The structure of grammaticalization chains Conclusion
The structure of pgn use in the Khoe languages can tell us a story of how gender agreement may arise and spread. Hiechware represents the first part of the story, where no gender agreement exists. Kxoe offers the second part, where certain patterns of optional gender agreement arise, in the case of the “nasal concord” also of obligatory agreement. Naro has reached an even more advanced part, but it is still far away from the final part, where gender agreement is complete, as e.g. in the Bantu languages surrounding the Khoe languages.
One may wonder what purpose there may be to identify cross-linguistic chains. It would seem that there are at least two goals, a diachronic-comparative one and a synchronic-descriptive one. According to the first goal, cross-linguistic chains are an immediate reflex of diachronic processes, and they provide a basis for reconstructing certain historical events. The reconstruction of such events is beyond the scope of alternative methods of historical analysis. The fact that an etymologically identical item occurs as a free form in one language but as an affix in another, or as a personal pronoun in one language and as an agreement marker in another can be accounted for in a systematic way with reference to the grammaticalization chain in question.
According to the second goal, cross-linguistic chains can help us to understand the structure of a given language with reference to more general processes of linguistic evolution. For example, the question of why there are pgn markers on demonstratives or nouns in Kxoe in some contexts but not in others, can be answered with reference to the overall evolution of personal pronouns as appositive specifiers: The situation found in Kxoe is the result of processes that have not (or not yet) taken place in Hiechware, but have already taken place in Naro. For a better understanding of the nature and functions of gender agreement in Kxoe it is essential to know more about the mechanism that can be held responsible for the present situation in Kxoe, and for a reconstruction of that mechanism, a systematic comparison between Hiechware and Kxoe seems indispensible. At the same time, by comparing the Kxoe situation with that of Naro it is possible to predict within limits what is going to happen to gender agreement in the future development of the Kxoe languages.
A characteristic of grammaticalization chains can be seen in their overlap ping structure, which can be represented in a simplified form as in (1).
(1)
The motivation for the presence of an overlapping structure can be seen in a structural property of grammaticalization according to which some components of a linguistic expression are grammaticalized earlier than others. Thus, all chains identified so far suggest that there is a principle according to which change in meaning precedes change in syntax and phonology (see Heine, Claudi and Hünnemeyer1991). For example, A may be the meaning of that exression which is found at Stages I and II but not at Stage III, while the syntactic form B associated with A does not occur at Stage I but rather is a characteristic of Stages II and III. While the data available for the Khoe languages is too scanty to allow for reliable reconstructions, it would seem to be in accordance with this principle: Thepgn’soccurassemanticallyfullfir stand second person pronouns in Hiechware and Kxoe, but as semantically empty agreement markers e.g. on demonstrative attributes in Kxoe.
But our Khoe chain illustrates yet another principle, according to which a grammaticalization processs is first restricted to specific contexts before it spreads to other contexts. For example, pgn use has become a common feature on demonstrative attributes in Kxoe and Naro but has not spread to adjectives as yet. That adjectives are more conservative than demonstratives in the evolution of gender agreement is reflected .g. in the cross-linguistic typology of gender systems.Amongthe500–plus African gender languages there are quite a few that mark gender agreement on demonstratives but not on adjectives, whereas gender agreement on adjectives but not on demonstratives appears to be extremely rare.
