Re: [lojban] Re: noxemol ce'u

[Rob Speer <rob@twcny.rr.com>]

On Sun, Oct 07, 2001 at 03:47:16PM -0400, pycyn@aol.com wrote:
> The basic problem with {nei} is that it is self-referential, so that it 
> cannot ever succeed in actually refrring to its referent because it referent 
> does not exist until it has succeeded in referring to it. Thus, it is the 
> whole expression {le nei} that functions as a unit, with the {nei} part there 
> just as a dummy, not a real referrer at all (but enough of a reality that it 
> cannot be used for the first argument in the BRIDI). With counting {le} 
> phrases as bridi in the meaning of the definition, {nei}'s work could be done 
> by {no'a} with a slightly less horrific result (though actually the same 
> problem can be recreated at every level, including {vo'a} -- this aprticular 
> anaphora technique is theoretically deeply flawed, for all that it works more 
> or less in practice). 

Oh. That. The problem is that some people refuse to understand recursion.

The phrase "something such that it sees itself" in English has the same
"problem": "itself" refers to what it is, but how do you know what it is until
you know what sees itself? Humans can make the tremendous mental leap to get
over recursion, however. Note the sarcasm on "tremendous mental leap", because
figuring out the meaning of {da viska lenei} is only as difficult as figuring
out what x is in (x = 2x - 3).

In (x = 2x - 3), a stupid computer program might decide it can solve it by
substitution:
x = 2(2(2(2(2(2(2(...) - 3) - 3) - 3) - 3) - 3) - 3) - 3

Whereas a better computer program or a human with any understanding of algebra
would know to consider both the "x"'s simultaneously and combine them onto the
left side.

When you say "does not exist until...", you imply that the sentence is parsed
in some temporal order. This might be the case if a very basic computer program
is doing it. However, both mathematics and human thought involve resolving
recursion. You can't resolve one part of the sentence before the other, but you
can resolve both simultaneously.

> <* What follows le is a bridi by your definition, but it is not the specific
> ? entry in the parser that we usually refer to when we say "bridi". For
> ? example, you can't put {mi klama le zarci} in {le}. {le} creates a bridi out
> ? of the pseudo-bridi which follows it, and takes the x1 out of it. This bridi
> ? is not part of the sentence>
> 
> Yes, it is not a BRIDI because it contains {be} and the rest -- not because 
> it doesn't have an x1. {le} doesn't make it a BRIDI out of the bridi tail 
> that follows it nor does it take a BRIDI and make something else out of it by 
> dropping x1 and replacing it with {le} (although that is closer). The result 
> of putting {le} in front of a bridi tail IS a part of thes entence -- where 
> else would it be? I think I am missing your point here. Whose "bridi" are 
> you talking about, not mine and not what I understood your to be.

Let me try this again with a specific example:
{mi viska lo broda be le brode}. (lo is easier to work with than le.)
The pseudo-bridi is {broda be le brode} (where did I say anything about not
having an x1 making it not a BRIDI? It's not a BRIDI because it's not parsed as
one.)
{le} makes this into an actual bridi: {ko'a broda le brode}. {ko'a broda le
brode} never becomes part of the sentence. It's a bit abstract, but if you
can't understand this concept we're at an impasse.
The imaginary {ko'a} does become part of the sentence, as the referent of {lo
broda be le brode}.

The result is that the sentence is {mi viska ko'a}, with that ko'a not really
there, but referring to the same thing as {ko'a broda le brode}.

I could do this without constructing a separate bridi, using poi and voi, but I
assume that's what And is doing which you object to.


> <poi contains a new level of the sentence. du'u contains a new level of the
> sentence. This is because both of these are followed by a BRIDI. {le} is not
> followed by a BRIDI. In some situations, as you enjoy pointing out, the thing
> that follows {le} could be a BRIDI on its own (an observative one, usually),
> but that does not matter.
> 
> This does not prevent rewriting a {le} phrase with {voi}, for example - but 
> you
> would have to put a subscript on {ce'u} if there was one in the {le} phrase.
> Situations like that are the reason subscripts for ce'u were proposed, after
> all.>
> 
> Actually, the {ce'u} would probably still not get a subscript, since it would 
> still have smallest scope, but you could move it up if you wanted -- as you 
> can in any case. Notice that what ahppens in And's rule -- one version 
> anyhow -- is that he makes the containing "sumti" disappear but exactly 
> replaces it with a new sumti which contains a BRIDI and the same internal 
> sumti and such that the whole means the same (officially) -- and this somehow 
> is to prove that the internal sumti, which now is clearly at a different 
> level from the sumti it is contained in was previously at the same level as 
> the sumti it was contained in. I just don't see how it follows, but in any 
> case the transformation that is done is trivial and has no bearing on the 
> issue at hand.

The conversation between you and And soared to great levels of abstraction. You
two couldn't understand each other, so I rather doubt I could.

If that is in fact what And meant, it's nonsense. A sumti does not contain a
BRIDI. It might contain words which would be a BRIDI if they weren't in a
sumti, but since they are in a sumti, they aren't a BRIDI.
-- 
la rab.spir
noi sarji zo gumri