This study investigates how language‐internal constraints regulate the future temporal reference (FTR) alternation across nine varieties of English around the world. We specifically marshal Variation‐Based Distance & Similarity Modeling (VADIS) to calculate distances between the varieties under study as a function of the non‐correspondence of the ways in which language users choose between FTR variants. Our linguistic data come from the spoken component of the International Corpus of English: the dataset covers 500 future marker observations per variety of English (Ntotal = 4500), richly annotated for seven language‐internal constraints (such as polarity and sentence type). VADIS uncovers a number of probabilistic grammar differences between the varieties of English under study, which are subsequently correlated with four language‐external distance measures: geographical distance, travel time distance, population size and GDP per capita. Mantel correlation analysis shows that language‐internal distances do not correlate with language‐external distances.