Computational studies of the thermodynamic properties of materials at the mesoscopic and macroscopic scales – involving lengths and times of at least $\mu$m and $\mu$s, respectively – rely on a coarse-graining approximation such that only a few relevant collective variables are treated explicitly. Those variables typically take the form of fields defined everywhere in space (e.g., a polarization field in dielectric or ferroelectric media) or macroscopic quantities when spatial inhomogeneities can be treated implicitly (e.g., the volume average of the electric polarization). The free energy is usually expressed as a Landau-type potential whose temperature-dependent minima track stable states, characteristic equilibrium fluctuations being implicitly accounted for. Further, the response of the system to external perturbations, and its relaxation toward thermal equilibrium, are described in terms of simple equations of motion governed by effective inertial and viscous-damping constants. There is considerable literature on the problem of deriving Landau-type free energy potentials, from either experiment or predictive atomistic simulations, including recent efforts to develop systematic machine-learning approaches that we denote "third principles". Much less attention has received the calculation of the effective constants defining the time-dependent Landau equations and the corresponding nonequilibrium macroscopic or mesoscopic dynamics. Here we tackle that problem, describing a protocol that allows us to compute the temperature-dependent inertial and damping coefficients associated to the electric polarization in representative soft-mode ferroelectric PbTiO${}_3$. Our scheme follows usual approaches to analyze experimental vibrational spectra. It relies on standard equilibrium molecular-dynamics simulations and is therefore readily applicable with modern machine-learned interatomic potentials. Our results also allow us to comment on common assumptions in the literature of effective dynamic treatments and phase-field simulations of ferroelectrics and related materials.