Date: Wed, 14 Feb 1996 03:30:59 -0500
From: Logical Language Group <lojbab@access.digex.net>
Message-Id: <199602140830.DAA10342@access1.digex.net>
To: pcliffje@crl.com
Subject: Re: TO PC ONLY, are you reading mail at all?
Cc: cowan@ccil.org
Content-Length: 12975

Re: TO PC ONLY, are you reading mail at all?

>In any case, drop me a note down the net about anything you want my 
>opinion on, flagging it as you did this one, and I will get back within 
>a couple of days, from wherever.

I think the way we hedged the ro inner quantifier on lo was that the
quantifier was "ro" if broda existed, and obviously no if it didn't.
But given this whol bit about empty universes, and empty sets in
non-empty universes, I am not sure it matters.  The whole reason we have
a default quantifier, of course, is because you put the bloody things in
there back in the 70s.  We can eliminate them most immediately by saying
that the inner quantifier is undefined/meaningless if not specified, in
the case of "lo".

>I can't think of what else I am unhappy about, except the apparent loss 
>of opaque places in things like _kaltu_ and _sisku_.

I don't know what an opaque place is, or how one specifies it.  JCB now
seems to think that loi broda is an opaque broda, patterned in turn
after your/his interpretation of TLI "lo predA" as an observative -
don't ask me how he gets from one to the other.  Nora favors using a
conventional quantifier like "ropa broda" = "any one broda" noting that
it solves the problem when you want a non-singular opaque referent
(which loi broda does only with a numberMEI tanru which is sloppy).
Thus ropano broda is any 10 of the brodas.  I am not sure if this is the
same as ro lo pano lo broda in terms of the opaque reference problem.

I'm not sure what change there has been in this area.  Before you were
on line or thereabouts, We would have simply said "mi sisku lo finpe"
and there was no question about opacity.  The year you got on, Iain
identified sisku in particular as a problem, and after consulting
brioefly with you by phone, Cowan had me change it to mu sisku loka
finpe [in set x3] - you don't search for things, but rather properties
of those things.  Everyone agreed this was clunky, but we thought it
necessary.  There was no serious discussion of opacity until Sept 94
when the any discussion started, and there have been NO changes at all,
as far as I am concerned because absolutely nothing was ever settled.
However somewhere in there, someone DID say that the sisku problem was
not unique to that gismu, and that whatever solution we found to the
opaque problem would also solve sisku, and that therefore we did not
need the clumsy place structure.  So I reverted it to the longstanding
norm.

>I see that fuzzitude is back (catching up with and passing even the 
>political volume). 

The political volume is of course a direct offshoot of the fuuzy
question, so getting back tot he technical issue will eliminate the
politics.

>I stay out of that until I can get Belknap et al to 
>state clearly (and, I think, get clear themselves) which of the half 
>dozen things called fuzzy logic (and dozens of sub systems) they want 
>in.

I can ask him directly and privately to state this.  My own feeling is
that we ought to be able to cover whatever McCawley says, since you have
come to use it as a standard in other areas or
logic-as-it-applies-to-Lojban.

I suspect that the primary issue to be dealt with is how to stick a
quanitfier between 0 and 1 onto one of several gramamtical constructs so
as to convey the scalar truth/applicability of that component to the
rest of the utterance.  The ja'a+quantifier solution is thus only a
special case of what Belknap is looking for, and he will not be
satisfied in the long term with it, which is why I oppose And's XVV
solution, and am not happy with the subscript solution (though another
member of XI that specifically was called the fuzziness factor would be
OK with me).

I am also unimpressed by the todo, but I figure that to ignore it risks
Belknap becoming another Carter, and I am already afraid that between
And and Jorge, we may already have more than enough Carters among us who
will apriori decide to ignore the prescription for their own aesthetic
or quasi-logical reasons.  It will happen eventually that the language
will drift from the prescription, but I want so badly for it to be a
natural drift and not a decision that people consciously make.

The third issue, and the one most immediately needing your feedback is
Goran's plea.  We have never addressed the issue about how to specify an
interval size or offset exactly, always opting for the fuzzy zi/za/zu
vi/va/vu ve'a/ve'i/ve'u ze'a/ze'i/ze'u.  I am sure that JCB never
addressed how to say "I went to Glasgow 6 months ago", and example of a
specific scalar offset.  And there is nothing in our design discussions
of Lojban about this, since whenever you and I talked, it was always
about the space time reference, and not about anchoring fixed length
intervals to same.

If you feel you know the current Lojban tense words/concepts well
enough, the obvious thing is to take up your challenge - that you can
find a way using the tense system we designed that clearly expresses
anything that is expressable in English (or other natlang).  How would
YOU express the above.

If we have a hole, how should we address it?  One obvious way that comes
to me just by stating the problem in the way I just did is to make a
construct that is XVV + time interval which is equivalent in the grammar
to each of the 4 selma'o VA ZA VEhA ZEhA.  But this gets tricky because
the time interval cannot be specified using only lexer-words and hence
must be handled at the non-lexer level, which is after the complete
tense has been assembled, yet we need a substitute for the words inside
the lexer where those four selma'o are used to assemble the tense.
(Hope this makes sense to a non-YACCer %^) Only by imposing an extremely
restricted grammar can we put an interval description into the lexer -
something like number+brivla - and that is misleading.)

The last issue, alluded to above, and I think in my post, is that JCB
has rewritten history and terminology, and done so neatly enough that
few will catch him on it, meanwhile possibly solving the opaque sumti
problem and a raft of other problems.  The last 2 LKs are critical to
understanding the stuff, and the main question is whether you can stand
to read JCB's word- twisting yourself, or whether you want to have it
filtered after Cowan attempts to digest it into plain English.

