BAI nightmare

[cbmvax!uunet!ee.mu.OZ.AU!nsn]

>From: jimc@math.ucla.edu
>Message-Id: <9106112350.AA26579@euphemia.math.ucla.edu>
>To: lojban-list@snark.thyrsus.com
>Subject: Response to Six Messages
>Date: Tue, 11 Jun 91 16:50:46 +0100

>If I get your drift, you ask "what does it mean when a non-abstract
>sumti turns up in an abstract slot"?  You say, "hundreds of brivla
>include vague default predicates and the listener is supposed to drop
>the non-abstract sumti into x1 of the default predicate".  I strongly
>recommend that such default predicates be discouraged.  A predicate
>should mean THAT PREDICATE, not some other relation contextually
>defined via poetic license.  The default predicates are too illogical! 

Again, jimc is right, though what this means is that lojbab hasn't expressed
the proposal as well as he should. Lojbab puts in {tu'a} as a calque for
NL speakers, who want the default meant. But in real lojban, such a default
is untenable, particularly as it can vary with different contexts. No, what
is meant when a concrete shows up in an abstract slot is the elliptical
selbri {co'e}, and only that.

tu'a mi rinka leti fasnu
lenu mi co'e cu rinka leti fasnu

OK? "I cause" means "My doing something causes". We should have no clue as 
to what that something is. I stress this because, later in the message, lojbab
tried to pull a swift one on us by saying that

mi djuno tu'a le catra

would mean

mi djuno lenu le catra cu du zo'ekau

("I know that the killer" means "I know that the killer is") The problem is
that djuno does not mean "to know", but "to know that"; and {mi djuno le
catra} means {mi djuno lenu le catra cu co'e} - I know that the killer
somethings - is tall, is hairy, even "is BOB" - but not necessarily "is BOB".
To cut a long story short, the phrase actually means "I know *about* the
killer"; and it's fortunate Nora stepped in and used {kau} properly.

OK? A sumti raising can be used as a calque for NL usage with a default predi-
cate; but what it should actually mean is an elliptic predicate. "I try the
door" should, in lojban, convey no information on whether you're trying to
open, close, blow up or eat the door; just that you're doing SOMETHING to
the door.

>(Msg 3)  Alternative #1: Athelstan (?) objects to this line of evolution:
>
>	gau mi galfi le bitmu			(bridi)
>	le jai gau galfi be *fi* le bitmu	(similar sumti)
>However, this is bogus because the first sentence should have been
>	gau mi galfi *fe* le bitmu		(bridi should be)

which is exactly what the first sentence means: with the x1 omitted, the
sumti after the selbri defaults to x2.

>So in both situations the user has to jump over a place.  

No I didn't. jimc, make sure you know lojban place structure transformations
before saying things like this.

>The change from fe to fi is simply a consequence
>of shifting x1-x5 to x2-x6, presumably a tolerable change.  

My arse it's tolerable. (ie it isn't). One type of conversion is enough. If
anyone defends pushdown again (especially after I used fai in my xai), I'll
be very cross.

>Actually, as a pan-predicatist, I would say to use the gismu which the
><BAI> is derived from, with the original main bridi becoming an
>abstract sumti of it.  Example: 
>
>	gau mi galfi fe le bitmu	
>		The wall was modified by me
>	le gasnu be le nu galfi fe le bitmu	
>		The actor in the wall modification

Nice of you to mention this. This is in fact how John will have to define
BAI when he gets to it. The problem is that the lujvo necessary for such
analysis (btw, this analysis is essential for us to find which BAI have
to be tagged on by a pe/ne: the underlying lujvo will either have an
abstraction as a place, or not.) can be freaky. This is how far I got yesterday:

ba'i: basti: x1 replaces x2: neither is an abstraction

thus {mi morsi ba'i do} meaningless: the bridi says {ma basti}, but not {ma
se basti}. When you say

mi ne seba'i do morsi
then you know that {mi basti} and {do se basti}, and the semantics flops beau-
tifully into the transformation

mi noi do se basti ke'a cu morsi.

I call this a weak {ne}-link, as there is no place in the lujvo for the current
predication itself. {mau}, on the other hand, is a strong {ne}-link, in that
there is a place for the current predication:

mi ne mau do necli ko'a:
mi se zmadu .i do zmadu .i lenu zo'e nelci ko'a te zmadu

Get it? The phrase translates as "I (and you more so) like him". The {do} goes
into the first place of zmadu (here not an abstraction), the {mi} into the x2
(and this is important: the current bridi is not the x2, so the {ne} link is 
necessary, else you wouldn;t know who in the hell is being exceeded (ma se
zmadu? "More than WHO?"). The current bridi, however, or a truncation of it,
*does* go into x3 of zmadu: the CONTEXT of exceeding is the utterance itself.

Then there is fair dinkum (true) sumti tcita. Because the x1 and x2 of zmadu
can be abstractions, when they are the current predication is assumed the x2,
and no {ne} is necessary. This is why lojbab left out the {ne} in the first
stanza of his "Language" translation - the x2 of se zmadu, the exceeded,
was the utterance's predication itself:

in a silence
More eloquent than any word could ever be
.i li'i nunsma semau ro valsi ...

ma zmadu? ro valsi. ma se zmadu? leli'i nunsma. No {ne} required, nor anywhere
for it to go, in any case.

the nicest true sumtcita are the causals: their gismu take abstractions in both
x1 and x2, so there's no problem with the x2 being {lenu no'a} - I go because
you said so: I go (your saying so causes that [= I go]).

Now try and analyse {bau} like this. You can't. The reason you can't is because
the place structure of {bangu} is inappropriate: the x2 of bangu is a people,
not a text or an event. le mi xatra pe bau la lojban: my letter in lojban.
This does not map to lemi xatra poi la lojban bangu ke'a. It can map to
bangu je velcusku (lujvo: banvelsku): is a language and a medium of communica-
ting of:

lemi xatra poi la lojban banvelsku ke'a

where ke'a is the selsku, the thing communicated -
here, the letter. But what of mi sanga bau la lojban? The above analysis
is no good, because {lenu mi sanga} is not a selsku, a communicated thing -
it is a nunsku, an act of communication.

So what an analysis of {bau} needs is a selbri that has as its x1 a langauge
and as its x2 a nunsku. Anyone got any ideas?

(btw, when you find this bridi, it'll be how you say "My singing was in
lojban", and probably also "this letter is in lojban" - something you can't
quite do right now as a jufra claim.)

Is this level of analysis too picky and prescriptive? Not when only lojbab
knew (until he told me %^) what {li'e} means. Does it make lojban a Carter-
esque panpredicatish nightmare? Not necessarily: you can put suitable caveats
on the types of transformations I'm talking about. 

More homework for John Cowan. No rush, John %^)