[YG Conlang Archives] > [jboske group] > messages [Date Index] [Thread Index] >
--- In jboske@yahoogroups.com, "Jorge Llambías" <jjllambias@...> wrote:
>
> On 9/20/06, John E. Clifford <clifford-j@...> wrote:
> > --- In jboske@yahoogroups.com, "Jorge Llambías" <jjllambias@> wrote:
> > >
> > > The move in question, I take it, is from "Fa" to "ExFx".
> > > That move is always valid, as far as I'm concerned.
> >
> > Actually, it is the more specialized one, in Lojban terms from {F lo
> > broda} to {da poi broda zo'u F da} from "F a dog" to "there is a dog
> > such that F it."
>
> Always valid too.
Alas, not so if you want that the "a dog" in "I want a dog" is not to
be treated as intensional (and even if it is intensional, you don't
get "there is a dog" which is why the shift in what is said is
recommended).
> > This works in extensional cases, not intensional
> > ones. The idea is to get it to work in all cases (by making the
> > places extesnional
>
> For me, all places are always extensional in that sense.
You may want it that way, but you can't get it. That dog just need
not be in the domain (and even if it is, it can't be identified in the
way required).
> > (though containing -- and thus quantifying on -- intensional objects).
>
> Is the domain of discourse for you partitioned into intensional objects
> and extensional objects? If yes, is this a metalinguistic or an object
> language partition?
It should be both. The problem is that often it appears only to be
metalinguistic although its effects appear in the object language and
then are inexplicable. the idea is to get it thoroughly into the
object language. Of course, there are object language intensional
expressions, set aside as such: propositions, properties and the like.
> > > The move from "Fa" to "Ex(Zx&Fx)" (where Z is the predicate {zasti})
> > > is not a move of logic, and does not always work.
> >
> > And, so far as I can tell, no one has said it does -- or even
> > mentioned it except you.
>
> It is you who keeps bringing up the "real world" domain. If the
predicate
> {zasti} is not to play some special role, then I don't see how the
"real world"
> is relevant in any of this.
As I keep pointing out, the real world is a legitimate model. If
someting fails in the real world case, then if fails outright.
> > > In the restricted domain where AxZx is true (the "real world"
domain),
> > > the move from "Fa" to "Ex(Zx&Fx)" will work. When using such a
> > > restricted domain we won't be able to refer to things that don't
exist,
> > > obviously, since they won't be in the domain.
> >
> > But here even the simple generalization does not work from intensional
> > contexts "I want a unicorn" does not permit "There is a unicorn I
> > want" though the first can be true in the "real world" domain and the
> > (let's suppose it's this real world" is not. Yet it appears to be of
> > the Fa to ExFx pattern.
>
> The generalization from {mi djica lo pavyseljirna} to {da poi
pavyseljyrna
> zo'u mi djica da} is quite acceptable in any discourse that takes
place in
> the real world (or in any other world). Uttering a true sentence {da poi
> pavyseljirna zo'u ...} in the real world in no way requires {da
zasti} to be
> true. For an utterance to be true in the real world, there is no
need that
> all the referents involved exist in the real world.
You apparently are working overtime not to deal with the example in
hand. The real world example requires that the "new things" be drawn
from the real world as well -- moving from the model to the situation.
So, the inference clearly does not work here, since no unicorns
exist. But the inference will fail in any domain that does not have
unicorns in it -- the first sentence is compatible with such domains,
the second is not. Nor -- contrary to your delusion -- does
mentioning unicorns get them int the domain when the mention is a
non-primary occurrence (and in wants it is definitely non-primary).
And, even if there are unicorns, the inference fails, since it is not
true of any of the unicorns that I want *it* (as the article says "any
one will do").
> > > > Showing
> > > > there is a domain in which it does not work (true premises, false
> > > > conclusion) is enough to show that it is not valid.
