Martin (1992: 574):
The first point that needs to be made is that the interpretation of language and context here is indeed multi-structural and polysystemic. System/structure theory has been re-involved in the description on a number of different levels — rank, stratum and plane — most of which involve metafunctional diversity and so can be analysed simultaneously as particle, wave and prosody; in addition, synoptic and dynamic perspectives on text as system and text as process have been introduced.
 As previously explained, context is interpreted here as types of language, rather than as context, with the result that, being context, these types of language are not considered language.
 This is misleading, in that it overstates what has actually been done. Almost none of the systems that Martin has provided specifies structural realisations. This is partially disguised by the fact that some networks do include realisation statements; however, these merely provide textual instances of the feature.
- Of Martin's 49 discourse semantic system networks, across four metafunctions, only 4 specify structural realisations, and all are confined to the interpersonal metafunction.
- Of Martin's 11 register system networks, not one specifies any structural realisations; that is, no register structures are specified by register systems.
- Martin provides 0 genre systems — only taxonomies of types (factual and story genres); that is, no genre structures are specified by genre systems.
 This is misleading, in that it overstates what has actually been done. Martin has not provided a rank scale for his planes of register and genre, and in the case of the stratum of discourse semantics, only one of the four metafunctional systems, the interpersonal, includes a rank scale: exchange and move. The ranks discussed in the experiential dimension of discourse semantics — the clause and group — are ranks of a different stratum: lexicogrammar.
 The level that does not involve metafunctional diversity is genre. Metafunction is thus another dimension in which the model is inconsistent with the architecture of SFL theory, which follows from the misinterpretation of genre as context.
 To be clear, these are the favoured modes of structural realisation only, varying according to metafunction. Significantly, these were introduced in the section on genre, the plane without metafunctional diversity.
 This confuses text with language. Text is only the instance pole of the cline of instantiation. Language is the entire cline, from systemic potential to actual instance, with every point on the cline providing a different perspective.
'Text as system', therefore, is the instantial system; i.e. the system of an actual text, not the system of the language as a whole. It is the instance viewed from the system pole. Halliday & Matthiessen (1999: 384):
If we look at logogenesis from the point of view of the system (rather than from the point of view of each instance), we can see that logogenesis builds up a version of the system that is particular to the text being generated: the speaker/writer uses this changing system as a resource in creating the text; and the listener/reader has to reconstruct something like that system in the process of interpreting the text — with the changing system as a resource for the process of interpretation. We call this an instantial system.
'Text as process', on the other hand, is the process of instantiation that occurs at the instance pole of the cline of instantiation during logogenesis. As previously explained, Martin misconstrues 'process' as structure, the syntagmatic axis, "viewed dynamically". That is, he confuses the instantiation of the system as instance with the axial realisation of the system as structure.