* Sunday, 2011-10-09 at 16:25 -0700 - John E Clifford <kali9putra@yahoo.com>:
> OK, that helps. You want not just quantified variables, but restrictively
> quantified variables, with the restrictions to be gotten from contexts in at
> least some cases. And the restriction is to be such that it has only one item
> (atom or plurality) meeting it. I just don't see the point of this. Why bring
> quantification, with the attendant problems of scopes, when all that is needed
> is a constant (one of the virtues of Skolem functions is just that they get rid
> of some quantification -- why bring it back?)
My point was that constants (or, more accurately, Skolem functions) are
handled as an extreme case of the "tight existential quantification over
a glorked domain", namely where the glorked domain is a singleton.
Of course if this were the only relevant case, it would be perverse to
consider it as quantification over a singleton domain.
In other words: I'm converting your binary ambiguity - between the Skolem
function "some things I (would) have in mind" and the existential
quantifier ranging over the whole universe - into a vagueness, the
glorked domain C interpolating between the two extremes.
> Nor do I see the point of trying to bring together anaphora (typically
> syntactic. occasionally semantic), deixis (always pragmatic, I think)
> and the "whatever I have in mid" sense (probably pragmatic), as well
> as the unfettered quantifiers of the indifferent gaps. It does provide
> for something that looks like a uniform explanation for the various
> treatment of gaps, but I think it would be better to admit that the
> treatment is a mistake and try to replace it than to fadge up
> a dubious coverup job.
I don't know if it's a mistake. I don't see it as being much more
conceptually repugnant that the binary ambiguity it interpolates, and it
explains more uses.
It is, however, incomplete as an explanation of unfilled places, because
unfilled places are often more like generics - e.g. in "adjectival"
predications like {ko'a se cinri} or {ko'a jdari}.
> Careful about the order of thing; if resolution of {zo'e} comes after anaphora
> (as it usually would, given your ideas), them {P(zo'e zo'e)} would not be the
> same as {P(zo'e ri)} because each occurrence of {zo'e} is evaluated on its own.
Yes, I think I was wrong to suggest that anaphora can be cleared up
before interpreting the {zo'e}s and {lo}s.
The real story is likely to be tricky.
To take a simple example: when the {lo} is read generically, what does
{lo remna cu prami ri} mean? There are two obvious possibilities
- "humans love humans" (both generic) and "humans love themselves". The
first is natural only if we admit kinds.
(For nastier a example, consider the apparently classic {ro te cange poi
ponse lo xasli cu darxi ri}... although I'd be happy simply considering
this to be meaningless)
> On Oct 9, 2011, at 13:03, Martin Bays <mbays@sdf.org> wrote:
>
> > * Sunday, 2011-10-09 at 12:11 -0400 - John E. Clifford <kali9putra@yahoo.com>:
> >
> >> On Oct 9, 2011, at 0:27, Martin Bays <mbays@sdf.org> wrote:
> >>
> >>> * Saturday, 2011-10-08 at 19:56 -0400 - John E. Clifford
> >>><kali9putra@yahoo.com>:
> >>>
> >>>> It seems the only logically sensible out is to allow unfilled
> >>>> spaces only for variables (the general case) and require something
> >>>> more specific for the rest,preferably the appropriate pronouns in
> >>>> those case and {zo'e} on the last, though I suppose that in most
> >>>> cases {zo'e} could do for all three.
> >>>
> >>> But unless I'm misunderstanding what you mean by variables, the other
> >>> three cases (which are arguably really just one case) are just special
> >>> cases of the variable case - namely, where the glorked domain of the
> >>> existential quantification is a singleton (whose single element might be
> >>> a plurality, of course).
> >>>
> >> Here is the problem, then. In standard semantics, the universe or
> >> domain of discourse is a given and all variables range over the items
> >> in that domain. There is no case of a special domain to be used for
> >> just one variable, separate from the domain that applies to all the
> >> others (there are complications here but none that bear on this
> >> point). I suppose some mechanism could be worked out to do something
> >> like this, but it seems a lot of work for no apparent gain.
