Re: What's a gender?

According to Greville Corbett, and I think according to most professional
linguists, a "gender" is a "concordial noun class".
"Concordial" means that other words (maybe adjectives, maybe adpositions,
maybe members of smaller "parts of speech" (word classes), or whatever)
must agree with the gender of the noun.  The noun's gender may not be
actually marked on the noun itself.

Whether a linguist uses the term "gender" or the term "noun class" seems to
depend mostly on tradition in the linguistic area or linguistic family of
the language on which he or she is working.  For instance, most Africanists
call them "noun classes", while most writers on languages in the Indo-
European family or in the "Standard Average European" linguistic area call
them "genders".

In areas where there is no established tradition, linguists seem to call
them "genders" if there aren't very many of them (say, five or fewer), and
to call them "noun classes" if there are many of them (say, eleven or more -
- most such languages I've seen seem to have more than ten).  This habit
obviously has a "fuzzy" cutoff.

Aikhenvald has recommended a change in terminology; she wants to call
them "gender" only if the following two conditions hold;
1) there are four or fewer of them, AND
2) at least one of them has, as its semantic core, one of;
2a) all male humans OR
2b) all males OR
2c) all female humans OR
2d) all females.
Otherwise she wants to just call them "concordial noun classes".

(You are familiar with the "semantic core" of a gender; in French, for
instance, the "semantic core" of the feminine gender is all females, but
many nouns that aren't by any stretch of the imagination female
nevertheless are in the feminine gender; and a few nouns that, technically,
are or could be female, are not.)