Re: Quine vs Montague, the deathmatch

At 01:03 AM 8/12/03 +1000, Nick Nicholas wrote:
>I also
>accept that Bob, when he fumbles for a reason to reject {lo} = {da
>poi}, is probably thinking intensions without realising it.

Lojbab doesn't know what he is thinking, in formal semantic terms, EXCEPT 
that from the beginning of Lojban until late in 1994 (as a result of 
another gadri argument involving Jorge, Cowan and pc with myself as 
befuddled watcher), there was a clear and explicit distinction between "lo" 
and "da poi" that was among other things justified by "lo -unicorn". I 
can't explain to anyone's satisfaction why this came about; I believe that 
the distinction was explicitly noted by Jack Waugh, who hasn't been active 
for more than a decade, within the first year of the effort, and confirmed 
at the time by pc, who has since waffled all over the place on the matter.

My own concept of "lo" (and I did invent the thing), was based on Loglan's 
article (le'u?) which was used for universal claims about a set, and was 
translated as "all those which really are members of the set". pc 
introduced me to the word "veridical" at that point and the essential 
feature of this concept was veridicality. There was no claim of existence; 
you could make claims about the set of unicorns. I personally avoid 
universal claims, and thought it would be more proper to make a veridical 
parallel to "le" NOT be quantified universally, but could be explicitly so 
as "ro lo"; I thought a similar parallelism would apply equally to lei and 
loi, and thus came up with the parallel trio of gadri (le/lei/le'i and 
lo/loi/lo'i, with le'i and lo'i used to talk about the sets 
themselves). Then the question immediately became a matter of what 
quantifiers should apply, and pc opined that, given what I wanted, "ro le 
su'o" was required for "le" and "su'o lo ro" for "lo". We agreed at the 
time that "da poi" was not the same thing because of the lack of an 
existence claim.

The semantic discussion stopped there, because I know neither Quine nor 
Montague, and was not that interested in constructing a semantic theory for 
Lojban (I was in fact trying to keep the language neutral with regard to 
such theories, and ignorance was bliss). After the intervening years, and 
the flip-flop that occurred in late 1994, the CLL version of lo already 
does not match my understanding, since I never bought into "lo" = "da poi" 
(but I wasn't the one writing CLL).

Thus, if one asks my opinion on what the essentials of the gadri are, it is 
that
"le" be "in mind referent" (which I think people have said means "+specific 
-veridical") and "lo" is "+veridical" and probably "-specific".

I personally don't like "lo" as equating to "da poi", and it implies to me 
that JCB's le'u would have been "ro da poi". I don't think that such a 
trivial abbreviation, saving only a syllable and concealing logical 
structure, would warrant assignment of a gadri, and if I had thought that 
was what I was doing, I probably would not have done so. But as I noted, 
people at the time we started saw that there was a distinction between "lo" 
and "da poi", though I won't pretend to be able to argue what that 
distinction is in semantic terms; I apparently failed to do that in 1994 
when the arguments were much fresher to my mind.

>This is not a minor and welcome fix, as I believe a cmene fix would be.
>This is a reversal of any understanding we have of Lojban; it annuls
>its basic semantic underpinnings.

I don't think that Lojban HAS any semantic underpinnings, which is why we 
have these arguments. People are insisting on erecting a foundation where 
I specifically tried to avoid one wherever possible, and I've accepted that 
they will do so, and that they find great need for a more solid foundation.

>Our usage may not be affected; but our understanding of the theory of 
>Lojban is tipped over. And I cannot
>accept it.

I can't accept a lot of things, but in order to get the language done, I've 
had to anyway. I have accepted that there will be changes in the gadri, 
but since I have no "theory of Lojban" to be tipped over, my line-drawing 
is not based on theory. The baseline will be changed in this area, unless 
your solution somehow protects it, and you seem less than sure that it will.

>Is this because I have a personal stake in having my own solution
>prevail? Perhaps, although my solution is yet to be articulated --- and
>if I continue to feel as I do now, may never.

I would rather see and judge your solution, since honestly, I trust your 
conservative instincts more than the collective jboske wisdom. You also 
think that you can rationalize the existing system, and I think that is a 
worthy enough goal that it should be documented as the best approximation 
to the status quo, even if we are going to decide that it isn't good enough.

>Am I letting my personal
>feelings exacerbate this? Probably: I've been defending And and Jorge's
>right to speak, apparently expecting in return a deference I have not
>proven I deserve.

You deserve deference merely for having undertaken the job of herding the 
cats, even if you did not already deserve it by simultaneous demonstration 
of both usage and linguistic background.

>But I do not recant the fact that Lojban has certain axioms; not
>everything is negotiable. I count extensionalism as such an axiom, and
>though I am reluctant to use my veto in general, on this, I feel
>compelled to.

