* Wednesday, 2011-09-21 at 19:08 -0300 - Jorge Llambías <jjllambias@gmail.com>:
> On Tue, Sep 20, 2011 at 10:15 PM, Martin Bays <mbays@sdf.org> wrote:
> > * Tuesday, 2011-09-20 at 21:16 -0300 - Jorge Llambías <jjllambias@gmail.com>:
> >> ro klesi be lo gerku cu gerku
> >> lo xanto cu bramau ro drata ke tumla danlu
> >> lo remna cu se tuple re da
> >
> > I would say that they are false.
> >
> > For the first, I don't think that's a problem.
>
> The first was kind of beside the point anyway. Presumably you would
> still allow "ro klesi" in other contexts, such as "ro klesi cu klesi",
> right?
Assuming we have mundane (i.e. non-kind) objects which klesi, then yes.
I'm not really sure what klesi should mean, but I'm taking it that you
want the kind corresponding to {broda} to klesi. That doesn't seem
unreasonable, and I'll accept it for now.
> The problem with saying it is false is that if "lo blabi gerku
> cu klesi lo gerku" is true, and "lo blabi gerku cu gerku" is also
> true, it's hard to say why at least "su'o klesi be lo gerku cu gerku"
> would not be true.
But we already have the same kind of weirdness with plurals:
lo gerku remei cu remei .i je lo gerku remei cu gerku .i je ku'i ro
remei na ku gerku remei
Generally: you can't quantify over plurals (assuming we agree to the
extent I'm under the impression we do on how plurals work); not being
able to quantify kinds is a similar kind of restriction.
> > For the third: it could be replaced by {lo remna cu ckaji lo ka se tuple
> > re da}, or by a tanru - {lo remna cu re mei se tuple} - or by an
> > explicit generic quantifier, which {so'e} arguably is: {so'e remna cu se
> > tuple re da}.
>
> It could. So in your system "lo du'u ko'a ckaji lo ka broda na nibli
> lo du'u ko'a broda" is true, right?
Depends what you mean... for any predicate broda, I would want that to
be false. But {se tuple re da} is not just a predicate in the above uses
- it introduces an existential, and (part of) the question is what scope
that existential has. Stuffing it inside a {lo ka} prevents it from
scoping over the {lo remna}.
> > The second is toughest, and a good demonstration of the power of
> > kind-quantification, but since it is in the end a case of generics,
> > I think it can be handled similarly:
> > {so'e xanto cu bramau so'e tumla danlu poi na xanto}
>
> > Correction: just using {so'e} isn't really adequate for this purpose, of
> > course, because default tense assumptions would have that making a claim
> > only about the present world.
> >
> > Maybe this is where {lo'e} could come in.
>
> With "lo'e" as "typical"? But the typical land animal is arguably an
> insect, so I could say "lo'e smacu cu bramau lo'e drata ke tumla
> danlu". With some other sense of typical I might judge the typical
> land animal to be some kind of cervid, but maybe that's just me. It
> still wouldn't capture what I want to say about elephants though.
Fair.
But wait, I was missing something obvious.
You can still use {lo}:
{ro da poi na'e xanto se danlu zo'u lo xanto cu bramau lo tumla danlu be da}.
> > That doesn't help with a pure kind predication, though. For that, e.g.
> > "I like all animal species", you'd have to be explicit about the kind
> > predication:
> > {mi nelci ro ka danlu ma kau}
> >
> > (assuming {ka} and qkau work so as to make that work)
>
> Would you really be happy with "so'i ka danlu cu se krasi lo
> frikygu'e"?
Not overly, because that would mean allowing {zo'e} to mean {ma kau}!
But {so'i ka danlu ma kau cu se krasi lo frikygu} seems reasonable -
i.e. this definition of krasi for properties in x2 seems reasonable.
> I think properties are things like propositions or numbers. perhaps
> you may like them, de gustibus non est disputandum, but I can't really
> say of them the kind of things we say of physical things.
I don't know. I don't see a problem with it.
"it" here, to be clear, being copying pure kind predications to
the corresponding properties.
I don't see any conflict with pre-existing uses of properties.
> > In general, it seems to me that kind predications resolve as one of
> > * existential quantification
> > * generic quantification
> > * property predication
> >
> > {lo broda} might allow you to be ambiguous between the three, but for
> > complicated sentences you'd have to say what you mean.
> >
> > Is that so bad?
>
> So you don't really buy into the Carlson-Chierchia non-ambiguity
> thesis. You favor what Chierchia calls "the A-approach".
For natural languages? I don't know enough to have an opinion. As for
lojban, I think both approaches provide useful ideas.
In any case, if I'm reading him right, Chierchia would agree that a kind
predication eventually resolves as one of three things - although he
would balk at the idea that the third is simple property predication,
because the iota operator part of the down operator is important to his
explanations of why plurals are used to denote kinds in appropriate
natural languages. This latter doesn't apply to lojban, and I see no
reason not to skip directly to the property (but am not sure there isn't
one).
> For me that approach is unsatisfying, but that wouldn't matter if we
> come to the same result about the meaning of sentences. At this point
> we seem to have different results for at least some cases.
You mean the question of whether a pure-kind predication can block an
existential or generic reading?
This does seem a fairly minor issue. If we could agree that {lo} is
ambiguous between those three things with the property case taking
precedence when it makes sense, I'd be happy.
(well, not wholly happy until I understood how the generic predication
case works, and how to unambiguously make generic predications... but
that's another issue)
This precedence issue aside, and the kind-quantification issue also
aside, are there any sentences for which your understanding of {lo}
gives meanings different from those given by the ambiguity-based
approach?
Attachment:
pgpoYN4koeMGj.pgp
Description: PGP signature