Dylan:
> I couldn't agree more! A central 'issue' which has plagued me since I hea=
> rd of=20
> the big C is the implicit assumption that all [word-order] variation is=20
> syntactic in origin. I feel very strongly that the extreme abstractness s=
> tems=20
> from trying to foist a syntactic solution onto a non-syntactic phenomenon=
> =2E
>
> Much of the abstract scaffolding (AGR heads and other malarkey, for examp=
> le)=20
> seems to me a fudge to synthesize a leanear order in an essentially=20
> non-linear framework. Linearity arises, to my mind, in the fact that the=20
> speech-stream is linear. Thus, the phonological component and the parser =
> can=20
> have just as much impact as syntax on areas such as rightward movement an=
> d=20
> head-complement ordering. For example, the two structures:
>
> a) [[give [the book]] [to Paul]]
> b) [[to Paul] [[the book] give]]
>
> are structurally (=3Dsyntactically) identical, so the ordering (reflectin=
> g=20
> English and Dutch, for example) must arise elsewhere. A strict Chomskyan=20
> would have to twist the syntax to generate the order.

Another way to say this is that the need for all the word-order
variations of different languages to come from one source is another
problem with TG syntax.  There have been various moments in the history
of TG when absurd solutions have been proposed: McCawley's English is a
VSO language, the all lgs are SOV of Haj Ross is another, and the one
which gave rise to the current garbage pit which Dylan is referring to
is the last (one hopes):  J.-Y. Pollock's Verb movement, Universal
Grammar and the structure of IP (LI 20: 365-424 (1989)).  This is the
paper which led to the various AgrPs and to TPs (Tense Phrases), and
the like.  (The seeds of this of course are in Bloomfield's innocent
and tentative suggestion that syntax need not stop at the word, i.e. in
addition to accounting for word-order variation modern TG seems to want
to account for morpheme-order variation as well.)

Chet