Martin (1992: 575):
The register and genre theory reviewed and developed above represents systemic theory's attempts to model heteroglossia and dialogism; it does this by formulating register and genre as social semiotic systems realised through text, thereby providing an account not simply of how one text relates to another (cohesion across products) but in addition of how one text relates to all the texts that might have been (product in relation to system). …
The interpretation does however need to be qualified in two important respects — namely heterogeneity and semogensis [sic] (i.e. semiotic change).
 This is doubly misleading. Firstly, 'the register and genre theory reviewed and developed above' does not represent systemic theory's attempts. On the contrary, it is inconsistent with SFL theory, and represents Martin's attempts only. Secondly, 'the register and genre theory reviewed and developed above' does not model heteroglossia and dialogism. On the contrary, what is claimed bears little or no relation to heteroglossia and dialogism; see further below.
 This is inconsistent, both with Martin's model and with SFL theory.
In terms of Martin's stratification model, register and genre systems are realised by the systems of language, not by text. In this, Martin confuses the system and instance poles of the cline of instantiation.
In terms of SFL theory, it involves two confusions. Firstly, the notion of register and genre systems confuses a midway point of variation on the cline of instantiation (register/genre/text type) with the system pole of the cline. Secondly, it misconstrues the relation between system and text as realisation instead of instantiation. (This in addition to the inconsistencies entailed by modelling varieties of language as context rather than language.)
 This is misleading in terms of both register and genre. In terms of register, any chance of providing an account how one text relates to another is undermined by Martin's numerous misinterpretations of field, tenor and mode systems, as demonstrated in many previous posts. In terms of genre, Martin merely provides two simple taxonomies of factual and story genres (text types). Martin nowhere presents any account of how his formulation of register and genre relates one individual text to another.
 To be clear, in SFL theory, cohesion is not a relation between 'products' (texts).
 This is the opposite of what is true. This is precisely what Martin's model does not do. In SFL theory, the relation of texts to text potential — of instances to system — is modelled as the cline of instantiation. Martin's model is inconsistent with the cline of instantiation, due to the fact that it misconstrues the midway point on the cline (register/genre), not as language, but as systems of context, and as such, as higher levels of symbolic abstraction than language. This follows from not understanding either stratification or instantiation, as demonstrated many times in previous posts.
 To be clear, Halliday & Matthiessen (1999: 18) identify three types of semogenic processes:
- logogenesis, the instantiation of the system in the text;
- ontogenesis, the development of the system in the individual; and
- phylogenesis, the evolution of the system in the species.