Re: your mail

[cowan (John Cowan)]

la stiv. rais. cusku di'e:
> In borrowing and deriving new words, Lojban doesn't seem to differentiate
> between the sounds of "a" in sAd, sAcred, and bAr, even though these represent
> quite distinct phonemes in English and (I think) Lojban.  Thus, the "a" in
> "badri" is equated with that of "sad"--or else the words have only one phoneme
> in common, which seems too tenuous for a mnemonic, let alone a derivation.
> this also occurs in "banli" (mnemonic "grand").  The sound in bAr appears to be
> the norm, so I won't give any examples for it.

This is correct.  In making the gismu, English [ae] was equated to Lojban
"a" rather than "e".  Either choice is defensible; even if "e" were
unambiguously inferior, the assumption is at least systematic and therefore
can be allowed for.  Similar choices were made in Lojbanizing words from the
other five source languages, some of which may not have been the best possible.

> "Censa" has the mnemonic
> "sacred"; again, unless the "a's" are the same, there is only one phoneme in
> common (unless the order of phonemes is considered unimportant).

"censa" has a zero score from English; its main components (in Lojbanized
orthography) are Chinese "cen" and Hindi "sant".  "Sacred" Lojbanizes as
"seikrd"; see below.

> Another case is the Lojban version of "Sally" ("salis").  The Loglan version is
> "selis", because /e/ is closer to the digraph [ae] than /a/ is.  Why does
> Lojban use /a/?

Either may be used, depending on the preferences of the name-bearer
(se cmene).  Lojbanist Nancy Thalblum, e.g., decided to be "la nansid."
The only final authority for the appropriateness of a name is the name-
giver (te cmene), which may be the same as se cmene.

> Likewise, I don't see more than one shared phoneme between "balvi" and
> "later"--unless "later" is transcribed /leitr/, for example, which is false to
> the English phonemic system.  While the glide is definitely there, it is
> generally unnoticed by the average English-speaker, so it wouldn't help one
> recognize or remember the word.  This leaves "balvi" with a single-phoneme
> match for mnemonic or derivational purposes, which would be unacceptable in
> Loglan.

"balvi"'s recognition score over all 6 languages is only 36 anyway:
it is most heavily influenced by Chinese "janlai" and Hindi "bavik".
Lojban's speaker counts are from 1980 rather than 1950, so Chinese has far
more weight than in Institute Loglan.  There are a number of Lojban words
with zero English scores as a result.

"leitr" is indeed the Lojbanization of "later".  When making the gismu,
we felt that if a natural-language sound had a Lojban equivalent, that
equivalent should be used in the algorithm.  To Lojbanize English
"try" as "tra" rather than "trai" seemed like too much of a distortion,
even if the vagaries of the gismu-forming algorithm ended up splitting the
"a" from the "i".

> I can only conclude that Lojban's system for deriving words is far
> more lenient than Loglan's, for about a fourth of the vocabulary (I've been too
> busy with Loglan to perform an exact count) seems to have the same
> peculiarity..
> 
> On a related subject, I noticed in a recent posting on sci.lang that (according
> to lojbab) Lojban has a phonetic writing system.  As a linguist, I've used
> phonetic orthographies, and they're extremely awkward for everyday pruposes.  I
> would suggest that you switch to a phonemic alphabet, such as Loglan has.

Lojban's alphabet is, of course, phonemic; in fact, it is the same alphabet,
except for the absence of "h" (replaced by "x"), "q", and "w" and the addition
of "'", a voiceless vocalic glide used only intervocalically.

> Another distinction is the method of borrowing national names and terms.
> Loglan uses the local word with as little change as possible. This is
> considered a courteous way to borrow.  Lojban seems to rely on international
> forms and particularly on the written form of words.  If I had to come up with
> a primitive predicate for "Argentinian culture", I would try ?hento rather than
> ?gento, because the word in the prevalent Argentinian language (Spanish) is
> pronounced "arxentino" (the "x" may be softened to an "h").  In fact, I would
> use a direct borrowing:  ?arhento (the "-in-" is an adjective suffix and would
> be dropped).  A similar case is "xebro" for "Hebrew"; the Hebrew equivalent is
> "`ivri".  Dropping the initial gutteral (which exists in neither Loglan nor
> Lojban, I think) would give the Loglan form ?ivrio. Does Lojban use
> international forms for such words?  [I've just seen a posting by Mark
> Shoulson, who evidently has noticed some of these points too.]

"xento" and "xivro" would doubtless have been better.  Overall there is
about a 4% error rate among the gismu as a result of transcription or
interpretation problems, not counting systematic difficulties (e.g.
Russian roots ending in voiced consonants were devoiced, whereas better
recognition would have been achieved by ignoring this purely morphophonemic
issue).  In our opinion, these problems are not worth fixing.  The investment
that people have already made in learning the words despite the flaws
is too large to discard.  In the history of the Loglan Project, this investment
has historically been discounted, forcing people to relearn and relearn
until language stability seems a chimera.  LLG therefore does not change
anything that people have learned unless it is demonstrably damaging to
the language's goals.

> Again, "jegvo" refers to "Jehovah".  The Name is now thought to have been
> pronounced either "iaue" or "iave".  I can see the problem in creating a
> primitive predicate for this word, because 1. "i" is not an actual consonant
> and 2. there is no consonant pair.  But if I for some reason wanted to produce
> such a form for this name, I would use the original vowels.  [For the curious:
> the Name was written with the vowels from another word ('adonai "Lord") as a
> sign of respect; when reading aloud, one would say 'adonai rather than Yahweh.
> When the Name was transliterated into Latin characters, it brought with it the
> vowels from 'adonai.  (The "a" changed to "e" because it was really a schwa,
> that is, an "y" sound, but was pronounced "a" following the gutteral aleph
> (').)]
> 
> What does Lojban do in these cases, and why?

"jegvo" is the best available answer for an inherently difficult word.
See my other postings for a discussion of what it means.

> Also, I've been puzzled about the derivation of the word "baxso"
> ("Malay-Indonesian...").  Could someone enlighten me?  (For that matter, why
> don't you just create a complex from "bindo" and "meljo"?)

"baxso" is derived from "bahasa", which just means "language" in Bahasa
Malay / Bahasa Indonesia.  This language is shared between two countries
as official, and as such is a numerically important language.  To base
the word for the language on either "bindo" or "meljo" would be an
unwarranted bias: by the same token, the word "xrabo" is used in
preference to a complex which combines all the Arabic-speaking nations.


-- 
cowan@snark.thyrsus.com		...!uunet!cbmvax!snark!cowan
		e'osai ko sarji la lojban