What Fawcett Means By 'Grammar'

Fawcett (2010: 37):

At this point I should clarify the sense in which I am using the terms "grammar" and "meaning". Let us take "grammar" first, since it is used in the caption for Figure 4. In "Categories", "grammar" was the name of a subcomponent of the level of form, but here its meaning has been extended in two ways. The most obvious is that a grammar now includes a level of meaning as well as a level of form. Thus a 'grammar' is essentially a model of the sentence-generating component of a full model of language and its use. The second extension — which is less obvious — is that the term "grammar" is regularly used as a short form for "lexicogrammar".

Blogger Comments:

[1] Contrary to the implication, this "extension" of the meaning of 'grammar' to 'lexicogrammar' is not Fawcett's.  It appears in Categories of the Theory of Grammar (Halliday 1961), where lexis is conceived as most delicate grammar.  Halliday (2004 [1961]: 54):

The theoretical place of the move from grammar to lexis is therefore not a feature of rank but one of delicacy.  It is defined theoretically as the place where increase in delicacy yields no further systems;

[2] To be clear, Fawcett stratifies grammar into meaning and form, whereas Halliday locates meaning in semantics, and distinguishes between grammatical functions and grammatical forms (the rank scale).  Crucially, it is the distinction between semantics and lexicogrammar that allows Halliday to account for grammatical metaphor in his model.

Fawcett		Halliday
(lexico)grammar	meaning	meaning	semantics
		function	lexicogrammar
	form	form

Fawcett's Introduction Of The Term 'Instance'

Fawcett (2010: 36n):

In the past SF linguists have usually referred to instances as "instantiations". But since we need the term "instantiation" to refer to the relationship between the 'potential' and the 'instances' that it generates, as here, it is preferable to use a different term for the 'product' of the 'process' of instantiation. Hence my introduction here of the term "instance". Note that some SF linguists (e.g., Matthiessen & Bateman 1991) have used the term "actualisation" in place of "instantiation". One reason for preferring "instantiation"' to "actualisation" is the fact that English has the corresponding noun "instance", which can be used to distinguish the 'product' from the 'process'. It is this pair of terms, then, that is used in the present theory.

Blogger Comments:

This is misleading.  The term 'instance' was "introduced" in Categories Of The Theory Of Grammar (Halliday 1961: passim), and many subsequent publications, such as Halliday (1978 [1974]: 40), Halliday (1977: 54) and Halliday (1987:152).

On 'The Main Components Of A Systemic Functional Grammar'

Fawcett (2010: 36):

The diagram in Figure 4 brings together these two pairs of concepts to define the four components that are essential for modelling any semiotic system. It provides a framework for thinking about language in terms of (1) the potential and (2) the many possible instances of that potential, and to do so at the two levels of (a) meaning and (b) form. The four components of the model are defined by the intersections of these two pairs of concepts. 

Blogger Comments:

On the one hand, Figure 4 presents a flow chart for text generation by computer, not a systemic functional model of grammar.  It is also inconsistent with SFL theory in that it presents the architecture of the theory in terms of interactions between modules instead of relations along dimensions; see Halliday & Webster (2009: 231).

On the other hand, Figure 4 is internally inconsistent.  Firstly, the 'potential' column presents realisation rules as the form that realises the system of meaning.  To be consistent with the level of meaning, the level of form would need to be a 'system network of choices in form'.  To be consistent with SFL theory, the realisation rules would need to be located in the system of meaning.

Secondly, the 'instance' column presents structure as the form that realises features of meaning.  That is, it presents the axial relation between paradigm (features) and syntagm (structure) as a stratal relation between meaning and form.

Thirdly, the 'form' row presents structure as an instance of realisation rules as potential.  That is, it presents the axial relation between paradigm (realisation rules) and syntagm (structure) as a relation of instantiation between potential and instance.

Any theoretical advantages afforded by such inconsistencies need to be demonstrated, and their value weighed against their consequences for the theoretical architecture as a whole.

Multiple Misrepresentations Of Halliday In A Footnote

Fawcett (2010: 35n):

Halliday often writes in a way that implies that the reason why human languages are more complex than many other semiotic systems is that grammar 'intervenes' between meanings and phonological and graphological forms (e.g., Halliday 1996:29), whereas it does not in simpler systems (such as a traffic light system). One difficulty about this view is that he also sees the grammar as having its own 'meaning potential' inside it, and this gives us a model in which the 'meanings' of the higher semantic level are detached, as it were, from their realisations in forms, such that they have to be mapped first onto the meanings within the grammar (or 'lexicogrammar', as Halliday prefers to call it). I take a different view, as this book shows, in that I regard the level of meanings within the 'lexicogrammar' as the key level of linguistically-realised meaning, such that it is realised in any one of (1) syntax, (2) intonation or punctuation (depending on the medium of discourse) and (3) items. (Notice that it is only one of these, namely items, to which the supposed 'third level' of segmental phonology/graphology is relevant, and this fact raises serious questions about its traditional status as a 'level of language'.) In my approach, therefore, any 'higher' meaning — such as those that Halliday includes under the umbrella concept of 'grammatical metaphor' — is to be accounted for by one or other of several different concepts. See Fawcett (1993) for a brief indication of some of these, and Fawcett & Huang (in preparation) for a much fuller picture. Many of the concepts introduced in this footnote will be amplified in later parts of this book.

Blogger Comments:

[1] Halliday does not imply this, he is quite explicit about it. Halliday & Matthiessen (2014: 25):

The ‘content’ expands into two, a lexicogrammar and a semantics (cf. Halliday, 1984a; Halliday & Matthiessen, 1999). This is what allows the meaning potential of a language to expand, more or less indefinitely.

The stratification of content makes grammatical metaphor possible, and it this, chiefly, that 'allows the meaning potential of a language to expand, more or less indefinitely'.  This is why the inability of Fawcett's model to account for grammatical metaphor is such a serious shortcoming.

[2] This is a bizarre misunderstanding.  For Halliday, the system of language is meaning potential — the potential to make meaning (semogenesis).   Halliday & Matthiessen (2014: 48, 76, 601):

Language has evolved as a fully systemic semiotic system: it is possible to posit and describe the overall meaning potential for a given language, interpreting this meaning potential as an aggregate of registerial subpotentials. … 

The functional categories provide an interpretation of grammatical structure in terms of the overall meaning potential of the language. … 

Logogenesis pertains to the entire meaning potential of a language – all the strata and all the metafunctions.

