Halliday (2002 [1979]: 212, 213, 215):
In the logical mode, reality is represented in more abstract terms, in the form of abstract relations which are independent of and make no reference to things. …
Since we are claiming these structural manifestations are only tendencies, such an argument is only tenable on the grounds that the type of structure that is generated by the logical component is in fact significantly different from all the other three. The principle is easy to state: logical structures are recursive. …
True recursion arises when there is a recursive option in the network, of the form shown in Figure 14, where A:x,y,z… n is any system and B is the option ‘stop or go round again’. This I have called “linear recursion”; it generates lineally recursive structures of the form a₁ + a₂ + a₃ + . . . (not necessarily sequential):
… Logical structures are different in kind from all the other three. In the terms of systemic theory, where the other types of structure – particulate (elemental), prosodic and periodic – generate simplexes (clauses, groups, words, information units), logical structures generate complexes (clause complexes, group complexes, etc.) (Huddleston 1965). … While the point of origin of a non-recursive structure is a particular rank – each one is a structure “of” the clause, or of the group, etc. – recursive structures are in principle rank-free: coordination, apposition, subcategorisation are possible at all ranks.
Postscript:
In his seminar paper (13/8/21) on 'construing entities', Martin claimed that there are no recursive networks anywhere in the SFL literature. The network above occurs in one of the references he provided.
