From phma@phma.optus.nu Tue Mar 09 14:00:49 2010
Received: from cdptpa-omtalb.mail.rr.com ([75.180.132.121])
	by chain.digitalkingdom.org with esmtp (Exim 4.71)
	(envelope-from <phma@phma.optus.nu>)
	id 1Np7Th-0006ot-7F
	for lojban-list@lojban.org; Tue, 09 Mar 2010 14:00:49 -0800
X-Authority-Analysis: v=1.0 c=1 a=y9PgM0RH17AA:10 a=ORa4HqFjfvEA:10 a=8nJEP1OIZ-IA:10 a=6I8iWPodvRE6TEUx_zMA:9 a=_fMOtG2K_ArVE5tWXifpnSyxsO4A:4 a=wPNLvfGTeEIA:10
X-Cloudmark-Score: 0
X-Originating-IP: 71.71.198.100
Received: from [71.71.198.100] ([71.71.198.100:48774] helo=chausie)
	by cdptpa-oedge03.mail.rr.com (envelope-from <phma@phma.optus.nu>)
	(ecelerity 2.2.2.39 r()) with ESMTP
	id 7F/1B-21054-BF4C69B4; Tue, 09 Mar 2010 22:00:28 +0000
Received: from localhost (localhost [127.0.0.1])
	by chausie (Postfix) with ESMTP id E2CB65EB7
	for <lojban-list@lojban.org>; Tue,  9 Mar 2010 17:00:26 -0500 (EST)
From: Pierre Abbat <phma@phma.optus.nu>
To: lojban-list@lojban.org
Subject: Re: [lojban] Re: stage-4 fu'ivla construction exception?
Date: Tue, 9 Mar 2010 17:00:21 -0500
User-Agent: KMail/1.9.6 (enterprise 0.20070907.709405)
References: <96f789a61003090826i27a6531eob88df0e3bfce7e14@mail.gmail.com> <925d17561003090917y2816542eh14ea3b879ea2259@mail.gmail.com>
In-Reply-To: <925d17561003090917y2816542eh14ea3b879ea2259@mail.gmail.com>
MIME-Version: 1.0
Content-Type: text/plain;
  charset="iso-8859-1"
Content-Transfer-Encoding: quoted-printable
Content-Disposition: inline
Message-Id: <201003091700.24760.phma@phma.optus.nu>

On Tuesday 09 March 2010 12:17:21 Jorge Llamb=EDas wrote:
> Yes, Pierre has pointed out this oversight in CLL before. I will add
> it to the errata page.

The example I came up with is "lerldjamo" (one of the Korean letters that a=
re=20
assembled into syllables), though one can avoid the problem for this word b=
y=20
saying "lerfrdjamo". "jikru" has no short rafsi.

> Independently of this situation, Stage-3 fu'ivla are not always
> unambiguously analyzable into classifier rafsi plus the rest. If you
> follow the CLL rules, any CVCrV or CVnrV gismu used as classifier is
> subject to this type of ambiguity. The "unwanted" reading, however, is
> very unlikely to occur, because the borrowed word would have to start
> with an unlikely consonant cluster. (like "ldjine" in your case).
>
> The way I have proposed to solve this is to add a rule that says that
> the lojbanized part of the fu'ivla, before adding the classifier, has
> to consist of valid lojban syllables (i.e., it has to start with a
> valid initial), so "jikrldjine" can only be analyzed as jikr-l-djine.

The way I think it should be resolved, and the way I think jbofi'e does it,=
 is=20
that if a stage-3 fu'ivla can be analyzed into either 3-letter rafsi +=20
foreign part or 4-letter rafsi + foreign part, use the 4-letter rafsi. If t=
he=20
rule specified the 3-letter rafsi, then there are some combinations of gism=
u=20
and foreign part for which it would be impossible to form a fu'ivla.=20
(badnrnsete, badrntriste, badrnabri) There are some foreign parts which beg=
in=20
with invalid initial clusters for which it would be nice to be able to form=
=20
fu'ivla, such as "gbe" ("bangrgbe"), a subfamily of African languages.

Pierre
=2D-=20
La sal en el mar es m=E1s que en la sangre.
Le sel dans la mer est plus que dans le sang.