Message-ID: <20060322180209.95713.qmail@web81303.mail.mud.yahoo.com>
Date: Wed, 22 Mar 2006 10:02:09 -0800 (PST)
In-Reply-To: <925d17560603220848w1479c333x344ef414cf0a92e1@mail.gmail.com>
MIME-Version: 1.0
Content-Type: text/plain; charset=iso-8859-1
Content-Transfer-Encoding: 8bit
Errors-to: lojban-list-bounce@lojban.org
To: lojban@yahoogroups.com
From: John E Clifford <lojban-out@lojban.org>
Reply-To: clifford-j@sbcglobal.net
Subject: [lojban] Re: semantic primes
Content-Length: 4679
Lines: 155


> > > > Well, there is the logical argument for
> there
> > > > being a single distinctive list of
> primes.
> > >
> > > A not very convincing one, for my taste.
> >
> > I'm never sure what will convince you of
> > anything.  this eems pretty
> knock-down-drag-out:
> > if there are no primes then all definitions
> are
> > ultimately circular.  To be sure, we can
> avoid
> > this in practical terms for a very long time,
> > maybe forever, but the threat is always there
> in
> > primeless systems.
> 
> But where is the argument that language, which
> is a practical
> system that obviously works in practice, is not
> primeless?

I thought that was what I gave.  I suppose that,
if language never involves its own definitions
(which is possible but rarely occurs in
practice), then the argument doesn't have any
force.  But I would think that the ability to
define itself is inherent in language and thus
the argument applies -- even as a practical
matter.


> Or, if it were based on primes, that there is
> one privileged
> set over all others?

Ah, that is one of the real problems.  I don't
suppose a good case can be made that it has to be
one set rather than another. The NSM set claims
to be universal, which would probably restrict it
somewhat, but I think that even then there are
alternatives which would work equally well.  the
advantage of the NSM set is that someone has done
a lot of the scut work and no one has for other
possibilities.

> 
> > > At first glance it seems rather inadequate.
> It
> > > is not very clear
> > > why they have so many pairs of opposites
> > > instead of just having
> > > OPPOSITE as a prime. (I'm not even sure how
> > > they define OPPOSITE
> > > in terms of their primes.)
> > >
> > I asked about that.  The simple fact is that
> many
> > languages (English, for example) don't have
> an
> > OPPOSITE that functions in the appropriate
> way
> > (like Esperanto mal- or aUI y-).
> 
> What would be the problem with defining "bad"
> as
> "OPPOSITE of GOOD"? Why would you need to have
> a preffix meaning "opposite"?

Oh, it doesn't have to be a prefix, just a fixed
expression of some sort.  And the problem of
defining "bad" as the opposite of "good" is that
this definition does not give an adjective in
form; that is, it is not a definition in NSM
terms.
 
> >  Defining
> > OPPOSITE may be a problem but it is
> relatively
> > insignificant compared to the problems with
> color
> > words, natural kinds and artifacts, most of
> which
> > are dealt with so far by unreliable verbal
> > pointing (green is the color of grass --
> without
> > using the words "color" or "grass" -- which
> only
> > works for people that have grass (and can
> > distinguish it in the definition, "things
> grow
> > out of the ground," from trees and
> mushrooms)).
> 
> I would think that's a good definition for
> green, once you
> have "grass" and "color" defined, and you could
> also add
> "the color between such and such in the
> spectrum", so that
> you have even more reference points.  I find
> "opposite", being
> more abstract, more difficult to define in
> simpler terms.

Well, as noted, this definition only works where
there is green grass (not everywhere by any
means) and it still doesn't get you "color".  To
go on from there as you suggest requires a
familiarity with the notion of the visible
spectrum which is not universally (indeed not
widely) available.  NSM builds a lot of
conditions into acceptability for definitions and
the like and this misses on several of these
counts, for all that it is offered by leading
NSMers.


> > The curious thing is that these problems have
> > been mentioned since 1972 for the project and
> > amazingly little has been done to solve them,
> > suggesting to the impatient observer that
> they
> > cannot be solved within the present
> framework.
> > And no obvious extension has turned up
> either.
> 
> Which makes the whole idea that everything can
> be gotten down
> to a few dozen primes very dubious.
> 
Yeah, NSM is not the generally accepted view in
Linguistics.  The only thing that can be said in
its defense -- aside from the fact that most of
the definitions presented are very illum8inating
-- is that generally the objections to it are so
abominably bad: off the mark, ad hominem, simply
wrong and so on. the few good ones are killers,
though, but they only so far demonstrate that
this particular set is not yet one of the ones sought.


To unsubscribe from this list, send mail to lojban-list-request@lojban.org
with the subject unsubscribe, or go to http://www.lojban.org/lsg2/, or if
you're really stuck, send mail to secretary@lojban.org for help.