The article focuses on how we can use contrastive analysis for translation purposes, i.e. search for and identify cross-linguistic correspondences, variants in translation. To find cross-linguistic variants, we used parallel texts matched in subject domain and text type. The cross-linguistic correspondences are viewed as functionally complete units that are natural and idiomatic for each language. These units were analysed by the theoretical language-independent model of the lexico-semantic set, both in the paradigmatic and syntagmatic dimensions. Cross-linguistically, they were compared across the semantic, language-independent categories. The results of the contrastive analysis were used as possible translation choices to achieve functional equivalence. To mark the difference between the polar types of translation equivalence, we used machine-based translation and functional translation. Contrastive analysis provides a wide range of idiomatic and natural language choices for translation.