This paper demonstrates the mutual relations between the various “voices” in Czech folk songs and the way in which these voices can be identified. The introductory section characterizes the specific features of the songs: the concision and condensation of their structure, as well as their emotionality of expression. The second section is devoted to the analysis of a selected corpus of folk songs from the perspective of heteroglossia. Their texture appears as a net of relations and functions which operate on different levels, stand in mutual complementarity, partly overlapping, and show differing relevance. Using methods of text linguistics, the author concentrates on two relations and functions – speech acts and compositional functions – as well as personal attitudes. The analysis focuses on utterances with the communicative function of appeal, i.e. on the interrogative and addressing turns which are quite frequent in Czech folk songs, on their interactive structure and modes of direct and indirect speech. The author shows that the song genre has two faces: it is a compact combination of verbal text and its musical part (the tune) which may appear to be uneven in their mutual relations. The last section deals with the phenomenon of intertextuality in the song structure.