[lojban] Re: A (rather long) discussion of {all}

[John E Clifford <lojban-out@lojban.org>]

Well, it is nice to find that I am not the only
person who can get involved in interminable
discussions with xorxes.  But I now find myself
sympathizing with those who do not participate,
for reading the sixty-odd salvoes fired while I
was away is very frustrating (yet I can?t bring
myself to just skip them). Part of the
frustration is in seeing people arguing past one
another, abusing language (both English and
Lojban), and so on ? and not being able to jump
in and set them straight.  Since I generally
agree with xorxes in this discussion (mirabile
dictu), I tend to see the flaws in Maxim?s
position most clearly, but I am sure the other
participant is guilty as well.  Herewith some
comments as I have been catching up (not that I
am done yet, so excuse me if some things I say
are dealt with more recrntly than where I read
the problems).

The shuffling around about ?context,? which is
often used, it seems to me, as a way not to deal
with issues by taking the term in one sense when
it clearly was meant (to the Gricean
collaborative interlocutor) in another, is
unfortunate.  The split between context and
setting does not seem to illuminate much and also
seems to leave out important factors (background
knowledge, for example, which is neither in the
words nor the physical environment).  Yes, to
humor Maxim, it would be strategic to try, when
relevant, to use ?setting,? but, on the other
hand, there is the general obligation to
understand what your fellow discussant is saying
and respond to that, even if the wording is not
up to your standard.

?Ambiguous? does not mean ?capable of more than
one interpretation;? it means ?having more than
one meaning,? which is similar in most cases, but
not the same.  It, like vaguenss, is situational:
an expression may be ambiguous or vague in one
situation and not in another, even though the
?ambiguous? one may be capable of several
interpretations in both (the capability just not
being called upon in one).  Vagueness is just not
being precise enough ? for whatever the prupose
at hand is. When you are planning what to do
against a threat, saying the threat is an animal
may be vague, but it is not ambiguous between
dogs and elephants, even if those are the
possible animals involved, since ?animal? doesn?t
mean ?dog? nor ?elephant? (nor is it capable of
these interpretations). (The dictionary citations
start out talking about interpretations but have
to get down to meanings almost immediately, since
?interpretation? gets them off into poetry, which
is not strictly ambiguous.)  Ambiguity is about
meaning, not reference.

In a similar way, {lo broda} is not ambiguous
between brodas predicated of collectively and
brodas predicated of distributively; it just
refers to the brodas.  How they are predicated of
depends upon what the predicate is and how it is
related to them. (I agree with Maxim that the
place for the information about how a bunch of
things is predicated of does not belong in the
gadri but in some attachment ot the sumti or the
predicate or somewhere in between (like
?together? and ?individually? in English).  As
far as I can tell, xorxes does not disagree.  But
the fact is that there is no way to do this in
Lojban (yet) and so we must hobble on with what
we have, despite the gaps and problems it leaves.
 If it is important to note that the predication
is collective and that is not obvious from
context, then use {loi}, otherwise {lo} is
sufficient.  To be sure, the rpredication will,
in fact,  be one way or the other, but we don?t
have to say so ? anymore than we have to say that
the dog biting us is black if want someone to get
it off us.  finding a good way to deal correctly
with predication type seems to me an issue worth
discussing, as the present one does not.)

{lo broda} refers to brodas (I would say ?a bunch
of brodas? but xorxes gets all metaphysical about
that expression, even though it does not force
the entity interpretation in English and the
logic of  both interpretations is the same).   An
internal quantifier says how many brodas are
involved (I know that CLL says that it says how
many brodas there are, but CLL was out of datr
several years before it got published and we are
just not correcting it.)  In {lo ro broda} that
number is everything that counts as a broda. 
This does not mean that {lo ro broda} means ?all
the relevant brodas? (or some such), since that
suggests that there are irrelevant brodas that
aren?t coverd.  But there aren?t.  To be sure, we
can bring in ? create within the world of the
conversation, if you will ? more brodas, but they
aren?t there until we do.  I take it that one of
Maxim?s problem is how to inform the partners in
this conversation that we have created these new
objects of discussion.  His suggestion for one
extreme case: using {lo ro broda} to bring in all
the brodas that are or might be or ?, does not
work, since, as he admits, 1) there does not
really seem to be such a supply of brodas to
bring in and 2) even if there were, it is
(almost?) never what we want to bring in.  The
question is legitimate, but the answer seems to
be that we do it in Lojban as we do in English
(etc.) : we say something that we thinmk will
trigger the understanding of the others.  If it
does not, we say something else, usually more
precise.  And we keep at it until we get the
result we want (or say ?To Hell with it!?). 
Mayhap a logical language ought to have a better
device, no better device has yet emerged, unless
it is to start off using the expression we would
have used that achieved the desired result.  But
that is hard to predict, so we work through the
possibilities in order, since saying more than is
needed is a uncooperative as saying too little
and stopping.

There is a kind of object-meta muddle going on
here, the above bit about the meaning of {lo ro
broda} (and, I suppose, even of {lo broda},
though less so) being an example.  What {lo ro
broda} means is ?all brodas.? Within the
conversation that is all the brodas within the
conversation, but we can only say it this way
from outside the conversation, where (in a larger
conversation, as it were) we recognize that there
are brodas that are not covered  -- one capable
of being brought into the conversation, perhaps. 
But this does not imply that {lo ro broda} means
?all the brodas in the conversation.?   Indeed,
it cannot, since ?in the conversation? is
meaningless within the conversation, where
meaning is first determined.  That is, you cannot
refer to the object language in the object
language (in any transparent way, anyhow) and so
cannot bring metalinguistic comments down into
the determination of meaning within that
language.  

