Misrepresenting Halliday's Earlier And Later Grammatical Theories As A Single Theory Of Syntax

Fawcett (2010: 92):

At this point in our exploration of Halliday's presentation of his ideas about syntax, we might reasonably conclude that Halliday has completely changed the "basic concepts" of his theory. This, however, would be a mistake, as we shall see in the next chapter. Indeed, before we draw our final conclusions about what sources it will be useful to consult in building a theory of syntax for a modern systemic functional grammar — and so before we draw our final conclusions about the extent to which the concepts of "Categories" are still valid today — we must bring certain other bodies of work into the picture. These are: (1) the theoretical concepts that underlie IFG, since this work constitutes the major manifestation of the 'text-descriptive' strand of the work in the Sydney Grammar framework, and (2) the fullest statement of the requirements of a theory of SF syntax yet made, i.e., Fawcett (1974-6/81), together with the subsequent revisions to that work. 

Blogger Comments:

[1] This is misleading.  As previously noted, Halliday has explicitly explained (1985, 1994: xiv) why his grammatical theory is not a theory of syntax.  In misrepresenting Halliday, Fawcett is falsely presenting his own theory, of syntax, as consistent with Halliday's theory.

[2] This is misleading, since it misrepresents two distinct theories, Scale and Category Grammar (Halliday 1961) and Systemic Functional Grammar (Halliday 1993), as a single theory.

[3] To be clear, the theoretical concepts that have been carried over into the current theory (Systemic Functional Grammar) from the previous theory (Scale and Category Grammar) are outlined in the work Fawcett has just reviewed (Halliday 1993).

[4] To be clear, the theoretical concepts that underlie IFG are outlined in the work Fawcett has just reviewed (Halliday 1993).

Misrepresenting Halliday (1993) On Delicacy, Realisation And Rank

Fawcett (2010: 92):

As for the three 'scales' found in "Categories", the concepts of 'delicacy' and 'exponence' (the latter now renamed "realisation") have changed as a natural consequence of the elevation of 'system' to model 'meaning potential'. And the term 'rank' (which has no meaning without 'unit', in its "Categories" sense) is re-interpreted in "Systemic theory" as a general statement about 'flat tree constituency', with no statement at all about the concepts of a 'rank scale' and the associated limitations on 'rank shift'. According to "Systemic theory", then, it would appear that all of the concepts that are presented in "Categories" as "fundamental" have either been dropped or been changed — many quite drastically. 

Blogger Comments:

[1] This is misleading, because it is untrue. The concepts of delicacy, exponence (as realisation) and rank have not changed across the two theories, as Halliday (1995 [1993]: 273) makes clear:

Systemic theory retains the concepts of 'rank,' 'realisation,' and 'delicacy' from scale and category grammar. 'Rank' is constituency based on function, and hence 'flat,' with minimal layering; 'delicacy' is variable paradigmatic focus, with ordering from more general to more delicate; 'realisation' (formerly 'exponence') is the relation between the 'strata,' or levels, of a multistratal semiotic system — and, by analogy, between the paradigmatic and syntagmatic phases of representation within one stratum. But in systemic theory, realisation is held distinct from 'instantiation,' which is the relation between the semiotic system (the 'meaning potential') and the observable events, or 'acts of meaning,' by which the system is constituted.

[2] This is misleading, because it is untrue.  As previously demonstrated, Fawcett's mistaken notion of 'system' being "elevated" to "meaning potential" arises from his (motivated) confusion of language as potential with meaning as stratum (semantics).

[3] This is misleading, on two counts: because it is untrue, as previously demonstrated, and because it is irrelevant (a red herring).  Moreover, the implication here is that new theories should be consistent with the theories that they replace.  In this way of thinking, quantum mechanics should be consistent with classical mechanics, and both should be consistent with the physics of Aristotle, and so on.

On The Differences Between Scale And Category Grammar And Systemic Functional Grammar

Fawcett (2010: 91-2):

Let me now summarise the differences between "Categories" and "Systemic theory". They could hardly be greater. The fact is that "Systemic theory" presents an almost completely new set of "basic concepts". These are: 'system' and 'system network' (but both in the 'meaning potential' sense), 'instantiation', 'selection expression', 'realisation' and 'structure' (the last being used in a highly generalised sense that is quite different from its precise sense in "Categories"). Thus the list of "basic concepts" in "Systemic theory" does not mention two of the four original "fundamental" categories at all ('unit' and 'class'), and the two that are included as "basic concepts" now have significantly different meanings ('system' and 'structure'). The concept of 'element' is referred to, as we have seen, but it is not presented as a "basic concept", and it has a significantly different sense from that of the term "element" in "Categories". 

