"The Generalisations That Halliday Gives Up"

Fawcett (2010: 333):

M&M make many valuable points in their "Response to Huddleston". The main weakness in the case that they present in the sections of their "Response" summarised above is this: they do not show how the generalisations that Halliday gives up in order to foreground the similarities between the 'hypotactic' and 'paratactic' analyses are to be handled in the grammar. If M&M are to 'defend' the position taken in IFG successfully, they need to address this question. But perhaps there should be less 'attacking' and 'defending', and more accepting of genuine 'problem examples', together with more exploring of comprehensive solutions to such problems, in the framework of a multi-component model?

Blogger Comments:

[1] To be clear, Halliday does not "give up" any generalisations in "foregrounding the similarities" of hypotaxis and parataxis. On the contrary, the distinction between hypotaxis and parataxis provides a generalisation, the system of TAXIS (interdependency), which Fawcett's Cardiff Grammar lacks. Importantly, the distinction between interdependency and embedding provides a more detailed specification of the different types of grammatical environments in which the generalised meanings of expansion and projection are manifested.

[2] To be clear, it is Fawcett who chooses to construe M&M's 'Response To Huddleston" in terms of 'attacking' and 'defending', the former being his own preferred modus operandi.

[3] To be clear, as previously demonstrated, the 'problem examples' that have been presented by Huddleston and Fawcett are not genuine 'problem examples' when examined with a sufficient grasp of SFL Theory.

Fawcett's Contention That The Notion Of A Rank Scale Depends On Hypotaxis

  Fawcett (2010: 333):

In other words, the analyses of dependent clauses that Huddleston advocates in his review of IFG are indeed possible ones within SFL. But the fact that SF linguists agree that we should explore the value of both 'hypotaxis' and alternative concepts within the theory does not absolve us from the obligation to try to decide whether 'hypotaxis' is needed in all or some or none of the cases for which Halliday proposes its use. The key point for the present debate is that, if we replace 'hypotaxis' by embedding in the examples discussed here (as Huddleston and I advocate), the case for retaining the concept of the 'rank scale' is greatly weakened. Thus it is not the case, as M&M suggest (p. 28), that "Huddleston's objections are descriptive, not theoretical".

Blogger Comments:

[1] To be clear, the theoretical value of hypotaxis is its explanatory potential. For example, with regard to projection, the distinction between hypotaxis and parataxis models the distinction between reports and quotes, and the distinction between hypotaxis and embedding models the distinction between reports and pre-projected facts.

[2] This bare assertion, unsupported by argument, is misleading, because it is untrue. Whether clauses are analysed as ranking (dependent) or rankshifted (embedded) can have no bearing whatsoever on the theoretical status of the rank scale. If hypotaxis were excised from the theory, and all dependent clauses were interpreted as embedded, such clauses can still be interpreted as rank units, shifted to the rank of another unit on a rank scale.

[3] To be clear, the sense in which this is true is that SFL Theory provides both hypotaxis and embedding as potential descriptions of language, and the linguist chooses which of these to deploy in the description.

Criticising Martin For The Wrong Reasons

 Fawcett (2010: 332-3, 333n):

As M&M very fairly point out, "Fawcett (1980), [...] working within a theoretical framework closely related to Halliday's, treats all of Halliday's hypotactic clauses through embedding, the very position which Huddleston espouses." After also indicating that Martin (1988) suggests an analysis of such examples in which the experiential analysis at clause rank is in these terms, they go on: "The theory can thus be seen to accommodate a range of approaches to the question of subordination" (M&M 1991:29). ¹⁹ 

¹⁹ However, I am surprised to find Martin advocating this idea, because it involves an insuperable problem. This is that the second part of his proposal is that there should be a "simultaneous analysis" (showing the 'logical' structure) at the 'rank' of the "clause complex". This is, as we have seen, a different 'unit' on the 'rank scale' from the clause, so that if Martin's idea were to be adopted there would even more serious problems for the concept of conflating these two structures than those already specified in Section 7.4 of Chapter 7. It may be that Martin would now wish to reconsider this proposal.

Blogger Comments:

[1] To be clear, from Fawcett's description, Martin's problematic analysis of he said he'd go would be something like:

[2] This is misleading, because it is untrue, as previously demonstrated. As the term 'clause complex' suggests, the rank at which this unit complex is located is 'clause'.

[3] This is misleading, because it is untrue. As previously explained in the examination of Fawcett's Section 7.4, in SFL Theory, it is only elements of structure that can be conflated, not entire structures. The notion of structure conflation is nonsensical because a structure is the relation between elements. 

Throughout Section 7.4, Fawcett confused the false notion of 'structure conflation' with the notion of the integration of the three metafunction structures through their realisation in a syntagm of clause constituents: groups and phrases.

[4] To be clear, even if Martin were the type to reconsider his own proposals, and capable of doing so, at 71 years, he does not have enough years left him to reconsider all the proposals he needs to reconsider. See, for example, the clarifying critiques of Martin (1992) and Martin & Rose (2007).

Projection As A Hypotactic Relationship, Locution As Phenomenon

Fawcett (2010: 332):

M&M then go on to consider, much more briefly, examples such as the clause he'd go in He said he'd go. For Halliday (and so M&M) this is a case of the 'hypotactic' relationship of 'projection', with he'd go being a 'beta' clause to the supposed 'alpha' clause he said. As we saw at the start of this section, for both Huddleston and me this example has an embedded clause that fills a Complement of the higher clause He said he 'd go. (Strictly speaking, for me it fills the Participant Role of Phenomenon, this being conflated with the Complement.)

Blogger Comments:

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

[2] This is misleading, because it is not true. To be clear, in SFL Theory, projection is not a hypotactic relationship. Projection is a logico-semantic relation, whereas hypotaxis is a type of interdependency.

