From LOJBAN%CUVMB.BITNET@UBVM.CC.BUFFALO.EDU Mon Nov 6 01:36:25 1995 Received: from VMS.DC.LSOFT.COM (vms.dc.lsoft.com [205.186.43.2]) by locke.ccil.org (8.6.9/8.6.10) with ESMTP id BAA02713 for ; Mon, 6 Nov 1995 01:36:21 -0500 Message-Id: <199511060636.BAA02713@locke.ccil.org> Received: from PEACH.EASE.LSOFT.COM (205.186.43.4) by VMS.DC.LSOFT.COM (LSMTP for OpenVMS v1.0a) with SMTP id 7574E1F3 ; Mon, 6 Nov 1995 2:31:36 -0400 Date: Mon, 6 Nov 1995 01:29:57 -0500 Reply-To: Logical Language Group Sender: Lojban list From: Logical Language Group Subject: rafsi To: lojban@cuvmb.cc.columbia.edu Cc: lojbab@access1.digex.net Status: OR Well, pc is far more charitable than I am feeling. The problem for me is that when someone puts this much effort into a proposal, no matter how bad I think it is, I feel compelled to respond. Must be that customer ethic that mark accuses me of not having %^) I have spared everybody my 1000 line response to mark's comments to me (anyone who wants it is welcome to ask, though), but I guess since pc felt this proposal was worth responding to, I will post my critique. I apologize in advance if this comes across rather denigrating. Thisis the type of proposal I see no good in, except for the possibility of ediucating people of all the factors that went into the current design, something I shouldn't spend time on now. Oh well. Here goes: "Mark L. Vines" >Okay. My "Rafsi Repair Proposal" is an alternative to the current >system of short rafsi & lujvo compounds in Lojban. My primary purpose >in proposing this alternative is to eliminate what I call homophone >affix ambiguity, a defect of the current system in which hundreds of >short rafsi are identical to cmavo with unrelated meanings. I'd >appreciate any _constructive_ criticism. The most constructive suggestion I can think of: Do something constructive with your time. Write a ckafybarja story or something. This is a waste of time. You may think you are changing something minor, but it is something quite fundamental to the language. You are requiring changes to the phonology - what diphthongs are acceptable, as well as where they are acceptable. You are requiring changes to the morphology, changing the definition of the shapes of Lojban lujvo, as well as their component rafsi. The ONLY things you aren't changing are the gismu and cmavo lexicon, and the grammar (at least I don't THINK you are). The morphology is baselined; the phonology is baselined - both have been baselined for 6 years. You are proposing MAJOR changes to both. They are not going to even be considered no matter how good your proposal might be as a whole, and I don't think it is all that good from what you have shown - to fix one "bug" that isn't widely acknowledged to be a bug, you introduce many "features" that ARE bugs. You also show that you don't really know the existing rules all that well, as I show below. Either that, or you don't respect them - which in my book usually means the same thing. Your proposal is so far outside the bounds of the language that it is NOT Lojban any more, and probably not Loglan even. You are proposing a splinter language, and luckily for us, no proposed splinter has ever drawn more than a couple of people. (Jim Carter's guaspi is thought by many people to be a satire of Lojban and conlang development, though he takes it quite seriously. Your proposal is nearly as extreme as his in my book - all you need is tones.) The bottom line is that the Loglan/Lojban project is NOT in the least bit interested in proposals to "improve" the language. We want to document the language, and use the language. Only most rarely and in the relatively far past, have we even considered significant proposals for change, and I can safely tell you that none has made it to first base except when the person was doing a MAJOR language project for us that proved his competence to the whole community beyond any doubt. You may consider us close-minded if you wish. You may even consider us "anti-customer". But we don't want our "customers" redesigning the language any more than Ford wants their customers redesigning their cars. I would rather lose a reformer to the community than tolerate a schismatic reform. >My goals are: to work with the existing gismu & cmavo; to provide at >least one short rafsi for every gismu; & to eliminate homophone affix >ambiguity. None of which have ever been high priority goals. > IU = any Lojbanic diphthong (16 variations) 14 last I counted. ai/au/ei/oi and 5 each iV and uV. But the latter are permitted only in VV cmavo, and are tolerated in names. Oh, I see down below that you've added in iy here and perhaps uy. These are not even used in VV cmavo. > EI = any