gadri

[Nick Nicholas <opoudjis@optushome.com.au>]

I am finding this more and more dis-spiriting. I'm giving Jorge a 
bollocking in the next email, but other than that, I am fast despairing 
of this whole venture.

Let me try and sum up the outstanding gadri issues as I see them. (This 
is a draft of the letter I will eventually have to send out to the 
BPFK. Remember that?)

1. loi

Lojbanmass conflates individuals, collectives (tuples of entities), and 
substances (uncountable agglomerations of entities or sections 
thereof.) This is in fact by founder design (which leads me, yet again, 
to despise the founders). Distinguishing between these three subsenses 
seems to me laudibly meritorious --- "can you or can you not in Lojban 
say "together" or "separately"; but quantification seems only partially 
a way of doing it.

A substance is loi tu'o broda
Individuals, collectives, and substances of collectives are all loi 
tu'o ro broda
I think the collective vs. substance of collective conflation is bogus, 
because a substance of collective can *still* be phrased as loi tu'o 
broda; and such second step abstractions are pragmatically more like 
loi tu'o loi ro broda, if you're going to put quantifiers in there. 
(Yes, that's grammatical.)

But I'm tired of these somersaults. Since Lojban already distinguishes 
between joi and jo'u, I'm back to thinking we should get a separate 
LAhE and be done with it for collective.

I note in passing that the gismu definitions do treat sets as n-tuples, 
just as Bob's addled recollection leads me to believe. (No, I don't 
like what he did. Because I'm stuck trying to clear up the mess. And I 
don't give a fuck if the mess is originally James Brown's.) Why else 
use sets in the gismu definitions where he did?

It is distressingly clear that Lojban is shortchanged of any tools to 
distinguish between collectives and substances, and I'll settle even 
for gismu at this stage.

Bob says the gismu for collectives might be kampu, The x2 of kampu 
is... a set! Ergo, Bob thinks of sets as collectives/tuples. Thanks a 
bunch.

2. lo'e

The candidate senses are: Statistical (e.g. mode), Prototype (mental 
definition), and Unique. The founders (addled once more) seem to have 
preferred Prototype (which matches stereotype), I've retracted 
Statistical, and And has retracted Unique (apparently), so this is 
resolved.

A secondary issue is whether anything inherited can be predicated of 
lo'e: do you study or draw lo'e cinfo, or is that only meaningful as a 
definitional trait? To keep our sanity, the latter.

3. Unique

And has fumbled this badly, as he will himself admit, but people are 
slowly starting to see the point to this construct. It was begotten of 
kludgery, and pressed into service to solve everything, but it is 
handy. It is a generic like the prototype. making an individual of a 
population by regarding the actual individuals of the population 
instantiations of the one underlying individual.

By speaking of Lions that are all underlyingly the same lion, And had 
made his interlocutors dispute his sanity. Speaking of Mondays, or the 
New York Times --- things we are used to abstracting an underlying 
individual out of instances --- would have helped him rather more. The 
Unique is an individuated version of the Kind (as in "I own that kind 
of car", treating all Porsches as the one thing), which treats the 
population as a single entity. As such, it corresponds to many an 
English use of the generic.

This looks headed for its own LAhE.

4. Intensional article

We may or may not need one.

If propositionalism rules the land, we won't need one for wanting, 
needing, or seeking. mi djica tu'a lo broda, mi nitcu tu'a lo broda, 
and mi sisku leka broda already mean exactly what we want. Because they 
have prenexes by default on the inner predication.

The problem arises with imagination preds, although it laps at the 
others as well. If the quantification is limited to this world (so 
that, for all {x|broda(x)}, the Any-x can be paraphrased as x1 .a x2 .a 
x3... , just as ro lo broda can be paraphrased as x1 .e x2 .e x3... and 
piro loi broda as x1 joi x2 joi x3... --- and hopefully, piro lu'oi 
broda as x1 jo'u x2 jo'u x3...),

then wanting a doctor can be interpreted as wanting one of the 5 
zillion doctors practicing today on this planet. Same with needing. And 
seeking \x:doctor(x) can range over the set of all doctors practicing 
today on the planet.

