jbocre: zo'e poi

From Lojban
Revision as of 16:18, 24 May 2015 by Cirko (talk | contribs) (formata gasnu)
Jump to navigation Jump to search

This discussion about zo'e, noi and poi took place on the #lojban IRC channel on freenode, 2014-09-04.

mukti I'm reminded of a related question I had when I first encountered the zo'e noi broda formula, which is to say, why noi broda rather than poi broda?
lukys Doesn't poi mean something that is essential for the identity, whereas noi is some incidental detail?
durka42 zo'e is always correct by magic, so it doesn't need to be poi
mukti If I say lo botpi, I'm referring to something contextually sensitive. That's the zo'e part, right? But the referent is restricted (or so it seems to me) among all the contextually available referents in so far as it satisfies botpi. That restriction seems to me more like a poi than a noi. Someone set me straight.
durka42 I think zo'e covers both of those, it's just not very useful for the listener to have a bare zo'e in every place...
mukti lukys: Yes, it's my impression that poi limits reference, whereas noi comments on referents.
xalbo I've never understood the problem with zo'e poi either. It still seems right to me. zo'e magically changes reference to find the contextually relevant thing, but zo'e noi broda cu brode seems, to me, to be saying that the thing we'd expect to be talking about if we saw zo'e brode also satisfies broda, where zo'e poi broda cu brode seems, to me, to be saying that the thing we're talking about that satisfies broda satisfies brode.
mukti zo'e noi broda seems to imagine a situation where the referent is pointed to independently of being described as broda. If lo broda is equivalent to zo'e noi broda, aren't we saying that the reference is independent of the description? That the description is incidental?
durka42 well, the reference is in the speaker's mind
mukti Ok, but is it a typed pointer or a void * ?
durka42 don't know how to answer that. I think it's typed
mukti What I'm trying to get at is whether or not the reference can be said to depend on the description.
durka42 that's what people keep going back and forth on
mukti If the reference does not depend on the description, then it is similar to a void * in C. It's the address of data which is unspecified in structure.
durka42 but the data's there, either way
mukti Well, it depends what you mean by "the data". :)
durka42 the referents
mukti If lo broda is zo'e noi broda, and the reference is *not* said to independent of the description, then zo'e must have some property of suspending reference until relative clauses are considered. Or so it seems to me.
durka42 I guess I'm saying they're independent then, but I don't know if we can't answer this question… perhaps zo'e is too magical...
xalbo Well, I think the point is that zo'e is nearly limitlessly magical. It does whatever it needs to, given the entire context (including relative clauses, and everything else possible) to make what you say true.
durka42 but you can still say things that aren't true...
xalbo I guess, then, to make it mean what you mean it to mean...I'm not sure.
durka42 if I say mi dunli lo merja'a, intending to lie, but zo'e undermines me by magically resolving to lo ka remna kei po'o, that's kind of annoying
xalbo Being xalbo, I'd say that's why you wanted to say mi merja'a in the first place. Or at the very least use mintu or du.
durka42 well yes
xalbo But you make a good point. If someone asks mo fa la tepcrida ("What happened to the dementor?"), then se citka mi is a lie, even if se citka mi would otherwise be a true statement (I did eat something). Though maybe that has more to do with place-filling and mo. "What I told you is true, from a certain point of view."
durka42 zo'e has to take the speaker's intentions into account, perhaps that's the same as saying zo'e takes the UD into account
xalbo I don't think those are the same. I'd say it probably has to take both into account.
selpa'i Don't take zo'e noi broda too literally. It's not a text replacement, but a referent replacement
durka42 uaru'e
selpa'i zo'e noi broda is lo broda if zo'e brodas.
durka42 but you can actually say zo'e noi in a sentence
selpa'i Sure, and then you need to know what zo'e refers to. Or not care.
durka42 or, every other week someone invents another experimental cmavo that means zo'e noi [bridi]
xalbo "<selpa'i> zo'e noi broda is lo broda if zo'e brodas." -- That seems like a really complicated way of saying that lo broda is/does the same thing as/mintu zo'e poi broda
durka42 xo'e da poi broda zo'onairu'e
selpa'i I understand zo'e poi broda as lo me zo'e je poi'i broda. poi as a definition for lo doesn't seem right.
xalbo Why not? Isn't the point of lo broda essentially that it gives you a referent that broda's?
selpa'i But zo'e is that referent already.s
xalbo I don't see mi pinxe lo ckafi as saying "I'm drinking something. BTW, it turns out it's coffee. Who knew?"
