| In a recent private exchange, rob called me to task for my use of {vo'a} for subjunctives and directed me to the places in the Wiki maze where discussions of the issues were to be found. In particular, he recommended {mu'ei} as the right way to go. I strongly diagree. {mu'ei} makes explicit reference to possible world in the object language. Possible world are a useful matalanguage device for explicating some notion in the object langauge, but, when introduced into the object language, they tend to create more problems than they solve (identifying worlds -- and things in them -- or (perhaps worse) distinguishing them (Prior's "The non-diversity fo the non-existent"), plotting connections, etc. etc.), not to mention the metaphysical freight they carry with them. Subjunctives-- and whatever else possible worlds are meant to do -- can be done perfectly well without possible worlds, as witness the fact that they are handled in all natural languages without once resorting to possible worlds. (The nearest thing to a successful use of possible worlds in an object langauge is the sentence labeling within "proof" in modal Dialogic, where the whole is merely a convenient device and exists only in proofs, never in sentences outside the proof context; sentences have only possibility and necessity operators. Happily there now are ways to get around even this use in almost every case.) Without possible worlds, what resources are left -- what do natural languages use? For the most part (and without making an actual survey -- other patterns would be welcomed), four patterns seem to predominate: tense, condition, causation, and imagination (well, some mental activity). Naturally, I prefer tense, but it is so SAE that I did not want to throw it to the groups that thinks that anything SAE is ipso facto not Lojbanic. Lojban has at least considered the other three: {va'o}, {bai} and {da'i} (and has put {mu'ei} in a tense-related selma'o}. Lojban is, of course, totally inspecific about the nature of time, but, if we wanted to do a metalinguistic explication of tense structure, we would almost certainly use one with linear past and branching futures. At least, the future would branch at kairoi,crucial events. The tense subjunctives tend to go back to such a kairos and then leave it along another temporal path, defined by a different outcome at the crisis, and come the same distance forward on that path as back on the other. From the viewpoint of this alternate now, the different world can be viewed. (The forms used for this are usually obsolete ones, so subjunctives are good digging ground for the history of tense systems.) This system then gets expanded to the general case (as do all of these) of any contrary-to-fact or hypothetical situation, even if it is not temporally connected to the present or a past kairos. So, {puba} would not make a goodcandidate for subjunctives in general, though something related to it might have the advantage of cutting off a particularly useless kind of speculation, the "If Socrates were a 17th century Irish washerwoman" sort, where all connections with anything recognizable are severed, making defensible continuations impossible (without a lot of assumptions not mentioned in the condition). Technically, a tense system subjunctive system needs two tenses: an establishing one for the kairos' other outcome, and one for the alternate now; the rest are normal tenses around these two. And, of course, the alternate now need not appear when it is clearly linked in with the kairos form nor in the subsequent sentences. But a comback form is needed to get out of the subjunctive situation (presumably the one for complex tense situations would do). Tense are obviously most natural for contemplating what might have happened if, HIBK and the like. They work literally only if we allow going back even to the Big Bang (the ultimate kairos) and coming out along alternate settings of the physical constants, say. The causal version has its natural home in general laws and their applications but is then expanded to cover cases where causality plays less to no role -- that can be used anyhow. "If I dropped this pencil it would accelerate pretty damned close to g" is natural for it; "If Socrates had escaped from prison, he would never have been famous" works with the loose notion of causality or law that holds in the social sciences; "If Scrates were a 17th century Irish washerwoman,..." is straining the idiom a bit, because no laws are apparently to be applicable (either Socrates or the woman's situation seem irrelevant). As with the tense case, we would need a sign that brings us back from the hypothetical realm to the real -- or, perhaps more naturally, a device that indicates that the hypothesis is still in play, since causal forms are typically one-sentence idioms (sumti of a causal predicate or BAI and a clause) or connectives that only connect two sentences, not easily more except in a cusal sequence: this and so this and so this. And, of course, the claim that this would happen if that had is a separate one from a strictly causal claim (in many cases); it may be true even if the causal claim is false. The choice of {bai} for this seems to me particularly unwise, since, while it is a very general causal notion, it is also a very strong one, inappropriate from many subjunctive uses, where only plausibility or, at best, probability are at play. The mental model has somewhat the opposite, supposing knows no limits from the real world -- we can suppose they did not apply. To be sure, except when one is composing fantasy for entertainment, there is usually a practical, this-world, point and so we cannot dismiss all constraints. But the at-home use of imaginational subjunctives is more open to the tests of plausibility and the like than to laws. And it lends itself more naturally to developing hypotheticals, adding new conditions on as the discussion moves along. It is ideal for brainstorming, then, but can, of course, be adapted to other patterns. While I like the framing {da'i} - {da'inai} for contained uses like indirect proof, they are a useful pattern for subjunctives of the imaginary sort as well. Every subjunctive starts by setting up a condition, around which the rest of the subjunctive discussion revolves. Thus, a device that introduces the condition is a natural beginning of any subjunctive. But that condition does not have to be temporal or causal or even imaginary (though not holding at the moment), so a general "what happens if this condition obtains?" seems the most complete basic subjunctive form, assuming there are ways to continue it and get away from the hypothesis generally. This is why I favored {va'o} for this purpose. I wasn't stuck on it, but it seemed to move in the right direction more literally than the other choices. On the other hand, I now appreciate the factual uses of {va'o} and would not like to lose them just to get a subjunctive. So, I think that, if we want a general subjunctive, something like {va'o} is the way to go. On the other hand, I am not sure we want a general subjunctive. As I have mentioned in the above paragraphs, there seem to be a number of uses for the subjunctive, similar in some ways, differing in others. And especially differing in what is going to constitute a true subjunctive claim, something that a logical language ought to take quite seriously. That thought seems to lead to sugtgestion that we open up a variety of subjunctives -- at least historical, causal and speculative -- and deal with them separately. It may turn out that not all of them need subjunctives at all (causal ones do seem to be pretty easily converted to indicatives, some would say), but for those that do, we can devise a starter (the condition), a continuer, and an ender. |