Math’n Pop versus Math’n Folk? a Computational (ethno)musicological Approach

According to a programmatic article by Philip Tagg on theoretical, methodological and practical aspects of popular music studies (Tagg (1982)), folk or traditional music is one of the three possible kinds of music, together with classical (or art music) and popular music. This typology constitutes what Tagg calls an axiomatic triangle of musical genres, each of which being characterized by criteria such as the usual or unusual mass distribution, the existence of a circle of professionals or a circle of amateurs who produces and transmits it, the principle modality of storage and distribution (ranging from oral transmission, in the case of folk music, to the recorded sound, in the case of popular music), the anonymous versus authorial character of the underlying compositional process, and so on. This typology, as well as any other traditional taxonomy, has been recently criticized by several scholars (see, e.g., Fabbri (2007)) stressing the existence of other music genres which are not included in Tagg’s axiomatics (e.g. jazz music or the so-called musiques actuelles) and suggesting the necessity of substituting this typology with a finer taxonomy based on computational models focusing on musical objects and making use of different theoretical approaches (or methods) in order to carry on computeraided music analysis. The real musical objects would not be musical pieces belonging to pre-established and universal taxonomies, but, as suggested by Franco Fabbri, “musical facts” and “musical events”, each being characterized by a set of properties turning out to be associated into highorder families called types (See Fabbri (2007, 2014)). Starting from this change of perspective in the study of musical genres, we discuss in this paper some possibilities of overpassing the main linguistic-based approach which seems to motivate Fabbri’s original formulation of types classification largely inspired by George Lakoff’s and Mark Johnson’s theory of metaphors (see Tendahl & Gibbs (2008)) to more mathematically-based formalizations of musical structures and processes. After recalling the principle construction of a computational linguistic-based approach aiming at symbolically notating the audio files of large data based of popular and folk music (Bimbot et al. (2012)), we show how a mathematical formalization of the harmonic system provides some additional structural criteria in order to apply the System & Contrast Model to symbolic music information retrieval and automatic genre classification. A new approach to genre classification has been discussed in Andreatta (2014).