I agree that there are axioms. Since I cannot keep straight in my mind 
intention/extensional and what they refer to, and they did not enter into 
the actual design of the language, I cannot call the distinction axiomatic 
to the design. Given that the contrast seems linguistically fundamental, 
the Lojbanic approach would *normally* be to have "lo" be neutral with 
regard to the matter, with a discursive cmavo and its negation (or a second 
cmavo) indicating the contrast when it is important.

>I will not give reformists everything, and I will not
>give them this; that is more change that I can justify to myself. If
>that's irrational, so be it; but rationality is not the only criterion
>we are working with --- if it were, we would accept all optimisations
>without question. The onus may still be on me to prove that a fully
>extensionalist system will work; but a fully intensionalist system, I
>repeat, is not Lojban as it is defined, *no matter what usage has
>done*. (I reiterate, to Jorge's use of my own usage against me, that on
>this issue, usage doesn't count for anything with me, the gadri system
>being so alien from English to begin with, and so poorly documented and
>understood.) And extensionalism is not a trivial detail of the
>definition of Lojban, but its cornerstone.

Except to the extent that it explicitly appears in CLL, extensionalism is 
not to me a cornerstone. I am resigned to change in the definitions of the 
gadri, so I want them to
a) have le and its series preserve "in mind" (meaning non-veridical and 
probably +specific) as its essential property, probably with ro as the 
default outer quantifier
b) have lo and its series preserve "veridical" (and probably -specific) as 
its essential property, probably with su'o as the default outer quantifier.
c) match usage insofar as is possible (with my own stronger attachment to 
usage of le/lei/loi)

I would prefer that the solution NOT have "lo" as a mere abbreviation for 
"da poi", but will accept it if necessary, if only to respect CLL. I 
certainly believe that this equation, even if it is in CLL, is less than 
fundamental, but derived from a certain semantic theory that is not 
necessarily the one that will be adopted. I would prefer that if there is 
an issue on extensionality/intensionality, that it be explicitly marked by 
cmavo, if this is possible, with the unmarked form being neutral, but 
possibly having a default value depending on the existence of the referent 
(if that last doesn't make sense, then ignore it).

My own usage has generally involved "loi" rather than "lo" among the 
members of that series, so I am more strongly attached to usage-preserving 
"loi"; I'll accept that "lo" usage has been inconsistent with any formal 
definition. (Even if your theory or Jorge's can be rationalized as 
consistent with usage, it is clear that the usage was not made with any 
such theory in mind). The only added guidance has been that English 
non-specific singular "a/an broda" was originally almost always been 
translated with "lo", though later "loi has been sometimes used. Lojban is 
not supposed to have a singular/plural distinction, but loi tended to be 
used for plurals and lo for singulars. JCB's language offers no guidance 
since his equivalent "lo" is closer to our "loi" (but late TLI Loglan usage 
of "lo taksi" for "a taxi" is closer to our "lo" although I think it would 
also work with our "loi").

I am more than a little intrigued by Cowan's claim that Lojban would not 
have such current trouble if JCB had made a certain pronouncement on the 
gadri 10 years earlier, but since JCB did not have "lo", I'm not sure that 
has bearing on the matter. If the gadri have to change, I am more 
sympathetic of a change that would bring us closer to the TLI language as 
respecting the roots of the language (which respect HAS been a fundamental 
axiom of the design), even recognizing that some are antagonistic towards 
JCB.

I'd like the result to be systematic with clear contrasts and examples of 
everyday usage of each form, so that it can be easily learned.

On this issue, I will probably be inclined to vote in favor of anything 
that Nick and Cowan can agree on, especially if it recognizes the 
fundamentals that I listed above (but I will be seeking the opinion of pc 
and his acquiescence, if not full acceptance, in this matter as well), as 
representing the best that "conservatism" can ask for in resolution of the 
issue. I respect the opinions, and especially the usage, of the others in 
the discussion, but none of them are in my mind committed to the history or 
the baseline and hence their solutions are more suspect.

While mentioning that there are lines in the sand, in the area of gadri, 
the turmoil has largely erased those lines, so I state the above only to 
indicate where mine still might lie. But having accepted that there will 
be change in this area, I've already indicated that in this area I cannot 
invoke a deathmatch based on the current prescription, however it may be 
described.

(I think that my insistence on defining the current language is relevant 
here. Nick clearly has some ideas of what is essential about the current 
baseline, and I have different ones, and Cowan possibly others. The point 
of defining the existing baseline before considering changes is to make it 
clear to all of us just what the axioms are that we cannot accept changes 
to. If we cannot even agree on the axioms, we cannot derive any consistent 
result.)

lojbab

-- 
lojbab [email protected]
Bob LeChevalier, Founder, The Logical Language Group
(Opinions are my own; I do not speak for the organization.)
Artificial language Loglan/Lojban: http://www.lojban.org