> > >
> > > There is no domain in which the move does not work, as far as I
can see.
> >
> > One just indicated.
>
> If the domain does not contain unicorns as a member, it is not possible
> to claim that unicorns are related to me by the relationship {se djica},
But, of course, "I want a unicorn" does not make such a claim. It
appears to and that is why the inference is paradioxical. But, if we
mention what the relation is really to, then we can avoid that
problem. In short, of course I can claim to want a unicorn in a
domain that contains no unicorns; to claim otherwise is to fly in the
face of linguistic evidence (try picking up that "a unicorn" with "it"
in a declarative sentence, for example, let alone by a quantifier).
To be sure, ther is a reading of "I want a unicorn" (and, if all goes
well, of {mi djica lo pavyseljirna} which does allow the inference.
But that is not the usual reading (nor the usual way to make the claim
it makes), so we are not talking about that reading (the specific one
in the article).
> and so there is nothing to move from. If the domain does contain
unicorns
> as a member, then the move works fine. There is no problem with unicorns
> being a member of the domain of a discourse that takes place in the real
> world. So you have not indicated a domain where the move would fail.
Since unicorns do not exist in the real world, there is a problem in
bringing them into the a real world domain. Nor need you do so to get
"I want a unicorn" or ? after the proposed correction ? {mi djica tu'a
lo pavyseljirna} ({tu'a} here astanding for whatever we decide to use
as a marker).
> > > Every referring term contributes its referent(s) to the domain of
> > > discourse. If we fail to assign a referent to a referring term,
we must
> > > resort to the metalinguistic {ki'a}.
> > > We must step out of the discourse, as it were, and clarify the
language,
> > > because the discourse is not working as it should.
> >
> > I think that this is where you get it screwed up. Every referring
> > term *in primary occurrence* does indeed contribute its referent (I
> > suppose we are doing something like discourse analysis where the
> > domain is built up as we go along). But secondary occurrences do not.
>
> I don't think there is a need to introduce secondary occurrences
> with my interpretation,
Which, of course, we are all waiting to see. So far you have said what
you expect it to do avoid problems with opaque contexts) but not how
it is going to work. Notice that, if it leaves the same expression in
for both the generic and the specific reading of {lo pavyseljirna),
it fails to solve the problem.
but in any case surely in {mi djica lo pavyseljirna},
> {lo pavyseljirna} is in primary occurrence.
It looks like it and that is why the problems arise. Generally
speaking the second place of {djica} is opaque, but occasionally it
may not be. Lojban, as a logical language, ought to distinguish the
two cases, since they make a logical difference.
> > Unfortunately,for most secondary occurrences, there are primary
> > occurrences that look on the surface just like them. In terms of that
> > article you pointed to, they have both a specific and a generic
> > reading -- and the generic reading does certainly not point to a
> > particular item but also does not guarantee any such item at all.
>
> The generic reading does not guarantee a particular referent, I agree.
> In my view, particular referents do not play any role at all in the
generic
> reading.
Well, of course, we agree about that, since that is a nice summation
of one part of the problem.
> >"A
> > bear did not do this" would ordinarily be taken as simply a denial of
> > "A bear did this" and the denial takes out both any specificity there
> > might be in "a bear" and also the existential import about bears. It
> > is just ~Db (or ~Ex(Bx & Dx).
>
> If you take "a bear" as {su'o da poi cribe}, yes. But that is not a
referring
> term. For me, all unquantified sumti are referring terms.
Well, yes, they are pragmatically distinct and that is one way of
putting it. But the problem remains however you read {lo cribe}.
For, if {lo cribe} is extensional, then it implies {su'o da poi cribe}
and, if it is intensional, then we have the solution already proposed
and are just arguing about the details. But, in that latter case, how
do we do something that actually refers to bears?
