David Rose On Realisation [Revised]

On 30/9/13, David Rose wrote on the sys-func and sysfling lists:

Well …the term realisation has been used for three types of relation - between strata, between axes (system/structure), and between ranks. So higher rank units are not simply composed of lower ranks units… higher rank functions (e.g. participant functions) are realised by lower rank units (e.g. nominal groups).

The whole area of realisation is not clearly resolved as far as my understanding stretches, e.g. Halliday treats KEY as a grammatical system, that is realised by the phonological system of TONE in the 'grammatical environment' of MOOD (Halliday & Greaves p123), but its features are stated as (graduated) speech functions, e.g. for declarative clauses, 'reserved statement: tone 4; insistent statement: tone 5...' (IFG3 p142). So are these interstratal, axial or other realisation relations?

The work on semantic networks of speech functions is accompanied by grammatical realisation statements, as Annabelle pointed out, e.g. [demand info:confirm:verify:reassure] is realised by major:indic:declarative:tagged:reversed mood. So are these axial realisation statements or interstratal or both at once? How are they like and unlike realisation statements for grammatical systems?

Im not suggesting these are wrong, but that our comfortable model of systemic features realised by function structures, or of 'meaning realised by wording realised by sounding' is not so simple. Particularly when 'the view from above' is considered, e.g. when is 'can you open this window' a question or command?

and then in reply to Brad Smith:

all strata contribute layers of meaning, that 'meaning realised by wording realised by sounding' is descriptively inadequate.

Blogger Comments:

[1] The term realisation has a precise, well-defined meaning in SFL.  It is always used in the theoretical architecture of SFL for the one type of relation: the relation between different levels of abstraction.  It is the nominalisation of the verb realise when serving as an 'intensive identifying' process that relates a lower level of abstraction (a Token) to a higher level of abstraction (a Value).  So wherever there is a Token-Value relation in the architecture of the theory, the term realisation is used.  For example, 

• it relates a lower (less abstract) stratum to a higher (more abstract) stratum;
• it relates the syntagmatic (less abstract) axis to the paradigmatic (more abstract) axis;
• it relates group rank (less abstract) forms to clause rank (more abstract) functions.

[2] Thus, the theoretical notion of realisation is "clearly resolved", at least to this extent, and it is indeed Rose's understanding of the theoretical notion that isn't.

[3] This is misleading, because it is untrue. The features of KEY are not stated in Halliday & Matthiessen (2004: 142) as "(graduated) speech functions". For example, in the following

(1) declarative clauses:

unmarked statement: tone 1

reserved statement: tone 4

insistent statement: tone 5

tentative statement: tone 3

protesting statement: tone 2

for the grammatical MOOD feature 'declarative', each of the first terms is a grammatical KEY feature, each of the second terms (statement) is a semantic SPEECH FUNCTION feature, and each of the third terms is a phonological TONE system feature.

The realisation relations, then, are obviously interstratal. Matthiessen (1995: 28):

The semantic system of SPEECH FUNCTION is realised through the grammatical systems of MOOD & KEY, and KEY is realised through the phonological system of TONE. So, for instance, if I select within the semantic system of SPEECH FUNCTION to give information, i.e., to make a statement, with some reservation about the validity of the information, I realise this by selecting within the grammatical systems of MOOD & KEY features such as declarative and reservation; and I realise reservation by selecting within the phonological system of TONE a falling-rising tone (tone 4).

[4] The realisation statements in semantic networks (such as SPEECH FUNCTION) that specify features in lexicogrammatical networks (such as MOOD) are clearly relating the (more abstract) stratum of semantics to the (less abstract) stratum of lexicogrammar, and, as networks, on the paradigmatic axis.

[5] The realisational relation between meaning and wording, and between wording and sounding, as different levels of abstraction, is simple enough for those who take the trouble to understand it, and are not motivated to misunderstand it. 

[6] It is precisely the distinction between meaning (semantics) and wording (lexicogrammar), and the realisational relation between them, that makes theorising about metaphors of mood systematic.

[7] It is true that all strata contribute to making meaning, that is: to semogenesis.  Layers of meaning, however, on the SFL stratification model, are all on the semantic stratum, and are created through grammatical metaphor, where congruent meanings (Value) are realised by metaphorical meanings (Token).

[8] The SFL stratification model incorporates the notion of metaredundancy; so strictly speaking:

• meaning is realised by [realisation of wording in sounding], or
• [the realisation of meaning in wording] is realised by sounding.

If it is indeed the model that is inadequate, Rose has yet to demonstrate the fact.