Message-Id: <199512151911.VAA00514@xiron.pc.helsinki.fi>
Date:         Fri, 15 Dec 1995 14:11:21 -0500
Reply-To: Jorge Llambias <jorge@PHYAST.PITT.EDU>
Sender: Lojban list <LOJBAN@CUVMB.BITNET>
From: Jorge Llambias <jorge@PHYAST.PITT.EDU>
Subject:      Re: tech harangue on le/lo
To: Veijo Vilva <veion@XIRON.PC.HELSINKI.FI>
Content-Length: 5767
Lines: 145

We are back in the mud about opaqueness:

> >  If the x2 of {kalte} was always opaque, we could not use
> > it for the transparent use of "hunt".
> There are, as I recall, at least three.  One is to be sure that that
> lion, Clarence, is introduced well outside the intensional context --
> another sentence or a prenex with appropriate anaphora will do.

I think we agree about that. For example:

        lo cinfo pu catra lo bakni i mi ba kalte cy
        A lion killed a cow. I will hunt it.

It is clear that there is a lion that I am going to hunt. I claim that
I could also say:

        mi ba kalte lo cinfo poi pu catra lo bakni
        I will hunt a lion that killed a cow.

If I say that, then I am saying that there is a lion that killed a cow,
such that I am going to hunt it. That is still transparent.

A different matter would be to say something like:

        mi ba kalte lo'e cinfo poi catra lo'e bakni
        I will hunt lions that kill cows.

In that case, I am not saying that there is a lion that killed a cow
such that I am going to hunt it. I am only saying that I am going
cow-killer-lion hunting.

>  The
> second is to mark the buried occurrence with the "external reference"
> mark (which, I recall was not needed for local descriptions and known
> names).

In my opinion it is not needed for {lo broda} either. Why should
it be needed? Why should {mi djica lo plise} be different in Lojban
from {mi catlu lo plise}? Just because "I want an apple" is different
in English from "I see an apple"?

> The third, which you might have been using but took you not to
> be (and so it now appears), is to use both tu'a and xe'e (or whatever it
> was) and then allow them to cancel out to an unmarked form.

I was not using it because I don't think it is needed. I don't assume
that {kalte} or any other predicate has any cloudy protection over one
of its places. In other words, {da broda de} is always equivalent to
{su'o da su'o de zo'u da broda de} = "there is an x and there is a y
such that x broda y", no matter what {broda} is.


> > I think that "he is hunting lions" in the opaque sense has the
> >same kind of feature as "he builds houses", in the sense of "he is a
> > house-builder". There need be no houses to instantiate that claim.
> > In those cases I use {lo'e zmadi} and {lo'e cinfo}.
>
> But the established logic and Lojban (even Loglan) is a tanru for these
> cases -- states rather than activities

According to some, every possible meaning of tanru can be expanded
into a non-ambiguous form. But whether that is true or not, it doesn't
matter. If you say that "house builder" is simply {zdani zbasu}, then
what goes in the x2 of that relationship? "x1 house-builds what?"

You could argue that it is a lujvo: {zdazbasu}. In that case, it
could have the meaning "x1 builds houses from material x2 for
inhabitants x3", but you would still be in trouble if asked to fill
the x2 and x3 places, unless we have {lo'e broda} available.

Is there really no established logic for "he builds houses" that
uses the predicate "x1 builds x2"? Is it really only possible
to use the predicate "x1 is a house builder" for that?

> > (1)     mi troci le nu da poi cinfo zo'u mi kavbu da
> >         I try that there be a lion such that I catch it.
> >
> > (2)     da poi cinfo zo'u mi troci le nu mi kavbu da
> >         There is a lion such that I try to catch it.
> >
> > (3)     da poi cinfo zo'u mi kalte da
> >         There is a lion such that I hunt it.
> >
> > I would say that (3) has the meaning of (2).
>
> 3 does mean 2, since the quantifier is nicely outside the opaque context.

What about:

(4)     mi kalte da poi cinfo
        I hunt something that is a lion.

Is (4) different from (3)? In my opinion, no. If it is, then the rules
for prenexing are more complicated than I ever imagined, because they
would depend not only on syntax but also on the semantics of the
particular selbri being used.

> The problem is having 3 mean 1 or inferring 3 (or 2) from 1.

I agree that that is a problem, and that one may erroneously do it
if translating naively from English or some other natlang. The naive
translation of "I want an apple" as {mi djica lo plise} is wrong,
but that doesn't mean that {mi djica lo plise} is not proper Lojban,
it just has to be properly understood as the transparent: "there is
an apple such that I want it".

> I think it
> would be very nice indeed if there were no gismu that had opaque places
> except those that took full lenu (etc.) sumti.

That's how I understand the situation to be. If there are such strange
behaving gismu, we should be told which places of which gismu have
that weird behaviour.

> But I also think that no
> one would tolerate losing all the familiar words (hunt, want, etc.) that
> have those places (and do in countless natural languages -- how do you
> think philosophers keep in business?

I don't see why we would be losing them:

        mi djica le va plise
        I want that apple.

        mi djica lo plise
        There is an apple such that I want it.

        mi djica lo'e plise
        I am in a state of apple-wanting.

These require no opaque places. There is no need for instantiation of
apples in the third case. The only way we would require an opaque place
is if we want to force {mi djica lo plise} to mean something other than
"there is an apple such that I want it", but I don't think anyone is
arguing for that.

> It may not be Lojban's job to
> solve philosophical problems, but, if the problem has a solution -- as
> this one does, Lojban should incorporate it and, to that extent, keep
> philosophitis at bay).

Certainly. But I'm not sure whether your solution and mine coincide.
What does {mi djica lo plise} mean in your scheme?

Jorge