Following van Heuven and Sluijter (1996), this paper suggests that the tradition of equating prosody with variations in fundamental frequency (F0), duration and intensity may have originated from a tendency to view prosody as the music of speech. This paper discusses several problems with this musical view and suggests that a functional perspective may be more appropriate. The four functions include signalling (1) lexical contrast (as in e.g., lexical tone and quantity languages), (2) postural settings such as overall rate of speech, F0 height and span, (3) non-lexical meaning (e.g., surprise), and (4) prosodic structure. As motivated elsewhere, prosodic phonology accomplishes the third and fourth functions, where prosodic structure includes prosodic constituent and prominence hierarchies. Intonational events such as pitch accents and boundary tones align with respect to these structures. Correlates of prosodic structure include, but are by no means limited to, F0, duration, and intensity. It is hypothesized that any physical parameter can be used as a correlate of prosodic structure as long as utterance recognition is not compromised. Although prosodic phonology plays a key role in relating language to speech, relationships between prosodic phonology and grammar, and between prosodic phonology and motor control, are not well understood.