From LOJBAN%CUVMB.BITNET@uga.cc.uga.edu Fri Nov  4 03:09:06 1994
Message-Id: <199411040809.AA16296@nfs1.digex.net>
Date: Fri Nov 4 03:09:06 1994
From: jorge@PHYAST.PITT.EDU
Subject:      Re: singular and plural
Status: RO

Bob Chassell says responding to me:

>    In any case, I'm glad that I finally understand where so much emphasis
>    on veridicality comes from. If that was the original distinction between
>    {le} and {lo} it was killed as the defining distinction when {lo} was
>    given the {su'o} quantifier.
>
> The {su'o} quantifier is strictly an *heuristic*.  It is a guide to
> what people are likely to mean when they talk about `that which is
> real' in the context.  In the context of everyday, practical
> conversation, people don't usually consider statements to be veridical
> if they are about that which is non-existant, even when in a context
> of strict logic, the statement may be non-false about a non-existant
> instance.

I'm not sure how this relates to whether {lo} is specific or not. The
quantifier {su'o} means "at least one", but in no way does it say which
one. If that is the outer quantifier of {lo}, then {lo} can't be specific.

> This need to distinguish or specify context also explains why Lojban
> gets defined as dialogue-based, rather than monologue-based;
> otherwise, the monologist gets to act as a sophomore and catch you on
> mis-defined contexts, like my friends who answer `Yes' when I ask them
> `Do you know the time?'

And then people say English is not very context-dependant. :) In some
contexts, it would be right to interpret the question literally, in
others, it wouldn't.

> It is worth repeating: in Lojban, in contrast to English, number is
> not automatically defined by the grammar, although the aspect of
> veridicality is.  Number is simply not a necessary part of the Lojban
> statement.  English almost always requires that you express number,
> although there are sentences that do not require that, such as `The
> deer ran from the sheep.'

It is also worth repeating that nobody is challenging that. Even the
specific {le} doesn't mark number, so the question of number and
specificity are really orthogonal. I'm sorry I used "singular" as
the opposite of "general". I repeat, I didn't mean "singular" as the
opposite of "plural". (In the context of my reply to pc, it should
have been clear, since I cited a few lines before where he defined
"singular". More context dependance in English :)


> Here is a practical example involving {lo}.  I say:
>
>     .i mi pu viska lo bunre gerku
>
> Given what little joint context we have, you figure: this is not a
> story; on the other hand, this fellow is not likely to be uttering
> logical truths involving non-existance.  Also, since he is using {lo},
> he is not talking about something that might be designated as a brown
> dog, such as a stuffed toy.  Therefore, he probably is trying to say
> that he saw _at_least_one_ real, brown dog.  If the number is
> important to us, we can also figure (less confidently), that he saw no
> more than one dog, unless he tells us otherwise.

I agree, he saw "at least one brown dog". In no way did he tell us which
dog he saw. If we were in a room, and there was a brown dog with us, we
still wouldn't know whether he was talking about that brown dog or not.
(Unless {lo} is specific, in which case it would almost surely be that
dog, but that would mean practically the same as {mi pu viska le bunre
gerku}.)

> Interestingly enough, this secondary regard for number arose in the
> program I wrote to format the new, big English/Lojban Dictionary.  In
> that program, I wrote an expression to find the beginning of the field
> containing the gismu and its rafsi.
>
> A part of the expression looks like this:
>
>     ...search-forward "/=/" ...
>
> This expression does not specify whether the program is looking for
> one or more than one marker.  But the program is looking for a
> manifestation of _at_least_one_ marker that `really' is "/=/". In
> Lojban, I would use {loi} to indicate this.  Depending on how I run
> the program, whether on a test case of one or two lines, or on all
> 15,000 lines, the program deals with the `more than one' case.

I'm not sure I understand the example. The problem with computer
languages is that they consist only of commands. The imperative mode
is tricky in regard to this quantification issue, so that makes it
even harder. To make things worse, "to search" is one of those verbs
that like opaqueness. I think I would translate it as

        ko crane sisku ro la'o xy /=/ xy
        Search-forward for every "/=/".

If you use {loi}, it would search for some portion of the mass of "/=/"
and then give up, but I suspect that you want it to search for each and
every one.


>     Robert J. Chassell               bob@gnu.ai.mit.edu
>     25 Rattlesnake Mountain Road     bob@grackle.stockbridge.ma.us
>     Stockbridge, MA 01262-0693 USA   (413) 298-4725

co'o mi'e xorxes