Re: [lojban] Re: Usage of lo and le

[John E Clifford <clifford-j@sbcglobal.net>]

<<> The implicit quantifiers on {le} are {su'o}
> internally and {ro} externally.  The implicit
> quantifiers on {lo} are just the reverse.  So
an
> explicit internal quantifier on {lo} gives the
> number of all the whatevers in the world, while
> one on {le} just tells how many thingies the
> speaker has in mind.  External quantifers are
> partitive, how many out of the totality given
by
> the internal quantifer are being spoken of
here.

I'm undecided on the outer, but I am firm in my
current belief that
{ro} should never be an inner quantifier by
default for any case.>>

Well, I think everyone (almost) is agreed on that
-- though for a variety of reasons.

<<> {lo,le,la} are about individuals taken
> separately, that is, what is predicated of a
> sumti of these sorts is predicated of each
> ultimate referent of that sumti taken
> individually. In contrast, {loi, lei, lai} are
> about "masses," one of those words that
> Loglan/Lojban has taken over from some fairly
> precise meaning -- I think "mass noun" -- and
> used differently and without a very clear
> meaning. Among the things that examples suggest
> as falling under this notion -- and which
others
> have elevated at one time or another to the
main
> meaning of {loi} etc. expressions are 1) {loi
> broda cu brode} says of some brodas that
although
> no one of them brodes, taken together they do
> (e.g. surround a building as the brode), all of
> them participating in the event.  2) the
> corporation of brodas -- like 1 in that no one
> member does it but unlike 1 in that {loi broda}
> may remain the same even if the brodas referred
> to change and the corporation may do things in
> which some -- or even all -- of its members do
> not participate (GM makes cars although many
> members of GM don't work on cars, the Red Sox
won
> the pennant although all management and some
> players on the roster did not ever play any
> baseball)(Species are either in this group or
> something very similar.).  3) The mass noun
> related to {broda} (which, in Lojban, is always
> count), the goo into which brodas dissolve
under
> pressure and of which they may be taken as
slices
> (the "gavagai" jokes and, after the accident,
> "there was dog all over the car).  There are
> probably others I have forgotten ("myopic
> individuals" or some such that I never
> understood, for example).  In any case, they
lVi
> sumti are not about individuals taken
separately.
> {lo'i, le'i, la'i} are for Cantor sets of
> individuals of the noted sort.  Like the lVi
> series they preserve the disntions among the
> simple e, o, and a gadri.

I am somewhat ignorant of 'Cantor sets' (reduced
into infinite
infinitely small sub-things..?), though I think I
understand enough
(of sets) to understand (what you're
explaining).>>

Cantor sets are the regular sets of usual set
theory.  The distinction is to differenntiate
them from 1) Lesniewski sets (alias mereological
sums) and 2) ordinary sets used with predicates
that connect to the members rather than just the
sets.
 
<<As for lVi, I think
(perhaps) that the most important thing is that
they all do it
together. Questions like "is it true that loi ro
countries fought the
country of Germany if the country of England has
fought it?" seem not
to affect the discussion.>>

Don't get the point of the discussion, but the
initial summary is pretty good for sense 1 and
for most of what has survived through the various
discussions.


<<> The way changes are going (this may not be a
> completely accurate presentation of all the
view,
> since I am a partisan here and also don't
really
> understand some moves by others).
>
> A. The lV'i series for sets was needed in the
> olden days because standard logic had (that it
> was aware of) no way of dealing with plurals
than
> by sets (which are singular but encompass
many).
> Of course, in that same standard logic talk
about
> sets had no (very straightforward) way to deal
> with the properties of the members of a set
while
> talking about the set explicitly.  The
appearance
> (or coming to attention) of plural
quantification
> (and reference) removed that problem and
> introduced a device (actually either of at
least
> two devices) which dealt with plurals in a way
> that covered both ordinary sumti (lV, lVi,
etc.)
> and did all the things that sets were
explicitly
> used to do.  In short, though lV'i remains in
the
> language, it has virtually no usefulness
outside
> of mathematics (and so does not need such a
> useful set of words).  I think everyone wants
to
> get rid of these altogether, but it will take
> some doing to actually make the change.
>
> Of the various uses of lVi, 1 is covered in
> plural logic by the notion of non-distributive
> (collective) predication.  As such it is not
> appropriately expressed by a gadri, since it
does
> not involve something different from a
> distributive predication but only a different
way
> of predicating on the same thing(s). It ought>>

