Martin (1992: 497-8):
Halliday (1961) similarly includes context as a level of language concerned with the relationship between form and extra-textual features of the situation. This is glossed as semantics in Fig. 7.4 (from Halliday, McIntosh and Strevens 1964: 18), although it is important to keep in mind that Halliday (1961) recognises both formal and contextual meaning, with contextual meaning "an extension of the popular — an traditional linguistic — notion of meaning" (1961: 245). As Halliday clarifies in Thibault (1987: 614),
"the whole system is meaning creating. Meaning is the product of the interrelations among the parts";
it is a mistake in other words to make
"too close a tie-up between 'meaning' and the notion of a specifically 'semantic' level."
 See the clarification in the preceding post.
 This is very misleading, especially when juxtaposed with this co-text. In this very early paper by Halliday, 'formal meaning' is the "information" of information theory, and 'contextual meaning' Halliday equates with 'semantics', as stated in Halliday et al. (1964). The following quote from Halliday (1961) clarifies the matter:
Language has formal meaning and contextual meaning. Formal meaning is the "information" of information theory, though (i) it can be stated without being quantified and was in fact formulated in linguistics independently of the development of information theory as a means of quantifying it, and (ii) formal meaning in lexis cannot be quantified until a method is found for measuring the information of non-finite ("open") sets. The formal meaning of an item is its operation in the network of formal relations.
Contextual meaning, which is an extension of the popular — and traditional linguistic — notion of meaning, is quite distinct from formal meaning and has nothing whatsoever to do with 'information'. The contextual meaning of an item is its relation to extratextual features; but this is not a direct relation of the item as such, but of the item in its place in linguistic form: contextual meaning is therefore logically dependent on formal meaning.
 This continues the confusion between stratification and semogenesis. To be clear, this quote from Halliday refers to semogenesis — 'meaning creating' — not to stratification.
 The notion of meaning as a semantic level is meaning as a stratum in the stratification hierarchy, which is a means of parcelling out the complexity of the meaning making process into levels of symbolic abstraction. Halliday (2008: 14) clarifies this narrower use of the word 'meaning':
Realisation is the relationship among strata… wordings realise patterns of meaning, which we refer to as the stratum of semantics. (Note that “meaning” is here being used in its narrower, more specific sense, to refer just to patterns in semantics.)