[YG Conlang Archives] > [jboske group] > messages [Date Index] [Thread Index] >
| In a message dated 10/9/2002 11:48:31 AM Central Daylight Time, xod@hidden.email writes: << Conceived to provide a quantitative counterpart to the qualitative ka, it >> Quite aside from the unjustified loaded langauge, this etiology is inaccurate in several ways, if indeed it is meant to give the history. As I have explained moe times than I can count off hand, the confusion (if there is one) about {ka} has to do with the English _expression_ "the quality of." but that aside, the worst matter here is the confusion of {ni} and {jei}. However obscure {ni} may be, {jei} is not at all. {jei} gives the evaluation of a proposition or sentence under a given assignment of truth values -- including a given number of possible truth vaues to be assigned. It has not direct connection with issues like the amount of redness (in either sense -- I suspect another confusion here), since these are not issues about sentences or propositions at all. The nearest they might come to such a connection would be if the amount of redness were what made for truth, so that truth values to were assigned onthings like {ko'a broda} on the basis of the amount of broda ko'a had (in some sense or other). But then, of course, the amount of broda and the truth value of {ko'a broda} are two different things -- because they are correlated as well as because they apply to different types of objects. And have different ranges. << Up the present, my discussion with pc on this had two outcomes: one was the question of whether jei ko'a xunre is actually an analog of the redness of ko'a. It's an interesting question but I don't see why a language that actually intended to be used would offer any other interpretation of jei. In any case it's an interpretive convention, and that's how I use it, and I don't know of any contrary usage. >> What was the outcome? I don't even understand this formulation of the question, nor, indeed, what xod's convention is. I think he means that the truth value of a claim is proportional to the degree to which the property mentioned in the claim applies to the objects mentioned in the claim, which has to be roughly correct: more gets a higher truth value, if one is avaliable, less a lower. But that doesn't seem conventional. What is conventional is 1) the number of truth values available, 2) the measure of the degree of application, 3) the exact relation between degrees and truth values (how the truth values partition the range of degrees). << And the other outcome was: does ka have any meaning without any ce'u; when the ce'u place is filled with a sumti, and ka's intended function is to abstract away some completely subjective, totally specified opinion that the observer has concerning the bridi in question. The concrete example used was that "ka la godziras. cu cadzu" could refer to the "earthshaking quality" that might be experienced should one be standing close to the event. Of course, it could also refer to just about any other impression or description of the event. (Now it's beginning to remind me of su'u lacking an x2!) Anyway, that ce'u-less ka was smacked down about a year ago, and such subjective impressions now need to come from li'i. Let the skeptics who have forgotten the sound beating that ce'u-less ka received enjoy the fine archives >> Well, this is not much to do with {ni} and not particularly accurate after that. {ce'u}-less {ka} seems to be alive and well, despite all the objections raised to it last time round. Indeed, what got smacked down last time was any attempt to regularize the thr relation between {ka} and {du'u}. As for {ka} with all its places filled (which was only touched on in the last go-round), nothing requires (or even suggests) that the properties described in that usage be subjective: earth shaking, for example, is clearly objective-- everyone can see it, instruments can record it, and so on. (li'i} does not seem to provide a meaningful alternative for many of these cases (to be sure, we might say that earthhaking was both a property of Godzilla's walking and a major part of my expreience of it). lojbab: << Now it happens that colors often are often defined such that truth value overlaps with quantity measurement. But most other concepts do not mix purely, and the measure of X-ness is usually NOT the same as "the truth value of (ko'a Xs)". To argue about "ni" solely on the basis of color seems to intentionally limit the concept. >> Yes, colors are not good for generalization BUT quantity measure does not overlap with truth value even in that case. Even the early fuzzy logics, which took the truth value of "x is red" to be identical numerically with the value of the membership function applied to x and the set of red things, kept the two distinct as objects. xod on lojbab: << > jei ko'a xunre du pimuvo > means that the truth value of "X is red" is .54 on some kind of fuzzy logic > scale. In other words, it is 54% true that ko'a is red. This may or may > not mean that it is a color blend which is 54% red and 46% na'e > xunre. That would constrain fuzzy logic usage too