From arosta@uclan.ac.uk Mon Oct 29 07:16:28 2001 Return-Path: X-Sender: arosta@uclan.ac.uk X-Apparently-To: lojban@yahoogroups.com Received: (EGP: mail-8_0_0_1); 29 Oct 2001 15:16:28 -0000 Received: (qmail 40788 invoked from network); 29 Oct 2001 15:16:27 -0000 Received: from unknown (10.1.10.27) by 10.1.1.224 with QMQP; 29 Oct 2001 15:16:27 -0000 Received: from unknown (HELO com1.uclan.ac.uk) (193.61.255.3) by mta2 with SMTP; 29 Oct 2001 15:16:26 -0000 Received: from gwise-gw1.uclan.ac.uk by com1.uclan.ac.uk with SMTP (Mailer); Mon, 29 Oct 2001 14:53:01 +0000 Received: from DI1-Message_Server by gwise-gw1.uclan.ac.uk with Novell_GroupWise; Mon, 29 Oct 2001 15:27:28 +0000 Message-Id: X-Mailer: Novell GroupWise 5.5.2 Date: Mon, 29 Oct 2001 15:27:17 +0000 To: pycyn , lojban Subject: Re: [lojban] a construal of lo'e & le'e Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII Content-Transfer-Encoding: quoted-printable Content-Disposition: inline From: And Rosta X-Yahoo-Profile: andjamin X-Yahoo-Message-Num: 11733 pc: #a.rosta@dtn.ntl.com writes: #> It's not an accidental error: his idea was to use the otherwise apparent= ly #> useless lo'e to solve an unsolved problem. I don't think lo'e is useless= ; I #> now think it is vital, but I think Jorge's usage still works. #>=20 #> The problem with {tu'a lo broda} is that it hasn't been established whic= h #> bridi's prenex the lo quantifier is in: is it in the bridi that tu'a lo #> broda is in, or is it in the imaginary bridi that would replace tu'a lo #> broda? Only the latter fixes the problem, but it (usefully) turns tu'a i= nto=20 #> an #> exception to the usual quantifier scope rules. # #I know it was deliberate, but I think it an error nonetheless. By that us= age=20 #what you are said to want is a typical whatever and so an atypical one wou= ld=20 #not do, presumably.=20=20 No. Jorge's usage was intended to be canonical/defining, replacing the meaning that has to do with typicality. #But, of course, an atypical one that satisfied your need=20 #or... would do fine: you any old whatever. As for the rest of it, the who= le=20 #point of {tu'a} is to prevent the quantifier on -- or one derived from -- = the=20 #sumti from rising to the upper level, so of course the quantifier goes on = the=20 #{le nu ... co'e} that {tu'a} indicates. Where has this been doubted?=20=20 I don't remember this ever having been the whole or even the partial point of tu'a. Tu'a marks so-called sumti-raising, and I don't recall the=20 quantification either having been settled or much discussed. My memory may be unreliable, though. #It thus is not an exception to the quantifier rules and, insofar as it app= ears=20 #to be, it is there precisely to remind you not to pull quantifiers out of= =20 #intensional contexts. It's a good exception, but you can't escape it being an exception, given the rule that sumti quantified in situ are interpreted as being quantified in the prenex of the localmost syntactic bridi. #<> OTOH , the claim is only true of 1 #> really and an abbreviation might well be useful for resolving some #> ambiguities efficiently (Swedes eat more yogurt than Danes). # #I'd resolve this as loi versus lo'e.> # #The sentence is said to be five ways ambiguous, at least three of which ar= e=20 #pretty obvious. {lo'e} seems to enter into at least two of them, since=20 #Lojban has lost (I just noticed) the distinction between typical and avera= ge,=20 #God knows when or why. Maybe there is another handy way to say "per capit= a"=20 #in these contexts. Come on, you chide others for saying "Lojban can't say X". Just make the appropriate lujvo. # # #Well, it will fail -- at least to be useful -- if it cannot be given some= =20 #meaningful content. Historically, it has been used as a magic wand to cov= er=20 #cases that could not be made to fit otherwise=20=20 you're talking about {lo'e}, I take it. Or prototype theory? #(this is not necessarily a=20 #strong criticism, since most theories in linguistics suffer from this prob= lem=20 #to a greater or less extent -- galloping adhocitis is a professional=20 #disease). So, given a class, even a natural kind, what is its category or= =20 #its prototype or its categorical individual? Is it a blueprint or a membe= r=20 #of the class or a way of talking about the class fuzzily or..... If we ar= e=20 #going to summon this thing, we need some clues about what it is. Note tha= t=20 #none of these things lives in Africa -- except perhaps a member of the gro= up.