Song variability of passerine birds is a result of the interaction of space and time. Analysis of the syntax and the structure of the song of different Palearctic passerines revealed variations at some hierarchical levels. This is the consequence of learned and innate song components. The high capability of learning finds expression on the one hand in mixed-singers (for example genera Sylvia and Phylloscopus). On the other hand we know genera (for example genus Emberiza s. str.), in which species differ distinctly in (innate) size and plumage and have similar (predominantly learned) songs. Cultural transmission of the vocalizations is in this case more conservative than the inheritable plumage. We found some species (Fringilla coelebs, Carpodacus erythrinus) which demonstrate the possibility of learning different song types not only during the so called 'critical period' but also over some years. On the basis of lifelong learning, and sometimes reinforced by geographical isolation, an intensive element and song-type sharing between individuals and populations exist. In this way local dialects ('regiolects') arise. Dependent on the actual behaviour, spontaneous changes of song patterns appear, including progressive and regressive genesis (for example Fringilla coelebs and Carpodacus erythrinus).