Misconstruing Hypotaxis As Embedding

 Fawcett (2010: 251):

Filling may introduce a single additional unit to the structure, or it may introduce two or more co-ordinated units. (For co-ordination see Section 11.8.2). For example, an Adjunct that expresses 'Time Position' may be filled by a nominal group such as the day before yesterday, a prepositional group such as on Friday, a quality group such as quite recently, or a clause such as when I was last in London. (In the last case it introduces a clause that is embedded in another clause; see Section 11.8.3 for 'embedding'.) Alternatively, an element may be filled by two co-ordinated units, as in (I lost it) either last Monday or last Tuesday. …

In the Cardiff Grammar, the realisation operation that introduces this relationship of filling to a structure is "Insert a unit to fill Element X". The most surprising fact about the Sydney Grammar's list of realisation operations, as stated in their theoretical-generative publications, is the lack of any equivalent to this crucial operation (as discussed in Section 9.2.3 of Chapter 9).

Blogger Comments:

[1] To translate this into SFL Theory: a functional element of one rank may be realised by either a unit or a complex of units of the rank below. However, only in SFL theory, the complex may be either paratactic ("co-ordinated") or hypotactic.

[2] To be clear, in SFL Theory, quite recently is an adverbial group.

[3] To be clear, in SFL Theory, the clause when I was last in London can be either rankshifted (embedded) or ranking:

Fawcett's Cardiff Grammar treats both of the above instance types as embedded, and so can not distinguish between them. Moreover, Fawcett's Cardiff Grammar treats such clauses as embedded (rankshifted) in cases of hypotaxis, but as co-ordinated (ranking) in cases of parataxis. That is, from the perspective of SFL Theory, Fawcett's model involves both reduced explanatory potential and theoretical inconsistency.

[4] This is misleading, because this fact is not surprising. As explained in the examination of Section 9.2.3 (here), Fawcett's realisation operation is unnecessary in SFL Theory, because a unit (clause, group, word, morpheme) is not "inserted" but selected from the rank scale in a system network.

Misrepresenting SFL Theory On Function-Form Relations

Fawcett (2010: 251):

Filling is the relationship between an element and the unit that 'operates at' it — this being the unit below it in a tree diagram representation. It can be argued that it is the fact that the Cardiff Grammar gives this concept a central position in the theory of syntax that enables it to solve a range of problems for which more complex solutions are proposed by Halliday. Thus it is filling that makes possible both co-ordination and embedding, and it is the extensive use of these that enables us to do without the somewhat problematical concepts of 'parataxis' and 'hypotaxis'.

Blogger Comments:

[1] To be clear, in SFL Theory, the relation between an element (e.g. Senser) and the unit that "operates at" it (e.g. nominal group) is the realisation relation between function and form. A clause rank function structure, such as Senser ^ Process ^ Phenomenon, is realised by a group rank syntagm, such as nominal group ^ verbal group ^ nominal group.

[2] This is misleading, because it falsely implies that this relation between function and form is not part of the architecture of SFL Theory.

[3] This is misleading because it is untrue. The realisation relation between function and form, in SFL Theory, is simpler, not more complex, than Fawcett's relation, filling, not least because that the latter is unnecessarily complicated by confusing formal constituency with function-form relations.

[4] To be clear, Fawcett (p272) relates his co-ordination and embedding to SFL tactic relations as follows:

To summarise: we treat four of Halliday's five types of 'hypotaxis' and two of his five types of 'parataxis' as embedding, and one type of 'hypotaxis' and his three 'expansion' types of 'parataxis' as co-ordination.

This "less problematical" approached will be carefully examined in the relevant future post.

The Absence Of Textual Structure In The Cardiff Grammar

Fawcett (2010: 250):

One difference between the Cardiff and the Sydney Grammars is that the former does not mark the 'Theme' function of a Subject explicitly in the structure, while the Sydney Grammar does. It is in fact redundant to show this, in that the Subject's status is as a type of Theme is directly inferable from the fact that it is the Subject — i.e., the grammar specifies that any PR that is conflated with the Subject is thereby automatically also a "Subject Theme'. The fact that it must be a PR excludes "empty Subjects" (as in the underlined parts of 

It was Ivy that did it, 

It's likely that she did and 

There's a fly in my soup), 

which are not 'Themes'. (See Fawcett in press for such 'enhanced theme' constructions.)

 

Blogger Comments:

[1] To be clear, this is a serious (and widespread) misunderstanding of Theme in SFL Theory. Importantly, Theme is defined textually, not interpersonally (Subject) or experientially (Participant Role). Theme and Subject are independent choices, and Fawcett's model fails to account for all the instances where Theme is not conflated with Subject — e.g. Theme/Adjunct, Theme/Complement, Theme/Predicator — and fails to account for both textual and interpersonal Themes.

Moreover, the absence of a Rheme element in Fawcett's model means that it does not provide a textual structure of the clause, since a structure is the relationship between elements.

[2] To be clear, in SFL Theory, because Theme is defined textually, the underlined parts of these three instances are indeed examples of Theme (and only in the final instance does the Theme not conflate with a participant).

See, for example, Halliday (1994: 60, 98).

[3] As previously noted, Fawcett (in press) is still unpublished, 21 years after the first edition of this work.

The Absence Of Experiential Structure In The Cardiff Grammar

Fawcett (2010: 249-50):

… every Participant Role (PR) is introduced to the structure by being conflated with an element such as Subject. A PR is simply a particular type of element that is generated from the experiential component of the system network, and it is not a different order of phenomenon from an 'element'. A PR may appear to be more 'semantic' than an element such as Subject, but it is not. The presence of each in the structure directly expresses a meaning, and the only difference is that the meaning expressed by a PR (such as Agent) is overtly referred to in the name of the features in the system network, e.g., as [overt agent] and [covert agent], as in Figure 10 in Chapter 7. … It is the concept of 'conflation' that expresses the multifunctional nature of language, and the same concept and notation are used in all versions of SFL.

