Perhaps if we tried to analyse speech by identifying the factors we do in music; pitch, rhthym, length of words etc. a pattern might emerge?