=20 # What properties does it really have? How are they relevant to properties= of=20 #members of the group? The main answer to all of this is that lo'e and its implementation of proto= type theory's categorial individual (=3D 'prototype') makes available an alterna= tive=20 ontology. Users who find that ontology useful can use it; those who don't needn't. All pretty Lojbanic. To answer your "Is it a blueprint or a member of the class or a way of talk= ing=20 about the class fuzzily", those are indeed answers given by weak forms of prototype theory, but the strong form is best understood as saying that either there are no categories, only individuals, or that everything is a category. Either way, the traditional member-of relation is replaced by the version-of relation. #<). The lV'e version implies a fictive #> element which is presumably not only irrelevant but flat wrong. # #I don't think it should imply a fictive element. It should imply only #an ontology consistent with prototype theory, so that instead of #Category and Member-of we have Individual and Version-of.> # #Unfortunately, the ontology of Lojban (yes, it has one) has sets and=20 #members, not individuals and versions, except in the st-worm-segment sense= .=20=20 This needn't be so. Nothing prevents Lojban from affording users alternativ= e ontologies. #And it is not clear what the version-ofs (or maybe it is the individuals) = would=20 #be like.=20 Personally I think it's easier to grock what the versions are like than wha= t the prototype is like: I find it easier to conceptualize a version of pc th= an to conceptualize the lion prototype, say. # I suspect that his is Mr. Rabbit come round again, and that has always=20 #failed to gain adherents precisely because it is too muddled to convince=20 #anyone=20 True. But it does have adherents among cognitive scientists, and Lojban oughtn't to be too ontologically prescriptive. #(or else it is some version of Platonism -- come to think of it, of=20 #course it is=20 I think it isn't what I would think of as Platonism. For example I can see touch and smell the pc prototype, so prototypes aren't inherently abstract (-- I understand Platonic categories to be inherently abstract). #-- in which case the question of connection -- which may not be=20 #important in the present sense -- remains unsolved,=20 I don't know what that question is. #and of course means that=20 #the individual has properties that none of its versions has and conversely= ).=20=20 I think this is held (by prototype theorists) to not be the case. I would f= avour going along with that view in the interpretation of {lo'e broda}, and takin= g {lo'e/tu'o du'u ce'u broda} to be the Platonic essence. # # #Well, ignoring 2500 years ago and ever since, but almost always so muddled= as=20 #to be useless. Given a is a member of A, represented (however that is=20 #intended) by *a, what is the reelation between properties of *a, of a, of= =20 #each member of A? That is the problem with {lo'e, le'e} and introducing *= a=20 #does not seem to get it any forrader, since the answers there are also=20 #unknown (I actually don't think the answers for {lo'e} are that obscure, j= ust=20 #very personal, but that is another matter).=20=20 I don't want to debate the ontology itself. I just want to argue that it is= important enough that Lojban should have a way of expressing it, and furthermore that it's useful if it is expressed by a gadri. # #I don't think there is one, since they attach always to unique objects. T= hat=20 #is, I think that, grammar aside -- but not really even that given their=20 #restrictions -- they are not gadri. Apparently KOhA gets quantified, not just LE, so (without my knowing the formal grammar) it seems plausible that the grammar allow "li" and "me'o" to be quantified. However, my ulterior question is: What should one do when the grammar requires a gadri but, as with "li" and "me'o", the sumti attaches always to unique objects? Answers so far are "tu'o" and "lo'e". #<> Maybe as a maxim of prudence, but it is always better to figure out #> what you really mean and say that, rather than just reduce your #> chances of saying something glaringly false or stupid. # #Not always easy to figure out, though. "The customer is always right", #"The postman misdelivered our mail (often/yesterday)"... Having some #foolproof recipes is useful.> # #But a foolproof recipe always gets it right, not merely usually keeps from= =20 #getting it hideously wrong. So stick with what you know is right, even if= =20 #there is something righter that you miss. (The first examle is directive = not=20 #descriptive, so less a problem, the second just suffers from time scope=20 #ambiguities, which Lojban ought to be able to do elegantly.) Those examples, on certain construals, take us into an imaginary realm where there is only one customer and only one postman, and situations in other realms where there are many customers and many postmen are versions of the imaginary one. To me this is not an unnatural mode of thought, and it is desirable that Lojban be able to express it relatively effortlessly. --And.