See comments to parid2178. Note further the confusion engendered by the notion of "special nodes." It seems clearer to say that there are nodes which participate in assertions and nodes that do not. Of those that do, the arcs within particular assertions attribute roles to the nodes in that assertion. A node that is an a-node in one assertion, may well be a t-node in another. It is insufficiently clear that nodes are not of some static "kind" in the current draft. It is only within a particular assertion and then by determining the roles that the nodes play in that assertion, that arcs can illuminate the various parts of the assertion.