Just the last bit?  A description refers to a
bunch of brodas (one way or another: "bunch" has
at least two realizations) A sentence involving
that description says something about those
brodas -- that they have a certain property. Now
it may say they have that property in either of
two ways (at least): either each of them has it
separately ("My students wear green hats" -- each
of them wears a green hat), also called
distributively, or they may have it collectively
(non-distributively) ("My students surrounded the
building" -- no one of them did, but acting
together they did).  In the two examples, "my
students" referred to exactly the same things in
each case, the kids in my classes.  What is
different is not in what is referred to but what
is said of it, so the distributive/collective
distinction belongs not with the referring
expression (the description) but with the
predicating part.  In addition, attaching the
predication type to the description means some
cases don't get dealt with: in "The people who
surrounded the building wore green hats" the
description is applied collectively (that is, it
is based on"these people collectively surrounded
the building" but the description is used
distributively ("They each wore a green hat"). 
In "The people wearing green hats surrounded the
building" the opposite is the case.  And, in "my
students wore green hats and surrounded the
building, I need "my students" to be distributive
and collective simultaneously -- one for one
predicate, the other for the other. Lojban has
nothing to mark these differences except the
gadri (nothing like "separately" and
"collectively" of the right size), so we continue
to use them when we can and the difference is not
obvious but is important.  Mainly, however, we
take it that it is clear from context which is
meant and then we can use {lo} (the least
specified gadri) throughout. 

<<> then to be somehow expressed in the predicate
not
> the arguments but there is presently no way to
do
> this in Lojban and no active suggestions how to
> do it.  For the nonce then the difference is
> still covered by the lV-lVi contrast, even
though
> this leaves some cases uncovered.  2, the
> corporate form, which is about a different sort
> of thing and so might be covered by a gadri, is
> also still covered by lVi, often without
noticing
> the difference involved.  Should a predicate
way
> of dealing with the collective/distributive
> distinction be devised, lVi might naturally be

I'm again lost.>>

Nowadays, {loi} etc. are used mainly for
collective predication, but also for the
corporate model.  If we get a way of getting the
collective notion attached to the predicate, then
{loi} could be used just for the corporate model.

<<> used for these cases, although they are
perhaps
> not common enough to deserve such a central set
> of words.  I thin that some people still use
lVi
> for the goo reading, 3, although it seems to be
> adequately covered by collective predication
over
> pieces of brodas and that locution seems to be
> about the right length for the frquency of this
> sort notion.  (Something like this may also
work
> forthe corporate model, 2, using the
appropriate
> one of a number of predicates for organizations
> of this sort -- if the right ones exist).
>
> Moving to lV, as far as I can tell {le} and
{la} are
> unchanged, except that the distributivity need
> not be assumed; rather whether distribution or
> collection is meant is mainly left to context,

What is distribution and collection (perhaps with
examples)?. It might
help to know that I'm very vague on the
distinction between {lu'o ro
lo ro cribe} and {ro lo ro cribe}.>>

On collective/distributive see above. On {lu'o}
and the like, I haven't seen enough of them to
have an opinion, but, since they seem to relate
to the old difference between {lo} and {loi}, I
think they are on their way out (except they
might work for the missing items attached to
predicates to mark distributivity/collectivity).


<<> with the lVi forms brought in where
collection is
> crucial and not obvious.  Presumably solving
the
> predication form of this would allow these
gadri
> to be neutral -- just referring to the brodas
> involved without limiting how they are inolved.
> Implicit quantifiers have been done away with,

I assume that implicit quantifiers are basically
an additional
assertion regarding how many there are such
that..., as I described
above, correct?>>

Implicit quantifiers aren't assertions and, as
noted, have been (in the new ideal) been done
away with.  What remains is that at least {le}
descriptions require that there be something
referred to (so as though there were an implicit
internal {su'o}) and distributive predication
claims that all of the referents have the
property involved (so as if an implicit external
{ro}).
 
<<> except that the very meaning of these two
gadri
> require that there be something they refer to
> (i.e., it is as if the implicit internal
> quantifier were {su'o}) and both distribution
and
> collection are about all the members in these
> cases, so something like explicit external {ro}
> is involved.  These readings off what is
involved
> in specifying seem to be the point which the
old
> implicit quantifiers were meant to cover).

