From phma@ixazon.dynip.com Wed Apr 30 15:32:23 2003
Received: with ECARTIS (v1.0.0; list lojban-list); Wed, 30 Apr 2003 15:32:24 -0700 (PDT)
Received: from 208-150-110-21-adsl.precisionet.net ([208.150.110.21] helo=blackcat.ixazon.lan)
	by digitalkingdom.org with esmtp (Exim 4.12)
	id 19B07d-0005hp-00
	for lojban-list@lojban.org; Wed, 30 Apr 2003 15:32:18 -0700
Received: by blackcat.ixazon.lan (Postfix, from userid 1001)
	id 137633E95; Wed, 30 Apr 2003 18:31:57 -0400 (EDT)
From: Pierre Abbat <phma@webjockey.net>
Organization: dis
To: lojban-list@lojban.org
Subject: [lojban] Re: nai in UI (was: BPFK phpbb)
Date: Wed, 30 Apr 2003 18:31:56 -0400
User-Agent: KMail/1.5
References: <Pine.LNX.4.44.0304300906370.7794-100000@simba.math.ucla.edu>
In-Reply-To: <Pine.LNX.4.44.0304300906370.7794-100000@simba.math.ucla.edu>
MIME-Version: 1.0
Content-Type: text/plain; charset="iso-8859-1"
Content-Disposition: inline
Message-Id: <200304301831.56527.phma@webjockey.net>
X-archive-position: 5070
X-ecartis-version: Ecartis v1.0.0
Sender: lojban-list-bounce@lojban.org
Errors-to: lojban-list-bounce@lojban.org
X-original-sender: phma@webjockey.net
Precedence: bulk
Reply-to: lojban-list@lojban.org
X-list: lojban-list

On Wednesday 30 April 2003 12:52, Jim Carter wrote:
> Hey, wait a minute!  Think modularity.  There are morpheme streams that can
> or can't be split into valid Lojban words.  There are word streams that can
> or can't be parsed into valid syntax trees.  And there are semantic
> constructs that are or aren't Carroll-esque.  The phrase "meaningful
> concept" is too vague to be a meaningful concept.  I would like to amend
> your pronouncement to say, a valid parser passes all valid word streams
> (putting out a correct syntax tree) and rejects all invalid word streams,
> where validity is judged from the grammar.  In other words, the parser does
> or doesn't truly realize that grammar.  (I.e. makes it real.)  (And as a
> separate module, the semantic analyser may have an opinion about
> jabberwockishness.)
>
> I'm sure we can come up with natlang examples where the parsing depends in
> an essential way on the meaning of the words (not just their syntactic
> category), but I can't think of one so early in the morning.  But that kind
> of a pain in the butt doesn't belong in Lojban.  Think modularity!

I agree (and for similar reasons separated lexing words from checking their 
validity). We have a context-free grammar that parses {kau} as UI, which can 
go anywhere. We can have another layer that checks whether {kau} follows a 
question word, whether it's in an abstraction, and whether it's on the 
indifferent side of {ju/u/gi'u}, and decides whether it makes sense there. It 
can also check whether a brivla has a sumti in a nonexistent place, whether a 
number string is valid, whether {ko'a} or a lervla has an antecedent, etc.

phma