Blogger Comments:

Reminder:

[1] To be clear, this is a serious shortcoming of Fawcett's Cardiff Grammar, since it is inconsistent with the notion of structure in SFL Theory. Halliday & Matthiessen (2004: 60):

The significance of any functional label lies in its relationship to the other functions with which it is structurally associated. It is the structure as a whole, the total configuration of functions, that construes, or realises, the meaning. The function Actor, for example, is interpretable only in its relation to other functions of the same kind — other representational functions such as Process and Goal. So if we interpret the nominal group I as Actor in I caught the first ball, this is meaningful only because at the same time we interpret the verbal group caught as Process and the nominal group the first ball as Goal. It is the relation among all these that constitutes the structure. In similar fashion, the Subject enters into configurations with other functional elements as realisation of the clause as exchange; and similarly the Theme, in realising the clause as message.

That is, Fawcett's Cardiff Grammar provides no experiential structure of the clause, since it does not provide a total configuration of functions in which Agent, Process and Medium, for example, are structurally related. By the same token, Cardiff Grammar provides no textual structure of the clause, since it does not provide a total configuration of functions in which Theme and Rheme are structurally related.

[2] This is not misleading, because it is true.

[3] This is true, but potentially misleading. In SFL Theory, both Subject and Agent (PR) are both semantic and lexicogrammatical. This is because lexicogrammatical forms, such as nominal groups, are analysed in terms of their functions in realising meanings, such as Subject and Agent. In the absence of grammatical metaphor, semantics and lexicogrammar are in agreement (congruent).

[4] This is misleading, because it is untrue. The crucial difference between Subject and a participant role is metafunctional: Subject is an interpersonal function, whereas a participant role is an experiential function.

[5] To be clear, this is seriously inconsistent with SFL Theory, in which all elements of syntagmatic structure — including both Subject and participant roles — are specified in paradigmatic systems.

[6] To be clear, Fawcett does not supply the system network from which Figure 10 is derived.

[7] This is misleading, because Fawcett's use of the concept of conflation is significantly different from its use in SFL Theory. In SFL Theory, conflation correlates elements of different metafunctional structures, whereas in Fawcett's Cardiff Grammar, conflation fuses different metafunctional elements into a single structure, thereby creating metafunctional inconsistency in that structure.

Why Nothing Is Gained By Modelling Dependency Relations In Syntactic Representations

 Fawcett (2010: 249):

The question is therefore: "What is gained by modelling these dependency relations in the syntactic representation too?" The answer is that nothing is gained. In a SF grammar it is simply not the task of a syntactic structure to show that the presence of one element 'depends' on the presence of another. To attempt to do so leads one to ask inappropriate questions, since this is not where dependency is located. The type of 'dependency' that is important in syntax is the relationship of componence, i.e., that between the elements and the unit of which they are the elements. Without this theoretical concept no SF grammar would be complete — and nor could it be implemented in a computer.

Blogger Comments:

[1] This is misleading, because it is untrue. What is gained is the structural realisation of the systemic options in the system of taxis (interdependency).

[2] To be clear, any 'SF Grammar' that does not show the structural realisations of interdependency relations is deficient on the syntagmatic axis, and so compromises its explanatory potential.

[3] To be clear, on the one hand, as previously demonstrated, "inappropriate questions" only arise from theoretical misunderstandings. On the other hand, in SFL Theory, interdependency relations are accounted for both systemically and structurally, since structure realises systemic choices.

[4] This is misleading, because componence is not a type of dependency. Moreover, as previously observed, Fawcett's 'componence' confuses formal constituency with function-form relations.

[5] This is misleading, to the extent that it implies that, unlike Fawcett's Cardiff Grammar, SFL Theory does not already include function-form relations in its architecture.

Why Dependency Relations Are Undesirable In Fawcett's Model Of Syntax

Fawcett (2010: 249):

However, it can be shown that it is not in fact necessary — nor even desirable — to model sister dependency relations in syntax. As I pointed out in Section 10.3.3 of Chapter 10, the supposed 'dependency' of a "modifier' on the "head' on a unit (e.g., in a nominal group) is a 'second order' concept, and ultimately an uninsightful one. In a SF grammar the relationship between the two elements is more appropriately seen as an indirect one, because each element realises its own aspect of the meaning of a referent, and the relationship is therefore at the level of meaning rather than form. In other words, in a SF grammar, any relationship of apparent 'syntactic sister dependency' is already expressed, in a natural manner, in the dependency relations by which one system is dependent on another in the system network of the language's meaning potential. For the systemic grammarian this is the true location of dependence (or 'dependency').

Blogger Comments:

[1] This is misleading. On the one hand, Fawcett's argument is invalid because it relies on his own misunderstandings of theory; see further below. On the other hand, interdependency is valuable for its explanatory potential across all grammatical ranks. This is demonstrated at group rank, for example, where the logical structure of the verbal group realises the system of TENSE in English. Halliday (1994: 198-9):

[2] See the examination of Section 10.3.3 — Confusing Structure With (Transcendent) Ideational Denotation — for a dissection of Fawcett's first-stated misunderstandings on this matter.

[3] Here Fawcett repeats his confusion of logical structure (dependency relations) with ideational denotation ("each element realises its own aspect of the meaning of a referent").

[4] To be clear, on the one hand, 'meaning and form' are levels in Fawcett's model (Figure 4), not in SFL Theory. On the other hand, in SFL Theory, dependency relations obtain at both levels of content, semantics and lexicogrammar — with mismatches constituting instances of grammatical metaphor — and are modelled in systems that specify their structural realisations.

[5] This is misleading. In SFL Theory, dependency relations are modelled as systems, not by dependency relations between systems. For example, 

Strictly speaking, system networks do not construe dependency relations between systems. However, it could be said that entry to a system "depends on" the selection of its entry condition.

Fawcett's Argument Against Dependency Relations In A Footnote

 Fawcett (2010: 248-9n):

¹¹ … However, in Fawcett & Davies (1992) and Lin & Fawcett (1996), my colleagues and I have demonstrated (in terms of both visual diagrams and a computer implementation) that sister dependency relations can always be interpreted in constituency terms without loss of information. 