I THINK that there are two major things covered in his writings that are
in need of consideration.

The first is that, starting from one person's questions about
subjunctive case, they have gone to none other than McCawley and his
discussion of modal logic, and "possible worlds", and there are at least
3 versions of major proposals being considered to implement a
subjunctive in TLI Loglan.  JCB states that he thinks that the import of
this will rival GMR, which means that HE at least thinks thatit is a big
deal - but because of the plethora of proposals, they plan to waiot at
least a year before deciding on one.  (They also may be waiting until we
baseline in hopes of pulling a technical coup on us, which is one reason
I am nervous.)  The two lognet issues have discussions of two issues
related to this, one being James Jennings original proposal based on
McCawley, and the opther being a more general discussion by Jerome
Frazee.  I think you may know both these people - Jerome was the
qudraplegic amateur logician in california.  I can;t remember anything
about Jennings, and may be confusing him with Jenner as to whether he is
an oldtimer (heard anything from Jenner in recent years???).

So the quick question is - do we handle subjunctives/possible worlds
adequately as defined in McCawley (I think Chap 11 and 15 were referred
to in Lognet) and more generally do we handle modal logics assuming that
is a different question.  The only real resources we have now are the
weak subjunctive UI "da'i" = supposing vs. da'inai = in fact, and the
suggestion that they are up to something big suggest that we aren;t even
seeing the whole problem.

This subjunctive thing is something I think we can let you do as a
working issue while you travel or whenever, rather than something we
need an immediate final answer on (though your short reaction to the
issue in general might stop my stomach from fluttering or make it turn
upside down, and either might make be feel better than uncertainty).

The other topic is that JCB has completely reexamined his temrinology,
and hence his gadri and concepts dealing with sets, multiples and
masses.  His article gives some new history, though you may have known
it - his original 1956 book among aother things attempted to code into
predicate logic some "grammatical curiousities" from Jesperson's 1937
Analytic Syntax.  I presume you are familair with this book and what the
"grammatical curiosities" are?  Is the book worth reading, BTW (I mean
should we sic someone not book-writing in the community to reading it
and summarizing what is interesting and relevant, or is the book and the
referenced "curiosities" woefully obsolete.)

Briefly, what I think he now says is

a) His ze (our joi) forms a collective/team/jointly, which is not a
Trobriand Islander mass - indeed ze entered the language after 1963 -
while his lo was already present - he thinks added in response to
Quine's "Word and Object" 1960 which discusses two world views, one
apparently the Trobriand Islander version and one the standard IE
version, and shows their equivalence (I assume you know this reference).
As a result JCB decided to make the Trobriand metaphysics an optional
usage in the language.  But he notes that he had already read about the
Trobrianders in stuff by Malinowski and Lee.

b) He uses "multiple" for sentences based on compressing many utterances
into a shorter one, as with  da .e. de broda expanding into 2 sentences,
and ci le broda cu brode expanding into 3 sentences.

c) He uses "set", harkening to the old "lea" to refer to what he calls a
human linguistic "set", which he says has nothing to do with logic's
definition of set.  The bottom line is that this is what I think we have
called a "collective", and he somehow mashes all this together and ties
it to lea and ze and a couple of new cmavo so that he uses our set
descriptors as collectives, and totally repudiates the logical set as a
linguistic construction.  At least I think so maybe it survives in terms
of "me" or something else.

d) as mentioned above, based on observative "lo" (our loi), he now
considers that construct to be an opaque reference, citing the example
of "waiting for lo taksi" which he claims is an opaque taxi.  But he
then seems to go on and contradict himself and say that that lo is very
much the Trobriander concept, and that (I think he is saying) a piece of
a taxi is also lo taksi, as well as many taxis.  (But then if this is
supposed to be his opaque reference, it is unclear how he gets 2 opaque
taxis.)

He also dabbles into some other heavy stuff that may or may not be
important, including discussions of quantifiers - and I am curious to
find out whether what he says has anything to do with what you have said
%^).  There are also a couple letters from Randall Holmes, his official
logician, that may be important or may not be - he is extremely hard to
read.  Holmes apparently is not a dominant force like you were, in that
JCB has several "pet logicians" including Jennings, someone named
Emerson Mitchell, Frazee, and Holmes, who form a "logic-working-group",
with Holmes at best only first among equals.  Holmes also did an
analysis on the logic of "respectively" in a recent Lognet that may be
important.

Summary - I think, is that his version of le'i is a jointly-collective
and not a set as we consider it, and hence two men carry a log are le'i
in his mind and not lei.  And loi is something else.

All in all, too much meat, and too arcanely worded, for me to judge the
importance, but it is scary that they seem to be able to talk in such
depth inthe first place - he has again got some solid people, and that
JCB seems to be able to pull off such an about-face, as he does (he
specifically nullifies what he wrote in 1989 L1, passing it off almost
as if it were a typo, rather than an about-face).  And I don't want him
to solve something that we haven't solved prior to our book publishing,
if I can help it.

Some of the stuff may be on-line on Jennings Web page, and assuming you
have a web browser, you can read it at your leisure.  I will give you
URLs when I check tonight.  I can send you snail copies otherwise if you
want to read stuff.

FINALLY, the last open issue is Jorge's request for an ordinal ROI, in
order to be able to talk about the second occurance of some event as a
parallel to the two occurances of some event.  This seemed plausible
enough that we even reserved a cmavo for it, but you never have said
anything, and it has not been written into the language.  I don't know
how much of an issue this is or needs to be.  Cowan has been ambivalent.

lojbab