There is the saying: never say never. Certainly this recent tragedy is once again a testament to this philosophy and a stark reminder to all engineers and scientists.
Given that disclaimer, these civilian nuclear reactors are designed with contingency in place to ensure they (for the most parts) do not go critical when they aggregate into a lumped mass corium at the bottom of a reactor containment and beyond. This is because firstly the nuclear fuel is of low purity in terms of fissile elements. Secondly, neutrons generated directly from fission of say uranium have too high energy - this results in a low probability of causing other fission events of other uranium atoms thereby eventually freezing out a chain reaction.
These reactors are purposely designed to have neutron moderators placed between fuel rods to slow down and reduce the energy of these neutrons from fission, so that they can cause fission and sustain a chain reaction. In other words, when fuel rods are melted down into a corium, there is presumably no more neutron moderators between the fuel rods and fuel elements (that is, the water), thereby from that point on you are primarily concerned with high energy neutrons generated from fission which have low probability of sustaining a chain reaction while thermal neutrons (lower energy) that can sustain a chain reaction are reduced in a hypothetical corium configuration.
This is the best case scenario of course. If it so happens that the civil structure of the plant itself unknowingly had neutron reflecting material nearby, it may be able to sustain a reaction to a degree. Theres also the possibility that the corium achieves localized critical mass configurations within the corium that can sustain a reaction (though not likely as the fuel is purposly designed to have low purity and is assumed to be homogenous in composition). All of this is purely hypothetical of course as are any other plausible scenarios people can draw up.
That is also not the end of the story of course, as life is never that perfect: it is not just 100% fuel configuration with control rods in place or 100% corium lump mass. If it is partially melted with a corium and some spent fuel assembies, presumably with neutron moderating water inbetween, this can again lead to criticality, especially since control rods that would have otherwise prevented criticality may be ineffective to block neutrons between the corium and the fuel assemblies though they will be far away in distance and the reaction rate should not be very high.