A theory of citations should not consider cited and/or citing agents as its sole subject of study. One is able to study also the dynamics in the networks of communications. While communicating agents (e.g., authors, laboratories, journals) can be made comparable in terms of their publication and citation counts, one would expect the communication networks not to be homogeneous. The latent structures of the network indicate different codifications that span a space of possible 'translations'. The various subdynamics can be hypothesized from an evolutionary perspective. Using the network of aggregated journal-journal citations in Science & Technology Studies as an empirical case, the operation of such subdynamics can be demonstrated. Policy implications and the consequences for a theory-driven type of scientometrics will be elaborated.