Martin (1992: 488):
The basic strategy adopted to explore this question [how do modules of the semiotic model interface?] was to look closely at the interaction of reference chains and lexical strings with specific lexicogrammatical and phonological variables — specifically TRANSITIVITY rôles, Theme, New and Subject. The texture deriving from these four types of interaction was discussed under the headings of cohesive harmony, method of development, point and modal responsibility.
 This confuses the content plane (New information) with the expression plane (phonology).
 The term and concept of 'texture' come from Halliday & Hasan (1976). Texture is the property of being a text, and is created by the textual metafunction. Here it is misapplied to the experiential metafunction (TRANSITIVITY rôles) and the interpersonal metafunction (Subject).
 The term and concept of 'cohesive harmony' come from Hasan (1984, 1985). Martin misunderstands and misrepresents Hasan's model and critiques these as if they were her model. See the evidence in previous posts here, here, here, and here.
 The term and concept of 'method of development' come from Fries (1981). Martin's 'method of development' is writing pedagogy masquerading as linguistic theory. He rebrands 'topic sentence' as 'hyper-Theme' — a term taken from Daneš (1974) but redefined — and 'introductory paragraph' as 'macro-Theme'.
 The term and concept of 'point' come from Fries (1981). Martin's 'point' is, again, writing pedagogy masquerading as linguistic theory. He rebrands 'paragraph summary' as 'hyper-New', and 'text summary' as 'macro-New'.
 The term and concept of 'modal responsibility' come from Halliday (1985), where it is the meaning of 'Subject': that on which the validity of a proposition or proposal rests. Martin first misunderstands 'modal responsibility', in proposals, as 'the constituent responsible for seeing that goods are exchanged or a service performed' (p461). He then misinterprets the meaning of Subject from a discourse perspective, in terms of exchange resolution (p462), as 'what is at stake' (p464) and as 'participant at risk' (p477).