Traditionally, since the days of Arthur Lloyd James and Kenneth Lee Pike, languages have been divided into two broad types: syllable-timed and stress-timed. French was considered the archetypal syllable-timed language (Lloyd James called this ‘machine gun rhythm’), in which each syllable had a similar duration, and English, probably the language whose rhythm has been studied most intensively, and mostly by native English speakers, the archetypal stress-timed language, in which stresses occur at approximately equal intervals of time. Doubt has been cast on this classification, because the measurements taken by phoneticians using ever more sophisticated machines have shown that neither syllables nor stresses are truly isochronous.
Phoneticians need to consider the case of music, which is, like language, a form of “organised sound”, and which also consists of variation in pitch, timing and intensity. Drum machines create mathematically exact rhythms, which humans cannot do. Human performers, on the other hand, produce interpretations of musical works which are not mathematically precise, but are still, nevertheless, rhythmical. Bach is considered to be one of the most scientific of composers, but electronic renderings of his music, while rhythmically precise, are lifeless. Performances by a human interpreter, however, can move the listener deeply. Composers using ‘Sibelius’, or one of the other suites of music software, find that they have difficulty using a (piano) keyboard to play in the rhythm that they want to be recorded, because the software recognises the minute differences of duration between the notes, and transcribes what it “hears”, leading to tiny fractions of beats being notated when they are not intended.
It is perhaps no accident that many phoneticians through the years have also been proficient musicians. As phoneticians, we need to learn more about the way in which we hear musical rhythm, and apply that knowledge to how we hear language. It has often been reported that interlocutors take up each other’s rhythms in a conversation. How can they do this if there is no rhythm to take up? Measuring durations of sounds in milliseconds will not work: rubato, accelerandi and rallentandi cannot be accounted for in such terms. Instead, a more impressionistic approach is needed, that will allow for the nuances of expression that are conveyed by rhythmical variation.
It may well be that there are more than the two types of linguistic rhythm, or that there is a gradient from extreme syllable timing to extreme stress timing, but I believe that it is our ears, not our machines, that will decide this in the long run.