2011/11/13 Jorge Llamb=EDas <jjllambias@gmail.com&= gt;

S =3D "su'o lo ctuca cu tavla ro le tadni"

The truth conditions of S are as follows:

- There are individual entities C whom Speaker denotes as "lo ctuca&qu= ot;
- There are individual entities T whom Speaker denotes as "le tadni&qu= ot;
- There is a predicate relationship (x1, x2) denoted as "tavla",<= br>
interpreted as the set of all ordered pairs of individual= s such that

x1 talks to x2 (full encyclopedia definition of a predicate can be cited as needed).
- Ex in C, Ay in T tavla(x, y) is true.=A0 In other words, for some x
among C, for all y among T, x talks to y.

I think that ctuca(x) needs to be an explicit par= t of the truth conditions due to the veridicality of "lo".=A0 On = the other hand, I don't think that "le tadni" entails tadni(y= ).=A0 Rather, the canonical explanation of le tadni as "something desc= ribed as tadni" seems to invoke somehow the pure _intension_ of tadni.= =A0 I.e. there really don't have to exist su'o tadni in order for l= e tadni to exist; rather, "tadni" seems to give a cognitive clue,= or provide an index, to identify and track certain entities based on some = salient relevance or resemblance.=A0=A0 Of course it is possible that le ta= dni cu tadni, but it's not necessary.=A0 That's how I understand it= at least.

The other important difference between the two gadri is specificity; th= e specificity of "le" can and probably should be formalized by us= ing some sort of choice function; David Hilbert's epsilon operator, whi= ch I think is the original example of a choice function, has been used by s= ome linguists to formally capture definiteness in natlangs.=A0

=A0

I doubt anyone has any issue with that. The issue in this round of
discussion seems to be about the "given model (universe of discourse)&= quot;
part. If you take that as a given, there is no discussion to be had.
If you don't take it as a given, then in some context the individual entities whom Speaker denotes as "lo ctuca" might be John, Alice = and
Mary, while in a different context they might be math teachers,
English teachers and biology teachers, or good teachers, bad teachers
and regular teachers, or ... And given that we can have different
domains, there is the issue of what happens when you want to jump from
one domain to another, when you want to deduce what happens in one
domain when all you have are facts from a different domain. Or how can
Speaker give enough information so that Listener can figure out what
the domain is without too much trouble.

Speaker can have many things in mind that he choo= ses not make explicit, but if "lo" is taken to be a nonspecific a= rticle, then I don't think that "su'o lo ctuca" ever impl= ies a truth-conditional commitment to the identities (specific) of the deno= ted entities, except in the trivial case in which the model contains exactl= y one ctuca.=A0

It's not 100% clear to me whether unquantified "lo ctuca"= generally implies a commitment even to _existence_, much less to identity,= of denoted entities.=A0

The strength of "le" is that it = implies a commitment to both existence and identity, though I hear this gad= ri may be on the wane in favor of "lo".

Over all, things seem very clear to me only when "le" and exp= licit quantifiers over "lo" are used, as in this example "su= 'o lo ctuca cu tavla ro le tadni".=A0 I think it would be interest= ing to analyze other examples in a similar way.

=A0

mu'o mi'e xorxes