Re: [lojban] Re: semantic primes

[John E Clifford <clifford-j@sbcglobal.net>]

--- Jorge Llambías <jjllambias@gmail.com> wrote:

> On 3/22/06, John E Clifford
> <clifford-j@sbcglobal.net> wrote:
> > No one claimed they were built up from others
> in
> > any sense but that it is possible to define
> all
> > concepts starting from just a few (relatively
> > speaking).  The alternative is to say that a
> > language (it would only take one to make the
> > point) cannot completely define its
> vocabulary.
> 
> Consider these two theses:
> 
> (A) Most concepts can be very well defined in
> terms of
> other concepts.

The objections below work pretty well for this as
well.  And the issue is whether something can
really be said to be defined if it is ultimately
in its definiens, as would often be the case
here.

> (B) Every concept (except a selected few) can
> be perfectly
> defined in terms of other concepts.
> 
> I don't think anyone would have much to argue
> against (A), it is
> pretty much an observable truth. 

What, even though false?

(B) is a much
> harder nut to
> swallow.

True, but that may just mean that definitions are
ultimately impossible.  Or, on the other hand,
that the list of primes is much larger than we
had hoped.
 
> And that's just about concepts. When it comes
> to words, things get
> much more muddied. Words generally point to a
> conceptual area more
> than to a strictly delimited concept, and the
> concept they bring up
> in a given use varies depending on other words
> used in their context.
> So defining a word is much more tricky than
> defining a concept.

Well, words are, as we say, polysemous, so
eachmay have several definitions corresponding to
slightly different concepts and f0or use n
slightly different contexts (how much context
interacts with a given definition to produce a
different meaning wihtout a different definition
is a matter that needs more investigation). But
this is only a practical, not a theoretical,
problem.

> 
> > The most one could get would be overlapping
> > partial definition sets, with the bottom
> level of
> > one set being at a higher level in some
> other(s).
> >  As a practical matter, given a finite span
> of
> > concern, this is sufficient pehaps, if we
> don't
> > get caught in a circle.  For a theory,
> however,
> > it is a disaster, since it means that a
> language
> > can only be completely defined in another
> > language and so on to an infinite regress.
> 
> And the thesis that a language can be
> "completely defined" must
> be taken as self-evident?

No, it is a hypothesis being tested.  the test,
of course, assumes that it is true and works on
from there.  Should the tests ultimately fail,
then the hypothesis would have to be abandoned. 
However, that part of the hypothesis has a good
deal of prior probability, given the arguments
above and our actual experience.
 
> > As
> > usual, it seems best to stop at the first
> step if
> > possible.  NSM holds that it is possible for
> each
> > language and furthermore that the initial
> step in
> > every language is the same (directly
> > intertranslatable).
> 
> Yes, it's an attractive thesis, but not a very
> convincing one from
> the evidence at hand.  And I don't see the
> "logical argument"
> for it yet.

What part of the argument do you not understand?
To run therough it again, trying not to skip over
even the most obvious steps: A definition of a
concept is giving an equivalent in terms of
other, more basic concepts.  Concept A is more
basic than concept B if, at a minimum, B does not
occur, however remotely, in the definiens to A. 
Now then, suppose every concept can be completely
defined.  A is defined in terms of B and C, each
more basic than A;  B, say, is defined in terms
of D and E, each more baisc than B or A, and so
on.  But then no concept is evr completely
defined, since, at each step, the definition
involves concpets which are, ex hypothesi, to be
defined in terms of yet further concpts.  That
is, there is always a yet-to-be-defined
component.  To be sure, as a practical matter, we
can stop off with a partial definition which may
be an adequate guide to usage and meaning because
we are (we think, any how) familiar enough with
the concepts then in play, just as even a
circular "definition" can guide our usage -- even
our understanding.  But the hypothesis that every
concept can be completely defined is false.  The
alternative is then that some concepts cannot be
completely defined.  Since they are functioning
concepts, we (well, somebody) understand(s) them.
 But, more strongly, it turns out that every
concept eventually leads to some undefinable
concepts (else some concepts would have infinte
definitions, which, as noted above, are not
definitions at all -- i.e., these are
indefinable). Thus, pooling these resources, we
get a set of undefinable concepts in terms of
which all the others are defined.  NSM's steps
beyond this are to claim that this set is small
(around 100, say), that it is the same in every
language, and that this [a given list] is it. 
All these later steps are open to challenge as is
the notion of a complete definition, but the
heart of the argument remsins.

> > > The definitions I've read for other
> concepts
> > > don't look much like
> > > fixed expressions.
> >
> > I don't follow this, I think.  Do you mean
> that
> > in other languages, the words that correspond
> to
> > the  concepts are not fixed expressions?
> 
> No, I haven't seen the proposed sets for any
> language other than
> English. I'd be interested to see the
> corresponding set proposed for
> Spanish for example, but I wasn't able to find
> it.

I can't give you a reference although I think
Spanish is one of the languages that has been
worked with fairly extensively.  If you can a
copy of the 2 vol. book from about 2002, there is
probably some Spanish material in it (I haven't
been able to find it in any local library and I
am not about to order it at megabucks, even from
Megabooks).  I've seen pieces of the list for a
number of languages, but none I know, so I can't
check them.  However, the attacks on NSM rarely
take the form of saying that the word they pick
for such and such a concept doesn't mean that
concept.

> > If you
> > can make that case for any language you have
> shot
> > down the NSM project in its present form,
> since
> > that fixed form is a requirement to be a
> prime.
> > If you mean that NSM definition (strictly
> > "reductive paraphrases") don't seem to be in
> > fixed form, I'd like to see a case.
> 
> Well, I've only seen two or three NSM
> definitions given as examples. Is
> there a comprehensive list of definitions
> somewhere?

I wish there were.  There are a few definitions
for English that get repeated over and over and
about a dozen more that I have managed to get
from here and there, plus maybe another dozen --
mainly about emotions, curiously -- for other
languages but in English.  I am sure there are a
lot more, but plowing therough the literature
(indeed, finding the literature) is too
time-consuming for  now and my present purpose.
  
> For example:
> 
> NSM definition of loves [2]
> Person-X loves Person-Y =
> X often thinks about Y
> X thinks good things about Y
> X wants to do good things for Y
> X wants good things to happen to Y
> when X thinks about Y, X often wants to be with
> Y
> when X thinks about Y, X often feels something
> good
> 
> (From this I now gather that the prime THINK is
> {pensi} and not
> {jinvi}. 

Actually it seems to be somewhere in between: "x
thinks y about z," which fits closer to {jinvi}
and has subject-raising.  This is actually
currently a matter of controversy, since
apparently there are languages which allow "x
thinks y" and "x thinks about z" but not both
claims in a single form.  Lojban may have the
drop on NSM here (if we get rid of subject
raising).  

>I also notice a few non-primes there,
> but I guess they
> have already been pre-defined.) 

That is either an item of faith or else the move
is obvious

>Now, given
> that, what would be
> the problem of defining:
> 
> X is bad =
> X is the opposite of good
> 
> How is that less of a fixed expression than the
> expressions used
> for "loves"?

Well, if you can work out a case for English, I
suppose the reason for rejecting it would be in
some Austronesian language.  But notice that your
definition is not one of the canonical form and
the NSMers insist that the sentential forms
allowed are as much a part of the system as the
concepts.  What would be the paradigm sentence
for for OPPOSITE (or THE OPPOSITE OF)?