[YG Conlang Archives] > [jboske group] > messages [Date Index] [Thread Index] >
Adam:
> de'i li 2002-11-10 ti'u li 13:19:00 la'o zoi. And Rosta .zoi cusku di'e
>
> >Quantifiers can be divided up into 'cardinals' and 'fractionals'
> >Fractionals mean "x per y": "x-per-y broda cu brode" means that
> >out of every y broda, x of them are brode. Y is always 1 or more
> >When an "x-per-y broda" claim is made about a world in which
> >there are fewer than y broda, the claim is meaningless in the
> >sense of uninformative, but is "deemed" to be technically true
>
> This is clearly not the case with how 'percent' is used in natural
> languages.
It basically is, but with the fairly trivial wrinkle that you apply
factoring and approximation, so that 33% is equivalent to 1/3.
> Suppose that there is a group of 30 students, 10 of which
> are male. People go around saying things like 'cici ce'i lei tadni cu
> nakni' in such situations all the time, and in a very meaningful and
> informative way. (Now, perhaps they *should* say 'pa fi'u ci', but if
> we're going to go that way, then there's no need for 'ce'i' at all.)
All true, but the details of how "percent" works in natlangs isn't
really relevant unless you acceept the basic model of fractional
quantifiers and want to formalize them more thoroughly so as to
reflect natlang usage.
> >(See below for more on what "deemed" might mean.) So "x-per-y
> >broda cu brode" strictly means "Either there are at least y
> >broda and x of them per y broda are brode, or there are fewer
> >than y broda". In other words, fractionals are 'nonimporting'
> >(Had they been importing, they would have meant "There are at
> >least y broda, and x of them per y broda are brode".) The
> >rationale for this is to make DeMorgan work more elegantly
> >Cardinals (other than no) are importing. {su'o, pa, re} are
> >cardinals. {so'e, ro, me'i ro} are fractionals. {no} neutralizes
> >the cardinal/fractional distinction (and is by deduction
> >nonimporting)
>
> We have 4 quantifiers: {ro, no, su'o, me'i}. I consider that ro = naku
> me'i and me'i = naku ro, and that su'o = naku no and no = naku su'o
> Thus, ro and me'i have opposite existential import, no matter what
> import we decide for either of them. I think that these equivalences
> are much more basic than DeMorgan's, and should be preserved in
> whatever system we finally come up with
I see no problem with no & su'o.
But various folks told me that if a set is empty then claims about
"99%" of it are as true as claims about "100%" of it. So I don't
yet grasp why it is that ro and me'i should have opposite import.
--And.