WikiDiscuss

WikiDiscuss


BPFK Section: gadri

I agree with the opening remark here, though for somewhat different reasons. The problem that arj is dealing with arises from the fact that one (or maybe the only) "meaning" assigned to {lo} is essentially vague. Trying to apply recise tests to discover the truth of a sentence in this case is bound to fail. Looking at the examples on the wiki sheet I found that most of them fluctuated (when applying precise tests) between "all" and "some" but with conditions. Thus, if read as "some," it often turned out that one or two — or any fixed number of — cases were not enought; if read as "all" then any number of exceptions had to be allowed without affecting truth. Nor would any of "may", "most," or such intermediate — and already rather vague — quantifiers work. This is out of the range of quantifers and into operators like "generally," "as a rule" and so on, applied at the argument level rather than the sentential. This seems to be one thing that xorxes means when he says that
{lo} expressions refer to kinds or species or maybe even Mr. The next step (and where the ambiguity comes in from my point of view) is moving from talking about things in a certain unspecified collective way to talking about one thing, the unspecified collect of those things — more or less like taking the typical x as being some paricular (though prehaps unidentified) x. This way of talking does result in an expression that serves much like a constant: it is usually negation transparent — if it is not the case that some unspecified collective of xs are ys, then such a collective is usually not ys (unles there are no xs at all) and, except for discrete properties, is such a collective is a and such a collective is b, then there will be such a collective that is both a and b. These are not valid moves, of course, but they usually work and thus can be applied in practical cases.

BTW, the test for {mi nitcu lo mikce} is probably misguided, the doctor you need may not be any actual doctor — you need one who can perform a memory-retaining brain transplant, for example. The {lo prenu cu tarci} works on any case. the {lo prenu cu sadjo} falls into the vagueness trap by working at the upper end of the range rather than the lower or middle (xorxes does encourage that with his examples, to be sure). Here it would be enough if some few (or many or...) people were Saudis (I am not sure whether one would be enough).
Arnt Richard Johansen <arj@nvg.org> wrote:
One of my problems with Jorge's proposal is that it seems to either
conflate two different meanings into one, or have one meaning that is
ill-defined or hard to formalize. What I want is a general algorithm fo=
r
determining whether or not a sentence that uses XS-lo is true or false.
These examples are taken from an IRC discussion I had with xod.

{mi nitcu lo mikce} — obviously true (in the right circumstances)
Here, I take every {mikce} in the world, assign it to {xy.} and for eac=
h
one asks if {mi nitcu xy.} is true. If and only if there is no {xy.} fo=
r
which {mi nitcu xy.} is false, then {mi nitcu lo mikce} is true. No
problem here.

{lo prenu cu tarci} — obviously false
We are then looping through every {prenu}. There are some (actually all=
)
instances in which {xy. tarci} is false, therefore {lo prenu cu tarci} =
is
false. No problem here.

{lo prenu cu sadjo} — obviously true.
We are again looping through every {prenu}. There are some instances in
which {xy. sadjo} is false, therefore {lo prenu cu sadjo} is false. Whi=
ch
is patently wrong, but fits with how I understand XS-lo.

So. What is it that I'm missing here.

--=20
Arnt Richard Johansen http://arj.nvg.org=
/
P=E5 1300-tallet kom tersen. F=F8r og etter det var det meste bare rot,=
men
s=E5 kom Sch=F6nberg og ordnet opp. Puh. Endelig litt system. S=E5 klar=
te Arne
Nordheim =E5 rote det til igjen. — Under Dusken 08/2=
001