Re: TECH: PROPOSED GRAMMAR CHANGE X6: Simplification of compound tenses

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

>That's not my impression at all. I would say that it is dominated by
>discussion about "how to say it", and "what does this mean". Very few

>new cmavo have been proposed, and even those are ususally in an already
>existing selmaho, so that there is no grammar change involved.

When "how to say it " implies that the language cannot cope with the
problem as is, then there is the implied threat of change.  Thta alone is
intimidating to some. A new cmavo in an existing selma'o is a dictionary
change.  It may or may not require work in the regrammar, depending on whether
it works semantically just like all the others of that selma'o.  John has
tried to be pretty comprehensive within selma'o as well as covering all
selma'o.

>> We can imagine the impression
>> this gives outsiders, when all we can tyalk about is what changes we are
>> thinking about.
>
>This is a matter of public relations.

And one that JCB managed horribly, and which we will manage horribly as well
if we cannot present the image of standing very firm against change.

>None of the changes that occured
>or were seriously proposed in the last three years (which is the time I've
>known Lojban for) has or would have had much effect on Lojban as used.

If they have no effect on Lojban as used, then they are unnecessary.  If
they merely don;t affect older text (what you probably meant) then they
might still be justifiable.  But my experience has been that even too many
extensions give a bad impression.

The problem is two fold.  Some people don;t want to have to learn and then
relearn.  Extensions don;t hurt them, so long as a) we convince them that
all changes will only be extensions (we've doine this with the gismu list)
and b) that all the extensions will be documented in the refgrammar and
rexrbook so people know what the h*** they are seeing when they are used.
The latter is why new cmavo have to be docuemnted in the refgrammar.

But an even greater number of people have I think a  stronger criterion.
They will not use the language until they feel that it is no longer a
plaything for conlang designers. To them, ANY CHANGE is us continuing to play
around.  To them, if an extension is justified, then that means that the
language before the extension was incomplete.  Well, if you have had 6
extensions tot he language proposed in the last month, just think how
incomplete the language design must be.  And these are only the proposals of
one Lojbanist.  What happens when And and all of the other conlang fiddlers
start  proposing THEIR XsubNs.

To these latter people, the requirements for them to accept the language as
done is a) the books done so we are on record as committed in print to some
design, and b) an absolutely firm guarante that we are out of the business
of considering changes for a long and/or indefinite future, BECAUSE we accept
that the language is sufficiently doen that we feel there will be no need to
consider such changes for a long time.

> If
>people are really not learning because of the impression of instability
>then they are making a mistake.

Are they?  How many conlangs that were even slightly successful did not
fall victim to a seriously sapping schism by people who wanted to "improve"
the language.  Even Esperanto, was hit by Ido, and I personally am
convinced that Ido did not fail because of the virtue of Esperanto, but
because of external factors like World War I, that damaged all conlangs and
left only the strongest surviving.

> I don't remember anyone posting to the list saying that they are
>waiting for the baseline to start learning.

That is because these are the people who either don't join the list in the
first place, or who sign off within a few weeks.  There is NOTHING on this list
for people who are waiting.  And the hypertechnical discussion is not only
unintelligible to them, but is a turnoff to their interest in the language.

>Anyone who thinks that learning
>will become easier with a baseline is mistaken, in my opinion.

WE disagree then, because *I* will learn the language easier when it is
 baselined.  Furthermore, I will again be confident enough to try to SELL the
 language,
which I am not now.  It will be easier because people can write teaching
materials that will not become obsolete.  It will be easier because old Lojban
text will remain "state of the art" longer.  And of course it will be
psychoilogically easier to make the commitement.  Mostof the people who will
learn Lojan will NOT learn it in a month or two of intensive work on LOjban
List, but rather in loose time over 2 or 3 or more years.  They want the
 language and language materials that they start with to still be accurate at
 the
end.

>We must be reading different lists! How many grammar changes were discussed
>in the last year? The only one I can think of, besides those made by John,
>is And's {xoi}, which Stivn likes so much. I have no idea where I would get
>material for 75 changes. pc's cmavo proposals for scope markers were never
>explicited as grammar proposals because he never suggested what grammar
>they would have. Probably they would be in UI, so they would not constitute
>a grammar change.

