Message-Id: <199511171119.GAA01901@locke.ccil.org>
Date:         Fri, 17 Nov 1995 06:10:20 -0500
Reply-To: Logical Language Group <lojbab@ACCESS.DIGEX.NET>
Sender: Lojban list <LOJBAN@CUVMB.BITNET>
From: Logical Language Group <lojbab@ACCESS.DIGEX.NET>
Subject:      Re: slinkui
To: John Cowan <cowan@LOCKE.CCIL.ORG>
Status: OR

Here was what I saved for the draft dictionary on le'avla/fu'ivla types.
I posted in in early 1994, so the longer discussion referred to is
probably in the 1993, or perhaps 1992 List archives.

>I won't repeat the long summary I posted last year.  The essential
>concept is that most le'avla represent things that are tatamount to
>names in that they tend to have fairly specific (non-abstract) meanings
>without much semantic loading (connotations other than possibly
>expressible through 'mabla' or 'zabna')
>
>Thus lea'vla are seen a s a hierarchy of successively Lojbanized names
>
>type 1 = non Lojban quoted with zoi/la'o delimited quotes
>
>type 2 = Lojbanized names normally marked with "la" or "mela" for
>predicates
>
>type 3 = most things called le'avla these days - being a Lojbanized name
>but ending with a vowel and meeting some other minor restrictions, but
>always marked with a classifier on the front that indicates the general
>field of reference of the word - the classifier is generally a CVC ro a
>4-letter rafsi for a gismu, glued on with a syllabic 'r' or 'n'.  It has
>the advantage of always giving a valid le'avla even if you know few
>rules of Lojban morphology.  The disadvantage is that all le'avla are
>polysyllabic and bear little resemblence to their original form, since
>the beginning of the word never has anything to do with that original
>form.
>
>type 4 - a fully Lojbanized word that fits in "le'avla" word-space, the
>set of valid brivla that are neither gismu nor lujvo. the rules for
>making these are sufficiently arcane that no one can do so infallibly in
>their head on the fly - there are too many rules that have non-obvious
>effects and can cause you to make a word that will break up into poeces
>in actual speech contexts.  Thus such word proposals will tend to be
>used for le'avl athat see actual heavy usage, get proposed to be
>assigned a type IV, and then people would use some computer verification
>to make sure that the word isn't flawed.  There is some debate whether
>type IV words could include gismu-space - the curr current policy is
>'no'.  There are no currennt type IV le'avla in the language.  Note that
>in many cases, especially the cultural words, type 3 le'avla are the
>natural final stage in that the classifier rafsi on the front can act
>like a first order lujvo, thus allowing us to talk of "bangr,taliano",
>and "gugdr,taliano".  A possible, unverified type IV for Italian might
>be "talnano" (it should be good - this is one wordform that has been
>consistently valid CVCCVCV with the CC not a permissible initial) for
>which le'avla lujvo comparable to the two tpye 3 words would be "talnano
>zei bangu" and "talnano zei gugde" which are most certainly NOT
>Zipfeanly shorter. le'avla lujvo are the only lujvo that are longer than
>their corresponding tanru because the morphology doesn;t work otherwise.
>
>===
>djarspageti is the classic example of a type 3 le'avla per the
>definitions I gave the other day.It has a classifier "dja" meaning food
>glued onto a Lojbanized root "spageti" with a vocalic consonant so that
>the result gives something that cannot be a lujvo even if preceeded by a
>cmavo which in speech stream might be taken as part of the word (i.e.
>ledjarspageti and leidjarspageti are not lujvo, primarily because the
>final CVCV construct never occurs except in a borrowing.  The
>now-preferred way to make type 3 le'avla, which requires no particular
>thought about the form of the borrowed root other than it start with a
>consonant and not end with one is to use the full 4-letter rafsi, as in
>cidjrspageti.  Vowel initial roots and CCV rafsi as classifiers take a
>little more care in making type 3 le'avla, so we do not recommend it for
>the typical nonce le'avla.  We also do not recommend using a multiple
>rafsi/lujvo as the basis for a classifier, since this can lead to
>problems too.)
>
>An example of a gismu-space le'avla for the above, might be "spagi", or
>the related longstanding proposal "pitsa".  These types of le'avla chew
>up gismu space.
>
>I disagree with Cowan that the average Lojbanist would not typically
>make his 'type V lea'vla', i.e. le'avla in gismu space, because the
>history of the Loglan project shows that people do precisely that.  No
>matter how abstruse the concept, when people Loglanized borrowings under
>the old morphology (which had all words in gismu 5-letter space or lujvo
>8,11,14,2 mod 3 space) they chose 5 letter forms.  A syrvey of 'The
>Loglanist' from 1977-84 reveals dozens of coined 'gismu' that hsould
>have been lujvo or borrowings and were not understandable without a
>summary wordlist.  My observation of the practice of TLI in the last few
>years is that they still tend to make borrowings in gismu space when
>they don't need to,and the existence of so many of them as examples only
>encourages the average TLI Loglanist to emulate the practice.
>
>People will do whatever we allow them to.  If making anything shorter
>than type 3 le'avla is easy, people will probably do so regularly, and
>they won't be prone to check for gismu concepts or malformed words as
>Cowan does.
>
>la lojbab. cusku di'e
>> I disagree with Cowan that the average Lojbanist would not typically make
>> his 'type V lea'vla', i.e. le'avla in gismu space, because the history of th
>> Loglan project shows that people do precisely that.
>
>"My theory falls apart in the face of his facts."  I concede this point.

lojbab