In this essay I'd like to follow up one of the ideas outlined in my reply to Peter Goldsworthy's stimulating article in PNR 114. What is rhythm for? Why are human beings apparently endowed with a potential for apprehending, and re-creating, highly organised rhythmic structures? What, in evolutionary and aesthetic terms, are these structures doing within human cognitive processes? One might agree with Stevens that 'It must be abstract... It must change... It must give pleasure...' Yes - but why?

Children know about, and enjoy, rhythm. Humans seem to be innately, and possible uniquely, endowed among the higher primates with the capacity to inhabit, and to recreate, three intersecting forms of aural structure. These forms are found in (i) sequence, (ii) contiguity, and (iii) hierarchy.

In a sequence, something recurs. Rhythmically, the something that recurs is a point of salience (in linguistic terms, of markedness). Languages differ in how they register salience. In some languages - Classical Greek seems to have been one - vowel length functions as salient, so that a long vowel or diphthong is perceived to be more prominent than a short vowel. In other languages, stressed syllables (not just vowels within syllables) function as salient, while unstressed syllables are non-salient. Modern English is an example. Here, contrasts of salience are optimized: stressed syllables occur predictably, and their vowels are uttered with full quality; unstressed syllables can be recognised as non-salient by their 'reduced' quality, their typical production with the 'murmur vowel', ...