[3] Strictly speaking, in SFL Theory, if he'd go were an embedded clause, it would serve as Verbiage (the Range of a verbal Process), not as Phenomenon (the Range of a mental Process). However, in SFL Theory, this is a dependent clause, and so a verbal projection, and so a reported locution.

Fawcett's Higher 'Rhetorical Structure Relations' Component

Fawcett (2010: 331-2, 332n):

However, M&M can still point out that the question remains of where, in the overall model, we should express the similarity between (1a) and (3a). My answer is that the place to handle the choice that is realised by these examples is in a higher component of the generation process than the lexicogrammar. This is the component which plans the rhetorical structure relations of the discourse, and so how best to present the relations between any two events. (For the key proposals for this component see Mann and Thompson 1987, and for a useful introductory discussion see Martin 1992.) Indeed, the choice that is realised in (1a), (2a) or (3a) must also be extended to include a realisation such as (4), so that for this reason too it is appropriate to handle it outside the lexicogrammar.¹⁸

(4) He left the room. Then they voted.

¹⁸ In fact, it is also at this stage in generation that the planner needs to consider choosing other conceptually equivalent choices realised in forms which M&M do not mention but which express the same basic temporal relationship of successivity between events, such as They voted after he left the room and After he left the room they voted.

Blogger Comments:

Reminder:

(1a) He left the room before they voted.

(2a) He left the room before the vote.

(3a) He left the room, then they voted.

[1] To be clear, in SFL Theory, the distinction between (1a) and (3a) is the distinction between hypotaxis and parataxis at the rank of clause, and it modelled in the grammar by the system of clause complexing. Halliday & Matthiessen (2014: 438):

[2] To be clear, Fawcett does not locate this higher component anywhere in the architecture of his model (Figures 4 and 12):

[3] To be clear, Martin (1992: 251-64) misunderstands Rhetorical Structure Theory.  See, for example, the clarifying critiques here, here, and here.

[4] To be clear, in SFL Theory, these are all different grammatical manifestations of the enhancement category 'time: different':

• in (1a), it is realised logically through clause complexing: hypotaxis;
• in (2a) it is realised experientially through clause transitivity: circumstantiation;
• in (3a), it is realised logically through clause complexing: parataxis; and
• in (4), it is realised textually through cohesive conjunction.

[5] To be clear, in SFL Theory, the logico-semantic relation between the clauses in both of these hypotactic complexes is analysed as 'time: different: later', with the first complex ordered dominant^dependent (α^β), and the second ordered dependent^dominant (β^α). Note again that Fawcett frames this in terms of a model of text generation by computers, rather than a model of language spoken or written by humans.

Misrepresenting An Argument Against Parataxis As An Argument Against Hypotaxis

Fawcett (2010: 331):

In the second part of their case, M&M suggest a second analysis of Huddleston's example, i.e., as two co-ordinated clauses with ellipsis in the second, thus: He left before the debate or (he left) (at least) before the vote was taken. But again, I am afraid, I have to point out a problem. While their analysis appears at first to be another possible one, we need to take account of the fact that one can insert either to the left of before the debate, so that it becomes He left either before the debate or (at least) before the vote was taken. And this fact demonstrates clearly that the grammar must allow for the possibility of generating the prepositional group and the clause as jointly filling an Adjunct.

Thus M&M suggest that two possible analyses should be allowed (the first being Huddleston's and mine). They therefore do not address the question of which of the two is systemically preferable, and why. Moreover, Huddleston's criticism of IFG stands. In other words, he is right that Halliday's decision to treat all clauses embedded directly in clauses as 'hypotaxis' means that the Sydney Grammar cannot handle examples such as Huddleston's.

Finally, to demonstrate that Huddleston's example is not a 'special case', consider the following example with a simple 'additive' Linker: On average, people died earlier in those days, both from diseases such as diphtheria and because they worked such dreadfully long hours. Notice that here (as in Huddleston's example) the item both prevents the M&M analysis in terms of two co-ordinated clauses with ellipsis. The conclusion, then, must be that we should treat all such 'beta' clauses as embedded clauses. … Withe respect to the areas of the grammar described here, then, M&M fail to rebut Huddleston's critical comments. 

 

Blogger Comments:

[1] To be clear, this is an argument about parataxis, and as such, not relevant to the question of hypotaxis. Moreover, this being the case, it could be equally directed against Fawcett's model of co-ordination.

[2] This is misleading, because it is untrue. On the one hand, Huddleston's criticism does not stand, with regard to hypotaxis, because it is an argument about parataxis. On the other hand, SFL Theory (IFG, Halliday) does not treat all such "embedded" clauses as hypotaxis. By distinguishing between embedding (rankshift) and hypotaxis, SFL Theory provides such explanatory advantages as distinguishing between

• defining relative clauses (embedded) and non-defining relative clauses (dependent), and
• pre-projected facts (embedded) and reported projections (dependent).

The conclusion, then, is the distinction between rankshifted and dependent clauses has explanatory advantages over the lack of distinction.

Misrepresenting Halliday (1994) On Rankshift

Fawcett (2010: 330-1):

Here we shall consider M&M's response to Huddleston's "co-ordination" argument. This begins, perhaps surprisingly, with M&M's agreement that his analysis is indeed a possible one (i.e., the analysis in which before the vote was taken is treated as a clause embedded as an Adjunct). However, they move quickly on from this apparent olive branch to a robust defence of what they take to be the IFG position on this matter — claiming that Huddleston is wrong to criticise IFG for ruling out his analysis, because "there is nothing in Halliday's system to block this analysis" (p. 27). 

But on this matter Huddleston is right and M&M are wrong, as is shown by Halliday's own specification of the types of rankshift that his model permits. Thus he states (IFG p. 242) that "the relationship of an embedded clause to the 'outer' clause is an indirect one" — i.e., the embedded clause must fill an element of a group, not a clause. (See Section 11.8.5 of Chapter 11 for a summary of Halliday 's statement on this matter.)

