Dick Hudson wrote:
>
> >-- and this is indeed how things work for 'true constructions'. But with
> >fixed expressions you'd expect that the greater the "textual deviance"
> >from the "textual prototype", (1), the less the acceptability. Now it would
> >be stupid to add rules to the grammar that rule out 2-9 case by case,
> >when the true explanation is the notion of "textual deviance" from
> >the prototype. So what WG needs to do is find a way to describe
> >a "textual prototype" that is distinct from a true construction (like
> >though-movement).
> ## That's an interesting distinction which I think I can understand in
> principle. A string of words could be either generated pairwise by the
> grammar, or remembered en bloc without any help from the grammar.
> The latter would be a textual prototype, and I suppose it could be
> used as the basis for forming other slightly different strings by
> applying "analogy". But in practice, I don't understand how textual
> prototypes could exist. After all, as soon as we recognise distinct
> words we're applying some kind of grammatical analysis.

On the basis of her experimental results, Bock argues that even set
phrases (e.g., idioms) analyzed syntactically.

 J. Cooper Cutting & Katherine Bock. 1997. That the Way the Cookie
   Bounces: Syntactic and Semantic Components of Experimentally
   Elicited Idiom Blends. Memory and Cognition, 25(1): 57-71.

This makes sense from a network perspective, since wholes activate
their parts.


> >I'll be interested to see your response between my attempt to distinguish
> >a "textual prototype" from a "true construction". Knowing you, lumper that
> >you are[*], you'll be wanting to deny the validity of the distinction. But
> I've
> >got my fingers crossed that you'll be able to see my point... Anyway, using
> >that distinction, I could say what I said before as: Joe wants to handle
> >all knowledge of language in terms of Textual Prototypes, while you want
> >to handle even the marginal fluff in terms of True Constructions.
> ## You're right. Lumper that I am, I think it's true constructions all the
> way down! I'll be interested to hear Joe's response, another lumper that he
> may be.

Not only is it constructions all the way down, but it's usage-based
constructions all the way down. And just as there are no outright
boundaries between syntax, the lexicon, and morphology (just think
of certain compounds, e.g., linguistics students), there are no
outright boundaries between rules and data (usage).

I think that a lot of people would agree that we only have
schematizations (rules) by virtue of all the concrete exemplars we
have acquired first. When a particular pattern is especially productive
(high type frequency), then the individual exemplars become much more
difficult to access if they have a low token frequency. In-between cases
are those expressions with mid-to-high token frequency that instantiate
a mid-to-high type-fequency schema. For example:

--Don't just sit (stand) there! Do something!
--You can say that again.

Surely these are stored as wholes, but that doesn't mean that we
don't also analyze them. In fact, even a "nondecomposable" idiom
such as "kick the bucket" is analyzable to some degree; otherwise
we wouldn't be able to say: "kick the *proverbial* bucket."

Constructions are usage-based in another sense: they are learned in
(situational) context. In other words, when we learn a construction
we bring along all the relevant contextual knowledge, sociolinguistic
or otherwise. "Core-grammar" constructions tend not to have concrete
pragmatics, because they are used (and learned) in many different
situation types. On the other hand, unusual syntactic configurations
tend to have special conventional implicatures, precisely because
they are used (and learned) in fewer types of situations.

In short, I don't think that the dichotomies And has been proposing
are dichotomies at all. In my opinion, what And is trying to save
is the notion of disembodied grammar. Despite their rhetoric about
doing "biolinguistics," Chomskyans are doing the same thing, arguing
not that grammar is comparable to biology, but that it's comparable
to physics. Having said this, And strikes me as intellectually much
more honest, since his position is in keeping with his skeptical
outlook on mentalist frameworks.

Joe
_________________________________________________________________
Home page:        http://lingua.fil.ub.es/~hilferty/homepage.html
__________________ http://www.ub.es/filoan/hilferty/homepage.html