But we don't KNOW what the grammar will be so it therefore is a potential
grammar change.  Until it is ruled out, it has to be treated as a potential
grammar impact.  This design team is a bunch of pessimists who are certain of
the worst case impacts, because far too often anything else has been
too optimistic.

And did not propose just one scope marker.  I think he made hios proposal a
couple of times and changed it.  Long-suffering Lojban Central has to
understand ALL versions of a proposal, not just the final one.  We have to knwo
 what was considered and rejected, and we have to truly understand WHY the
change is being proposed.

ONe other thing is that some non-proposals are still proposals.  Your proposal
to use ke'a for lambda coverage may not require a change to the YACC grammar,
but it requires both dictionary and refgrammar work.  And  how many varieties
of subscripting and prenexing proposals appeared JUST TO DEAL WITH lambda.
Maybe only a half donze, though it seemed like twice that.  Any change that
affects the books affects John's and my work now.  And as John just mentioned
to be today, the problem in writing these thinsg is not the writing itself,
but in deciding what to write about.  That last year's discussion of lo/le
and family will have ENORMOUS impact on what he evenutally writes on the
logic paper, i am sure.

>> And they were only considered because it made it easier to write.
>
>My contention is that my changes do that too. Of course, it would require
>a little rewriting now,

Then they by definition do not make it easier to write nbecause they add to
an already excessive workload.  Not to mention that John has to spend time
reviewing the proposals that could be spent writing.

>but from the point of view of the learner they
>would make it easier to present a simpler language.

I don't think any extension will make the language easier to learn.  You
are presuming that regularities you see are necessarily regaularities that
everyone else will see and understand in just the way you do.  Our continuing
debate over whetheryour proposed ke'/lambda  is consistent with the existing
ke'a should convince you otherwise.  I simply haven;t the foggiest idea why
you see the two any more rrelated than ke'a and zo'e are.
>> BTW, on extension vs change.  Yes we are more permissive of extensions, but
>> they still have been justified in general by saying thats omething COULD NOT
>> be said, and not that it simply had to be said in a more roundabout way.
>
>Are you sure? I would bet that most or all of John's changes were
>unjustifiable by that standard. What is an example of a change that
>allowed to say something new that could not be said in a more roundabout
>way?

Well, the new ju'e proposal allows afterthought causals that do not imply
logical connection.  Since the existing langauge PRESUMED the use of logical
connectives, the eexisting language was likely to MISLEAD.  There might be
roundabout ways to make this clear, but people will choose the shortest and
in this case potentially erroneous way.

The relative clauses proposal (chnage 19) definitely met this criterion.
The current proposal, not yet numbered, O fVeijo;s altering preposed relative
clause grammar (this one has been hanging since Aug 94),remedies that fact
that the "natural" way to use the existing preposed clauses with nested
relative clauses generates center-embedding.

SEveral of John's other propsoals were fixes to the MEX system, which was not
propoerly done when we did the first baseline.  I'd have to look
at the lsit to find other examples.

I think you are interpreting my wording abotu "roundabout way" a little
 differently
thatn I intended.
>> John and I have found, and presume that others find it as well, that the
>> image of instability affects those of us working on learning or using the
>> language FAR MORE than the changes themselves.
>
>Then let's change that image. You should not react so violently to changes
>that have no effect whatsoever on learning

All changes affect learning.
> Or rather, that they make
>learning easier by removing unjustified exceptions.

All such chjanges affect the refgrammar, which generally mentions the
 exceptions.

>That debate was mostly of the "what does this mean" kind, not of the
>"let's introduce new stuff" kind.


No, it was a how to say it debate,w here every new proposal on how to say it
met with some objection.  By the tine the debate was 2 weeks old, I was
 convinced that the existing language was going to turn out to have some
 irremediable
hole in it - that there was somewhere in t6he "this doesn't work" arguments
that were being posted that would sooner or later lead to the conclusion
that there was a real gap in the language.  I can;t say to this day that this
deabte was resolved with the consensus being that allavanues are covered
by the current language.  It is thus for me an open ande throbbing isisue,
one that can;t be put behind us until we know that the isses are all resolved
or with the books out so we can say that the language is done and working
de jure regardles of endless de facto questions.

