Message-Id: <199601190616.IAA28744@xiron.pc.helsinki.fi>
Date:         Fri, 19 Jan 1996 01:14:33 -0500
Reply-To: Logical Language Group <lojbab@ACCESS.DIGEX.NET>
Sender: Lojban list <LOJBAN@CUVMB.BITNET>
From: Logical Language Group <lojbab@ACCESS.DIGEX.NET>
Subject:      Re: TECH: tosmabru test rationale (PU: TECH QUERY: variant
              fu'ivla)
To:         pbarreto@UNISYS.COM.BR
Cc:         lojban@cuvmb.cc.columbia.edu
Content-Length: 1223
Lines: 28

>cu'u la dn. joi mi
>
>>>the proposed form cannot be confused because of the
>>>anti-tosmabru hyphen
>>
>>.i tosmabru ki'a .i .uanaisai .i lo mabru cu danlu .i zo tos. rafsi ma
>
>That's not Lojban. It's an instance of the flaw Nora LeChevalier found in
>Loglan, motivating the hyphen rule in section 11 of the morphology paper.
>
>I'm not surprised at your question. In fact, I'm glad about that (.ui).
>Without the hyphen rule, "tosmabru" would be ambiguously analyzed into
>either "to sma,bru" or "tos,mabru" (the correct form, if there existed a
>"tos" rafsi at all, would be "tosymabru").


Minor correction - JCB found and knew about the tosmabru/tosymabru problem
with TLI Loglan (though he was not too good at explaining it).  Nora found
ANOTHER word form that broke apart even though it appeared to meet the
already defined-by-JCB tosmabru test.  This is a wordform like
CVCCVCyCVCCV which is intended to be compounded from CVC+CVC+y+CVCCV
but instead breaks down into CV CCVCy+CVCCV.

(At least I think that is the wordform.  See discussions of the lujvo
making algorithm and there should be TWO cases dealt with, and Nora found
the second one, which JCB has never publically acknowledged or dealt with.)

lojbab