Re: [lojban] Re: {lo} down

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

I too am pleased to see that there is a growing
focus on creating Lojjban text and other vehicles
(hopefully including effective teaching aids). 
On the other hand, after 30 years in the
business, I am concerned that the language being
taught and used is neither defective nor
mislabelled.  Calling it the logical language
carries along some commitment to a special
relationship with Logic.  And so, I speak out
when I see some breech of that relationship.  In
most cases, a perfectly good language can be
constructed in the way I oppose, but then it
ought not be called "logical."

So, to the issue of {lo}.  

First of all, here is the easy rule for {le} and
{lo}.  If you have a particular thing (or things)
in mind and want to refer to it by reference to
some property (with the intention of halping your
hearer identify what that thing), then you may
use {le}.  Because it is your selection of the
object, not its relation to the property which is
vital, the preoperty need not be one the object
actually has, so long as it helps others read
your mind.  (The classic good use is for the end
of The Crying Game -- or an old Mike Hammer novel
-- {le ninmu cu nanmu}).  Otherwise, to refer to
something by it sproperties, use {lo} (but now,
since it is the preoperty that counts for
reference, the object actually has to have it).
Everybody agrees about this rule, with various
minor caveats.

The disagreement is about what {lo broda} refers
to: a bunch of brodas or Mr. Broda.  On the first
view (it's mine, so I put it first)to say {lo
broda cu brode} is to say directly that some
brodas (make this as indefinte as need be -- no
clue about how many and certainly not about which
ones)are brode, in the straightforeward way in
which Rex is a dog, and so on.  On the other
view, {lo broda cu brode} is to say of an
individual, Mr. Broda (using this name is meant
to have a certain denigrating effect, but it is
the most convenient handle for this thing in any
case), is brode (exactly how is unclear -- saying
it is just like Rex is a dog creates some problem
as does saying that it is in some special way). 
The thing about Mr. Broda is that his having a
property is not completely independent of the
properties of ordinary brodas: if some brodas are
brode, then so is Mr. Broda, for example.

For the msot part, the choice between these two
interpretations makes no difference in practice
(hence the lack of interest, perhaps, in today's
Lojbanists).  There are a few cases where they
differ in what is correct to say.  1) {lo broda
cu broda} and {lo broda cu brode}, where {brode}
is part of the definition of {broda} and {lo
broda cu brodi} where {brodi} is related to
{broda} by well-known generalizations.  In prelo
(the first position), all of these are
technically false when there are no brodas --
there can't be a bunch of things that don't
exist.  In xorlo (the other view) these are all
true (well, the third one is open to dispute: 
unicorns are unicorns and unicorns are horselike,
but are unicorns white?). In prelo, the fact that
we often accept sentences of this sort even when
they are strictly false is covered by the Lojban
permission to leave out all manner of details if
they are "obvious."  In these cases, the details
are matters of mode.  The first case is just an
instance of a tautology and so, under some
formalist mode is detached from any actual
reference.  Something similar happens with the
second, which is just stating a part of a
definition, again without any reference.  The
thrid is trickier, but can be covered by claiming
an implicit appeal to cultural knowledge.  In any
case, their truth can be accounted for.  In
xorlo, these truths are part of the rules for
assigning properties to Mr. Broda, for these go
beyond just the cases of the actions of ordinary
brodas, though hopefully not in such a way as to
lead to difficulties or falsehoods.

Another (and, indeed, the only other one I know
of) case of difference is in opaque contexts,
contexts from which sumti cannot be fronted or
generalized upon without risking going from a
true claim to a false one.  So, if I hit a dog,
there is a dog which I hit.  But if I want a dog,
there is not (generally) a dog that I want -- any
dog (within some limits probably) would do.  In
prelo, this problem is solved by putting in that
place of "want" ({djica}) not a reference to the
sort of things wanted but to an intensional
object, in this case an event, which involves
those kinds of things: {mi djica tu'a lo gerku}
(what event is left very vague in this example). 
The non-fronting is now accounted for by the fact
that {le gerku} refers not to actual dogs in this
world but the dogs in a (possibly) remote world,
defined by the intensional context of the event
phrase.  We can, obviously, not take reference to
them (or generalizations from them) into this
world with any guarantee that they will succeed
in referring to something here (that the worlds
overlap in just the right way).
   

In xorlo, these intensional cases are treated as
extensional for Mr. Broda.  That is, that Mr.
Broda is in the extension of {mi djica} is just a
given as it is for any other predicate.  To be
sure, in this case the connection to ordinary
brodas is more remote, since no ordinary broda
may be what I want.  On the other hand, there
must be some connection with ordinary brodas if
Mr. Broda is to do his job.  The connection is
counterfactual: "if such-and-such were the case
for some ordinary brodas, the my desire would be
met" (and similarly for other such cases).  The
"if" cause here corresponds to the event phrase
in the prelo sentence and the consequent clause
is something about the gap (or whatever) that
raises the intensional issue being filled (or
whatever -- there are a lot of these predicates
and the exact apodasis ("then" clause) for each
needs to be laid out). The protasis ("if" clause)
also needs to be laid out for each situation more
clearly: it is generally clear that "brodas
exist" won't do, but what precisely is needed
often takes some work -- if it is important.  The
general "something involving brodas" usually is
enough here as in prelo.

The virtue of xorlo is that it seems perfectly
natural to someone whose home langauge has
something like the "a" of English, that can be
used in almost ever situation (varying with
plurals, "sm," and mass-noun constructions);  for
aprticular cases and generalities, extensional
and intensional.  So it is easy to get used to.

The vices are that it introduces a mass of new
entities of a sort unknown to Lojban before and
tha, in the process it complicates the semantics
by having a mass of new rules for one kind of
sumti that have no parallel for any others
(including just the quantified forms of {lo}
expressions.  I have not, in fact, been able --
after several years of trying -- to construct a
completely adequate semantics for xorlo, though I
have come more or less close several times, only
to have my attention called to some details that
did not fit. Xorlo also hides logically
significant facts very throroughly -- especially
the counterfactual connections buried in
"intensional" occurrences of {lo}.

Prelo is the negative image of xorlo: its virtue
is that it has a standard semmantics, uses only
standard objects in the same way as they are used
for all other sumti, and leaves all logically
significant moves out in the open (the mark of a
logical language, after all).  On the other hand,
it is harder to learn (leaving out {tu'a} in
{djica2} and the like is a very common beginner
-- and some more advanced -- Lojbanist's text).