(Accusing Halliday Of) Confusing Axial Realisation With Instantiation

Fawcett (2010: 61-2):

Passages in Halliday's recent writings such as the following two seem to be attempts to reconcile the two senses in which he now finds himself using the term "realisation" — senses which in the new model have in fact become incompatible. Is the concept of 'realisation' interstratal or intrastratal — or is it really possible that it can be stretched far enough to be used for both, without losing its integrity? In the first of the two passages Halliday writes:

realisation" is both "the relation between the 'strata', or levels [i.e .the original sense of "realisation"] of a [...] semiotic system — and, by analogy, between the paradigmatic and the syntagmatic phases of representation within one stratum [my emphasis]. (Halliday 1993:4505) 

And in a slightly later work Halliday describes the use of the term "realisation" in the 'intrastratal' sense as an "extension" of the concept of 'realisation', saying: 

Realisation is [...] extended to refer to the intrastratal relation between a systemic feature and its structural (or other) manifestation [my emphasis]. (Halliday 1996:29) 

If Halliday had not used the words "within one stratum" and "intrastratal" in these two passages, they could have been interpreted perfectly satisfactorily as describing the relationship of realisation that holds between the two levels of instances in the model represented in Figure 4 of Chapter 3, i.e., (1) the selection expression of features and (2) the structure at the level of form. And in this case the use of the term "realisation" would not be an "analogy" or an "extension". But Halliday did use those words, and under a strict interpretation of their meaning, the relationship must, in his current framework, be surely be said to be one of 'instantiation' rather than 'realisation'. 

Blogger Comments:

[1] This is misleading.  Halliday has only ever used the theoretical term 'realisation' in one sense; as an identifying relation between two levels of symbolic abstraction.  What varies is the dimension along which the relation obtains; e.g.

• globally between strata,
• locally within strata, between function (Process) and form (verbal group), and
• locally within strata between paradigmatic and syntagmatic axes.

[2] This is misleading on two counts.  Firstly, there is no "new" model, as previously explained.  Secondly, the 'interstratal' and 'intrastratal' senses of realisation are not incompatible.  Both involve an identity relation between two levels of symbolic abstraction.

[3] This makes it clear that Fawcett does not understand the meaning of 'realisation'; he merely takes it as a label for the relation between strata.  It thus also suggests he does not understand that strata represent levels of symbolic abstraction, as the inconsistencies in his flowchart (Figure 4) also suggest.

[4] To be clear, both of the Halliday quotes are concerned with the realisation relation between the paradigmatic and syntagmatic axes on the stratum of lexicogrammar.  As such, they do not describe the relation the two "levels of instances" in Fawcett's flowchart:

As previously demonstrated, a selection expression of features is an instance of systemic potential, whereas structure is the realisation of system, not an instance of it. That is, the flowchart confuses instantiation with (axial) realisation.

[5] To be clear, here Fawcett confuses axial realisation (the subject of the Halliday quotes) with instantiation, and does so on the basis of the confusion in his own model.

In short, Fawcett projects his own misunderstanding onto Halliday.

Confusing Axial Realisation With Stratal Realisation

Fawcett (2010: 61):

The specific problem is that, in the earlier stage of the development of the theory (when the system networks were regarded as being at a higher level than the structures that manifest them), Halliday had identified the set of operations that change the selection expression into the structures as 'realisation statements'. So are the outputs from the system networks of TRANSTIVITY, MOOD, THEME and so on really at the level of form, or are they are the same level as the system networks, i.e., at a level that has been "pushed [...] fairly far [...] in the direction of the semantics" (Halliday 1994:xix)?

Blogger Comments:

[1] This is misleading.  As previously explained, system networks are at a higher level of symbolic abstraction than the structures that realise them, and, contrary to Fawcett's misrepresentation, this is as true now as it was in earlier stages of Halliday's development of his theory. This is the relation between paradigmatic axis and the syntagmatic axis, and it applies to every stratum of language: semantics, lexicogrammar and phonology.  That is, different levels of symbolic abstraction (system and structure) can be construed within a given level of symbolic abstraction (stratum).  This is why the term 'realisation' applies to both axial and stratal relations.

[2] This is correct.  In SFL theory, a realisation statement specifies how a systemic feature is realised structurally.

[3] Again, Fawcett's term 'outputs' blurs the distinction between the theoretical dimensions of axis (structure as the "output" of system) and instantiation (instance as the "output" of potential).

[4] To be clear, on the one hand, 'form' is not a level in the SFL hierarchy of stratification.  On the other hand, grammatical systems and structures are at different levels (axes) of symbolic abstraction, within the same level (stratum) of language (lexicogrammar).

[5] To be clear, the level that has been "pushed fairly far in the direction of semantics" is the stratum lexicogrammar (Halliday 1994: xix).

Misunderstanding And Confusing Realisation And Instantiation

Fawcett (2010: 61):

Taking this position brings with it a serious problem. We shall address it now, because it is a problem which, like the introduction of the higher level of meaning, has the potential to make it impossible to compare the representations at the level of form of the Sydney and the Cardiff Grammars. 

The problem is as follows. In his earlier descriptions of the grammar (1969/81) and (1970/76b), Halliday showed the relationship between the system networks and the output structures to be one of realisation. But in his this [sic] second approach to the representation of meaning the relationship must logically be regarded as one of instantiation. This is because system and structure are presented in the new model as being at the same level of language.

Blogger Comments:

[1] This is misleading.  As previously demonstrated, Halliday has only had one view on this matter, and Fawcett's misrepresentations derive from his own misunderstandings of stratification.

[2] This is true.  In Halliday's model, the relation between the paradigmatic axis and the syntagmatic axis, between system and structure, is realisation.

[3] Fawcett's use of the word 'output' is the source of his confusion, since it confuses two distinct dimensions of SFL theory:

• syntagmatic structure as the "output" of paradigmatic system (realisation),
• instance as the "output" of systemic potential (instantiation).

[4] Fawcett's argument is as follows:

Premiss (reason): because system and structure are presented as being at the same level of language

Conclusion (result): the relationship must logically be regarded as one of instantiation

To be clear, the conclusion does not follow from the premiss.  Presenting system and structure as being of the same level of language does not logically entail that the relation between them is necessarily one of instantiation (the relation between system and instance).

