Message-Id: <199309081819.AA20206@eli.CS.YALE.EDU>
Date:         Wed, 8 Sep 1993 11:19:20 -0700
Reply-To: jimc@MATH.UCLA.EDU
Sender: Lojban list <LOJBAN%CUVMB.bitnet@YaleVM.YCC.YALE.EDU>
From: jimc@MATH.UCLA.EDU
Subject:      Re: Event contours and ZAhO tcita
To: Erik Rauch <rauch>
In-Reply-To:  Your message of "Mon,               06 Sep 93 01:25:36 EDT." <9309060526.AA04778@julia.math.ucla.edu>
Status: RO

Veijo Vilva <VILVA@VIIKKI21.HELSINKI.FI> writes:
> The purpose of a sumti tcita is to make visible a sumti place which
> originally was left away from the definition of a gismu. When the
> place structure of a gismu is defined, all the possible places are
> considered and those which seem most relevant are retained. This
> doesn't mean, however, that the rest disappear from the underlying
> relationship. They still are part of it and the sumti tcita are
> the mechanism which makes it possible to define -- at least a part
> of -- the sumti which have no permanent sumti place.

For a long time I had the same opinion, but I find myself wavering.
To restate the opinion, the numbered places of a relation and the
tagged places have no essential difference.

Now the number of different selbri is potentially infinite.  Large
numbers of them have officially defined sumti tcita, and all the rest
can be transmogrified into sumti tcita with the construction
{fi'o (bridi) fe'u}.  Each of this infinite set of tags can be used
to glue on a separate place to any / every bridi.  Thus the relation
represented by any selbri has infinitely many places.

I'm not comfortable with the idea of relations with infinitely many
places.  (What is this, some kind of Hilbert space? :-)  So I waver
towards a less elegant but more tractable formalism, namely that each
selbri has a prespecified set of places, finite in number, which are
the numbered places.  To deal with the tagged places you backmap them
to the corresponding {fi'o (bridi) fe'u} and you interpret that as
a kind of restrictive subordinate clause that restricts the main
bridi referents to only those fitting the tag relation.  <PU> tags
may have slightly different syntax but are similar semantically.

Example, deliberately avoiding event contours:

    In colloquial :-) Lojban
        mi citka pu le nu mi xalpinxe   I eat before I drink (alcohol)
    Equivalent with fi'o
        mi citka fi'o [ke'a] purci le nu mi xalpinxe
    The restriction stated inside-out as the main bridi
        lo nu mi citka cu purci le nu mi xalpinxe

> The contour is something external, not part of the structure of
> the relationship. All the possible contours are inherent in the
> relationship but NOT expressible using sumti. The sumti places
> can be filled with references to external events and thus make
> it possible to specify relationships between the outer bridi and
> these external events -- whether the external event descriptors
> are inserted into the existing sumti places or tagged on with any
> kind of sumti tcita.

Event contours are like (tagged) places but, having no particular
sumti involved, they aren't actually places.  That insight is very
interesting.

This sounds like a fi'o restrictive clause where the bridi has only
one place [ke'a] occupied or, as for contours, only the one place
is available for occupancy.  Let's give it a try; maybe the choice of
selbri is not the best, but you get the idea.  For mo'u (natural end
point) let's use {fanmo} = "x1 is an end of (thing/process) x2".

    Colloquial
        mi mo'u citka                   I've finished eating
    With fi'o
        mi citka fi'o [ke'a] se fanmo
    Inside-out restriction
        se fanmo fa lo nu mi citka      My eating terminates

(Actually the gloss "I've finished eating" implies a present tense
on mo'u-fanmo which isn't in the Lojban; "my eating terminates" is
the better translation for both, I think.)

I think you can get a lot of mileage out of backmapping sumti tcita
(including event coutours) into fi'o clauses, rather than trying to
understand them as places of the relation equivalent to the numbered
places.

                -- jimc