For a long time I thought of it as quite a peculiar phenomenon: grown-ups who, the moment they spot a baby, start talking in a curious dialect. A dialect that has unclear semantics, little or no grammar, and is full of exaggerated rhythmic and melodic diversions. Nevertheless, babies love it. They react, cooing with pleasure, to melodies that are not unlike pop songs as ‘De do do do, de da da da’ of The Police or ‘La la la’ by Kylie Minoque. This babbling, or, more formally, infant-direct speech (IDS), differs from normal adult speech by its high pitch, exaggerated melodic contours, a slower tempo, and more rhythmic variation. A kind of ‘musilanguage’ indeed.
IDS is a widespread phenomenon that is —as far as we know— present in all cultures and has more similarities than differences, even when some characteristics of IDS conflict with the rules of the adult language, like Chinese. Hence, it is unlikely that IDS is ‘just’ a preparation for language -- until recently the most common interpretation.
Laurel Trainor, and her team at McMaster University (Ontario, Canada) suggests that IDS is essentially a tool to communicate emotion. The decoding of the speech patterns into their emotional meaning is something infants can do easily, and long before they learn about language. In that sense, it seems more likely that language makes use of faculties special to music then that it emerged as a side effect of language (as as suggested once by a well-known cognitive psychologist).
Laurel J. Trainor, Caren M. Austin, Renee N. Desjardins (2000). Is Infant-Directed Speech Prosody a Result of the Vocal Expression of Emotion? Psychological Science, 11 (3), 188-195 DOI: 10.1111/1467-9280.00240