Indo-European and Nakho-Daghestani languages: a short typological comparison
The grammatical categories of class, number and case are characteristic for Indo-European (IE) nouns in the same way in which they are for North-East-Caucasian (NEC: Nakho-Daghestani, the family comprising the languages of Daghestan, as well as the neighbouring Chechen and Ingush).
Nouns in both IE and NEC are inflected for class (that we traditionally call grammatical genders), number and case. I will restrict the comparison to IE on the one side, and to the Avar and Dargwa language families on the other side. In Avar and Dargwa, as well as in Indo-European (IE), nouns in the singular are divided into three grammatical classes.
The class I contains names of male beings, or of notions assimilated to masculinity:
IE: equus (horse, Latin); anthropos (man, Greek)
NEC: waša (boy, Ghodoberi); urši (boy, son, Dargwa)
Names of female beings or assimilated as such belong to class II:
IE: femina (woman, Latin); gunē (woman, Greek)
NEC: ila (mother, Ghodoberi); rursi (daughter, girl, Dargwa)
The other animate and inanimate nouns and abstract notions belong to class III, conventionally called neuter in IE linguistics:
IE: templum (temple, Latin); zugon (yoke, Greek)
NEC: muču (branch, Ghod.); awl (field, Darg.)
This division of the realia into classes is an areal feature which is very rare on the planet. Most languages do not share this feature. The historical expansion of Indo-European and Semitic (mostly through Arabic) should not obscure the fact that it is an extremely rare feature.
In the Caucasus-Middle East this is a trait shared by: Indo-European, North-East-Caucasian and Semitic (where it is reduced to two classes: masculine, feminine). Languages possessing grammatical genders, or nominal classes, on the IE model, tend also to be locally defined. Such are the Bantu languages in Africa, the Pama-Nyungan languages in Australia, and some language families in the Middle East and the Caucasus: Semitic, Indo-European and Nakho-Daghestani.
These are the only regions on the planet where languages display the classificatory system of the grammatical genders (or nominal classes). No indigenous American language on both continents, no Asian or Siberian language share this trait. It thus appears only in Africa, in the Bantu languages, and in Australia in the Pama-Nyungan family. It is thus an areal feature, shared as I said in the Old World by Indo-European, North-East-Caucasian and Semitic.
Classes are usually marked word-initially in NEC languages and word-finally in IE. The first (male, active) class is thus marked by an initial w– or u– in NEC: waša (boy, Ghodoberi); urši (boy, son, Dargwa); by the final –s in IE: equus (horse, Latin); anthropos (man, Greek).
A language from the Avar family such as Ghodoberi allows both initial and final markers. When we compare the formation of adjectives with class indicators in final position, we find thus:
CLASS I CLASS II CLASS III
IE nov-us nov-a nov-um (new, Latin)
NEC bečeda-w bečeda-j bečeda-b (rich, Ghodoberi)
A noun with no apparent class marker will thus be defined by its corresponding adjective. Nothing in homo indicates morphologically the class, but it becomes immediately apparent with an adjective: homo novus. The same happens in NEC.
The plural of the third class (neuter) is identical with the feminine singular: templum novum (sg.) – templa nova (pl.). The same happens in NEC.
We thus arrive at the already proposed theory about the PIE (proto-Indo-European) ergative *-s theory. IE marks what we conventionally call the nominative case with an *-s. This goes against the fact that in most languages nominative cases which mark the subject of a sentence are unmarked. The proposed ergative theory (whereby PIE *-s is the marker of an ancient ergative, such as in the NEC languages, would help solve this problem, but it hasn’t been generally accepted so far. (Ergative is the case in which the actant, the subject of a transitive verb, takes a special termination.)
We now find that when used with transitive verbs (that is: when they acquire an active function), NEC nouns have in languages such as Avar (and the distantly related Chechen) the termination -s. This is the ergative case. When, for instance, the Avar či (man) is used with a transitive verb it takes the termination -s: čiy-as. The Indo-European active-masculine termination of the nominative (*-s) could thus appear to be the survival of a previous areal ergative case and would not go against the fact that in most languages nominative cases are unmarked.
A hand or a tree could be seen by speakers as active, being thus able to function in the ergative case with the termination becoming permanent regardless of the actual class (gender) of the noun. I personally met instances in which a Chechen boy would naturally say kafes, in a sentence in which the coffee had burned his hand, with the termination of –s the ergative case that the otherwise inactive word kafe wouldn’t normally take. Kafe-s was formed in such a sentence in the same way in which da (father) would be in the ergative: da-s (cf. IE: equus, anthropos, etc.).
That would explain why the nominative IE *-s was not rigorously used for masculine. The Romans themselves could never decide whether dies (day) was masculine or feminine. It would solve the problem if we take it as a survival of another areal trait: an old ergative, such as the Avar-Chechen ergative case in *-s.