Blogger Comments:

[1] This is misleading, because it is untrue. In SFL Theory, if a clause is shifted to the rank of group, then it functions as a group and so serves as ("fills") an element of a clause. For example:

On the other hand, it is when a clause is shifted to the rank of word, that it serves as ("fills") an element of a group. For example:

[2] See the post on Section 11.8.5: Seriously Misrepresenting Halliday On Embedding.

"Two Problems Unaddressed"

Fawcett (2010: 330):

Even though Halliday may be right that language is ultimately "ineffable" (Halliday 1984/88), it seems to me that, as SF linguists, it is our task to carry out the research programme outlined earlier, i.e., to assemble the available evidence; to decide which relationships between examples should be given systemic priority in the model of the lexicogrammar; and to explain our decisions. As we have seen, M&M accept this goal too, recognising that the question is that of "what to treat as the basic agnation" (M&M 1991:25). So how far do they do this? 

On the 'thematisability' evidence for treating (1a) as being like (2a) rather than (3a), M&M simply state that "Halliday points out that the thematic principle is not limited to [elements within] the clause; it is also in operation in the clause complex" (1991:26). But one's inevitable response is that, since they start from a position of commitment the concept of 'hypotaxis', they are bound to take this position, so that this it is not independent evidence in support of their argument. Their reply to Huddleston leaves two problems unaddressed. 

The first is that, if we do not treat both after they voted in (1a) and after the vote in (2a) as Adjuncts, our grammar will require two different rules to model a generalisation that patently invites expression in a single rule. 

And the second is that it is not clear how such a rule can in fact be formalised (unless the 'alpha' and 'beta' clauses are admitted as full elements of an as yet unnamed unit, so that the 'beta' element can be thematised in the same way that an Adjunct is.

Blogger Comments:

Reminder:

(1a) He left the room before they voted.
(2a) He left the room before the vote.
(3a) He left the room, then they voted.

[1] This is misleading. M&M's reply specifically addressed Huddleston's claim that the thematisability of the dependent clause in (1a) was evidence that it is embedded as an Adjunct, demonstrating why, in theoretical terms, why this was not the case.

[2] To be clear, if this is a problem, it is only a problem for Fawcett's Cardiff Grammar, not for SFL Theory. In SFL Theory, the general meaning of 'time' — like all categories of expansion and projection — is realised across several domains of the grammar. In (1a) it is realised through clause complexing, whereas in (2a) it is realised through the transitivity system of the clause.

[3] To be clear, the "as yet unnamed unit" in which a 'beta' clause can be thematised is not so much a unit as a unit complex: the clause complex. Halliday (1994: 57):

Fawcett's Epistemology

Fawcett (2010: 329-30):

The main rhetorical thrust of M&M's "Response to Huddleston" is a forceful rejection of almost all of what he says. Yet often, as we shall see, they do not show why we should reject arguments such as these, even sometimes accepting Huddleston's analysis. 

Instead, their major point is (following Halliday in his 'reply' to Matthews in the first stage of the 'rank scale' debate) that it is a virtue of Halliday's model that it raises questions about grammatical structure, rather than to establish that the IFG approach is 'right' and that the more traditional analysis that Huddleston offers is 'wrong'. This approach is fully justifiable at the exploratory stage, but systemic functional linguists have now had well over a quarter of a century to explore English in the systemic functional framework.

Even though Halliday may be right that language is ultimately "ineffable" (Halliday 1984/88), it seems to me that, as SF linguists, it is our task to carry out the research programme outlined earlier, i.e., to assemble the available evidence; to decide which relationships between examples should be given systemic priority in the model of the lexicogrammar; and to explain our decisions.

Blogger Comments:

[1] To be clear, here again Fawcett makes a bare assertion, unsupported by evidence, about an argument to presented later in order to prime the reader to prejudge the argument when it is finally presented.

[2] To be clear, this is a serious epistemological misunderstanding that pervades Fawcett's work. The notion that theories are right or wrong presumes that 

• there is one true model of phenomena, if only we could find it, and so that
• there are criteria for assessing theories that are themselves independent of theory.

Clearly, theories are contingent on the initial assumptions on which they are developed. Theories can be compared on the basis of explanatory potential, but this too is contingent on the assumptions on which each theory was developed, which include the functions the theory was designed to serve.

On the other hand, what can be assessed as "right" (consistent) or "wrong" (inconsistent) are interpretations of theory.

[3] To be clear, this is a serious misunderstanding. Halliday argues that it is grammatical categories that are ineffable, and in a specific sense. Halliday (2002 [1984]: 303, 306-7):

The meaning of a typical grammatical category thus has no counterpart in our conscious representation of things. There can be no exact paraphrase of Subject or Actor or Theme – because there is no language-independent clustering of phenomena in our experience to which they correspond. If there was, we should not need the linguistic category to create one. If language was a purely passive partner, ‘expressing’ a ‘reality’ that was already there, its categories would be eminently glossable. But it is not. Language is an active participant in the semogenic process. Language creates reality – and therefore its categories of content cannot be defined, since we could define them only by relating them to some pre-existing model of experience, and there is no model of experience until the linguistic categories are there to model it. The only meaning of Subject is the meaning that has evolved along with the category itself. … 

But a language is an evolved system; and evolved systems rest on principles that are ineffable – because they do not correspond to any consciously accessible categorisation of our experience. Only the relatively trivial meanings of a natural language are likely to be reducible to (meta-)words. Fundamental semantic concepts, like those underlying Subject, or Theme, Actor, New, definite, present, finite, mass, habitual, locative, are, in an entirely positive way, ineffable. 

Fudging The Data

 Fawcett (2010: 326):

(I have slightly altered the wording of Huddleston's examples to create 'minimal pairs' that make the relevant contrasts fully explicit.)

(1a) He left the room before they voted. 

(2a) He left the room before the vote. 

(3a) He left the room, then they voted.

