Re: TECH: Nested relative clauses

[Logical Language Group <lojbab@ACCESS.DIGEX.NET>]

>>John has made {re xirma} and {re lo xirma} nonequivalent in either one
>>or two ways (they're different with respect to dogbiting behaviour
>>("2 men bite 3 dogs" - how many dogs?) and possibly with respect to
>>existential import.
>
>djer:  Did I miss something about the installation of this vague change
>in an era of five years of no change?  Maybe I should go to lojfest.
>
>>But anyway, if {re xirma} were equivalent to {re lo xirma}, and
>--More--
>>deserves your criticism, it would be a very mild offender, since the
>>rule by which it is abbreviated is so straightforward.
>
>Straightforward, yes, and far reaching.  It's just the difference
>between having the number system be based on traditional logic or on set
>theory. And a lot more.

doi djer

This is one of the few things that seems to have come out of the endless
discussions over the last two years on "any" and opacity.  No - nothing was
decided on this at LogFest (though you are indeed welcome to come %^).
John just reached the conclusion that this interpretation is more "natural".
I won't pretend to understand or explain why, because I don't understand
myself %^).  I concurred weakly only because "re broda" is an abbreviation
intended for naturalistic use, and is not intended as a model of logic or
set theory or anything else, but a short cut.

On the other hand, if you have strong feelings onthis, feel free to speak up.
This type of semantic equivalence argument in my mind is not the sort
of thing that has to be decided by the baseline, so long as we do not need
new grammar.  (there is some effect on the refgrammar though, so I would
prefer a decision, so John knows whether to stick with what he has written
or weasel-word to indicate that the semantics is not yet settled.

When one wishes to be absolutely true to logic, one must use prenexes and
quantified variables.  "lo" seems to be turning out to have logical pitfalls,
and "le" of course makes a claims only so much as the referent is agreed
upon.

If you want to find the discussions where this came about, good luck.  I do
recall that there was one point where pc pointed out that the way logic handles
quantifiers other than ro, su'o and pa can be very unwieldy.  But I don't
recall the discussion revolving around set vs logic versions of quantifiers.

lojbab