The childish form of language may undergo a similar structural change, though, in this case, some items may be biased to prefer a functional state. The structural change would result from a first order phase transition instead of a second order phase transition such as percolation.
Children learn grammar from adults. They register sentences that involve functionalized vocabulary. It is not clear how these vocables are processed but it is common knowledge that when a child is asked if she wants an apple or a banana she will say yes, rather than specify which. It often requires a certain experience with language before functionalized vocabulary can be used expertly. It may be that the cortical territory must reach a point of overload to exploit low connectivity structures, characteristic of functionalized vocabulary. It may be that sentences involving functionalized vocabulary must lose the temporal aspect of immediate sensory activation before the distributed nature of the functionalization process will be fully exploited.
We can imagine that the synaptic patterns of thousands of sentences intersect on only one point, say, the soundor. This sound could be the vocable or or it could be the or in ordinary, other, orbital etc....
Suppose that all structures that hold the or sound generate stable or attractors that are widely distributed. Other of these or attractors may be easily dismantled and readily co-opted into other structures but generate week entrainment non-the-less. As cortical space becomes restricted the bias generated by the strong hold of or attractors may percolate and entrain all distributed - non spatio-temporal - or attractors, the stable and less stable ones. This kind of phenomenon may explain the emergence of the connectival use of or and how its functional aspects may be readily exploited.