Embedding and modular embedding are two well-known techniques for measuring and comparing the expressiveness of languages—sequential and concurrent programming languages, respectively. The emergence of new classes of computational systems featuring stochastic behaviours – such as pervasive, adaptive, self-organising systems – requires new tools for probabilistic languages. In this paper, we recall and refine the notion of probabilistic modular embedding (PME) as an extension to modular embedding meant to capture the expressiveness of stochastic systems, and show its application to different coordination languages providing probabilistic mechanisms for stochastic systems.