[YG Conlang Archives] > [jboske group] > messages [Date Index] [Thread Index] >
Jordan:
#On Tue, Nov 26, 2002 at 02:19:52AM -0000, And Rosta wrote:
[...]
#> Okay, but if ka'e = kakne, then we need to establish how to talk
#> about imaginaries as imaginaries.
#>
#> su'o mu'ei
#> su'o mu'ei je ca'a
#> su'o mu'ei je nai ca'a
#>
#> would do the trick, but they are extremely cumbersome. We don't
#> often need to talk about imaginary dogs or even detectives, but
#> we do often need to talk about imaginary events. And it has been
#> said by Reputable Sources (JWC) that CAhA does this; yet that
#> does not square with the actual definition of ka'e or its
#> apparent relationship with kakne.
#
#I'm not sure if I followed the above exactly. Doesn't {nu'o}
#get you the ability to talk about things which aren't {ca'a}? Or
#is that not what you mean...
For things like "Sherlock Holmes was a detective", "mi djica lo nu"
-- we variously need to indicate whether the predication (is a detective,
is a nu) holds in This World, or not in This World, or not necessarily
in This World. It has been suggested by several people that ca'a,
nu'o and ka'e serve this function: Sherlock Holmes nu'o is a
detective; mi djica lo ka'e nu klama la brada plise.
It's not the same as saying "In some possible future (of Now) SH is
a detective", "In most possible futures of Now, such and such would
be the case". (Those are the meanings of cumki/lakne, su'o/so'e
ba'oi.)
#> > WRT mu'ei (and somewhat off topic), I just realized (while trying
#> > to think of a way it could be done as a NU) that all that ledu'u
#> > .... kei goo (or for whatever reason some prefer lo'edu'u, though I
#> > maintain that a statement is itself and trying to think of a typical
#> > version of a set of (itself) is weird, and seems to imply the set
#> > isn't singular) can be avoided by using a forethought stag:
#> > ro mu'ei gi do gerku gi mi citkygau do
#> > Which is kinda nice
#>
#> Can you give a couple more examples? I don't quite get what you
#> mean & can't work it out for myself because I haven't properly
#> learnt the construction.
#
#Since PA+ROI is a valid tag, you can put it after I before BO. Like:
# mi klama le zarci .isemu'ibo mi xagji loi cidja
#You can do this in forethought like so:
# semu'i gi mi xagji loi cidja gi mi klama le zarci
#(which is the same as the first).
#
#So, since mu'ei is ROI (and thus a valid tag) we can do the
#same stuff.
# mi klama le zarci .i romu'ei bo mi xagji ==
# ro mu'ei gi mi xagji gi mi klama le zarci ==
# romu'ei le/lo'e/whatever du'u mi xagji kei mi klama le zarci
Ah. That is a *very* neat idea.
#> Regarding {lo'e du'u}, the Jboske consensus was that {lo'e} is
#> only tangentially -- or rather epiphenomenally -- concerned with
#> typicality, and that actually {lo'e broda} means "In a world
#> like this one but with only one broda, the one broda is...". In
#> the light of that, {lo'e du'u} makes perfect sense. (Statements
#> explicitly about typicality can be made using {fadni}, {so'e roi}
#> and other such devices.)
#
#That seems like a pretty bad analysis of {lo'e} to me. What possible
#reason could there be for abandoning the idea of it relating to
#typicality? Something which relates lo'e broda to most broda would
#be better...
I think the best thing for me to do is look through my archives when
I get onto my home PC, and locate the message where I presented
the summary that led to the consensus. That would help you locate
the relevant portion of the Jboske archive, so you could do a quick
browse to see what was said and then articulate your position in
the light of that.
But to answer your point briefly, we felt that:
(a) true statements about typicality could be expressed in other ways,
(b) the mahoste gloss was only an approximation, and lo'e was in fact
a *generic* gadri, (and a generic gadri makes much more sense
systemically, as a gadri, than something that merely encodes a
notion of statistical typicality),
(c) the 'bad analysis' of {lo'e} is the best analysis of the meaning of
a generic gadri
(d) there was a class of examples where generic reference made
sense but not any notion of statistical typicality,
(e) there was not a class of candidate examples of use of lo'e where
a statistical typicality but not a generic notion was clearly required.
#> Xorxes and I had a very interesting exchange about the most
#> appropriate default gadri to use when the category is singleton.
#> I favour {lo'e}; he approves that on semantic grounds, but
#> prefers {le} on phonological grounds.
#
#You can say lodu'u (which tu'o du'u is of course a variant of),
#though that is strange because lo usually implies you don't know
#which member of lo'i, but here there's only 1 member of the lo'i,
#and furthermore you're about to say it right after the du'u, so you
#clearly sanji it. lo'e or le'e have similar strangeness. Mass and
#set articles, as well as li/me'o are completely wrong. So that
#leaves le... (which, yes, is also 1 sylable shorter, which is nice
#because du'u takes two).
I'll check my archives to try to locate the messages where this
was discussed. There were a few messages from me and xorxes
probably worth putting on the wiki, not because they decided
anything but because they spelt out all the issues and
considerations pretty clearly.
The argument against {le} is that it implies that the +/-specific
e-gadri/o-gadri distinction amd the distributive/collective distinction
lV/lVi are relevant (why use a gadri that encodes one value for
an attribute when the attribute is itself redundant?), which implies
that the set is nonsingleton.
--And.