In cities in the modern world we can observe the process of reallocation in authentic linguistic data collected from representative samples of speakers and speech situations. In early modern Paris, since the quantity of empirical data giving access to the vernacular is minimal, we are left with the indirect, fragmentary and uncontrollable evidence provided by literary texts and metalinguistic works of various kinds. The effects of refraction inherent in contemporary representations of past sociolinguistic states are severe, so what can material like this tell us? Firstly, it provides information about the existence of sociolinguistic variants at that time, which otherwise would have been irretrievably lost to us. The range of forms which grammarians and literary authors represent is selective, what they say about their frequency and distribution is not entirely trustworthy, but the forms they attest almost invariably prove to be authentic. Secondly, material like this provides useful insights into the social value of the variants in question. When it was believed that what was relevant to change was the linguistic function of a structural unit, social information of this kind was irrelevant. It is now clear, however, that reallocation of the community's linguistic resources takes place less on the basis of linguistic function than on that of the symbolic value conventionally attached to particular variants. In this chapter we will bring together the information provided by metalinguistic works and literary authors on those morphological and phonological variables that were sociolinguistically sensitive during the proto-industrial period.