Message-Id: <m0pexK8-0000R2C@xiron.pc.helsinki.fi>
Date:         Thu, 10 Mar 1994 21:53:14 EST
Reply-To:     jorge@PHYAST.PITT.EDU
Sender:       Lojban list <LOJBAN%CUVMB.bitnet@FINHUTC.hut.fi>
From:         Jorge Llambias <jorge@PHYAST.PITT.EDU>
Subject:      Mad Proposals II: The watered down version.
To:           Veijo Vilva <veion@XIRON.PC.HELSINKI.FI>
Content-Length: 4427
Lines: 122

Thanks to Art, And and Matthew for their responses. It seems we
all agree that it would be nice, but maybe too big a change at
this stage of the project.

I agree with And that the proposals don't stand a chance, but
something as trivial as that won't stop me from proposing a
second version which is not quite so elegant, but still is
worthwhile.

Two things convinced me that the full proposal might not be a
good idea (appart from the political aspects, I'm just looking
at the language-in-itself for now, not the lenguage-and-its-users,
if you see what I mean).

First was Matthew's comment that {gu} changes its meaning, and thus
old texts that use it become wrong. (Old texts using the really
eliminated cmavo do not become wrong, but archaic. With a cmavo
used in a new function however, it's worse.) This can be solved
by using another cmavo instead of {gu} in that function, but it has
to be a CV'V cmavo, and the pleasing symmetry of {gi}-{gu} is lost.

The second reason came out of some comments that lojbab sent me.
This problem already exists for the {joi}'s, but nobody seems
to mind because nobody seems to use them in forethought.
It's this:

When you start a sentence with

        .i joi ...

there are two possibilities: either you're connecting this sentence
non-logically to the previous one, or you're starting a non-logical
forethought connection, in which case you'd continue:

        .i joi gi ...

This, of course, won't confuse the parser, but if the gi takes some
time in coming, it will probably make the human listener have to make
some jumps in mid-understanding, and this can be confusing when you're
at the same time working out truth tables in your head.

The GAs save the day for the logical connectives, because

        .i je ...

won't be confused with

        .i ge ...

If we add this two reasons together, we are forced to admit that the
GAs have to stay. Thus MAD PROPOSAL NUMBER 1 part B is hereby withdrawn.

I have not heard convincing arguments (lenguage-in-itself-wise) against
the other four proposals, but I will argue now that No.3 may not be
worth bothering with.

This one dealt with the GUhAs, which (I don't know if I've mentioned this
before) serve a pretty useless function, namely forethought tanru
logical connection. I'm now persuaded that the best thing is to ignore
them and let them disappear from the language simply from not being used,
rather than find a replacement for them. Forethought tanru non-logical
connection, which my proposal made possible, is almost as useless, and
therefore nobody will miss it if it's not provided.

With this explanation, MAD PROPOSAL NUMBER 3 is hereby withdrawn.

I'm still in favour of the remaining three, which I'll repeat here
(without giving all the arguments I gave before):


MAD PROPOSAL NUMBER 1:

Extend the grammar of {je} connectors to that of {joi} connectors.

COMMENT: This means that GAs and As have an equivalent JA form, but
         both forms are legal. No relearning required whatsoever.
         Simply a natural extension.


MAD PROPOSAL NUMBER 2:

Replace {gi'e}'s by {gije}'s, and allow {gijoi}'s for the same function.

COMMENT: The relearning effort for this is almost zero.


MAD PROPOSAL NUMBER 4:

Place {ji} in selmaho JA

COMMENT: Just to make things nice. The meaning of {ji} does not change
         by doing this, it's simply extended. No relearning, other than
         to remember those {ku}s.

----------------------------------------------------------------------------

The connectors would then be

        proposed:               today's:

        ....   je ....          ....   je ....          for tanru
        ....   je ....          ....   .e ....          for sumti
        .... .ije ....          .... .ije ....          for bridi
        .... gije ....          .... gi'e ....          for bridi-tails

        ....   joi ....         ....   joi ....         for tanru
        ....   joi ....         ....   joi ....         for sumti
        .... .ijoi ....         .... .ijoi ....         for bridi
        .... gijoi ....          not possible           for bridi-tails

         je gi.... gi....           ge .... gi....
        joi gi.... gi....       joi gi .... gi....

{.e} and {ge.... gi...} would also be allowed for convenience.


(And GUhAs don't exist, as far as I'm concerned :)


All comments wellcome, especially if there's some obvious flaw I missed.

Jorge