John E Clifford, On 09/09/2012 03:26:
> *From:* And Rosta <and.rosta@hidden.email> John E Clifford, On
> 08/09/2012 19:35:
>> So, then, how do we tell if la mlta xkra is true? We squint,
>> apparently, but at what? I don't even know where to look.
>
> You take the logical form, add in the content from the encyclopedia
> entries (from l-, mlt- and xkr-), and see how closely the
> proposition matches the world. That won't be a formalizable process
> (in the sort of loglang in question).
>
> Neato. Now what is the encyclopedic entry on lV?
Within the rules for logical forms are rules of equivalence between logical forms. Some rules of equivalence serve to define lV, and I think the entry on lV would be empty. The equivalence rules would probably involve sV and rV, whose encyclopedia entries are what one would expect.
> If all you want is predicate-argument and operator-variable
> structure, then what exactly is your problem with Lojban, which has
> that. The obvious answer is that it is not consistent in its use of
> variables, replacing them sometimes by repetitions, sometimes by
> pronouns, and so on.
The problems with Lojban are firstly that it doesn't in fact unambiguously encode logical form, secondly that it is full or cruft, garbage and bodging, and thirdly that it is insufficiently ergonomic, for various reasons especially including longwindedness.
Xorban even at this early stage is fixing most of those problems.
> So far, the Xorban design is doing nicely on being consistent (if
> occasionally obscure) but not so good on being speakable (a common
> problem in loglangs, although more often for parentheses than
> variables).
Once a loglang has met the key criterion of unambiguously encoding logical forms, the main criterion for choosing among them would be usability -- how speakable and hearer-comprehendable it is.
I only know two loglangs to have met the key criterion, and the only one visible in public is Xorban. it would be good to have alternatives that are (even) more usable, so this we could work on usability.
>>>> I've given you one plausible interpretation, that l is an ever
>>>> leftmost quantifier that picks out a bunch of items and sticks
>>>> with thm throughout its scope.
>>>
>>> I don't see problems with it. Crucially, if Y is true of
>>> something that is that bunch, then Y is true of everything that
>>> is that bunch -- given that it's a single bunch.
>>
>> This seems to me a rather different point than what has been said
>> formally. The formalism has been la Ra Pa <=> re RePe and la Ra Pa
>> <=> se Re Pe. But what you are saying seems to be la Ra Pa <=> re
>> e=a Pe and similarly for s.
>
> I don't understand what your "re e=a Pe" means, or how it differs
> from "re Re Pe".
>
> We seem to agree that in "lA bcdA fghA", A is a single thing, and is
> a bunch of bcds, and we also seem to agree that maximal leftmost
> scope is fine, but for a case where it contains a variable bound by
> something outside the scope of lV. So if we disagree about anything
> it must be on what a bcd is. Or am I missing something?
>
> Well, I am having second thoughts about that leftmost scope idea and
> seeing other problems with it, especially for some of the predefined
> cases of Xorban. But I may be wrong about those worries and and even
> find a qorkaround for leftmost rules generally. My main problem
> comes from your explanation for the equivalences between the lV
> sentence and both the rV bound and the sV bound, with the same
> restriction and the same predicate. You seem to suggest that, given
> la bcda fgha, re bcde refers only to the bcds among la bcd and
> similarly for si bcdi. That is the peculiarity that I was trying to
> point out in my rewrite (and actually, I am not sure whether it is
> a=e or e[epsilon]a that is involved). That does seem to be what you
> say and I don't see how to make the equations work otherwise.
OKay, thanks for explaining. Yes, I was saying "la Ra Pa" <=> "re [is-A]e Pe" and "se [is-A]e Pe". (Not epsilon and probably not "=".)
--And.