> Can you give a specific reference (one that does not ascribe countability
> as an implication of the default quantifiers)
No. But as John explained some weeks ago, the gismu were defined to
be neutral wrt countability, with uncountable interpretation encoded
by the application of loi and countable interpretation encoded by the
application of lo. The issue of default quantifiers shouldn't matter,
because any quantifier on lo will involve counting.
It now occurs to me, though, that perhaps lo is not supposed to force
a countable interpretation, in which case lo broda = da poi broda can
stand. That is, perhaps the idea is that lo broda by virtue of its
quantifiers involves counting, but does not require that each variable
bound by the quantifier be *a single* broda. Perhaps Lojban does not
have a way of forcing a countable interpretation.