[3] This misunderstands stratification.  Meanings are modelled as a higher level of abstraction than the forms of wording.  There are no "meanings within the grammar"; the wordings of grammar realise the meanings of semantics.

[4] Halliday "prefers" the term lexicogrammar because the stratum encompasses both lexis and grammar, related systemically along the scale of delicacy.

[5] This is not a different view to Halliday's; it is merely different from Fawcett's misunderstanding of Halliday.  Halliday sees lexicogrammar as the dynamo of language, and as the source of the type of meanings that differentiate language from other (bi-stratal) semiotic systems.  Halliday & Matthiessen (2014: 22):

Grammar is the central processing unit of language, the powerhouse where meanings are created …

[6] This misrepresents the SFL hierarchy of stratification, which distinguishes meaning (semantics) from wording (lexicogrammar) as different levels of symbolic abstraction.  Grammatical functions (wordings) are congruent (agree) with semantic functions (meanings) in the absence of grammatical metaphor.

[7] This blurs the distinction between two distinct realisation relations in SFL theory.  On the one hand, grammatical functions are realised in the grammatical forms of the rank below; e.g. clause functions are realised by groups ± phrases.  On the other hand, the realisation of grammar in phonology or graphology is an interstratal relation.

[8] This confuses realisation with delicacy.  In SFL theory, lexical items are the synthesis of the most delicate features of the systems of lexicogrammar.  They do not "realise" grammatical functions because lexis and grammar are of the same level of symbolic abstraction (wording).

[9] Here Fawcett presents his own misunderstanding of the place of lexis in SFL theory as a challenge to modelling some of the systems of the expression plane as a lower level of symbolic abstraction than lexicogrammar.

[10] This misrepresents Halliday's notion grammatical metaphor, which is not 'higher meaning', but an incongruent realisation of meaning (semantics) in wording (lexicogrammar).  Grammatical metaphor is an integral part of semogenesis, and the inability of Fawcett's model to account for it is a very serious shortcoming indeed.

[11] The promised concepts to account for grammatical metaphor are said to be in:

Fawcett, Robin, 1993. "Language as program: a reassessment of the nature of descriptive linguistics". Language Sciences 14.4. 623-57. 

Fawcett, Robin, & Huang, Guowen, in preparation. Explaining Enhanced Theme (the 'Cleft' Construction): a Test Case for Systemic Functional Linguistics - and Every Theory of Language. London: Continuum. 

The latter publication is still, apparently, "in preparation" in 2017, but there is:

Huang, G.-W. (2003). Enhanced theme in English: Its structures and functions. Taiyuan: Shangxi Education Publishing House.

However, the abstract of Fawcett (1993), at least, makes no mention of grammatical metaphor:

The purpose of this paper is to challenge one of the basic tenets of linguistics. The traditional view is that it is the business of linguists to describe language and languages—and/or to theorise about it and them. It is essentially a view of language as an object. While I remain solidly based in linguistics, I draw here on my experience in natural language processing (i.e. of putting English in computers as part of a model of a communicating mind) as I advocate the view that it is more insightful to view language not as object, but as procedure, as process, as program. Note that ‘program’ has no final ‘me’ to signify the sense of ‘computer program’. 

I invite you to come with me on an exploration of what is involved when someone produces (‘generates’) a sentence, to show why I advocate a model of language that is not static, but dynamic. The model is also, I shall argue, essentially generative (in the sense of ‘productive’) rather than interpretive. Thus a program whose essential design is as it is to enable it to run in one direction—i.e. to turn meanings into sounds—is consulted as the person trying to understand a text attempts the converse task. This is not just ‘to turn sound into meaning’, but to turn the performer's sounds back into the performer's meanings. Hence the logical priority of generation over interpretation. 

In this new view of language as program, it is no longer possible to draw a clear line between the material of linguistics and the algorithms that determine how we choose between linguistic options; one needs a more complete framework, which is inevitably also more complex, e.g. as described here.

[12] If any of the "concepts introduced in this footnote" are "amplified in later parts of this book", each will be noted on this blog, and cross-referenced to this post.

Misunderstanding Interstratal Realisation And Confusing It With Instantiation

Fawcett (2010: 34-5):

 

The relationship between the set of meanings and the set of forms is one of realisation. Thus the arrow pointing downwards shows that the meanings of a language are 'realised by' linguistic forms, and the arrow pointing upwards shows that forms 'realise' meanings. 

However, contrary to what some formal language theorists think, the two processes are not the reverses of each other. This is because the problems faced by someone who is trying to produce a text that will be effective and appropriate to a particular point in discourse are not the same as the problems that face someone who is trying to work out the meaning of an incoming text. (It is only in a very impoverished, form-based view of language that grammars may appear to be 'reversible'.) This principle applies both to abstract psychological models of how humans process incoming and outgoing language texts and to computer models of language processing (1993 and 1994a, Weerasinghe & Fawcett 1993). Thus, while the fuller model of language that we shall develop in the next two sections may appear to be oriented more to the 'production' of text than to analysing it, the term "realisation" is in fact directionally neutral. 

It is also too general to be useful, so that we need to build a picture of the steps by which this model turns 'meanings' into 'forms'. For this we need to move on to a second pair of concepts.

Blogger Comments:

[1] This confuses the realisation relation between strata with the instantiation relation between potential (system) and instance (text).  Halliday & Matthiessen (2014: 51):

But ‘text’ is a complex notion. In the form that 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] This again misunderstands the relation between strata in SFL theory.  The relation is an intensive identifying relation of 'symbol' ('realise'), not one of 'time phase' ('turn into').

[3] The second pair of concepts that Fawcett introduces, system and instance, does not bear on the identifying relation between meaning and form.  The source of Fawcett's confusion (encapsulated in his Figure 4) will be identified in the post of 1 October 2017.

Confusing Realisation And Instantiation

Fawcett (2010: 34):

In other words, just as an individual sign (such as a red 'stop') sign has both a form and a meaning, so too a sign system such as a natural human language has the two levels of form and meaning. Figure 3 can therefore be seen as a very simple model of a language as a whole.

The two levels of 'meanings' and 'forms' in Figure 3 are linked by two arrows. These indicate that a sign system is not a static object, but a device for turning meanings into forms, and forms into meanings. I have argued that a language is best thought of as a 'procedure' (Fawcett 1980:54f.) or, in computing terms, as a 'program' (Fawcett 1993).