of the following diphthongs (5 variations): > ai au ei oi ui; possibly a subset of IU > (unless we have to add eu for a total of 6 variations) eu is not even a defined Lojban diphthong at all. ui does not fit with the other 4, since it does not occur after a consonant. > ND = any consonant pair that can end a monosyllabic rafsi > (76 variations); a subset of DL; table at end of posting; > we could call ND pairs "permissible finals," but as a rule > they terminate only non-word-final syllables, not words You suggest a list and give no phonological basis for that list. It has no relevance in Lojban EXCEPT in your proposal. The permissible initial and medial apply to names and fu'ivla as well as gismu and lujvo. > In later postings, I'll also discuss: > XR = any DL consonant pair which is not ND (103 variations) > TK = any XR consonant pair which is not ST (not yet counted) > IE = any Lojbanic IU diphthong which is not EI (11 variations) > IA = a not-yet-defined subset of IE (about 7 variations) > SP = any ST consonant pair which can precede any IA diphthong And some people call the existing system complex! >= = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = = > >Okay. In this proposal, short rafsi can take the following FORMS: > > CIU => these rafsi are reserved for those cmavo identical to them; > gismu are _not_ eligible for CIU rafsi 1/3 of these are not permissible as cmavo since Cui and Ceu are not valid cmavo forms. > STAC => reserved primarily for the corresponding STACA gismu > > CAND => reserved primarily for the corresponding CANDA gismu These are ALL reserved for the current gismu. Indeed all CCVC and all CVCC are reserved, against the possibility that any given other gismu form word might be assigned as a gismu. It becomes obvious from your later suggestions, though never stated explicitly, that your primary goal is to reduce lujvo syllable counts presumably by making most n-part lujvo n syllables (plus hyphens if you still have them). This is not particularly desired nor desireable. It is ADVANTAGEOUS for lujvo to be longer in syllable counts than gismu (since they are more complex words, and generally also less common words - hence Zipf demands that they be longer), and having a lot of syllables creates and maintains the current spoken rhythm of the language. Your proposal would thus drastically change the whole sound of the language. > STEI => gismu are eligible Again, 1/3 of these are invalid in brivla. The others are part of fu'ivla space, which you would be cutting significantly by this proposal. > In later postings, I'll also discuss: > SPIA => gismu are eligible All of these are invalid in brivla. I will inform you in passing that JCB DOES permit all 14 diphthongs in his cmavo and rafsi and lujvo. When we remade Lojban we introduced the apostrophe partly *because* CEI diphthongs are non trivial for a variety of speakers, and the effects on differing speakers varies. So in this sense your proposal is a step backwards from what we now have. >In the future, perhaps some of the remaining 3749 possible STAC rafsi can >be assigned to new gismu, or to le lebna valsi. Meanwhile, we still have >1009 gismu left without any short rafsi. These are all CADLA gismu >(equivalent to CVCCV gismu in conventional notation.) fu'ivla are not assigned rafsi. If a fu'ivla concept warranted rafsi, it would be assigned to a gismu. >The total CAC "rafsi space" allows for 1445 variations. Several of >these are reserved for short cmavo such as se, te, ve & xe. Out of the >CAC rafsi not assigned to cmavo, 532 are reserved for the remaining >(CAXRA) gismu. At this point, every gismu has at least one monosyllabic >rafsi, & there are still several hundred CAC rafsi left over, which can >be assigned to any eligible STACA & CANDA gismu whose ST or ND pairs >make their STAC or CAND rafsi difficult to pronounce. We will IMO want >to use all the CAC forms. Nice idea and we tried. But some CVC rafsi cannot be derived in any way from any current gismu, especially those with 'o' in them. I dare say that you will not come up with any mapping that goes substantially beyond the current percentage in use (something like 80%). If you do, it will be by making even stranger convolutions than kerfa->kre. >The total STA "rafsi space" allows for only 240 variations, & we'll want >to use all the STA forms. Again, we are using all that we can, and even broke one of your rules (kerfa -> kre) to get that much coverage. >Obviously, one great weakness of my "Rafsi Repair Proposal" is the small >number (present maximim 528) of word-final short rafsi. We will have to >be very selective in deciding which gismu are entitled to monosyllabic >word-finals, because 