Re: Existing ways to fuzz

[Logical Language Group <lojbab@ACCESS.DIGEX.NET>]

>lojbab:
>>I'm not sure how onw would substantiate it.  Even Steven said that there
>>were ways to express fuzzy stuff in 15 syllables.  I am as yet unconvinced
>>that there is need for anyuthing shorter.  How many will use it?  If only
>>Steven, we are working hard for a small portion of the audience.

>A valid concern. The few participants in this discussion are not a
>statistically valid sample of all potential speakers on the planet. We need
>to think carefully about what the needs of these speakers are. Since the
>participants in this discussion are all heirs of the
>Western/Aristotlean/Zoroastrian tradition, we must take special care to
>respect the future needs of persons from other traditions.


Yes indeed.  We have made considerable effort to do so.  I should note
that pc not only is well-versed in Western logics, he has done fairly
extensive study in Sanskrit and Chinese logics.

>Does lojbab
>believe it is enough to pack the language with Mandarin gismu?

Hardly.

>Kosko and
>others have reported that Chinese & Japanese cultures are essentially
>fuzzy-logic based rather than two-valued logic based.

REference for benefit of pc??

>Quick, lets rush the language into baselining before we understand the
>needs of Chinese speakers!


Common sense tells us that if we wait until we understand the linguistic
needs of all the world's languages, we will never have a language.

In any event, it is not my choice.  I have been directed by the membership
to "rush thelanguage into baselining" as fast as we can get the books done.

The point is, that we would need not only to know that some other language
had a fuzzy logic implementation built in, we would need a fairly
complete description thereof in order to model it in Lojban. Nebulous
references to the difficulty of translation are not enough - what are the
 LINGUISTIC details of Chinese fuzzy linguistics.

>>Reiterating, pc and I have said that there are resources in the language
>>to express most anything.  (IF he has only said it obliquely recently,
>>be sure that he has said it directly before; e.g. on tense issues).  The main
>>arguments seem to be whether we need short forms to make the distinctions.
>>And it is harder to justify new grammar or new cmavo merely to shorten a
>>little used are aof the language.
>
>Perhaps suffficient resources are present. But every other question I've
>asked about "how to say something" was readily answered almost immediately.

And there is something wrong with this - that we can answer "how to say it"
queations?  Or are you saying that fuzzy expressions alone are the
exception.  In which case I will cite 1 1/2 years of opaque discussions
of how to express opacity to show where the REALLY hard questions are.

>If fuzzy logic can already be expressed, show us an *non-idiomatic,
>non-conflicting, general* way of doing it. Where's the beef?

I don't know what means "non-idiomatic", "non-conflicting-with-what?".
I have already said that you can say damn near anything "generally" using
sei metalinguistic sentences.  If "non-idiomatic" means anything for
Lojban, then you are INSISTING that all expressions of fuzzy logic values
be full bridi, even at the expense of wordiness, because there is very
little in Lojban that isn't in some way "idiomatic" that is a condensation
from a longer, more fully logical, bridi expression.  All of tense grammar is
"idiomatic" in a sense.  Most of MEX is similarly so.  pc has pointed out that
to logic, the use of quantifiers other than the traditional some/all/none/one
are abbreviations for very long-winded logical expansions.

In any event, if it turns out that some language, or even one we think we know
about, is later analyzed by linguistics to show a feature like fuzziness to
exist, then I contend we should design a solution at that point to solve a
known problems, rather than to try to solve an ill described problem with
a solution that does not necessarily reflect the needs of other languages
any more than the status quo.  Because if there is one thing I am fairly sure
of, any Chinese fuzziness is NOT going to be of the mathematically precise
sort that you have been seeking, probably is not open-ended as to degrees of
fuzziness (in other words the so'a/so'u quantifiers might be more than
suffcient, as might the scalar indicatiors cai/sai/ru'e/cu'i etc.

>This is the core of the problem. You really don't understand even the bare
>essentials of fuzzy logic. Please read Kosko's book, "Fuzzy Thinking". (I
>disagree with some of Kosko's wild extrapolations, but his book is an
>entertaining & painless way to get an intelligent layman's overview of
>fuzzy logic in a weekend. Worth reading, even if you are *not* designing a
>logical language. Essential if you are.)

I am NOT going to do so, because I am really not designing a language any more,
 merely documenting a design that is completed, subject to noise level.
I don't have time to read McCawley, who also covers fuzzy logic, and probably
from a more balanced persepctive since he isn't trying to advocate, as
your brief description says that Kosko is attempting to do.  There are many
other linguistics books I haven;t read as well, that I feel I should read.
I DO NOT rely on my personal expertise in linguistics, logic, or other
languages to justify my design efforts.