This also confirms — along with his flowchart (Figure 4) — that Fawcett does not understand the theoretical dimension of instantiation.

[5] To be clear, in SFL theory, system and structure are modelled as the same level of language, whether at the level of semantics, lexicogrammar or phonology.  However, they are modelled as different levels of symbolic abstraction within that level of language, such that the lower level Token, structure, realises the higher level Value, system.

This also confirms — along with his flowchart (Figure 4) — that Fawcett does not understand the theoretical relation of realisation.

Halliday's Refusal To Add To Fawcett's Flowchart

Fawcett (2010: 60-1):

However, adding to the model in Figure 4 in this way has not been Halliday's response to the situation. Instead, his later works present the view that the system networks of meaning potential and their outputs at the level of form shown in Figure 4 are all at the same level. 

Blogger Comments:

Reminder:

[1] This is misleading.  The reasons Halliday has not added to Fawcett's model in Figure 4 are, as previously argued:

• it is not Halliday's model, 
• it is not a model of language consistent with SFL theory, and
• it is not a model consistent with itself.

[2] This is misleading.  To be clear, Halliday's later works do not present "system networks of potential and their outputs at the level of form at the same level", not least because "outputs at the level of form" is not Halliday's model.  One reason for this is that Fawcett's "outputs at the level of form" confuses two distinct dimensions of SFL theory: axis (local) and instantiation (global); see Halliday & Matthiessen (2014: 32).

The relation between system and structure is the identifying relation of symbolic abstraction between the paradigmatic and syntagmatic axes.  For a given stratum, syntagmatic structure (Token) realises paradigmatic system (Value).

On the other hand, the relation between system and instance is the attributive relation of instantiation, such that an instance is a token (Carrier) of the type (Attribute) system.  See Halliday & Matthiessen (1999: 145).

In the next post, it will be seen that Fawcett accuses Halliday of confusing realisation and instantiation.

The First Of Two Possible Effects Of Semantic Systems On Fawcett's Flowchart

Fawcett (2010: 60):

It would be reasonable, if this higher level of system networks really is necessary, simply to add the higher level to the model illustrated in Figure 4 of Chapter 3. This would require the addition of the following sub-components: (1) a new higher system network (a second 'meaning potential'), which would generate its own set of features after each traversal of the network), (2) the selection expression of the 'instances' that have been chosen, (3) a 'realisation component' that states what the effect of each choice is (i.e., 'preselection' rules, which would specify that if Feature X is chosen in the higher network, Feature Y must be chosen in the lower network), and (4) the output from the operation of that component, i.e., the list of features to be chosen in the lower network. The result of this change would be to add the components decribed [sic] here above the model shown in Figure 4 of Chapter 3. This is the first of the "two possible effects" referred to in the title of this section.

Blogger Comments:

[1] This higher level of system networks (semantic stratum) really is necessary if grammatical metaphor is to be systematically accounted for, and if knowledge is to be included in the SFL model of language.  Halliday & Matthiessen (1999: 429):

… grammatical metaphor is a central reason in our account for treating axis and stratification as independent dimensions, so that we have both semantic systems and structures and lexicogrammatical systems and structures. Since we [unlike Fawcett] allow for a stratification of content systems into semantics and lexicogrammar, we are in a stronger position to construe knowledge in terms of meaning. That is, the semantics can become more powerful and extensive if the lexicogrammar includes systems.

[2] This is misleading.  It would not be reasonable, because, as previously demonstrated, the flowchart depicted in Fawcett's Figure 4 is inconsistent with the architecture of SFL theory.

[3] This is misleading.  Here Fawcett describes how he would change his flowchart to accommodate the SFL distinction between meaning (semantics) and wording (lexicogrammar), as if such changes are relevant to the architecture of SFL theory.  That is to say, Fawcett's presumption is that Halliday's stratified model of language has to conform to Fawcett's self-inconsistent flowchart.

Misrepresenting Fawcett's Flowchart As The Architecture Of SFL Theory

Fawcett (2010: 60):

I claimed in Chapter 3 that Figure 4 (in Section 3.2 of that chapter) represents the way in which both the Sydney and the Cardiff grammars operate. As we have seen, Halliday's more recent writings have increasingly strongly taken the position — which he first explored in the 1970s — that there is a 'higher semantics' as well as the 'meaning potential' within the lexocogrammar (e.g., Halliday 1993 and 1996). How would this affect the overall model of language?

Blogger Comments:

Here again Fawcett makes multiple use of the logical fallacy known as proof by (repeated) assertion. 

[1] To be clear, Fawcett's Figure 4 (p36) is:

As first demonstrated here, and many times since, the model represented in Figure 4 is invalidated by its own internal inconsistencies, arising from misunderstandings of axis, stratification and instantiation.  Fawcett's flowchart does not represent the architecture of SFL theory, not least because the architecture is dimensional, not modular.  Halliday & Matthiessen (2014: 32):

[2] This is misleading, because, since the inception of SFL theory, Halliday has always posited a level of semantics (meaning) above a level of lexicogrammar (wording).  For example, Halliday & Hasan (1976: 5):

[3] Here Fawcett confuses meaning potential (the system pole of the cline of instantiation) with lexicogrammar (the level of wording in the hierarchy of stratification).

[4] On the one hand, the level of semantics is already a part of the SFL model.  On the other hand, by 'the overall model of language', Fawcett means his own flowchart (Figure 4), which he is about to use as the benchmark for assessing Halliday's model, as will be seen in the next few posts.  

Misrepresenting Halliday As Inconsistent With The SFL Model Of Language

Fawcett (2010: 60):

Nonetheless, there has been a knock-on effect of Halliday's adoption of the second position which leads him to describe the grammar in a way that appears to be in conflict with the general SFL model of language proposed in Chapter 3 and in expressed in Figure 4. The next section describes this problematical change of position by Halliday, and the simple alternative.

Blogger Comments:

[1] This is misleading, because it misrepresents Halliday, since, as previously demonstrated, Halliday only ever had one position on this matter (the stratal location of his grammatical networks).

