True in principle, but IMO unlikely to make a difference. We are, in effect, looking at the moment whan all the KE is absorbed by the two girders, which will be a local maximum of force and deflection. If that local maximum hasn't broken anything yet (with the connection being the obvious candidate to fail first), the girders will both recoil/unload elastically - and that goes to their other ends, too. Elastic waves will travel into and through the two columns and then mostly away. That's a real energy loss to our local system. If oszillations build up beyond the original amplitude, they better do so on the first cycle, or one of the first few. It seems very unlikely that this would actually happen from a singel fall-and-impact event.