… the 'sister-dependency' approach has certain disadvantages. Firstly, it lacks the concept of 'class of unit' (channelling all 'constituency' relations through the 'head'). Secondly, the sister dependency approach raises questions about which element 'depends on' which, when there is frequently not in fact any 'dependency' between them — even between the features in the system network from which the elements are generated.

Blogger Comments:

[1] This is misleading, because it is untrue. In SFL Theory, dependency is modelled as the univariate iterative structure of the logical metafunction, whereas constituency is the basis of the multivariate segmental structure of the experiential metafunction. Halliday (1994: 36):

Because of this, interpreting dependency in constituency terms "loses" the logical dimension of ideational meaning.

[2] This is misleading, because it is untrue. Since interdependency relates formal units — clause, group/phrase, word, morpheme — in complexes, it does not "lack the concept of class of unit", as demonstrated, for example, by nominal group complexes, verbal group complexes, etc. The element Head, on the other hand, identifies the function of one of the formal units in a complex.

[3] This is misleading, because it is untrue. SFL Theory distinguishes two types of interdependency: parataxis and hypotaxis. Parataxis involves units of equal status, whereas hypotaxis involves units of unequal status. The dependency of a unit in a hypotactic complex is demonstrated by its inability to stand alone, as in the case of the dependent clause because they trusted Fawcett without question… .

Componence vs Dependency

Fawcett (2010: 248):

Throughout this book I have assumed, as most linguists do, that the dominant relationship in modelling relationships in syntax is 'constituency'. And in spelling out the concepts that make up 'constituency' I have assumed that it is the concept of 'componence' that expresses the 'part-whole' relationship. However, there is an alternative concept that has at times attracted the support of a number of fine linguists. It is often called 'dependency grammar' — but, strictly speaking, it should instead be called "sister dependency grammar". … The question is: "Is it either necessary or desirable to show 'sister dependency' relations in the syntactic representation of a text-sentence?" If it is either necessary or desirable, this would have a profound effect on the proposals set out here, and in particular on the place of 'componence' in the theory.¹¹

Blogger Comments:

[1] To be clear, SFL Theory distinguishes four types of structure, only one of which is based on constituency. Halliday (1994: 36):

[2] As previously demonstrated, Fawcett's notion of componence confuses formal constituency with the relation between function (element) and form (unit).

[3] To be clear, in SFL Theory, interdependency is used to relate units in complexes at all ranks.

[4] To be clear, whether or not interdependency is "necessary of desirable", componence has no place in SFL Theory because it is invalidated through its confusion of function with form.

Confusing Formal Constituency With Function Structure

Fawcett (2010: 247):

The main notation used by Halliday in IFG consists of several lines of analysis, as exemplified in Figure 7 in Chapter 7. If we showed just the line that is the equivalent of what is shown here (the top line of the MOOD analysis), it would be as in Figure 15.

This gives essentially the same amount of information as the previous notation, and it has the same limitation that it does not lend itself to showing in the same diagram the lower layers of the structure. Moreover, it is not clear why it should be thought worth using the extra time and work of drawing boxes around categories in the diagram, rather than simply drawing single lines between categories, as in the first notation, i.e., (a) above.

Blogger Comments:

[1] Fawcett's Figure 7, which misanalyses Mrs Skinner as Goal instead of Scope, is reproduced below:

To be clear, Fawcett's tautological claim is that if everything in this diagram is removed, except that which corresponds to the previous notation, then it "gives essentially the same information as the previous notation". Clearly, if this is not done, it provides considerably more information than Figure 15.

[2] This is misleading, because it confuses formal constituency ('lower layers') with function ('structure'). Though units of higher ranks consist of units of lower ranks, the structure of a lower rank (e.g. group) is not a structure of a higher rank (e.g. clause).

[3] To be clear, one practical advantage of using box diagrams for function structure and tree diagrams for formal constituency is that it provides a visual means of distinguishing the two theoretical principles.

Misrepresenting Halliday On Notation

Fawcett (2010: 246):

Diagram (a) in Figure 14 illustrates the two types of componence [branching and singulary], and Diagram (b) illustrates an alternative notation used by some grammarians for multiple branching (e.g., Halliday (1969/81:143), Hudson (1974/81), and Halliday (1994:17f.), in the chapter of IFG that discusses 'constituency'). 

Blogger Comments:

This is seriously misleading, because it falsely implies that Fawcett's notion of componence is consistent with IFG (Halliday 1994). Such notations play no part in Halliday's theory of grammar. Instead, Halliday (1994: 29) makes pedagogical use of constituency diagrams to distinguish class labels from function labels, as a means of guiding the reader from formal constituency to function structure:

Misrepresenting Halliday (1993) On Place

Fawcett (2010: 245):

We saw in Section 9.2.3 of Chapter 9 that the Sydney Grammar, according to the theoretical statements in each of the three summaries of its theory, uses two operations to accomplish the placing of an element in the structure of its unit, i.e., "Insert" and "Order". However, it seems from "Systemic theory" (p. 4505) that the "Order" operation may now have been extended to include "Order an element to some defined location", and this may signal the use of the concept of 'place', as originally introduced to SFL in Fawcett (1973/81). If this is so there is now little difference between the two models in this respect.

Blogger Comments:

[1] This is misleading, because it is untrue. Of the seven realisation statements listed in Halliday (1993), three can affect the ordering of elements:

[2] This is misleading, because it is untrue. Halliday does not define structural locations in terms of Fawcett's concept of place (p279):

…the numbered position (or 'slot') in a unit at which an element is positioned.

Componence: Functions As Parts Of Forms

Fawcett (2010: 244):

In the next four sections we shall examine the three crucial relationships of componence, filling and exponence into which the 'consists of relationship between units must be broken down — and also the concept that models in the syntax itself the multifunctional nature of language, i.e., conflation.

