Return-Path: <lojbab>
Message-Id: <m0jbVZs-0002HqC@snark.thyrsus.com>
Date: Fri, 10 May 91 07:24 EDT
From: lojbab (Bob LeChevalier)
To: lojban-list
Subject: le'avla history in response to jack Waugh
Status: RO
X-From-Space-Date: Fri May 10 07:25:07 1991
X-From-Space-Address: lojbab

In-Reply-To: from "Jack Waugh" at Apr 29, 91 5:52 pm
             and John Cowan's later response

>> Was any work done on chemical nomenclature before the
>> Lojban schism?  Perhaps reported in The Loglanist?
>
>As I noted, JL2 (actually UL2, "me la Uacintyn Loglytuan" issue 2),
>contained an earlier version of this list.  The split did not occur
>until JL4.

Here is a brief history of le'avla:

For those with 1975 dictionaries, you will find that every known element
was included twice, as a name, and as an S-prim (read le'avla that looks
like a gismu if you are newer to the project).

THough JCB disagrees, I believe it was a conversation between him and me
in 1980 that led to the "3rd lineage" of borrowings (translation - the
creation of a third form of brivla besides gismu and lujvo - the
le'avla).

His response proposal - the birth of le'avla, was reported in TL3/4.

{For historical record here was my argument:

In translating "Man of La Mancha", which as I've reported was my own
first attempt to learn Loglan, there was no word for "trumpet",
"gauntlet", etc.  Even if there had been, in the context of the song,
these words convey specific cultural values that are not inherent to the
musical instrument or the piece of medieval armor.  I tried to make a
lujvo for each, but we're obviously talking 6-7 terms - ugly!

I had also noted that many/most of the gismu proposals being made were
for plants/animals, etc. and that the limits on this set were
effectively infinite, but that the gismu space wasn't.

I thus proposed to JCB in this conversation a series of 4 steps to
borrowing words.  I still in effect stand by these steps, though we
haven't gotten past the third for any words yet, and shouldn't.  The
element words and the culture words are the most likely candidates to
get to the fourth step:

1. Most borrowings are little more than names, and indeed are used as
   sumti.  Thus to use a current example, la kromium. will do for most
   instances of the concept "Chromium".  In a rare instance where you need
   a predicate, you have "me la kromium."

2. When a borrowing needs to be used as a predicate more frequently, you
   want to coin a word, but don't want to go through the 6/8 languages
   effort.  So you just make up a word on the fly, and then OVERTLY mark it
   as a borrowing.  At the time, I was thinking only of standard gismu
   forms of 2 mod 3 letters in length (the pre-GMR standard).  The marker
   was to be an unassigned cmavo, probably from the then partially unused
   'hV' set.

   This proposal survived into the initial Lojban design.  I had a cmavo
   "le'a" which would mark the following word as a nonce borrowing.  This
   particular version lasted until a couple of months ago when John Cowan
   proposed the generalization to mark ANY nonce word usage using (instead)
   "za'e", lexeme BAhE.

   This current design says that you coin a word, which must be a legal
   brivla and not break down into multiple words.  Your marking this word
   by preceding it with "za'e" means that you have just now coined the
   word, it may or may not conflict with another "official" meaning of the
   wordform.

   While it isn't "approved" (indeed no word marked by "za'e" is), you
   >could< make le'avla in the form of gismu or lujvo.  The permitted
   word-forms for le'avla are defined primarily by exclusion (it can't
   break down into two words, it can't be a lujvo, or a gismu, it can't
   fail something called the "slinkui" test), and coining nonce words is
   difficult, so this freedom is worth something for sponteneity in the use
   of Lojban by non-fluent speakers.  I do not recommend intentionally
   invading lujvo space with le'avla because even in a nonce lujvo, the
   listener will presumably try to take the word apart into component
   rafsi.  But lets face it; the people making nonce le'avla will be
   less than expert, and "za'e" allows some slack.

