Re: descriptions and quantification

[John Cowan <cbmvax!snark.thyrsus.com!cowan>]

Colin Fine asks about the semantics of LE and related stuff.
I am posting some general information for him and others to mull over.
It's a bit summary, I'm afraid -- eventually there will be a paper on
the subject.

LA, LE

These selma'o contain cmavo called descriptors, which corresponds somewhat
to the English articles: "the", "a(n)".  LA has a slightly less restricted
grammar than LE: only members of LA can be found in front of a name.

Each LE descriptor has a pair of associated default quantifiers, one "inside"
(representing the size of the set described) and one "outside" (representing
the size of what the descriptor refers to).  LA descriptors do not have
"inside" quantifiers, because everything after the LA is part of the name.
Thus "la ci cribe" means "that which is named 'Three Bears'" and does not
imply that the thing named has three components (it might be a rock group
or a restaurant).  The table indicates the default quantifiers for each
member of LA and LE.

There is a general pattern to LA/LE cmavo:  the central vowel is "a" for
things-named, "e" for things-described-as, and "o" for things-which-really-are.
This is followed by: nothing for individual(s), "i" for masses, "'i" for
sets, and "'e" for representative individuals (but "la'e" does not fit
here and belongs to another selma'o).  The distinction between individuals,
masses, and sets is an obligatory category of Lojban.  Members of LUhI
can be used to convert among these types, and members of JOI can assemble
sumti of one type into a collective of another type.

cmavo   quantifiers     full English translation

la      su'o la         "at least one of
                                 the thing(s) named"
lai     pisu'o lai      "some part of the mass consisting of
                                the thing(s) named"
la'i    piro la'i       "the whole of the set consisting of
                                the thing(s) named"
le      ro le su'o      "all of
                                the at-least-one thing(s) I describe as"
lei     pisu'o lei su'o "some part of the mass consisting of
                                the at-least-one thing(s) I describe as"
le'i    piro le'i su'o  "the whole of the set consisting of
                                the at-least-one thing(s) I describe as"
lo      su'o lo ro      "at least one of
                                all the things which really are (if any)"
loi     pisu'o loi ro   "some part of the mass consisting of
                                all the things which really are (if any)"
lo'i    piro lo'i ro    "the whole of the set consisting of
                                all the things which really are (if any)"
le'e    ro le'e su'o    "all of the at-least-one thing(s)
                                I stereotype as having the property"
lo'e    su'o lo'e ro    "at least one of all the things
                                which typically are"


> INDEFINITE SUMTI
>
> We are allowed the kludge "quantifier selbri". I take it this means
> pre-quantified "lo":
>
> ci cukta = ci lo cukta
> three of the things in the universe which actually is-a-book

This is correct.  Interestingly, this construct was present in Loglan as
long ago as 1960 (in the >Scientific American< article) but then dropped out.
It was later reinstated as a concession to natural-language habits, and because
it was grammatically safe.

> The grammar also allows "quantifier quantifier selbri", eg
> ciboi vo cukta
>
> It is tempting to conclude this means
> ci lo vo cukta
> 3 of the 4 things in the universe which actually is-a-book
>
> but that would mean that the [vo cukta] in [ciboi vo cukta] has a
> different meaning from a [vo cukta] tout court. This suggests that it
> means
> ci lo vo lo cukta
> 3 out of some 4 of the things .....

This analysis is probably correct.  Nobody has ever used one of these
double-quantifier constructions; indeed, nobody has used the other construction
which Colin quotes above:  LE + quantifier + full sumti.

> RELATIVE CLAUSES

I omit Colin's examples in favor of the following general remarks:

Relative clauses may follow many kinds of simple sumti, not just descriptors.
The relative clause is therefore outside the scope of the descriptor:

        le cukta [ku] poi se nelci me
        the( book ) such-that [it] is-liked by-me

may or may not be a book, due to the nature of "le", but it must really be
liked by me.  Since

        le do cukta
        your book

is defined as equivalent to:

        le cukta [ku] pe do
        the( book ) associated-with you

the "do" is in effect outside the descriptor scope also, though physically
within it.  Lojban "poi" relative clauses always restrict the description
truthfully, making

        le mlatu [ku] poi gerku
        the "cat" which is-a-dog

a reasonable sumti.

There is another cmavo, "voi", which is to "poi" as "le" is to "lo"; it makes
non-veridical restrictive relative clauses:

        lo mabru voi finpe
        a/some mammal(s) which-I-describe-as a-fish

perhaps referring to whales.  Here the "mabru" is accurate, whereas the
"finpe" only refers to what I have in mind.  "voi" was introduced at the
most recent pass over the cmavo list, and will not appear on most copies.

(Historical note:  in JL10 p. 39 it was said that "le mi cukta" is identical
in meaning to "le cukta po mi".  "po" in this case was an error for "pe".)

--
cowan@snark.thyrsus.com         ...!uunet!cbmvax!snark!cowan
                e'osai ko sarji la lojban