Message-Id: <199308271728.AA09280@eli.CS.YALE.EDU>
Date:         Fri, 27 Aug 1993 08:45:33 BST
Reply-To: I.Alexander.bra0125@oasis.icl.co.uk
Sender: Lojban list <LOJBAN%CUVMB.bitnet@YaleVM.YCC.YALE.EDU>
From: Iain Alexander <I.Alexander.bra0125@OASIS.ICL.CO.UK>
Subject:      Re: TECH: Desperately seeking [properties]
To: Erik Rauch <rauch>
Status: O

zoi glico. RE:  TECH:  Desperately seeking ... .glico
.i la djan. spuda tu'a mi zoi glico.
> > One of the things I want to be able to say is
> >
> >         mi sisku la djan.
> >         I'm looking for John.

> The easy way to do that with the new place structure is:

>         mi sisku tu'a la djan.
>         I'm looking for an object with the property of being ...
>                 (insert some abstraction about John here).
.glico .i mi spuda fi zoi glico.

I _think_ I buy that.

Obviously you and lojbab came away with different perceptions
of what was to happen to the place structure of {sisku}.
When this all settles down we'd better make sure he understands
the same as the rest of us this time. :)

.glico .i la djan. cusku zoi glico.
> This sounds a little strange.  If you are a detective, then you want
> "lo catra be ...", the one who actually did the deed: the sparrow.
> (Maybe there were other killers too, but you want at-least-one of them).

> OTOH, if you are just using "le catra be ..." as a conventional tag for the
> sparrow, then you >can< identify him.  Indeed, saying "le catra" >is< an
> identification.
.glico .i mi spuda fi zoi glico.

Hmm, you could be right there.  Still, I think there's a discussion
in here about what it means to _identify_ something, but it's
a bit deep for me to cope with at the moment.  I may come back
to it some other time.

.glico .i la djan. cusku zoi glico.
> You never have anyone particular in mind unless you use an in-mind article
> like "le", or an in-mind relative clause beginning with "voi", or something
> equivalent to one of those.
.glico .i mi spuda fi zoi glico.

Let's cut to the chase.

If we _have_ to use {le} or the equivalent when we have
someone or something specific in mind, then we _can't_ also
have the additional implication that I expect you to know
which I mean.

(Well, we _could_ make a distinction between {le} and {voi},
but I don't think that was the intention when {voi} was
introduced.  I'm not sure what other equivalents you
@had in mind@ (oops).)

But then that's why Colin persuaded us to have {bi'u}.

Hmm, I've probably seen this stuff explained several times,
but it's just beginning to sink in.  Time for another rethink.

Let's try a few more examples.

So, "I'm looking for something" (specific, although unspecified :) )
would have to be
.glico ce'o lu

        mi sisku tu'a le co'e

li'u ce'o zoi glico.
right?

And "I'm looking for a princess", meaning I'm looking for a specific
person, and I start the process of describing that person to you,
to help you identify who I mean, by saying she's a princess,
would be
.glico ce'o lu

        mi sisku tu'a le pa da ku noi nolraixli

li'u ce'o zoi glico.
or possibly just
.glico ce'o lu

        mi sisku tu'a le bi'u nolraixli

li'u ce'o zoi glico.

And my rant about {djica} earlier in the year was totally misguided.
.glico ce'o lu

        mi djica lo plise

li'u ce'o zoi glico.
means something like "There are some apples which I desire", leaving
open the possibility that there might be some kinds of apple which
I wouldn't want.

This still doesn't feel right with {djica}.  I still think the x2 of
{djica} ought to be an event.

It's all right with {nelci}, and I can't even see any difficulty
allowing the x2 of {nelci} to be either an object or an event,
but there's something about the semantics of {djica} which causes
a problem unless there's an event there.

.glico .i la djan. cusku zoi glico.
> Back when we introduced "tu'a" to settle the sumti-raising question, we
> decided not to allow polymorphic place structures, because they easily lead
> to bad reasoning.  If you aren't very careful, you end up substituting
> abstract sumti for concrete ones at the wrong moments and destroying your
> chain of reasoning.
.glico .i mi spuda fi zoi glico.

But of course the gimste people are using still has several
gismu with polymorphic places, and in fact so does the one
lojbab's working on, as his recent posting showed.

.glico .i la djan. cusku zoi glico.
> Forcing you to mark sumti-raising with "tu'a" makes
> for convenience but compels clarity.  See the JL15 discussion.
.glico .i mi spuda fi zoi glico.

I haven't read that yet, but I'm hoping JL15 will arrive soon.

.glico .ita'o mi cusku zoi glico.
BTW, am I right in thinking that the parser takes '/'
to be a synonym for {fa'o}?
.glico co'o mi'e .i,n.