Huddleston's grammatical analysis of (la) — which is broadly similar to mine — is to treat it as a single clause in which the embedded clause before they voted functions as an Adjunct that identifies the 'time position' of the event of 'leaving' by relating it to an event that is already known to the addressee (the 'voting' event), in the same way that before the vote does in (2a). Indeed, Halliday and M&M would agree with Huddleston and me that, when the event of 'voting' is nominalised as in (2a), it serves this function and is therefore an Adjunct. So why, we might ask, do they not also treat before they voted in (la) as an Adjunct? Essentially, their approach is to interpret (la) as a relating of two events (rather than as a 'main' event that is located in time by relating it to another event) — and to claim that this 'relating' can be achieved either 'paratactically', as in (3a), or 'hypotactically' as in the second interpretation of (la).

Blogger Comments:

In SFL Theory, the three instances are analysed as follows:

[1] This is misleading, on two counts. On the one hand, Fawcett's slight alteration of the data does not create 'minimal pairs', and on the other hand, it does not make the relevant contrasts fully explicit. Instead, the contrast it makes explicit is one of time: 'earlier' ((1a) and (1b)) vs 'later' (1c), which is the irrelevant to the distinction between hypotaxis and embedding.

[2] To be clear, this is bare assertion unsupported by reasoned argument.

[3] This is not misleading, because it is essentially true.

The Concept Of 'Unit' Has No Rôle In Systemic Functional Syntax

Fawcett (2010: 280):

The first "fundamental category" in "Categories" was that of a 'unit', but this concept, as will by now be abundantly clear, has no role to play in the theory of syntax proposed here, because it is inherently bound up with the concept of 'rank'. The word "unit" is used here, however, as a short form for the concept of 'class of unit'. Surprisingly, the concepts of 'unit' and its partner 'rank' occur only rarely in the recent writings of Halliday and Matthiessen. Yet it is clear that this pair of concepts still provides the general framework for the description of English set out in IFG — just as they underpinned the theory of syntax presented in "Categories".

Blogger Comments:

[1] To be clear, this is nonsensical, because the concept 'class of unit' includes the concept 'unit', just as the concept 'breed of dog' includes the concept 'dog'. Moreover, Fawcett's fear of acknowledging the concept is entirely unnecessary, because 'unit' is only "bound up with" the concept of 'constituency', which can be modelled in ways other than rank (e.g immediate constituent analysis).

[2] This is misleading, because it is untrue. SFL Theory models grammatical form as a rank scale of units and assigns functions (e.g. Token) to the constituents (e.g. nominal group) of each unit (e.g. clause). However, grammatical form is backgrounded in SFL Theory, because, as a functional grammar, rather than a formal grammar, it gives priority to function over form.

The Fourth Fundamental Category Of The Cardiff Grammar: Place

Fawcett (2010: 279):

A fourth category that is required in the present theory is that of place, in the sense of the numbered position (or 'slot') in a unit at which an element is positioned. Interestingly, Halliday refers in passing to 'place' in "Categories", but it is not presented as a significant category, and nor is it given any role in his later work. Yet this concept has come to play an essential role in the generative versions of SF grammar — especially in providing the conceptual framework, with 'class of unit' and 'element of structure', for explaining the phenomenon known as 'raising' (as noted in Section 11.7 of Chapter 11). The first appearance of the concept of 'place' in the sense defined here was in Fawcett (1973/81) and it was later described formally in Fawcett, Tucker & Lin (1993). However, the concept was also used, it appears, in Mann and Matthiessen's computer implementation of Halliday's grammar (as explained in Section 10.4.2 of Chapter 10).

Blogger Comments:

[1] This is potentially misleading, because Halliday (1961) and Fawcett do not mean the same thing by 'place'. For Fawcett (p279) 'place' refers to

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

For Halliday (2002 [1961]: 46):

A structure is thus an arrangement of elements ordered in places. Places are distinguished by order alone: a structure XXX consists of three places. Different elements, on the other hand, are distinguished by some relation other than that of order: a structure XYZ consists of three elements which are (and must be, to form a structure) place-ordered, though they can be listed (X, Y, Z) as an inventory of elements making up the particular structure.

[2] This is misleading. The only "generative versions of SF Grammar" that feature Fawcett's concept of 'place' are those of Fawcett's Cardiff Grammar.

[3] See the previous post The Key Rôle Of 'Place' In Handling 'Raising' Phenomena.

[4] To be clear, Fawcett has produced no evidence that Mann & Matthiessen used his concept of 'place' in the Penman Project. But, in any case, the functionality of an adaptation of a theory to the limitations of computers is not an argument for its incorporation into a theory of human language.

Victory, Defeat And Humility

Fawcett (2010: 276-7):

If after making this small investment of time the reader's decision is that the original approach to, let us say, the 'rank scale' is superior to the approach taken here (in which the nearest equivalent is 'filling probabilities'), it would be helpful to future generations if the reasons could be given for maintaining that approach, in the light of the evidence offered here (and in Appendix C for the 'rank scale'). 

Indeed, we should not assume that the outcome to an exchange of views will necessarily be 'victory' or 'defeat'. One possible type of outcome is what has been called a 'transcending solution', i.e., the emergence from the discussion of a better idea than either party held at the start (so not a mere compromise), and this may well be a possibility in some areas. At other times, of course, we must hope for the humility, in both ourselves and our interactants, to say "Yes, I now see that the weight of the evidence is on your side, so I shall change my position."

It is my hope, then, that the effect of the publication of the work that has been developed in the framework described here will not be to separate further these two versions of the theory, but rather to provide ways of helping the theory as a whole to make the further improvements over the standard model (as represented currently by Halliday 1994 and Matthiessen 1995) that are clearly needed to make the theory stronger and more usable for the new century.

Blogger Comments:

[1] To be clear, this blogger has made a huge investment of time and provided reams of evidence demonstrating that Fawcett's theory is internally inconsistent and based on misunderstandings of Halliday's two theories. It can only be hoped that this will prove "helpful to future generations".