>But the debate had to do with interpretation of what's there. What sort
>of proposal do you expect?
>

Pc's scope markers.  Changes to the default qiantifiers of the various gadri.
Changes to the definition of "re prenu" that affect a dictionay entry
based on such a phrase.

>> How many changes will the other active users of the language have once
>> they get to the point where they decide to post formal proposals.
>
>Not very many, I would guess.


With only Goran as an excpetion, every Lojbanist who has reached any level of
comfort expressing n the language has made some proposal or another for change.
Oops, maybe Sylvia also.

>> If I had
>> finsihed the dictionary and published it, and you had made these proposals
>> then they would have been flat-out rejected.
>
>I don't see why. Change X6, for example, requires zero rewording in the
>dictionary.


the dictionary has cmavo compounds.  If you greatly increase the number of
varieties of cmavo compounds that can exist, i may feel it appropriate to
add them to the dictionary.

>> Luckily the refgrammar covers a slightly smaller amount of the language than
>> the dictionary, so new lexicon could be proposed that has no grammatical
>> significance.
>
>I beg to disagree. The refgrammar covers a different part of the language
>than the dictionary. There is a lot in it that you would have no way
>of working out from the dictionary.
That may be true, but any new or changed cmavo or compound potentially affects
the refgrammar.  The new interpretation of "me" for example, affects both
refgrammar and dictionary.

>> I guess I don't follow this, but if you are saying it needs "na" to be
>> accepted why couldn't you just use "ja'a" instead of "nana".
>
>I used {nana} because it is more ridiculous, so it makes the point better
>than {ja'a}. They are both two syllables long anyway.


But one is a single camvo with a succinct grammar.  The other necessarily
implies two levesl of nested grammar.

>> >so that I don't have to say {co'a nana ze'a broda} instead of {co'a ze'a
>> >broda}
>> I guess I don't follow this, but if you are saying it needs "na" to be
>> accepted why couldn't you just use "ja'a" instead of "nana".
>
>I used {nana} because it is more ridiculous, so it makes the point better
>than {ja'a}. They are both two syllables long anyway.
>
>You can have as many tags attached to a selbri as long as you put some
>NAs in between them. I think that should not be allowed, but that would
>be a real change, not an extension, so I will leave it to John to propose
>it. (I don't remember whether he explains this "feature" in the papers.)

The "feature" is a side effect of the intension for the current rule.

The question at the tie was whether you needed to have the na first or the
tense/stag first.  We couldn;t make up our minds, and we found out that
YACC would allow both.  So the existing grammar is itself one of those
"extensions" you say has no impact on the language.  But it turned out that it
did, because by permitting this, it caused co'a naze'a broda to
be allowed but not co'a ze'a broda.  Now you propose a new change to extend the
extension stil further, but this extension might elad to some other
irregularity later, as someone thinks of something that you did allow, but
also comes up with an option you did not allow because they see a different
"big picture" than you did.

REmind me BTW, why it has to be co'a ze'a broda and not ze'a co'a broda.
If I clearly understand the problem, i may see an alternate solution.

>> > My proposal
>> >is to tell the parser that this means what humans will think it means,
>> >a compound tense, and not {co'aku ze'a broda}, as the parser currently
>> >believes.
>> Except that BOTH mean "co'aku ze'aku", at kleast they used to %^).
>
>If {co'a ze'a broda} means {co'aku ze'aku broda} and
>{ze'a co'a broda} means {ze'aku co'aku broda}, then why does the
>parser read them differently?

Because that is what YACC does.  Sone things YACC can handle; other things
it cannot.  But if the things that YACC handles transform consistently
into forms parallel to thge ones that cannot be handled (as was the intent)
Then the co'a na ze'a form is just an accidental abbreviation short cut that
fell out of the rules but was not especially necessary.

>> I could perhaps test your proposal by putting in unlikely values and seeing
>> how much nonsense I can generate %^).
>
>Please do. You will see that you can only generate benign nonsense, of
>the sort of "long short stick", which makes sense given an appropriate
>context, and not malign nonsense of the {li pipai} sort.
>

We used to let the random sentence generator do this, but it proved too much
work to update it even for the much rarer changes John proposedm and is still
back at something like the 2.08 grammar.

lojbab