Componence is the part-whole relationship between a unit and the elements of which it is composed. Thus the componence of the nominal group the man with a stick is dd h q; the componence of the prepositional group with a stick is ρ cv, and that of the nominal group a stick is qd h.

Blogger Comments:

[1] This is misleading, because it is untrue. The 'consists of' relationship between units — in SFL Theory: the rank scale of forms — can not be broken down into componence, filling and exponence, because each of these three is concerned with the form-function relations, not with the constituency of forms (units); see further below.

[2] To be clear, Fawcett's notion of componence confuses form (units) with function (element), and posits functions as parts of forms. The reason this is theoretically invalid is because form and function are different levels of symbolic abstraction.

[3] To be clear, as previously demonstrated, Fawcett's notions of filling and exponence are both the relation between function (element) and form (unit and item, respectively). That is, both involve the relation of different levels of symbolic abstraction, which in SFL Theory, is the relation of realisation.

Misrepresenting "The Main Use Of" The Rank Scale

Fawcett (2010: 242):

In practical terms, then, the main use of the 'rank scale' concept has been as a model that makes predictions that guide the text analyst as to how the units of a text-sentence relate to each other — though these have sometimes caused problems for the analyst. However, statements of 'filling probabilities', as in Appendix B, meet the same need in a more effective manner.

All of the probabilities discussed so far are instantial probabilities, i.e., probabilities that certain patterns will occur in instances, i.e. in text-sentences. They are moreover probabilities at the level of form. In contrast with these are the probabilities on features in system networks, which we might refer to as potential probabilities, these being at the level of meaning. See Section 2 of Appendix C for a discussion of the relationship between the two.

 

Blogger Comments:

[1] This is misleading, because it is untrue. In SFL Theory, the rank scale is the means by which formal constituency is modelled. Most importantly, in a systemic functional grammar, each rank is the entry condition for the systems that specify function structures at that rank.

[2] This is misleading, because it is untrue. As previously demonstrated, all the "problems" raised by Fawcett derive from his confusing the rank scale of forms with function-form relations.

[3] This is misleading, because it is untrue. As previously demonstrated, Fawcett's 'filling probabilities' are concerned with the relation between function and form, not with formal constituency. As such, they do not "meet the same need in a more effective manner".

[4] This is misleading, because it is untrue. As previously demonstrated, Fawcett's 'filling probabilities' are concerned with the relation between function (element) and form (unit) — not with form alone.

[5] To be clear, as Section 2 of Appendix C (p315) explains, in Fawcett's model, potential probabilities are paradigmatic probabilities, whereas instantial probabilities are syntagmatic probabilities. That is, this reflects Fawcett's confusion of axis (paradigmatic/syntagmatic) with instantiation (potential/instance), as previously demonstrated in the examination of Figure 4 (p36):

Seriously Misrepresenting Halliday On Embedding In A Footnote

 Fawcett (2010: 240n):

⁵ The clear implication of Halliday's concept of 'rank shift' that, when it occurs, the 'natural order' has in some sense been disrupted. The extensive use that he makes of 'hypotaxis' for modelling relations between units (and especially clauses) can be seen as a way of minimising the role of embedding in the grammar. (See Section 2.6.1 of Chapter 2, Section 11.9 of this chapter and Section 3 of Appendix C for discussions of 'hypotaxis'.) I have never understood why Halliday should see embedding as something to be avoided in a model of language and its use. I suspect that a strong influence on the formation of his position has been the difficulty that users of any language have in processing texts — in either production or understanding — when the depth of embedding (especially non-final embedding) becomes a strain on short-term memory. 

So far as I know, Halliday has never discussed the reason for his distrust of the concept of embedding in any publication. This is perhaps not surprising, given the fact that he would regard himself, if pushed to choose, as a sociolinguist rather than a psycholinguist. Indeed, the substantial literature on embedding takes the viewpoint of the cognitive processing of language (and for good reasons). Halliday, then, leaves psycholinguistics and cognitive linguistics to others. However, this has not prevented some psycholinguists from using his insights in their own work in modelling language and its use, e.g., the excellent psycholinguistics textbook by Clark & Clark (1977), now sadly out of print.

Blogger Comments:

[1] To be clear, as previously noted, in SFL Theory, rankshift (embedding) is seen as a powerful semogenic resource. Halliday & Matthiessen (1999: 10):

The units below the clause on the rank scale are all groups (nominal, verbal adverbial, etc.) or phrases (prepositional phrases), or else clauses that are shifted downwards on the rank scale to serve as if they were groups or phrases. Such down-ranking is known as rankshift. This has the powerful effect of expanding the resources of grammar by allowing the meaning potential of a higher-ranking unit to enrich that of a unit of lower rank. … Such rankshifted clauses construe what we call macro-phenomena.

[2] This is very misleading indeed. Hypotaxis is not a means of minimising the role of embedding in the grammar. Halliday's use of the traditional term 'embedding' excludes hypotaxis, which is the main reason why he prefers the term 'rankshift'. In IFG (Halliday 1994: 242) he makes the important theoretical distinction between the two:

[3] This is very misleading indeed. The important place of rankshift in SFL Theory puts the lie to Fawcett's claim that Halliday sees embedding as something to be avoided in grammatical theorising.

[4] This is misleading, because it is untrue, though, in this case, harmlessly so. For how Halliday views cognitive semantics and cognitive science in relation to SFL Theory, see Chapters 10 and 14 of Construing Experience Through Meaning: A Language-Based Approach To Cognition (Halliday & Matthiessen 1999).

[5] This is doubly misleading, because, on the one hand, as previously demonstrated, Halliday does not "distrust" the concept of embedding, and on the other, Halliday explains his distinction between embedding and hypotaxis in IFG (Halliday 1994), the very same book that Fawcett makes constant reference to in this work.