[2] To be clear, here Fawcett frames the relative intellectual value of theories in combative terms of victory and defeat (of a personal position).

[3] To be clear, the close examination of Fawcett's Cardiff Grammar on this blog suggests that this is not a possible outcome.

[4] The reader is invited to consider whether or not Fawcett has himself in mind in his hope for humility.

[5] To be clear, the SFL Community has, for the most part, ignored Fawcett's model, so the question of a further separation of SFL Theory and the Cardiff Grammar is largely irrelevant. Indeed, it could be argued, on the evidence here, that the continued existence of the Cardiff Grammar crucially depends on it being ignored instead of critically examined.

On the other hand, from a personal point of view, this blogger can confirm that the effect of closely examining Fawcett's book has been to deepen his understanding of SFL Theory, its development, its self-consistency, and its explanatory potential.

A Common Framework For Comparing SFL Theory And The Cardiff Grammar

Fawcett (2010: 274-5):

The starting point for the summaries that follow must be the framework for a modern SF grammar that we established in Chapter 3. As Figure 4 in Section 3.2 of that chapter showed, its two principle characteristics are (1) that it consists essentially of the two levels of meaning and form, and (2) that there is at each level (a) a component that specified the potential, and (b) an Output' from the grammar, i.e., the instance at that level.

As we have seen, an additional advantage of this formulation of the model is that it is at a sufficiently high level of generalisation to provide a common framework in which we may compare the Sydney the Cardiff Grammars. Moreover, its ability to provide this common framework is not affected by Halliday and Matthiessen's increasing commitment, culminating in Halliday & Matthiessen (1999), to the idea of having a 'two-level' model of 'semantics' (as we saw in Section 4.6 of Chapter 4). In that model, you will recall, there is both the level of 'meaning potential' that Halliday recognised in the early 1970s as the semantics (e.g., Halliday 1971/73a:41-2), and a level of 'semantics' that is higher than this, roughly equivalent to Martin's (1992) 'discourse semantics'. The proposal that this common ground exists follows directly from statements of Halliday's from the late sixties to the present, such as:

In a functional grammar, [...] a language is interpreted as a system of meanings, accompanied by forms through which the meanings can be expressed" (Halliday 1994:xix).

Indeed, it can be argued that all of the concepts that are required in a modern SF grammar follow from accepting the need to recognise the appropriate 'division of labour' between the two levels of semantics and form in such a model.

Blogger Comments:

Reminder:

[1] To be clear, as previously demonstrated, it is the theoretical architecture of the Cardiff Grammar, as represented in Figure 4, that both invalidates the theory and illustrates the extent to which Fawcett does not understand the dimensions of SFL Theory. For example, Figure 4

• misrepresents the axial relation of realisation (paradigmatic system-syntagmatic structure) as the instantiation relation between potential and instance;
• misrepresents a selection expression as an instance of a system network; an instance of a potential system network is an actual system network; a selection expression can be both potential and instance, as demonstrated by the selection expression for the phoneme /b/: [voiced, bilabial, stop] which characterises both the phoneme as potential and the phoneme in a text;
• misrepresents structure as an instance of realisation rules, despite the structure being specified as the realisation of the realisation rules;
• misrepresents a system network and realisation rules as different levels of symbolic abstraction (meaning and form), despite the fact that both the network and the realisation rules include the features of the same level of abstraction (meaning); and
• conflates content (syntax) and expression (phonology/graphology) as the same level of symbolic abstraction (form).

[2] This is misleading, because it is untrue. Even ignoring all the theoretical inconsistencies embodied in Figure 4, it cannot provide a common framework for comparing SFL Theory with the Cardiff Grammar, if only because the two models assume different principles of stratification: meaning/wording/sounding (SFL Theory) vs meaning/form (Cardiff Grammar).

[3] This is misleading, because it is untrue. As pointed out in the examination of Section 4.6, SFL Theory has a 'one-level' model of semantics (Halliday & Matthiessen 1999) and a 'one-level' model of lexicogrammar (Halliday 1985, 1994, Halliday & Matthiessen 2004, 2014). The motivation for this misrepresentation is Fawcett's desire for his model to replace the SFL model of grammatical structure.

[4] To be clear, while it is true that Martin presents his discourse semantics as a stratum above lexicogrammar, the truth, nevertheless, lies elsewhere. As demonstrated here, Martin's (1992) discourse semantics is largely Halliday & Hasan's (1976) lexicogrammatical cohesion, misunderstood and rebranded as his own invention.

[5] This is misleading because it is untrue. The Halliday quote — which does not appear on the cited page — does not suggest that Fawcett's Figure 4 offers a common ground for comparing his model with SFL Theory. It merely acknowledges that in SFL Theory, lexicogrammatical forms are interpreted in terms of the meanings they realise. Cf Halliday (1994: xvii):

[6] To be clear, in SFL Theory, the 'division of labour' is not between meaning and form, but between meaning, wording and sounding, with wording being the interpretation of lexicogrammatical form (e.g. nominal group) in terms of its function in realising meaning (e.g. Senser).

Fawcett's 2 Reasons Why Adding The Cardiff Grammar Syntactic Representation To SFL Clause Structures Cannot Reconcile The Two Theories

Fawcett (2010: 273-4):

However, we also saw in Chapter 7 that we cannot reconcile the two versions of the theory by simply adding the syntactic representation of the Cardiff Grammar to the 'multiple structures' of the Sydney Grammar, as a way to integrate them in a single structure. 

The first reason is that in the Cardiff Grammar it is simply not necessary to have any such 'intermediate' instantial representation between (1) the selection expression of features that are the output from the system networks and (2) the single, integrated structure that must be the final structural representation of any text-sentence (e.g., as shown in the upper half of Figure 10 in Chapter 7) — a fact that is demonstrated by the successful operation of the computer version of the Cardiff Grammar. 

