Bilingual speakers of Spanish and Juchitán Zapotec (JCH), two languages that have been said to differ substantially in their semantics for expressing information about space, offer a fresh perspective on the classic problem of linguistic relativity because they allow us to test the extent to which cognitive styles may be related to linguistic codes or to other socio-cultural variables. This dissertation aims to combine the theoretical and methodological approaches of linguistic anthropology with some more recent quantitative innovations from other disciplines in order to probe further the question of why cross-cultural variation in styles of cognition exists, and how people come to adopt their particular cognitive styles. In this dissertation, I argue that the typology of the language a speaker knows or is using at a given moment does not reliably predict patterns in spatial thinking. Instead, I argue that different conceptualizations of space emerge in interaction on the basis of different conceptualizations of multiple layers of context. I present evidence from a battery of semi-experimental tasks that demonstrates a mixed profile of spatial reasoning strategies in Juchitán. The distribution of these strategies did not correlate with language dominance or language used on the task. However, variability in frame of reference use by children participating in the elicitation tasks indexes cultural differences among the children, related not obviously to spatial conceptualization but to different narrative and gestural styles. I propose that the particular ways in which language is used and thought of in modern settings, especially the school, may driving conceptual change in Juchitán. This manifests both as language shift from JCH to Spanish, but also as conceptual shift independent of code. Variation in the use of “spatial frames of reference,” then, is potentially indicative of different ways of conceptualizing language in relation to context, and self in relation to world.