528 is less than half of the 1340 existing gismu. >Of course, this weakness is a direct consequence of my refusal to assign >CIU & CA'A rafsi forms to any gismu. But some such refusal is required >for the elimination of homophone affix ambiguity. In short you are straining one important part of the language for the benefit of another unimportant part. If you are so upset about "homophones" that aren't (since one is a word and the other is not) that mean diffferent things, why aren't you campaigning for better correspondence of cmavo derived from gismu to be more mnemonic with the gismu. This one is at least a more familiar complaint, one people including me agree is aesthetically unpleasant, albeit one equally intractible. >We may also want to consider adding another word-final short rafsi form, >namely, the SPIA rafsi, where SP is a subset of ST (the 48 initial >consonant pairs permitted in Lojban), & IA is a subset of IU (the 16 >diphthongs). I haven't yet identified the SP consonant pairs or the IA >diphthongs, but SPIA would be a pronounceable region of the STIU "rafsi >space." > > (Not everything in STIU can be easily pronounced as a single >syllable. Try saying "mlua xriu jvio kluy vliy" real fast a few times, >saying the vowel pairs as diphthongs, & you'll see what I mean.) And not surprisingly, none of these are allowed in Lojban brivla. There is a reason why we limited our primary diphthongs to 4. Note that ANY change to the language that introduces the risk that some speakers will pronounce one syllable as two in word final position completely screws the morphology because you have lost consistency in determining the penultimate syllable to be stressed. >Such a new rafsi form would add maybe 252 or so to our list of >word-final short rafsi, for a new maximum of 780: one word-final rafsi >each for more than half the existing gismu. Fewer than the current set of CVV rafsi we currently have - all for the sake of an aesthetic principle, for which you cannot cite any ambiguity that results from it. >Once the word-final short rafsi have been assigned, reassignment of some >CAC rafsi may prove to be desirable. After all, a gismu which has a STA, >STEI or SPIA rafsi may not need a given CAC as much as another gismu >does. But even to speak of such trade-offs is to acknowledge that my >"Rafsi Repair Proposal" is not especially elegant or consistent. Its >only real virtue, besides eliminating homophone affix ambiguity, is its >compatibility with the gismu & cmavo already existing in Lojban. You don't even acknowledge the primary selling point, which is the cut in syllable count, which of course in my book makes it quite unLojbanic. >Okay. Here's a TABLE of ND pairs ("permissible finals" -- consonant >pairs which are not word-final but merely internal-syllable-final): > > -b -c -d -f -g -j -k -l -m -n -p -r -s -t -v -x -z > > b- bd bj bv bz > p- pc pf ps pt > v- vb vd vg > f- fk fp ft > g- gd gj gv gz > k- kc kf ks kt > d- dj dv dz > t- tc tf ts A bit of an assymetry in v and f, and d and t, compared to the others, isn't there? This usually means English bias. > c- ck cp ct > j- jb jd jg > s- sk sp st > z- zb zd zg Well these are symmetrical at least. > m- mb mc mf mp ms > n- nc nd nf ns nt But these aren't for some reason. > l- lb lc ld lf lj lm ln lp ls lt lv lz > r- rb rc rd rf rg rj rk rm rn rp rs rt rv rz You are putting a very heavy load on l and r - 1/3 of your total), and a asymmetric one at that. Linguist and those from non-English native languages have correctly criticized Lojban in that we should have considered eliminating one of these two letters. Putting more load on them is not acceptable. > x- No x clusters - malglico. >This table is certainly open to revision. At present it includes 76 >pairs. Please note that acceptance of the proposed ND table & CAND rafsi >will require extending the phonological rules of lujvo compound formation >(& -y- or -r- hyphenation) to cover ND,C triples & ND,ST quartets. (All >consonant triples currently take the form C,ST; also, under the current >system, quartets occur only in some cases of -r- hyphenation.) Quartets NEVER occur in lujvo. You may be looking at the r/n/l/m glue used to hold categorizers onto Type III fu'ivla, but in fu'ivla you could have 10 straight consonants if you follow the rules. The existing system has been exhaustively tested for unambiguity. You seem to want lujvo of the form CCVC-CCVV and the like. I would envision that proving the unambiguity of the result would be a nightmare, and the Slinkui and Tosmabru tests would have to be greatly expanded when they are already hard enough to do in your head. In short, you are fiddling with the unfiddleable. lojbab