From @uga.cc.uga.edu:lojban@cuvmb.bitnet Tue May 30 23:59:52 1995
Received: from punt2.demon.co.uk by stryx.demon.co.uk with SMTP
	id AA3179 ; Tue, 30 May 95 23:59:45 BST
Received: from punt2.demon.co.uk via puntmail for ia@stryx.demon.co.uk;
          Tue, 30 May 95 11:36:28 GMT
Received: from uga.cc.uga.edu by punt2.demon.co.uk id aa13751;
          30 May 95 12:35 +0100
Received: from UGA.CC.UGA.EDU by uga.cc.uga.edu (IBM VM SMTP V2R2)
   with BSMTP id 7457; Tue, 30 May 95 07:33:40 EDT
Received: from UGA.CC.UGA.EDU (NJE origin LISTSERV@UGA) by UGA.CC.UGA.EDU (LMail V1.2a/1.8a) with BSMTP id 8247; Tue, 30 May 1995 07:33:40 -0400
Date:         Tue, 30 May 1995 07:33:36 -0400
Reply-To:     Logical Language Group <lojbab@access.digex.net>
Sender:       Lojban list <LOJBAN%CUVMB.BITNET@uga.cc.uga.edu>
From:         Logical Language Group <lojbab@access.digex.net>
Subject:      inner quantifiers and clauses - cimei again
X-To:         lojban@cuvmb.cc.columbia.edu
To:           Iain Alexander <ia@stryx.demon.co.uk>
Message-ID:  <9505301235.aa13751@punt2.demon.co.uk>
Status: R

You guys are probably way past me on this discussion, but I'll keep
playing:

>Hold on, let's back up.  I was intending to give alternatives for "There
>are _exactly_ three men in the room", but I think most of my
>alternatives fell short, saying, at most, "there are _at least_ three
>men in the room".  Which of the following work?  What the best way to
>say this?
>
>        lo ci nanmu pe ne'i le kumfa cu ce'o
>        .i le ni lo nanmu cu nenri le kumfa cu du ci
>        .i piro loi nanmu pe ne'i le kumfa cu cimei

How about
         lo'i nanmu poi nenri le kumfa cu se cimei
         The set of men which are in the room are a set with 3 members.


>(This leads me to wonder what the consequences of changing
>sumti-tail-1<112> = [quantifier] selbri [relative-clauses] | quantifier sumti
>to
>sumti-tail-1<112> = [quantifier] selbri | quantifier sumti
>in the BNF would be, other than making {le se cusku pe mi ku}
>ungrammatical.)

I hate to try to figure consequence of changes in the BNF as opposed to
the YACC grammar.  The latter is the 'real thing', while the BNF is
derived.


<description_110         :  LA_558  sumti_tail_111  gap_450
<                        |  LE_562  sumti_tail_111  gap_450
<                        ;
<
<sumti_tail_111          :  sumti_tail_A_112
<                           /* inner-quantified sumti relative clause */
<                        |  relative_clauses_121  sumti_tail_A_112
<                           /* pseudo-possessive
<                              (an abbreviated inner restriction);
<                              note that sumti cannot be quantified */
<                        |  sumti_F_96  sumti_tail_A_112
<                           /* pseudo-possessive with outer restriction */
<                        |  sumti_F_96  relative_clauses_121  sumti_tail_A_112
<                        ;
<
<sumti_tail_A_112        :  selbri_130
<                        |  selbri_130  relative_clauses_121
<                           /* explicit inner quantifier */
<                        |  quantifier_300  selbri_130
<                           /* quantifier both internal to a description,
<                              and external to a sumti thereby made specific */
<                        |  quantifier_300  selbri_130  relative_clauses_121
<                        |  quantifier_300  sumti_90
<                        ;

Among other things, it would appear that it eliminates all postposed and
afterthought relative clauses - not just "pe", but "goi", and "poi" and
"noi" - *within* a description.  Note that you can still put them on
"outside" the descriptive sumti per the rules in sumti_94 thru _96.

This distinction between outer and inner restrictive clauses is subtle
and a bit too much to repeat here.  You can almost certainly find an
extensive discussion of it in the List archives (and if Cowan has the
appropriate reference grammar paper on relative clauses done, it will be
in there).  It was far enough back that it probably is in the WWW index
archives.

>As you see from the list, I backed off from much of what I said.  But in
>any case, I wasn't proposing anything as drastic as what you seem to
>think:  I wanted to rewrite the grammar, preserving all (or almost all)
>observed behaviour.

That sounds like quite a challenge; I won't say it is impossible, but
you probably need to have a lot of practice doing computer language
grammars with YACC to have a shot - the people who have gone before you
put in lots of hours to get it where it is today, and the sumti grammar
was among the hardest to get right.  You need a version of YACC that
will handle the full grammar (not all versions will necessarily do so).
And you need a LOT of patience and practice to get any given Lojban
grammar to be unambiguous (no s/r or r/r errors).  Don't waste time with
trying to rewrite the E-BNF.  It isn't clean unless YACC says it is
clean, and the EBNF tells us nothing about what YACC will say.

And, of course, as Cowan noted, if there are any differences between
what you come up with and what is current, in terms of either what is a
valid expression, and how that expression groups/parses, you need to be
able to assert that the old form is in some way invalid (not generally
useful is probably insufficient - it must NEVER be useful, and must be
somehow invalid).  This takes considerable mastery of the subtleties of
the language.

On ke'a in abstractions:
>{pe'i} The only issue is then a pragmatic one:  whether this will
>conflict with the relative clause use of {ke'a} too often.

I would rather assign a new cmavo.  Better safe (and unambiguous) than
sorry.  This is an example of one of those subtleties of the language.
ke'a might rarely be ambiguous, but it could be in some cases, and if
the frequency of occurance of the problems is rare enough and the
solution not intuitively obvious, my instincts tell me that it is not
right for the language.  If there needs to be a pronoun to serve the
role Jorge wants for ke'a in abstract sumti, the justification is
logical precision, and a use of ke'a which could have two meanings is
logically imprecise.

For the same reason, I do not think people will use subscripted ri/ra/ru
in contexts of logical precision, because the counting back is not
defined in a logical precise manner.  Likewise tanru are not suitable
for logical precision.

>The position of an event can be obtained with jai, if I read the
>intention OK.  The place I thought at is lo jai bu'u mi pensi.  If
>that's what you meant by position abstract.

That isn't grammatical, and since you gave no translation, I am not sure
what was intended. lomi jaibu'u pensi does work. jaibu'u has the grammar
of SE in a description, and so what you had was grammatically something
like *le se mi pensi.

ni'o
>Sometime I'm going to try doing some serious mathematics in Lojban; then
>we'll see how well the principle holds up...

Actually, you need to compare TALKING ABOUT mathematics in English vs.
in Lojban.  Writing the equations symbolically will always be preferable
in print.  Any mathematical text you attempt to write in Lojban should
be matched by the same effort in English text (and then read your text
aloud to someone and see if they can reconstruct the symbolics).


lojbab