>>I do know the underlying philsophy of Lojban; i.e. that version of Loglan
>>thatwe are implementing, since I think I can claim to have largely defined
>>it.
>
>Your grasp of the underlying philosophy of Loglan as described by JCB is
>not apparent in your statements regarding fuzziness. Fuzzy logic is not
>some esoteric thing separate from logic. In fact, true/false logic is a
>subset of fuzzy logic. Fuzzy logic is a *more general* logic than
>two-valued logic.

Fine well and good, but JCB has no more embraced fuzzy logic than we have.
JCB has in fact done a poorer job of covering traidtional logic than we have.
I am trying to complete JCB's language; I am NOT rrying to tread new ground
to make some hyperlogical language.  We don't handle 2nd order predicate logic t
too well either, but neither pc nor Cowan nor I really consider this a flaw
wirth going back to the drawing boards over.  Lojban cannot be all things to
all people.  We have doe the best we can given our resources.  I cannot
justify continued puttering with the design.

>A larg body of evidence suggests fuzzy logic is common in natlang
>utterances, despite the clumsy mechanisms in natlangs forhandling fuzz.

Fine.  But Lojban is trying to model natlangs except in certain well-defined
ways.  Handling fuzz is not one of those ways.  Therfore we are not
justified in "improving upon" the clumsy mechanisms of natlangs.  We WANT the
language to feel natural.  We are NOT trying for "spoken logic" with
built in thorough Montague semantics associted with every concept of the
language.

>Principle: Cultural Neutrality
>Assessment:Lojban design ignores inherent fuzziness of languages of >1.5
>billion potential speakers.


UNproven that any language has fuzzy features that Lojban does not have.

JCB did not design anything more fuzzy than the status quo Lojban into
any version of Loglan.

>Principle: Syntactic Unambiguity
>Assessment: Lojban design relies on idiomatic conflict-inherent mechanisms
>such as metaphors, subscripting, and overloading previously defined
>operators to express fuzziness.

This is not a vioaltion of syntactic unambiguity.  It is a possible
violation of semantic unambiguity, which Lojban does not disdain but indeed
revels in.  But of course Lojban has a more firm grasp of rules of metaphor
than JCB's versions.

I think that the ke'a arguments show that I am quite averse to overloading
operators, and indeed you will note that I have indicated that I am less
than happy with the XI proposal.  But I don't view the XI proposal as
being contradictory to any other usage in the two areas it is proposed
for use (je'a and ja'a), since no one has proposed any other use for
subscripts in that context.  ALL use of subscripts is "idiomatic" as to
what the semantics are, and I don;t see this particular application to be
inconsistent with the range of uses for subscrpts.  If it is, then we could
add another cmavo to XI with same grammar but some more explicit fuzzy
semantics.  But I don;t think that a need for this has been demonstrated.

For all you have claimed that you don;t like what we have, we HAVR been able to
come up with a way to express all the actual linguistic usage problems you have
posed.  One can come up with theoretical problems that would be difficult,
but I need to see a natural language problem that we cannot convey easily.
I am not impressed by complaints that our solutions rely on idiomatic
interpretations, unless those idioms go against the grain of other Lojban
idioms.  But still, there has been no problem that we have not solved more
clearly than the English that you took to express the problem.

>Principle: Parsimony
>Assessment: Clunky, long-winded operators needed for even simple fuzzy
>expressions. Wouldn't even attempt to make complex fuzzy expressions.


You apparently don't understand what metaphysical parsimony means and it is
ONLY parsimony of the metaphysical sort that JCB claimed as  a design
principle.  parsimony has NOTHING to do withj "clunkiness" or "long-windedness"
and has only to do with with features of the language are obligatory when they
are not metaphysically necessary.

Asto whether you wouldn;t attempt to make a complex fuzzy expression, that
is your choice.  But can you cite a complex natural language fuzzy expression
from CHinese or any other language that we cannot handle at least as clearly
or concisely?

>Principle: Logically Powerful
>Assessment: Implements only special case of two-valued logic in a general,
>elegant fashion; kluges and gimicks used to handle more general case of
>fuzzy logic.

This was JCB's design.    Compared to natlangs it IS logically powerful.
Indeed probably too powerful for natural language users toever master all
the logical features of it.  Whether it is more powerful than some theoretical
abstraction of logic is irrelevant.  Loglan/Lojban is logically powerful on
a natural language scale.

