Condon and Sanders published seminal work in 1974 showing that the whole body movements of an infant are synchronised to the rhythms of adult speech. Infant movements start, stop, or change synchronously with the articulatory segments of the adult's speech. These data showed probable interactional participation of a unique kind in language acquisition. However, their methodology was questioned and their results were refuted by others, further experiments failing to reliably replicate them (e.g. Drowd & Tronick, 1986). The controversy sits undecided. Recently, I acquired high-precision infant movement data together with digital audio records of both parent and baby. In these data, the synchrony phenomenon is again very apparent. I would like to show these data and question how best to proceed with the analysis. Of particular importance is the notion of 'articulatory segments', or stresses in speech, because it is a robust analysis of these that will enable a precise measurement of the adult-infant synchrony, if indeed it exists. The notion of synchrony is attractive, because if it active, then it goes some way to showing a direct perception-action coupling between the infant and the other, it touches on the importance of the mirror neuron system in early language acquisition, and it exemplifies the importance of movement as a foundation for language.