Message-Id: <199502080151.AA28186@nfs1.digex.net> From: "John E. Clifford" Date: Tue Feb 7 20:52:04 1995 Subject: ago24 & replies X-From-Space-Date: Tue Feb 7 20:52:04 1995 X-From-Space-Address: LOJBAN%CUVMB.BITNET@uga.cc.uga.edu Sorry about that last try; I am shifting computers and my Lojban files are on the other one from the modem. When I try to do this stuff from memory, I get random products from four stages of Loglan, with '76 being the most likely. Please relex and bear with me. That aside, the last exercise was to find plausible Lojban for "ago," "plausible" being defined as expressions that are true in the relevant situation and for which Griceans could find an acceptable explanation for the fact that the expression did carry the "ago" message, whatever its literal meaning was (I assume that Griecan pragmatics is not the Freudian psychology of lin- guistics, but that that some possible turns of phrase cannot be explained to carry some meanings). So far as the responses go, I see no evidence that the exercise was a failure. All three of the expressions presented are demonstrably true in the situation in question and at least the first and second (with -mei and -moi -- correct places, too, please) are patterns which work in natu- ral languages and so, presumably, have good pragmatic explana- tions. The only criticisms are that the first confounds the origin and the magnitude of the displacement and that they are kludgy. Actually, the first uses a device for indicating an origin in order to specify an magnitude of displacement, a trick used in several natural languages.As some examples in another thread show, the differences between the types of arguments will usually sort matters out correctly. (And, incidentally, coming up with the same solution as Lojbab in Lojban is hardly a criticism or a grounds for dismissal.) As for being kludgy, they are a least meaningful Lojban and true, whereas the proposed alternative is questionable on both counts. I am not sure what "a medium tempo- ral distance from the set of three years" means (if I understand _za_ correctly) and so am unsure that it is true of an event three years ago. Now, of course, saying something obviously false or hopelessly unintelligible is one way to trigger conven- tional implicatures and so this barbarism may work and even work better than my versions. But I do not see any evidence provided that it will. To be fair, the alternative proposed is not meant to mean "ago" in Lojban-as-of-now but is rather a proposal for a "small, cheap" change to Lojban-new-and-improved. One of the effects of living with an ancient cmavo list is that I regularly see how much has changed in the last several years. Some of these changes have no doubt been useful and needed. Some of them have arguably improved things somewhat (adding a place for the sequence to n- moi may be one of these, though, thinking back to the fight to get the origin added as a place, it does seem that the sequence probably is less often referred to and so should be the third place rather than the second -- I assume the origin is the third place, so "_befi la Cac_ to be on the safe side"). But a number of them seem to me to have been "small cheap" changes to solve problems for people unable -- or unwilling -- to solve them within the Lojban they were given. And, since these problems tend to turn up randomly and piecemeal, these "solutions" have been inserted in the same way, often without regard for broader pictures. Now, I do not know what happened to change _zai_ from a tensor (presumably a metric on a vector, though the description is not very clear) to whatever it is now, but it seems to have been done without regard to the need for such tensors. Similar- ly, the move from _ze'e_ meaning "during an indefinite (i.e., not specified as short, medium or long) interval" to (can this really be right?) "an infinite time interval" (when would we use that?) was not carefully planned (and it is hard to imagine what could have prompted it -- even if it turns out that I did it myself). Finally, some of the shifts seem just pointless, if not counter- productive. Some of the above may be of that sort, but the shift of _lo_broda_ from "all broda" to "some broda," both of which are redundant for Qda poi constructions and much less efficient in that role, strikes me as an especially clear case of change for change sake and without regard to further consequences. I am not sure that the proposed changes in _za zi zu_ (why all three, by the way?-- since the metric applied gives the more exact size) is as pointless as these or even that it reflects the frequent quick-fix-rather-than-working-in-the-system attitude that some changes surely have. The claim that the set are not used (i.e., that no one has yet had occasion to use or, perhaps, has yet figured out how to use) shows at least some concern with the effects of such a shift, though the systemic ramifications have not been dealt with yet - what does this do for the set as tense affixes, for example, or for the meaning of other tense affixes that also serve as tags? Barring some clarification on those points, I would have to say that the proposal had not yet earned acceptance. (BTW, what is _bu'u_? I still have it as a bound predicate variable, the only second order part of this system. In the same way, _ne'a_ is given as a non-restrictve relative clause giving membership in a set and _to'o_ is a toggle for print case on words. Losing the last two does not seem a loss, assuming they went to a good cause, but the first one would cripple the logical nature of the language if it went completely. Has it been re- placed?) Some Notes on Related Threads 1. I see _xo'u_ is still alive. Good! The move to the heqad of the highest prenex is simply the simplest rule and the one that seems to be involved most often in natural languages (besides, every Q is defined ultimately from its highest prenex position, so all we really affect is order here). If there are good reasons for a different view, we can still adjust (obviously, last in highest prenex is not useful, since that we can obtain in after- thought mode already). Some of the comments on _xo'u_ seem to be more appropriate to whatever it is that marks terms in opaque contexts that can shine through the opacity and be taken to have external reference. Incidentally, most of the examples of _xo'u_ are not from unlikely meanings but from cases of English "any," which dfunctions in English in just the same way. 2. The use of _vi_ for "at" is already in Urloglan, c.1960, and probably is ineradicable. Nor is getting rid of it necessarily desirable: "at" just is not a precise term at all and the dis- tinction between it and "right up close to" is not going to be more than one which varies with purposes. See Mad Ludwig on "Stand just there." 3. The thing And wants for this sibling problem, one from column A and one from column J is a Cartesian product, for which we once had a cmavo in JOI, though I cannot now lex it. 4. I am not sure that a house can be NO color at all and it it is several different colors, they come down to a single JOI color combination, a suitable instantiation for _da_. But some things can be no color, so that part doesn't count for _dakau_. Of course, if x2 of _skari_ is a set, then even the 0 case works. Still, I find _makau_ and its ilk very crisp. 5. If we are to have lambda variable, we need a slough of them, since the whole point of lambdacism is that different ones can be replaced differently. In particular, _lo ka kea mamta kea_ is not the mother relation but the self-mother relation (one that rarely holds except for the odd goddess), since _kea_ must be replaced by the same term in all its occurrences on each applica- tion. (authoritative utterance of someone who studied the lambda calculus with Church his own self.) 5. Yes, functions are predicates, but special ones. To be a function, a predicate has to verify the appropriate form of AxAyAz(Fxy & Fxz => y=z), i.e., that for every set of "arguments" the "value" is unique. With that, you can use either notation to do mathematics. But function notation is much simpler; compare (x+y)*z = (x*z)+(y*z) with (Sxyw & Pxzv & Pyzu => (Pwzx1 <=> Swvx1)). Of course, allowing functions into the language as primitives opens the path for more quick ways to get undecidable sentences.