It is widely recognized that grammar can be meaningfully analyzed as a variable system that is probabilistically conditioned. Over decades of research, the variationist paradigm has established that this probabilistic conditioning is evidenced in the covariation between linguistic elements, as seen in the relative favoring or disfavoring of a variable form across different contexts. Here, Catherine Travis proposes that valuable insights about morphosyntactic variation can be gained by expanding the evidence we draw on to infer the variable grammar to include both the variable context, or the set of environments in which there is actually a choice among competing variants, and contextual distributions, or the relative distribution of competing variants across conditioning environments. Considering the use of a set of variables in spontaneous speech corpora from different Spanish varieties, she demonstrates how these dimensions are relevant for cross-varietal or cross-language comparisons and for reliably identifying grammatical change.