"Regular lujvo", the ideology that the place structures of lujvo should be derivable from the place structures of their components in some regular and predictable fashion, in order to make lujvo have realistically usable and guessable place structures. Under this approach, it should be possible to at least constrain what the x2, x3, x4... etc. of a lujvo may be, and the semantic values of those places should be drawn from the semantic values of the places of the component brivla.

The alternative (which is probably still LLG policy) is that the place structure of each lujvo has to be determined from usage.

The (hardliners-like) objection to this is

  1. This multiplies the years of inconclusive debate over gismu place structures by several orders of magnitude.
  2. In what may be termed the atismo principle of conlangs (see http://groups.yahoo.com/group/lojban/message/9783, 6d) [or may be termed something else: suggestions welcome], if there isn't a readily available place structure, language users will just avoid specifying non-trivial lujvo place structures entirely. In real life, this is actually what happens: few nonce lujvo see more than x1 and x2. [seljvajvo would like to change that.]


seljvavo were initially championed by Jim Carter. The cause was then taken up by Nick Nicholas, who expanded on Carter's work, and John Cowan, who refined and simplified Nicholas' scheme (and further tempered its hardlinerism, which Nick Nicholas wishes to state he is grateful for.) Although the cause has never enjoyed Bob LeChevalier's approval, Cowan's statement of seljvajvo principles was incorporated into the refgramm, Chapter 12, though not as a binding part of the grammar.

There is much wiggle-room left within seljvajvo, too. For example, the issue of Lean Lujvo.

Formerly (and frequently still, for those with long recall) termed dikyjvo.



How many of the hundreds of lujvo in Nora's list are seljvajvo, and how many are not? It's a bit late to start issuing proclamations on how to build Lujvo, isn't it? --xod

The place structures in Nora's list were mostly devised by me, in 1994, as seljvajvo. Outside the seljvajvo program, non-trivial place strctures of lujvo (e.g. past x2) have been sporadic, and exceedingly rare. The 1994 does not document existing place structures, because overwhelmingly, there were no existing place structures, outside x1 and occasionally x2. In that regard, the list was very much a 'proclamation', tempered with (I hope!) enough common sense not to be overly rigid about the rules.


The 'proclamation' was incorporated (suitably watered down) into The Book, so it is not 'too late', it is a fait accompli. I have no doubt most Lojbanists, including you, will ignore it; all I can say in that case is, good luck when you hear someone use an x4 for a lujvo.

That aside, my April proposal for the dictionary is that only x1 and x2 be included as recommendations in any dictionary lujvo place structures, and any trailing places should have a less certain status. — nitcion.

Since above you said the LLG policy was different than si'o seljvajvo, I figured plenty of the existing lujvo violate sy. — xod.

One, I know for a fact I tampered with. Lojbab had proposed in a JL jdaselsku 'prayer': x1 is a prayer to deity x2 by person x3. But a selsku has the place structure: x1 is expressed by x2 to x3. So I changed it when it came time in 1993 to compile the lujvo list (which I certainly understood not to be a documentation of existing place structures - there were almost none such proposed or used - but a proposal of new ones, for the purposes of writing a dictionary.) This was fascist of me, yes. But is the inconsistency between selsku and jdaselsku a good thing?

You totally should have created the much simpler 'jdasku'! Then you keep the places of cusku, which is what people are more without-thought familiar with than the places of selsku--xod

  • *smile* I did create the simpler jdasku. But I still had to provide a provide a place structure for jdaselsku, because I had to provide place structures for all lujvo coined as of 1994. Obviously, se jdasku is the only sensible way to say 'prayer'. — nitcion.


(If your answer is yes, or at least, that it's tolerable for the greater good of determining place structures individually according to context - then seljvajvo are not for you, and Lojbab will feel vindicated :-) . I've said amply what dire consequences will attend you; but the seljvajvo have never been accorded regulatory power.) nitcion

But I am still confused because most of what you just wrote seems to be discussing lean lujvo, not regular lujvo. The fact that a lujvo is regular has nothing to do with how many places there are.--xod

It constrains how many places, because it encourages you to eliminate redundant places. It can't determine ahead of time "this lujvo shall have 4 places", true, but it does say "this lujvo shall have less places than the sum of its components" (i.e. that at least one place will be redundant, and following the proposed templates helps you pick which one.) In any case, what seljvajvo are really about, in my opinion, is not the number of places, but the order. It doesn't ultimately matter all that much if you've put in 4 or 7 places; it matters far more if what you might reasonably have expected to be the x2 place, you see turn up in the x4 place. — nitcion.

seljvajvo and the Book's algorithm are probably very good for a default. There may be cases where they create a really inappropriate result. I can't give any examples. The idea of understanding the places of a never-heard-before lujvo is alluring, but the elimination of redundant places adds tedium to that process, making it once more a hard task. (What's the ve of a 3-rafsi lujvo? You'll be there a while! You will resort to context, not any lujvo algorithm.) Perhaps if you want a lujvo completely recognizable to the unfamiliar, it should be a tanru. Lujvo may have to be learned just like gismu; out of common usage, and out of dictionaries, whether or not their places are seljvajvo. --xod

  • All I can say is, I counterproposed a seljvajvo to Jay's non-seljvajvo in Astronomy (ninsisli'u); it was three-part, and it took me 30 seconds. I would rather lujvo not be learned just like gismu, but I've already said my piece, so everbody, do as you see fit. As you'll have noticed, I tend to use tanru more for precisely this reason of unfamiliar-recognisability. (For the same reason, I tend not to use the shortest rafsi.) — nitcion


My problem with this is that it seems to limit the semantic space of lujvo drastically. They can't ever really mean anything not already expressible with gismu (and drastic amounts of fi'o and zi'o) My preference would be that you try to make lujvo seljvajvo. Nothing wrong with them, certainly. But one should also not be afraid to junk the concept. IMO: The components of a lujvo should only be guide posts suggesting the meaning to you, not restricting the meaning. Also IMO: In the end, lujvo have to be learned from a dictionary (or your community of speakers if you're native), not every lujvo's meaning can be obvious from its makeup. --jay

The Gismu Deep Structure Hypothesis, which 'Tweeners will remember, states precisely that "They can't ever really mean anything not already expressible with gismu"; seljvajvo are an outgrowth of this. — nitcion