Blogger Comments:

The general confusion here is between the realisation relation between strata and the process of instantiation.  (This will be clarified further in the discussion of Figure 4 on 1 October 2017.)

[1] A sign system is not "a device for turning meanings into forms, and forms into meanings" because meanings are not turned into forms, and forms are not turned into meanings.  Meaning and form are different views, at different levels of abstraction, on the one phenomenon.  The type of intensive identity relation between them is one of 'symbol' (Halliday & Matthiessen 2014: 269), whereas 'turn into' signifies an intensive identity relation of 'time phase' (ibid.)

On the other hand, as vectors, the arrows could be understood as representing the two directions of coding between the two levels of abstraction; see Halliday & Matthiessen 2014: 280).  That is, the symbolic identity encodes meanings by reference to forms, and decodes forms by reference to meanings.  But this is not Fawcett's understanding.

[2] To be clear, the notion of language as a procedure rather than a "static object" is modelled in SFL as the process of instantiation, the selection of features and the activation of realisation statements during logogenesis (the unfolding of text at the instance pole of the cline of instantiation).  The term for the potential pole of the cline, 'system', is short for 'system–&–process' (Halliday & Matthiessen 1999: 507).

Locating Both Expression And Content Within One Level Of Form

Fawcett (2010: 34n):

In the current model, however, 'segmental phonology' is regarded as being the internal specification of items, i.e., as a phenomenon that occurs within the level of form (rather than being a 'realisation' of (part of) it). See (Fawcett 1980:59) for a fuller statement of this position.

Blogger Comments:

To be clear, Fawcett's model locates both expression (phonology) and content (syntax) "within the level of form". Without the realisation relation between the two, this is equivalent to locating both the phonemes that realise the word 'stop', and the word 'stop', at the same level of symbolic abstraction.

Moreover, it is only one of the four ranks of phonology, phoneme, that is modelled this way.  The other three ranks, tone group, foot and syllable are not included.  This creates an internal inconsistency within this theoretical inconsistency.

On The Interpretation Of Signifier And Signified As Form And Meaning

Fawcett (2010: 33):

Saussure's central concept is the 'linguistic sign'. As is well known, he emphasises that a 'sign' consists of a 'signifier' and a 'signified', and we shall interpret this as saying that a sign has both a form and a meaning. And to this we can add the important concept — which is long familiar to functionally oriented linguists — that form and meaning are mutually defining. 

Consider, as an analogy, the type of simple traffic control system that has just two 'forms' of display — a red disk and a green disk. It also has just two 'meanings' — which are, in crude terms, 'stop' and 'go'. A human language is of course very much more complex than this in both its forms and its meanings — and indeed in the relationships between the two. But it is nonetheless helpful to recognise that the two primary levels of a language — as in any semiotic system — are those of meaning and form.

Blogger Comments:

[1] To be clear, Saussure's distinction between signifier and signified is the general distinction between lower and higher levels of symbolic abstraction, as between:

• expression and content
• form and function
• wording and meaning
• sounding and wording.

[2] To be clear, the relation between two levels of symbolic abstraction is realisation, which is a relation of intensive identity between a Token and a Value.  In terms of coding, the identity encodes a Value by reference to a Token, and decodes a Token by reference to a Value.  For example, the identity of expression and content encodes content (e.g. 'go') by reference expression (e.g. green traffic light), and decodes expression (green) by reference to content ('go').  This is the sense in which they are 'mutually defining'.

An Unsurprising Finding

Fawcett (2010: 32):

Surprisingly, perhaps, we shall find that the concepts of "Categories" appear to be more relevant to the theory of syntax found in the Cardiff Grammar than they are to Halliday's presentations of his own version of SF grammar.

Blogger Comments:

This unwittingly acknowledges the degree to which Halliday's theorising continued to evolve after his 1961 publication, in comparison with Fawcett's understanding and use of Halliday's ideas.

Misrepresenting Halliday On Hypotaxis [2]

Fawcett (2010: 30):

There is a further problem about the proposed relationship of 'hypotaxis'. This is the question of what it actually means to say that the relationship is one of 'dependency without embedding'. The answer lies in Halliday's use of the terms 'modifier' and 'head' (IFG p. 217) to describe the relationship. Essentially, the relationship of 'hypotaxis' is the same the traditional 'modifier-head' relationship in a unit. The only differences are that Halliday narrows the definition, such that (1) each element must be filled by the same unit, and (2) the relationship between each element and the sister elements on either side of it is always the same. (And neither of these is the case, it can be argued, for the relationships between the modifiers and the head in the English nominal group, for which see Section 10.2.5 of Chapter 10.) Thus, despite this narrowing of the definition, 'hypotaxis' is still a relationship between sister elements — and this, you will recall, is essentially what a 'multivariate' structure is. So the distinction between 'multivariate' and 'univariate' structures is not in fact very clear.

Blogger Comments:

[1] To be clear, dependency is a relation between units of the same rank, whereas embedding is the functioning of a higher rank unit in the structure of a lower rank unit.  That is, dependency and embedding are mutually exclusive.

[2] This is misleading because it is not true.  Hypotaxis is not essentially the same as the traditional Modifier–Head relationship.  The Modifier–Head relationship in the nominal group additionally includes the relationship of subcategorisation.  Halliday & Matthiessen (2014: 388-9):

We now need to consider the structure of the nominal group from a different, and complementary, point of view; seeing it as a logical structure. This does not mean interpreting it in terms of formal logic; it means seeing how it represents the generalised logical-semantic relations that are encoded in natural language. These will be discussed in greater detail in Chapter 7; for the purposes of the nominal group we need to take account of just one such relationship, that of subcategorisation: ‘a is a subset of x’. This has usually been referred to in the grammar of the nominal group as modification, so we will retain this more familiar term here.

[3] This is misleading because it is not true.  The Modifier-Head relationship in the nominal group is not one in which 'each element must be filled by the same unit'.  Like all univariate structures, it is generated by the iteration of the same functional relationship.  Halliday & Matthiessen (2014: 390):

We refer to this kind of structure as a univariate structure, one which is generated as an iteration of the same functional relationship (cf. Halliday, 1965, 1979): α is modified by β, which is modified by γ, which is ... .

[4] This is only half true.  It is true that the Modifier-Head relationship in the nominal group is the iteration of the same functional relationship, subcategorisation, but the subcategorisations themselves vary according to the logico-semantic relations of expansion (elaboration, extension, enhancement) and projection.  See Halliday & Matthiessen (1999: 183).

