Date: Tue, 20 Dec 1994 11:54:56 -0500
From: Logical Language Group <lojbab@ACCESS.DIGEX.NET>
Subject: Re: fyfyfyfy
In-reply-to: <199412200920.AA14766@access1.digex.net> from
 "Logical Language Group" at Dec 20, 94 04:20:55 am
Sender: Lojban list <LOJBAN@CUVMB.BITNET>
Reply-to: Logical Language Group <lojbab@ACCESS.DIGEX.NET>
Message-id: <01HKVGR27OTU0000JE@FIPORT.FUNET.FI>
Content-transfer-encoding: 7BIT
To:         Lojban List <lojban@cuvmb.cc.columbia.edu>
MIME-Version: 1.0
Content-Type: text/plain; charset="us-ascii"
Content-Transfer-Encoding: 7bit
Content-Length: 1405
Lines: 30

la lojbab. cusku di'e

> I tried this on the parser, and suspect a problem.
> It does not break up fyfyfyfy before the selbri, nor implies that ti takes
> it as a sumti.  When followed by a separate fy "fyfyfyfy fy klama", it blows
> up.  But "fy fy fy fy klama" parses fine, and with a single sumti, unless
> split up by BOI.

There are two problems here: one is with the current machine parser's
feeble morphology algorithm, one with what I said.

The parser distinguishes brivla from compound cmavo by looking for
consonant clusters, defined as two successive consonants optionally
separated by "y".  By that standard, the word "fyfyfyfy" appears to have
a y-hyphenated consonant cluster in the first 3 letters.  Thus it is
lexed as a brivla, with resulting problems.  IMAO, this is no worse that
the treatment of "secmene" as a brivla rather than as "se cmene",
or the fact that the lexer breaks up stuff within "zoi" quotes as if it
were Lojban.  Eventually, there will be a proper morphological preprocessor
that handles all cases.

However, I was wrong to say that "fy fy fy fy" was four sumti; it is a
single sumti, because the lerfu-word pro-sumti actually consist of a string
of lerfu words.  To get four instances of the "f" pro-sumti, we need
"fyboi fyboi fyboi fy[boi]".

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