poi'i, se/te/ve ka (was Re: [jboske] sane kau?)

[And Rosta <arosta@uclan.ac.uk>]

Jordan:
#>>> fracture@allusion.net 12/13/02 05:30am >>>
#On Thu, Dec 12, 2002 at 11:11:12PM -0600, Jordan DeLong wrote:
#[...]
#> poi'i gives you the identity abstractor, and it is one of the few
#> of And's (extreemly numerous) experimentals which could maybe be
#> useful. I don't like that it reuses ke'a though.
#[...]
#
#Actually, xorxes mentions on the wiki that if you accept x2 of ka
#you can just use se ka instead of poi'i for this. Which is a bit
#nicer imho:

I agree. I let the poi'i proposal stand because currently poi'i is
baseline conformant and se ka is not. 

#I would actually like to see the byfy maybe consider modifying the
#place structure of ka so that x2-xn are places for filling in lambda
#variables. Which variables are filled by which place could be
#specified by subscripts (ce'uxipa is x2, ce'uxire is x3, etc) or
#assumed left to right otherwise.

I like this idea. Is it on the wiki yet? 

#A ka with every variable filled would the same as a du'u.
#
#Also, note that if any identity-capable method for NU is accepted
#it will put to rest the issue of people complaining when I use
#tu'a/jai for metonymy. (klama fu tu'a mi == klama fu leseka ce'u
#karce mi)

Did those people complain that it was wrong, or just hard to guess
what you meant? I can't see how it could be wrong.

#One other point this made me think of: "le" wouldn't be the default
#on a "se ka" like it is on normal "ka", as there's probably a number
#of things which can fill in the variable---so "le" is just used if
#you know which thing.

Of course. {se ka ce'u broda} is equivalent to {broda}, so each takes
the full range of gadri.

--And.