[YG Conlang Archives] > [jboske group] > messages [Date Index] [Thread Index] >
| In a message dated 10/3/2002 4:23:00 PM Central Daylight Time, jjllambias@hidden.email writes: << I think it would be useful to work out how we can make these >> xod to the contrary not withstanding, I think that there are several different things going on here and that we need 1) to keep them separate and 2) to figure out how to say them in Lojban -- preferably (as always) with the resources we have, but, that failing, with judicious additions. The first distinction relevant right now is that among truth values and truth functions and comments on certainty or success (or failure), which could easily get mixed up in xorxes discussion above. Truth values: assignments to sentences of values from a fixed range {T,F}, [0,1] or even (0,1), fuzzy or not. The assignment is meant to reflect some feature of the event described by the sentence: whether it occurs or not and -- with other than {T,F} values anyhow -- some relative notion of this (non)occurrence of the event with other possible ones in regard to some further notion (traditionally, for fuzzy values, the value of the membership function on the arguments for the specifed set -- and here the relation is usually direct: fuzzy identity). A truth value assignment to a sentence is metalinguistic, not a part of the sentence -- or the event the sentence describes -- but a separate event and described by a separate sentence (in Lojban, officially number + ni + original sentence, but variations are possible). In particular, it is not a modifier within a sentence leaving the original sentence. [Note: just to add to the confusion, the claim of a truth value for a sentence is also a claim, and thus has a truth value, and so on ad inf -- another source of fuzzitude.] Truth functions. These are modifiers within a sentence, setting up different truth value assignments systematically related to the basic ones. Thus, using just bivalent cases for the moment, the function "very" applied to "is tall" shifts the cross-over zone of the assignment rightward: if the zone was around 5'10" for "tall" (for American men), that for "very tall" might be around 6'5" and "extremely tall" probably above 7'. (I suspect these can be related roughly to standard deviations in this case, but that is just a handy correlation for this case and would not work -- nor make sense even -- in some other cases). Comments. These also go into the sentence itself but are not so systematically related to truth values. Something like "barely" might be viewed as a function but then as a very different one: low on both ends (i.e., false in a bivalent system), high (true) in the middle (where the original passes the cross-over zone). I could equally well, however, be taken as a comment outside the truth assignment itself, allowing that the sentence passes the truth test, but commenting on how near a thing it was. In the multivalent cases, fuzzy or not, it becomes a comment on how close to the lower limit of designated/acceptable the value will lie, without itself being evaluated in the original -- a metalinguistc comment of a sort, then: in the sentence but not of it. Obviously, the line between the latter two types is not perfectly solid and opens some possibilites for confusion. Lojban presumably will find ways of doing them distinctly. Even the line from first to last is open to some leakage, since, if we can metalingusitically comment that the truth value is closs to the cross-over zone, we should be able to comment metalinguistically that the truth assignment is just such-and-such. I would take it that xorxes' {ja'acai} etc. are of the class of comments, with "barely". I am less sure where to put {ja'axipiPA}. They look like more precise comments, but are translated as functions, like "very." Since I think that functions should be represented in the selbri, not off to the side, I will take them too as comments, though a good translation is not always obvious. |