Date: Thu, 20 Nov 1997 13:35:29 -0500 (EST) Message-Id: <199711201835.NAA12108@locke.ccil.org> Reply-To: bob@rattlesnake.com Sender: Lojban list From: bob@MEGALITH.RATTLESNAKE.COM Subject: How {lo} works X-To: lojban@cuvmb.cc.columbia.edu, bob@rattlesnake.com To: John Cowan In-Reply-To: <199711191923.LAA26950@fern.math.ucla.edu> (message from Jim Carter on Wed, 19 Nov 1997 11:23:57 -0800) X-Mozilla-Status: 0001 Content-Length: 4958 X-From-Space-Date: Thu Nov 20 13:35:39 1997 X-From-Space-Address: LOJBAN@CUVMB.CC.COLUMBIA.EDU The {lo} issue is not one of scope, at least not directly. It is a question of how {lo} works. I have been working under the rubric that `what is not forbidden is permitted'; and come up with one way that fits the requirements for a {lo} expression standing alone that causes disagreeable results in larger utterances. Others have been arguing under the rubric that `Lojban is a logical language and {lo} should be restricted so that logic operates in a simple rather than a complicated way'. If {lo} operates as you and most others presume it does, then the scoping is clearly as you and others have said and none of the funny things I am talking about happen. But I am suggesting that {lo} may operate differently, and legitimately so. John Cowan defined "veridical description" to mean "a description whose truthful applicability to its referent is *essential* to the truth-claim of the surrounding sentence". I agree. In turn, this means that {lo} presupposes a joint procedure between speaker and listener to discover what referent meets that criterion. Clearly, there are a great number of different procedures. I think we are all agreed on this. Next comes a proposal for one procedure that I don't think you will actively disagree with, although it may not a favored procedure: One way (not the only way) for {lo} to act is as an operator that returns true instances of the class, some number of them, one or more. Speaker and listener examine the instances and agree they are `true' instances of the class referred to. Thus, if I were to say: lo mlatu one or more of all the things which really are cats. we could look at one or more cats and test the truth of the expression by seeing whether we agreed they are truly cats. Now, consider my statement that: .i mi nelci lo mlatu This requires two tests. Firstly, whether {lo mlatu} are truly cats and, secondly, whether I like them. You can test whether I like {lo mlatu} by presenting me with some cats and seeing whether I like them. Now suppose I make the statement a second time, and you decide to test the predication a second time by presenting me with cats again to see whether I like them. Heer we come to the crunch: Are you going to present me with the same cats or with different cats? If you present me with the same cats as in the first test, then the outcome for global negation is as you and others state. If, on the other hand, you present me with different cats than from the first test, the outcome for global negation is potentially surprising. Let's walk through the procedure: Suppose you twice test the overall proposition that {.i mi nelci lo mlatu} by presenting me with two groups of cats and seeing whether I like them. * In both tests, you and I, speaker and listener, agree that the cats you present are really cats. * However, in the first case, I like the cats you present and in the second I do not. The first statement of the overall proposition is true, the second is not. How can this be? In the statement {.i mi nelci lo mlatu}, my liking is not an essential characteristic of true cats, it is an `accidental characteristic' that is not relevant to the veridicality test of the {lo} expression. So some true cats that you present to me may be liked by me; and others may not. Whether or not I like the cats does not determine whether the {lo} expression on its own is true, it only determines whether the overall expression is true. But now we are back to the puzzling consequence that *both* .i mi nelci lo mlatu and .i naku zo'u mi nelci lo mlatu can be true at the same time! If you reject this and say, "this cannot be the case", then you are forbidding certain procedures associated with {lo}, such as using different cats in the second test rather than the same cats. Incidentally, even with two groups of cats, it cannot that both .i mi viska lo mlatu and .i naku zo'u mi viska lo mlatu are true at the same time. This is because {viska} is concerned with a different kind of property than {nelci}, with whether even one cat exists in the visual field rather than with whether a cat is liked by me. So we have a `permissive' understanding of what procedures people could agree on for truth checking and a `contrained' understanding. Now, I have to agree that the `permissive' understanding makes for a more complicated world than the `contrained' understanding. On the other hand, I don't see what is intrinsic to {lo} that requires such constraint. It seems to me that {lo} is and should be simply a gadri that says that `in this conversation, {lo} indicates "a description whose truthful applicability to its referent is *essential* to the truth-claim of the surrounding sentence". -- Robert J. Chassell bob@rattlesnake.com 25 Rattlesnake Mountain Road bob@ai.mit.edu Stockbridge, MA 01262-0693 USA (413) 298-4725