   It turns out that our design made "za'e" form le'avla a bit useless
   anyway.  "za'e" will now be used more with lujvo than with borrowings
   because it turns out that it is virtually as easy to make step 3 lea'vla
   as step 2. nonce forms. as described next.  Because step 3 forms are
   limited to specific fields, and the method for making them is so simple
   it is not necessary to mark these with "za'e" (though it is permitted
   and may be recommended if you are a using a word in a field in which you
   are non-expert - this is like putting quote marks around the word to
   show that you may be being non-standard in the technical terminology.)

3. Most of the le'avla you see nowadays are step 3 le'avla.  These are
   names for concepts in specific semantic fields, lojbanized into
   brivla-form by a most trivial process, and then marked with a classifier
   rafsi.  Originally I proposed that classifier rafsi go on the end,
   making things look like lujvo:  kromium-xuki (chromium-chemical) gives
   the general idea, though it may not be correct - I've forgotten the
   specifics.  This is the proposal that was printed in UL2, the early
   version of JL from before the Institute/LLG split because Lognet had
   just folded and TL was dead.  The reaction to UL2 was that people did
   not like the ending rafsi partly because there rafsi were CVCV form
   which meant a second type of rafsi had to be memorized.  I then came up
   with the current design, which is described lightly in the Synopsis.
   John Cowan has codified (and I think posted his codification) of the
   current process for step 3 le'avla, and the chemical words he recently
   made are (presumably - I haven't checked) valid by that process.

   Simply that process is to use a standard rafsi, or even more than one
   like a lujvo with some restrictions, as a classifier on the front of a
   lojbanized form of the word to be borrowed, which must have a final
   vowel, no letter 'y', and only permissible medial consonant pairs.  The
   classifier is 'glued on' with a vocalic consonant 'l', 'm', 'n', or 'r',
   which also incidentally makes the first consonant cluster in the word
   NOT a permissible initial, avoiding many of the problems in le'avla
   coining (such as the aforementioned "slinku'i" test - "slinku'i" is an
   invalid le'avla because if you use it with "pa":  "pa slinku'i", the
   sound stream is ambiguous and the listener hears paslinku'i, a valid
   lujvo - since le'avla are defined as NOT causing conflict with gismu or
   lujvo, this form of le'avla is invalid).

   The virtue of step 3, is that almost anyone can make nonce le'avla with
   minimal learning, the resulting words are flagged so that a listener
   knows he/she is hearing a le'avla, that it is somewhat a nonce word, and
   that it is restricted to a specific jargon field.  This is often all
   that is needed - since someone familiar with the jargon field will
   recognize the borrowed portion, and someone who doesn't can ask.

   Step 3 and step 4 le'avla can be used in lujvo.  However they are always
   joined to adjacent rafsi by the hyphen syllable "iy", and NO letters are
   deleted - the "rafsi" form of a le'avla is the le'avla itself.  JCB did
   not have this restriction, and the 4th edition of Loglan 1 (1989) went
   to press with rules that permitted the sample le'avla compound
   "protonynukli" for "protoni-nukli" - but this breaks into rafsi as
   "pro-ton-nukli" an is thus invalid - so standard hyphenation can't work.
   Nora found this counter almost instantly, and we included it in my
   review of Loglan 1 (I think in JL10).  JCB, without acknowledging who
   found the error, has back-pedaled and adopted a strategy similar to
   ours, after two or three issues of discussion in the reborn Lognet in
   the last year.

4. There are as yet no step 4. le'avla.

   These words are made by people skilled in Lojban word-making, familiar
   with the previous body of such words to prevent conflicts.

   The words need no classifier rafsi, and may utilize any of the valid
   le'avla wordform space.  As stated above, there is no simple algorithm
   for this space, and making these words correctly is a
   trial-and-error-aided-by-growing-experience process.  More on this in
   the history below.

