How does a language acquire gender markers?
In a seminal paper, Greenberg (1978a) presented an account of how demonstrative markers are grammaticalized in certain languages. He proposed an evolutional scale roughly as sketched in (1) below (the reader is referred to Green berg [1978a:61ff.] for further details. For example, by the ‘specific article’, he refers to” an article in contexts in which a specific but unidentified item is referred to, that is, there is a presupposition of reference” [1978a:62].)
(1)
What Greenberg’s scale suggests is that demonstrative attributes may undergo an evolution whereby they develop first into definite articles, subsequently into markers of specific reference, and finally into semantically more or less empty markers of nominality. In languages where the erstwhile demonstrative is marked for gender, this evolution leads to the rise of a new set of nominal gender markers at Stage III, since gender distinctions tend to be retained even if the demonstrative has lost its original function.
The evidence adduced by Greenberg also includes Khoe languages: In the “Hottentot” languages (= Nama, !Ora, and Xiri) and in Naron (= Naro) he finds evidence for a scale of evolution from unmarked to gender-marked nouns (Green berg 1978a:59). Whether Greenberg’s reconstruction is correct in every detail may be a matter of dispute; on the whole it would seem that what applies to the Khoe khoe (“Hottentot”) languages is immediately relevant to the present situation.
But there remain a few problems. First, Greenberg’s evolutional scale (21) has a demonstrative as a source for gender markers, whereas in the Khoe languages there is no evidence to suggest that a demonstrative was involved in the evolution of pgn’s. Rather, the pgn’s closely resemble the personal pronouns of Kxoe. Still, this fact is not necessarily at variance with Greenberg’s thesis. What he has in mind is what in the theory of grammaticalization is called polygrammaticalization: demonstratives can give rise to either definite articles or personal pronouns, or even to both, as has happened e.g. in the development of French: French il (‘he’) and le (‘the’ or ‘him’) are etymologically related in that both can be traced back to one and the same item, namely the Latin demonstrative ille (‘that’). But the pgn’s do not show any etymological relationship to the demonstrative stems found in the Khoe languages (see Kilian-Hatz & Heine 1997). We are therefore led to conclude that the reconstruction proposed by Greenberg, while being supported by a number of well attested evolutions in other language families, does not take care of the situation in the Khoe languages.
But there is an alternative account. The evolution of pgn’s appears to have been determined by a pragmatic strategy commonly employed in the Khoe languages whereby a noun or other element is further specified by adding a marker providing more information on semantic characteristics of the referent in question, especially information on the gender, number, and personal deixis of that referent. This marker has the status of an afterthought or appositive constituent. The following observations made in Kxoe may illustrate the effect of this strategy:
a. While kin terms in Kxoe take the vocative enclitic ’è without any linkage, generic human nouns require the pgn to be placed between noun and vocative marker:
(2)
There is an obvious reason for treating kin terms and generic nouns differently. Kin terms are likely to be inherently associated with appropriate information on the person, gender, and number of the referent concerned; the English term uncle, as in my uncle, for example, stands for a single male, third person referent. Generic nouns, on the other hand, are less specific in this respect and, accordingly, pgn’s are a convenient means of specifying generic nouns for person, gender, and number.
b. The vocative enclitic is also used to form what Köhler (1973:54) calls the emphatic imperative, in which case a second person pgn is placed between verb and vocative enclitic è, e.g.
(3)
c. The vocative suffix can also be suffixed to the demonstrative stem à- and, once again, the pgn precedes the vocative enclitic:
(4)
d. But even without a vocative enclitic, pgn’s are frequently added to nouns for person, gender, and number specification, e.g.
(5)
e. Specification by means of pgn’s goes even one step further: it can have the effect of introducing what may be called redundant predications. The interrogative attribute màá ‘which?’ can occur without specification, as in (6a). But it may also have a pgn after it even if a virtually redundant question results, as in (6b).
(6)
In (27) there is even one more instance of pgn use:
(7)
To summarize, pgn’s can be added to various kinds of word to specify the person, gender, and/or number characteristics of the referent concerned.
