From a.rosta@lycos.co.uk Mon Dec 02 03:02:44 2002 Return-Path: X-Sender: a.rosta@lycos.co.uk X-Apparently-To: lojban@yahoogroups.com Received: (EGP: mail-8_2_3_0); 2 Dec 2002 11:02:44 -0000 Received: (qmail 42503 invoked from network); 2 Dec 2002 11:02:44 -0000 Received: from unknown (66.218.66.216) by m13.grp.scd.yahoo.com with QMQP; 2 Dec 2002 11:02:44 -0000 Received: from unknown (HELO lmsmtp03.st1.spray.net) (212.78.202.113) by mta1.grp.scd.yahoo.com with SMTP; 2 Dec 2002 11:02:44 -0000 Received: from oemcomputer (host81-7-58-125.surfport24.v21.co.uk [81.7.58.125]) by lmsmtp03.st1.spray.net (Postfix) with ESMTP id C67293CF43 for ; Mon, 2 Dec 2002 12:02:38 +0100 (MET) To: Subject: RE: [lojban] Re: Why we should cancel the vote or all vote NO (was RE: Official Statement- LLG Board approves new baseline policy Date: Mon, 2 Dec 2002 11:04:47 -0000 Message-ID: MIME-Version: 1.0 Content-Type: text/plain; charset="iso-8859-1" Content-Transfer-Encoding: 7bit X-Priority: 3 (Normal) X-MSMail-Priority: Normal X-Mailer: Microsoft Outlook IMO, Build 9.0.2416 (9.0.2910.0) In-Reply-To: <5.1.0.14.0.20021201175116.031410b0@pop.east.cox.net> X-MimeOLE: Produced By Microsoft MimeOLE V5.00.2615.200 Importance: Normal From: "And Rosta" X-Yahoo-Group-Post: member; u=122260811 X-Yahoo-Profile: andjamin Lojbab: > >To be concrete, I would make the following proposals, which I realize > >would not currently have majority support > >(1) All monosyllabic cmavo whose usage to date falls below a fairly high > >threshold should be replaced by a phonologically similar disyllabic > >cmavo > > There are some areas of the language (like Mex) that simply haven't seen > enough usage at all to get useful statistics. If one argues that therefore > nothing in Mex should ever be monosyllabic, then one is arguing that there > should not be a Mex, since there is in fact no usage that demands a > Mex at all I don't follow your reasoning. I would argue that only high-frequency cmavo deserve monosyllabicity. But for cmavohood in general, frequency should not be a criterion (except maybe for those cmavo that could easily be replaced by a lujvo). Anyway, I definitely think that nothing in the never-used portion of Mex should be monosyllabic. As for whether it should exist at all, I refrain from expressing a view. > >These freed up monosyllabic cmavo, along with all other unassigned > >monosyllabic should not be assigned. After enough text has been written > >after this point -- a million words of good-quality writing, say -- the > >corpus should be examined and monosyllabic cmavo assigned to the highest > >frequency forms. (In hindsight, I think no monosyllabic cmavo should > >have been assigned before a million words of quality usage, but it's > >too late now for this.) > >(2) As described on the wiki at "Exploiting the preparser". The idea is > >that once there is enough text to generate high quality statistics, > >new cmavo could be introduced that rewrite as high-frequency cmavo > >sequences. If this were to happen, it would be many years hence, and > >I simply think we should not at this stage constitutionally prohibit > >it from ever occurring > > The bottom line is that if there is any significant possibility of this > happening in the future, then large numbers of people will refuse to learn > Lojban, and we will never GET the usage needed to make Zipfean > decisions. You have NO idea how painful the revisions to TLI Loglan were, > over the years, to the TLI community. EVERY change, no matter how minor, > lost people, and the major changes lost lots of people even when it made > the language much better. Even today, 15 years after we last used TLI > Loglan in any significant way, Nora more than I (but both of us) still > occasionally pull up a TLI gismu rather than a Lojban gismu for a > concept. Every change you guys make to the CLL baseline with your jboske > debates (and I won't pretend any longer that your debates haven't > effectively changed the baseline, which is one reason why the new policy) > makes it that much more likely that I never will speak Lojban as much as I > did in 1997. If the byfy result is planned to be modified, by any criteria > set down like planned Zipfean adjustments, then I won't bother to learn the > language. And if I won't, I'm sure a lot of others won't I have the sense that everyone who has become active in Lojban in recent years is a newcomer, not an old Loglander returner. The old guard have more of the force of a dead weight, an impediment, and it is the voices of the likes of Adam, Craig, Jordan, xod, Robin and (horribile dictu) Jay that we should be paying the most attention, as