My position regarding outer quantifiers is
undecided. It's the
difference between
{xu do pu viska lo cribe ca lo nu do vitke le
dalpanka} meaning:

{xu do pu viska su'o lo cribe ca lo nu do vitke
le dalpanka} - "did
you see some"
{xu do pu viska ro lo cribe ca lo nu do vitke le
dalpanka} - "did you
see all (surely meaning did you see all that were
in the zoo)"

...and given this example to oppose others that
exist, I can't say
which is better.>>

I guess I still don't get the point or is it just
that if there are no external implicit
quantifiers we don't know how to take {lo broda}.
 I think one version, which takes {lo broda} as
somehow about all brodas, would say that this has
to be read  without quantifiers at all (and how
that works goes into metaphysics) while the
other, which takes {lo broda} as being about some
brodas, would say that the assumption is that all
of the some referred to are intended.


<<> The case of {lo} is somewhat more complex. 
The
> basics are clear enough: it is unmarked for
> specificity and for distributivity. And the

Again (as always, since I think that it's my
major point), I wonder
what is meant by specificity. Distributivity is
another matter, and I
need to give it more consideration (specifically
in terms of the
second non-ro-outer suggesting X-for-each").>>

I don't see the connection.

<<> explicit external quantifiers are clear, that
is
> how many brodas we are attributiing the
predicate
> to (and, probably, distributively since
> quantifiers tend to individualize rather than
> mass).
> After that comes the separation.  On one view,
> the unmarked form is just the unspecific form
of
> {le}, brodas that get caught up in this case by
> context and intent, but not specified.  An
> explicit internal quantifier says how many
there
> are as such in this case, and an external
> quantifier says how many of them get the
current
> prdicate.  And, by the way, {lo broda} in
primary
> usage entails that there are broda (not in the
> scope of negations, world altering modals,
> absttractions or opque contexts).  I am less
> clear what the other version says about simple
> {lo broda} except that on occasion at least, it
> is said to yield true claims from primary
> occurrences even when there are no brodas and
to
> authorize external generalization from opaque
> contexts.  To do these things, it can no longer
> refer to brodas as such but moves to something
at
> a different level (I've tried a number of
> suggestions, none of which worked apparently).
In

I'm somewhat lost here.>>

Welcome to the club.

<<> addition, internal quantifiers become part of
the
> defining predicate: {lo ci broda} is not three
> brodas bu some (or maybe no?) broda triads. 
{mu
> lo ci broda} then is five broda triads --
between
> seven and fifteen brodas.

I disagree with this. {ci lo broda} is the triad
(formed out of
members of some here-unrestricted group), and the
{mu} (five triads
of) is given by whatever-it-is earlier in the
sentence. The quoted
version would be inconsistent with what I've
described, and I'm very
sure also inconsistent with whatever is the
current usage.>>

Well, I agree that this is disagreeable, though I
am not sure I follow your objection.  Presumably
a  broda triad is {lo broda cimei}, not {lo ci
broda}

<<> Now, against that background, I wonder if
Maxim
> can provide some clarification of his
suggestions.

The biggest aspect of my suggestion is that
{lo}-types are capable of
handling all cases thus-far provided, and that
{le} is /not/ a subset
of {lo}. >>

Well, any time {le} is appropiate, {lo} may be
used instead, but the opposite is not true.  {lo}
cannot be used to specify referents.

They are completely seperate. It may as well be
that {le}
didn't exist. 

Well, we could get along without the distinction
(and maybe should) but for now we need {le}
because {lo} can't do its job, which is built
into Lojban.

<<And, with that in mind, this lets us
re-introduce the
{le}-types as a compliment to the {lo}-types,
with the very same
usage, except that with {le} you get "by my
definition", while with
{lo} you get "by common definition". 

This strikes me as a bad idea, but if you want
that distinction there are other ways to do it in
Lojban -- and {le} doesn't do it.

<<I'd then imagine that {le} would
be used for more casual speech where you're not
explaining /
discussing / arguing anything, and so have the
liberty of using your
own definitions (just in case, so why not?), and
it allows for
comments like "by-my-definition-the prince of
Wales tore down the
curtains" (in reference to a chimp).>>

I don't see the advantage of this, certainly not
for a logical language.  Nor for one for ordinary use.