Misrepresenting Halliday On Rankshift (Embedding)

Fawcett (2010: 239-40):

It follows naturally from the statements which I have just made that there is no implication in the present theory of syntax that a unit is functioning in a highly marked manner when embedding occurs. In this theory it is expected that a clause will quite frequently occur as an element of another clause, or as an element of a group. This position is almost the opposite of that presented in IFG, where the picture is one of very severe limitations on the embedding of units within each other. Indeed, in IFG Halliday stipulates that it is not possible for a clause to fill an element of another clause (except indirectly, by filling the head of a nominal group that fills an element of the higher clause (IFG p. 242).

Blogger Comments:

[1] To be clear, in stark contrast, in SFL Theory, rankshift (embedding) is seen as a powerful semogenic resource. Halliday & Matthiessen (1999: 10):

The units below the clause on the rank scale are all groups (nominal, verbal adverbial, etc.) or phrases (prepositional phrases), or else clauses that are shifted downwards on the rank scale to serve as if they were groups or phrases. Such down-ranking is known as rankshift. This has the powerful effect of expanding the resources of grammar by allowing the meaning potential of a higher-ranking unit to enrich that of a unit of lower rank. … Such rankshifted clauses construe what we call macro-phenomena.

[2] To be clear, in SFL Theory, a rankshifted (embedded) clause does not "occur as" an element of a clause or group, it realises it. This is because clause (form) and element (function) are different levels of symbolic abstraction.

[3] To be clear, Halliday & Matthiessen (2004: 427) identify the following types of embedding in SFL Theory:

[4] This is very misleading. In SFL Theory, a nominal group with a rankshifted clause as Head (or Postmodifier) is "directly" serving as an element of clause structure — just like any other nominal group. Moreover, IFG (Halliday 1994: 97) explicitly provides the following example of an embedded clause serving as Subject:

and Halliday & Matthiessen (2004: 156) add the syntagm of groups to the analysis:

Misrepresenting SFL Theory On The Rank Scale And Function-Form Relations

Fawcett (2010: 239):

In this theory, then, there is no expectation that an element of a clause will necessarily be filled by a group. (For 'filling' as a theoretical concept, see Section 11.5.) Some clause elements are frequently filled by groups, a few are sometimes filled by groups, and some never are. And the same is true of groups. The frequencies vary greatly, so that in the question of what element a unit may fill is often better stated as a probability rather than as an absolute rule.

A further feature of the new theory is that, with respect to the groups, what matters most is not the fact that the unit is a group (as it is in the 'rank scale' model), but what class of group it is. Indeed, in the present framework the differences between the different classes of group (nominal, prepositional, quality and quantity) are just as important as the differences between them all, considered together as groups, and the clause. Indeed, each of the clause and the four classes of group recognised here for English may fill any one of various elements of structure in various classes of unit. Some of the constraints on what may fill what can be expressed by absolute rules, of course, but many others are better expressed as probabilities. In this version of SFL, then, the fact that a variety of different classes of unit may fill many of the elements of many on the units of a language such as English is not regarded as a problem (as it is in a 'rank-based' grammar) but as one of the great riches of human language.

Blogger Comments:

[1] To be clear, throughout this extract, Fawcett confuses the relation between function (element of structure) and form (unit) with constituency relations between forms (the rank scale).

[2] To be clear, from the perspective of SFL Theory, Fawcett uses 'filling' for the relation between elements of clause structure (functions) and the groups (forms) that realise them. From the perspective of Fawcett's source, Halliday's Scale & Category Grammar (1961), this realisation relation was termed 'exponence'. Fawcett also uses 'exponence', but only for the relation between elements of group structure and the items (words/morphemes) that realise them. That is, Fawcett inelegantly uses two distinct terms, filling and exponence, for the same relation between function and form.

[3] This is misleading, because it misrepresents SFL Theory. On the one hand, the rank scale and the class of group are distinct issues. For example, it is possible to model formal constituency as a rank scale and classify groups either 'from above' (as in SFL Theory) or 'from below' (as in Fawcett's Cardiff Grammar). On the other hand, classes of group are no less important theoretically merely because they are not predetermined by the use of a rank scale. However, what is even more important, in a functional grammar, is the function served by such formal units.

[4] This unsupported bare assertion is misleading, because it misrepresents SFL Theory. The fact that an element may be realised by different classes of unit is not a problem for a 'rank-based' grammar, as demonstrated by circumstantial Adjuncts that may be realised by either adverbial groups or prepositional phrases, or by Agents that may be realised either by nominal groups or prepositional phrases, or by topical Themes that may be realised by nominal groups, verbal groups, adverbial groups or prepositional phrases.

The "Principle" Concepts Of A Replacement For 'Rank'

Fawcett (2010: 238-9, 239n):

The principle concepts of the alternative approach

In the approach to 'constituency' proposed here the two key concepts are:

1. that the predictions are made in terms of the relationship of filling that holds between a unit and an element of structure in a higher unit in the tree (rather than being about relations between units), and

2. the use of filling probabilities.⁴ (We shall look at the precise nature of 'filling probabilities' in Section 11.2.2.)

 

⁴ It may be significant that, although the concept of 'filling' was indirectly present in the S&C model (and so is still implicitly there in IFG), it has never been presented as one of the 'basic concepts' of the theory, as it is here (in Section 11.5 of Chapter 11). As we shall see, it is the concept of 'filling' that gives us a principled way to handle co-ordination as a phenomenon that is different from the usual 'componence' relationship of elements in a unit — a difference that all good grammars recognise but for which few have an adequate notation.

Blogger Comments:

[1] To be clear, this is not an alternative approach to modelling the rank scale — constituency relations between forms (units) — because it is instead concerned with the relation between function (element) and form (unit).

