Dick:
> >#What isn't that doing here?
> >#What is that not doing here?
> >#What is that probably doing here?
> >#What is she doing sitting outside and doing reading a book?
> >
> >(cf.
> >What the hell is that doing here?
> >What is that suddenly doing here?)
> ## Good points again. I imagine some of these exclusions can be explained
> pragmatically, in terms of the special meaning of WXDY - it presupposes
> that XY is true (and odd), so why would one want to make it negative or
> only probable?
I agree. Only the last example (coordination) really calls for
an explanation.
> >When the idiosyncratic properties of a particular construction are special
> >properties lacked by the more general constructions, then I think the
> >WG approach is correct. But when the particular construction lacks,
> >for no good reason, general properties that you would otherwise expect
> >it to inherit, I think we're looking at a completely different phenomenon
> >-- the phenomenon of idioms, fixed and formulaic expressions, frequent
> >collocations, and so forth.
> ## As usual, I prefer a unitary account so I'd need very good evidence
> before dividing the grammar into qualitatively different areas such as
> general vs special. I don't think WXDY makes the case.
I'm not advocating dividing the grammar on the basis of general vs
special. Rather, I think that external to the grammar is our memory of
usage 'module', where we remember common interpretations, common expressions
and so on. Idioms and constructions that have no special syntactic properties
exist only in the memory-of-usage module.
I see this module and the grammar module as different becuase the grammar
is made up of abstract rules and is generally governed by principles of
simplicity etc, while the mem-of-usage module is just an absorbent memory
and statistical analyser, the source of our sense of style, register,
naturalness, etc.
Contrast this view with Joe's, which wants to shift all of the grammar
module into memory-of-usage, and with Dick's, which wants to shift indefinitely
large amounts of memory-of-usage into grammar.
--And.
|