[5] This is misleading because it is not true.  In the promised discussion, Fawcett misconstrues the Epithet/Head of the nominal group very bright as Thing.  This misunderstanding will be explained in more detail in the critique of section 10.2.5.

[6] This is misleading because it is not true.  A "relationship between sister elements" is not "essentially what a multivariate structure is"; it describes structure in general, whether multivariate or univariate, and within univariate, whether hypotactic or paratactic.

[7]  This is misleading because it is not true.  The distinction between multivariate and univariate structures is very clear and simple.  Continuing from the quote in [3] above, Halliday & Matthiessen (2014: 390) explain:

By contrast, the type of structure exemplified by Deictic + Numerative + Epithet + Classifier + Thing we call a multivariate structure: a configuration of elements each having a distinct function with respect to the whole.

Fawcett's Argument Against Verbal Group Complexes

Fawcett (2010: 30n):

For a demonstration that the concept of the 'hypotactic verbal group complex' is not needed, and for an account of how the Cardiff Grammar provides for the phenomena for which Halliday sets up the concept of the 'verbal group complex', see Fawcett (forthcoming c). That paper also shows why the concept of a 'paratactic verbal group complex' is not needed, as well as showing, incidentally, how most types of text-sentence that Halliday would analyse as showing 'hypotactic' relations between clauses can be handled elegantly as embedded clauses.

Blogger Comments:

[1] The question of whether or not the concepts of 'hypotactic verbal group complex' and 'paratactic verbal group complex' are needed in SFL theory is decided on many interrelated factors, including on the basis of their explanatory power, and how they fit within the overall architecture of the theory — as well as the consequences, for the theory, including its self-consistency, if they are dispensed with.

[2] Fawcett's promised paper In Place Of Halliday's Verbal Group Complex was 'forthcoming' in the first edition of this work (2000: 342), still 'forthcoming' a decade later in the second edition (2010: 30n), and still unpublished another seven years later at the end of July 2017.

[3] This is another bare assertion.  The promised argumentation is still unpublished, seventeen years after the original declaration.

With regard to 'elegance' in this context:

In the philosophy of science, there are two concepts referring to two aspects of simplicity. Elegance (syntactic simplicity) means the number and complexity of hypotheses. Parsimony (ontological simplicity) is the number and complexity of things postulated.

Bare Assertion Without Reasoned Argumentation

Fawcett (2010: 29-30):

In pages 35-39 of Halliday (1965/81), Halliday discusses the complexities of alternative interpretations of Example (a) and similar examples, with the intention of showing that 'hypotaxis' represents the structural relationships more clearly than an analysis with embedding would. However, that discussion does not throw up any problems that are not equally well represented in the framework proposed in Part 2. Indeed, in many cases such relations are better represented by embedding. Thus the analysis of If you'd telephoned before I left, I'd have come would simply show if you'd telephoned before I left as a thematised Adjunct. And examples such as If before I left you 'd telephoned, I'd have come can be handled equally straightforwardly (though they would cause discontinuity in Halliday's model). See Section 11.9 of Chapter 11 for my analyses of the set of examples for which Halliday uses 'hypotaxis', and see Fawcett (in press) for the full presentation of this alternative approach.

Blogger Comments:

[1] Fawcett's report of Halliday's intention is inconsistent with Halliday's epistemological position.  The issue is not which analysis "represents the structural relationships more clearly", but which analysis is the more functional; which analysis has the more explanatory power.

[2] An alternative analysis that creates "discontinuities" in the theory is not "equal" to one that doesn't.

[3] This is merely a bare assertion, made without any supporting evidence.

[4] Section 11.9 of Chapter 11 provides no arguments in support of Fawcett's alternative analyses.

Not Demonstrating The Greater Utility Of An Alternative Analysis

Fawcett (2010: 29):

In the light of the discussion of Example (b), we can see that the difference between Examples (a) and (b) is the difference between a Complement and an Adjunct. To put it in explicitly SF terms, it is the difference between a Participant Role (a role that is 'expected' by the Process expressed in the Main Verb), and a Circumstance (a role that is not). In Example (a), therefore, the Cardiff Grammar would model before I left as a clause that fills an Adjunct in the clause if you'd telephoned before I left, and this longer clause would be shown as filling an Adjunct in the clause I'd have come if you'd telephoned before I left.

Blogger Comments:

Example (a) is Halliday's example: I'd have come if you'd telephoned before I left.
Example (b) is Fawcett's example: I said to her that I believed that you'd come.

Here Fawcett merely presents his alternative analysis in which clause complexes are construed as single clauses, without providing any argument to demonstrate any greater explanatory power of his approach.

Fawcett's argument against treating these as clause complexes is applicable to one of these examples only — Example (b) — and so, does not support his analysis of Example (a).  In any case, the argument for Example (b) is invalid, because it mistakes the low incidence of certain types of instance for disconfirmation of the general potential, as explained in the previous post.

Fawcett's Argument Against Hypotaxis [2]

Fawcett (2010: 28-9):

In my view, however, examples of 'hypotaxis' such as those in Figure 2 are more insightfully analysed as cases of embedding, where a unit fills an element of the unit above. Indeed, in using Example (a) in Figure 2 to illustrate the concept of 'hypotaxis' Halliday has chosen the most favourable type of example. It is one in which it is possible to interpret the 'main' clause in each 'hypotactically' related pair as a complete clause. And it is this that enables one to think first of each such pair as two separate clauses, and then to go on to ask how they are related to each other. However, Halliday also treats clauses that report speech or thought in the same way, and his analysis of Example (b) in Figure 1 would be as shown there. And here the line of reasoning used to justify the hypotactic analysis of Example (a) is simply not possible. In other words, I said to her is clearly an uncompleted clause that is "expecting" (to borrow Firth's metaphor) another element (which we may call a Complement). And that I believed similarly expects a Complement. In the Cardiff Grammar — as in virtually all grammars other than Halliday's — these would be treated as cases of embedded clauses that fill an element of a higher clause.

Blogger Comments:

(Example (a) is Halliday's 'enhancement' example: I'd have come if you'd telephoned before I left;

example (b) is Fawcett's 'projection' example: I said to her that I believed that you'd come.)

[1] Opinion is not argument.  The text that follows does not provide a grammatical argument that supports this opinion.  Removing the distinction of hypotaxis vs embedding reduces the explanatory power of the theory, as explained in previous posts.  As Marshall McLuhan (1962) cautioned:

A point of view can be a dangerous luxury when substituted for insight and understanding.

