Return-Path: Message-Id: <9205191938.AA08565@relay1.UU.NET> Date: Wed May 20 10:57:14 1992 Reply-To: Logical Language Group Sender: Lojban list From: Logical Language Group Subject: calling all nay sayers, we want opinions quick! X-To: lojban@cuvmb.cc.columbia.edu To: John Cowan Status: RO X-Status: X-From-Space-Date: Wed May 20 10:57:14 1992 X-From-Space-Address: cbmvax!uunet!cuvma.bitnet!LOJBAN How about a controversial proposal? It must be decided quickly, since it affects stuff going into JL. It is controversial for among other reasons because it affects a great deal of previous Lojban text, and requires relearning among people like me who have learned to speak Lojban. Obviously, there has to be considerable virtue to such a proposal for it to even be seriously considered. I am of mixed minds in this regard. I am presenting what I think is the strongest case for change below. The primary argument AGAINST change is that it is a change in well-known usages. A secondary argument might mean that a change could create constructs of obscure semantics - but we have a lot of these in the language %^) The question is whether "nai" should continue to serve the mixed purpose of indicating contrary or scalar negation in some constructions, and contradictory negation in other constructions. The most well-known such constructions are, for contradictory negation, the logical connectives. For example, in "ganai" means that the following sentence or sumti is contradictorily negated before evaluating truth conditions. Non-logical connectives are also contradictory when modifed by "nai": joinai means that it is not true that the two components are conceptually mixed or massified in evaluating the meaning; it does not necessarily imply that some different connective might apply, as a scalar negation might. On the other hand, the attitudinals modified by -nai produce a contrary negation, along a fixed scale defined for each attitudinal. .oinai is the scalar negation of complaint, which can be comfort, or pleasure. It does not merely mean the absence of complaint. No one has had any trouble keeping these two most familiar usages of -nai straight. The problem comes in the more obscure usages. Each such usage was defined in the negation paper as having either a scalar or a contradictory meaning. Most of these are fairly intuitive IF you memorize what one particular example usage means. Thus "ki'unai" is a contradictory negation meaning "in spite of reason", as compared to "ki'u" which means "because of reason". "punai" has been defined since time immemorial as claiming only that the relationship modifed is not true in the past, and does not claim that it is true in the present-or-future, hence it is a contradictory negation. "ru'inai", also found within tense structures, in contrary. Whereas "ru'i" means "continuously", "ru'inai" means "discontinuously", or "intermittently". It thus claims more than the absence of continuity, it claims that the relationship holds in a >different< manner than continuously. But the clean division of the negation paper has undergone a lot of stress, as we've rigorously analyzed other components of the language. No one remembers, for example, what "nai" means on abstractors, because no one has ever used these in actual usage. There are only two theoretical usages that have been derived to even justify having negation on abstractors, and they will probably need to be taught as idiom before people learn them well enough to try productive new uses of the construct. The analysis can get obscure. "paroi" means "exactly once", as a measure of occurances of an event. Does "paroinai" mean some number other-than-exactly-once, one version of a scalar negation, or does it mean "never" = "not once". If a contradictory negation, it turns out to probably be identical to the former, since it claims only that it is false that exactly one instance is the count - it might be two, three, or none, but it is false that "once" applies. Thus it is inclonclusive whether "nai" in this instance is contradictory or contrary unless we define the contrary to mean the same as English "not once" = "never". More significantly, and the proximate cause of this proposal, are the FAhA space directions. "fa'a" means towards a specific point". Does "fa'anai" mean "directly away from the point", "other than towards the point (maybe veering off to the left a little)", or simply that it is false that "towards the point" applies (which might include that the event is not happening at all, since contradictory negation denies the claim as made rather than directly making a different claim)? Originally, when we created the set of words, the first meaning was used. However, John Cowan pointed out that "directly away from" assumed not merely a contrary negation, but a polar opposite - a direct antonym. Thus we added a new word for "directly away from", and I assumed the middle meaning. However now, as a result of writing the detailed analysis in the tense paper, John has observed that "FAhA" as a space direction is exactly analogous to "PU" as a time direction. If "punai" means merely that a claim that an event occurred in the past is false (while making no claim about the present or future) it would be extremely inconsistent to assume that "fa'anai" meant anything more than that the claim of "towards the point" is false. It has been suggested several times that we change "punai" to mean "other than in the past", a scalar negation. The arguments against this were several. First, it is a direct contradiction of a usage that JCB has taught for Loglan since he first introduced tense constructs. His classic example of a complex