[2] This is misleading, because it is untrue. In Scale & Category Grammar, what Fawcett calls 'filling' is termed exponence, which Halliday (2002 [1961]: 41, 55) explicitly identifies as one of the three scales of abstraction in the architecture of the theory. Halliday (2002 [1961]: 57):

The fact that by moving from structure to class, which is (or can be) a move on the exponence scale, one also moves one step down the rank scale, is due to the specific relation between the categories of class and structure…

However, from the perspective of SFL theory, the term 'exponence' covered both realisation and instantiation (the relation of theory to data). So in SFL theory, the relation is explicitly one of realisation, as illustrated by Halliday & Matthiessen (2004: 52):

[3] To be clear, Fawcett's componence is 'the part-whole relationship between a unit and the elements of which it is composed' (p244). That is, it confuses form (unit) with function (element). On the other hand, Fawcett's co-ordination is largely* the relation of paratactic expansion between units; *but see further in the upcoming examination of co-ordination.

Fawcett's 2nd, 3rd & 4th Reasons For Dispensing With The Rank Scale

Fawcett (2010: 238, 238n):

The second reason for dispensing with 'rank' is that in the present theory there are only two major candidates for 'ranks of unit' in the present description of English (and other languages): i.e., 'clause' and 'group', in contrast with the five 'units' of "Categories".³ Can a 'rank scale' of which only two of the supposed 'ranks' occur in most instances really be regarded as a 'scale'?

The third reason for dispensing with the concept of 'rank' is that, while the 'rank scale' model raises problems even for Halliday's own description of English (as described in detail in Appendix C), the problems are far greater when we also take into account the extensions to the internal structure of groups that have been introduced in the Cardiff Grammar'. And the difficulty of sustaining [t]he 'rank scale' hypothesis becomes even greater with the abolition of the 'verbal group' (the reasons for which are given in Fawcett 2000 and forthcoming b, and summarised in Section 4 of Appendix C).

The fourth reason for abandoning the 'rank scale' model is a positive one. It is that there is now an alternative model of relationships between units that enables us to predict much more accurately what units will occur within what other units in natural texts. Let us now see what this alternative model is.

³ There is also the 'cluster', as we saw in Section 10.2 of Chapter 10, but this occurs only infrequently and its existence is therefore an embarrassment for the 'rank scale' concept rather than a support for it. See Section 11.6.2 for the concept of 'variation in depth of exponence', which is often used in conjunction with the concept of the 'cluster' in the Cardiff Grammar.

Blogger Comments:

[1] This reason for dispensing with the rank scale is misleading, because it is untrue. SFL Theory posits four ranks: clause, group/phrase, word and morpheme. In any case, the number of ranks is not grounds for dispensing with the rank scale.

[2] This reason for dispensing with the rank scale is misleading, because it is untrue. As previously demonstrated, the rank scale does not raise problems for Halliday's description of English. Instead, Fawcett's critique relies on confusing the rank scale of forms with relations between form and function.

[3] These reasons for dispensing with the rank scale are comical. Here Fawcett claims that the rank scale should be dispensed with simply because his own model is inconsistent with it. See other posts on this blog for the problems with Fawcett's model.

[4] To be clear, Fawcett's positive reason for dispensing with Halliday's rank scale is that his own model has a predictive advantage in modelling form. Evidence for this claim will be examined when it is tendered.

Fawcett's First Reason For Dispensing Wth The Rank Scale

Fawcett (2010: 237-8, 238n):

Let us begin with the last of the four assumptions listed above. We have already seen in Section 10.2 of Chapter 10 the reasons why, in the new theory, we prefer to determine the class of a unit by its internal structure. This leaves just the concept of the 'rank scale' itself. There are four interrelated reasons for dispensing with this concept.

Firstly, the idea that there is a "consists of' relationship of 'constituency' between the units — which is what the concept of the 'rank scale' states — is not sufficiently precise to be useful. A unit does not in fact function directly as a constituent of another unit; as later sections of this chapter will demonstrate in detail. Instead, the concept of 'constituency' must be broken down into a number of other relationships. In the present theory we shall say that a unit is composed of a number of elements, and that any such element will be either filled by another unit or expounded by an item. (For the concepts of 'componence', 'filling' and 'exponence' see Sections 11.3, 11.5 and 11.6 respectively.) In other words, there is not in fact a 'consists of' relationship between units, but a rather more complex series of relationships.

Blogger Comments:

[1] As previously observed in the examination of Section 10.2, classifying units by their internal structure ('from below'), rather than by their function ('from above'), is inconsistent with the notion of a functional grammar.

[2] To be clear, this is irrelevant to the notion of a rank scale, because the rank scale is a model of form, not function.

[3] To be clear, the notion of a unit being composed of elements confuses formal constituency (unit) with function structure (elements). In SFL Theory, a unit (e.g. clause) consists of lower ranked units (e.g. groups) each of which serves a function (e.g. Process) in the structure of the higher ranked unit.

[4] This is misleading, because it is untrue. Clearly, units of form do consist of other units of form, as demonstrated by any clause consisting of groups. The "more complex series of relationships" are those between form and function, not between form and form.

"The Claims Of The 'Rank Scale' Hypothesis"

Fawcett (2010: 236-7):

For Halliday, it is upon the concept of the 'rank scale' that all of his generalisations as to what units may function at what elements rests. Since we are about to meet the set of concepts which is proposed as a replacement for the 'rank scale', it may be useful to remind ourselves of just what the claims of the 'rank scale' hypothesis are. They are as follows:

1. that the elements of clauses will be filled by groups, the elements of groups will be filled by words and the elements of words by morphemes (though it is clear from the passage cited above that Halliday recognises that there will be at least some exceptions);

2. that 'upward rank shift' does not occur (except, apparently, in the case of the 'Finite', as mentioned above);

3. that 'downward rank shift' is only permitted in the case of clauses and groups functioning as elements of groups (i.e., it is not the case that clauses may fill elements of clauses);

4. that the class of a unit is determined by the element or elements at which it occurs in the unit above it on the 'rank scale'.

(See Section 11.8.5 for the details of the very limited amount of embedding that Halliday allows, as specified in IFG.)

Here I propose a model in which none of the above four hypotheses has a place. The question therefore is: "What is wrong with the above set of generalisations, and what more useful alternative generalisations should replace them?"

