Bruno Nettl outlines this theory of the origin of music in Primitive Culture, first published in 1956.
My comments are in square brackets.
My own theory is based on the assumption that an undifferentiated method of communication existed in remote times, one which was neither speech nor music but which possessed the three features that they hold in common: pitch, stress, and duration. Fixed pitch was not present in this early form of communication; durational values were relative and never consistently demarcated; stress patterns were irregular, like prosaic speech today. There was no definite distinction between vowels and consonants. The sounds produced were grunts, cries, wails – things which hardly sound to us like either music or language, but which embodied pitch, stress and duration. This kind of communication exists today although the ontogenetic implications of infant prespeech have been examined by Heinz Werner in the realm of music and by Roman Jakobson in linguistics, the study has not progressed far enough to be related to the origins of music and language.
From this early method of communication, whose exact symbolism we cannot speculate on but whose structure we can postulate, the two specific media, language and media, developed. There must have been a long, gradual stage of differentiation and specialization in culture, during which the two became distinct, language taking on vowels and consonants as its chief trait and music taking on fixed pitch. Again, we cannot even guess much about this intermediate stage, but from it arose the existing types of music – of which there are basically very few, despite enormous variegations in detail.
This theory then, postulates three stages in the development of music: (1) undifferentiated communication, (2) differentiation between language and music, with music still in a highly elementary stage, and (3) differentiation between various musical styles. The last stage is, of course, the only one for which we have any data at all, and even that data is fairly recent.
According to this theory, even the simplest primitive styles – those with litany-type forms [where a short phrase is repeated with little or no variation] and ditonic scales [scales with only two notes] – have a long period of development behind them. Certainly “primitive” music as we know it now is a far cry from man’s earliest musical experiences, just as the simplest extant cultures cannot be assumed to resemble those of prehistoric times. The dynamics of music change in a culture is one of the most interesting phases of ethnomusicology and one about which practically nothing has yet been discovered.
Nettl, Bruno, Music in Primitive Culture, Harvard University Press, Cambridge, Massachusetts, 1969, SBN 674-59000-7, pp.136-137
Theory of Music 