Why Fawcett Thinks SFL Requires A Separate Component For Realisation Rules

Fawcett (2010: 64-5):

4.8 Why a SFL model of language requires a separate component for the realisation rules 

Finally, we must consider the implications for the model of language summarised in Figure 4 of another interesting change in Halliday's representations of his model of language. It is a change that correlates with the view that we have just been examining, i.e., the view that the system networks of TRANSITIVITY, MOOD, THEME etc. belong at the same level as their outputs. 

The reason why it is important to examine this change too is that it carries with it an implication that is even more drastic in its consequences for the model of language outlined in Figure 4 than the change discussed in the last section. This is because it implies that the grammar has no separate component for the realisation rules. If this concept were to be sustained, there would be no 'form potential' in Figure 4 that corresponded to the 'meaning potential'. And this in turn would have serious consequences for the picture of language to be presented here, and especially in Chapters 5 and 9, where we shall make a major distinction in the theory of syntax to be presented there between 'syntax potential' and 'instances of syntax'.

Blogger Comments:

Reminder:

[1] To be clear, here Fawcett is considering the implications of his own misunderstanding of SFL — see [2] below — for his own model (Figure 4) which, as demonstrated many times on this blog, is internally inconsistent and confuses the system-structure relation (realisation) with the system-instance relation (instantiation).

[2] To be clear, in SFL theory, the system networks of TRANSITIVITY, MOOD, THEME and their "outputs" are both located on the same stratum of symbolic abstraction, namely: lexicogrammar.  Because Fawcett's "output" confuses structure with instance, it is necessary to take each in turn.

For system and structure, while both are located on the same stratum, they differ in terms of symbolic abstraction within that stratum, with the higher level, system, realised by the lower level, structure.  For system and instance, on the other hand, both are most obviously located on the same stratum, since a grammatical instance is an instance of a grammatical system.

[3] To be clear, in SFL theory, realisation statements that specify structural realisations are located in systems, at points that satisfy the conditions for their activation (instantiation entails both the selection of features and the activation of realisation statements).

∞

In short, Fawcett's reason why SFL requires a separate component for realisation rules is that it will have serious consequences for his own model if it doesn't.

Fawcett's Powerful, Theoretically Well-Motivated And Computer-Tested Models Of Language

Fawcett (2010: 64):

Thus Figure 4 expresses the powerful, theoretically well-motivated and computer-tested models of language that have been implemented in both the Sydney and the Cardiff Grammars. In contrast, the view of language summarised in Figure 5 (which is what Halliday's re-interpretation of his earlier insight entails) loses precisely the major insight of his revolutionary changes in the 1960s, as summarised in Sections 4.1 to 4.4 of this chapter — i.e., the insight that there is a relationship of realisation between the system networks of meaning potential and the structural outputs.

Blogger Comments:

[1] The word 'thus' here is misleading, because it gives the false impression that the misunderstandings that preceded it validate the conclusion that follows it.

[2] As demonstrated over and over on this blog, Fawcett's flowchart (Figure 4) misrepresents the architecture of SFL theory and misunderstands the notions of instantiation and realisation, both axial and stratal.



[3] As demonstrated in preceding posts:

• Halliday did not reinterpret his earlier insights,
• Figure 5 does not represent Halliday's interpretation of SFL architecture,
• Figure 5 is Fawcett's own invented 'Straw Man' which he fallaciously argues against on the false pretext that he is arguing against Halliday's model.

[4] Here again Fawcett, like his Figure 4, confuses system in relation to instance (instantiation) with system in relation to structure (axis).  In Halliday's model of his own theory, the relation between system and structure is realisation.

Misunderstanding The System Network

Fawcett (2010: 64n):

In any case, an adequate model of the full process of generation requires other ways of modelling decision-making in the higher stages of planning. System networks can be regarded as a special type of 'decision tree' that is incorporated within the semiotic system of language itself.

Blogger Comments:

[1] This is a bare assertion, unsupported by argument or evidence.