Ilmen Maybe defining zo'e from lo would be wiser than the other way round
xalbo zo'e and lo co'e seem really, really close to me.
selpa'i But do you see lo ckafi as a restricted reference?
xalbo Intuitively I think i do.
selpa'i Part of why it may seem weird is that zo'e does two different things. It can be "it" or "something"
Ilmen Saying zo'e = lo co'e is probably not more bad a definition than lo broda = zo'e noi broda
durka42 what a ringing endorsement
selpa'i Do you start with a bigger reference set and then restrict it to coffee? lo ckafi goes right to coffee. zo'e poi broda takes zo'e as a start and then restricts to those among it that also satisfy broda
durka42 but zo'e is magic, so it changes to be the restricted set as soon as you restrict it?
selpa'i lo broda -> zo'e noi broda is only true for a zo'e that refers to brodas
xalbo And zo'e noi broda takes zo'e as a start, and then says that it already satisfies broda. Which seems far odder to me.
selpa'i Which is why it's not a literal equality
selpa'i But that oddness comes from zo'e doing both unspecified reference and definite reference
xalbo But the way you say "and this is only true for a zo'e that satisfies broda" is to use poi. That's pretty much exactly what poi does.
selpa'i The equality is only true if zo'e refers to brodas. And zo'e takes its value from context
xalbo It just seems like you keep saying things that sound, to me, entirely consistent with zo'e poi broda, while rejecting "zo'e poi broda". zo'e, but only if it satisfies broda.
selpa'i That's a meta-statement about the equivalence.
xalbo To me, the difference between restrictive and incidental is that I would expect the former to change the referent. If I saw zo'e poi broda cu brode, I would expect to find some zo'e that satisfies both. If I saw zo'e noi broda cu brode, I would expect that the very same zo'e I'd get if I just saw zo'e brode would also happen to satisfy broda. Which, in fact, is a whole lot closer to what I'd expect for lo brode cu broda.
selpa'i The equivalence "lo broda == zo'e noi broda" holds only when zo'e refers to brodas. That's different from saying that lo broda means a zo'e that only refers to brodas. zo'e poi starts with a bigger referent set
xalbo That seems like a bizarre equivalence. What's the point of the noi broda, then? Why not just say "lo broda == zo'e, but that only holds if zo'e refers to brodas"?
selpa'i Exactly. And noi broda only comments on the referent.
xalbo But poi broda adds the bit about requiring it to satisfy broda, but takes it out of the metalanguage English qualification and into the actual equivalence. Why comment on a referent you've already restricted externally?
selpa'i Why restrict that referent again if it's already the referent that makes the bridi true?
xalbo Because we're adding part of the bridi that we also want to say is true (that the referent must also broda).
selpa'i We already know that it does. In the definition of lo broda (remember the "don't take it too literally"). That definition starts by knowing the referent of zo'e. zo'e poi broda is like saying "The contextually obvious things that also broda (two properties need to be satisfied)", while zo'e noi broda is more like saying "Those contextually obvious things, and those things broda" (only one property, namely broda)
xalbo I'm saying that if we define lo broda to mean zo'e poi broda, we'd need a lot less of the "don't take this too literally", "this only applies if it already broda", and other provisos. I don't understand. What are the two properties that must be satisfied for zo'e poi broda?
selpa'i me zo'e and broda
xalbo And zo'e noi broda only requires broda, not me zo'e‽ Or am I misinterpreting "(only one property, namely broda)"?
selpa'i The referent in the poi case includes only those individuals that satisfy both properties, whereas in the noi case the referent is zo'e, and it's (incidentally, that is, it has no effect on a quantifier) broda. This is quite similar to the difference between ro ko'a poi and ro ko'a noi. One has a logical conjunction imposed on the referent, the other asserts both independently. This second part is hard to explain but the quantifier example is hopefully helpful
Ilmen BPFK: "ko'a poi broda" = lo me ko'a je broda ----> According to this definition, defining "lo" from "zo'e poi" would lead to a circular definition, wouldn't it?