But when you include Dr Quinn Medicine Woman among your doctors, you're 
ranging beyond this planet. You're getting into imaginary doctors. You 
don't want a made-up doctor when you want one, need one, or seek one. 
Whatever the 'objective' existence of what you put into those preds, 
you the speaker do believe that a referent exists in this world. 
Usually. So, by Gricean default (and using the resources of Classical 
Lojban),

mi djica tu'a lo mikce

means

mi djica lenu
le ca'a munje
su'o da poi mikce
zi'e poi se vasru le ca'a munje
zo'u: mi penmi da

Usually, I say, Because we do also have exchanges like this:

"I want a doctor"
"Yeah, and I want a gold toilet, but it ain't gonna happen."

Since people can want the impossible, there is the possibility that you 
can actually be saying

mi djica lenu
su'o de poi munje
zi'e po'u na'ebo le ca'a munje
su'o da poi mikce
zi'e poi se vasru de
zo'u: mi penmi da

That's dispreferred for wanting, needing, seeking. But for depicting 
it's entirely possible, and for imagining it's almost mandatory.

When you depict a mermaid, you know that she's not in this world, but 
in a made-up world. So you can't say

mi te pixra lo fipni'u

because that's tied up with the initial prenex, so it comes out as

su'o da poi fipni'u zo'u: mi te pixra da

which means

su'o da poi fipni'u zi'e poi se vasru leca'a munje zo'u:
mi te pixra da

Even if we had propositionalism here (and it would not kill us), we'd 
still be defaulting to quantification in this world:

mi te pixra lenu
su'o da poi fipni'u
zi'e poi se vasru le ca'a munje
zo'u: da co'e

mi te pixra le ka
ce'u noi se vasru le ca'a munje
cu fipni'u

But when we know we're imagining stuff, we can go to imaginary worlds:

mi te pixra lenu
su'o de poi munje ku'o
su'o da poi fipni'u
zi'e poi se vasru de
zo'u: da co'e

mi te pixra le ka
su'o de poi munje zo'u
ce'u noi se vasru de
cu fipni'u

Maybe this kind of propositionalistic solution could make it a 
pragmatic default of the predicates involved that they leave the world 
of the quantification open --- whereas wanting/needing/seeking default 
to this world.

And with that, we'd have a solution. It would be butt ugly, but we 
could live with it.

da'i and ka'e may do things with imaginary worlds that would have to be 
worked out, and that may end up being conscripted in this story.

There are a couple of outs.

The Lojbanmass will do for wanting/needing/seeking, at a pinch, because 
it includes in its denotation the Any-x: mi nitcu loi mikce is true of 
{x1 .a x2 .a x3...}. It's no good once you go outside of the real world 
into imaginaries, though. So it doesn't help with imagining or 
depicting.

When you draw a doctor, you could be said to be drawing a depiction of 
the prototype in your mind. Now, the Cowan solution to the problem of 
inheritance, which And has accepted, is that you can't make secondary 
claims of prototypes, only generic claims: you can't say {mi kavbu lo'e 
cinfo}. You also can't say, for that reason, {mi te pixra lo'e cinfo}, 
because that would make the lion defined as something you draw. But you 
can claim {mi te pixra la'e lo'e cinfo}

Another alternative is the Unique: the Unique is a population-based 
generic, and abstracts away differences between the individuals in the 
population. So the picture becomes a picture of lion-dom.

If we accept that le broda need not have a referent in this world (so 
we can speak of {le fipni'u} or {le xavlerfu gismu} meaningfully}, then 
{le} is already non-commital as to whether its referent exists in this 
world or not, by virtue of its nonveridicality. What would serve as an 
intensional article then would be a counterpart to {le}, with no 
specificity. And would invent a new paradigm of gadri; I would be 
rather happier with a UI kludge.

And says that Unique is not veridical either. It is true that in 
English we would say "Lions live in Africa" and "Elves live in Mordor". 
But I think the assumption that prototypes and Uniques in Lojban are 
non-veridical is rash.

The intensional reference ranges over individuals, lojbanmasses, and 
sets. Therefore a new gadrow (to use And's terminology) for the 
intensional is ill-judged: there cannot be just one intensional 
article. It will be either a whole suite, or the intensional shall be 
done by other means. For obvious reasons, I prefer the other means. 
These include both propositionalism and continued kludges involving 
ka...ce'u, possibly incorporating something like ka'ai.

###
ki egeire arga ta sthqia ta qlimmena;#Nick Nicholas, French/Italian
san ahdoni pou se nuxtia anoijiata # University of Melbourne
thn wra pou kelahda epnixth, wimena! # nickn@unimelb.edu.au
stis murwdies kai st' anqismena bata.# http://www.opoudjis.net
-- N. Kazantzakhs, Tertsines: Xristos#