A modest proposal #2: verdicality

[Dylan Thurston <dpt@abel.MATH.HARVARD.EDU>]

{ga'inai loi satcyselsku zo'u stidi}

[The attitudinal paper mentions {galtu} w.r.t ga'i, which seems
malglico.  {tolbanli}, maybe?]

[I'd like to stick in a tense for "prematurely", but I couldn't find
one.  Why isn't there a distinction:  natural beginning vs. real
beginning?]

If I understand correctly, the usage of {le broda} as "that which I
choose to say brodas" was introduced to mimic [with reason] a feature of
natlangs:  one can say "the cat in the painting", even though that cat
is not, strictly speaking a cat, but only an image of a cat.

There is a looseness here (clearly necessary for succinct
communication).  Lojban analyses it as being a looseness in the
description.  I'd like to argue that, instead, it's a looseness in the
predication "cat".

Consider another example: a painting of a man walking across a field.
How would one express the English sentence

        The man is carrying a heavy load.

in Lojban?  The man is not a man, but that is covered by the use of {le
ninmu}; but neither is he carrying anything.  The appropriate Lojban
gismu, {bevri}, requires a source, destination and route and hence
movement (IMHO anomalously, but that's another issue), of which there is
none in the static painting.

But this is just the ordinary loose way of speaking, expressed in Lojban
by the cnima'o {sa'enai}.  Just because Lojban is intended to be a
logical language doesn't mean that speakers have to be pedantically
strict (unless, of course, they say so, with {sa'e}).

So I'd like to see some justification for why the verdical/non-verdial
distinction must be made with every reference to {le/lo toldi}.  Other
distinctions make sense:  the difference indivual/mass/set/ideal affects
the semantics of the rest of the sentence (though I don't understand the
difference between mass and set yet), and it's clearly necessary to make
the distinction between quantified (existential) and specific
references.  But I don't see why +/-verdical distinction has to be there
when it can just as easily be made with a cnima'o following the brivla.

If this has been discussed before, my apologies.

c'o mi'e. dilyn.



>From lojbab
To: dpt@abel.MATH.HARVARD.EDU
Subject: Re:  A modest proposal #2: verdicality

To answer the question about veridicality takes some history.  Old TLI
Lojban does not have "lo".  It has only "le".  It also has confusion
%^). pc and I made a kind of table of all the distinctions that were
embedded implicitly in the Loglan articles, and came up with what we
have now.  Specifically (caps refer to TLI Loglan words, not selma'o)

LE   corresponds to le or lo, and is definite and specific, except when it
     isn't %^)
LO   is Lojban loi - the mass of all things meeting the description.  Except
     for the one example translated as "the two men (together) carried the
     log across the field, for which JCB used LO implying loi, but Lojban
     would use lei.

This was all there was in early TLI Loglan (~ 1974).  Then was added
restrictive relative clauses for what might be called veridical
descriptions.  This goes well with Jorge's argument that "lo" shoudl be
euqivalent to "da poi".  For many years the equivalent of da poi broda
was the way to express a veridical individual.  The problem came with
numbers of individuals, or the "SE SORME" (= ze mensi) question.  JCB
was convinced that quantified descriptions were natural without an
explicit article like "le".  He waffled back and forth between allowing
and disallowing "ze mensi" (equiavlent).  He also was wishy washy as to
what it meant when it was used:  sometimes "le ze mensi" sometimes "ze
le mensi" sometimes ze da poi mensi.

In the late 1970s, pc and others attempted to nail down default
quantifiers for the existing Loglan articles, at a time when SE SORME
was not a legal sumti, and came up with what is essentially the system
today.  When JCB decided that he could not live without the SE SORME
concept, he made it work with YACC grammar, but never defined what it
meant in terms of quantifiers.  Then in 1986, when at the first DC
LogFest we were reviewing JCB's draft essay on the state of the language
(which became _Notebook 3_, 1987), we decided that it had to be "ze da
poi mensi".  All this was pre-Lojban.

Meanwhile in the late 70s, there was proposed some auxiliary articles
LEA (the set of all those meeting the description) and others.  But the
usage of LEA was not as a "Set", but rather was identical to what we now
use as "ro lo" - used in making universal statements about veredicals.
He also added an article LOE for "the typical".

In Oct 86, when I visited pc still working on the TLI Loglan dictionary,
I was trying to make sense of all these articles, and especially of the
reasoning behind the default quantifiers.  "LEA" was clearly dealing
with veridicality, and we recognized that the two usages of LO were
respectively veridical and non veridical.  Thus, when JCB's copyright
claim on the words was made a couple of months later, and I started
reinventing the words, the first iteration was what is now lo/le
loi/lei, but with the outer quantifier on "lo" being "ro" and not the
current "su'o".  Meanwhile we had also recognized that there was still
the issue of "sets" - the supposed defintion of "LEA" that did not match
JCB's usage - that would be used in talking about set properties like
cardinality.  So we added "lo'i/le'i" recognizing that we had a good
pattern going.

We tried to fit the other Lojban articles into this scheme, and found
they worked well.  Hence where JCB had only "LA" (= la), we added
la/lai/la'i.  And where JCB had "lo'e", I had to decide whether this
belonged in the veridicals or the non-veridicals (le'e or lo'e) and
relaized that there was a plausible meaning for either, so we added
both.

pc and I then hammered out default quantifiers for all of these in
several 2 hour long phone calls, with Nora by then getting involved as
well since she had moved in with me.  Until Jorge came along, people
thought that our systenm was perhaps overkill, but there weren;t any
logical holes in it (though we kept getting the default quantifiers
mixed up, and had to work them out repeatedly from first principles).

We chose the current quantifiers for "lo" when I found in actually
trying to use the language that I hardly ever wanted to make a universal
claim about all brodas - that may be something logicians do, but if you
are concerned about making true statements about the real world, it is
not wise to make universals.  Far more natural seeming was the default
quantifier su'o lo ro.  It also conveniently sidestepped the "SE SORME"
question because most such usages were indefinite veridical, and "ze
mensi" thus would equate to "ze lo [ro] mensi".  It was only later that
we had people argue that "ze mensi" might refer to nonveridicals "= le
ze mensi".

"lo" thus drifted towards being similar to English indefinite, but with
that stressed veridicality that was LEA's primary reason for existence,
so that you could make a universal claim about brodas succintsly with
"rolo".  It was not until Cowan and I had an argument about "lo xanri"
(which led to a change in the definition of xanri) that anyone ever
questioned the usefulness and/or importance of veridicality as the
central criterion of the "lo" series - even more than the
non-specificity.  Indeed, usage came to allow "lo" to be used rarely as
a specific, with some fats-talking words about implicit restrictions.
Nowadays, I think the argument has been made that such implicit
restrictions constitute a limitation of the universe of discourse so
that "lo broda" becomes "su'o da poi broda" to all intents and purposes.
And the SE SORME question has come full circle so that it can mean
either "ze da poi mensi" or "ze lo mensi" because the two are
essentially the same in meaning.

But if we eliminate veridicality as the central tenet for the lo/lo'V
series, I am not sure that there is any justification for it to exist at
all.

lojbab