[2] Here Fawcett presents his own model (Figure 4) as the benchmark by which to assess Halliday's version of Halliday's theory.  As previously demonstrated, Fawcett's model is invalidated by its internal inconsistencies with regard to the theoretical dimensions of axis, instantiation and realisation.

[3] The question of whether Halliday's version of Halliday's theory is problematical, and the question whether Fawcett's proposed alternative is valid, will be examined in the review of the next section.

Misrepresenting Halliday In A Footnote

Fawcett (2010: 59n-60n):

I should point out that, when Halliday adds a further level of system networks above the existing level, he neces[s]arily also add[s] a further component to the model, to enable the grammar to 'map' the choices made at one level onto the choices available at the lower level. However, he does not describe what this extension to the model entails, so I shall attempt to provide a summary of this in the next section. A further problem is that Halliday uses the term 'preselection' in two senses. The first is the standard sense of the 'preselection' that occurs in a realisation rule which 'pre-selects' a feature to be chosen on a subsequent traversal of the network (as we shall see in Chapters 5 and 9). But his second use is for the 'preselection' of a choice in a network that results from a choice in a higher component. It is of course important to keep a clear distinction between the levels (or strata) of language and the layers of structure within syntax. In the Cardiff Grammar, therefore, we use "preselection" only in the established sense of the relationship between layers of the tree structure. We use the terms "predetermine" and "predetermination" for the relationship between any higher component in the process of generation and the choices in the system network. 

Blogger Comments:

[1] This is misleading, because it is manifestly untrue.  The semantic networks Fawcett alludes to are presented in Halliday & Matthiessen (1999), which devotes 618 pages describing "what this extension to the model entails".

[2] This is misleading, because it is manifestly untrue.  There is only one sense of 'preselection' in the two usages cited by Fawcett: the selection of one feature entails the selection of another.  That is, the probability of their co-selection is 1.  The theoretical location of the features involved is irrelevant to the meaning of the term.  Halliday & Matthiessen (1999: 375):

More specifically, inter-stratal realisation is specified by means of inter-stratal preselection: contextual features are realised by preselection within the semantic system, semantic features are realised by preselection within the lexicogrammatical system, and lexicogrammatical features are realised by preselection within the phonological/graphological system.

[3] This is misleading, because it presents Fawcett's reinterpretation of a term introduced into SFL by Halliday as the established sense.

[4] To be clear, in SFL theory, there is no causal relation between strata.  Halliday & Matthiessen (1999: 25):

In any stratal system (i.e. any system where there are two strata such that one is the realisation of the other) there is no temporal or causal ordering between the strata. … the relationship is an intensive one, not a causal circumstantial one.

[5] To be clear, in Fawcett's model (Figure 4), because system networks are located at the highest level, meaning, there is no higher component.

[6] To be clear, in SFL theory, this process is the process of instantiation: the selection of features during logogenesis, the unfolding of text at the instance pole of the cline of instantiation.

Confusing Language As System With Language As Stratified

Fawcett (2010: 59):

From the viewpoint of our concerns in this book, however, this major difference of view about what is required 'above' the meaning potential that belongs within the grammar is relatively unimportant. As I have argued in this section, both of the two models have networks for TRANSITIVITY, MOOD, THEME etc. that are at the level of 'meaning potential'. Moreover, the features in both are realised by essentially the same types of 'realisation operation', as we shall see in Chapters 5 and 9.

Blogger Comments:

[1] Here again Fawcett confuses the SFL notions of language as system ('meaning potential') and language strata as levels of symbolic abstraction ('above', 'level').  The theoretical dimensions confused are thus those of instantiation and stratification.

[2] The validity of this claim will be examined in the reviews of Chapters 5 and 9.

Misrepresenting Halliday And Deploying A Logical Fallacy

Fawcett (2010: 59):

I am therefore as confident as it is ever possible to be in science that it is indeed possible to make a reality of Halliday's original proposal that the system networks of TRANSITIVITY, MOOD, THEME and so on should be, to adapt Halliday's metaphor (1994:xix) "pushed all the way to the semantics". We who work in the framework of the Cardiff Grammar — together with all of the many other systemic functional grammarians who take the view that the system networks of TRANSITIVITY, MOOD, THEME etc. constitute the major level of meaning in language — have therefore come to a different conclusion from Halliday on this matter.

Blogger Comments:

[1] This is misleading, since it misrepresents Halliday.  As previously explained, Halliday's systems have, from their beginnings, been presented as grammatical systems, not semantic systems.

[2] The citing of others who share Fawcett's view is here presented as an argument in its favour.  In terms of logical fallacies, this is another instance of the fallacy of relevance known as an appeal to popularity:

Appeals to popularity suggest that an idea must be true simply because it is widely held. This is a fallacy because popular opinion can be, and quite often is, mistaken.

[3] To be clear, on the SFL model, there is only one level of meaning in language: semantics.  Lexicogrammar constitutes the level of wording, the lower stratum of the content plane.

The Hypothesis That Grammatical Systems Can Be Developed Into A Model Of Semantics

Fawcett (2010: 58-9):

In the work done over the last fifteen years by my colleagues and myself at Cardiff, one of our main goals has been to develop a new model of a SF grammar that tests the hypothesis that the system networks of TRANSITIVITY, MOOD, THEME and so on are capable of being developed into a fully adequate model of semantics. Indeed, the very large computer implementation of a grammar (including lexis, intonation and punctuation) that we have built at Cardiff operates on precisely these principles. The first stage of this work was described in Fawcett, Tucker & Lin (1993) and related papers, and many aspects of the later stages have been reported in the many other papers listed in Fawcett (1998).  As a result of all this work by the many members of the team, I am convinced that, in Halliday's own words (1994:xix), the "choices in the grammar [i.e., the system networks] can be essentially choices in meaning without the grammar thereby losing contact with the ground".

Blogger Comments:

[1] To be clear, the theoretical motivation for distinguishing semantic systems (e.g. Halliday & Matthiessen 1999) from lexicogrammatical systems (e.g. Halliday & Matthiessen 2014) is the resultant ability to systematically account for grammatical metaphor as an incongruence between the two levels of symbolic abstraction.  As Halliday & Matthiessen (1999: 237) explain:

Of course, what we are recognising here as two distinct constructions, the semantic and the grammatical, never had or could have had any existence the one prior to the other; they are our analytic representation of the overall semioticising of experience — how experience is construed into meaning. If the congruent form had been the only form of construal, we would probably not have needed to think of semantics and grammar as two separate strata: they would be merely two facets of the content plane, interpreted on the one hand as function and on the other as form.