selpa'i zo'e poi broda "those things among zo'e that broda"
xalbo I still don't understand. What's wrong with restricting our referents to only those that broda? That seems to be a fundamental thing to what lo broda does, and it seems that even you are doing that, you're just doing it in English separately with "it's only true if zo'e satisfies broda"
selpa'i I tried to make it very clear that that last part is *not* part of the definition; it is a comment *about* the definition
xalbo I don't see the difference between "those things among zo'e that broda" and lo broda. To my mind, noi broda adds completely incidental information. That is, we could replace noi broda] with goi ko'a, and then add a separate sentence ko'a broda, and get the same result (scope issues and grammar issues notwithstanding).
selpa'i I do see a difference between "The dogs" and "The things among those things that are dogs" (though the latter reads a bit ambiguous)
Yes. noi broda adds a separate statement. It seems your trouble is actually with the step from zo'e noi to lo, not vice versa. Or maybe you think it doesn't matter
xalbo I'm not sure.
selpa'i You can go from lo broda to zo'e noi broda in the gadri definition because the definition can choose that this zo'e refers to lo broda. Thus you can go from any lo brodi to zo'e noi brodi as long as you have a zo'e in mind that already refers to exactly what you want. However, in the other direction, it's less true that you can simply replace the strings. Going from zo'e noi broda to lo broda requires the zo'e to refer to lo broda. But not every zo'e refers to lo broda, it takes a special context. If zo'e is tea, then zo'e noi broda won't be lo ckafi. And that's why you cannot take it as a literal replacement.
xalbo Then that makes using zo'e noi broda to explain lo broda less than worthless. You have to already have lo broda as context for zo'e, the noi broda adds literally nothing, and it only works when it already works.

Continued discussion

2015-05-21

omni___ Has anyone who uses le managed to formulate a Lojban definition of it that they are entirely happy with?
zipcpi I don't know how to have such metalinguistic conversations in Lojban, unfortunately
xalbo I was fine with le broda = zo'e voi broda, but others disagree with every bit of that for technical reasons that I don't understand.
zipcpi Problem is almost no one uses voi. But yeah in translations I tend to use ra (non-contextual translations)
nejni-marji .i do gletu zo voi .ije coi nai do, .i do za'o cupra lo mrofoi
durka42 one problem is I've never seen anyone define voi except "something to do with le", so if you define le in terms of voi then everything explodes
omni___ Yes. You first have to define voi. And if it's restrictive, then it becomes weird with zo'e
gleki xalbo: but then you need to explain what is veridicality and provide a bunch of examples where lo and le should be used.
durka42 that is, it's fine but it doesn't answer the question of "what does le mean to you" :)
zipcpi I can only think of one use-case for voi; that of "titles" affixed to a person's name. But I'm sure there's another way to do that. Like maybe no'u la'au ... li'u
xalbo Remind me again what's wrong with restrictives applying to zo'e?
nejni-marji Personally, I'd think to use voi when describing an event I'm too far from to see clearly. And such, describe what I think is happening.
omni___ zo'e poi broda == lo me zo'e je broda, meaning zo'e can refer to things that don't broda
durka42 me zo'e je broda would seem to require an x1 that does broda
omni___ zo'e alone.
durka42 zo'e alone can refer to anything
durka42 is that relevant? you said zo'epoi :)
omni___ Yes. The zo'e is a bunch of stuff (which needn't only include things that broda), and poi restricts those things.
omni___ That came out a bit wrong. It restricts the referents of the *resulting* sumti, not zo'e itself. If voi is anything like that, it becomes weird.
nejni-marji example lu doi lo pulji mi zgana lo clapre voi zerlebna lo jdini pe lo tolcni prenu zi'e poi klama lo stici li'u. "Officer, I saw a tall person <voi stole money from an old person> zi'e <poi heading west>."
xalbo I'm still understanding the problem with zo'e poi. Is there some concrete example or something?
nejni-marji durka42: me zo'e ~ co'e
zipcpi One thing I don't like is how three of the four cardinal directions have the same initial letter
omni___ Yes, and thus the FAhA don't match
xalbo zipcpi: be'a, du'a, ne'u, and vu'a have distinct initials, so those tend to be used.. But yes, it's silly.
zipcpi * zipcpi nods
xalbo omni___: So you object to mi citka zo'e poi plise, because the zo'e could contain things that don't apple, before being further restricted? Or...I still don't see the issue.
nejni-marji It's more annoying than lo rafsi be ma'oi se. But it's not a huge deal because I rarely use them.
omni___ Make new brivla then that match the FAhA. I don't "object" as much as it just sounds weird.

