This study examined the influence of sex, social dominance, and context on motion-tracked head movements during dyadic conversations. Windowed cross-correlation analyses found high peak correlation between conversants' head movements over short ([approximately equal to]2-s) intervals and a high degree of nonstationarity. Nonstationarity in head movements was found to be positively related to the number of men in a conversation. Surrogate data analysis offsetting the conversants' time series by a large lag was unable to reject the null hypothesis that the observed high peak correlations were unrelated to short-term coordination between conversants. One way that high peak correlations could be observed when 2 time series are offset by a large time lag is for each time series to exhibit self-similarity over a range of scales. Multifractal analysis found small-scale fluctuations to be persistent, [tau](q) less than 0.5, and large-scale fluctuations to be antipersistent, [tau](q) greater than 0.5. These results are consistent with a view that symmetry is formed between conversants over short intervals and that this symmetry is broken at longer, irregular intervals. (Contains 8 tables, 1 footnote, and 9 figures.)