I leave my first reply below, but after writing it I realised something
which may render it obsolete.
I'm not sure that the basic assumption I was making here (and I'm not
sure to what extent you doi xorxes or anyone else was going along with
it) - namely that we can think that a given proposition holds on some
(not necessarily very well defined) subset of space-time(-possibility),
and tenses involve claims about the nature of this set - is actually
a very good fit for lojban.
I think it may be better to make the switch from functions to
distributions.
So rather than model merely telling us whether, given a proposition
P and a point w in the space of worlds, P holds at w, it instead tells
us whether, given a proposition P and a *subset* W of the space of
worlds, P holds at/in/on W.
(For anyone familiar with the terminology of distributions: think of
W as corresponding to a boolean-valued test function; just like we
generally only allow sufficiently nice test functions, we only care
about "nice" sets W. For anyone familiar with the language of types:
don't think of it as a type, even if it looks a bit like one. For anyone
familiar with measures - a bit like that, but (crucially) don't make it
even finitely additive. Or you could try having it real-valued (like
we're literally integrating our test functions) with some mapping to
{T,F} at the end... but probably it's best not to overthink it.)
So e.g. a lion doesn't have to cinfo at any particular point (it is,
after all, a bit bizarre to think that it does); but if I take a blob
around it (perhaps using VA) and ask if it cinfos there, the answer is
probably yes. If I take a small blob around its head, some might say it
cinfos there and some may say it doesn't; it may depend e.g. on how
crucial claws are to our conception of a lion.
So does this move help matters? I had thought that it wouldn't much,
which is why I was avoiding it; but actually it really does.
Suppose ko'a is assigned to a bunch of three lions, one of which is in
my garden, one of which is in my living room, and the third of which is
in Nairobi (I don't live in Nairobi).
With this new setup, it's now consistent to have:
xu ko'a vi le mi purdi cu cinfo .i na go'i
xu ko'a ne'i le mi surla kumfa cu cinfo .i na go'i
xu ko'a ne'i la nairobis cinfo .i na go'i
xu ko'a ga'u la terdi cu cinfo .i go'i
So the upshot is that there's no need to make {ro lo broda cu broda} be
a tautology to have {lo vi cinfo} get only nearby lions; it wouldn't get
ko'a in the above example.
So let me add an option (iv) to my list:
On Sun, Oct 30, 2011 at 6:25 PM, Martin Bays <mbays@sdf.org> wrote:
> That does raise (again) the possibly important question of where it is
> that a bunch of lions cinfos.
>
> Obvious answers:
> (i) everywhere at least one of them zvatis;
> (ii) at some specificish locale, such as their centre of mass;
> (iii) everywhere.
(iv) on those domains which contain all of the lions.
I think this has to be the right answer. Perhaps everyone else already
knew this.
Now I need to rethink JC's maximal-bunches idea.
Martin
* Monday, 2011-10-31 at 18:49 -0300 - Jorge Llambías <jjllambias@gmail.com>:
> On Mon, Oct 31, 2011 at 5:09 PM, Martin Bays <mbays@sdf.org> wrote:
> > Ah, so you meant something like the minimum *convex* volume containing
> > them all?
> >
> > Still pretty icky.
>
> Forget I said minimum then, just some volume that contains them, not
> too much larger than the minimum.
...and not necessarily connected, although it might often be? That's
more-or-less what I intended by (i).
> "Those three lions we were talking about are now at the zoo" sounds
> normal enough to me.
Sure.
> >> If I were pointing at your spleen, I should probably say "bu'u ta".
> >
> > Even if you were touching it?
>
> Probably, unless I'm talking to someone else, not to you. The ti/ta/tu
> distinction is supposed to be "close to me"/"close to you"/"away from
> you and me",
So it is. Cool. I was thinking that ti/ta/tu was an exact parallel of
vi/va/vu.
> and any part of your body is more likely to be closer to you than to
> me.
>
> >> But I agree it is a matter better left to philosophers, which is to
> >> say that there is no single right answer. Same thing would happen if I
> >> asked you when you remna. Did you remna yesterday and remna again
> >> today, or is there just one long when that you remna in?
> >
> > Hmm, there I would be inclined to answer definitively that I remna
> > continuously.
>
> But "continuously" could mean "at each point in the interval" or "on
> the interval as a whole". In one case you remna an infinite number of
> times, in the other case just once.
I'd peg that as different ways to look at the event(s) associated with
the predication. Roughly all I meant was that the subset of space-time
at which I remna would be connected. How that translates to counting
events is another matter.
> Similarly the lions may lion many times (in many places) or just once,
> in one place.
>
> >> I obviously would say that the tautology is "lo vi cinfo cu vi cinfo",
> >> which of course does not exclude the (non-tautological) possibility
> >> that "lo vi cinfo cu vu ji'a cinfo".
> >
> > Huh. So {lo vi cinfo} just means (ignoring kinds) "some lions at least
> > one of which is here",
>
> But I wasn't ignoring kinds! "lo vi cinfo" for me is just "the lions
> here".
Well ok... if it's getting a kind, the bunch is a singleton bunch, so
non-distributivity is not an issue. But you do allow {lo cinfo} to
sometimes get a bunch of lions rather than a kind, right?
> In some contexts "the lions here" are the same lions that are
> over there, in other contexts they are not. If they are not the same
> lions there, then they are not there.
>
> > and you'd have to say {lo cinfo poi ro ke'a vi zvati} to mean "some
> > lions here"?
>
> "some lions here" would have to be "su'o vi cinfo".
>
> > If common predicates like {vi cinfo} are non-distributive, are you still
> > sure it's best to have {lo} work this way? If so, could you explain why?
>
> By "non-distributive" you mean "not necessarily distributive" or
> "necessarily non-distributive"?
The former.
> In general, I would say that "vi broda" is not necessarily disributive
> over the referents of "lo vi broda".
>
> In the case of "vi cinfo", I would say it does distribute over the
> referents of "lo vi cinfo". So I would say "ro lo vi cinfo cu vi
> cinfo" is probably true.
Now I'm confused. Do you mean that this holds for the trivial reason
that you have {lo vi cinfo} having only one referent, namely a kind
"lions here", so {ro lo vi cinfo cu broda} == {lo vi cinfo cu broda}?
If not, and you're now talking about the case that {lo vi cinfo} gets
a bunch of lions (or indeed a bunch of kinds of lions), this seems to
contradict what you said before about where a bunch of lions cinfos.
Probably I'm misunderstanding you somewhere?
> (But ro lo vi broda cu vi broda" need not be.) Our disagreement is not
> about distributivity, we would still have it even if we were talking
> about a single lion.
I think you must be disagreeing with me about something I didn't know
I was saying (in this subsubsubthread, anyway).
> I don't have a problem with "They have this lion in the Bronx Zoo
> too", but you probably would object. (Or rather you may say that it
> resolves into something else, and that "this lion" does not refer to
> a lion, as it appears to do at first sight.
Quite; presumably you're talking about a kind of lion (which technically
can under certain circumstances (although it's generally a marginal
interpretation) be referred to as "a lion" in english... which does make
english a bad metalanguage for this conversation. I suppose you wouldn't
appreciate it if I suggested we disambiguate by reserving 'a lion' and
'some lions' for actual individual lions which aren't kinds (sometimes
meaning generic such), and reserve 'a kind of lion' and 'some kinds of
lion' for kinds? In any case, I will continue to abide by that rule (as,
for what it's worth, I believe I unthinkingly have in the vast majority
of speech acts I've made thus far this life.))
> )
Attachment:
pgptSfH1g6Qlk.pgp
Description: PGP signature