> > There is also the specific reading (I
> > don't actually like these terms but they come from the article which I
> > assume you have read)which \x~Dx(b) or Ex(Bx & \y~Dy(x)), which in
> > either reading says that there are bears and furthermore that some
> > particular one of them did not do this (though another one may have --
> > consider the grade-school riddle:"I have two coins that add up to 30
> > cents and one of them is not a nickel.") There are similar cases for
> > most secondary occurrences (the article gives some for intensional
> > direct objects and some adjectives ("imaginary"). The point is that
> > the secomdary occurrences of referring expressions need not refer to
> > anything in the domain -- and so do not automatically add something
> > when they first occur.
>
> If you interpret descriptions a la Russell, I can follow all that. But I
> don't find that interpretation of descriptions at all compelling. For me
> {lo broda} are always referring terms.
This is not about descriptions a la Russell or anyone else. This is
just about natural languages ? including, I hope, Lojban. I take my
examples from English because I know it best __ and most of the
articles are about it ? and because the facts about it are not open to
discussion.
> > And what they need not add is not just a
> > specific thing but even the whole predicate extension: "I want a
> > unicorn" adds to the domain not only not a particular unicorn but not
> > unicorns at all. To think that secondary occurrences add particular
> > references or even generic ones (guarantee that a certain predicate
> > has a non-null extension) generates one array of either paradoxes or
> > misconstruals of ordinary -- and ordinarily understood -- sentences.
>
> I don't think generic terms add particular tokens to the domain, if
that's
> what you mean by particular references. I don't find any paradoxes in
> requiring all (unquantified) terms to refer.
Well, it depends upon what you want them to refer to. If you want to
enter some intensional object like a type as the reference of a
referring term in certain contexts, find. But an extensional one will
not do. And, if you are using {lo broad} for the intensional one, you
need either to spell out when it does that or provide another term for
reference to actual brodas.
> > It is, as you note, exceedingly difficult to have a primary
> > occurrence of a referring expression that fails to refer. It can
> > basically only happen when we have in mind a holistic model into which
> > the developing situation is to be fitted to test for truth and that
> > model does not contain anything to which the expression might refer.
> > We have then either to scrap that background model or reject the
> > referring expression (Oh, we're talking about..., I was thinking of
> > ..." or "But we're not talking about ...").
>
> Right, stepping outside the discourse for a moment and making
> some metalinguistic comment about the language being used.
Of course. No argument there (though technically we could go ahead and
assign all clause with non-referring terms a conventional truth-value
? actually "False" is forced by most models).
> > It is unfortunate that
> > "primary occurrence" seems always to be defined negatively ("not in
> > the scope of....") and that, as investigations proceed, the list of
> > things in the diairesis gets longer and longer, but the distinction
> > remains basic to semantics of natural languages for all of that.
>
> In some theory.
I don't know of one where this does not occur. Can you spell it out
or direct me to a case.
> > > > > > We say that two things are identical when we refer to the same
> > object
> > > > > > using expressions that have different sense.
> > > > >
> > > > > Why do you say "two things" then? Can two things be one and
> > > > > the same object?
> > > >
> > > > We tend to say there are two things when we mean there are two
> > > > denoting expressions. I think I said it carefully at least
once, but
> > > > that is tedious and hardly anyone who is a cooperative
> > > > conversationalist misunderstands the other expression.
> > >
> > > I agree it's easy to understand. I'm not convinced that the
theory that
> > > says that this way of speaking is just sloppy usage is better
than the
> > > theory that says that this way of speaking is perfectly justifiable
> > uage.
> >
> > I am afraid that I can't even figure out what a literal reading of the
> > sentence would be; as you point out, there can't be two things that
> > are identical in the same ways as there are two of them. They can be
> > two in one sense, identical in another, but that is not the issue here
> > (or, if it is, then there is no issue).
>
> Two things can be the same F and different G's, that's the only
> significance of relative identity.
