I'm aware that a number of classifer languages have what might be called "plural classifiers" which -- unlike "normal" classifier -- force a plural, count interpretation, instead of being ambiguous between a single interpretation with a mass-y quantification. That is, in the general case, we see things like:
Onek-Ta rasta many-Cl road "a lot of road" or "many roads"
nei-ge ren that-Cl person "that person" or "those people"
But, there are also "plural" classifiers for these languages that disambiguate, such as:
Onek-gulo rasta many-Cl.PL road "many roads"
nei -xie ren that-Cl.PL person "those people"
Cool. But, I also know that Bengali and Mandarin both disallow the "plural" classifier with numerals (and I think other weak quantifiers), so
*tin-gulo rasta tin -Ta rasta three-Cl road "three roads"
*san-xie ren san -ge ren three-Cl person "three people"
Are you familiar with a classifier language that has a "plural" classifier? If so, does it only occur on things that aren't numerals?
I'm aware of languages in which numerals co-occur with the stem of the noun and not the plural form (cf. Finnish kissat "cats" vs. kolme kissaa "three cats"), but this is obviously parameterized since in English we say "three cats", not "*three cat".
The reason why I ask this is I've been concerned about the fact that all(?) classifier languages seem to place a classifier with the word "one", so you get Mandarin Chinese:
*yi ren yi-ge ren one-Cl person "a/one person"
Yet, there's no "plural" marker that occurs with the numeral "one" in languages that have a morphological singular/plural distinction. This should be surprising on the Doetjes/Borer-type story in which classifiers and number markers are all but the same thing, just morphophonemicized differently. Hence, I'm curious if the lack of co-occurrence with plural classifiers with numerals is a universal, or if that is parameterized in the same way as the Finnish/English distinction.