semantics ...

[C.D.Wright@solipsys.co.uk]

coi rodo

.i mi pu djica lenu mi cusku di'e rodo
.iku'i mi na kakne lenu mi cusku
          lei pluja sidbo bau la lojban.


la pablov. puzi cusku di'e:

> First, sorry about the misspelling of lujvo. I guess
> I confused the "jvo" with the "jbo" because of the
> rafsi {jbo} for the gismu {lojbo}. (bad excuse!)

Easy to do, easy mistake to make.  I still push many,
even most, of my lojban utterances through the parser
and then through my lujvo splitter and lookup program
to make sure that what I say is grammatically correct,
and that the words mean what I think they mean.


la robin. puzi cusku di'e:

> Lojban, {ninmu}  means  "a female humanoid person"
> English "woman"  means  "a female human adult"
> Greek  "gineka"  means  "a married female human"
> Turkish "kadIn"  means  "a female human non-virgin"
(snip)
> Leaving _everything_ up to everyday use would result
> in different meanings of {ninmu} similar to some of
> the translation errors we see between natlangs.

.ice'o la pablov. cusku di'e:

> I would say the problem is that there's never an
> exact definition for a lujvo,

{pe'i}  The stronger statement is that *all* words in
*every* language can never have *exact* definitions.  To
a greater or lesser extent, the semantics associated with
every word in every language will always be fuzzy.  The
only thing that we or anyone can ever hope for is that
the fuzziness is not so great as to cause significant
misunderstandings.

How do we express to someone else the meaning of a word?
We use more words, and the fuzziness can never disappear.


> It's never perfectly clear what "xxx type-of yyy"
> means. In fact, I believe the claim for semantic
> precision is illusory, since, well, there's always
> a level of "tolerance" for interpreting a certain
> word.

It's not clear to me that there has ever been a real
claim that there is semantic precision.  The explicit
claim is that a tanru is always a metaphor, and that a
lujvo is intended to have one meaning, not withstanding
the fuzziness that will always accompany any word.

It's likely to be the case that  { relxilma'e }  will
always explicitly mean bicycle, rather than the more
general concept of a two wheeled vehicle, which could
include a dicycle (which has the wheels side-by-side,
not inline (and yes, I have ridden one)).


> Every word is, strictly speaking, a metaphor.

Hmm.  I think I disagree with that, because to me, the
word  "metaphor"  means  "an application of a name or
descriptive term to which it is not literally applicable."
As such, any word that *is* literally applicable in some
situation cannot possibly be a metaphor.

However, what I think you mean to say is effectively what
I was saying earlier.  Every word has a greater or lesser
degree of fuzziness.

Back to lojban, however.

> if we want to stick to our claim of "unambiguity", to
> go on and create new words ad infinitum.

As I say above, I don't think the claim of a total lack
of ambiguity is sustainable, and I think that anyone who
thinks that such a claim has been made has misunderstood,
as indeed has anyone who makes such a claim.  The lack of
ambiguity has always been in the realm of parsing,
grammar, etc., and not in the field semantics.

In essence, I'm saying that you're perfectly correct in
almost every respect.  I only disagree with your belief
about what claims have actually been made, and the
consequences of them.


> the problem I perceive with lojban is that it gives
> fixed definitions for words (gismu), a thing which
> is just opposed to the essence of semantics, since
> words are understood by their use, and at the same
> time they are defined by their use, so that the real
> meaning is the complex sum of all its everyday uses.

Hmm. I do agree that lojban gives fixed definitions for
gismu, but I don't believe that doing so is opposed to
the essence of semantics. Words are understood by their
use, yes, but why is that a problem?  If the gismu in
lojban are always used according to their original and
intended definitions, then usage agrees with definition.
Linguistic semantic drift will be an interesting thing
to watch in the case of lojban.


> A dictionary only tries to approximate that, but it's
> always aimed to a person that have already learned his
> vocabulary in a "natural" way.

This isn't always true.  I use both a French dictionary
and a German dictionary, and I've learned neither sets of
vocabulary in a "natural" way.  Please note that I freely
admit that this example is probably not very helpful in
clarifying the situation, however.


> So, we have a clear contradiction here: lojban -

> defines the gismu from english words already in
> existence, which were and are used in this "natural"
> way;

> wants to "freeze" the meanings and create new words out
> of them (lujvo), which are supposed to have a priori
> definitions.

