indefinites - Lojbab's phonecon with Cowan (finally!)

[Logical Language Group <lojbab@ACCESS.DIGEX.NET>]

>> Another possibility might be that indefinite descriptions, and maybe
>> even quantified lo descriptions, need to be treated as singular
>> collectives for scope purpose, even when the number is plural.  This
>> makes indefinites more like masses
>
>That would make them exactly like masses! What would be the difference?

No, because a) the implicit outer is "ro"/"piro" b) the individuals do
not lose their identity and amalgamate into the mass.  "ci remna" in no
way talks about pieces of humans.

>> ("ci broda" = "piro lei ci da poi
>> broda" XOR maybe "pa lo broda cimei" - not sure what the difference
>> would be)
>
>The first one is a singular term, there is only one {lei ci broda},
>the one that I have in mind. The second is just one of all possible
>{broda cimei}, there are many broda triplets.

Nora pointed out one other distinction vs. masses that must persist -
the one that established our current mass setup:

le re remna/lei re remna/re lo remna/re remna cu bevri le pa mudri le
bancu be le foldi

Two men carry the log across the field.

The mass "lei" means they do it together, in a mass.  There is one event
of carrying.

le re remna says that there are two carryings each by one man,
presumably at separate times.  Likewise re lo remna.

The first thought is that re remna should probably also mean that there
are two carrying events, though perhaps there might be a stronger
implicature that the two carryings are related.

But apply this to "re tuple cu tuple le pa remna", and you get two
events of a man having a single leg, or rather two events of a leg
having a single man, with no hint that they are both being legs at the
same time.  A lizard with bad luck therefore has 5 tails (cu se rebla mu
rebla), but you may not have an easy way of saying he only has one at a
time.  I don't think we want this.  Indeed the goat's leg rule prohibits
it, I think.  If "le pa kanba cu se tuple vo da", presumably this means
it is not true that "le pa kanba cu se tuple pa da" and you thus cannot
expand a "voda" sentence into 4 "pada" sentences.

Note that however we play this game, the traditional interpretation of
the above log-carrying example has ALWAYS been that there is only one
log being talked about.  The distinction among the various values for x1
may make a difference on who carries the log on any given crossing, or
on how many crossings occur in toto.

Now for the biggie - my phonecon with Cowan the other evening, with Nora
standing by.  We didn't use the log carrying example - I wish we had,
but rather the men touching dogs that has dominated the discussion for a
while.

If we say "le ci remna cu pencu le ci gerku", we are talking about
exactly 3 men, and exactly 3 dogs, but 9 events of touching.

Because the contrast in usage between le and lo was only intended to be
one of veridicality and in-mindness, the corresponding

"lo ci remna poi co'e cu pencu lo ci gerku poi co'e"
which I have chosen as preferable to "voi" in restricting the universe to
the in-mind individuals.

I contend that this should also be equivalent to:

"ci lo remna cu pencu ci lo gerku"

which should also refer to exactly 3 men, 3 dogs, and 9 acts of touching,
although we have lost the definiteness.

I think this means that I favor the two sumti having identical scope,
but not being sure I know the implications of this last, I'll just stand
by the examples and anlaysis and let you-all figure out what scope is
meant %^).

Cowan's words:
"Descriptors refer to some particularized members of a group."
"Quantifiers delimit how many are in the group.  They do not multiply
 under scope."

Cowan thinks this insight is closely related to something pc has said
about the distinction between quantifiers and descriptors that he and
others had failed to understand before.  He was unable over the phone to
come up with a quote or reference that would help me find it and see if
the now absent pc also agress with us.

In passing, BTW, we decided that the goat legs quantifier rule is, like
djuno in another argument, unnecessarily complicated by an irrelevant
semantic factor - that legs are inalienable possessions in our mindset.

So we instead looked at the sentence
re jubme cu se tuple ?xo tuple

If quantifiers multiplied under scope, then the answer "vo" would give 8
legs.  But try to make this definite by putting it back into a "le"
sentence"

le re jubme cu se tuple le vo tuple

and we have the two tables somehow sharing the same 4 legs (perhaps they
are stacked tables, one above the other on the same set of legs, like a
freestanding bookshelf?)

So Cowan thinks that

re jubme cu se tuple vo tuple is not what we normally want to say in a
natural language, but rather:
pa jubme cu se tuple vo tuple
re jubme cu se tuple bi tuple

etc.

I basically agree, as does Nora, though our agreement is more specific to
the "lo" form of the above:
pa lo jubme cu se tuple vo lo tuple
re lo jubme cu se tuple bi lo tuple

Since the conversation with Cowan, Nora and I have talked further about
indefinite descriptions.  The following has NOT been run by Cowan, who
will be back on Monday the 10th, and probably swamped trying to read the
backlog for a couple of days.

Nora and I have gone similar but varying directions in regards to what
to make of indefinites "re jubme" and "bi tuple".  I think we both now
feel that equating them to the same thing with "lo" inserted may be too
simplistic.

Nora feels that "re jubme" should be ambivalent between "re lo jubme"
and "le re jubme" - the critical thing is the number and the Zipfean
abbreviation, and you might have some specifics in mind.  But then since
"le re jubme" is a superset of "re lo jubme", you get "re jubme"
equating to the former.

I lean at minimum towards saying that "re jubme" is veridical, though it
may be specific and definite as well.  This is as much based on English
idiom as anything else though (but then the whole inclusion of
'indefinite descriptions' is based on natural language idiom).  I note
that "indefinite description" isn't a very good label for the sumti
class, if indeed they are really 'definite', though I think the
indefiniteness really refers to the fact that we aren't specifying the
restrictions rather than the other meaning of 'definite' that has
occurred in this discussion.

But then I also might like to see the "indefinite" become truly
'indefinite' in this sense, i.e. ambiguous with respect to descriptor
type. re tuple could mean any of:

re lo tuple
lo re tuple poi co'e
le re tuple (possibly with veridicality also specified)
re le tuple
AND
(piro) lei re tuple
piro loi re tuple poi co'e
and maybe even
le'i re tuple

so
re lo remna cu bevri le pa mudri
means 2 carryings, but
re remna cu bevri le pa mudri
could either be one or two depending on context

Since indefinites are a Zipfean shortening, some increased ambiguity
would not be too much of a problem.

Even if we don't want to specifically include the masses (and sets) as
alternate meanings of an indefinite description, I think we definitely
want to insist that they somehow generate one event.  If re remna
carries pa mudri, we get a single event with a substructure of two
related subevents of carrying.

Now the question is whether the rest of you find any of these flags
worth saluting.

lojbab