Blogger Comments:

[1] To be clear, "Categories" (Halliday 1961) and "Systemic theory" (Halliday 1993) expound two different theories: Scale And Category Grammar and Systemic Functional Grammar, respectively.  Fawcett's comparison of them in setting up the context for his own theory exemplifies the rhetorical strategy known as the red herring logical fallacy.

[2] To be clear, as previously demonstrated, Fawcett confuses meaning potential (language as system) with meaning as level of symbolic abstraction (the semantic stratum).

[3] This is misleading, because it is untrue.  The sense of 'structure' is the same in both theories, but it is expanded in Systemic theory, due to the innovation of the notion of 'metafunction'.  In the earlier theory, structure is limited to what will become interpersonal structure in the later theory, whereas it is supplemented with the structures of the other metafunctions in the later theory.

[4] To be clear, the terms 'unit' and 'class' are formal categories, whereas Systemic Functional Grammar gives priority to function, as Halliday (1995 [1993]: 272) makes clear. On the other hand, the two concepts are inherent in rank, which is listed as a basic concept (op. cit.: 273).

[5] This is misleading, because it is untrue.  As Halliday (1995 [1993]: 272) explains:

[Systemic theory's] immediate source is as a development of scale and category grammar. The name 'systemic' derives from the term 'system; in its technical sense as defined by Firth (1957); system is the theoretical representation of paradigmatic relations, contrasted with 'structure' for syntagmatic relations. In Firth's system-structure theory, neither of these is given priority; and in scale and category grammar this perspective was maintained. In systemic theory the system takes priority; the most abstract representation at any level is in paradigmatic terms. Syntagmatic organisation is interpreted as the 'realisation' of paradigmatic features.

[6] This is misleading, because it is untrue.  On the one hand, 'element' features in the basic concept 'realisation statements', and on the other hand, it is used in precisely the same sense as in the earlier theory.  As previously explained, Fawcett's misunderstanding arises from confusing (structural) 'element' with (rank) 'constituent.

On Halliday's And Fawcett's Versions Of Halliday's Theory Sharing The Same General Framework

Fawcett (2010: 91):

There would be widespread agreement among systemic functional linguists — and especially among those who are interested in the theoretical-generative strand of work in SFL — that "Systemic theory" provides an excellent (though necessarily highly compressed) summary of the essential concepts of Halliday's SF grammar. Indeed, the model described in Chapter 3 and summarised in Figure 4 (in Section 3.2 of that chapter) can be seen as an alternative statement of broadly the same set of concepts — subject to the qualifications expressed above and in Section 4.7 of Chapter 4. 

From the viewpoint of the topic of this book, this difference is not crucial, since we are focussing here on the theory of syntax, i.e., the theory of both the potential and the instances at the level of form. In Section 4.6 of Chapter 4 we established that the difference between the levels of the system networks in the Sydney and the Cardiff versions of SFL, while significant in some cases, did not invalidate the view that the two share the same general framework, and this view is supported by the broad similarity between the 'realisation operations' in the two frameworks that we have noted. This means that we are indeed in a position to make a direct comparison between the theories of syntax presented in each of the two theories. 

However, from this last perspective "Systemic theory" has one great weakness. This is that it does not provide a specification of the "basic concepts" of the part of the theory whose task it is to account for the status of the instances at the level of form. It is these concepts with which IFG is concerned.

Blogger Comments:

[1] This is misleading, because it is untrue.  As previously demonstrated, Fawcett's model (Figure 4) is not only inconsistent with Halliday's theory, it is invalidated by its own internal inconsistencies, including the confusion of axial realisation with instantiation.

[2] To be clear, a systemic-functional theory, by definition, gives priority to system and function over syntagm and form.  In Halliday's theory, grammatical form is modelled as the rank scale, with each rank as the entry condition for grammatical functions.

[3] To be clear, Fawcett's potential at the level of form includes functions in realisation rules, and his instances at the level of form are the realisation of these functions in structures.