Let me say again that you're making valid and valuable
points here, and much of what you say I agree with.
However, I think the  "contradiction"  that you claim
here is rather less clear than you seem to be saying.
It seems that my understanding of these matters differs
significantly from yours, so let me explain what I think
is true, and we can see where the difference lies.  I'll
start with the basics that we all agree with, and develop
my points as I go.


Lojban has structure words, {cmavo}, and root predicate
words, {gismu}.  Cmavo have many purposes, from simple
bracketing of linguistic constructions, through to the
complex modifications of existing constructions to give
new meanings.

The root predicate words are each intended to occupy a
single point in semantic space. In other words, they are
intended to have one specific meaning.  However, it is
recognised that it is in the nature of language and its
usage that there will always be a greater or lesser
degree of "fuzziness" associated with each gismu.  This
fuzziness is inevitable, and is intended to be only as
large as necessary for communication.  Metaphorical use
of gismu is generally to be discouraged.

In order to warp or modify the meaning of a given gismu
so as to to match the desired meaning,  gismu (and more
generally {brivla} )  can be gathered together into a
{tanru}.  Each of the brivla in a tanru modifies the next
in a manner that depends upon the context.  The order of
modification is specifically defined by the grammar - the
words in a tanru always associate to the left.  B1 B2 B3
is always to be understood as  (B1 B2) B3.  This is one
specific example of where lojban is completely
unambiguous.


The meaning of a tanru is largely determined by context,
and a given tanru could have several potential meanings.
The place structure is taken from the last brivla in the
tanru, and the meaning is generally a restriction of the
meaning with that brivla alone.

Gismu can also be glued together to make a compound word,
called a {lujvo}.  In principle, a lujvo can have just as
many possible meanings as the equivalent tanru.  However,
a lujvo is intended to enter the language with one of
those meanings chosen as  "the"  meaning for that lujvo.

Given this description we can see that it is not always
possible  /a priori/  to know for certain the one true
meaning associated with a given lujvo.  However, it is
guaranteed that the meaning will be one of the possible
meanings of the corresponding tanru. It is expected that
it will usually be obvious, but perhaps not always.  It
is, however, expected that the meaning of a lujvo should
be generally guessable, and if a listener knows that a
guess has been made then no deep-seated misunderstandings
should happen.

What I do not know at the moment is how that one true
meaning for a given lujvo will be decided.  So far it
seems to have been the case that a lujvo is coined, and
then people have agreed that it's reasonable, and so it
has been adopted officially.  Er, whatever that means.


Finally to return to your posting about lujvo:

> we always get in trouble when we try to give a lujvo
> for a certain word, because the definition must be
> contained by the word itself.

This is largely true, although not entirely.  The
definition must be one of those possible from the
associated tanru, yes, and if that's what you mean
then fine.


> In this way, it will be never be possible to have
> complex concepts, since we would need endless lujvos
> to define them.

Now, this seems to be your final conclusion, and while it
has some merit  (which I'll come to in a moment)  I think
it goes just a little too far.  Complex concepts can be
embodied in tanru with their meanings pinned down by the
specific context.  This is what happens with natural
languages, and I think every usable language will have
this same property.  If I have an idea that you haven't
ever come across, I will have to shape that idea in your
mind using words you already know.  In English we don't
(often) create entirely words for ideas we have.

Well, not often.  It does happen on occasion, and I have
explicitly used the idea in both my previous research as
a mathematician and in my current work as principal
researcher for a commercial company specialising in
radar.  ( Side note: the word "radar" originally came an
acronym - "RAdio Detection And Ranging", but it is now a
word that has come into English in its own right.)

In my research I and my colleagues often discuss several
complex ideas over a period of a few hours, or even days.
To each of these complex ideas, constructions or concepts
we give nonsense names, invented words.  Concept builds
upon concept, and we need to be able to refer to each one
quickly and effectively, and so we deliberately build a
dictionary of nonsense words.

At the end of the session, those nonsense words whose
definitions we can still remember magically correspond to
the important concepts, because that's the way the brain
works.

We could do the same thing in lojban by inventing proper
names all the time - la glob., la tat., la kanet., and so
on, but in the end we may want a new word in the language
to cover the new concept, a concept not easily covered by
a tanru, and hence not easily covered by a lujvo.

Tricky.


mu'o

co'o mi'e. sidirait.
====================
" If you never go off at a tangent
  you will forever run in circles. "