classifiers and number

[Suzanne E Kemmer]

I have been wondering throughout the discussion of
classifiers whether the direction of implication
that is being talked about is simply reversed.

Instead of looking at the hypothesis,
"if a language does not have plural inflection or
plural words, then the language will most likely have classifiers",

consider "if a language has classifiers, it
does not have number inflection as a general property of
noun phrases". (typically the number inflection in
classifier lgs. is either restricted or optional or both)

The latter hypothesis was discussed by Greenberg
in "Numeral classifiers and substantival number"
(in Selected Writings of Joseph H. Greenberg,
ed. by K. Denning and S. Kemmer). He suggested
that this generalization has a pretty strong empirical
correlation; and also attempted to generalize
over the apparent exceptions: where a classifier
lg. does have number inflection, it actually
has a collective/singulative opposition instead
of a singular/plural one.

He explains these generalizations in terms of the semantics of bare
nouns in numeral classifier lgs. If I can reconstruct this off the top
of my head, it goes something like this: bare nouns in numeral
classifier lgs.  are generic or collective; numerals don't go directly
with such nouns because numerals count individuals; so the classifier is a
unitizer that intervenes and makes the nouns countable.  Collectives
in general are semantically at odds with plurals. Plural forms
are therefore not general in these lgs.; only a subset
of nouns where individuation is very high (e.g. humans)
might take plurals.

The arguments for all of these things are pretty involved
so please don't take my simple statement above
as the whole story.

--Suzanne Kemmer