[2] Here Fawcett suggests that Halliday has used the logical fallacy of incomplete evidence, also known as cherry picking. It will be seen below that this accusation is unfounded.  For Fawcett's use of logical fallacies in his own argumentation, see below, or click here.

[3] This is untrue; see [4] and [5] below.

[4] Here Fawcett employs the logical fallacy known as the argument from incredulity: "I cannot imagine how this could be true; therefore, it must be false."

[5] The expectation comes from the system probabilities established by previous experiences of instances.  Verbal clauses are more likely to include Verbiage or to project a locution clause.  Low probability is not disconfirmation.

[6] Here Fawcett employs two versions of the 'red herring' fallacy.  On the one hand, it can be interpreted as the logical fallacy known as the appeal to tradition (also known as argumentum ad antiquitatem, appeal to antiquity, or appeal to common practice): a conclusion supported solely because it has long been held to be true.

On the other hand, it can be interpreted as the logical fallacy known as argumentum ad populum (appeal to widespread belief, bandwagon argument, appeal to the majority, appeal to the people) – where a proposition is claimed to be true or good solely because many people believe it to be so.

Misrepresenting Halliday's "Motivation" For Hypotaxis

Fawcett (2010: 28):

Halliday's claim is that the introduction of the 'hypotactic' relationship avoids what he describes, revealingly, as "a somewhat artificial increase in 'depth' in number of layers [in the tree structure]" (Halliday 1965/81). Indeed, it seems that the desire to avoid the embedding of units within units of the same or a lower 'rank' (i.e., 'rank shift') was the major motivation in the rapid extension of the use of 'hypotactic' relations to analyse a wide range of phenomena in the structure of language (as is shown in Section 4 of Appendix C).

Blogger Comments:

[1] The claim here is that Halliday's theoretical notion of hypotaxis was motivated by his desire to avoid embedding.  The claim is disconfirmed by the presence of both embedding and hypotaxis in the theory.  As previously explained, the distinction between hypotaxis and embedding provides insights such as what distinguishes, grammatically, a projected report (hypotaxis) from a pre-projected fact (embedding), and non-defining relative clause (hypotaxis) from a defining relative clause (embedding).  The absence of the distinction in Fawcett's model is a reduction in the explanatory power of the theory.

[2] This claim of a future supporting argument is untrue, since Section 4 of Appendix C bears no relation to this discussion.  Instead, it is a critique of the SFL notion of the verbal group, the merits of which will be examined later on this blog.

Misrepresenting Halliday On Hypotaxis [1]

Fawcett (2010: 27-8):

This leaves just the concept of 'hypotaxis'. Halliday describes 'hypotaxis' as "a chain of dependencies [between units]" (Halliday 1965/81:34). In Halliday's example I'd have come if you'd telephoned before I left, the clause before I left is said to be 'dependent on' if you'd telephoned — but not to be embedded within it and so not to fill one of its elements. And the clause if you'd telephoned is similarly said to be 'dependent on' I'd have come without being embedded in it as an element. This concept of 'dependence without embedding' is shown by the use of letters from the Greek alphabet, i.e., "α β γ" etc. to represent the 'elements' of such structures. … 

In IFG, however (e.g., pp. 375-81) Halliday shows the relationship of 'hypotaxis' as a set of horizontally linked boxes below the text, each labelled "α β γ" etc., exactly as in the 'box diagram' representations of structure in Figure 7 in Chapter 7. In these diagrams the relationship looks like sister constituency, the only remaining expression of 'dependence' being the Greek letters.

Blogger Comments:

Fawcett's argument against hypotaxis here is that Halliday's use box diagrams to label elements in a complex makes the structural relations look the same as constituency.  This is not an argument against hypotaxis, for the simple reason that it confuses the meaning of hypotaxis with the way it is expressed in a diagram.  More importantly, however, it is misleading in two ways, both of which serve Fawcett's own position, as will be explained.

Firstly, in IFG (1994: 223-4), at the beginning of the discussion of parataxis and hypotaxis, Halliday presents two very different diagrams that contrast the dependency structure of a clause complex (Figure 7-5) with its constituency structure (Figure 7-6), and then presents another different diagram (Figure 7-7) that combines both principles, showing how the two relations differ.  He (ibid.) also provides the same constituency and dependency relations simply as notations:

This can be represented as at the foot of the tree:

α ^ ββ1 ^ ββ2α ^ ββ2b1 ^ βαβ ^ bαα

or, using brackets (and showing type of interdependency), as:

α ^ " β (x β (1 ^ + 2 (α ^ " β (1 ^ + 2))) ^ α (x β ^ α))

The notation that is used here expresses both constituency and dependency at the same time: constituency by bracketing (using either brackets or repeated symbols), dependency by the letters of the Greek alphabet.

Secondly, the constituency relations here are of the clause complex, not the clause.  Fawcett's argument is that dependent clauses are "more insightfully" modelled as constituents of the clause.  That is to say, interpreting interdependency in a clause complex as constituency is not an argument that supports the interpretation of dependent clauses as embedded constituents of a clause.

Misconstruing Parataxis As Coordination

Fawcett (2010: 27):

A 'univariate' structure, then, is said to be a 'structure' that is composed of two or more categories that are the same.  More specifically, the 'taxis' is between two or more classes of unit — predominantly between clauses, but also between groups, words and even morphemes. As we shall see when we meet the concept of 'co-ordination' in Section 11.8.2 of Chapter 11, the Cardiff Grammar models a 'paratactic' relationship between two or more units, as in cats, dogs and horses, as three co-ordinated units that jointly fill an element of structure in a higher unit in the tree. There are a number of problems to consider in relation to co-ordination, but we shall delay the discussion of these to Section 11.8.2 of Chapter 11. We can therefore set aside the concept of 'parataxis' (as we have already done with 'multivariate' structures).

Blogger Comments:

[1] This is misleading.  To be clear, a univariate structure is 'an iteration of the same functional relationship' and is 'unique to the logical mode of meaning' (Halliday & Matthiessen 2014: 451).

[2] This is untrue. Tactic relations obtain between units of the same class and rank — e.g. between nominal groups — not between different classes of unit — e.g. not between nominal and verbal groups.

[3] This confuses co-ordination — paratactic extension — with parataxis in general.  Parataxis also combines with the other logico-semantic types: elaboration, enhancement and projection.  These have not been considered at all in this discussion of parataxis.

