Message-Id: <m0pe2eX-0000R2C@xiron.pc.helsinki.fi>
Date:         Mon, 7 Mar 1994 22:32:51 EST
Reply-To:     jorge@PHYAST.PITT.EDU
Sender:       Lojban list <LOJBAN%CUVMB.bitnet@FINHUTC.hut.fi>
From:         Jorge Llambias <jorge@PHYAST.PITT.EDU>
Subject:      The Mad Proposals
To:           Veijo Vilva <veion@XIRON.PC.HELSINKI.FI>
Content-Length: 6820
Lines: 202

This posting contains proposals regarding some extensions and/or
replacements to the logical and non-logical connectives system.

Lojban Central is unanimously opposed to it, but does not mind my
posting it for your perusal. If public opinion is strongly in
favour they may consider it again (they say), so get paper and pen
ready and start writing to your senator.

I will also present some of the counterarguments to the proposal, but
you can be certain that it will be in a biased manner.

I will use {joi} to refer to all the non-logical connectives of
selmaho JOI, {je} for all of selmaho JA, etc. Since by definition
all the members of a selmaho have the same grammar, this makes
the discussion easier.

(The proposals are called "mad" for historical reasons.)


MAD PROPOSAL NUMBER 1:

Allow {je} everywhere that {joi} is allowed.

RATIONALE:

{je} and {joi}, the basic logical and non-logical connectives
(BIAS ALARM: {je} is being called the basic logical connective),
have very similar grammar. However, there are a few places where
{joi} is allowed, but {je} is not.

This restriction doesn't stop us from being able to say anything,
because there are other structures provided to cope with those functions.

For instance:

        le xunre joi blanu bolci (the red mixed-and blue ball)
        le xunre je blanu bolci (the red and blue ball)

are both legal, but:

        mi joi do
        *mi je do

The second one is illegal. The corresponding grammatical structure is:

        mi .e do

Why is the {je} form illegal? Because to link sumti in general, we'd
have to use lots of {ku}'s:

        le ninmu ku joi le nanmu
vs.
        le ninmu .e le nanmu

with {je} we'd have to use {ku}, just like we do with {joi}:

        *le ninmu ku je le nanmu

My point is that, since we have to use {ku}'s with {joi}'s anyway,
why not allow the {je} versions to be legal. The {.e} version would
still be there when needed.

(At this point I should say that John has ran the proposals through
the YACC, and there were no problems with that.)

One argument against, is that people will generalize from {mi je do}
to {le ninmu je le nanmu}. My response to that is that then people
will generalize from {mi joi do} to {le ninmu joi le nanmu}, so that
is not a new problem. The reply that {joi}'s are less central to Lojban
is not convincing to me, because I think that in real speech, logical
connectives are not more significant than non-logical ones.

(BTW, the ku-less form is illegal only because the parser can't handle
it, not because it generates any ambiguity.)

In short: Proposal 1 is not a change, but simply an extension that
removes an unnecessary restriction. It goes well with the many stones
for one bird philosophy, because it allows more than one way to say
the same thing. It doesn't introduce any weird interpretation of
anything, it's a natural extension that I bet fluent speakers will
make, whether the parser likes it or not.


MAD PROPOSAL NUMBER 1 part B:

Eliminate selmaho GA.

RATIONALE:

They become redundant. Just like {joigi} serves currently as the
forethought non-logical connective, {jegi} would do for the logical one.

The trade-off for the simplicity (we'll end up with only one series
of logical connectors when we're through, instead of the current 5)
is that they have two syllables instead of one.

My argument was that since they're forethought connectives, that
doesn't matter: people are supposed to think more and thus take
more time when using forethought. This didn't seem to be convincing
enough, though.


MAD PROPOSAL NUMBER 2:

Replace {gi'e}'s by {gije}'s, and extend the grammar to allow {gijoi}'s
in the same function of bridi-tail connection.

RATIONALE:

A whole selmaho with 5 cmavo, of type {gi'e}, is eliminated, and replaced
by the almost identical compound cmavo of type {gije}, with the same
number of syllables. This requires almost no relearning.

As a bonus, afterthought bridi-tail connection (this is what {gi'e}'s do)
is also possible for the non-logical connectives.

Also, the afterthought form is made to look just like the one for whole
bridi:

.ije  (for a whole bridi)
gije  (for bridi-tail)

same as

.ijoi (existing)
gijoi (currently not possible)


MAD PROPOSAL NUMBER 3:

Eliminate {gu'e}'s and replace them by {guje}'s. Also allow {gujoi}'s
to fullfill the same function for non-logical connectives.

RATIONALE:

Again, 5 cmavo eliminated. This time a new one {gu}, parallel to {gi}
is introduced (we eliminated its previous function in 1b, right?)

In case you don't remember, {gu'e}'s serve as forethought connectives
within tanru. IMHO a useless construction, but since they're there, we
provide them with a substitute.

{guje}'s would be most similar to the current form, but {jegu}'s
would be the logical choice. Then forethought connectives would all
be of the same form, instead of today's variety:

je gi .... gi ....  (instead of    ge ... gi ...)
joigi .... gi ....  (as is now  joigi ... gi ...)

je gu .... gu ....  (instead of  gu'e ... gi ...)
joigu .... gu ....  (no current equivalent)


MAD PROPOSAL NUMBER 4:

Change {ji} from selmaho A to selamho JA

RATIONALE:

This one is just to round everything off. The question connective becomes
regular.


-------------------- END OF PROPOSALS ----------------------

The net result is:

- Instead of five series of logical connectives: A, JA, GA, GIhA, GUhA
  we are left with a single one: JA (plus A in case we want to spare ku's,
  but cut the flow of our speech)

- We eliminate 15 (yes, you read correctly: fifteen!) cmavo:
  ga, ge, go, je'i, ge'i, gi'a, gi'e, gi'i, gi'o, gi'u, gu'a,
  gu'e, gu'i, gu'o, gu'u.

- We lose no expressive power. Every feature of the current language is
  preserved, and the only small drawback is two syllables instead of one
  for the forethought logical connectives.

- We gain two previously non-existing forms for the non-logical
  connectives: afterthought bridi-tail, and forethought tanru.

- As a result, logical and non-logical connectives are equal, and there
  is complete regularity between them. They could almost be put into the
  same selmaho if it wasn't for a small difference in how {na} and {nai}
  affect them.

The price to pay is that people who have already learned the complicated
system have to forget it and learn the simple one. (What? that's not
an unbiased way to say it? :) For those who are still learning, and for
the future generations, it would be a great gain.

If you are in favour of regularity and simplicity, now is the time to
speak up!

(I'm considering becoming a preacher.)

Jorge

(The fact that this proposal is presented while Colin is off the list,
and I suspect he would be against any changes, is purely coincidental.
The fact that Nick is also not connected is also coincidental. Really!)