[2] The citing of work carried out in Fawcett's framework is here presented as an argument in its favour.  In terms of logical fallacies, this is the fallacy of relevance known as an appeal to popularity:

Appeals to popularity suggest that an idea must be true simply because it is widely held. This is a fallacy because popular opinion can be, and quite often is, mistaken.

Misrepresenting Halliday's Grammatical Systems As Semantic

Fawcett (2010: 58):

The perception that the system networks of TRANSITIVITY, MOOD, THEME etc. represent the meaning potential of a language is, in my view, the most significant of all of Halliday's insights. As I pointed out above, one important result of accepting this major claim is that it challenges us to develop our system networks further, with the explicit goal of making them represent choices between meanings (rather than forms), and so to model the level of semantics in language.

Blogger Comments:

[1] This continues the misrepresentation of Halliday's grammatical systems as semantic systems, and the misrepresentation of meaning potential (language as system) as semantics (meaning as a level of symbolic abstraction).

[2] To be clear, in Systemic Functional Linguistics, the system network is the theoretical formalism for modelling all strata.  In SFL, language is stratified as meaning (semantics), wording (lexicogrammar) and sounding/writing (phonology/graphology) — not meaning and form — with grammatical form incorporated through the rank scale of clause, group/phrase, word and morpheme.

For system networks that represent choices between meanings, see Halliday & Matthiessen (1999); for system networks that represent choices between wordings, see Halliday & Matthiessen (2014).

A False Claim Invalidly Inferred From A False Claim

Fawcett (2010: 58): 

In terms of the change from "Categories" to a modern SF grammar, we may say that the effect of the fundamental change in the theory in the late 1960s was that the concept of 'system' was removed from the account of language at the level of form, and made the central concept at the level of meaning. As a result there were now the two levels of 'instances' shown in Figure 4 of Chapter 3: the selection expression at the level of meaning and the richly labelled tree structure at the level of form.

Blogger Comments:

[1] This continues the misrepresentation of Halliday's grammatical systems as semantic systems and meaning potential (language as system) as semantics (meaning as a level of symbolic abstraction).

[2] This is a non-sequitur, since the latter claim is not entailed by the former, regardless of the validity of either claim.  That is, the relocation of grammatical systems to the level of meaning does not necessarily result in Fawcett's model of "two levels of instances".

[3] As Figure 4 shows, Fawcett's model confuses the dimension of instantiation: the relation between system and instance (selected features) with the dimension of axis: the relation between system and structure (tree structure).  (It also proposes that systems are realised by realisation rules.)

See also the critique of Figure 4 here.

Misrepresenting Halliday's Grammar As Semantic

Fawcett (2010: 58): 

For the purpose of a general comparison between the two models, then, we may treat the level of 'meaning potential' in Halliday's grammar (i.e., the level at which TRANSITIVITY, MOOD, THEME and so on are located) as roughly equivalent to the semantic system networks of the Cardiff Grammar. In other words, Halliday's adoption of the second position on levels of meaning makes no significant difference to the components of the model of language that we shall assume to be common to all of those who work in the framework of SFL. (But see the next section for a caveat to this claim.)

Blogger Comments:

[1] This continues the confusion between meaning potential (language as system) and meaning as a level of symbolic abstraction (semantic stratum), and the use of the confusion to misrepresent Halliday's grammar as semantic.

[2] This continues the misrepresentation of Halliday as having two "positions" on the stratification of the content plane as meaning (semantics) and wording (lexicogrammar), and the wrongful attribution to Halliday of recasting his grammatical systems are semantic.

[3] This is misleading.  On the one hand, the relocation of Halliday's grammatical systems to semantics creates serious theoretical inconsistencies and reduces its explanatory power (grammatical metaphor), as explained in previous posts.  On the other hand, the relocation is precisely what Fawcett needs in order to make theoretical space for his theory of syntax.

[4] But see the review of the next section for an examination of this caveat.

The Claim That The Outputs Of System Networks Are At The Level Of Form

Fawcett (2010: 58): 

Notice, moreover, that the outputs from any grammar with system networks of either type must be considered to be at the level of form, because they specify the sequence of the items that constitute the 'final' output ("final", that is, apart from specifying the output's spoken or written shape). And this is true of both the Sydney and the Cardiff Grammars, despite the differences between the types of representation that are found in each. (However, there are theoretical problems about the status of the Sydney Grammar representations, as we shall see in Chapter 7.)

Blogger Comments:

[1] As the term 'output' suggests, this misunderstands the notion of system in the architecture of SFL theory.  The process of selecting systemic features and activating realisation statements during logogenesis, the unfolding of text, is the process of instantiation, and so the relation between system and text is one of instantiation.   Halliday & Matthiessen (2014: 51):

But ‘text’ is a complex notion. In the form in which we typically receive it, as spoken and written discourse, a text is the product of two processes combined: instantiation, and realisation. The defining criterion is instantiation: text as instance. But realisation comes in because what becomes accessible to us is the text as realised in sound or writing. We cannot directly access instances of language at higher strata – as selections in meaning, or even in wording. But it is perhaps helpful to recognise that we can produce text in this way, for ourselves, if we compose some verse or other discourse inside our heads. If you ‘say it to yourself’, you can get the idea of text as instance without the additional property of realisation.

[2] See the review of Chapter 7 for the theoretical misunderstandings on which this falsehood is based.

