A model of phase-coupled delay equations for the dynamics of word usage

Word use presents regular oscillations mounted over slowly varying trends. These oscillations have been recently interpreted in terms of fashion-like cycles of interest and saturation, and modelled using a logistic equation with distributed delay. Here we show that the communities of semantically related words are partially synchronized. To account for this, we model the words of each community using logistic equations connected with a Kuramoto coupling. In this way, we test the simple hypothesis that the change in the occurrence of a word depends linearly on the occurrence of its semantic neighbours. We show that this simple model reproduces the coherence observed in the experimental communities using a single global coupling across multiple languages, regardless of the network topology. Our results build confidence on a universal model of language usage based on the interaction between cognitive forces and the sociocultural context.