[4] This is misleading in what it implies.  In SFL theory, not just the Cardiff Grammar, a paratactically extending group complex like cats, dogs and horses serves as an element of structure of a higher unit, the clause.  For example, in the clause they have cats, dogs and horses, the nominal group complex serves as the possessed Attribute of clause structure.

Fawcett's Argument Against Hypotaxis [1]

Fawcett (2010: 27):

It is within the category of 'univariate' structures that a new pair of concepts are to be found. The question is whether their introduction leads to better descriptions of languages, or whether they cause difficulties. I shall argue that one of them — the concept of 'hypotaxis' — is a source of problems in three ways. Firstly, it is not in fact as easy as one at first thinks to work out what it actually means in structural terms to say that one unit is hypotactically related to another. Secondly, Halliday has given it great prominence in his theory, and this prominence has been at the expense of another concept — that of 'embedding'. The concept of embedding is of course tied to the general concept of 'constituency', but it is not necessarily linked to formal grammars, and its use in SFL leads to the insightful analysis of long and complex texts. The third problem with 'hypotaxis' is related to this: in practical terms, it has often led to analyses of text-sentences that are plainly counter-intuitive (e.g. treating He said as a clause in He said that he would be there).

Blogger Comment:

[1] Fawcett's first argument here against the concept 'hypotaxis' is a version of the logical fallacy known as the argument from personal incredulity which might be termed an argument from personal incapacity: because I can't apply the concept, there must be a problem with the concept.  Applied to the field of physics, this could become: because I can't apply the notion of wave–particle duality, there must be a problem with the concept.

In IFG2, Halliday (1994: 221) clarifies "what it means in structural terms to say that one unit is hypotactically related to another" by distinguishing hypotaxis from parataxis, logically, in terms of two parameters, symmetry and transitivity:

Parataxis is the linking of elements of equal status.  Both the initiating and the continuing element are free, in the sense that each could stand as a functioning whole.

Hypotaxis is the binding of elements of unequal status.  The dominant element is free, but the dependent element is not. …

In principle, the paratactic relation is logically (i) symmetrical and (ii) transitive. This can be exemplified by the 'and' relation.

(i) 'salt and pepper' implies 'pepper and salt', so the relation is symmetrical; 

(ii) 'salt and pepper', 'pepper and mustard' together imply 'salt and mustard', so the relationship is transitive.

The hypotactic relation is logically (i) non-symmetrical and (ii) non-transitive. For example, 'when':

(i) 'I breathe when I sleep' does not imply 'I sleep when I breathe'; 

(ii) 'I fret when I have to drive slowly' and 'I have to drive slowly when it's been raining' together do not imply 'I fret when it's been raining'.

[2] Fawcett's second argument here against the concept 'hypotaxis' is an untruth.  It is simply not true that the introduction of hypotaxis has been at the expense of embedding.  In SFL theory, both concepts are used, and the explanatory advantages provided by the distinction include the ability to distinguish:

• in expansion, between non-defining (hypotaxis) and defining (embedding) relative clauses, and
• in projection, between projected reports (hypotaxis) and pre-projected facts (embedding).

[3] Fawcett's third argument here against the concept 'hypotaxis' is another instance of the logical fallacy known as the argument from personal incredulity: because I can't believe it, it can't be true.  No grammatical argument is proffered in support of the intuition.

"Setting Aside" Multivariate Structure

Fawcett (2010: 26-7):

Halliday's first distinction is in fact between two broad types of structure that he calls "univariate" and "multivariate" structures. A "multivariate" structure is simply a unit that is composed of one or more of a set of different elements of structure, such as a clause or a nominal group, essentially as distinguished in "Categories" and as recognised here. The term "multi-" is intended to mark the fact that the sister "variables" (as Halliday calls the 'elements', in what seems an unnecessarily abstract terminology) are different from each other. So for our present purposes we can set the concept of 'multivariate structure' on one side, since it does not introduce a new concept to the theory.

Blogger Comments:

[1] To be clear, a multivariate structure is defined as 'a configuration of elements each having a distinct function with respect to the whole' (Halliday & Matthiessen 2014: 390), whereas a univariate structure is 'an iteration of the same functional relationship' and is 'unique to the logical mode of meaning' (op. cit.: 451).

[2] To be clear, groups have both a multivariate structure and a univariate structure.  It is the latter that motivates the term 'group', rather than 'phrase'.  Halliday & Matthiessen (2014: 437):

Although groups are word complexes, they cannot be fully accounted for as complexes. Groups have developed their own multivariate constituent structures with functional configurations such as the Deictic + Numerative + Epithet + Classifier + Thing of the nominal group in English. Here the elements are

• (i) distinct in function,
• (ii) realised by distinct classes, and
• (iii) more or less fixed in sequence.

A configuration of such a kind has to be represented as a multivariate structure. Treating the group simply as a ‘word complex’ does not account for all these various aspects of its meaning. It is for this reason that we recognise the group as a distinct rank in the grammar.

[3] To be clear, this is not "unnecessarily abstract terminology", merely misunderstood terminology.  The term 'element' identifies a part of a whole (structure), whereas the term 'variable' identifies the choice involved (system).

Misconstruing Parataxis As Coordination

Fawcett (2010: 26):

The first two concepts are a pair of ideas that have come to play a major role in Halliday's later model of grammar — and especially in the framework that he uses for analysing text-sentences in IFG. These are 'parataxis' and 'hypotaxis'. 

In what follows, I shall use 'scare quotes' around these terms as a sign that they — or rather one of them, i.e., 'hypotaxis' — will have no role to play in the framework presented here. However, since the two terms are so closely bound up together in Halliday's theory, it will be safer to avoid 'parataxis' too. The types of relationship between units that 'parataxis' provides for are essentially the same as those covered by co-ordination, in a broad sense of the term that includes 'asyndetic co-ordination' (as described in Quirk et al. 1985:918) as well as co-ordination with overt markers such as and. I shall therefore normally use the term "co-ordination" rather that "parataxis".

Blogger Comments:

This is misleading.  The traditional notion of coordination is not equivalent to parataxis, but to the combination of parataxis and the logico-semantic relation of extension.  Halliday & Matthiessen (2014: 472):

The combination of extension with parataxis yields what is known as co-ordination between clauses.  It is typically expressed by and, nor, or, but.  We can recognise three major subtypes of paratactic extension, (i) addition, (ii) variation and (iii) alternation.