Conceding That The Cardiff Approach Has Been Invalidated In Principle

Fawcett (2010: 57-8):

Since our purpose in this book is to establish the theory that is required for modelling syntax at the level of form, we must ask the question: "Does Halliday's adoption of the second position on the levels of meaning in language mean that the general framework within which we are comparing the Sydney and the Cardiff approaches to analysing syntax, as summarised in Figure 4 of Chapter 3, becomes inval[i]d?" The answer is that it might in principle have been invalidated by Halliday's recent decision, but that in practice it does not. The reason is that, whether or not we add a higher layer of 'meaning' to our model of language, there is still a level of meaning potential within the grammar. It is a level of description that Halliday describes as having "been pushed [...] fairly far [...] in the direction of the semantics" (Halliday 1994:xix). From the 'theoretical-generative' viewpoint (a concept that will be introduced in Section 5.2 of Chapter 5), there seems to be no very significant difference between the mode of operation of a grammar in which system networks that have been pushed "fairly far in the direction of the semantics" and one in which the system networks have been pushed all the way. 

Blogger Comments:

[1] To be clear, SFL theory models grammatical form as the rank scale.  Paradigmatically, each level on the rank scale serves as the entry condition for functional systems of that rank; syntagmatically, functions at one rank are realised by forms of the rank below.  As far as the term 'syntax' is concerned, Halliday (1994: xiv) explains its inappropriateness for SFL theory as follows:

[2] As previously explained, since the inception of Systemic Grammar, Halliday has only had one position on stratification of language.  The notion of a second position arises from a (motivated) misunderstanding on Fawcett's part: confusing meaning potential (language as system) with semantics (meaning as stratum).

[3] As previously demonstrated, the Cardiff model is invalidated by its own internal inconsistencies, arising from misunderstandings of axis, delicacy, stratification and instantiation.

[4] This is misleading.  On the one hand, Fawcett uses his misunderstanding of 'meaning potential' to misrepresent lexicogrammar as a level of meaning, rather than wording.  On the other hand, Fawcett confuses the notion of a grammar 'pushed in the direction of semantics' with semantics.  Halliday & Matthiessen (2014: 39) explain:

Being a ‘functional grammar’ means that priority is given to the view ‘from above’; that is, grammar is seen as a resource for making meaning — it is a semanticky kind of grammar. But the focus of attention is still on the grammar itself. Giving priority to the view ‘from above’ means that the organising principle adopted is that of system: the grammar is seen as a network of interrelated meaningful choices.

Not Addressing The Theoretical Motivation For The Stratification Of The Content Plane

Fawcett (2010: 57n):

In my view the addition of this new level of meaning is an unnecessary complication to the theory. It has the considerable disadvantage that it requires a whole new level of system networks, which together must cover the same broad range of types of meaning as the existing ones. We need to be absolutely sure that this very large new level of system networks really is needed, before we commit ourselves to a vast amount of new work, the result of which will be to complicate even further what is already a very rich and complex model of language. I believe that the evidence is that this vast extension of the theory is neither desirable nor necessary — so long as we actually carry out the implications of Halliday's proposal that the system networks of TRANSITIVITY, MOOD, THEME and so on should model choices between meanings. I consider that the phenomena that have led Halliday to adopt his latest position are to be explained in other ways (one being to further semanticise some of Halliday's networks, e.g., that for MOOD). See Fawcett (forthcoming a) for a set of such semantic system networks.

Blogger Comments:

[1] Here Fawcett studiously ignores the reasons given for the stratification of the content plane in the work he has just cited, Halliday & Matthiessen (1999: 237):

Of course, what we are recognising here as two distinct constructions, the semantic and the grammatical, never had or could have had any existence the one prior to the other; they are our analytic representation of the overall semioticising of experience — how experience is construed into meaning. If the congruent form had been the only form of construal, we would probably not have needed to think of semantics and grammar as two separate strata: they would be merely two facets of the content plane, interpreted on the one hand as function and on the other as form.

[2] As previously explained, this level of meaning is not new, and the "old level of meaning", lexicogrammar, is wording, not meaning.

[3] These clauses project bare assertions, unsupported by reasoned argument.

[4]  This promised work was "forthcoming" in 2000, and is still "forthcoming" in 2018.

Misrepresenting Halliday And Misunderstanding Stratification

Fawcett (2010: 56-7):

Thus Halliday himself did not take on the task of a thorough re-working of the existing systemic descriptions that the revolutionary new model logically called for. However, the few new networks that emerged from that period such as those for 'modality' (Halliday 1970/76a) are more clearly oriented to meaning than most of the 1964 networks reproduced in Halliday (1976). The position remains that some of Halliday's networks (e.g., TRANSITIVITY, the network for generating Participant Roles, etc.) have been pushed very much further towards the semantics than others (e.g., the MOOD network). 

In recent years, however, Halliday appears to have reached the decision that it really is necessary to add a second and higher level of 'meaning'. This is the position that is expressed in Halliday (1996), Matthiessen (1995) and in Halliday & Matthiessen (1999) — the latter being the exploration of a possible 'experiential semantics'. As a consequence of this decision, Halliday now uses the term "semantics" for this new level of 'meaning' — while continuing, however, to describe the system networks of TRANSITIVITY, MOOD, THEME and so on as "meaning potential" (e.g., Halliday 1993:4505).

Blogger Comments:

[1] To be clear, the task that Fawcett thinks was "logically" called for was the reconstrual of Halliday's grammatical systems as semantic systems; see below. Here yet again Fawcett employs the logical fallacy of proof by assertion: repeating a false claim as if multiple repetitions of the claim have established it as valid. 

[2] To be clear, Halliday's system of modality is a grammatical system, and, like all grammatical systems, including TRANSITIVITY and MOOD, it was theorised from the viewpoint of semantics, that is: in terms of the meanings it realises.

[3] This misunderstands the SFL stratification hierarchy.  Semantics is not "a new added second and higher level of meaning".  On the one hand, it is misleading to claim that semantics is a "new added" level, since it was theorised as a level in, for example, Halliday & Hasan (1976).  On the other hand, on the SFL model, there is only one level of meaning, semantics, and this is distinguished from the level of wording, lexicogrammar, with these two constituting the two levels of the content plane.

[4] As previously explained, Fawcett continually mistakes 'meaning potential' (language as system) for 'meaning' as a level of symbolic abstraction (semantic stratum).

On The Need To Incorporate Martin's Discourse Semantics Into Semanticised Clause Systems

Fawcett (2010: 56):

As will perhaps be obvious, the position taken here is that the semanticisation of the system networks for TRANSITIVITY, MOOD, THEME and so on that constitute the meaning potential should incorporate many of the types of meaning covered in Martin's 'discourse semantics', so enabling the overall model of language to remain as it is in Figure 4 of Chapter 3, rather than becoming increasingly complex.