The second reason why we cannot simply add the Cardiff representation of syntax to an IFG-style 'multiple structure' representation is that there are major (and probably insuperable) theoretical problems for the generative version of a model of language that is intended first to generate a set of five or more different structures for a clause and then, by the application of some type of 'structure conflation' rule that no SF theorist has yet attempted to formalise, to integrate them all into a single structure. 

It seems from the experience of those who have tried (in the Penman Project as reported in Matthiessen & Bateman 1991, and in the early stages of the COMMUNAL Project as described in Fawcett, Tucker & Lin 1993) that it is just not possible to incorporate 'multiple structures' in a generative SF grammar. The clear conclusion is that such grammars should be based on the concept of 'element conflation' rather than 'structure conflation'.

Blogger Comments:

[1] For the detailed arguments that demonstrate that Fawcett's claims in his Chapter 7 are based on multidimensional misunderstandings of SFL Theory, see the 103 posts here.

[2] To be clear, the unquestioned assumption here is that it would be theoretically advantageous to reconcile the Cardiff Grammar and SFL Theory. As this blog has demonstrated, over and over and over, unknown to Fawcett, the Cardiff Grammar is inconsistent — in terms of both theoretical assumptions and architecture — with both SFL Theory and itself.

[3] To be clear, as demonstrated in the examination of Chapter 7, there is no theoretical requirement that the three metafunctional structures of the clause be integrated into any single structure, let alone that of the Cardiff Grammar, which is itself a confused hybrid of form (Main Verb) interpersonal function (Subject, Complement, Adjunct).

[4] To be clear, Fawcett's first reason for why the Cardiff Grammar syntactic model cannot be added to the SFL model of clause structure is that the architecture of Cardiff Grammar does not require the SFL model of clause structure. This is analogous to arguing that the model of alchemy cannot be added to the model of chemistry because the model of alchemy does not require the model of chemistry.

[5] To be clear, Fawcett's second reason for why the Cardiff Grammar syntactic model cannot be added to the SFL model of clause structure is that the SFL model of clause structure has major theoretical problems. However, as demonstrated in the examination of Chapter 7, this problem arises from Fawcett's misunderstanding of SFL Theory, especially his false claims that (i) the metafunctional clause structures are not syntagmatically integrated, and that (ii) the metafunctional clause structures need to be integrated in a single structure. As previously explained, the metafunctional clause structures are integrated in a syntagm of clause constituents.

[6] To be clear, the reason why 'no SF theorist' has attempted to formalise a structure conflation rule is that structure conflation is not a feature of SFL Theory. Here, also, Fawcett misleads by shifting terminology from structure 'integration' to structure 'conflation' to set up his final misleading point in this extract; see [8] below.

[7] To be clear, on the one hand, as previously demonstrated, this misrepresents Matthiessen & Bateman (1991), and on the other hand, any adaptations of theory to the limitations of computers is not an argument about the validity of the theory itself, since the theory describes what humans can do, not what machines can do.

[8] To be clear, on the one hand, this is a non sequitur, because Fawcett has not presented an argument weighing up the relative theoretical values of structure vs element conflation. On the other hand, it is misleading, because it falsely attributes structure conflation to SFL Theory, and falsely claims that element conflation is not a feature of SFL Theory.

Misrepresenting Halliday (1994) On "Filling" Notation

Fawcett (2010: 251-2):

Interestingly, there is an equivalent gap in the Sydney Grammar's notation for representing the outputs from the grammar. This arises from the surprising fact that there is no diagram in IFG — or in the equivalent diagrams in Matthiessen & Bateman (1991) or Matthiessen (1995) — that shows how such a relationship should be represented in the full analysis of a text-sentence. In all of these works each unit is analysed in its own terms, almost as if the way in which they are to be related to the units above and below them in the structure is self-evident and has no complications. Filling is in fact a complex matter, and it very often happens that the possibilities as to what class of unit may fill an element depends, either in absolute or in probabilisitc [sic] terms, on choices in the generation of the unit above. The most obvious example is the restrictions on what may fill the Complements of particular Main Verbs (for which see Fawcett 1996).

Blogger Comments:

[1] This is misleading, because it is the direct opposite of what is true. Halliday (1994: 109) provides the following diagram illustrating both clause experiential function types and their realisations by classes of forms at the rank of group/phrase:

Halliday & Matthiessen (2004: 169) further elaborates the model for all three metafunctional structures:

[2] This is not misleading, because it is true. In SFL theory, each rank provides the entry condition to the systems of that rank, in which the structures of each rank are specified.

[3] This is misleading because, in SFL Theory, formal constituents are related to each other by the rank scale, and the relation between function structures at a higher rank and formal syntagms at the lower rank is specified as realisation.

[4] To be clear, in contradiction of SFL Theory, Fawcett here gives priority to the view 'from below', classes of form that realise functions, instead of the view 'from above', the functions that are realised by forms.

[5] To be clear, in SFL Theory, Complement is an element of interpersonal structure at clause rank, which may be conflated with most, if not all, types of participant in experiential structures. Any restrictions on the class of unit that realises a Complement thus depend on the type of participant with which it is conflated.

The SFL Method Of Sequencing Elements In A Unit

Fawcett (2010: 221):

Systemic functional linguists have explored two ways of locating elements in an appropriate sequential relationship to each other in a unit. The first — which may at first sight appear to be the simplest — is to locate each element in its 'place' by relating it to some previously located element. This approach depends crucially on the existence of what we might term an 'anchor' element in each unit, i.e., the existence of a 'pivotal' element that is always present. It was this method that Halliday used in his seminal first description of a generative SF grammar (Halliday 1969/81). Surprisingly, references to this as a method of sequencing elements in a SF grammar are still found in current descriptions by Halliday (1993: 5405) and Matthiessen (1995: 23-4).