Blogger Comments:

[1] This is seriously misleading, because the 'rank scale hypothesis' makes only the second of these 4 claims; but see further below.

[2] This is misleading, because the rank scale is not concerned with functional elements. The rank scale is a model of formal constituency. It posits that a clause consists of a whole number of groups, each of which consists of a whole number of words, each of which consists of a whole number of morphemes.

[3] This is misleading, because the Finite is not a case of any sort of rank shift, because the Finite is not a rank unit, but a functional element,.

[4] This is misleading, because it misrepresents IFG, as demonstrated, for example, by the following instance from IFG (Halliday 1994: 266) of an embedded clause serving as the Carrier element of a clause:

[5] This is misleading, because, on the one hand, the rank scale is not concerned with elements, and on the other hand, a unit does not "occur at" an element in the unit above in the rank scale. On the one hand, Fawcett again confuses the formal constituency of the rank scale with the realisation relation between function (element) and form (unit), and on the other hand, by "occurs at", Fawcett miscontrues these two distinct levels of symbolic abstraction as the same level of abstraction.

[6] See the upcoming examination of Section 11.8.5 for the misunderstandings on which this false claim is based.

The Rank Scale As An Old Habit Of Thought

Fawcett (2010: 236):

The saying "Old habits die hard" holds for our habits of thought as linguists as well as for other sorts of habit, and the habit of thinking in terms of a 'rank scale' has been with most systemic functional linguists for all of their working lives. It seems to me that what Halliday did in setting up Scale and Category Grammar in 1961 and what other did in accepting it was to take over what was already to a large extent a tacit assumption in traditional grammar, and to formalise it as part of the new theory — just as he did with a number of other traditional concepts.  The concept of a 'rank scale of units' is therefore one which we may find particularly hard to hard to let go of. Indeed I allowed vestiges of it to remain in my own theory long after it had stopped playing any practical role in it (as I admitted in Chapter 8). So the general concept that there is indeed some sort of 'rank scale' — even if it is not exactly as Halliday describes it — is one that has been a background assumption about language for many linguists for most of the last half century — and probably longer.

 

Blogger Comments:

[1] To be clear, Fawcett is here suggesting that linguists use the rank scale merely out of habit, rather than from understanding its place in SFL theory and its explanatory potential.

[2] This is potentially misleading. In his now superseded theory, Scale & Category Grammar, Halliday (1961) formalised the relation between grammatical units and constituency as a rank scale. In his successor theory, Systemic Functional Grammar, Halliday (1985: 22-30) distinguished two models of constituency: immediate constituent analysis (maximal bracketing) and ranked constituent analysis (minimal bracketing), associating the former with class labelling, and the latter with function labelling.

[3] As previously demonstrated, unknown to Fawcett, his Cardiff Grammar does assume a rank scale, though one in which form (unit) is confused with function (element).

[4] This is a serious epistemological misunderstanding. The question is not whether there is a rank scale — whether a rank scale "exists" — but whether a rank scale has sufficient explanatory potential to warrant its inclusion in theory.

Misrepresenting Halliday (1993) And Halliday (1994) On Rank

Fawcett (2010: 235-6):

Encouragingly, Halliday displays his usual willingness to allow for alternative approaches when he adds the following to the above passage:

Such issues will be decided empirically. [...] The issue is whether, in a comprehensive interpretation of the system, it is worth maintaining the global generalisation because of its explanatory power, even though it imposes local complications at certain places in the description (Halliday 1994:12).

This is sound advice on the question of how to go about describing a language. Let us see how it applies in the present case. The Cardiff Grammar certainly counts as a "comprehensive interpretation of the system", and there have been decades of "empirical" work by researchers using this framework. It has involved the exploration of alternative SF description of English (and other languages) for both the large scale analysis of texts and for the computer generation of language. The view to which it has led is that the "global generalisation" that the 'rank scale' was intended to express has less "explanatory power" than the theory of syntax described here. So we who work in the framework of the Cardiff Grammar have, following the principles suggested by Halliday, "decided empirically" that the 'rank scale' is no longer needed. (For a fuller account of the 'rank scale' debates, see Appendix C.)

Is there a hint in the passage cited above that Halliday would now make a less strong claim for the 'rank scale' concept? If so, this might explain why it is omitted in "Systemic theory", and why it receives so little space in IFG.

Blogger Comments:

[1] To be clear, this is a bare assertion unsupported by argument. Moreover, there are many reasons for thinking the opposite to be true, as demonstrated by the arguments presented on this blog.

[2] To be clear, this is a bare assertion unsupported by argument. Fawcett has nowhere demonstrated that his model of syntax, his development of Scale & Category Grammar (Halliday 1961), has more explanatory power that SFL Theory.

[3] This is misleading. Fawcett and his colleagues reject the rank scale on the basis of their misunderstandings of the principle, as demonstrated on this blog, not on the basis of empirical findings. Clearly, the validity of empirical findings depends on having an accurate understanding the principle being tested. See the examination of Appendix C for more of the misunderstandings involved.

[4] These are still misleading, because they still untrue. On the one hand, as previously observed, rank is not omitted from the (very brief) encyclopædia article, Systemic Theory (Halliday 1993), where the following point is explicitly made:

On the other hand, IFG (Halliday 1994) is organised on the basis of rank, with 4 chapters devoted to clause rank and 2 chapters to group/phrase rank. Moreover, the third edition of IFG (Halliday & Matthiessen 2004: 20) explicitly puts the lie to Fawcett's claim:

Misrepresenting Halliday (1994) On The Rank Scale

Fawcett (2010: 235):

The situation is therefore that Halliday seems on the one hand to wish to maintain the general principle of the 'rank scale' — and with it the view that "the guiding principle is exhaustiveness at each rank" (IFG p. 12 — and on the other to allow for exceptions. Indeed his statement of the "exhaustiveness" principle in IFG is closely followed by these words:

At the same time, there is room for manœuvre: in other words, it is an integral feature of this same guiding principle that there is indeterminacy in its application (Halliday 1994:12).

