lerfu pro-sumti, and why ko'a sucks
|This page contains discussions of experimental/scientific/philosophical/logical aspects of Lojban that are|
non-official and not for everyday usage. You've been warned.
On the mailing list, And Rosta writes:
- Mark writes Type 4 fu'ivla:
- "we can use any string of lerfu as a ko'a-style sumti variable (which makes me
- think that there's practically no reason ever to use the ko'a series at all)"
- -- can you explain?
To which I (Mark) answered:
Um, I can try.
As I understand it, any old lerfu-string in a sentence can be used as a sumti (yes, it's also a mex, usable with li, and so forth. I mean just a bare string of lerfu as a sumti). It's considered a "variable" pro-sumti, assignable with goi (well, you can assign any sumti with goi, but I mean it's semantically and pragmatically sensible in general with lerfu-strings). If unassigned, they default to the most recent sumti with the appropriate initial letter(s), if any, or so they tell me. So I can say "le ctuca .e le vecnu cu prami le speni be cy." for "the teacher and the merchant separately love the spouse of the teacher," relying on the default that "cy." goes back to "ctuca." Or, more conservatively, you can say "le ctuca goi cy. .e le vecnu..." with the same result, using cy. exactly like ko'a (with the slight difference that an unassigned ko'a is meaningless, while an unassigned lerfu-string will at least try to find a meaning, possibly the one you wanted). So basically, a lerfu-string can do anything a ko'a can do, and then some. The two and-then-somes are: (a) as mentioned, if it's unassigned, it will try to snag a nearby meaning, and if done properly, this is not so bad and pretty reliable (see below). (b) lerfu-strings have far more potential for mnemonic power than ko'a. It's a lot easier to keep track of pronouns for la bab., la djan., la .alis., and la fred. as by., dy., .abu, and fy. than as ko'a, ko'e, ko'i, and ko'o (has anyone ever used ko'o? I had to look that one up to double-check that it really was still in the series). (mi pilno mi'e maikl.)
I like using these, though I don't trust the implicit assignments completely (but I do use them, just not in all cases). The mnemonicity is a real brain-saver. I don't like the implicit assignment with descriptors, though, for some reason. Somehow it doesn't seem that reliable. Maybe it's because there's nothing about lo ctuca that particularly would associate her with cy; I might just as easily have described her as lo ninmu and use ny. This is not a valid argument I'm making, just something that sort of affects my thinking (after all, there are endlessly many names for everyone). Also, with all the brivla that get used in a sentence, I could easily forget that there was another intervening cy-sumti that would thus get misassigned. But with names I have no such qualms, especially if I have a two-word name and can use two initials. This saves me the assignment step, and works quite well. So in  I refer to la mark. okrand. and then in the next sentence use "my.obu." Perhaps that's counting on a little too much, that the reader should know I'm taking initials (as opposed to, say, the first two letters in the name), I don't know. I think it works. Or recently I was writing something about the Phillip Morris company, referring to it as "la filip moris" and then as "fy.my." (or "fymy.") On the other hand, when I introduced the KLI as "la klingon. zei bangu ckule", I assigned it with "goi kybycy." and used that throughout (I didn't trust the assignment to a tanru/lujvo/zei-thingy of indeterminate initials). Multi-letter strings are even less prone to confusion than single letters, so I think they work quite well. I used "xy." for the KLI's journal HolQeD, but assigned it with goi, since I used la'o quotes and can't count on people to know the pronunciation.
(As an aside, here's a bizarre consequence... since cmene+bu is syntactically a lerfu, you could use ".mark.bu" as a sumti, leaving it up to the listener to somehow work out who it was... which isn't all that much worse than saying "la mark." and hoping they know. The best guess for "mark.bu" is probably somebody named Marck, right? (And no cracks about how "mark.bu" is some "letter Mark." That's true, when mentioned, but not when used this way. "li mark.bu" is a mathematical expression, like using "area" as a variable. To get "letter-mark" you'd need a letter context, not a sumti context. Just like I can use "by." for la bab., I can use "mark.bu" for la markonilentironafilos.]
- But if you haven't assigned .mark.bu, then the listener would have to figure out what sort of name begins with the letter Mark.
- True. But such a listener, absent any other ideas, could do no worse than to guess that the name is Mark, or starts with it. Or, more likely, would wonder if he missed the introduction of the name earlier. I think you could get away with using mark.bu as an alternative to la mark.... but only with a very clever and playful and fun-loving audience. It's more a curiosity, I think.
The only advantage ko'a-series does still have (aside from history) is the fact that three of the fo'a series (fo'a, fo'e, fo'i) have rafsi and can thus make lujvo... but has anyone really done this? So that's why I say I think the ko'a-series doesn't have much utility. Anything they can do, lerfu-strings can, and more, and easier.
Okay. Since posting that, And Rosta and pycyn both asked about the problem of telling fy.my. is one sumti and not two (fy and my). The answer is the little-known lerfu-string terminator boi. So if they were really two sumti, I'd have to say fy.boi my.boi, but in practice I don't see that happening all that much. Well, at least, it hasn't in my experience. I don't know there'd be all that many lerfu pro-sumti at once, even though I think that the mnemonicity involved would enable far more pro-sumti than the usual half-dozen or so that people can keep track of at once. But even beyond that, another curious feature of Lojban is that even though gismu have up to five places, and lujvo can have even more, it's really comparatively uncommon to see more than two of them filled in. Someone should do an analysis someday. And if there are only two, usually the selbri separates, so boi is unnecessary. And even if you do have to say it now and then, I don't think it's much of a hassle, on the whole.
(The problem with this is, this was meant for casual usage; and as soon as a second gismu beginning with that letter appears, doesn't it render the assigned lerfu ambiguous?)
(a) That's not a problem if you use two-lerfu abbreviations, but that's a cop-out. (b) I'd say that an explicitly bound lerfu pro-sumti should have great enough staying-power to withstand implicit binding by an intervening gismu. I don't even think that's controversial: I believe the official word is that lerfu-as-sumti take the nearest same-letter sumti as their referent only if they are not otherwise bound. --mi'e .mark..
You don't just have to use the first letter, obviously, which leads to non-obvious solutions like goi la .y'y. - since nothing begins with ' in lojban. True, but that blows away all the advantage of using lerfu in the first place: the mnemonicity is a big plus. Oh, and it would be "goi .y'y.": the lerfu IS the pro-sumti, not a name.