Re: [lojban] zipf computations & experimental cmavo

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

>>> Rob Speer <rob@twcny.rr.com> 09/26/01 10:17pm >>>
#On Wed, Sep 26, 2001 at 04:57:08PM -0400, John Cowan wrote:
#> Rob Speer wrote:
#> >(For those who don't see the problem with symmetry: names are assignable.
#> >Pro-sumti are assignable. What gets assigned if you say {la djan. goi 
#> >ko'a}?)
#> 
[1]
#> If you know what ko'a means, then la djan. is defined to mean whatever
#> ko'a means.
[2]
#> If you know what la djan. means, then ko'a is defined to mean whatever
#> la djan. means.
[3]
#> If you don't know what either means, then they mean the same, but *what*
#> they mean will arrive in future.
[4]
#> If you know what both mean, and they already mean the same thing, the
#> goi-phrase is unnecessary.
[5]
#> If you know what both mean, and they mean different things, *bzzzzzt*,
#> semantic error.
#> 
#> This is called "unification" in Prolog.
#
#Hmm. That actually makes sense. I think I'll stop touting asymmetrical goi.
#I suggest you put that on the Wiki, too, because I don't think it's clarified
#anywhere else.

I reject symmetric goi because:

(1) Even if ko'a has already been assigned a meaning, you may want to reassign
a different meaning to ko'a
(2) You may want to assign the name la djan to something regardless of whether
anything else in the world of discourse could plausibly bear that name. From the
hearer's perspective, the hearer has no way of knowing whether to treat the
name as simply a label assigned to ko'a, or whether to set off round the universa
of discourse in search of a plausible referent for la djan and then assign that
referent to ko'a.
(3) symmetric goi = no'u

#(Incidentally, I don't need 2 copies of each e-mail - just reply to the list.)

Everyone is entitled to a vice. This is John's.

--And.