most representative of the future of Lojban. AFAIK, Michael is the only Lojbanist over 40 who is actively and publically producing Lojban text. Furthermore, we can see in these young Turks a dynamism in trying to move Lojban forward, in sundry ways, that is in marked contrast to the current inertia of the oldies who in the days of their own youngturkhood founded Lojban, and I expect it won't be long before the new generation take over the board and decide to forget about the old guys who were forever stating the conditions under which they would refuse to learn Lojban without ever showing signs of a real intention to learn it. If setting in place the preconditions for possible future changes to the language will improve the lot of future Lojbanists, but alienate people who at some time in the past said they might learn Lojban but haven't, I will vote for giving future Lojbanists freedom of manoeuvre, and if that leads to you not bothering to learn the language that will be sad, but probably would amount just to giving you an excuse for not doing what you wouldn't have done anyway. > > > And: Baseline > > > > > > I assume what you mean is, the baseline encompasses semantics as well > > > as syntax. I whole-hearted agree, but I do not understand what the > > > statement should concretely say to assert this, and I would have > > > thought it was obvious anyway > > > >It's not as easy as this. It's easy to determine whether a text contains > >ungrammatical sentences, but much harder to determine whether intended- > >meaning matches baseline-meaning, since the gricean principles that > >natural language exploits can allow such great mismatches between > >baseline-meaning and communicated-meaning > > I deny that the (existing) baseline encompasses semantics because the > semantics has not been defined. The new baseline, to the extent that the > cmavo are finally defined rather than merely listed and keyworded, MAY > encompass whatever semantics is included in those definitions. We won't > know if it will be easy or hard to determine matches between intended and > baseline semantics until we have documented baseline semantics We will know, because some of have studied language beyond Lojban, and are informed by that experience. The closest analogy from natlang experience would be to 'correct' the writing of foreign learners (of, say, English) without being stylistically prescriptive. That's a very difficult exercise. In reading a text written by a foreigner we can fairly easily identify what is plain ungrammatical, and we can identify stuff that a native speaker wouldn't say, but it's very hard to separate out from that the stuff that is bad for semantic reasons. > > > And: Unintelligible cmavo > > > > > > We will go with the supplicatory model before we decide we don't know > > > what a cmavo is. The current statement of the baseline does not allow > > > cmavo deletion, because all cmavo are documented in CLL, one way or > > > another. To erase cmavo would be a major techfix, and I am opposed to > > > such ventures on principle > > > >In my experience, the supplicatory model doesn't really work, because > >-- to put it hyperbolically -- we end up with Lojbab's half-baked > >understanding or recollection of what something meant 20 years ago > >in Loglan or what a gang of now-invisible and uninterrogable > >Lojbanists came up with 15 years ago > > Possibly. In which case the byfy should feel free to reject it. In many > cases we have evidence and not merely my half-baked > recollections. Including a TLI rep in the byfy discussion may also provide > an independent source for "what Lojbab remembers trying to do" because in > most cases, TLI is still doing it > > >I agree that deleting these cmavo is quite a radical step, but it is > >also a refreshingly honest one > > It is not a practical one, unless you know someone who wants to buy several > hundred copies of CLL that would suddenly be worthless. We made a serious > commitment to NOT changing the language, and backed it with a $17,000 > investment which has not yet been repaid. And while it might be honest, > there also might be hundreds of potential Lojbanists that would go > elsewhere, for fear that in another 5 years we might be "honest" again and > throw out a bunch of stuff that they've spent time learning Must we pander to irrational fears? Nobody can have spent time learning the unintelligible cmavo, since there is nothing to learn. Anyway, I subsequently changed my suggestion to that we handle unintelligibles by, if they are included in the minidictionary at all, a note saying nothing but that they are to be defined through