One man's kludge or gimmick, is another man's elegant solution.  I am not
impressed by loaded words.  I will admit that some probablems have been
solved by serendipity rather than predetermined design.  But we were not and are
 not attempting to go back and redesign the language fundamentals - that goes
against the ground rules ofour revolt against JCB - the language must still
be recognizable as JCB's language, and we are NOt reinventing wheels.

>Principle: Suitability for Testing the Sapir-Whorf hypothesis
>Assessment: Hypothesis is that natlangs contribute to depression by
>encouraging cognitive distortions (an invariable feature of depression).
>Depressed people *frequently* utter false dichotomies and selective
>abstractions. lojban fails to capitalie on an opportunity to test an
>important possible instance of Sapir-Whorf effect.

Lojban fails tocapitalize on many opportunities to test the Sapir-Whorf
hypothesis.  JCB built in only one such test by intent - the underlying
predicate grammar.  That >IS< the be-all and end-all of what is necessary
for a SWH test according to JCB's intent.  Indeed it could be argued that
any other feature that is not strictly according to natlangs OTHER than
standard predicate logic, introduces unnecessary variables.

Now I contend that JCB unintentionally introduced several other features
into the language that are more suitable for and more likely to reveal a
sapir-Whorf effect thanthe mere introduction of a predicate logic grammar
and logical operators.  And I have done my best to clean those up to make
their potential use clearer.  But there is NOTHING in JCB's mandate to
go loading up the language with new features to test other possible Sapir-
Whorf effects.
>lojban without fuzzy logic is like arithmetic without rational numbers. Our
>slogan ought to be:
>
> lojban: The COBOL of logical languages.

Actually, my wife has been more afraid that we are becoming the PL/1 of
logical languages - and she was a skilled PL/1 user, becasue of all the bells
and whjistles that havebeen hung on the language.

But of all the programming languages, COBOL for all its clunkiness is among the
most natural language-like of them, so your insult is not necessarily as
devastating as you migfht have planned.  At least we are not the APL of
logical languages.  Nor a logical Assembly Language.  %^)

>Who is the one ignoring history here? If you won't reread JCB's books, at
>least check out the back covers!

I have read JCB's books at least as many times as you have.  I also have
read the less available Loglan 2 where he elaborates on his rationale.
And I have talked with the man for hours about the philsophy of language
design.

>You are *not* adhering to JCB's design
>principles regarding fuzzy logic, leaving a gaping hole in the language.

You have yet to show me any JCB quote that mentions fuzzy logic much less
in the context of a design principle.

Nor have you yet demonstarted a gaping hole in the language, since you have
not come up with something not clearly expressible.  I don;t give a damn
about aesthetics when it comes to functionality.  If you will accuse us of
having a hole in the language, it had better be a real hole, not something
that is aesthetically displeasing if you make different assumtpions, not
something that with nary a new cmavo I can render invisible.

>You admit that you lack the knowledge to decide whether fuzzy logic is
>important, and apparently have no intention of correcting this defect.

I am human.  I have limits.  mastering all subjects that MIGHT be important
is beyond them.  You have yet to make a case that your subject is MORE
important or more likely than most to reveal a deep flaw in the language.
There is as yet NO evidence provided that natural language uses any more
than a rudimentary version of fuzziness.  Guttman scales are a far more
elegant anlaysis, and I think we have them covered as well, but ther is
no evidence that natlangs use anything as sophisticated as a Guttman scale
in ordinary expression.  I don;t care what might be POSSIBLE.  What do real
speakers actually do.

>If
>lojban goes to baseline without fuzziness elegantly supported, we will all
>come to regret it.

There goes the elegnace argument again.  Elegance is a luxury.  We are not
trying to redesign a language for elegance; we are trying to complete
JCB's original design with minimal additional effort.  JCB has yet to
complete his language, for all the books he has written, he doesn't have a
fraction of the documented definition that is inthe refgrammar.  Our
goal wqas originally, and remains, to get a documented language out there and
use it.  If it has some weakenesses, then we will address that problem
when we know what it is.

>That's perhaps just my opinion, but at least it is an
>*informed* opinion.

You are informed about one area of the language problem that is very important
to you.  pc is informed about other areas, Nick about others, Cowan about
others.  My job is to put the pieces together, not to form opinions.  I
am not, cannot be, and will not try to pretend to be, someone expert on all
aspects of language, much less things like fuzziness that are not proven to be
as fundamental as you are arguing that it is.  The latter will take some
scholarly peer-reviewed papers in Chinese linguistics and not a thus-far
unsubstantiated claim from awriter who you admit is prone to overextrapolation.

lojbab