Unlike coordination, parataxis also combines with the other two relations of expansion, elaboration and enhancement, as well as the other major logico-semantic relation, projection, as in the case of a quoting nexus. Traditionally, coordination contrasts with subordination, which largely corresponds to the SFL category combination of hypotactic enhancement.  The distinction between taxis and logico-semantic relations is an important one that potentially obtains between units in complexes at all ranks in the grammar: clause, group/phrase, word and morpheme.

Misrepresenting Halliday (1985)

Fawcett (2010: 25):

It was only with the eventual publication of the first edition of IFG in 1985 that large numbers of examples of analyses in SF terms became generally available. By then, however, Halliday had introduced a completely different type of representation, as we shall see in Chapter 7. 

Even so, IFG has the same lack of integrated representations of structure as "Categories", i.e., representations that shows [sic] how the analysis of, let us say, a nominal group fits into the analysis of a clause.

Blogger Comments:

This negative appraisal is unwarranted and misleading.  Halliday (1985) provides a grammatical rank scale which sets out explicitly how the analysis of each rank relates to the analysis of another.

For example, in terms of constituency, a higher rank unit, such as clause, consists of one or more units of the rank below, group/phrase.

In terms of levels of symbolic abstraction, each element of the function structure of a higher rank unit, such as the Senser of a clause, is congruently realised by a unit of the rank below, in this case, by a nominal group.

The elements of nominal group structure function within the nominal group, not the clause, whereas the nominal group, as a whole, serves a function within the clause.

Misrepresenting Halliday (1961)

Fawcett (2010: 24):

Reading "Categories" is demanding work, and the reader's task is not eased by the fact that only rarely does Halliday illustrate the abstract concepts that he is presenting with examples. Ironically, by far the fullest exemplification of the concepts comes in the 'grammar' of meals (viewed as another social construct) that is presented in Section 9! There are occasional indications as to what the elements of the units of the clause in English would be like (e.g., in Section 4.3), but there is no attempt to provide a full set of descriptive categories for any aspect of the grammar of English or any other language, other than the list of units on the 'rank scale' of English. For example, there is no indication of what Halliday considered at the time to be the full set of classes of group.  

Perhaps the most surprising omission is that there is no visual representation of the full structure of a clause-length text that has been analysed in terms of the categories proposed in this key paper. One direct result of this is that there is no indication as to how the various units in a the description of a clause are to be related to each other in an integrated representation of structure.

Blogger Comments:

This series of negative judgements, complaining of the absence of structural representations in Halliday (1961), is misleading because it studiously ignores the explicitly stated purpose of 'Categories of the theory of grammar'.  Halliday (2002 [1961]: 37-8):

My purpose in writing this paper is to suggest what seem to me to be the fundamental categories of that part of General Linguistic theory which is concerned with how language works at the level of grammar, with brief reference to the relations between grammar and lexis and between grammar and phonology. … 

No excuse is needed, I think, for a discussion of General Linguistic theory. While what has made linguistics fashionable has been, as with other subjects, the discovery that it has applications, these applications rest on many years of work by people who were simply seekers after knowledge. It would not help the subject if the success of these applications led us into thinking that the theoretical problems were solved and the basic issues closed.

Misconstruing Functional Grammar As Semantics

Fawcett (2010: 23):

While Halliday continues to use the derived concept of 'secondary structures' in IFG (as we shall see in Section 10.3.4 of Chapter 10), he took the innovatory step in "Categories" of interpreting 'secondary classes' as 'systems' (1961/76:67) — and so of relating the two paradigmatic concepts of 'class' and 'system' on the scale of delicacy. Thus, the elevation of the concept of 'system' to the level of 'meaning' (which we shall discuss in Section 4.3 of Chapter 4) also implies the elevation of the concept of 'class'.

 Blogger Comments:

[1] This is misleading.  What Fawcett labels 'secondary structures' turn out (in his Chapter 10) to be three distinct phenomena mistaken for one:

• univariate logical versus multivariate experiential structure of the nominal group,
• interpersonal structure of the clause: Mood + Residue versus Subject + Finite + Predicator + Adjunct,
• element of textual structure of the clause: Theme versus textual/interpersonal/experiential subtypes versus further subtypes of each.

The first involves two complementary metafunctional perspectives on the nominal group, the second involves the two layers of the interpersonal structure of the clause, and the third involves the elaboration of subtypes of the Theme of a clause.

[2] This is confused.  Class and system are not related on the scale of delicacy.  The scale of delicacy is a dimension of system.

[3] The term 'elevation' is misleading here, and is the source of the false inference.  In Section 4.3, Fawcett misinterprets Halliday's clause systems of TRANSITIVITY, MOOD and THEME as semantic instead of lexicogrammatical.  This is because Fawcett confuses semantics with a view of the grammar from the perspective of semantics.  Halliday & Matthiessen (2014: 49) clarify:

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.

Misunderstanding Lexicogrammatical Delicacy

Fawcett (2010: 23n):

7. In creating and interpeting [sic] system networks, the concept of 'dependency' is in fact more fundamental than 'delicacy', as I have shown in Fawcett (1988b). It is quite widely assumed that systems that are realised in lexis are inherently more 'delicate' than systems that are realised in syntax, and that syntactically-realised systems are therefore never dependent on lexically-realised systems. But see Fawcett (1996) for a demonstration that this assumption is wrong in relation to certain classes of lexical verb, and Tucker (1998) for a similar demonstration in relation to certain adjectives and manner adverbs.

Blogger Comments:

[1] In SFL theory, delicacy is a major organising principle of system networks.  It is the dimension from the most general choices to the most specific.  In terms of the fractal types (Halliday & Matthiessen 1999) that feature throughout the theory at various scales, delicacy is the expansion subtype elaboration.

In SFL theory, dependency is the hypotactic relation between units in unit complexes, as between clauses in a clause complex.  The closest relation to 'dependency' in the architecture of system networks is the conditional relation between systems, as specified by the entry conditions to systems.  In terms of the fractal types, condition falls within the expansion subtype enhancement.

[2] In SFL theory, the traditional notion of 'syntax' can be seen in the syntagms of forms that realise grammatical function structures, and lexis is part of a unified lexicogrammar.  Halliday & Matthiessen (1999: 198-9):

… we can differentiate both processes and participants into finer and finer subcategories, until we reach a degree of differentiation that is associated with the choice of words (lexical items). Note that it is not (usually) the lexical items themselves that figure as terms of the systems in the network. Rather, the systems are systems of features, and the lexical items come in as the synthetic realisation of particular feature combinations. Thus lexis (vocabulary) is part of a unified lexicogrammar; there is no need to postulate a separate “lexicon” as a pre-existing entity on which the grammar is made to operate.

