enumeration
of a set by some set of formulae in an explicit description language
on the other---are legitimately conflatable, even though the two are
entirely orthogonal to each other in terms of what they are
*about*. Try the following analogy: we have a distinction
`rational/irrational' on the one hand and a distinction
`classical/Intuitionistic' applying to logic on the other. A
Brouwerian of unusual zeal argues that there is evidence that the way
people actually think does not employ classical negation but rather
something corresponding to Intuitionistic negation; therefore, the
latter may be classified as rational and classical logic as
irrational. Would anyone seriously argue that this is a legitimate use
of the term `(ir)rational', or that it would be if `most people', at
least most Intuitionists, equated `classical' (logic) with
`irrational'??

Bob