This pointed to some major theoretical differences between 3 SFL models (that "make different and often opposing assumptions”)
In model a) structure and function are treated stratally, structure as grammar and function as semantics
In b) structure and function are treated by rank - grammatical structures realise functions at higher ranks, and semantics is the meanings realised by grammar
Blogger Comments:In c) grammatical structures realise higher rank functions, but structure and function are properties of both grammatical and semantic strata
 The 3 SFL models that Rose has in mind are Halliday's original model (b), and two variations derived from it: Fawcett's Cardiff Grammar (a), and Martin's Discourse Semantics (c).
 'Structure' and 'function' are each one half of two separate dichotomies, each involving the relation of 'realisation'. On the one hand, 'structure' contrasts with 'system' — this is the relation between the syntagmatic and the paradigmatic axes.
On the other hand, 'function' contrasts with 'form' — this is the relation of symbolic abstraction that obtains between strata, and, in the rank scale, between elements of function structure, such as Senser^Process, and syntagm that realises them, such as nominal group^verbal group.
 In Fawcett's derived model, content system and structure are distinguished stratally. Halliday & Matthiessen (1999: 429):
In Fawcett’s model, there is only one system–structure cycle within the content plane: systems are interpreted as the semantics, linked through a “realisational component” to form, which includes items and syntax, the latter being modelled structurally but not systemically…
 In Halliday's original model, system and structure are modelled at both strata of the content plane. Halliday & Matthiessen (1999: 429):
…in our model there are two system-structure cycles, one in the semantics and one in the lexicogrammar. Terms in semantic systems are realised in semantic structures; and semantic systems and structures are in turn realised in lexicogrammatical ones.
 Grammatical 'structures' do not "realise higher rank structures". Syntagms (forms) of a lower rank realise function structures of a higher rank.
 Rose's claim that only Martin's derived model posits "structure and function" on both content strata is demonstrably false. For structure on both strata in the original model, see ; for function on both strata, consider, inter alia, the metafunctional organisation of both semantics and grammatics.