repost of my response to Nick of last month

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

Nick asked me to repost this, saying it was badly garbled.  Let's hope
it comes across better this time

lojbab

Nick wrote last month:

>From: nsn@mullian.ee.mu.OZ.AU (Nick Nicholas)
>Subject: Re: Some how do you say it's
>Date: Sun, 15 Aug 1993 21:29:36 +1000 (EST)
>To Logical Language Group respond I thus:
>#Not to say that I favor or disafavor it as a solution, why does kabryselmre
>#require a measuring agent place.  If it isn't interesting leave it out - it
>#is a lujvo, after all.
>
>I learn more about lujvo --- and my attitude towards them --- every day.

This was my argument that we are still probably premature in being
prescriptive about lujvo places.  If YOU, who have analyzed the problem
the most, are having your attitudes change day by day with relatively
little competing analyses from the rest of us laggards, what will happen
when we get a group of people all of whom know and use the language as
well as you do, but who might have different analyses, on the order of
the current ZAhO discussion.  At that point, we will either be stuck
with a prematurely set in concrete status quo, or some people like you
who have sunk a lot of work into a particular analysis are going to have
to deal with competing or changing ideas that will render much of the
work obsolete.

Thus my argument is that people who make lujvo should propose place
structures for them based on what the concept they are trying for is,
and we should be doing more reconciliation of your theoretical framework
with what people actually do...

>I have indeed favoured the throw-the-place out attitude, but I'm rather
>more hesitant to apply it to tertanru in the veljvo.  The reasoning is
>that a kabyselmre is a selmre is a se merli, and a se merli must have a
>merli.  The attitude I've taken is that, if you have to delete a
>tertanru place, and it's not because it overlaps with a seltanru place
>(Lean Lujvo), then I'm using the wrong tertanru.

..., which might of course result in suggesting a different lujvo.
...

>Thus in brulu'i, I have le lumci cu brulu'i le se lumci --- I leave out
>the te lumci, because we know it to be a broom, and if we need to name
>it explicitly, we can just say le lumci cu lumci le se lumci le burcu.
>
>Come to think of it though, this distinction between omitting tertanru
>places when they're overlapped by the seltanru, and leaving them in when
>they aren't, is a bit artificial, isn't it.  Hm.  I think Lojbab may be
>right after all.  Any ideas?

...  But instead, what it might reveal is what places are important to a
concept.  And this is the essential point that I see being lost.  In my
comment about kabryselmre, I was arguing that the true concept covered
by karyselmre probably entails that there be a lo merli.  BUT, we have
the option, in making a lujvo (unlike in making a tanru), of deciding
that the essence of the CONCEPT WE WANT TO CLAIM, will not require such
explicit places.  Thus we are doing a defacto "zi'o" deletion through
ellipsis - the place may or may not have a value, but is not important
to the claim, so we ellipsize it.

Of course we also ellipsize places that we DO feel are important to the
concept in actual usage.  Thus the problem is in distinguishing what
kind of ellipsis is taking place.  This is something I've always felt is
best done by the person who invents the word, but when that person
doesn't do so, or when we add words to the list in a hurry like we do
without necessarily checking to see what place structure the person used
in the text (or proposed in commentary), we are somewhat forced into
doing something like you have been doing if we want to have any sense.

>This comes up all the time, btw:  the regularity in place structures
>gives you a slightly different concept to what you originally intended,
>but usually I wear it.  Thus for an agentive 'change', I want x1 changes
>x2 into x3; but my algorithmic gafygau gives me x1 does x2, which
>changes x3 into x4.  I'm prepared to wear that.  Of course, the places
>need reordering, but that can be done with 'jai':  jai galfi has the
>place structure:  agent x1 changes x2 into x3 by doing x-fai.
>
>Hm.  Thanks for pulling me up on this one, Lojbab.  I've got some more
>thinking to do...

What I would like to see is to use the systematic procedures to come up
with the theoretical m,aximum set of places, and then use pragmatics -
why someone would create a lujvo for this concept rather than leaving it
a tanru, for example, or what the essentials are in related concepts in
other languages - use these to decide which places to "zi'o" out, or
Lean-Lujvo out if you prefer, as being places that are so unimportant or
obvious to most or all pragmatic applications of the concept that it
makes no sense to require them.  Deleting them DOES have truth-functional
and semantic import - the leaner concept is inherently more broad since it
presumably could include related concepts where the omitted place has an
unexpected or important value, as well as the basic concept for which
the omitted place is unimportant and expected.

I am talking about pragmatic deletion of places, and not place strutures
wherein the place being omitted is being so omitted because THERE IS NO
POSSIBLE VALUE which makes the statement true.  I might accept some of
these as lujvo, but think that is a rare occuramce, and proibably the
omission of the place is sufficently important that we would WANT to
have it made explicit in the lujvo that we are doing something
nonstandard woth the lujvo.

lojbab