Blogger Comments:

[1] To be clear, for the semantics of TRANSITIVITY, see Chapter 4 'Figures' in Halliday & Matthiessen (1999: 128-76), and the semantic system realised by MOOD is SPEECH FUNCTION.   This distinction between semantics and lexicogrammar provides the means of modelling grammatical metaphor systematically.  For the semantics of the textual metafunction, see Halliday & Matthiessen (1999: 398-414).

[2] Here Fawcett again misconstrues 'meaning potential' (language as system) as the stratum of meaning (semantics) — a misunderstanding that supports his position.

[3] As demonstrated in great detail here, Martin's discourse semantics is based on misunderstandings of SFL theory at all scales and is wholly inconsistent with it.  Incorporating it into the semantic counterparts of grammatical systems would therefore severely compromise the theory overall.

[4] As previously demonstrated here (and in subsequent posts), Fawcett's Figure 4 ('The main components of a systemic functional grammar') is invalidated by internal inconsistencies arising from misunderstandings of the theoretical dimensions of realisation, instantiation, delicacy and axis.

Misrepresenting Martin (1992) And Halliday & Hasan (1976) In A Footnote

Fawcett (2010: 56n):

The reason why the coverage in Cohesion in English was limited was not, of course, that the authors were unaware that other factors also contribute to the 'cohesion' of a text, but because they explicitly confined their goals in that work to covering those aspects of 'cohesion' that are not realised in structures — and Halliday takes the position that TRANSITIVITY, MOOD, THEME and so on are all meanings that are indeed realised in structures. (But see my discussion of the reasons for disagreeing with this view in Chapter 7.) Yet the fact is that TRANSITIVITY, MOOD, THEME etc. can also contribute to the 'cohesion' of a text, as Martin (1992) clearly demonstrates. Perhaps one of the reasons for the popularity of Cohesion in English is the fact that its ideas can be applied to the analysis of texts without having a full understanding of the SFL approach to understanding language. The re[s]ult is that most studies of cohesion do only part of the job. This has in turn had an unfortunate effect on work in some areas where SFL can usefully be applied, such as psychiatric linguists, where the two major areas of study are cohesion and syntax (the latter being of the 'phrase-structure' type).

Blogger Comments:

[1] This is misleading and misrepresents Halliday & Hasan (1976: 7), who explicitly define their term 'cohesion' as only referring to non-structural text-forming resources:

In other words, a text typically extends beyond the range of structural relations, as these are normally conceived of.  But texts cohere; so cohesion within a text — texture — depends on something other than structure.  There are certain specifically text-forming relations which cannot be accounted for in terms of constituent structure; they are properties of the text as such, and not of any structural unit such as a clause or sentence.  Our use of the term COHESION refers specifically to these non-structural text-forming relations.

[2] See the critique of Chapter 7 for the misunderstandings on which  Fawcett's views are based.

[3] On the one hand, this is misleading because it gives the false impression that Fawcett uncovered something unacknowledged in Halliday & Hasan (1976: 6-7):

Structure is, of course, a unifying relation.  The parts of a sentence or a clause obviously 'cohere' with each other by virtue of the structure.  Hence they also display texture; the elements of any structure have, by definition, an internal unity which ensures that they all express part of a text. … In general, any unit which is structured hangs together so as to form text.  All grammatical units — sentences, clauses, groups, words — are internally 'cohesive' simply because they are structured.

On the other hand, it confuses the general notion of 'cohesive' with the technical term 'cohesion' that refers only to non-structural text-forming relations; see [1] above.

To be clear, Halliday (1994: 334) identifies the structural and non-structural resources of the textual metafunction — theme, information and cohesion — as the creators of texture:

 

[4] This is misleading because it misrepresents Martin (1992), in as much as Martin has nothing at all to say about these grammatical systems contributing to the cohesion of a text, as demonstrated here.

[5] These are bare assertions, unsupported by evidence, and based on misunderstandings of cohesion, including those outlined above.

Misrepresenting Martin (1992)

Fawcett (2010: 56):

However, the major effort to provide system networks that could function as the 'higher semantics' in the framework of the Sydney Grammar has been Martin's development of what he calls a "discourse semantics", as described in his valuable English Text (Martin 1992). Martin's theory of language constitutes a separate 'sub-dialect' of the Sydney Grammar that is significantly different from that of Hasan and indeed Halliday — but he too, like Hasan and Cloran, works on the assumption that his 'discourse semantic' system networks will be 'realised' by a grammar such as that of Halliday (1994) and Matthiessen (1995). However, even Martin's 620-page work does not provide a full coverage of the proposed higher level of system networks. This is understandable, both because of the size of the task and because the focus of English Text is not on the grammar as a whole, but on providing a much more complete coverage of the topic of 'cohesion' than the limited coverage in Halliday & Hasan's Cohesion in English (1976).

Blogger Comments:

[1] As demonstrated here in meticulous detail, Martin (1992) is based on fatal misunderstandings of SFL theory across all scales.

[2] To be clear, the commonly used geographical description of Martin's misunderstandings of SFL theory is 'The Sydney School'.

[3] This is true.  Martin (1992) is not only 'significantly different' from Halliday's theory, but also, more importantly, wholly inconsistent with it, as demonstrated here.

[4] This is misleading.  Martin's networks provide no realisation statements that specify how discourse semantic features are realised in lexicogrammatical systems (evidence here).  Moreover, Martin (1992) misunderstands strata as modules (evidence here), and demonstrates that he does not understand the theoretical term 'realisation' (evidence here).

[5] This is an empty claim in the absence of a means of determining what would constitute "full coverage".

[6] This misunderstands both SFL theory and Martin (1992).  In SFL theory, cohesion is a non-structural resource of the textual metafunction at the level of lexicogrammar.  As demonstrated here, Martin (1992) misunderstands these systems, rebrands them, distributes them across different metafunctions, and relocates them to the level of (discourse) semantics.

[7] This is misleading in as much as it falsely implies that Martin (1992) complements the description of Halliday & Hasan (1976) in a theoretically consistent way.