[2] To be clear, system networks do not model decision making, and thus there is a very important sense in which a system network is not a decision tree:

A decision tree is a decision support tool that uses a tree-like model of decisions and their possible consequences, including chance event outcomes, resource costs, and utility. It is one way to display an algorithm that only contains conditional control statements.

A system network does not represent a sequence of decisions. A system network systematises the relations between options in terms of types of expansion:

• delicacy = elaboration
• disjunction = extension: alternation
• conjunction = extension: addition
• entry condition = enhancement: condition
• realisation statement = elaboration (+ identifying)

For instance, a lexical item is not the output of a sequence of decisions.  A lexical item realises all the relations between all of the features that specify it, from the most general systems of the grammar to their most delicate elaboration.

Summarising The Main Point Made In This Section

Fawcett (2010: 63-4):

Let me summarise the main point that is made in this section. This is that, even if Halliday turns out to be right about the need to add another level of system networks above those in the 'meaning potential' of the lexicogrammar (which I do not think he is), it does not necessarily follow that we must deny that there is also realisation between the established levels of the generative apparatus that I have described in Chapter 3 (this generative apparatus being exemplified both in Appendix A and in Halliday's own early generative grammars). To assert that the only relationships involved in Figure 5 — and so in Figure 4 — are ones of 'instantiation' would be to sacrifice the great insight of the later 1960s and early 1970s that the system networks of TRANSITIVITY, MOOD, THEME etc. are choices in meaning.* And for what would this sacrifice be made? Ultimately, it would be for the abstract (and undesirably limiting) notion that the specification of the 'potential' at every level of language necessarily has the form of a system network. The more desirable alternative, as our work in the COMMUNAL Project at Cardiff has shown, is to allow that a full model of language in use may require different ways of specifying the 'potential'. Indeed this concept is illustrated in the outline proposed in Chapter 3, in that we saw there that it is the role of the realisation rules to specify the 'form potential'.

* It would not solve the problem to label them as 'formal meaning' in contrast with 'semantic meaning' ; they are still patently more semantic than their syntactic correlates.

Blogger Comments:

[1] To be clear, as already explained, the distinction between semantic networks and grammatical networks provides a systematic means of accounting for grammatical metaphor.

[2] Here again Fawcett confuses the systemic potential of language (meaning potential) with the stratum of lexicogrammar.

[3] To be clear, the 'generative apparatus' is Fawcett's flowchart (Figure 4).  As previously explained many times, it is based on misunderstandings of axis, stratification and instantiation, as well as being internally inconsistent.

[4] This is potentially misleading. To be clear, the exemplication of Fawcett's flowchart in Appendix A involves a system network and realisation rules.  It is this that it has in common with "Halliday's own early generative grammars".  The system itself confuses semantics (a system for 'thing') with lexicogrammar (nominal group features); see the earlier critique here.

[5] As previously explained, this is Fawcett's misunderstanding, deriving from the misunderstandings inherent in his own model.

[6] On the one hand, this is a non-sequitur inferred from a false premiss. On the other hand, in SFL theory, the system networks of TRANSITIVITY, MOOD, THEME etc. are choices in wording that realise meaning, and in the absence of grammatical metaphor, the wording and meaning agree (are congruent).

[7] On the one hand, 'undesirably limiting' is merely interpersonal attitude (negative appreciation) falsely presented as if logical argument.  On the other hand, the system network is the formalism of Systemic Functional Linguistics, embodying the fundamental principle of meaning as choice.

[8] Again, 'more desirable' is merely interpersonal attitude (positive appreciation) falsely presented as if logical argument.  The claim that work has shown this to be so is merely Fawcett's own assessment, made without providing the evidence on which it is based.

[9] This is potentially misleading.  The only alternative to system networks for representing potential that Fawcett proposes is realisation rules/statements (Figure 4), and in SFL theory, these are located in system networks.

[10] To be clear, in SFL theory, this contrast is the stratificational distinction of wording (lexicogrammar) and meaning (semantics), these being two levels of symbolic abstraction modelling the same phenomenon: linguistic content.