[4] This misleading.  The view that Fawcett's theorising is of the same general framework, SFL, is not at issue.  What is at issue is if Fawcett's version of Halliday's theory is itself valid, both in terms of SFL theory and in terms of internal consistency.

[5] To be clear, in Halliday's version of his own theory, realisation statements are located in system networks at the levels (strata) of semantics, lexicogrammar and phonology.  In Fawcett's version of Halliday's theory, realisation operations are located only at one of two levels, the level of form, which is, incongruously, a lower level of abstraction than the systemic features to which they apply.  Moreover, Fawcett's realisation operations are incongruously held to specify instances, rather than realisations.

[6] This is misleading, because it is untrue.  Halliday's theory is not a theory of syntax.  Halliday (1985/1994: xiv):

[7] This is misleading, because it is untrue.  To be clear, 'instances at the level of form' do not feature in Halliday's theory, and so their theoretical status in Halliday's model does not need to be accounted for. However, since Fawcett's 'instances of form' translates to 'structures' in Halliday's theory, Halliday (1993) identifies syntagmatic structure as the realisation of paradigmatic system, with the basic concept of realisation statements in systems specifying structural realisations.  Halliday (1995 [1993]: 272):

The system has one further component, namely the 'realisation statement' that accompanies each option. This specifies the contribution made by that option to the structural configuration; it may be read as a proposition about the structural constraints associated with the option in question.

Falsely Presenting A Non-Sequitur As A Conclusion Of Two False Propositions

Fawcett (2010: 90-1):

Let us now summarise the place in "Systemic theory" of the more specifically 'structural' concepts from "Categories". The categories of 'unit', 'class (of unit)' and 'element' are not included in the presentation of the "basic concepts". Moreover, while the term "element" is used in presenting the realisation statements, it has a different sense from that in "Categories". 

On the other hand, "Systemic theory" includes a set of seven 'realisation operations'. While the latter are related to the "categories" that are missing in "Systemic theory" — in the sense that they generate the structures that exemplify the missing categories — the relationship is not self-evident. In Section 9.2.1 of Chapter 9 we shall see exactly how a revised set of realisation operations can generate all of the specific categories and relationships that are needed to specify the instances at the level of form in a modern systemic functional grammar.

Thus two sets of concepts are required in a full theory of syntax: (1) the theoretical concepts that specify the syntax potential, and (2) the theoretical concepts that specify the instances. The theory of 'syntax potential' will be presented and discussed in Chapter 9 of Part 2, and the theory of 'instances of syntax' will be set out in Chapters 10 and 11.

Blogger Comments:

[1] This is misleading, because it is untrue.  As previously demonstrated, Halliday's term 'element' retains the same meaning across the two theories, and Fawcett's misunderstanding on the matter arises from confusing 'element' with 'constituent'.

[2] This is misleading, because it is untrue.  As previously demonstrated, the "categories" that Fawcett mistakenly thinks are missing in Halliday (1993) are those that serve as the entry conditions to the systems at each rank: clause, group/phrase, word and morpheme.

[3] To be clear, Fawcett's 'instances of form' (in Figure 4) refers to 'one layer of a richly labelled tree structure'.  As previously noted, this confuses instance (of potential) with the structure (realising system).

[4] This is misleading, because it falsely presents this non-sequitur as a conclusion entailed by the preceding discussion.

[5] To be clear, Fawcett (p172) identifies 'syntax potential' as part of 'form potential', which, in turn, he (p36) identifies as 'realisation rules/statements' (Figure 4).  However, it turns out (p175) that 'realisation operations' and 'potential structures' are the two parts of 'syntax potential'.  This compositional inversion will be examined further in the analysis of Fawcett's Chapter 9.

