Message-Id: <199310131708.AA00614@eli.CS.YALE.EDU>
Date:         Wed, 13 Oct 1993 13:05:45 -0400
Reply-To: Logical Language Group <lojbab@ACCESS.DIGEX.NET>
Sender: Lojban list <LOJBAN%CUVMB.bitnet@YaleVM.YCC.YALE.EDU>
From: Logical Language Group <lojbab@ACCESS.DIGEX.NET>
Subject:      Re: TECH: Lean Lujvo and fat gismu
To: Erik Rauch <rauch>
In-Reply-To:  <199310131036.AA04293@access.digex.net> from "Logical Language               Group" at Oct 13, 93 06:36:04 am
Status: RO

mi'e .djan. .i la lojbab. cusku di'e

> Now, given the formulaic convention, replace "cau" above with "zi'o" and you
> have a convention that no longer has 2 interpretations.  This solves the
> problem and gives a much friendlier use for zi'o than bare places.  But then,
> this causes me to re-examoine my opposition to using zi'o directly on the
> converter, and I now chnage my mind and decide I like "selzi'oklama" better
> since it puts the focus of the deletion on the term being deleted.
>
> What the place structure of selzi'oklama?  Since the zi'o is joined to the
> "sel" first, it is deleted and klama is not converted, so x1 is kl1, x2 is kl3
> x3 is kl4, x4 is kl5.

This might make sense if "zi'o" belonged to UI, or otherwise attached to the
previous item.  But it doesn't and it can't.  I can only read "selzilklama"
with place structure k3 k2 k4 k5, with k1 down the tubes.
(It can't be "selzi'oklama" anyhow, because "-zi'o-" is the rafsi for "dzipo".)

Instead, we established a convention of using numeric rafsi after "-zil-",
to serve as quasi subscripts.  Thus "zilpavklama" is k2 k3 k4 k5, whereas
"zilrelklama" is k1 k3 k4 k5.  Yes, it's ugly, but it works for arbitrarily
many places: "zilxavnunklama" gives n1 k1 k2 k3 k4, with k5 lost.  "zilklama"
then becomes a vague lujvo: some place is dropped, but we don't know which.

--
John Cowan              sharing account <lojbab@access.digex.net> for now
                e'osai ko sarji la lojban.