Date: Thu, 14 Jul 1994 00:53:48 -0400
From: Logical Language Group <lojbab>
Message-Id: <199407140453.AA18930@access1.digex.net>
To: nsn@vis.mu.oz.au
Subject: Re: cukta
Cc: lojbab@access.digex.net, lojban@cuvmb.cc.columbia.edu
Status: RO
X-From-Space-Date: Thu Jul 14 00:54:19 1994
X-From-Space-Address: lojbab

Well, regarding cukta, it is precisely one day too late to make any change
even if I agreed, because I have declared the dictionary baseline of the
gismu list .uocai.o'aru'e.u'u

In addition, I determined by inspection that a physical book is selpapri
a be-paged document - it was already in the gismu list.  I still like
cukta to be a conceptual book which is a copy/manifestation of some work
or collection of works in some medium.  Being essentially conservative,
this was sufficient to let me feel that the issue is not urgent enough to
delay the dictionary until we can debate it.  If the existing definition
doesn't work, someone can propose or start using a different definition
in time for the 2nd edition of the dictionary %^)

I'm going to ponder the rest of what you wrote, and maybe we can discuss
it at LogFest. BUT, I wish to quickly respond on:

  >Lojban
NS> (I claim) is built on a particular view of semantics: one in which {le jei
NS> da broda} is taken as 0 or 1 --- one in which you can make black or white
NS> judgements. X either is a book, or isn't. In that context, it is
NS> *meaningful* to ask "is a collection of blank pages bound together a book
NS> or not?"
NS> 
NS> Now prototype semantics, which is a more, I dunno, postformal view of 
NS> semantics, would go for a fuzzy logic approach to le jei da broda, rather
NS> than a truth-conditional approach. What is a book? Well, a book has
NS> certain prototypical properties: it has pages, it has text printed on
NS> these pages, it conveys recorded discourse. If something has all these
NS> properties, it's a book. If it has none of them, it's no book. 
NS> 
NS> What if it has only some? Like a CD ROM, or an empty book, or an Ionesco
NS> work? Well, though Lojban needn't inherently do so, the underlying bias
NS> of its predicate bias is to make a categorical judgement: this *is* a
NS> book, that *isn't*. Prototype semantics would take an attitude which I
NS> think corresponds more closely to what people actually do --- and to how
NS> meaning changes over time, which I now think is what I'm going to do my
NS> PhD on. It says --- just like a layperson would --- "It's *sort of* a
NS> book." If it's bold enough, it'll go explicitly fuzzy-logical, and say
NS> "It's 0.6 a book," or something. 

As a principle designer of Lojban, and the inventor of "jei", I can testify 
that I introduced "jei" precisely because I considered that most usage of
language requires the acceptance of fuzzy-logic values.  "jei" for me USUALLY
isn't either black or white; I tend to see the world in shades of gray.

Furthermore, at the first DC LogFest in Sept 1986, before the Lojban split was
even conceptually possible, we spent a LONG time debating the meanings of
4 gismu (now tutci, cabra, minji, zukte) to cover the spectrum of machinedom
vs. agenthood, as it might apply to arttificial intelligence, coming up with
words for computers and the various 'semi-intelligent tools and machines that
seem likely to emerge in the near future.  During this discussion, essentially
the first Lojban semantics discussion, built around the concept of partitioning
up a semantic space in a fuzzy-logic kind of way (one of the participants 
had recently gotten into fuzzy logic in a big way, and his arguments were
VERY convincing and I think germinal on a lot of what has transpired in Lojban
semantics).

In my readings of lexicography, I have seen that an almost identical concept
of fuzzy boundaries to the semantics space of eachj word is part of the
conceptual framework for dictionary definition writing - you try to come
up with the definition that makes all instances of the word meet some kind
of fuzzy logic threshold for inclusion in the set meeting the definition.

I further contend that JCB, having built the cornerstone of Loglan usage around
"le", the intensional descriptor, rather than something like "lo" or "da poi",
also recognized and considered paramount the intensional and hence fuzzy nature
of hukan language semantics.

The fact that predcate logic itself was designed around absolute truth values
has tended to be irrelevant to Loglan/Lojban, and indeed, we have found that
the heavy use of intensionals has largely supplanted a tendency towards
formal logical syntax (though I think I actually use it more than most since
I am reasonably comfortable with prenexes in my grammar, and prefer them to
figuring out the logical manipulations).

lojbab