But then it is odd to talk about a different kind of identity, since
it is only about comparing things in different way or or different
things in the same way. It is just identity and the issue is what it
is identity between. This does seem to catch something of what we
actually do when we come across a real case, but that catching can be
done as well with just one identity and ? as here ? different ways of
comparing things. Creating a new kind of identity where none is needed
or used (except to confuse people) seems a bad idea.
>
> > > > > Have you read
> > <http://plato.stanford.edu/entries/identity-relative/>?
> > > > >
> > > > Yeah. It didn't help since you seem to have reproduced its way of
> > > > talking fairly accurately and it is that way of talking that
does not
> > > > make sense.
> > >
> > > To me it makes pretty good sense. It is in fact the only account of
> > > the "paradoxes of identity" that I have found satisfying. Is
there any
> > > particular part that does not make sense to you?
> >
> > Well, the notion that there are two kinds of identity, rather than
> > comparisons between two different things is a good starting point.
> > This seems simply a misuse of language and to fly in the face of what
> > we actually do.
>
> That we actually do often say things like "two identical things" is,
I hope,
> patently true, so it certainly does not fly in the face of what we do.
> That "two identical things" is a misuse of language requires a theory
> that obviously results in lots of misuses of language in ordinary
> speech.
That is not the problem; the problem is saying that the "identical"
and the "different" arise from two different identities rather than
from comparing two different sorts of things. I suspect there is also
a problem of partial completion of sentences here, but that is harder
to show in the usual examples.
>
> > > so as to make clear to Superman/Clark that his "me" must be
interpreted
> > > in accordance with his current guise.
> >
> > Indeed so, And your point would be?
>
> That there is no ultimate Superman/Clark token independent of context.
I didn't realize we were talking about tokens and types here, but
there clearly is an ultimate token of each of these (assuming for the
moment that this way of talking makes sense) ? that token happens to
be the same being, though viewed in different ways. I don't suppose
there is any problem (since it is built into the notion of a sense)
with saying that being a token of different types involves being
looked at in different ways. So, in that sense , what type a thing is
a token of ? or at least recognizing this ? does depend on context.
> > We are now comparing Superman
> > slices with Clark slices and finding them not identical, whereas, if
> > we compare the wholes from which the slices are cut, they are
> > identical. But there is only one sense of identity here.
>
> Right. But then we have to be careful with the claim
> {la superman du la klark kent}. It is very often false. And if we can
> see that that claim is false, there is no failure of substitution of
> co-referring terms since we don't have co-referring terms to begin
> with.
I am not sure that it is ever false, but it is true that it is, then,
yes, the problem does not arise. So the trick is to know when it is
false and that would be handled logically by having different
referring expressions for referring to the various things that are now
all referred to by "Superman" and "Clark Kent."
>
> > > There is
> > > already the predicate {mintu} which is pretty much what relative
> > > identity is.
> >
> > Well, then we need something for absolute identity.
>
> {du} should be reserved for that.
I thought {du} was reserved for mathematical expressions, but that is
clearly just a convention. The point remains that either {du} or
{mintu} mean the same thing or at least one of them does not mean
identity. More likely, it appears, expressions in identity statement
(etc.) are often ambiguous and the ambiguity may actually be played upon.
> > And, note, we
> > need the stuff for sense-reference anyhow, in some form or other.
> > And, if we do other things right, the "problems" that the
> > two-identities theory is meant to solve won't arise.
>
> I'm not convinced we need any grammatical marker for sense. So far,
> all can be done with simple reference.
You have yet to show that *any* of the problems can be solved by
simple reference. If you mean by that that every expression in all
contexts refers to its referent, that is precisely the situation that
I am trying to get in Lojban. For some expressions that referent will
be intensional (as it is for some expressions already), for other
extensional.
By the way, it seems to me that the right thing to use for "qua" is a
modality (which I can't find, suggesting another screw up) for "by
standard." For the case of Clinton, something derived from {te xamgu}
seems to be just right.