* Thursday, 2011-08-25 at 21:49 -0300 - Jorge Llambías <jjllambias@gmail.com>:
> On Thu, Aug 25, 2011 at 8:19 PM, Martin Bays <mbays@sdf.org> wrote:
> > * Thursday, 2011-08-25 at 19:11 -0300 - Jorge Llambías <jjllambias@gmail.com>:
> >> On Thu, Aug 25, 2011 at 6:06 AM, Martin Bays <mbays@sdf.org> wrote:
> >> >
> >> > Firstly: there is the question of whether Kinds are in our domain of
> >> > quantification.
> >>
> >> My answer is that sometimes they are and sometimes they are not,
> >> depending on what "our domain of quantification" happens to be at the
> >> time.
> >
> > Meaning that quantification can be over ordinary individuals, and it can
> > be over Kinds, but it can't be over a mixture of the two?
>
> Not exactly. I think kinds are just as ordinary as any other
> individual. It's not a binary distinction.
Well, it seems *possible* to make the distinction; the question is
whether it's a good idea to hardwire it in the semantics of the
language. I think it might be (if Kinds are to be admitted at all).
> > So our
> > universe has multiple sorts, and {ro da} can be quantifying over any one
> > of them - but only one at a time? I'd be happy with that, and it seems
> > to deal with your "two favourite desserts" issue.
> >
> > It would be nice to have a way of explicitly indicating that
> > quantification is over usual individuals and not Kinds.
>
> But "usual individuals" is just whatever can be the value or values of
> a variable. In some contexts there is and can be one and only one
> letter "e". In some other contexts there can be two and only two
> (written "e" and "E".) In other contexts there can be many, but such
> that there are only two in the word "letter". Even if I write "letter"
> and "letter" twice, that one word has two "e"'s. But then in another
> context, when it's two words that I wrote, there are four "e"'s that I
> wrote. And perhaps if we count not just the two words I wrote but the
> (tens? hundreds?) of computer screens those words I wrote have
> appeared in, we may have to admit that there are hundreds of 'e's that
> have been read. I don't see how we can reduce all that to just two
> levels.
Well. I think I'd analyse that with just one level, with a mix of
ambiguity of "letter" and a varying universe aka domain of discourse.
But wouldn't you agree that the domain of discourse should be mostly
static? So we can't solve the problem of {ci da panzi mi} by just
declaring that such a statement indicates that the Kind {lo'e panzi mi}
probably isn't in the domain of discourse, because it wouldn't be at all
remarkable for {lo'e panzi mi} to be explicitly talked about in a nearby
or even the same sentence.
> > I'd say that {da
> > poi du} would do that (or {da poi zilmintu} if we want {du} to be
> > magic), as long as we ignore Kinds made from singletons.
>
> I'm not sure it's such a good idea to think of kinds as being made of
> something.
Well... they come from unary predicates (no?), and sometimes those are
true of only one thing. e.g. {lo'e me mi}, maybe.
> > You mention below "[not] just two levels of abstraction, concrete and
> > abstract, but lots and lots of levels with different degree of
> > abstraction"; could this mean more sorts? If so, what are you thinking
> > of? I guess you could have Kinds of Kinds, though maybe that's more
> > trouble than it's worth...
>
> It may be trouble, but I think it's inevitable. I think making
> generalizations and particularizations is an ordinary part of
> language, and it is not limited to two levels.
Yes, I suppose so.
> >> Would you object to "natural numbers are equal to something (namely
> >> themselves)"
> >
> > No, but I don't see how to analyse it with generics - I'd say that's
> > a clear case of quantification, i.e. that it's
> > {ro da poi mulna'u cu du de ne da}.
>
> You don't agree with Carlson's analysis of the English bare plural
> then. I think he presents a very good argument that a unified analysis
> makes more sense than thinking that it is sometimes the plural of
> "a(n)" and sometimes a universal/generic.
But what would 'themselves' mean if 'natural numbers' refers to the
Kind? No good just having it be the same Kind - that doesn't handle
"natural numbers are less than something (namely their successors)".
> > You can't say
> > "something (namely themselves) is/are equal to natural numbers"
>
> How about "One thing are natural numbers and some other thing are
> rational numbers".
No, I think that's entirely ungrammatical.
> > (the problem isn't the pronoun position - you can say
> > "something (namely itself) is equal to any natural number"
> > )
>
> I think "something is/are natural numbers" is weird due to the
> difficulty in making any sense of why someone would want to say
> something like that. "Something is a dog" is almost as strange, and
> there we don't have to fight against the plural morphology.
>
> >> and "poets write some poems, but most poems are written by non-poets"?
> >
> > I don't think 'poets' is a generic there, any more than 'non-poets' is.
>
> Carlson would disagree. (Or rather,I think he would say that bare
> plurals are always the same thing, I'm not sure he would call it a
> generic.)
Hmm. I think he would call 'poets' and 'non-poets' indefinite plurals,
and analyse them as existentially quantified
instances/realisations/stages of the corresponding Kinds (as in section
4.2 of the paper you reffed).
> >> But I want to be able to say "dogs have been known to eat carrots"
> >> even when the set { (x,y) | dog(x) /\ carrot(y) /\ eat(x,y) } does not
> >> seem to be Large in { (x,y) | dog(x) /\ carrot(y) }
> >
> > Again, I don't think 'carrots' is a generic there.
> >
> > Actually, doesn't it just mean
> > {se zgana lo nu su'o gerku su'o najgenja cu citka}?
> > If not, what did you mean by it and how would you like to Lojbanise it?
>
> Maybe something along the lines of:
>
> va'o su'o da lo gerku cu citka lo najgenka
> "Under some circumstances, dogs will eat carrots."
Hmm. But I don't think that that lojban captures the most obvious
interpretations of that english, which would have the circumstances in
question referring to the dogs and/or carrots involved.
e.g. "When they're forced to, dogs will eat carrots" doesn't validate
{va'o su'o da lo'e gerku cu citka lo'e najgenka}.
But if you wanted to e.g. translate
"In Soviet Russia, carrots eat dogs!"
as
{bu'u la sofnai lo'e najgenka cu citka lo'e gerku}
then I agree this wouldn't agree with the semantics I suggested.
> Much like:
>
> va'o du'o da la djan tavla la alis
> "Under some circumstances, John will talk to Alice."
>
> I would not bring in time slices of John and time slices of Alice in
> order to interpret that sentence.
>
> > More generally, could you indicate (however vaguely) what you think the
> > truth conditions for {lo'e broda lo'e brodi cu brodu} should be?
>
> I'm not really sure about "lo'e",
(In case this wasn't clear, I'm using {lo'e} here to unambiguously refer
to the Kind reading of (xor){lo})
> but for "lo broda lo brodi cu
> brodu", just the same as for any other "ko'a ko'e broda". I just don't
> think that "lo broda" fixes by itself any level of abstraction.
So you think we should just consider it as another part of the ineffable
and irreducible semantics of brodu?
Maybe that's the only way, though I'm not going to find it very
satisfactory, and I worry that it might be a recipe for confusion -
english speakers are likely to interpret Kinds as working exactly like
generic bare plurals in english, while I guess (without knowing any
examples) other languages don't have generics which work the same way.
Meanwhile, one last translation question: do you have a way of rendering
"humans have two legs" in your lojban, using the Kind 'humans' but
without using anything like the {ckaji lo ka...} trick?
Attachment:
pgpDtNj4z68Bf.pgp
Description: PGP signature