Misrepresenting Halliday (1984)

Fawcett (2010: 55-6):

In just one area of meaning Halliday provided a small "semantic" network which 'preselected' options in what has always been the rather 'form-centred' MOOD network (Halliday 1984:13). This little network for the semantics of MOOD only had eight pathways through it, but Hasan & Cloran (1990) and Hasan (1992) have developed very much fuller system networks which can be regarded as expansions of it for use in describing children's language, all within the Sydney Grammar framework.*

* See Fawcett (1999:247-9, 258-9) for a discussion of some of the differences between Halliday's 'grammatical' network for MOOD and my much richer MOOD network, which is explicitly at the level of semantics. See Fawcett (forthcoming a) for the full version of the computer-implemented network for MOOD (which replaces that in Fawcett 1980:103).

Blogger Comments:

[1] This is misleading on at least four counts.  Firstly, the semantic network in Halliday (1984) is that of SPEECH FUNCTION (Figure 2):


Secondly, the semantic network is clearly distinguished from the grammatical system of MOOD (Figure 3):


Thirdly, the term 'preselected' does not appear in Halliday (1984).  Its use here serves to blur the distinction between the two networks.

Fourthly, the use of the term 'semantics of mood' and the avoidance of the term 'speech function' also serves to blur the distinction between the two networks.

The reason for these attempted deceptions is that the validity of Fawcett's model — the Cardiff Grammar — depends on the MOOD network being semantic, not grammatical.

[2] To be clear, what Fawcett refers to geographically as 'the Sydney Grammar framework' is Halliday's version of his own theory.

[3] A MOOD network at the level of semantics mistakes wording for meaning and, more importantly, is unable to account for grammatical metaphor, as when a command (speech function, semantics) is incongruently realised as an interrogative clause (mood, lexicogrammar).

[4] To be clear, Fawcett (forthcoming a) is Functional Semantics Handbook: Analyzing English at the Level of Meaning. London: Continuum, and it is still "forthcoming" 18 years after the first edition of this publicaton.

Misrepresenting Halliday (1977/78) And Halliday (1973)

Fawcett (2010: 55):

For very many years the major new component of the second possible model of language was represented in Halliday's writings solely by the use of the term "semantics" as a placeholder in his summaries of what a language is like (e.g., in Halliday 1977/78:128). His nearest related work in this period was on the development, with Bernstein and Turner, of the concept of "socio-semantic" system networks (e.g., Halliday 1973:48-102) — but these only applied in very specific contexts of situation and consequently did not constitute a generalised semantics, as I pointed out in Fawcett (1975).

Blogger Comments:

[1] This is misleading.  What Fawcett dismisses as a mere "placeholder" is elaborated by Halliday (1978: 128) as follows:

[2] This is very misleading, because the wording 'as I pointed out in Fawcett (1975)' gives the false impression that Fawcett uncovered something unacknowledged in Halliday (1973). The initial context of language use under consideration in Halliday (1973) was announced at the beginning of the article and clarified as such with regard to his first semantic network (Figure 1), and its later developments (Figures 2 & 3). Halliday (1973/2003: 324-5, 328-9):

Let us take as an example the use of language by a mother for the purpose of controlling the behaviour of a child. …
Figure 1 is a first attempt at a semantic network for this context. …
Figure 2 More generalised version of earlier network …
Figure 3 Revised version of Figure 1 …

Subsequent semantic networks for other specific contexts are similarly qualified (op. cit.: 332-3, 335, 337-41):

Figure 4 Network for move in pontoon …
Figure 5 Pontoon move network showing recursive option …
Figure 6 Revised version of Figure 5 …
Figure 7 System network for greeting …
Figure 8 Network for agency in physical threat …
Figure 9 Network for physical threat …
Figure 10 Non-physical threat …
Figure 11 Network for warning …
Figure 12 Revised network for threats and warnings …

Misrepresenting Halliday On Lexicogrammatical Systems

Fawcett (2010: 55):

In the first approach, then, the system networks of TRANSITIVITY, MOOD, THEME and so on are held to be choices in meaning, and so "the semantics" — very much as in the model proposed here. 

In the second approach, which we shall call the 'two-level model of meaning', the existing networks are held to be choices within the grammar (or the 'lexicogrammar') — albeit in a grammar that is described as having been "pushed [...] fairly far" towards the semantics, and whose networks Halliday still describes as modelling "meaning potential" (Halliday 1994:xix). These lower level choices in 'meaning' are said to be "preselected" by choices in a higher system network, which is itself the level of semantics.

Blogger Comment:

[1] As previously explained, the 'first approach' is Fawcett's misrepresentation of Halliday's grammatical networks as semantic networks.

[2] As previously explained, the 'second approach' is Halliday's actual model, wherein the said networks are grammatical systems.

[3] As previously explained, the wording 'two levels of meaning' misunderstands the principle of stratification.  To be clear, in SFL theory, the upper level of the content plane of language is meaning (semantics), and the lower level of the content plane of language is wording (lexicogrammar).  Lower level choices are choices in wording, not meaning.

[4] As previously explained, by the wording 'meaning potential', Halliday refers to language as system, not the stratum of semantics.  Halliday (1994) is titled An Introduction To Functional Grammar, and its introduction includes a discussion headed Grammar and Semantics (xix-xx), and the stratal distinctions of semantics, lexicogrammar and phonology (xiv).

(Trivially, the wording 'meaning potential' does not appear on the page cited by Fawcett: Halliday (1994: xix).)

∞

The reader is invited to consider why Fawcett repeats these misrepresentations of Halliday's model over and over and over and over, instead of just getting on with setting out his own theorising.

Misrepresenting Halliday's Writing As Indecisive And Lacking Specificity

Fawcett (2010: 54-5): 

In evaluating the lack of specificity that is sometimes found in Halliday's writings — which at times risks being interpreted as indecisiveness — it is important to understand the spirit in which he 'does his linguistics'. Essentially, he is an explorer. His typical practice is to suggest some new idea, and then to explore its possibilities in text-descriptive terms to see how far it fits in with other concepts in the theory, rather than to present the world with new 'truths'.