much, IMO. I'd like a concrete example showing why this is broken. >> The quickest is to take the usual bivalent logic, where the {ni} is .54 and the {jei} is 1 (or, maybe 0, but certainly not .54, a value not even available.) << > ni ko'a xunre du pimuvo > means that there is X's redness is measured as .54 units on some scale, > which may or may not be a unitary one. You take ni to refer to the first sumti. What about ni ko'a ko'e broda? >> {ni}, like {ka} in the comparable usage, refers not to any particular part of the event (event, note, not claim) but to the whole. lojbab: << . That would constrain fuzzy logic usage too much, IMO. > >I'd like a concrete example showing why this is broken. I'd really need one of the fuzzy logic people to do that, like Belknap. But moving the discussion away from colors would help. >> Yes, it would. But I don't see anything inherently fuzzy in all this discussion. Like early fuzzy logic, this is at best truth value system with an infinite number of values, all those in [0,1], say, and those behave quite normally. The fun comes when the numbers get fuzzy, too, and, for example, {P &~P} gets some fuzzy value other than 0. << >You take ni to refer to the first sumti. What about ni ko'a ko'e broda? I'll accept the need for ce'u in ni at least as much as in ka. (I'm not sure I like using ce'u in jei.) >> Notice that {ce'u} in {ka} clauses creates a very different critter from {ka} without {ce'u}. Whether the same would happen (and what it would be) with {ni ce'u} is an open question. One would expect {ni ce'u broda} to be a function from objects to quantities of the appropriate sort for brodas. Similarly, {jei ce'u broda} should be a function from sumti (linguistic expressions) to truth values: from {ko'a} to the truth value of {ko'a broda}. << But where is the ce'u in "ni ko'a xunre"? If you agree with me on ce'u, you agree that "ni ko'a xunre" is as bad as "ka ko'a xunre"; really bad. >> Non sequitur. {ni ko'a broda} has an established use in the langauge, while {ni ce'u broda} does not. The situation is almost the opposite with {ka}: {ka ce'u} is well defined, {ka ko'a}, while mentioned in CLL, is not well established. The two "new uses" arise equally out of the apparnt parallelism here. So, if {ni ce'u} is OK, then so is {ka ko'a}, by the same argument. << I certainly agree that ce'u in jei is an abomination. >> Not quite what lojbab said. But also not obviously worse that {ni ce'u}, though the justification would be different, since the parallelism with {ni} and {ka} does not hold. & << Nuel Belknap the fuzzy logician, or Stephen Belknap the lojbanist and physician and fan of fuzzy logic? >> Did Nuel do fuzzy logic? He was into so many things (relevance, questions, ...) that I wouldn't be surprised, but I can't remember a case. << > Conceived to provide a quantitative counterpart to the qualitative ka, Was it? How do we know this? >> It wasn't so we don't (if anything it goes tother way round -- {ka} with a full bridi was conceived as a qualitative counterpart to already existing {ni} with a full bridi). << I agree that the ni/jei distinction can be collapsed, but each degree of ni/jei must also be associated with a property indicating whether it counts as a true-making degree or a false-making degree. In other (& hopefully clearer) words, the possible values of ni/jei can be grouped into those yielding True, those yielding False and those yielding Sorta. >> I am sorry to hear that. Fortunately, you turn around and disclaim it in the next breath by distinguishing truth-making and false-making (and sorta-making) properties -- i.e. partitions of the values of {ni} based on the truth values {-, 0,+}, and so not collapsing at all, since {-, 0, +} are, in some disguise, values of {jei}. << But as far as I can see there's nothing ugly about ni per se. Rather, just as the ka/du'u distinction does not exist, so it can be argued that the ni/jei distinction can be dispensed with. We just end up with two redundant cmavo. >> Note that the {du'u}/{ka} distinction is still maintained in the controversial case, since {ka} with a full bridi is not a proposition, as {du'u} with a full bridi is. But even without this separation, {ni} does not collapse to {jei} or conversely -- as the distinctions you make above show. xod: << What do you mean? jei, at least, really is limited to [0, 1]. ni is unlimited, which makes it hard (but not impossible!!) to represent truths. >> Correct! And from xod at that! Although I take it that his point is probably not that {jei} and {ni} are the same, but that only {jei} has any use, so {ni} should be done away with altogether. << At least ka means a special case of du'u; one with su'o zo ce'u. ni doesn't even offer us that much. And it's interpreted in all crazy ways: if I were to describe real usage, I'd have to admit it's usually used to count xo ko'a! >> Probably true, but not relevant at this stage. The fact that we are not yet very good at using {ni} and often use it flat wrongly (though counting participants is sometimes th right thing to do, too) doesn't mean that it doesn't have a correct use that we can learn. |