Grammaticalization has been defined, again and again, as a diachronic or even a historical process (s. Lehmann 2004, § 2.1 for discussion). Given the distinction between diachrony and history, the latter kind of definition makes no sense at all; only a particular grammaticalization process in the history of a particular language can be a historical process (see relevant conceptual distinction). As for grammaticalization as a process of diachronic change, it is to be noted first that every diachronic change has its counterpart in a synchronic variation. Consider as a simple example.

The reduced variant of the auxiliary in #b developed out of the full variant in #a. However, it is not the case that the reduced variant instantly superseded the full variant. This would mean that the entire speech community overnight stopped using the full form and started using the reduced variant in its place. Instead, the reduced form started its life as a synchronic variant of the full form. It became gradually the preferred variant in the colloquial variety and has all but replaced it in informal speech. The sole survival of the reduced variant is a possibility which has not yet been realized, but which has occurred in other cases like the development of the Portuguese perfect auxiliary hei.

These observations may be generalized: Diachronic change and synchronic variation are two sides of the same coin. Every change, be it a phonological change like assimilation, a morphological change like analogical adaptation, a lexical change like lexicalization or a grammatical change like grammaticalization, manifests itself both on the synchronic and on the diachronic axis. The question of whether variation precedes the change or vice versa is a hen-and-egg problem.

As grammaticalization is a process, it is dynamic. However, language activity as a whole is dynamic, not static. It is not therefore diachronic rather than synchronic. Instead, it may be viewed synchronically and diachronically. The same goes for grammaticalization, an integral component of linguistic activity.

There is a methodological side to this theoretical position. Phenomena instantiating grammaticalization are documented in the history of many languages. If one follows their chronology, one takes a diachronic perspective on them. The historical sequence determines the direction of grammaticalization: the earlier form is less grammaticalized, the later form is more grammaticalized. Such historical data play the methodological role of evidence for grammaticalization and its direction.

On the other hand, two forms of a synchronic state of a language may be related in the sense that one is a grammaticalized variant of the other, as in . Then no direction of variation is visible. In a sufficient number of cases, we know the historical background of such a synchronic variation; i.e. we can see which variant developed out of which one. is such a case. From such cases, we know that grammaticalization does have a synchronic side, that it does manifest itself in the synchrony. Whenever we have no historical records documenting the sequence – which is our methodological situation in many languages with no documented history –, only the theory of grammaticalization can tell us which of the two variants is the original one and which is the grammaticalized one. Consequently such cases are no evidence for grammaticalization.

In cases of this latter kind, the theory of grammaticalization determines the fact and the direction of grammaticalization on analogy with numerous historically documented cases. Thus, although such cases of synchronic variation do not have the status of evidence, their theoretical backup is sufficient to plausibilize their analysis. Many analyses of grammaticalization presented in the linguistic literature are, in fact, based on synchronic data. Analyzing such synchronic phenomena as cases of grammaticalization is instrumental in understanding them (Givón 1979).

Reference

Givón, Talmy 1979, On understanding grammar. New York etc.: Academic Press (Perspectives in Neurolinguistics and Psycholinguistics).