Thus the "exhaustiveness" principle seems to be one that can be breached. But the examples that Halliday cites are all cases of borderline judgements, and there will always be such cases in text analysis. The greater problem is the existence of many clear cases of exceptions, such as those cited above and in Appendix C. And even if the principle is only breached to the extent that I argue to be required in Fawcett (2000) and (forthcoming b), it is in effect dead.

Blogger Comments:

[1] This is misleading. What Halliday (1994: 12) actually says is that there is indeterminacy in the application of this principle, and the "borderline judgements" that he cites is the question of whether there is one sub-sentence layer (group/phrase) or two (phrase and group):

[2] This is doubly misleading, because it is untrue in at least two respects. On the one hand, indeterminacy in the application of the principle is not a breach of the principle, and on the other hand, the "many clear cases of exceptions" that Fawcett cites are not exceptions to the notion of a rank scale and the principle of exhaustiveness because Fawcett confuses the rank scale of formal constituency — clause, group/phrase, word, morpheme — with the relation between function (e.g.Finite) and form (e.g.word), as demonstrated in the previous post. Accordingly, reports of the rank scale's death have been greatly exaggerated.

Halliday's "Most Spectacular Breach Of The 'Rank Scale' Principle"

Fawcett (2010: 234-5):

However, Halliday's most spectacular breach of the 'rank scale' principle is his treatment of the 'Finite' element (e.g., did in Did he like it?). Throughout IFG his analyses of the MOOD structure of clauses show the Finite as an element of the clause. The effect is that single words such as the 'modal verbs' and forms of do, be and have all regularly function as direct elements of the clause. In Fawcett (2000) and (forthcoming a) I demonstrate that the way to resolve this and the various other problems of the 'verbal group' is to promote not just the Finite but all of the elements of the supposed 'verbal group' to function as elements of the clause. This two-part paper sets out important evidence against the concept of the 'verbal group, and so against the concept of the 'rank scale'. (See Section 4 of Appendix C for a summary.)

Blogger Comments:

[1] To be clear, the Finite element is irrelevant to the theoretical validity of the rank scale principle because the Finite is a function whereas the rank scale is a model of form. Halliday (2002 [1961]: 43):

The relation among the units, then, is that, going from top (largest) to bottom (smallest), each consists of one, or of more than one, of the unit next below (next smaller). The scale on which the units are in fact ranged in the theory needs a name, and may be called rank.

That is, Fawcett critiques a model of formal constituency by confusing it with the relation between function and form. This fundamental misunderstanding invalidates his entire argument.

[2] This is not misleading, because it is true. However, what it strategically omits is that the Finite is also an element of the verbal group where it is realised by a word (finite verb).

[3] As previously noted, Fawcett (forthcoming a) is still unpublished, 21 years after the first edition of this publication.

[4] To be clear, Fawcett's solution to this non-existent problem is to mistake a formal unit, his 'Main Verb', for a functional element of the clause.

[5] As foreshadowed by this discussion, the "important evidence" against the concepts of the verbal group and rank scale arise from theoretical misunderstandings on the part of Fawcett. This will be further confirmed in the examination of Section 4 of Appendix C.

Misrepresenting Halliday On Conjunction Groups

Fawcett (2010: 234):

Let us be clear from the start of this discussion that the general concept of 'rank' and the specific concept of 'total accountability at all ranks' are interdependent. In other words, the concept of a 'rank scale of units' makes no theoretical claim if it does not imply 'total accountability at all ranks' — or at the very least 'accountability at all ranks, with only a few justifiable exceptions'.

However, the strict application of the 'total accountability' principle leads to problems with certain classes of word. This has led to what might loosely be termed 'the rank scale debate', and I provide an account of this debate in Appendix C. Here I shall restrict the discussion to just those points that are central to establishing why I myself have abandoned the concept of the 'rank scale' in favour of a different concept.

As Matthews (1966) and others have pointed out, Halliday's principle of 'accountability at all ranks' requires that, in a clause such as after we left Henry's, the word after must be treated as a group, since it is an element of the clause. And the same would be true of the word and in ... and we left Henry's and therefore in we therefore left Henry's. In IFG Halliday introduces the concept of a 'conjunction group' to model structure within a Linker or a Binder, but see Section 1 of Appendix C for a dismissal of this concept as a possible solution to the problem of satisfying the principle of 'accountability at all ranks'. (For a start, it is highly unlikely that Halliday would wish to analyse every one-word Binder as a group with one element.) See Section 10.2.8 of Chapter 10 for how the Cardiff Grammar handles Binders with an internal structure (when they occur), and see Section 11.6.2 for the relevant concept of "variation in depth of exponence".

Blogger Comments:

[1] To be clear, the principle of exhaustiveness holds that everything in the wording has some function at every rank. Halliday & Matthiessen (2014: 84):

The general principle of exhaustiveness means that everything in the wording has some function at every rank (cf. Halliday, 1961, 1966c). But not everything has a function in every dimension of structure; for example, some parts of the clause (e.g. interpersonal Adjuncts such as perhaps and textual Adjuncts such as however, play no role in the clause as representation.

[2] For context, Matthews is on record as describing the Chomskyan revolution as "the best thing that has happened to linguistics in the past 2500 years". Matthews' misunderstandings will be identified in the examination of Appendix C. In the meantime, see The Concept Of Rank: A Reply (Halliday 1966).

[3] This is not misleading, because it is true. In SFL Theory, the word after in this instance serves as the Head element of a conjunction group.

[4] This is misleading, because it misrepresents Halliday's model. The conjunction group does not "model structure within a Linker or Binder". Instead, linker and binder are subtypes of conjunction, and a conjunction serves as the Head of a conjunction group.

[5] See the forthcoming examination of Fawcett's Section 1 of Appendix C for the validity of Fawcett's dismissal. In the meantime, consider the serious misunderstandings identified in this post.

[6] To be clear, this is misleading, because it is the direct opposite of what is true. For Halliday, every "one-word binder" serves as the Head element of a conjunction group.