Message-Id: <199308261724.AA07080@eli.CS.YALE.EDU>
Date:         Thu, 26 Aug 1993 10:36:06 -0400
Reply-To: John Cowan <snark!cowan@GVLS1.VFL.PARAMAX.COM>
Sender: Lojban list <LOJBAN%CUVMB.bitnet@YaleVM.YCC.YALE.EDU>
From: John Cowan <snark!cowan@GVLS1.VFL.PARAMAX.COM>
Subject:      Re: TECH: QUERY on ZI & ZEhA
To: Erik Rauch <rauch>
In-Reply-To:  <9308251854.AA28865@relay1.UU.NET> from "Jorge LLambias" at Aug
              25, 93 02:50:43 pm
Status: O

la lojbab. pu cusku di'e

> > I just called up the parser to check this one, and that is indeed what it
> > parsed as:
> > mi ba'o ze'u klama
> > is
> > mi ba'oku ze'u klama
> >
> > This should be a bit easier to interpret:  I go for a long time, in the
> > aftermath of some event.

Actually, the question of what ZAhO+KU means is vexed, and I deliberately
didn't discuss the construction in the tense paper.  The difficulty, of
cours, stems from the semantic difference between ZAhO as tense and ZAhO
as sumti tcita.

Consider PU.  PU as a tense has a clear relationship to PU as a sumti
tcita: the PU tense can be converted to the sumti tcita by supplying an
event which is at the speaker's time reference: "le cabna" is the traditional
one.  So

1)      mi pu klama le zarci

can be interpreted as:

2)      mi klama le zarci pu le cabna

and this is the tack taken by the Lojban textbook, although I didn't use
this explanation in Imaginary Journeys.

Therefore, the form

3)      puku mi klama le zarci

can be seen either as a version of Example 1 with the tense moved from its
usual position (and a "ku" appended for unambiguity), or else as a version
of Example 2 with the sumti elided (and replaced by a "ku").
Imaginary Journeys used the first explanation, the textbook used the second,
but it didn't really matter semantically.

With "ba'o", however, there simply is no sumti which can be tagged with
"ba'o" such that the meaning is the same as when the selbri is tagged:

4)      mi ba'o klama le zarci

doesn't equate to:

5)      mi klama le zarci ba'o zo'e

for any interpretation of "zo'e".  So whether

6)      ba'oku mi klama le zarci

is equivalent to Example 4 or Example 5 is simply not determined.
I agree that this result is bogus, but it represents where we currently stand.

la xorxes. cusku di'e

> Then my interpretation of "mi pu'o co'u citka" as "I'm about to finish
> eating" is wrong. It means something like "I finish eating, in the
> 'beforemath' of some event".

Well, no.  Multiple ZAhOs do fuse (no "ku" implied) and represent sub-events,
just as you thought.  You can't generalize from one case to the other without
checking the grammar.

This is a good place to point out that there is a bug in the current BNF
(but not YACC) version of the grammar related to interval modifiers:
the line:

interval-modifier<1050> = interval-property & ZAhO

should read:

interval-modifier<1050> = interval-property & [(ZAhO [interval-property]) ...]

as in the first baseline BNF (file bnf.28).  I don't know how this error
slipped in.

--
John Cowan      cowan@snark.thyrsus.com         ...!uunet!lock60!snark!cowan
                        e'osai ko sarji la lojban.