Indian grammarians knew of linguistic variation: in Pāṇini's Aṣṭādhyāyī we already find rules to describe Vedic usages, marginal and preferred usages, even regionally restricted usages. Still, this awareness conflicts with a likewise deeply rooted belief in the intrinsic inalterability of language that Dehspande calls the ‘theology of Eternal Sanskrit’. This leads, as far as Pāṇini is concerned, to a description of Sanskrit language as a panchronistic flatland, i.e. as a totality which includes “all known diachronic and synchronic facts of Sanskrit” (Deshpande 1985: 124).

Pāṇinian grammar is based on a common set of rules which represents all the shared facts of Sanskrit, including what we would consider Vedic facts, as long as they are not exclusively Vedic. Of course Pāṇini and his commentators distinguish restricted domains inside this totality; these are generally accounted for in rules limited by what later commentators call ‘a locative of domain’ (viṣayasaptamī), i.e. rules which apply exclusively in the domain of mantras (mantre), in the domain of the Vedic hymns (chandasi), in the domain of ordinary language (bhāṣāyām) and so on. Nevertheless these domains are not identified—through historical, regional or social criteria—as different linguistic systems, but simply as sub-sets of rules deviating from the set of common rules. Already Renou pointed out that “the chandas […] do not represent in any way a state of language as distinct from the normal; it is, in the interior of a common language, a fringe of archaism or ‘diversities’ which, for some reason or other mark the margin of the system” (Renou 1941: 248-249).