* Wednesday, 2011-09-07 at 23:58 -0300 - Jorge Llambías <jjllambias@gmail.com>:
> On Wed, Sep 7, 2011 at 11:03 PM, Martin Bays <mbays@sdf.org> wrote:
> > * Wednesday, 2011-09-07 at 21:47 -0300 - Jorge Llambías <jjllambias@gmail.com>:
> >>
> >> I was still thinking in terms of possible answers to "xu do klama lo
> >> zarci". In such a context, I would take the referents for "zo'e" in
> >> "ro ma'a klama [zo'e]" to be the same as for "lo zarci".
> >
> > Ah. So in other contexts, {ro ma'a klama} could be true without the
> > destinations being the same for different referents of {ma'a}? But only
> > because you'd have the zo'e mean the generic "destinations"?
>
> Right.
>
> > If you introduce such generics, it seems that it becomes impossible to
> > unambigously specify order of quantifiers. This, surely, is a Very Bad
> > Thing.
>
> Scope of quantifiers is always unambiguous. What I think is impossible
> is to fix one domain of discourse for all discourse independent of its
> context.
>
> > I mean: you seem to be suggesting that for any broda(x,y) and any domain
> > of discourse M, there should be another plausible domain of discourse *M
> > extending M and an element *y \in *M such that
> > \forall x\in M. (\exists y\in M. broda(x,y) => broda(x,*y) ).
>
> I'm not sure I follow. If *y is meant to be a generic term for some
> things already existing in M
Yes
> , then I would consider the extended domain *M implausible rather than
> plausible. I would think that in most plausible domains of discourse
> the generic doesn't coexist with its manifestations. Making them
> coexist takes some special effort.
Sorry, yes, you did say that in the other thread. But it doesn't really
affect the logic here.
> > But then if I do say {su'o de ro da zo'u da broda de}, it could be that
> > I'm working in M and really mean to make the strong assertion
> > M satisfies \exists y. \forall x. broda(x,y) ,
> > or I could be working in *M and hence be claiming only
> > M satisfies \forall x. \exists y. broda(x,y) .
>
> Yes, if I say "there's some fruit that everybody ate" I could be
> saying (and probably would be saying) that everybody ate apples, for
> example. Is that what you mean?
Yes, precisely.
The ambiguity would be widerspread than in English, of course. You can't
say "someone loves everyone" and mean that the generic lover does, but
you could with {da prami ro de}. And if this generic lover is invoked by
innocuous phrases like {ro se mamta cu se prami}, it wouldn't be
a particularly exotic reading.
> > The only way to tell which I meant would be informal rules about saying
> > things in the least confusing way.
>
> Yes, if you thought there was some risk of confusion as to what the
> domain of discourse was, you would need to be more precise: "there's
> some kind of fruit that everybody ate" vs "there's some individual
> fruit that everybody ate". But that's unavoidable, since the domain of
> discourse is not uniquely fixed for any and all contexts.
>
> > So no, I don't think such tricks should be resorted to unless absolutely
> > necessary - and if they do prove necessary, I'd think it a problem with
> > the language.
>
> You call them tricks, but I think it's an ordinary part of language.
You may be right that it's impractical to grammatically distinguish
between generics and mundanes (and perhaps also that there isn't
a coherent distinction between the two), though I'm not really convinced
yet. But even if so, introducing with zo'e these generics which
demonstratedly *aren't* necessary, because they don't really exist in
natural languages, seems unhealthy.
> >> > It sounds like you might be giving it longest scope rather than
> >> > shortest, which gets around that kind of issue... though it still has to
> >> > scope inside the da in {ro da zo'u broda zo'e noi brode da}.
> >>
> >> I don't give it any kind of scope, since I don't think constants have
> >> scope. But if you do need to force constants to be quantified, then
> >> yes, I would have to favour longest over shortest.
> >
> > What's the alternative to scope?
> >
> > I thought we agreed earlier today that zo'e isn't literally a constant
> > in general, e.g. it has to scope inside {da} in the above example.
>
> But functions don't have scope either, it is quantifiers that have
> scope. If a function happens to take a bound variable as an argument,
> then its value will be determined within the scope of the quantifier
> that binds that variable, yes.
I think this is just different language for the same thing - i.e. you're
Skolemising away the existential quantifiers
(http://en.wikipedia.org/wiki/Skolem_normal_form).
Attachment:
pgpWZm2TdPT7k.pgp
Description: PGP signature