Observational data on the internal structure of superassociations are analysed. Basing our assumptions on several well-studied examples we try to clarify the characteristic features of these large-scale regions of ongoing violent star formation. One of the most impressive examples of this kind is provided by the superassociation NGC 206=OB 78 in M31. It reveals a binary internal structure: there are two definitely separated components of more or less equal (or at least comparable) sizes, with hundreds of OB stars in each. The age of these stars is less than 10 Myr. In one of the components, HII regions are observed in its external edge. In the other component, Cepheid stars with ages of about 40-50 Myr are seen. Between the components, a dust strip stretches to a length comparable with the sizes of the components. The space-time structure of the superassociation suggests that there were at least two events or bursts of star formation in the evolutionary history of this system: one about 50 Myr ago and the other less than 10 Myr ago. It is the latter star formation event that caused the system to become a superassociation. The very intensive formation of massive stars proceeds over the whole area of the two-component system almost simultaneously. Binary (or triple, or composite) structures are always observed in superassociations and also, generally, those violent star formation regions that can only be studied with a high angular resolution (for instance, with the Hubble Space Telescope). The binary space-time structure may provide a new insight into the origin and the physical nature of the superassociations. The approach we develop here is based on the observational data of the internal structure of the superassociations, and supports violent gas dynamics processes in the interstellar medium of spiral and irregular galaxies as being the physics that might be responsible for the origin of these systems. A key physical mechanism of violent star formation in the superassociation is assumed to be related to the formation and collisions of large-scale shock fronts involving gas masses of about one million solar masses or more. The non-linear dynamic structures produced by violent shock-shock collisions are at the focus of our discussion here. To study them, we use a set of computer models developed at Ioffe Physico-Technical Institute. The main question with regard to the nature of the superassociations concerns the physical cause of the very effective formation of massive stars on a spatial scale of 1kpc simultaneously over the whole area of the system. The gas dynamic processes we address here seem to provide a starting point in the search for an answer to this question. This approach suggests new observational studies of the superassociations, and we give some theoretical predictions which can - in principle -be tested by observations.