2015-05-22

durka42 http://mw.lojban.org/papri/jbocre:_zo%27e_poi oh no I am preserved in this conversation for perpetuity :(
rutytar durka42: welcome to the internet
Vanlamigu Yeah we made it into a wiki page since this topic keeps coming up
durka42 would be great to have a TL;DR at the top
Vanlamigu Yeah, but is the tldr? Whose opinion is the right one :P
rutytar i know whose opinion is the wrong one
durka42 Vanlamigu: at least a summary of the different camps, then. I was _in_ the original conversation and I can't follow it
Vanlamigu Hummm
rutytar as i noob, i would really appreciate a summery of the different camps on different issues
durka42 as a jbocre, go'i ra'o
xalbo Vanlamigu: Well, I got in the final word, so I guess that makes me the right one :
Vanlamigu I just cut it off at that point :)
xalbo Rereading that conversation, I still don't understand fpcalep's views any more than I ever did. I mean, they make *no* sense to me. I would really like to understand, but I'm completely unmoved.
Vanlamigu I don't know what else fcpalep can do.
durka42 * durka42 is still not sure who Vanlamigu is but is pretty sure he is fpcalep
xalbo Can you give any bridi where, for some predicates bu'a and bu'e, zo'e noi bu'a cu bu'e, zo'e poi bu'a cu bu'e, and lo bu'a cu bu'e are not all truth-conditionally identical?
Vanlamigu Then again, xorxes has been unable to get PC to understand his ideas for some 20 years, so I guess I'll just hang in there.
durka42 * durka42 would also not mind such an example
Vanlamigu The point is that zo'e noi broda cu brode has zo'e ("all" of them (though not necessarily distributively)) brode, while zo'e poi broda cu brode does not.. That's all.
durka42 "all" of *them* <— them = ?
Vanlamigu zo'e
durka42 je'e
Vanlamigu The first zo'e might be just one thing, but the latter is most certainly more than one.
durka42 because poi narrows a set while noi jungaus things about the set without changing its members (please excuse my use of the word "set")
Vanlamigu Right. But I don't know if that formulation is helpful (?)
xalbo I'm sorry, but what is "the latter" there? Are you saying that zo'e poi broda cu brode means that there must certainly be more than one referent of zo'e?
durka42 that's how I think about noi/poi, but I dunno if it's useful either
Vanlamigu poi implies a non-singleton domain. Applying it to a singular constant implies (at least pragmatically) a restriction over instances of that singular referent (though it does not logically entail it)
xalbo So lo mlatu poi blabi cu citka is false if there's only one cat, even if it's white and eats?
Vanlamigu No that's a different sentence. Add a ku before poi
xalbo Point. lo mlatu ku poi blabi cu citka
Vanlamigu Yes, so what I said applies there.
xalbo Ahh, pragmatic constraint instead of logical. That makes a bit more sense. lo mensi be mi ku poi speni cu jibni implies (pragmatically) that I have more than one sister, and that at least one is married and at least one is not married, even though in reality I have exactly one, she is married, and the statement is literally true.
Vanlamigu Yes, otherwise why would you use poi, a restrictive relative clause.
xalbo Is that what you're getting at?
durka42 "you're my favorite son!" "…I'm your only son."
xalbo "<sadly>Stop reminding me."
Vanlamigu "You're the five greatest Lojbanists!" "... we're the only Lojbanists."
xalbo I think we're thinking of poi and noi somewhat differently. In particular, I feel like noi only adds extra information, but doesn't change the referents of anything else. But you're asking it to. zo'e citka and lo mlatu cu citka are very different things.
Vanlamigu No, noi doesn't change referents.
xalbo zo'e noi mlatu cu citka seems to me give something rather different. That is, zo'e citka .i lo go'i cu mlatu.
durka42 I would say those *are* the same, if it's obvious to everyone involve that we are taking about mlatues. ko catlu lo mlatu .i citka
Vanlamigu Not a bad example, since there zo'e and zo'e noi mlatu do the same exact thing, because we know what zo'e is.
xalbo lo mlatu cu citka, though, doesn't already assume that we were talking about cats beforehand. Among all the things that zo'e could potentially mean, it limits it to only those that mlatu. Therefore, zo'e poi mlatu cu citka
Vanlamigu And the same applies to the xorlo definition
Vanlamigu zo'e is not "everything in the UD". zo'e is what makes the bridi true. What you say sounds more like ro da poi mlatu
xalbo Agreed. I think we disagree about what noi does.
