Re: TECH: new cmavo "ju'e"

[Logical Language Group <lojbab@ACCESS.DIGEX.NET>]

> nothing except experimentation (thought or computer) can pick
>> up such ambiguities.
>
>Are Lojban Central's feelings really a good reason to oppose "stag BO"?

Yes.

>Why not do the experimentation?

Because it is toolate to experiment.  Anything we write into the grammar now is

IN, and we are stuck with it.

>I don't see what the hidden ambiguity
>could be.

I don't either.  We just have certain hunches.  Indeed looking at this
proposal allowed Cowan to instinctively catch the BO prefix ambiguity in
MEX.  Only THEN did he try to come up with a senence proving his hunch.

>> The complex grammar of JOI makes errors in use and resulting scope likely
>> unless the proper terminators are used.
>
>You keep making this claim, which seems unfounded to me. Could you give an
>example sentence showing how this could happen?


I have.  I cited Nick forgetting "ku" on JOI connected descri
ption sumti.
The parser chokes if you do that.  So does Lojbab.  I am VERY intolerant
of certain types of errors in speech and writing - it stops my understanding
cold, and unless I have enoiugh pragmatic information to guess the likely
intent, I will not figure it out.

I can't give better examples, because JOI thankfully hasn;t seen much
other use.  I don't propose encouraging the use of that grammar, bec
ause it
encourages non-logical thinking.  In my classicv sentence of
"John went to the window  and ..." you usually can figure out the grammar
from what follows.  Presumably you co
uld do the same with Lojban even idf you
arbitrarily violated the grammar and used JA for ALL logical connectives.
But then you wouldn;t be speaking Lojban.

>> The proposed usage of stag or stag-BO as a connective, without a preposed
>> connective scope marker, would add an infinite set of cmavo-compounds to JOI
>
>Not infinite, unless you mean use of subscripts.


Well, i had to go back and see what stag was, since it is a creation of the
EBNF nd not the YACC grammar.  And yes, it can be infinite.  In the EBNF
there are 3 different constructs within stag that have infinite repeat
expressd with . . .
But i doing so, I also notice that fi'o constructs are not stags, and hence
you are making it such that fi'o cpnstructs  cannot be used to conjoin
in an "stag-BO" construct.  You would need "tag-BO" and that now moves you out
of the preparser into the regular LALR grammar.

>> and they would be difficult to resolve on the fly since they make use of
>> the BAI words that already have two other usages with different scopes
>> (sumti tcita and selbri inflection).
>
>How do you explain then that nobody has had any trouble understanding me
>when I use them?

I don't recall you using them at LogFest?  When since then have you attempted
to use them in speech?

That someone can eventually puzzle out the stuff you do in text on the list pr
oves nothing much.  I don;t even know if the stuff you write parses or whther
>I< can understand what you write, but I know it isn;t trivial to do so.

>> Just as Jorge's proposal to allow JE the full range of usage of JOI is
>> unacceptable then, so is a major increase of JOI-scoped connectives,
>
>But why is it unacceptable? You take it as a given that it is unacceptable
>but you haven't explained why. You say it would be dangerous and confusing
>but you don't give an example of how it could cause danger or confusion.


Because I have to teach the  *** language, and >I< get confused.  Lojban is
already very-low redundancy.  Opening up JOI to more usage strains that redundan
cy further.  It also mixes logical and nonlogical connectives, which is
confusing to the untrained learner.  Just because you don;t make mistakes
doesn't mean that people working a little less intensely than you  won't do
so.

WE generally have NOT been putting "let usage decide" experiemnts into the
grammar unless we were convinced that even if the experiment failed it would

cause no problems.  This proposal to let usage decide is risky, because if it

causes problems, it cannot be easily removed and the problems caused would be
of the type almost impossible to root out of the language.

Beyond that, you are simply going to have to believe me.  I am standing
very firm on this one, and am not particularly interested in debating it
further.  The proposal is rejected, since there is no consensus in favor of i
t.

>> I would normally have done this using ".e ri'a bo", but Cowan convinced
>> me last
>> night that using a specifica logical connective like ".e" is logically
>> risky - a causal does not necessarily want to claim ".e" truth conditions.
>
>Other BAIs are even less likely to want {.e}. For example {mau}, if you
>want to make only the comparative and not the two absolute claims.

I have generally presumed that using a BAI in there rendered the  ".e" as
not especially llogically relevant - .i.e. not necessarily expandible into
an ".ije" sentnce.  But I could always mark with da'i or something similar
if I did not want to make the claim.  It is easy to non-claim in Lojban.

>>   Using JOI, with its
>> overstretched scope rules is dangeorous and confusing, ESPECIALLY when the
>> problem being posed is specifically one of scope.
>
>Again I would like to see examples. Why is it dangerous and confusing?
>I'm not sure how scopes enters into all this, either.

WE have a whole bunch of different logical connectors for different scopes.
All of the ones thgat we have were nbecessary when we first YACCed the
grammar, and it hasn't gotten any simpler.  JOI jumps across two different
scopes - that of JA and that of A.  It does this by fudging some
 rules and
requiring that the speaker never elide the terminator.    (The rule tha
t is
fudged is noted in the YACC grammar after rule 957.)  Not ever being able
to elide temrionators is simply a pain.

To use JE with JOI grammar in effect makes it also usable where A comnnec
tives
are.  Now the question is why yopu choose one over the other.  Indeed, why
JA ande not A, if you are going to bastardize the grammar in order to
 suit
your aewsthetics.

>> I am thus RELUCTANTLY conceding to John's proposal for ju'e BECAUSE it makes
>> no grammar change and uses only one cmavo, and more so, because it preserves
>> the existing principle of clear scope marking in afterthought connection.
>
>I'm not sure what that existing principle is, could you ellaborate?

sai

A/JA/GIhA/I+JA and the old ZIhA that no longer is a logical connective set.
We use different connectives for different scopes.  Doing so HELPs the
listener keep track of what exactly the speaker wants connected.  This is
MUCH more important in speech than in text, because in text you can reread and
ponder.  If you have a multipart nested sumti joined to another multipart
nested sumti, having clearl indicatoirs of scope may make the sentence
understandable when otherwise it is not.  It is therefore hoped that spoken

Lojban and written Lojban can be similar in level of complexity.  (You
know audiovisual isomorphism)

> If it was up to me I would eliminate
>the guheks altogether.

It's not up to you.

lojbab