The article discusses a problem of analysis of a polycode (multimodal) text as a hybrid phenomenon. Complex semiotic systems, which imply simultaneous use of codes on the levels of graphics, sound and traditional text format, present a problem for analysis on the basis of standard, traditional methods typical for linguistics, text analysis and literary criticism. The article is focused on the genre of the cartoon. We made an attempt to trace and analyse the ways of transformation, adherent to the cartoon as a genre, and to define this phenomenon in terms of genre theory. The theoretical grounds were formed on the basis of foundational research and findings in the theory and typology of speech genres contributed by M. Bakhtin, V. Dementyev, V. Karasik, G. Molchanova, E. Sheigal as well as the research of multimodal or polycode text. All this served as the ground to formulate typologically universal and specific criteria of such a text. As the basic criteria, which can serve to define cartoon as a genre, we dwell upon the criteria of primary and secondary genres, theme or topic, style (stylistically marked and stylistically neutral), plot structure (including conflict and position of objects in space), aim and conditions of communication. All these factors depend on specific means of cartoon execution and technique, linguoculture, predominance of verbal or iconic components, event, public order, and data channel. We also discuss point of view and focus, tone of narration, type of explication, cultural and/or ideological coding, pragmatic and axiological criteria. A separate group of features typical of this genre is constituted by its folding and compression, ability to become viral. As some additional criteria we may rely on genre composition, conceptual density, presence or absence in linguoculture, historical dynamics, being native or borrowed, inner and outer attitude to this genre, realisation of it in the world outlook, key phrases and other codes, ways of manifestation.