[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [lojban] Re: coi rodo - mi'e .aulun.



At 09:16 AM 05/28/2000 +0000, Alfred W. Tüting wrote:
--- In lojban@egroups.com, "Bob LeChevalier (lojbab)" <lojbab@l...>
wrote:
> If we had it to do over again, we would map "ong" to "on(g?)" and not to
> "yn".

IMHO this mapping to lb: "yn" most probably is based on a
misunderstanding, maybe going back to pronunciation standards in
southern Chinese dialects like Cantonese: Mandarin (pinyin) -eng (lb:
-yn(g)) very often is pron. -ong (=ung) or -un in
southern dialects (e.g. BIG5 Â× pinyin: feng1, "fruitful/abundant"
in dialect is *written* and pronounced "fung" - pron. in
'virtual' pinyin about: fong -, or BIG5 ­· pinyin: feng1, "wind"
etc. in dialect written and pron. "fun" as in Cantonese taifun, BIG5
¤j­·, pinyin: dafeng!)
Pinyin "-ong" should be transcribed to lb: "un(g)" *not* -yn (e.g.
mau.dzydun.)

I now have the source in front of me:
_The Pinyin Chinese-English Dictionary_
Editor in Chief Professor Wu Jingrong, Beijing Foreign Languages Institute
Published by Commercial Press (Hong Kong and Beijing) and John Wiley and Sons
ISBN 0-471-86796-9

You are correct and I recalled incorrectly - ong is listed as [un(g)]

The short form of what I can eventually put up is that /e/ generally mapped to schwa which by the pattern of other languages we mapped to Lojban a - we should have instead mapped it back to Lojban e for Chinese. /a/ mapped to Lojban a; /i/ mapped to Lojban i (but should have mapped differently after z/c/s/zh/ch/sh/r) /io/ and /u:/ mapped as Lojban u. I have several special cases including /iu/ and /iou/ mapping to Lojban iyu, /iong/ to Lojban un(g) /uo/ to Lojban uy (which may be why I remember the schwa for "o").

The other problem, one specific to the gismu making process, is the mapping of both /n(g)/ and /n/ to Lojban /n/ in addition to the mapping of all the fricatives and affricate consonants to the only 4 Lojban counterparts. Thus, if I recall correctly, j and zh both mapped to Lojban j, because d alone is useless as a clue, as did r
q and ch both mapped to Lojban c, because t alone is useless as a clue
z mapped to z
c mapped to s
sh mapped to c

If this seems completely illogical (or backwards or something) you are probably right, since I am reconstructing this on sight of the IPA chart after being up all night. The actual table is packed away rather deeply at the moment and likely will not be dug out before LogFest in August.

> The g is questionable because Lojban maps the /ng/ consonant to
> /n/.  As someone noted, if the g is present it is pronounced
separately
> from the n.  But the real problem is that in gismu making we could
have
> ended up with the g and not the n in the Lojban word, and the g by
itself
> without the n is probably useless to a Chinese speaker for
recognition.

In Mandarin there can be no g-ending except in -ng, so this would not
necessarily lead to much misunderstanding to a Chinese
speaker familiar with lojbanization.

That is true, but in gismu the g is always medial, and it could have crept in from one of the other 5 languages - the point is that we had it on authority from both a linguist and a Chinese native speaker that they would not see any sound relation between g and (ng), and the point of the gismu making algorithm was to provide sound patterns that could serve as hints. We used the rules we inherited from JCB, the inventor of the language, and those rules by and large treated all languages equally. This was not wise for Chinese for one reason (the bad sound mappings) and Arabic for a different reason (in Arabic, the vowels have little sound/meaning significance, while the consonants and their order are vital). Russian suffered from its tendency towards long words, even after we dropped declension endings. As a result, Hindi, English, and Spanish are somewhat more effectively represented (unfortunately our Hindi scholarship was probably the weakest of our 6 languages though).

 BTW, in southern dialects, there
are lots of consonant endings, e.g. in Hakka language BIG5 «È
®a (=Hakka or "guest families"), pinyin: ke4 jia1, it is "hag-ga".

I had a Cantonese dictionary, but we decided to stick firmly to Mandarin as Pinyinized, since that was nominally official.

One principle question:
For what reason cmene have to follow the strict rules of Lojban
phonology??

If they do not, then presumably they will be heard by an ideal Lojban speaker differently than they are spoken. In other words, since we do not allow mixing of voiced and unvoiced consonants in clusters other than lmnr, a cluster like tj would be heard either as tc or dj, probably depending on the norms of the speaker's native language.

If it just were for computers to understand the
language, isn't it sufficient to enclose cmene (first of all personal
names etc.) in the structure beginning with the cmene indicators
"la" and the "."?

No. Because what comes within the rules must fit Lojban rules or it might not recognize the delimitors. For example, the Arabic gutturals might be taken by a computer as a glottal stop, which in turn would be a pause. We have a means for quoting non Lojban for a name, using dual identical full-word delimiters with pauses before and after each delimiter, so that you would be la'o .doc. Alfred W. Tüting .doc. where it is mere convention to use something like the language rafsi as the delimiter in order to suggest the basis for pronunciation. This convention was originally devised for scientific nomenclature such as Linnean binomials, where it simply is unacceptable to Lojbanize the name (and even Chinese does not do so, I have been told).

When you use "la" however, you are promising that the name has already been sufficiently Lojbanized that such precautions are unnecessary.

 Shouldn't it be the main goal to giving the reader
an idea of what the name word is pointing to and what's its
common/real pronunciation? So, e.g. writing la maozedong. (instead of
la mau.dzydun.) would function a lot better in this sense.

The tradition in linguistics is that the spoken language takes precedence over the written, and we followed that tradition in setting down our prescription. Presumably someone pronouncing the Lojban as written will come closest to the pronunciation that would be recognized by a native speaker. The ultimate authority as to the correct form of a name is of course the person who names (the parent or in some cultures the person himself upon becoming an adult); Unfortunately, we have no way of asking Mao how he would prefer his name written in Lojban %^).

Or,
writing the the cluster "ng" in (e.g. Chinese or German) names could
imply that it's not pronounced like in British/American
English as two separated consonants (Engl.: fin-ger, Germ.: Finger),
so the reader is not obliged to pronounce it the lb way.

If it is "la" you pronounce it the Lojban way. If it is "la'o" anything goes. If it is "la" you are, absent explicit context, without information as to the kind of name it is and the pronunciation rules that apply. The only rules we know about in Lojban are Lojban rules, and you cannot count on a random Lojban reader knowing any other ones.

In Glosa language, all 'foreign' words (like 'french', 'english'
etc.) had been 'incorporated' (i.e. left untouched). I wouldn't go
that far, though.

With la'o, you have that option.

lojbab
----
lojbab                                             lojbab@lojban.org
Bob LeChevalier, President, The Logical Language Group, Inc.
2904 Beau Lane, Fairfax VA 22031-1303 USA                    703-385-0273
Artificial language Loglan/Lojban:                 http://www.lojban.org