This paper presents a contrastive analysis of adpositions in English, Nepali and Russian corpora. Two sets of highly frequent adpositions, those with broadly locative and broadly ablative meaning, are contrasted. The ‘quantitative-distributional’ analysis is based on identifying patterns across the most statistically significant collocations of the words in question; it is undertaken using 1 million word comparable multi-genre corpora of each language. The results suggest that, while in all three languages the adpositions are characterised by two collocational patterns (one of subcategorisation and one of semantic congruence), the former pattern is substantially more prominent in English than either Russian or Nepali.