These quotes are from Simha Arom’s African Polyphony and Polyrhythm
…. for each type of procedure observed, there must be an explanation of how it is organized, both vertically (in terms of the ‘con-sonance’ and inter-weaving of the parts) and horizontally (the temporal organization and syntax). In other words, its internal coherence must be displayed……
the description must include a detailed analysis of each piece, i.e. show how its parts are divided into segments (on the basis of the principles of recurrence and substitution), what variations are possible, and where the points of substitution are located….
finally, for each part, it must determine the ultimate reference used by the musician as a model; this is a condensed (in fact ‘minimal’) formula with respect to which all variations are produced and which summarizes all of the part’s characteristic features, and only these. Modelisation thus allows us to apprehend the relations prevailing between the spontaneous production of a musical event and the idea it springs from. The synthetic model of the polyphonic or polyrhythmic structure of any piece is, of course, obtained by placing the minimal formulae for each of its constituent parts one above the other. p. xxi Arom
[This looks to me like how you describe a language and how its speaker accesses it]
On the contrary, in societies where there is practically no conceptualization or abstract speculation about the substance of the music, the (in every sense of the word) unheard-of experience of listening to the parts of a polyphonic construction, first separately, then in combinations of increasing numbers of parts (in successive layers, as it were) provides a strong stimulus to the verbalization of the musicians’ reactions, i.e. to the emergence of the (often implicit) vernacular musical terminology, and thereby makes those who know the tradition aware of the individuality and the complexity of their musical heritage. I may add that the method involves active experimentation, insofar as the musicians themselves decide how it should be applied at all stages prior to transcription. The set of operations leading from performance to modelization of a given piece, like the steps which allow the model to be checked by deriving new versions of the piece from it, are not speculative in nature; rather, they take the material form of recordings made by the traditional musicians themselves. p. xxiv Arom
[this makes me think of a naïve native-speaker having the underlying grammar of his language revealed to him and then using that grammar to construct sentences which he can then verify as well-formed or not.]