tense, the equivalent of Lojban "pujecanaijeba" (in the past, and not in the present, and in the future" is a contradictory formulation useful in the phrase "The Once and Future King". We've never overridden JCB unless we had a damned good reason. The best reason that we can come up with is this inconsistency issue. JCB never considered this because he never recognized the distinction between contrary and contradictory negation, even though the distinction is fundamental to logic. Given that JCB's classic example can be expressed just fine by tranferring the contradictory negations to the logical connectives "pujenaicajeba, there is no loss of expressivity in making punai contrary. (But note that you have to be careful with logical connectives, which are left grouping - and there is no way to change that grouping inside a tense construct: pujecanajeba means "not true (in the past and the present) and true in the future", which does not rule out it being true in the past and the fututre but not the present.) The current Lojban grammar almost makes the problem go away. When used as a tense on a selbri, "na'e pu" gives the scalar negation, and "na pu" gives a contradictory negation. But these negations apply to entire tense constructs, and this may not be what you want. "na'e puparoi" means "other than (exactly one time in the past)", but "punaiparoi" means "na puparoi" (false that it occurs exactly one time in the past) and not "exactly one time other than in the past" - the latter seems only to be expressible by the compound tense "na'epujoiparoi". The intricacy of the rules in this area of the tense grammar will discourage people from creatively adding to the set of Lojban tenses, for fear they will misanalyze, and there may be some tenses that are simply impossible to express because some interaction of rules makes it impossible to put a contrary negation on exactly the piece that you want. A counter argument is that, used as a sumti tcita, you can have "na'epu" but you cannot have "na pu" - it would fail the LALR1 rules to permit this construct. Thus there is no way to express contradictory "not before ..." except with a contradictory interpretation of "punai", while a contrary interpretation would give two means to say the same thing. By the same token, of course, the contrary interpretation of "fa'anai" can be expressed with "na'efa'a", but there is no other way to express the contradiction of "fa'a" as a sumti tcita. There are basically two choices that seem viable. One is the status quo, which works well enough, with the nooks and crannies somewhat difficult to teach, but all the basic expressions of the language stable and the negation paper essential accurate as written (if incomplete in documenting some of the choices and their implications). The other, which Nora suggested lightly before realizing the implications, is the subject of this message. This alternative is to introduce a second word to selma'o NAI. For relearning considerations, the word must be fairly close to "nai", and I therefore am suggesting "nei", which is a thus-far-almost-never-used word in GOhA for the current bridi; "nei" has also been changed once before, and probably no one except Beta test users of LogFlash 3 is likely to have learned it. It would be given some other cmavo, probably a CV'V form. I would recommend that contradictory negation continue to use "nai", and that scalar negation would use "nei", since nai has a similarity to na and nei has the 'e' from "na'e". Thus, contrary negations would use "nei". "fa'anei" would mean "some other direction than towards", "fa'anai" would mean "false that it is towards". "punai" would mean that it was not in the past, "punei" would mean that it occurred some time other than in the past - i.e. the present or future. The major impact is to attitudinals - ".oinai" would become ".oinei", and ".uinai", ".uinei". I would presume that ".uinai" would be denying happiness, which is not the same as expressing unhappiness , but doesn't rule it out either - it would not be the same as ".uicu'i" (no particular happiness) which rules out both happiness and unhappiness. Thus, even in the area most affected, "nei" would add some expressive power. Indeed it was this realization that led me to write this message. It is NOT clear, though, what "nei" might do in places that have hitherto been sacrosanct contradictory negation. What does "mi .anei do" mean? "Me and someone other than you", I would presume. On the other hand, tanru logical connection becomes a real loose end. In existing usage, we've realized that tanru connection is not currently subject to the rules of other logical negations - you can't necessarily break them up into separate sentences. How does the "nai"/"nei" distinction affect these constructs? I won't even try to guess. (John, care to tackle this?) So the pros are increased expressivity, including some identifiably useful forms, and some increase of regularity. The cons are that this is a change to a well-learned language feature, a real relearning problem that can be minimized but not eliminated, and that we may open some semantic nightmares that we might wish to avoid. The original problem gets solved, though I'm not necessarily recommending elimination of the possibly redundant na'epu form, which scalar negates a whole tense construct. Do you see value in a change, or let the status quo prevail at this late date? If you are already a Lojban user, will the relearning affect you more severely than the increased analyticity benefits you? Let's hear from the audience. Soon. Whether you are a nay-sayer, or a nei-sayer, I want your opinion. (Groan!) lojbab