Misrepresenting Halliday (1961) On The Rank Scale

Fawcett (2010: 21):

Halliday then points out that "the theory allows for downward 'rank shift' : the transfer of a [...] given unit to a lower rank" [i.e., it allows a unit such as a clause to occur at an element where, in the unmarked case, a lower unit such as a group or a word would occur].  Moreover the theory "does not allow for upward 'rank shift'", i.e., a word cannot function directly as an element of a clause. The claim that elements of a clause must be filled by groups rather than by words (which Halliday expresses as "no upward rank shift") has attracted particularly strong criticism, both from outside SFL (e.g., Matthews 1966) and from within it (e.g., Hudson 1971 and Fawcett 1973/81).

Blogger Comments:

This is misleading.   Excluding (downward) rank-shift, structural elements of a higher rank are realised by units of the rank below — elements of a clause by groups and phrases, and elements of groups and phrases by words.

An element of clause structure, such as Subject, may be realised by a nominal group with only one structural element, Thing, realised by one word, as in the clause Cretans are liars.

The reason for distinguishing the group Cretans from the word Cretans is that each unit affords different systemic potential.  For example, the group Cretans can be replaced by another group, such as the richest corporate executives, whereas the word Cretans can only be replaced by another word, such as politicians.

Misrepresenting Halliday (1993)

Fawcett (2010: 12):

Is it necessary, you may ask, to read all of Part 1, in order to understand the presentation of the theory in Part 2? The answer is that, strictly speaking, it is not. However, it is in Part 1 — and in particular in Chapter 7 — that I explain why the theory presented in Part 2 is needed.

But there is a further reason to read Part 1. It is that Halliday describes Systemic Functional Linguistics as a theory of language that has remained essentially unchanged in its lexicogrammatical 'core' over the last forty years, with the developments in the theory coming through "expansion" rather than change (e.g., 1993:4507). However, Chapters 3 to 7 of Part 1 demonstrate clearly that he has in fact introduced a whole series of changes between the sixties and today. Thus the fact that I am proposing certain further changes to the theory in Part 2 is not so revolutionary as it might at first appear. This further stage in the evolution of the theory is necessary, I believe, in order to enable it to meet the demands that will be made on Systemic Functional Linguistics in the 2000s.

Blogger Comments:

[1] This is misleading.  What Halliday (1993: 275*) actually wrote, after describing the origins, basic concepts and the development of SFL theory, is as follows:

A feature of systemic work is that it has tended to expand by moving into new spheres of activity, rather than by reworking earlier positions.

That is, Halliday contrasts developments in the theory with the expansion of its applications to new fields.  The reason this trivial point is worth clarifying is that Fawcett's claim constitutes another strategic misrepresentation, rather than an accidental misunderstanding.

[2] It will be seen that creating theoretical consistencies — which, as previously demonstrated, Fawcett's proposals would do —will not improve the functionality of any theory.

* Halliday (1993) can be viewed online here.

Misrepresenting IFG (Halliday 1994)

Fawcett (2010: 10):

But it is the analysis of texts in IFG, of course, that constitutes the major evidence as to how Halliday sees the structure of what we shall later call 'instances of syntax'. And in evaluating these representations we shall find — contrary to what you might expect — that they raise serious theoretical problems. Moreover, in the course of Chapter 7 it becomes clear that, even if you feel completely happy about the representations of structure in IFG, Halliday's model additionally needs an integrating syntax of the sort proposed here in Part 2.

Blogger Comments:

[1] In SFL theory, there is an important distinction between function structure and formal syntagm (syntax).

[2] It will be seen in the examination of this later discussion that Fawcett confuses the theoretical dimension of instantiation with the theoretical dimension of delicacy.

[3] It will be seen in the examination of this later discussion that the function structures do not raise serious theoretical problems.  For example, Fawcett will present the double layering of interpersonal clause structure as a theoretical problem.

[4] Fawcett's 'integrating syntax' depends on an incompatible theoretical architecture (semantics as system, grammar as formal syntagm), and so cannot be included in the SFL model without sacrificing theoretical consistency.

Misrepresenting Halliday (1961, 1993, 1994)

Fawcett (2010: 10):

Interestingly, however, neither Halliday nor any of his close colleagues has made a detailed statement about a modern theory of SF syntax that can be compared with that in "Categories".  The best summary of the "basic concepts" of the theory as Halliday sees them today is in his paper "Systemic theory" (1993), and this is summarised in Chapter 5. A second obvious source of insights is the major description of English that he provides in IFG, and this is examined in Chapter 7. Surprisingly, there are considerable differences between the theoretical concepts presented in these two works by Halliday, both of which were published in the early 1990s, and this clearly requires comment and explanation.

Blogger Comments:

[1] This is yet another negative appraisal (judgement) that is entirely without foundation.  Halliday (1961) set out the categories and dimensions of a theory of language, not syntax, and developed a theory from that framework.  In the course of developing what became known as Systemic Functional Grammar, syntax — the syntagmatic arrangement of form — was backgrounded in favour of function, in line with the requirements of a functional, rather than formal, theory.  Form is modelled in SFG as a rank scale of clause, phrase/group, word and morpheme, and a function structure at a higher rank is realised by a syntagm of forms at the rank below.  This was all set out in the first edition of IFG (Halliday 1985).

[2] This is manifestly untrue, and will be demonstrated to be so in the critique of Chapter 7.

Misrepresenting Halliday's 1961 Grammar As Syntax

Fawcett (2010: 10):

We shall begin, in Chapter 2, with a summary of Halliday's seminal paper "Categories of the theory of grammar" (1961/76) — which is itself essentially a theory of syntax. After outlining Halliday's overall model of language as it was in 1961, I shall summarise the seven main concepts in what was at that time an exciting new theory of syntax (or "grammar", as Halliday would term it). However, during the sixties Halliday developed what was essentially a theory of syntax into the rich theory of language as a whole that is known today as Systemic Functional Grammar.

Blogger Comments:

[1] As previously demonstrated, and as the title suggests, the theory of grammar in Halliday (1961) goes well beyond mere syntax.  The reason this repeated misrepresentation is worth noting again is that it strategically serves Fawcett's position.

[2] Trivially, to be clear, the theory of language is Systemic Functional Linguistics, whereas Systemic Functional Grammar is the theory of grammar.