I say "surprisingly" because, when Mann and Matthiessen were working on the large-scale computer implementation of Halliday's SF grammar in the Penman Project in the late 1970s, this approach caused problems. When faced with the additional complexities of building an [sic] large, generative grammar, they found that in practice they had to turn to the second method — to which we shall come in a moment (Matthiessen, personal communication).

Blogger Comments:

[1] To be clear, the first method of sequencing elements is Halliday's theoretical approach, whereas the second is Fawcett's method for adapting SFL Theory to the limitations of computers for the purpose of text generation.

[2] This is potentially misleading. In terms of the realisation statements that specify the sequencing of elements, what is required is the insertion of the elements that are to be sequenced. The notions of 'pivotal' and 'anchor' are irrelevant, as demonstrated by realisation rules like Finite^Subject, where neither element is 'pivotal' or an 'anchor', even in Fawcett's terms, since the 'anchor' or 'pivotal' element of the clause in the Cardiff Grammar is the 'Main Verb' (p201).)

[3] To be clear, the cited instances are as follows. Halliday (1995 [1993]: 272):

(c) 'Order' an element with respect to another, or to some defined location (e.g., order finite auxiliary before subject);

Matthiessen (1995: 23-4):

Structuring statements determine the organisation within one layer of a function structure:

(1) the presence of a grammatical function (insertion; e.g., "insert Subject," symbolised +Subject), 

(2) the relative sequence of two grammatical functions (ordering; e.g., "order Subject Finite," symbolised Subject^Finite), and  

(3) the constituency relationship between two functions (expansion; e.g., "expand Mood Subject", symbolised Mood(Subject)).

[4] The word "surprisingly" here is misleading, because it is the opposite of what is true. The need to adapt a theory of human language to the limitations of computers for text generation would not be surprising to anyone engaged in such a project, unless they were unable to distinguish between theory and its context-specific application.

Fawcett On Classes Of Clause

Fawcett (2010: 201, 201n):

We have noted that, in the "Categories" framework, there is a 'unit' on the 'rank scale' called the "clause". From the 'rank scale' viewpoint it is surely odd that, unlike the 'rank scale units' of 'group' and "word', there is only one class of the "clause", namely the clause. Yet this fact is never commented on.⁷ 

⁷ Clauses can of course be classified in terms of the features in the network that generates them, e.g., as 'independent' or 'dependent' clauses, and as 'action' or 'mental' or 'relational' clauses — but this should not lead one to set up "classes of clause". These differences are quite unlike the distinction between a nominal group and a prepositional group; rather, they are like the distinction between a nominal group that has a modifier and one without one — i.e., they are different from each other, but they are still the same 'class of unit'. If a grammar did set up 'classes of clause' along these lines, there would be as many different classes of clause as there are combinations of semantic features realised in the clause — i.e., millions of millions. Surprisingly, Matthiessen (1995:77) describes the following as "grammatical classes" of clause: 'major' and 'minor' clauses and, within 'major clauses', 'free' and 'bound' clauses. Yet these are simply early features in the sub-network for the clause.

Blogger Comments:

[1] This is misleading. As Fawcett knows, since he has been arguing against it, Halliday classifies units in terms of the functions they serve at a higher rank. Halliday (2002 [1963]: 95-7):

I have assumed, for the purpose of the main points made in the paper, that this category of “class” is to be defined syntactically. By this I mean that the concept is introduced into the description of a language in order to bring together those sets of items that have the same potentiality of occurrence; in other words, sets of items which are alike in the way they pattern in the structure of items of higher rank. 

Clearly, since the clause is the highest ranked unit, it cannot be classified in terms of its functions at a higher rank.

[2] This is not misleading, because it is true. In treating 'major' and 'minor' as classes of clause, Matthiessen (1995: 77-8) is inconsistent with the criteria that Halliday uses to distinguish classes of unit in SFL Theory. Instead of using the scale of rank, Matthiessen has used the scale of delicacy to classify the clause by its features.

In 201 pages, this is Fawcett's first valid critique. Happy New Year!

Misrepresenting Fawcett On Class Of Unit

Fawcett (2010: 197):

Halliday takes a very different approach to the criteria for recognising a 'class of unit'. While the criterion used here is the unit's internal structure (together with semantic criteria, as described in Section 10.2.1), for Halliday the criterion is the unit's ability to fill elements of units at the 'rank' next above it on the 'rank scale' (as we saw in Section 2.3 of Chapter 2). Thus Halliday's definition of 'class of unit' is dependent — like so much else in his theory — on the concept of the 'rank scale'.

It is interesting to note that Halliday later (1963/76) introduced to the theory a concept that he termed "type". It was introduced as a complement to 'class', in a sense that is exactly equivalent to the concept of 'class of unit' as it is used here. In other words, in Halliday (1963/76) a unit's 'type' is defined in terms of a unit's internal structure. Interestingly, while Berry's introduction to the theory gives a clear account of the difference between this concept and Halliday's "Categories" sense of 'class of unit' (1975:76-7, 124-6), Berry makes no further use of it, and the concept has not been used in most later accounts of the theory. Thus it is not mentioned by Halliday in either "Systemic theory" or IFG, nor by Matthiessen (1995). 

Moreover, in both IFG and Matthiessen (1995) units continue to be defined in the "Categories" manner. Halliday is clearly using this criterion when deciding to treat very lucky in You're very lucky as a nominal group (p. 194 of IFG), and Matthiessen emphasises the correlation between "grammatical units of different classes" and their "functional potential [in the unit above]" (Matthiessen 1995:22).

Blogger Comments:

[1] To be clear, this is not misleading, because it is true.

[2] To be clear, this is misleading, because it is untrue. As previously demonstrated, Fawcett's classifies his units differently according to the syntagmatic structure of each unit ('from below'), not on the basis of the meaning it realises ('from above'). For example, as previously discussed, he classifies the groups over sixty and very clever differently — as quantity vs quality group — despite the fact that both congruently realise the same meaning: Attribute.

