WikiDiscuss

Let me stress that my comments were directed at the use of {nitcu lo} and {djica lo} in place of {nitcu tu'a lo} and {djica tu'a lo} as stock examples, not at other problems with the new {lo}. As a solution to the opaque context "problem" in Lojban, the new {lo} is both inadequate and ineffective. It is inadequate because it does not "solve" the "problem" for any type of sumti other than one introduced by {lo}, but the same problem occurs with every other type of compound sumti (LO + (bridi)) or names. It is ineffective because the problem persists: the inference from the new {lo bridi} in situ to {su'o da poi (bridi) go'i ... da...} works with the new {lo} as well as with the old — lacking the blocking action of {tu'a}.

It is also superfluous since Lojban does not have a problem with opaque contexts; {tu'a} and the abstract phrases it covers deal with these contexts in an exemplary fashion. The problem is with Lojbanists who refuse to use these devices, either because they have not taken the time to understand them or just persist in following their home language habit of not making the Lojban distinction — often even claiming that that habit rather than the Lojban (logical) format is the correct view of reality.

To be sure, we could ignore opaque contexts if we were willing to make some modifications to Lojban. But I think the modifications are much worse than just learning to use the language as she is given. We could, for example, insist that only existent objects culd be needed, wanted, dreamed about and so on. But this would limit our ability to plan and invent and countless other things we do — and want to — do. Unless, of course, we quantify not over existents but over non-existents (even impossibles) as well. But then we have to introduce the distinction between existents and non- explicitly, since everything (in the outer domain sense) is available for quantification, but not all are available for acting upon. Alternatively (or additionally, for that matter) we could deny that generalization was an inference that Lojban supported and take it that the logic it copied was a free logic (parallelling but slightly different from the view that "every" does not have existential
import). Along a different line, we might insist that {nitcu} and {djica} and an indefinite number of other predicates have (unmarked) opaque places, places which interfere with generalization (and Leibniz's law) and then just learn to recognize those places when the need arises (this is, of course, what natural languages do for the most part) and drop the need for explicit markers on ordinary places. This solution has regularly been rejected, even by people who act as though it were in place.

On the larger issue — which I did not mention except in passing — of the new definition of {lo} in general, I have said my say on that many times. Briefly, the definition is simply muddled since it tries to deal with a number of separate (and already fairly well separated in Lojban) concepts in a single device. The result is that the whole is at least incoherent and, I think I have shown, inconsistent. It does point to a couple of possible additions for Lojban: a way of talking about kinds separate from talking about the collective of its members or unspecified members or (stereo)typical members and a way of talking about stuffsubstancegoo. Note that, insofar as the new {lo} does either of these things, it fails in its other purposes, since neither the kind nor the kind-stuff does what an unspecified individual does. As for "generic" individuals, I have trouble seeing them as other than either the genus itself or a concrete individual without specifying which one, a
particular quantifier — with short scope, perhaps. The notion of Mr. Whatsis — at least the one that turns up most often — as a single object doing whatever any whatsis does really is contradictory and turns out to be nothing more than minimum-scoped quantifiers loosely disguised.

Ahah, another possibility for {lo}: it is just such a minimum-scoped quantifier and thus different from {su'o} which is, presumably, capable of being fronted. {lo} would then not be capable of being fronted and would — by itself — block generalization. Further, it would not be anaphorizable, since that would illegitimately extend its scope. (Actually, it would be anaphorizable with pronouns that repeat the phrase but not the referent of the phrase — as happens in English, for example). That might be a useful change to make, but I suspect that even it would go against much previous usage.

The possibility that {lo} really does refer to the genus — as opposed to the collective or the set or the members taken separately — might also work, but it would require a rewriting of the semantics of virtually every place since most places have been taken to deal with members of the genus, not the genus itself. And then the rewrites have to be adjusted to account for sumti not introduced by {lo} (or all the other gadri have to be redefined as well).

Frankly, I think the best solution to all the problems that the new {lo} is meant to solve is just learning to deal with Lojban itself, not Lojanized English or Spanish or whatever.

pc