If you have a probe pressing against a material, its force and position are functions of time. If you’re using a spring attached to an actuator, you can try to jam it in harder or less hard, which will tend to increase or decrease the force, respectively; but the material can then yield. Viscoelastic materials will yield to different degrees at different time scales, and every material is somewhat viscoelastic; this gives us a complex spectrum in which the real part at a given frequency is the elasticity (or resonance) and the complex part is a frictional loss.
If you apply a Heaviside step function to the probe, in theory this contains all frequencies and so you can determine the material’s entire viscoelasticity spectrum from its response to the step function. In fact, though, your step function is going to be bandlimited, and the material’s response at low frequencies may be lost in the noise. By applying a sequence of such step functions, sometimes in opposite directions, at random times, you can get more data points, which will allow you to estimate not only the viscoelastic spectrum of the object but also the frequency and Q of its vibrational modes.
A piezoelectric actuator can straightforwardly produce frequency components up to a megahertz or so, and a strain gauge or piezoelectric force sensor can measure its special mix of force and displacement at similar speeds.
You can use the same approach for dielectric spectroscopy of time-dependent permittivity and magnetic spectroscopy of time-dependent permeability.