* Saturday, 2011-10-15 at 20:52 -0300 - Jorge Llambías <jjllambias@gmail.com>:
> On Sat, Oct 15, 2011 at 8:22 PM, Martin Bays <mbays@sdf.org> wrote:
> >> (In fact I'm never quite sure about what to do with "poi" when it is
> >> not being used to restrict the domain of a quantifier.)
> >
> > Quite. If it does do anything, there's also the issue of which gets
> > priority in {ro lo broda poi brode}.
>
> Does it make an actual difference? "ro lo broda poi brode" = "ro da
> poi me lo broda zi'e poi brode".
It does make a difference if {ro} is a singular quantifier.
{ro ko'a poi broda cu brode} -> "for every atom x among the (plural)
referent of {ko'a} such that broda(x), brode(x)"
So (forgetting about kinds for a minute, since this is purely an issue
of plural semantics) in {ro lo broda poi ke'a brode}, the question is
whether the relative clause is evaluated with a potential plural
referent of {lo broda} for {ke'a}, or whether it's evaluated once per
atom below the plural referent of {lo broda}, with that atom for {ke'a}.
I think the latter does make most sense.
> > It could *suggest* that the {lo xasli} should be interpreted more
> > specifically, I suppose, but I don't see why it should do so any
> > more than the {noi} clause.
>
> "noi" makes perfect sense when applied to a singleton. "poi", while
> still interpretable, doesn't make that much sense because the only
> thing a singelton could be restricted to is itself, so no real
> restriction. It's the same situation that occurs with quantifiers: you
> can quantify over a singleton, but since doing so is rather pointless,
> the mere presence of a quantifier suggests the domain should not be a
> singleton.
So you mean {lo xasli poi da darxi} gets interpreted as {su'o lo xasli
poi da darxi} (interpreted according to the second of the two options
above)? That seems reasonable.
Or maybe as {pi za'u lo xasli poi da darxi} (assuming that {pi za'u}
is a plural existential quantifier), which maybe is even more
reasonable.
> >> "lo speni be da" is "zo'e noi speni da". I don't see how you could get
> >> rid of the unbound variable there. There's no referential "lo speni"
> >> in the non-referential "lo speni be da".
> >
> > OK, but if {lo speni be da} == {zo'e noi speni da}, then we have
> > a situation analogous to that above - with {zo'e} in place of {lo
> > xasli}. {zo'e} can be taken to referential, for example with referent
> > the kind 'humans', which does indeed satisfy {ke'a speni da} for each
> > da.
> >
> >> > Given this, I'm now slightly surprised that you're willing to allow {lo}
> >> > to ever give a Skolem function rather than a constant!
> >>
> >> If the selbri that "lo" transforms into a sumti contains an unbound
> >> variable, then I don't see how "lo" can create out of it anything
> >> other than a function.
> >
> > So am I taking "{lo} -> {zo'e noi}" too literally?
>
> Hmm... Maybe you are right, and it never need be a function, or at
> least not always. It needs more thought.
This trick of generalising out of a quantifier only works with kinds, of
course; so requiring that it always be a constant would make it even
rarer that {lo} gets interpreted as some mundanes.
Martin
Attachment:
pgpA5XmUqoI12.pgp
Description: PGP signature