[YG Conlang Archives] > [jboske group] > messages [Date Index] [Thread Index] >
xorxes:
> la and cusku di'e
>
> > > > Here's how it might be done:
> > > >
> > > > (tu'o) lu'i ko'a e ko'e
> > > > = the set that includes ko'a e ko'e and that excludes ro da poi
> > > > ko'a e ko'e na du ke'a
> > > > = the set {ko'a, ko'e}
> > > (where na is naku). Right.
> [...]
> > > The problem is that it fails for {lu'i ko'a a ko'e} and
> > > therefore also for {lu'i su'o broda}, which is the set that
> > > includes at least one broda but excludes everything which
> > > is not equal to every broda, i.e. it excludes everything if there
> > > is more than one broda.
> >
> > You're right, but it's still probably the most useful definition of
> > lu'i/lu'o, in that it yields something useful for ro & e.
>
> Testing some other cases, {lu'i ko'a u ko'e} gives just {lu'i ko'a},
> but I think we would want it to cover {ko'a} and {ko'a, ko'e}, i.e
> include ko'a and may or may not include ko'e.
>
> I propose instead the following heuristic definition:
>
> su'o lu'i <sumti>: a set X such that <sumti> cmima X and
> nothing not mentioned in <sumti> cmima X.
>
> This gives the same results as before for e and ro, but it
> also gives the right intuitive results for a, u, and su'o:
> {su'o lu'i ko'a a ko'e} is {ko'a}, {ko'e}, or {ko'a, ko'e}.
> {su'o lu'i ko'a u ko'e} is {ko'a} or {ko'a, ko'e}.
> {su'o lu'i su'o broda} is some set of some number of broda, but
> containing nothing that is not a broda.
>
> I'm not sure how to formalize it better, but I think that's
> what we want.
Given the laudable effort to give LAhE a strictly compositional
interpretation, it is necessary to formalize this heuristic definition.
I had already tried and failed.
--And.