usage. > > > At the very most, if noone has ever ever ever used lau, I might accept > > > turning lau into say xu'e, and releasing lau. But for a one syllable > > > cmavo to be necessary for something Zipfeanly, it has to be something > > > so urgent and obvious, I'd have thought the heavens would be clamoring > > > for it by now. In my book, xa'o and mu'ei are the only ones even close > > > to this (with mu'ei endangered by sumtcita ka'e); and they're not that > > > close > > > >You will see from what I say above that I favour deassigning monosyllabics > >that are not heavily used, and reassigning them only after a LOT of > >skilled usage > > The whole area of alphabets and lerfu is tied up with Mex. We cannot say > how useful Mex will be, but it certainly will not be useful if we make it > more difficult to use Avoiding making mex harder to use is not a good reason for not making the rest of the language easier to use. I am proposing (and I think Jordan is too) that mex and other stuff that has never seen substantial usage be made more longwinded so that future generations of fluent lojbanists can decide where shortwindedness can most efficaciously be applied. > >As for disyllabics that currently clamour for monosyllabics, I personally > >crave them for {du'u} above all, and also {lo'e}, {le'e} and perhaps > >{ke'a} and {ce'u} > > I don't crave ANY change to any cmavo that I already know and use. I want > the bloody language to stop changing long enough for me and others to > really learn it and BECOME skilled speakers Perhaps this is partly because you don't use the language much and partly because you use it without scrupulous regard for meaning and so say things like {le nu} when {lo'e du'u} would be more correct. But I also concede that you have always advocated a static Lojban design over an improved Lojban design. But I have the impression that within Loglan/Lojban there has always been the more whorfian Conservative faction and the more logical/engineerist Progressive faction. Lojban was founded by Conservatives, but it never advertised itself as the Conservative alternative to Progressive Loglan; it advertised itself as the thriving and publicly owned alternative to the moribund and privately owned Loglan. So I don't think it is fair to argue that Lojban is intrinsically Conservative. > >Certainly I find myself using {lo'edu'u} constantly > >and find it extremely irritating -- infuriating, even (given that the > >language design could have reduced it to two, one or even zero syllables) > > Whereas "la'edi'u" was a common phrase from the earliest versions of > Lojban, and no one ever suggested that it deserved a shorter form. Nor do > I want one now It is not true that noone has ever suggested that it deserved a shorter form. Xorxes, by far and away the most competent Lojbanist there has been to date, has said that "di'u" should have meant "la'edi'u" (hence "la'edi'u" would be said as "lu'edi'u"), and he has also railed often against the criminal squandering of monosyllabic cmavo. > >Regarding the existing experimental cmavo, I suppose we could have a > >poll about which, if any, we would like to make official. But I > >would prefer to get rid of the notion of officialness > > That rejects the idea of a baseline No it doesn't. The baseline would list the cmavo that have a defined meaning, and would treat all other cmavo as undefined (which is how most current 'official' cmavo are), but not as unofficial. An unfrozen baseline could add new definitions as they are decided upon. > >and instead > >simply say that the mini-dictionary fixes the meaning of the cmavo it > >lists. A proper syntactic parser should not have the mahoste built > >in to it, but should instead take input from a community-maintained > >mahoste that can be updated with cmavo not listed in the mini-dictionary > > Then write one I have (collaboratively) written one for cmavo that are not in the official mahoste. It is on the wiki. It is easily adaptable (with about 1 minute's work) by anyone writing a parser to take input from a mahoste. > Meanwhile, in how many natural languages are the set of structure words > really an open set? Openness vs closedness is a much more robust taxonomic criterion than structure words vs content words. (IM-scholarly-O.) But there are open 'structurey' classes (e.g. pronouns in some lgs) and closed 'contenty' classes (e.g. adjectives in some lgs). The real issue here is whether we can tell at this stage what set of cmavo future Lojbanists are going to find most useful, and whether we should allow future Lojbanists to determine for themselves which cmavo they require. --And.