sumti qualifiers

From Lojban
Jump to navigation Jump to search

The proposed interpretive convention is that every LAhE and every NAhE+BO is simply a short form of lo broda be for some suitable broda. The proposed broda are:

la'e = lo selsni be
lu'e = lo sinxa be
tu'a = lo datni be / lo jazyfau be / lo selkai be / lo datni ja jazyfau ja selkai be ???
lu'a = lo cmima be
lu'i = lo selcmi be
lu'o = lo gumselcmi be
vu'i = lo porselcmi be
na'ebo = lo drata be
to'ebo = lo dukti be
no'ebo = lo dutnu'i be
dutnu'i = x1 is neutral on a scale in which x2 is opposite to x3
je'abo = lo du be

The first two are given explicitly in CLL, with le instead of lo.

One virtue of this scheme is that issues of scope of quantifiers and connectives in the sumti complement are resolved in one swoop.

This conflicts, however, with some other understandings of LAhEs. For example, that lu'i ko'a e ko'e is the set that contains ko'a and ko'e only, or that lu'a lo mlatu is equivalent to lo mlatu, and other conceptions.


  • And Rosta:
    • I strongly endorse the general principle.
    • I won't comment on the particular definitions but we must be sure that they're not understood as "this LAhE means whatever this predicate means". That is, if, say, sinxa got its definition changed, that oughtn't to affect lu'e. I know that that's not your intention, but experience shows that it's a possible misinterpretation.
  • The equivalence LAhE = lo broda be should hold when the LAhE is bare (not preceded by a PA). When preceded by a PA, it should be equivalent to just broda be. Thus lo PA LAhE should = lo PA broda be and not lo PA lo broda be.
    • Good point. I think defining it as lo pa broda be, with XS inner pa, might be even better. Then lo PA LAhE will be lo PA lo pa broda be, which does reduce to lo PA broda be, as wanted, while we make sure that lu'a ko'a for example refers to a member and not a group of members of KOhA. In that case, the equivalence would always hold. pe'ipei?
      • Yes, your revised proposal is good - simple & robust.
  • pc:
    • This looks good in principle, but, as the last paragraph notes, has some difficulties in practice. Part of the problems is that, grammar --and even spelling to the contrary notwithstanding - these critters are not all the same logically. tu'a for example, is not a gadri qualifier at all - it creates an abstraction (usually event) description with all details except the selected sumti blanked out: maybe lo nu ... bu'a (this is all CLL, of course - I don't know what y'all have done to tu'a and your proposed replacements don't help much). That is, the "thing" it actually refers to is at least a step more abstract than the sets and groups of LAhE.
  • tu'a is sometimes lo nu ... co'e and sometimes le du'u ... co'e. (I hesitate between bu'a and co'e. Is tu'a specific or nonspecific? Does it refer to a certain relationship that the speaker has in mind, or just a claim that there is some relationship that will hold? The latter seems too vague to be useful.) Besides the bu'a / co'e issue, there is the nu / du'u ambiguity, which is normally not a problem because context will make sense with only one of the two. The two proposals I make for tu'a follow these two possibilities:
  • The last one provides an intensional context just as well as the nu version, since it is equivalent to:
  • Maybe se lifri is not the most general broda to replace nu. Maybe jai fasnu -> jazyfau? Also, I should probably throw selkai in too, as in:
  • I am inclined to think that tu'a is usually nonspecific, given the generality involved. Why settle for any old nail and yet require that it be present in some specific way? The vagueness that results is inherent in the usage; we focus on the object and don't care about the rest - including what kind of abstraction is technically involved (hiding technicalities is what makes language speakable in large part). So, as far as it goes your proposal is fine, but it needs refinement to meet the fact that it is a proposal about a number of superficially similar but logically quite different things. though the logical different may be little more than whether what is referred to is existents or possibilia.
    • pc:
      • I assume that this is XS lo, the kind (SS loi). I'm not sure that is where you want to end up in every case. lu'i, for example, seems to require lo'i, a set, and lu'a, the members, seems to be distributive rather than collective (CLL lo or, more likely le).
        • Yes, I should write lo pa with XS lo.
          • I'm not clear what you are clarifying here. Isn't XS lo pa broda one subkind of broda-kind? It still doesn't get the right sorts of things out, so far as I can see.
      • lo pa broda is "Mr Single Broda". I want lu'a ko'a to give "Mr Single Member of ko'a", so that quantification will be over real avatars of "Single Member of ko'a".
        • I think I don't have a handle on Mr. Single Broda, how it differs from Mr. Broda, in particular. Is this perhaps about singletons, so present wherever there is a set whose only member is a broda? I can't at the moment think of any use for this and I am pretty sure it would be too rare to justify using this compact form. So, I assume it means something else, but I don't know what. I do take it that the avatars of Mr. Single Broda are individual brodas (not their singletons), but I don't see how that is related to Mr. Single Broda -- in a way that contrasts with Mr Double Broda, say.
          • "I receive Mr One Newspaper at home every day" vs. "I receive Mr Two Newspapers at home every day", for example. "I receive Mr Newspaper at home every day" is more general.
            • Isn't that just one avatar vs two avatars vs some avatars? Jumping up to a different kind (not necessarily in any technical sense) to deal with this seems unnecessary and strange. The subdivisions of Mr. Newspaper would seem to be of the Mr. New-York-Times and the Mr. Four-Star-Final and the Mr. Of-January-14-2003 sort (and probably many others - my NYT is printed in Chcago, for example). And, of course, the intersections (as it were) of these. Not Newspaper-Monad and Newspaper-Dyad and so on.
              • You could go down to the level of avatars, of course. In that case, in Lojban you would have to move the "everyday" to the front, so that you don't claim that it is the same two avatars every day. lo re karni is more general. I agree that Monad vs Dyad is probably not among the most frequent subdivisions of Mr Newspaper. But it is one possible subdivision.
                • Well, as you have explained (or at least as I understand what you have said) Mr. Two-Newspapers is not a subdivision of Mr. Newspaper, all of which are presumably second run objects whereas Mr. Two-Newspapers is a third run object. Unless, of course, you mean something other than singletons by pa mei. It is the case that there are at level two a number of Mr.s that bear the Mr.-avatar relationship to exactly two newspapers (concrete copies on the doorstep). And, at the third level, there is a line that bears the Mr.-avatar relationship to exactly such second level things. And at the end of a possibly infinite process there is a Mr. Newspaper that bears the ancestral of the Mr. avatar relationship to all the newspapers and Mr. Whatever-Newspaper at any level at all. And it would probably turn out, if the Mr.-avatar relation were simply treating collectively, that for every Mr. in that hierarchy has there is a second round Mr. such that anything true of one is true of the other. But in that process the two-ness of third level Mr. Two-Newspaper is lost. That requires that the first step be a set step and that would not be a subdivision of that highest Mr. Newspaper. So, if this is really what you want rather than just two avatars, you need to explain moe what is going on. And how we get from the dyad to the two newspapers (lu'a?). We already know that tense information has to go out front with na, so that is not a problem.
    • pc:
      • In the case of moving in and out of signs, it is even less clear, since the same word purports to deal with denotation and designation and probably the other relations, too, so the appropriate output is going to depend a lot on the word that is the sign or the thing that is signed.
        • Those two at least have CLL pedigree.
          • True, alas; this is why we talk about extensions (or just plain corrections).
            • But the prospect of solving all quantifier problems at once is pretty appealing.

See also discussion of the Conservative Extension.