[YG Conlang Archives] > [jboske group] > messages [Date Index] [Thread Index] >
This discussion has (surpeise!) gotten way off track. The situation
at issue is the following. There are several pairs of arguments --
one valid, one not -- that are represented in English and other human
lanuages (including most spoken by people who are interested in
Lojban) and in Lojban as it seems usually to be written (though not as
it is recommended) by the same surface sentences. Linguists,
Logicians and Philosophers have spent a couple of millennia trying to
work out how this could be and have come up with a variety of answers,
which, though they differ in detail, follow the same broad outlines.
On this basis, Logicians have devised logical systems that give the
correct results for these -- and other relatd -- arguments by showing
that, despite surface appearances, they have different forms. The
question is whether these differences are of sufficient generality and
commonality to be worth incorporating into a constructed language and,
if os, how. For Lojban, so long as it claims to be a logical
language, the works of Logicians decides the issue: the distinction
should be incorportaed. That just leaves the question of how. Now
you seem to have worked out an explanation of the common form these
arguments take in usual Lojban, such that at least one of these
arguments is valid. This amounts to saying either that you decided
that the usual argument presentation is a presentation of the valid
member of the pair or you are using the form of this argument for
another, new -- or at least undiscussed -- argument separate from the
pair of interest. In the first case, we still need to know how the
other member of the pair -- the invalid one -- is to be represented.
We also need to know whether your move in this case will carry over to
other argument pairs that have been taken historically to be related
to this pair. In the second case, we need to know why we should be
interested in this new argument and whether it shed any light at all
on the arguments we are discussing.
--- In jboske@yahoogroups.com, "Jorge Llambías" <jjllambias@...> wrote:
>
> On 9/22/06, John E. Clifford <clifford-j@...> wrote:
> > --- In jboske@yahoogroups.com, "Jorge Llambías" <jjllambias@> wrote:
> > >
> > > There is no model in which {mi djica lo pavyseljirna} is true
> > > and {da poi pavyseljirna zo'u mi djica da} is false.
> >
> > Sorry, I was reading into your statement. You mean that his is how to
> > do the broad-scope specific reading, not the usual English one. I
> > agree with that. How, then, do you do the narrow-scope, generic
reading?
>
> No, that's not what I mean. There is no broad-scope/narrow-scope
> distinction if {lo pavyseljirna} is a referring term. The distinction
> is only possible when dealing with a quantified term, then the
quantifier
> can be outside or inside some other operator.
So you are talking about a new and different argument, yet another one
that uses the same form of words as the first two but has a new
(unexplored, so far as I can tell) reading of some key terms. Why
should we be interested in this?
> With my interpretation
> of {lo pavyseljirna}, the two readings you mean will not correspond to
> two scope possibilities within the same model, but to two different
models
> altogether. (Of course the two-scope distinction is still available when
> using quantified terms and some other scope operator for the quantifier
> to interact with, it is not available when using just the simple
referring
> term.)
Let's see. {lo pavyseljirna} is a simple referring term -- it gets its
meaning directly from the interpretation in the model. There are two
different kinds of interpretations, one that makes {lo pavyseljirna}
act like a broad scope case and onwe that makes it act like a narrow
scope case. Which of these are you intending for your model? Or is
there a third type not yet mentioned? You claim that on your model
the argument is valid, so it cannot be the model where {lo
pavyseljirna} acts like a narrow scope, since the argument is invalid
on that. So, it must be the other one, broad scope. But, since we
know that there are models in which the argument is not vlaid, the
argument is not valid simpliciter. It is, however, valid on ,odels
that emulate the broad scope quantifier. So you are claiming that,
contrary to the empirical evidence, the meaning of {mi djica lo
pavyseljirna} is always the broad scope reading. So, how do we do the
narrow scope one, which surely turns up from time to time?
> As for "the real world model", what exactly is it?
>
> I would define *a* real world model as any model such that:
>
> (R1) The real world is a member of the domain of discourse.
> (R2) There is a referring term (say {le zasti munje}) whose referent
> is the real world.
> (R3) For every member x of the domain of discourse, if x exists in
> the real world and the referring term <x> of the object language refers
> to x, then the sentence {<x> zasti le zasti munje} of the object
language
> is a true sentence of the model.
> (R4) For every member x of the domain of discourse, if x doesn't exist
> in the real world and the referring term <x> refers to x, then the
> sentence {<x> zasti le zasti munje} is a false sentence of the model.
Here is where your model is not a real world model. If something is
not in the real world, it is not in the range of variables or the
interpretation function for referring terms. That is, if <x> doesn't
refer to something in the real world, <x> does not refer at all. To
be sure, as I would do it, that does make {<x> zasti le zasti munje}
false, but other might treat non-referring terms differently. In any
case, the claim would not be true.
> With that definition, there are lots of different real world models,
> depending on what the universe of discourse contains.
>
> If we want to identify one special real world model as "*the* real world
> model", we might define it as something like this:
>
> *The* real world model is a real world model such that:
>
> (R5) Every thing that exists in the real world is a member of the
> domain of discourse.
> (R6) Every member of the domain of discourse exists in the real
> world.
That is about it. Of course, we would usually just work with embedded
situations, so that not all the real world need be in at the beginning
-- though we can pull more stuff in as the need arises.
> But the real world model so defined is not a very useful model for
> most discourses. It is too wide in one dimension (the domain
> contains lots of things that are totally irrelevant to most discourses)
See above
> and it is too narrow in another dimension (the domain lacks many
> things that are often required in a discourse).
It is surprising how seldom we really do go beyond the real world. To
be sure, in the course of a conversation, we may occasionally (or even
often) range out into hypothetical worlds of various sorts. But We
always come back to the original domain one the semantic and logical
excursions are over; they are bracketted and do not affect the domain
in any significant way (except to provide support or explanation for
various claims made in primary space).
> Is that what you mean by "the real world model"?
See above.