Vanlamigu How?
xalbo I feel like noi only adds information, but doesn't get to change around the referent. So zo'e noi mlatu cu citka is only true if zo'e citka would already have been true in context, and additionally says, of that same contextually found zo'e, that it mlatu.
Vanlamigu Yes, I agree.
xalbo Not that you pick a single zo'e based on both parts, but that the noi doesn't participate in the choice of zo'e, it only tells you more about it after you've already chosen it.
Vanlamigu Yes. I never claimed anything different.
xalbo Then if lo mlatu cu citka, zo'e poi mlatu cu citka is necessarily true, and vice versa (even if poi seems to be misleading, it's still valid). But there are many cases where lo mlatu cu citka and zo'e noi mlatu cu citka diverge in meaning.
ldlework xalbo: is there an article like po'o that means 'other than'?
Vanlamigu na'e[bo]
Vanlamigu In the previous discussion, I mentioned that lo broda is only equivalent to zo'e noi broda if you already know that zo'e broda.
Ilmen even worse, if "zo'e .olkai lo ka broda" :p
xalbo And my point is that that makes using zo'e noi broda worse than useless. You've literally added nothing to the conversation (if you already know that zo'e broda), except to make things more confusing.
ldlework well
Vanlamigu But it's not something to say in a conversation as much as it is a definition.
ldlework its like a subtext
Vanlamigu It's trying to paraphrase lo broda into something that doesn't use lo
ldlework Especially if you already have a substantial description, like "The monkey who always steals the straws from my drink"
Vanlamigu And in the context of the definition, we already know that zo'e brodas
xalbo If we already know that zo'e brodas, then why prefer zo'e noi broda over just plain zo'e?
ldlework ko'a noi mi ro roi mi fraxu cu morji
xalbo Why add the extraneous noi broda that we already know has to apply?
ldlework I wouldn't just
ldlework say
ldlework lo se fraxu be mi cu morji
ldlework pei
Ilmen maybe I could use a similar trick for defining lo mlatu = lo co'e. zo'o
ldlework xalbo: pei
Vanlamigu It'd be funny to define lo mlatu as just zo'e. The zo'e tells us it's a constant, and the noi-clause makes it extra clear that the zo'e (and lo mlatu) is satisfying mlatu
xalbo ua nai
ldlework xalbo: you're saying, instead of ko'a noi broda just say lo broda. I'm saying, if ko'a is already defined as something meaningful, then the noi acts as a subtext, not the main (and usefully identifying) description reference
xalbo No, I'm saying define lo broda as zo'e poi broda and be done with it.
ldlework oh sorry
Vanlamigu It's a meta definition. The noi is not restricting anything, just explaining to the person who wants to know what lo does
ldlework I didn't realize there was a deeper discussion
xalbo They're true in exactly the same situations, and we don't need an extra "as long as zo'e already broda" metadefinition.
ldlework xalbo: what if the thing you're relativizing is already described?
ldlework IE, its still subtext
xalbo That is a valid point.
ldlework xalbo: I think your point about definition is apt tho
Ilmen zo'e poi broda --> lo co'e poi broda ---> lo co'e je broda. ie nai pei
mensi ba'e mi na tugni
Ilmen re'i la mensi .i ko sa'umnai cusku
Vanlamigu na ru'e drani. lo co'e poi broda -> lo co'e je poi'i broda
ldlework xalbo: I think using zo'e, da, ko'a or any other unquantified thing with noi or poi is ultimately useless though like you were saying
Vanlamigu Else you get wrong alignments of x2 x3 etc
Ilmen .i'a
xalbo So the problem with zo'e poi broda is that it implies (weakly, pragmatically) that there are other zo'e around that don't broda (even though it actually works either way). The problem with zo'e noi broda is that it breaks *entirely* if there are.
Ilmen at any rate for me lo co'e je poi'i broda ≠lo broda, in most cases
xalbo To me, that makes the former a far better choice for a definition than the later.
ldlework xalbo: ie
xalbo Ilmen: Defining lo in terms of lo is ... co'e.
ldlework u'i
xalbo co'e ge'e
Vanlamigu Different strokes. Perhaps you would prefer: lo broda cu brode == ko'a broda .i ko'a brode. The .i is equivalent to noi
ldlework le co'e => lo co'e poi mi ka'e termi'u
ldlework * ldlework bajra
Vanlamigu Using two sentences, you don't have to bother with relative clauses