> >>
> >>> So I'm understanding you as having {zo'e} force the domain to be
> >>> a singleton, but otherwise to work like an unfilled place
> >>
> >> No. Domains don't change like that. {zo'e} is simply a constant. Is
> >> this strange ad hoc domain what you mean by "close-scope"? Rather
> >> than its effect in the structure of the sentence?
> >
> > I didn't mean to do anything funny with the domain of discourse. By
> > 'domain', I meant the domain of this particular quantification - so in
> > {da poi broda}, the set of (atomic) brodas is the domain of that
> > quantification.
> >
> > So having {zo'e} give existential quantification over a glorked
> > singleton domain is equivalent to having it give a constant.
> >
> > To be more precise about how I'm suggesting zo'e works / should work:
> >
> > If we have a predication P(zo'e noi broda, zo'e noi brode), it resolves
> > as:
> > EX (X1,X2). (C(X1,X2) /\ P(X1,X2))
> > where C is a context-glorked relation which depends on any quantifiers
> > (including ones over worlds) which the current predication is in the
> > scope of, and which is such that C(X1,X2) implies broda(X1)/\brode(X2).
> >
> > (X, X1, X2 all plural mundane variables, i.e. not allowed to take kinds,
> > but not restricted to atoms)
> >
> > (Here I've made C a relation rather than a set, which is a subtle
> > difference but I think an improvement)
> >
> > Furthermore, I'm suggesting that at least some uses of {lo} follow this
> > pattern - i.e. that P(lo broda, lo brode) means the above, at least
> > sometimes.
> >
> > Something else which might not be obvious: I think this resolution
> > of zo'e-terms happens *after* most other processing, in particular after
> > resolution of anaphora. So e.g. {broda zo'e ri} is just equivalent to
> > {broda zo'e zo'e}.
> >
> > More generally, I think we can split semantic analysis of lojban into
> > two broad stages - a pre-pragmatic stage, in which there is no
> > vagueness, ambiguity or glorking, but which leaves behind tanru,
> > zo'e-terms, non-anaphoric prosumti like {ti}, and perhaps some other
> > such things; and a pragmatic stage which applies glorking to handle
> > those leftovers. We're talking here about how the pragmatic stage
> > handles zo'e-terms.
> >
> > The prepragmatic stage should return a sentence in a logic something
> > like Montague's IL, but with basic terms and relations having some
> > structure, like zo'e-terms and abstractions and tanru. I think this is
> > quite doable, and that doing it is the best way to specify the logical
> > parts of lojban.
> >
> > But that's branching from the point.
> >
> >>> It seems reasonable to want a word for that. Maybe it should be {zo'e},
> >>> I'm not sure. If it were, we'd need to find another word with the
> >>> meaning of an unfilled place, say {zo'e'e} - if only because {lo broda}
> >>> would then be {zo'e'e noi broda} rather than {zo'e noi broda} (to
> >>> whatever extent that equivalence ever works).
> >>>
> >>>> So, back to the question case: the appropriate negative responses to
> >>>> the question { xu do klama le zarci} are {na}(or should that
> >>>> be{naku}?), {na go'i}, {mi na klama zy} ( or some more official
> >>>> pronoun), and the basic {mi na klama le zarci}, with {mi na klama
> >>>> zo'e} as a marginal possibility.
> >>>
> >>> And {mi na klama} as a definite possibility, yes?
> >>
> >> I would say, no, because that would have me going nowhere, not merely
> >> not to the store.
> >
> > Right. But I don't think having the quantification always be over the
> > whole domain of discourse, rather than a glorked portion thereof, is
> > very usable. For example, one arguably is always klamaing somewhere -
> > even if just to the place one already is at.
> >
> > Martin
>
> --
> You received this message because you are subscribed to the Google Groups "lojban" group.
> To post to this group, send email to lojban@googlegroups.com.
> To unsubscribe from this group, send email to lojban+unsubscribe@googlegroups.com.
> For more options, visit this group at http://groups.google.com/group/lojban?hl=en.
Attachment:
pgpg9fVN7vpRb.pgp
Description: PGP signature