Sets, masses and the like.  I agree with xorxes
that C-sets have no role to play in Lojban
outside of discussions of set theory (something
that has yet to occur) and that {lo?i} and the
like either have no role to play correlative with
their length or have some use not yet established
(I would suggest corporate entities, if {loi}
can?t be shifted over to them once the
predication issue gets dealt with).  The same
seems to me to apply to ?mass,? although in this
case because the word may mean any of several
things and is actually ambiguous in most uses in
the word lists and CLL.  For most of the uses of
set-language and mass-language, reference to (a
bunch of) brodas is sufficient, if the
corresponding logic is understood and applied. 
Indeed, reading both ?set? and ?mass? in this way
is about the only one that makes sense of many of
the things said. The use of ?set? was a response
to the need to deal with plurality when there was
no better means available (or at least known) in
logic.  The use of ?mass? is too muddled to
account for but most of the apparent factors have
been pulled out in the notion of collective
predication once the fuller logic became
available (known).

Outer quantifiers.  I am conflicted on this.  CLL
makes it clear that, just as the primary referent
of {lo broda} is all brodas, in context {lo
broda} refers to some (not necessarily all) of
those brodas.  I now think that {lo broda}
primarily some brodas (no necessarily all) and
that {lo robroda} in context refers to all of
them (cf. {le broda} in CLL).  I do not think,
however, that that means that {lo broda} has an
implicit quantifier -- either internal {su?o} or
external {ro}, though I have less trouble with
the internal one.  The external quantifier, if
necessarily present, would force me to take {lo
broda} as being distributively predicated (in
fact, distributivity is defined by means of the
universal quantifier) and I want {lo broda} just
to refer to the brodas involved, regardless of
how various things are predicated of them.  It
seems to me that any other external quantifer
would have the same effect, forcing
distributivity, whereas the internal quantifier
does not: it just counts the brodas referred to. 
So, {ro lo broda} is very different from {lo ro
broda}: the first is distributive and need not
refer to all brodas, the second is not yet
committed on mode of predication andd does refer
to all brodas.

The argument about whether there is a thing over
and above the students when the students surround
the building is the worst kind of vacuous
metaphysics against which the Logical Positivist
were ever warning us.  The point is: it doesn?t
matter which way you say it or even which way you
think it, the logic is exactly the same (though
what picture you have in your head when you see
certain expressions may be different, but
language is not about the pictures in your head  
-- the fact that you see it as a new entity
doesn?t mean there is a new entity there andy
more than the fact you don?t see one means that
it isn?t there).  In natural languages we
regularly switch back and forth between the two
locutions: ?several students?/  ?a group of
students,? without any significant change in
meaning.  If you ask a ordinary person whether ?a
group of three students walked down the street?
meant that four things (the students and their
group) walked down the street, he would quite
rightly think you were a bit more than tetched. 
To be sure, the two locutions ?three students?
and ?a group of three students? have different
grammars (one is plural, the other singular, for
example) but they do not have different logics
(as McKay eventually admits, after having talked
entity language: ?plurality? and the like, for
many chapters).

It has dawned on me slowly that the problem I am
having with Maxim is that he is trying to teach
Grandma to suck eggs, a proverbial form of
chutzpah.  As far as I can tell, he has been
working with Lojban for only a couple of months;
xorxes has been producing paradigm Lojan text for
a decade or so.  So Maxim telling xorxes that
xorxes doesn?t understand what the Lojban says or
is saying something improper in Lojban
immediately strikes one as simple arrogance. 
That does not, of course, mean that it is wrong,
but it does mean that Maxim has to give really
good arguments for his case.  Of course, he was
under that obligation from the beginning, since
he is the one proposing changes.  So far as I can
see, his arguments have not met the challenge. 
The one that brought this home to me seems
particularly bad.  Briefly it seems to be
collective predication creates an entity that
does the work of the members together and that
entity cannot be referred to by {lo broda}, since
that is needed to refer to the distributive
situation, and so a sentence which uses {lo
broda}as subject of both a distributive and a
collective predicate is improper.  But, of
course, neither of the first two clauses has been
established (and, indeed, it looks almost as if
the claimed impropriety of the sentence were
taken as evidence for them) and thre conclusion
holds only if the first two are established (but
not even then necessarily).  To be sure, CLL,
written in antiquity, logically speaking, does
tend not to be clear about where the
distributive/collective distinction lies, placing
it in the entities, as we might now say, rather
than in the predicates and so Maxim is to that
extent justified in speaking of these entities as
significantly different.  But he seems not to
have absorbed the lesson that this difference is
merely a manner of speaking, not a reality that
forces some linguistic feature.  Indeed, even if
he does hold that the students form a new entity,
he has not shown that that new entity cannot be
predicated of in two different ways and thus
allow one reference to it to serve as subject to
two predications of different mode (the fact that
set theory doesn?t allow these other predications
directly does not mean they don?t occur and
deserve representation: indeed, we could do the
whole of the {lo} discussion in terms of
non-empty C-sets as well as plural logic or
L-sets).  The argument actually goes tother way
round: since the sentence with {lo broda} as
subject to both predications is proper (ones of
this sort are in the accepted lojban corpus), it
follows that {lo broda} of itself does not
indicate mode of predication.



To unsubscribe from this list, send mail to lojban-list-request@lojban.org
with the subject unsubscribe, or go to http://www.lojban.org/lsg2/, or if
you're really stuck, send mail to secretary@lojban.org for help.