[3] To be clear, this is slightly misleading. It is not so much that the 'definition' of class of unit is dependent on a rank scale, but that the rank scale provides a grammatical means of modelling the natural relation between meaning (e.g. thing) and grammatical form (e.g. noun). See, for example, Halliday & Matthiessen (1999: 15, 18ff).

[4] This is very misleading indeed. Halliday (1963) only outlines this alternative method in order to identify it as the approach he is not taking — which is why it is not taken up later by Berry, Halliday or Matthiessen. Halliday (2002 [1963]: 95-7):

I have assumed, for the purpose of the main points made in the paper, that this category of “class” is to be defined syntactically. By this I mean that the concept is introduced into the description of a language in order to bring together those sets of items that have the same potentiality of occurrence; in other words, sets of items which are alike in the way they pattern in the structure of items of higher rank. Thus, to take a typical instance from grammar, we may have morpheme classes defined by word structure, each such class being one set of morphemes having a given value in the structure of words: as, for example, the morphemes of inflexion in Latin nouns. Likewise we might have word–classes defined by group structure, or clause–classes by sentence structure.

This use of the term “class”, to name a category defined in some way by its relationship to a higher structure, is by no means universal in linguistics; but it would probably be granted that some such category is necessary to linguistic description whatever name we choose to adopt for it. Syntactic classification (sometimes referred to as “functional classification”, in what is perhaps a rather misleading opposition of “form” and “function”) is a central feature of linguistic method, and one which it seems appropriate to discuss in the present context.

The alternative to this use of the term “class” is to consider morphological classification. Here “class” would be the name given to a set of items which are alike in their own structure: that is, in the way that they themselves are made up of items of lower rank. A word–class would then be a set of words having a certain similarity in their own formation out of morphemes. In this usage there are no morpheme classes, since “morpheme” is the name given to the smallest unit in grammar, which by definition has no structure: its relation to items abstracted at other levels, such as phonemes, is not one of structure, but involves the interrelation of different dimensions of abstraction.

It is important to notice that this is in the first instance a terminological alternative, not necessarily implying a different theory. It is not the case that the linguist has to choose between two different kinds of classification, the syntactic and the morphological; he has in fact to recognise both kinds of likeness. Moreover, the sets of items identified on these two criteria often coincide: we may recognise a syntactic class “noun”, for example, defined as “that class of word which operates as head of a nominal group”, and find that the items grouped together on this criterion will be the same set as would be grouped together on a morphological criterion such as “that type of word which is made up of a stem morpheme followed by a morpheme of case and a morpheme of number”. Indeed other things being equal, it is usually accepted as desirable that the two should coincide: when the linguist is faced, as he often is, with a choice between two descriptions, both theoretically valid and both accounting for the facts, one in which the two assignments coincide and one in which they do not, he will normally, and “intuitively”, choose the former. For example, groups in English such as this morning operate in clause structure both as Adjunct, as in “I came this morning”, and as Subject (or Complement), as in “this morning promises to be fine” (or “I’ve set this morning aside for it”). The syntactic class defined by operation as Adjunct is the adverbial group; that defined by operation as Subject or Complement is the nominal group. Syntactically, therefore, this morning could be assigned to either or both of these classes. Morphologically, however, it clearly resembles other nominal groups (the morning, this man, etc.) rather than other adverbial groups (quickly, on the floor, etc.), and this can be allowed to determine its primary syntactic assignment.

There are, however, clear instances where syntactically defined sets do not coincide with morphologically defined sets; and it would probably be generally agreed that, whatever the status accorded to the latter, the former cannot be ignored. Syntactic likeness must be accounted for. Moreover, even where the two sets do coincide, the criteria on which they have been established, and therefore their theoretical status, is different; and it is desirable that they should not be called by the same name. It seems to me appropriate that the term “class” should be reserved for the syntactic set (the morphological set may then be referred to as a “type”), and I propose to adopt this usage here. It is also true, in my opinion, that the class thus defined, the syntactic set, is crucial to the whole of linguistic theory, since it is required to give meaning to the basic concepts of “structure” and “system”; whereas the type, or morphological set, is more a descriptive convenience whose theoretical implications are largely internal to itself.

In the remaining sections of this paper, therefore, I should like to discuss two aspects of the syntactically defined set, which I shall refer to henceforward simply as the “class”.

Misrepresenting Halliday & Matthiessen On 'Unit' And 'Rank'

Fawcett (2010: 192-3):

To summarise: the two mutually defining concepts of 'unit' and 'rank' have no explicit role in the theory of SF syntax that is set out here. Moreover, although they are heralded as central concepts in most works that describe the Sydney Grammar, neither Halliday nor Matthiessen make much use of them in practice — either in their theoretical statements or in their descriptions of English (except in their accounts of 'rank shift', where Halliday's preference is now for the term "embedding"). The twin concepts of 'unit' and 'rank' play no part in the operation of the grammar, and the centrality of the concept of 'unit' in "Categories" is replaced by the centrality of the concept of class of unit (as described in Section 10.2 below). The concept of the 'rank scale' is replaced by a statement about the probabilities that a given class of unit fills a given element of the same or another class of unit, as discussed in Section 11.2 of Chapter 11 and as exemplified in Appendix B.

Blogger Comments:

[1] This is misleading, because, as previous posts have demonstrated, Fawcett explicitly ranks his units on a scale from higher to lower.

[2] This is very misleading indeed. In SFL Theory, the rank scale provides the formal units that are modelled in terms of their functions in realising meaning. Paradigmatically, each unit on the grammatical rank scale is the entry condition for the systems that are realised as structures. Syntagmatically, each function structure at a higher rank is realised by a syntagm of forms at the lower rank. For example, a clause structure such as Sayer ^ Process ^ Verbiage ^ Location is realised by the syntagm nominal group ^ verbal group ^ nominal group ^ adverbial group.

[3] To be clear, if there is a class of unit, there is a unit of which there are classes.