Return-Path: From: cbmvax!uunet!math.ucla.edu!jimc Return-Path: Message-Id: <9108231620.AA26859@luna.math.ucla.edu> To: lojban-list@snark.thyrsus.com Subject: Re: brivla unique meanings (was: loopholes po la korant.) In-Reply-To: Your message of "Fri, 23 Aug 91 02:25:00 EDT." Date: Fri, 23 Aug 91 09:20:09 -0700 Status: RO X-From-Space-Date: Fri Aug 23 17:29:06 1991 X-From-Space-Address: cbmvax!uunet!math.ucla.edu!jimc There's a scale here: each person acquires the meaning of each brivla through unique experience, but the language hardware/software drives everyone's meaning in the direction of being the same. Lojbab thoughtfully discusses the unique end of the range. But one of the basic philosophies of a "language artifact" or "constructed language" is that the language architect prescribes the basic design of the language. While it is possible to explicitly decline to prescribe (as Lojbab does in the case of today's lujvo), I think this choice is not in the spirit of "constructing" or "designing" a language. In fact, his position is that there hasn't been time to deal with lujvo, and guidelines (his word) will appear eventually. On the other hand, professional linguists study natural languages, for which there is no effective architect and the only credible rule is "let usage decide". We're engineers, not observers, and the world of normal linguistics is turned upside down for us, particularly in the dimension of observing usage. This is yet another scale: is usage determined by observation or by design? I see a third scale perpendicular to the "individual <-> uniform" range, intersecting at the "uniform" end to make a T shape. (Of course it can slide along that axis but it is more clearly associated with the uniform meaning position.) The endpoints are: A. The language architect knows in advance what's going to work, and prescribes it unalterably in advance of testing on live users. B. The language architect tests various versions of a feature and selects the best one for final prescription. In recent discussions with Lojbab about diklujvo, I have been trying to make similar our understanding of my position on this latter axis and to negotiate a position acceptable to him. But I believe he has been negotiating a position on one (or both?) of the other two axes. As could be expected, the result was excessive energy spent, no results, and a desire on both our parts to abandon the effort. I still feel, though, that: A. The Loglan family of language artifacts is unique in that a very small set of gismu can be combined into lujvo that cover an extremely broad range of semantic space. B. Lojban can be made even more useful and easier to learn if the meanings of the lujvo are derived algorithmically from the meanings of the gismu, rather than letting each lujvo have a unique meaning that has to be learned the same way the meaning of a gismu is learned. Then a speaker or listener would only have to know the gismu meanings and the rules, and could create or interpret lujvo he had never heard before. It is understood that some meanings cannot be achieved this way so there is a place for "creative lujvo". It is understood that a number of technical problems have to be solved before such an approach can work in Lojban. It is also understood that a number of people wonder whether such an algorithm is even possible (i.e. whether I'm deluding myself that I've accomplished it in -gua!spi). I hope Coranth's comments have helped us realize upon which axes negotiation needs to be done. -- jimc