* Sunday, 2011-12-04 at 10:13 -0300 - Jorge Llambías <jjllambias@gmail.com>:
> On Sat, Dec 3, 2011 at 10:49 PM, Martin Bays <mbays@sdf.org> wrote:
> I'm pretty sure the intention was that implicit quantifiers on "da"
> were mere elisions, so "ge ko'a gi da da broda" would have to be
> either "(su'o da zo'u) ge ko'a gi da da broda" or "ge ko'a gi (su'o)
> da (su'o) da broda". It couldn't be something that needs an expansion
> before being made explicit.
>
> > If we're to let scope jump out of geks, why not also out of NOI-clauses
> > or NU-clauses?
>
> I don't think anything should be jumping out of anywhere, it's just a
> matter of where the elided quantifier is in the first place.
You seemed to be suggesting rules which would give {ge da broda gi de
brode} -> {su'o da zo'u ge da broda gi da brode}, no? I'd call that
jumping.
And presumably this rule actually *would* have us jumping, in this same
sense, out of NOI or NU? e.g. {mu'i lo nu da kukte ku mi citka da} ->
{su'o da zo'u ...}?
I suppose my main complaint is just that this rule is severely
garden-pathy, in a way that's even worse than {i na je}. Consider the
case that a sentence starts with something like {mu'i lo nu na ku da
kukte}. Then we've set up a semantic bomb - and we won't know whether
it's going to be triggered by a later bare {da} - pulling the first past
the negation and wholly changing the meaning - until the end of the
whole statement. This seems unnecessarily cruel!
> There are two separate issues to consider: (1) Where is the binding
> quantifier when an expression contains apparently unbound variables?
> (2) What is the scope of an explicit quantifier which is not presented
> in prenex form?
>
> I don't think there's more than one reasonable answer to (2). Saying
> that "no da blabi .i je no de xekri" means something different from
> "ge no da zo'u da blabi gi no de zo'u de xekri" seems just
> unreasonable. No quote from CLL can make it reasonable. The fact that
> you need to use tu'e-tu'u if you want to move "no da" to a prenex
> while maintaining the non-prenex ijek connective form is just
> incidental.
So you really consider ijeks as just a different notation for giheks?
So e.g. you'd have {ro da muvdu .i na ja ko celgunta da}
mean "if everything moves, shoot something" rather than "if anything
moves, shoot it"?
I don't know. Certainly this is against CLL, however little you may care
about that; and I don't see that the CLL is unreasonable here. If we
want your meaning in your example, we can use {no da blabi .i no de
xekri}.
> Question (1) may admit more than one reasonable answer. The simplest
> answer seems to be that the elided "su'o" is right in front of the
> first instance of the apparently unbound variable, with scope as in
> (2), and any instance outside that scope will require new binding.
> Another perhaps reasonable answer might be that the elided "su'o" has
> scope wide enough to capture as many instances of the apparently
> unbound variable as possible. I don't find it so reasonable that there
> be no possible place to make "su'o" explicit in the expression as
> presented.
I can see that. To be clear (because given your example at the top, I'm
not sure we already have clarity), I'm effectively disputing this only
in very rare edge cases - those of the form {[PA] da .A ko'a da}, and
those of the form {GA [PA] da broda gi brode vau da}. I doubt the
designers had these cases in mind.
In both cases, the issue is that the "simplest answer" you mention
doesn't give an answer - I would replace the indicated "[PA]" with
{su'o} if it's empty, but then there's no answer to the question of
whether the second {da} is in the scope of the first. So should it also
get a {su'o} and hence be rebound on both arms of the connective? Or
only on one? I don't see that there's an obviously-correct answer.
Martin
Attachment:
pgpgBmYLtQG52.pgp
Description: PGP signature