The present article is intended to sum up the past ten years of our research concerning the timing of speech and the development of a coupled oscillator model (COM) for speech timing. This enterprise started with our paper (O'Dell and Nieminen 1998) presented at the 1998 Finnic Phonetics symposium in Pärnu-the first of the series to be held in Estonia-and has since continued through several conferences and conference papers.Our initial focus Nieminen 1998, 1999) was on the so-called rhythm dichotomy (Pike 1945) and particularly Eriksson's (1991) re-interpretation of it. Eriksson had noted that in both stress-timed and syllable-timed languages, a simple linear relation holds between the duration of the stress group and the number of the syllables in it, the only difference being in the constant term of the linear regression equation (y-intercept). This intriguing observation provides simple mathematical tools with which to describe the classic rhythm dichotomy and we proposed the COM as a possible explanation of the observed mathematical relation.Later, we have expanded the scope of the model and applied it to additional data and other types of timing phenomenon (O'Dell and Nieminen 2001, 2002a,b, 2006. Most recently we have been developing statistical tools to account for rhythmic variation in spontaneous speech (O'Dell et al. 2007(O'Dell et al. , 2008.In the following we first present, in Section 2, a brief introduction to the terminology and the basic concepts of the COM. We also discuss its relationship both to dynamical systems theory, this being the general (i.e. not speech-specific) point of view, and to the phonetic study of speech in particular. In later sections, applications of the model are presented mainly from our own studies but partly taken from studies conducted by other researchers in the field: Section 3 describes the interaction of speech units at different levels, Section 4 deals with intralevel effects, and Section 5 provides an example of applying the model to a specific case.