As we have explained, functional language plays the role it does because it is structurally different. We assume that in the long run - across a millennium - no vocabulary is immune from structural changes since low connectivity involves an optimization of cortical space. Of course, not all vocabulary becomes functional but all vocabulary becomes extinct eventually. It makes sense to assume that, as the cortical space is hosting an increasing number of cloning competitions, there must be strategies to optimize limited resources. Phase transition is one way by which cortical space may be optimized. Furthermore, we suggest that phase transition in proto-language may have promoted grammatical forms without the necessity of major restructuring of neural resources at the genetic level.
The percolation phenomenon - remember the Kerplunk! example - would make it possible for grammaticallization to emerge from a system that has not displayed that particular feature at any time in the past. Moreover, constituents in a proto-language system do not have to be biased towards a grammatical arrangement in order to develop that feature. Grammaticalization may emerge (to sustain a causal flow in a proto-linguistic transaction), from the exploitation of low connectivity structures in cortical territories. The level of generalization in functionalized items may be just what is necessary to accommodate increasing complexity in vocables.