Linguists usually study the consequences of the sixteenth-century invasion of Mexico and the Caribbean by Castile through the constructs of the language (Nahuatl, Spanish, Yoruba, etc.) and the dialectitalic (Old Castilian, Andalusian, New World koine, pluridialectalism, etc.) and in terms of the contactitalic between these constructs. In contrast, contact is studied here at the level of individual speakers whose inventories of lexical and structural features change and evolve, as new features from other speakers are differentially acquired. These disaggregated processes crucially involve inter-speaker and intra-speaker variation dependent in part on the different frequencies of lexical exemplars. We stress the socially invented nature of named communal languages and argue that our focus on variable contact between idiolects and the disaggregated conception of lexical exemplars can overcome theoretical limitations that are unavoidable when contact is seen in terms of languages and dialects. The data come primarily from the well-documented history of Castilian /s/ in so-called loan phonology in sixteenth-century New Spain, supplemented by a secondary look at nineteenth-century Cuba.