   Institute Loglan ONLY uses these, never having accepted my 4 step
   proposal before the split, since there was no meaningful process at the
   time to approve such major proposals.  Examples include "protoni", as
   mentioned above.  (Institute Loglan permits le'avla in gismu space,
   allowing "nukli" as well.  We currently don't, on aesthetic principles
   only since some gismu like the culture words are really such lea'vla -
   the key advantage in having gismu length is that the gismu-forms have
   shorter rafsi, and may be more easily and briefly made into lujvo.  We
   don't have useful standards yet for deciding that a word deserves this
   privelege enough to invade gismu space, other than the class decisions
   that were made for culture words and "cmavo" and "lujvo" which are
   themselves borrowings from malglico Lojban tanru - "cmalu-valsi" and
   "pluja-valsi" for JCB's English "little words" or "LWs" and "complexes"
   or "Cpxs".

   The qualification for a step 4 le'avla must be that it is a word used
   sufficiently often, probably outside of a single field of endeavor, that
   it violates Zipf's law to have such a frequent word be as long as step
   3. le'avla must inherently be.  Rather than have Lojban suffer some sort of
   irregular shortening - the historical linguistic response to long words that
   become frequently used - like "teevee" for "television".

   With no usage history yet, we've never bothered to make standards for
   step 4 le'avla.  The current discussion of culture words suggests that
   any culture whose name acquires any signicant use in Lojban will get a
   step 4 le'avla, giving more equality with the historical culture gismu.
   This answers most, if not all of the criticisms of the cultural gismu.

End of interlude}

At the time of GMR, JCB moved MOST of the S-prims into borrowing space.
The exclusion 'algorithm' for le'avla was discussed in TL6/1 (1983).
JCB then launched what he called the "Sciwords" project, to massively
borrow words from many fields into the language.  If there were any
volunteers at the time, their work was never reported because TL folded
followed by Lognet a year later after the 1983-4 political squabbles.

JCB continued to work on the borrowings, and translated some paragraphs
of Scientific American (reprinted in 4th edition Loglan 1) that were
heavy in scientific jargon to be borrowed.  He also reported making
borrowings for 50 kinds of cheese one night after reading an article on
the subject.  There have been some reports in recent Lognets that others
have made some le'avla and that the Sciwords project finally
accomplished something but there has been no list because of the
Institute's trade secret policy.  They don't want LLG to borrow their
borrowings ;-)

JCB also introduced borrowing-and-name-only lerfu for W, Q, X, and Y to
make visually recognizable borrowings easier.

When Rebecca Bach and I visited JCB is May 86, we discussed borrowings,
and specifically JCB's then current effort on remaking the element words
into le'avla as a test for his attempts to devise "fast-tracks to
borrowing" that would evade the awful le'avla test.  They didn't.  It
turned out that we went through all of the elements and remade them, but
found that there were few simple guidelines (we did notice that CVCV
endings more frequently give good le'avla, as well as that it is easier
to avoid "slinku'i" problems by making the initial consonant cluster not
a permissible initial.  Rebecca, a Loglan novice, proved better than
either JCB or me at detecting flaws in le'aa'vla-making, but none of us
were really good at it.

JCB at this time made clear that a standard for scientific lea'vla
making, unlike gismu, was visual recognition rather than aural
recognition, since technical words are used in written language more
than spoken language.  It was this that justified the additional lerfu.
JCB also felt that the beginning of the element words should reflect the
international symbol - the closest thing to an international 'word' for
the elements to be borrowed from.  Unlike John Cowan, I still subscribe
to this philosophy, since as has been noted for German (and also for
Chinese), national languages have sometimes made non-international forms
of the elements.  The "Latinate forms" are really the English/French
forms since those two languages have dominated the scientific
publication field during the time of internationalization of science.
We can't get around this entirely, but if a truly international standard
exists, IMHO we should use it.

I went home, and reworked the element words, which were left hanging.
The UL2 publication was 4 months later.  Other than a discussion in
JCB's Notebook 3, and Loglan 1 4th edition, and ensuing responses to my
criticism of the latter, there has been minimal discussion of le'avla
until the present - although as I've noted, the culture words have been
questioned by many new Lojbanists (who have generally been satisfied
with my answers - again, until now).

-lojbab