The idea that the system networks of TRANSITIVITY, MOOD, THEME and so on are choices between meanings — and that they are therefore essentially semantic choices — was initially just one such tentative proposal, as we saw in Section 4.3, but the warmth of its reception by many systemic linguists at the time (though not all; compare Hudson 1971) did not prevent Halliday from exploring, in parallel, an alternative approach to the representation of meaning. It is significant that his 1973 book, in which the two alternative positions on 'meaning' are discussed, was entitled Explorations in the Functions of Language. From the early 1970s onwards, then, Halliday has consistently held open the possibility of exploring two alternative models of the stratification of meaning. But the fact that each makes 'meaning' central to understanding language means that they can easily be confused and the distinctions blurred.

Blogger Comments:

[1] Here Fawcett misrepresents his own inability to understand Halliday's writing as deficiencies on Halliday's part, appraising it in terms of negative appreciation, 'lack of specificity', and negative judgement, 'indecisiveness'.

[2] Here yet again Fawcett repeats his own misrepresentation of Halliday's grammatical networks as semantic networks, as if multiple repetitions of the claim have established it as valid (the logical fallacy of proof by assertion).  To be clear, in SFL theory, grammatical networks realise semantics; that is, as networks of wording, they are construed as a lower level of symbolic abstraction than meaning.

[3] This is misleading.  To be clear, the approach that Fawcett labels as "alternative" is Halliday's only view on the matter: that the system networks of TRANSITIVITY, MOOD, THEME are grammatical systems that realise semantics, rather than semantic systems.

[4] The wording 'stratification of meaning' betrays Fawcett's misunderstanding of stratification.  It is language that is stratified, not meaning; meaning is but one stratum in this hierarchy of symbolic abstraction.

[5] As demonstrated over and over here, the confusions and the blurring of distinctions in this regard arise solely from Fawcett's inability to understand Halliday's writing.

Misrepresenting Halliday On Lexicogrammatical Systems

Fawcett (2010: 54):

Whatever the reason, the fact is that Halliday felt justified at the time in presenting the existing networks as at least a first approximation to what was needed for a representation of the meaning potential of English. Thus the existing system networks had an ambivalent status between being at the level of form (for which they had been developed) and being at the level of meaning (which they were now said to represent). 

It may have been the ambivalence of the status of the system networks of TRANSITIVITY, MOOD, THEME and so on that allowed Halliday both to think of his existing networks as 'semantic' and at the same time to explore the alternative approach to the representation of meaning to which we shall come shortly. But the point to note here is that Halliday himself never embraced fully the revolutionary change described above in Section 4.4 — despite the fact that it was his own proposal. 

Blogger Comments:

[1] See the previous post.

[2] To be clear, 'meaning potential' is Halliday's characterisation of language as system, as opposed to language as text.  It does not refer solely to the semantic stratum as a level of symbolic abstraction.

[3] This is misleading.  To be clear, Halliday's grammatical networks of TRANSITIVITY, MOOD, and THEME did not, and do not, have ambivalent status with regard to stratal location.  This is Fawcett's misunderstanding, deriving from his mistaking 'meaning potential' for the stratum of meaning; see [2] above.

[4] This is misleading.  The claim that these networks are at the level of meaning is Fawcett's, not Halliday's.

[5] This is very misleading.  To be clear, 'Halliday himself never embraced fully the revolutionary change' of relocating his networks to semantics because the fact is that it was not his proposal, merely Fawcett's misunderstanding, as explained above.  The logical fallacy deployed by Fawcett continues to be proof by assertion.

[6] Trivially, this was discussed in Section 4.3; Section 4.4 was concerned with the notion of metafunction.

Misrepresenting Halliday On Lexicogrammatical Systems

Fawcett (2010: 53-4):

So why, we may wonder, did Halliday not carry out such a programme of semanticising his system networks? We may make a number of guesses at the possible reasons. These might have included:

1. The enthusiastic welcome already given to the existing networks by new converts to systemic linguists;
2. the lack of serious criticism of the networks by his immediate colleagues — a lack that is perhaps not surprising, given that Halliday's main collaborators at the time were Hudson and Huddleston, both of whom were more 'form-centred' than Halliday himself (as they have continued to be); 
3. Halliday's preoccupation in that period with various other aspects of the burgeoning work, both in the theory and its applications in many fields, to many of which he contributed personally; 
4. the concern that the features in the new networks would be so much further removed from their realisations at the level of form that the new realisation statements (to use Halliday's term) would be very hard to write; 
5. sheer lack of time to undertake this task, given its size and his other commitments.

Blogger Comments:

As previously explained, the system networks in question were and are grammatical networks, and that is the reason why Halliday didn't "semanticise" them.  This is another instance of the logical fallacy of proof by assertion.

Misrepresenting Halliday And Matthiessen On Lexicogrammatical Systems

Fawcett (2010: 53): 

We saw in Section 4.3 that in the mid-1960s Halliday changed the theoretical status of the system networks, so that they came to be seen as modelling choices between meanings. And yet a close study of the system networks in Matthiessen (1995) shows that many are essentially the same as Halliday's 1964 networks (as published in Halliday 1976). In other words, system networks that Halliday had originally developed on the assumption that they were at the level of form were re-interpreted as being at the level of meaning. Is it really possible, one has to ask, that networks that were developed for one level of representation should be able to be transported, unchanged, to function at another level of representation? After major theoretical changes of the sort described above, the next logical steps should surely be a critical reexamination of the existing networks to discover where they were and where they were not already sufficiently 'semanticised', followed by the careful semanticisation of those parts that needed it, in order to turn a brilliant insight into a practical reality.

Blogger Comments:

[1] This is misleading, because it is manifestly untrue.  As demonstrated in the critiques of Section 4.3, Halliday did not reinterpret his grammatical networks as semantic networks.  This was merely a misinterpretation of Fawcett and his colleagues.  The continual repetitions of this falsehood are instances of the logical fallacy known as 'proof by assertion'.

[2] This is misleading, because it is manifestly untrue.  The networks in Matthiessen's Lexicogrammatical Cartography (1995) are grammatical networks, and so, not re-interpretations of Halliday's networks as semantic networks.  See also (the title of) the immediately preceding post.

[3] Here Fawcett is chiding Halliday for not acting according to Fawcett's misunderstanding.