From: Nick Nicholas Message-Id: <199302091034.AA23540@munagin.ee.mu.OZ.AU> Subject: Re: GENERAL: language stability and *mo'e To: lojbab@GREBYN.COM Date: Tue, 9 Feb 93 21:34:49 EST Cc: lojban@cuvmb.cc.columbia.edu (Lojban Mailing List), nsn@mullian.ee.Mu.OZ.AU (Nick Nicholas) X-Mozilla-Status: 0011 *smile* What, *me*, stir up a mini flame-war? :) I suppose Lojbab has thrown down a gauntlet of sorts on the changes I've mentioned: I'll dwell a bit on them. Accuracy VUhU: I don't quite see the problem. {xa'i} will be a function taking two arguments, the nominal value and an accuracy, and returning a 'fuzzy' value. 1.0 +/- 0.1 = {li papino xa'i pipa}. This doesn't seem problematical at all; am I missing something? MEX Change: current: = MAhO letteral-string /BOI#/ proposed: = MAhO mex No ifs, no buts; as became clear in Iain's Lambda Calculus piece, Lojban currently has no means of making functions arguments of other functions, including inverting, squaring, composition, and nesting. We cannot currently say f(g(x))! The grammar is broken, and requires fixing; unfortunately this may necessitate extra terminators. traji: x4 to become x2: *shrug* It'd be consistent with {zmadu}, and analogy is pretty important in formulating these place structures. {xe'e}: *sigh*. It's not surprising the usage hasn't been picked up. Consider: our NLs do not have unclefted place structures. Everything we want to talk about has its own place. The uncleftings are 1) recent, and 2) extensive; the give us a language in which, to talk about the things we want to talk about, we have to dig them out of clauses. We do so with {tu'a}, {jaigau}, and {voi}; but none of them do what would be the simplest transformation: to recover from the unclefted place structure the clefted place structure. That is, to turn event x1 {rinka} event x2 into something approximating agent y1 {rinka} patient y2 to do action y3 by doing action y4 We want to talk about y1; we can currently do so by saying: le jaigau rinka tu'a da rinka da voi rinka We want to talk about y2; it is, after all, the best way to say "patient" in this sense in Lojban. tu'a da se rinka da voi se rinka but no {le broda} form. With {xe'e}, which costs no grammar change, we can say le xe'e rinka [y1] befai le rinka [y4] le xe'e se rinka [y2] befai le se rinka [y3] Maybe it doesn't add capability that doesn't already exist (clumsily) with {voi} in constructing sumti. But the clumsiness *is* there, and won't go away. And as long as these unclefted place structures are around, people will either naturally cleft them as the language evolves, because they *do* want to talk about those hidden-way sumti, and not only the ones hidden in x1 that you can get at with {jaigau} --- OR take advantage of a less bothersome means of extracting the places than {da voi} --- which has a *huge* conceptual difference from a {le} sumti. The reason, I believe, {xe'e} hasn't seen more use is that the full extent of unclefting hasn't sunk in to the lojban users yet, and this is an issue which will turn out irritating only with very extensive, idiomatic usage. The reason I've kept it at xVV is that I didn't believe I had the authority to propose an unassigned cmavo. But really, the proposal costs nothing in terms of grammar, and is a quite straightforward transformation: lenu da bu'a cu broda -> da xe'e broda fai lenu bu'a As for *mo'e, my main gripe remains, what do we do with BAI words like {fi'i} --- do they take {*mo'e} or {ne}? Is it the case that all {*mo'e} phrases have a deep structure {lenu... co'e lenu} sentence, as in the examples in John's posting? Apologies for abruptness. The grammar snags John reports are bad news, and I'll get to them next mail item. "Kai` sa`n swqh~kan t'akriba` piota`, N N O nsn@munagin.ee.mu.oz.au kai` sa`n plhsi'aze pia` [h [w'ra te'sseres, I I L IRC:nicxjo RL:shaddupnic sto`n e'rwta doqh~kan eutuxei~s." C C A University of Melbourne. K.P.Kaba'fhs, _Du'o Ne'oi, 23 E'ws 24 Etw~n_ K H S *Ceci n'est pas un .sig*