Martin (1992: 579):
Hasan's major innovation has been to base her study on semantic variables, essentially by conceptualising system networks as variable rules and coding her data on the basis of selections from delicately elaborated discourse semantic networks she devised.
This is misleading. The unwarranted intrusion of the word 'discourse' falsely implies that Hasan shares Martin's misconstrual of semantics as discourse semantics. Hasan's networks are semantic networks.