From LOJBAN%CUVMB.bitnet@YaleVM.YCC.YALE.EDU Sat Mar 6 22:44:39 2010 Received: from YALEVM.YCC.YALE.EDU by MINERVA.CIS.YALE.EDU via SMTP; Sun, 6 Jun 1993 22:06:31 -0400 Received: from CUVMB.CC.COLUMBIA.EDU by YaleVM.YCC.Yale.Edu (IBM VM SMTP V2R2) with BSMTP id 4076; Sun, 06 Jun 93 22:05:32 EDT Received: from CUVMB.COLUMBIA.EDU by CUVMB.CC.COLUMBIA.EDU (Mailer R2.07) with BSMTP id 4575; Sun, 06 Jun 93 22:06:54 EDT Date: Sun, 6 Jun 1993 22:03:48 EDT Reply-To: Logical Language Group Sender: Lojban list From: Logical Language Group Subject: responses to Frank Schulz and Jeff Prothero (delayed) X-To: conlang@buphy.bu.edu, lojban@cuvmb.cc.columbia.edu To: Erik Rauch Status: RO X-Status: X-From-Space-Date: Sun Jun 6 18:03:48 1993 X-From-Space-Address: @YaleVM.YCC.YALE.EDU:LOJBAN@CUVMB.BITNET Message-ID: Frank Schulz wrote: >and gismu to be algorithmic. One way to do this is to restrict gismu to >CCVCV form, insist on uniqueness in the first 3 letters, and strip off >the last 2 letters to form the rafsi. That is: > jbama -> jba > klama -> kla > >This does require generating lots of CCV forms. I would rather learn to >generate and recognize new phoneme patterns, than to deal with the >complex lojban morphology. Easy C form additions are velar nasals and >voiced velar fricitives. Rereading my mail, I noted that I didn't complete answer this, observing only that Lojban's current phonology would allow only 240 CCV forms, because it has only 48 permissible initials. You seem to be suggesting that we rememdy this by some additions to the set of consonants. But, even if we could agree on this AND we allow all consonants to be clustered with 'l' and 'r' (the only ones I think you can be SURE of without looking closely at phonological effects), you get only 10 addtional cluster for 1 additional consonant. If you go so far as to make 'n' and 'm' liquids that are allowed to pair with all consonants, you can nearly double the number of clusters. So add 6 new consonants to get 300, then add 'n' and 'm clusters to get 1200, and you are still short of matching all the current Lojban gismu, even if you eliminated the culture words, as some would want, and packed the words tightly in the available space without any attempt at recognition aids (which I realize haven't been too helpful to you, though they have helped others). Of course, finding 6 consonants that you could add might be tough. I could live with a voiced velar fricative, now that I know how to say one, but we have plenty of people who already think they are going to choke saying Lojban 'x' initially, which is defined as the unvoiced frictaive but allows non-phonemic voicing if it helps, since it doesn't have the voiced counterpart. You would have an easier time adding vowels, which is what English has done, but there are many people who think Lojban already is pushing the limits of acceptability for number of vowels. (Jack Waugh, one of the original supporters when we first split Lojban from JCB, constantly tried to argue us into having one vowel fewer than whatever number we currently had in mind %^) Actually, if you've read the discussions on conlang for a while, you will find that people like Harrison and Morneau that try to come up with ideal prescriptions for conlangs say that conlangs should avoid consonant clusters completely. I defend the current system by noting that we considered a lot of tradeoffs (one of which included continuing historical Lojban traditions where possible). The set of consonants, vowels, and clusters was the very first issue we looked at, and we looked at it long and hard. We had linguists helping out, and the arguments ranged from minimizing (as Waugh wanted us to, after the manner of Polynesian languages) to increasing (as you suggest, but also supported by Gary Burgess, our Russian linguist, who jokingly suggested that we adopt the permissible initials of Georgian, since they are obviously speakable (at least by the Georgians). Jeff Prothero answered Frank: >| One way to do this is to restrict gismu to CCVCV >| form, insist on uniqueness in the first 3 letters, and strip >| off the last 2 letters to form the rafsi... > >I suggested this to Jim Brown during GMR, and his objection at the time >was lack of sufficient CCV forms. I failed to respond that by promoting >some 'effective' vowel such as 'r' to C status, and shifting to CC*V >(any number of Cs followed by a V) as the affix form, one could have an >indefinitely last (large) space of trivially resolvable morphemes... >I'm not sure if he would have been swayed, he tends to stick with his >own ideas, and he's very fond of classifying Loglan words using his >lenghth-mod-3 schemes, but the idea remains workable. This idea was proposed to JCB by others in several variations, the most recent public version of which was by Rex May a couple of years ago when Rex was editing LogNet. JCB gave a long, fairly rationale answer, published in LogNet that showed difficulties in meshing the idea with other principles of the Loglan design. Length mod 3 was not one of those principles, although keeping cmavo (little words) distinct from predicates was important (though it doesn't affect what Jeff talks about in this message directly). JCB's conclusion was more or less that the idea might work, and Rex was welcome do develop it into a conlang if he chose, but that such a morphology would not be 'Loglan'. As for mod 3 classifications, I have to note that I am to blame for that aspect of JCB's focus during GMR (the restructuring of Lojban from 1979-82). Loglan originally had only predicate words of 2 mod 3 length because that was necessary to JCB's originally very simple schemes for keeping cmavo separate from predicate words. When it was first contemplated that compounds should be resolvable AND distinguishable from both gismu roots (prims) and from cmavo (LWs), he obviously had to look at the other 'mod 3' word spaces to see how they could be contrasted. Right about then I first got involved in Loglan, and in a long evening discussion where he was trying to explain what he was trying to do in GMR (which I didn't understand at the time), and I was trying to find out how to get a word like "trumpet" into Loglan (I was trying to translate the "trumpets of glory" from _Man of La Mancha_), we came up with the scheme of 2 mod 3 for prims, 0 mod 3 for compounds and 1 mod 3 for borrowings. Whose idea it was, I can't remember, but JCB got into an real "aha!" mode late in the evening about this. He then went on to develop his ideas into an essay that 'responded to me', which was published in TL 3/4. While the scheme did not survive to be the final form of Loglan GMR, it certainly dominated JCB's thinking for a long while. (JCB now denies my participating in any of the thinking about GMR, or stimulating the TL 3/4 essay that set the framework for the GMR solution. This is either politics or his poor memory, since what happened made such an impression on ME - the fact that I, with no linguistic knowledge or real understanding of Loglan, could still have such a meaningful contribution, was what led me to become emotionally committed to the project.) lojbab