It is well known that a recipient perceives verse and prose in different ways: we extract from prose first and foremost its “objective content”, while verse conveys to the recipient a “(lyrical) mood.” This mood is of great importance in verse: it can even obscure the “objective content”. The general problem dealt with here is the modelling verse perception. While the issue clearly has a psychological aspect, this paper focuses only on linguistic issues, and our hypothesis is as follows. The “mood” of verse is created by certain lexical semantic components which occupy “semantically strong” positions in a verse line. These positions are distributed in a verse line according to certain rhythmic rules. Such semantic components usually repeat in a poem; the repetition of these components in accordance with meter and rhythm affects the mind as it perceives verse. The repeating semantic components can be identified by means of semantic analysis of a lexeme. The paper proposes a two-level structure for the semantic representation of a poem. One level represents the “objective content” of a poem, while the other represents the poem’s “mood” as sets of repeating semantic components. In fact, these sets denote the “objective content” of a poem in a very simplified form, isolated from concrete details and therefore applicable to any emotionally similar situation. The approach is illustrated through an analysis of Pushkin’s poem “Anchar”.