[YG Conlang Archives] > [jboske group] > messages [Date Index] [Thread Index] >
xorxes:
> la and cusku di'e
>
> >A further problem of sorts is that {da} is of course not a licit
> >antecedent for an anaphor on the other side of a scope boundary,
> >but {le'i} is a licit antecedent, and it would be good to have
> >an anaphor that would take the le'i as its antecedent, even if
> >{ri} is not the anaphor for this job. The sort of anaphor that
> >would be helpful would be something like a {go'i} that picks up
> >its antecedents by backcounting through sumti
>
> I think the solution should be something like this, the
> same one for all/most pro-sumti:
>
> 1- If the pro-sumti is overtly quantified, then the new
> quantification is restricted to the same set over which
> the antecedent's quantification was restricted
>
> 2- If the pro-sumti is not overtly quantified and is still
> within the scope of its antecedent's quantifier, then it
> is a variable bound by that quantifier
>
> 3- If the pro-sumti is not overtly quantified and it is
> outside the scope of its antecedent's quantifier, then
> it is taken to have a default quantifier (ro?, su'o?)
> that starts a new quantification over the same set over
> which the antecedent was quantified
It's a good suggestion (of course). I had been thinking
about 1 yesterday (1 & 2 are documented on the wiki), and
3 makes 1 more compelling.
My two reservations are these:
A, Might we ever want to get interpretation 2, bound variable,
but apply a quantifier or relevant LAhE to it? (E.g. if
the the variable already expresses a za'umei.)
B. Is it a problem if the bound variable is reused so
often in mode 2 that the speaker/hearer forgets what the
underlying set was?
But unless you think these, and A in particular, are serious
problems, I think your solution is likely the best one.
BTW, I have drafted but not posted a wiki page arguing that
vo'a-series sumti anaphors should behave like ri, & hence
be eligible for the scheme you propose, but that nei/no'a
sumti anaphors should always follow pattern 1 (precisely
because they always do have an explicit or implicit
quantifier or LAhE.
--And.