The paper offers a language-based hypothesis for the emergence of a Bavarian folk dance called the <i>Zwiefache</i> (‘twofold’). A corpus analysis revealed that rhythm, metrics and song are closely connected in this genre. It is suggested that natural musical systems, as a default, are based on phonological systems (cf. Maxim of Natural Text Setting (Vetterle 2003)). The conclusion, which contrasts with many aspects of previous research on the interface between rhythm, metrics and song, is corroborated by evidence from German rap songs, data from other musical cultures, and literary history.