Misrepresenting Halliday (1993) On Realisation And Instantiation

Fawcett (2010: 90):

Let me summarise the "basic concepts' of "Systemic theory". These are: 'system', 'system network', 'selection expression', 'realisation' and 'structure' — the latter, however, only being used in a highly generalised sense. In addition, Halliday makes a fundamental distinction between 'realisation' and 'instantiation', exactly as we have done as in Chapter 3. However, he then he goes on to blur the distinction by saying that the term "realisation" is not only used for "the relation between strata" but also, "by analogy", for "the relation between the paradigmatic and the syntagmatic phases of representation within one stratum". I find this an unfortunate formulation, because it suggests that there is an immediate relationship between the system networks ("the paradigmatic [...] phase") and the output structure ("the [...] syntagmatic phase"). In other words, this way of describing matters overlooks (1) the relation of instantiation between the system networks and the selection expression, (2) the selection expression itself, (3) the realisation rules (which are triggered by the features in the selection expression), and finally (4) instantiation relation between these and the output structures that they generate. In other words, at this point in "Systemic theory" Halliday's second view of 'meaning' (as described in Section 4.6 of Chapter 4) appears to be dominant — i.e., the one in which the system networks are assumed to be at the same level as the final output structures. Apart from this short passage, however, the theoretical model of language presented in "Systemic theory", with its two components of the system networks and the realisation statements, is essentially the same as the general systemic functional model proposed in Chapter 3.

Blogger Comments:

[1] To be clear, Halliday (1995[1993]: 272-3) presents the system, which includes realisation statements and selection expressions, as the basic concept of Systemic theory, and lists rank, realisation (stratal and axial), delicacy, instantiation and metafunction as the other basic concepts.  Structure, on the other hand, is only mentioned in relation to metafunction:

These metafunctions define the dimensions of semantic space; and since they tend to be realised by different structural resources — experiential meanings segmentally, interpersonal meanings prosodically, logical meanings in iterative structures, and textual meanings in wavelike patterns …

[2] This is misleading, because it is untrue.  As previously demonstrated, Fawcett's model (Figure 4) confuses the realisation relation between system and structure with the instantiation relation between potential and instance; see further below.

[3] This is misleading, because it is untrue.  It is Fawcett who "blurs the distinction", not Halliday.  The relation between the paradigmatic axis (system) and the syntagmatic axis (structure) is realisation, not instantiation.  Instantiation, on the other hand, is the relation between potential (system) and instance (text).

[4] To be clear, this is neither unfortunate, nor a suggestion.  It is a statement about the architecture of Systemic theory: paradigmatic systems are realised by syntagmatic structures.

[5] These four claims are misleading, because they are untrue.

• A selection expression is not an instance of system; it is the bundle of features that define a unit, whether as potential or instance, as in the case of the features [voiceless, velar, stop] defining the phoneme /k/, as potential or instance.
• As such, a selection expression is irrelevant to both instantiation and axial realisation.
• Realisation statements are activated by feature selection, not selection expressions.  Here Fawcett is presenting his own logically-inconsistent model as the benchmark for assessing Halliday's.
• The relation between realisation rules and the structures they generate is, as the name implies, realisation, not instantiation.

[6] This is misleading, because it is untrue.  Halliday has one view on meaning.  As previously demonstrated, Fawcett's misunderstanding on the matter derives from his confusing meaning potential (language as system) with meaning (the stratum of semantics).

[7] This is misleading, because it is untrue.  Moreover, it is contradicted by Fawcett's own report (above) of Halliday (1993):

"realisation" is not only used for "the relation between strata" but also, "by analogy", for "the relation between the paradigmatic and the syntagmatic phases of representation within one stratum".

Clearly, Fawcett does not understand that realisation is a relation between different levels of symbolic abstraction.

[8] This is very serious misrepresentation indeed.  As previously demonstrated, Fawcett's model (Figure 4) is not only inconsistent with Halliday's theory, it is internally inconsistent, as shown again here, where Fawcett presents selection expressions as instance of systems, and syntagmatic structures as instances of realisation rules.

Misrepresenting Halliday On Semantics

Fawcett (2010: 89-90):

One reason may be that he was not sufficiently confident of its place [i.e. the semantic stratum] in the overall theory at the time of writing "Systemic theory" to give it this status. Another possible reason may be that he limited himself, in what was necessarily a short paper, to just those concepts that he believed to be common to all 'dialects' of SFL. In other words, he may have omitted the concept of a 'higher semantics' on the grounds that some other systemic functional linguists (including those working in the framework of the Cardiff Grammar) consider that the existing system networks (or replacements for them that are more explicitly semantic) are all that is needed to model those aspects of 'meaning' that it is appropriate to model as lying within language. Either way, the absence of this concept from this key summary of the theory seems to signal that at the time of writing Halliday was less confident of its centrality in his view of language than he appears to have become in subsequent works, such as Halliday & Matthiessen (1999).

Blogger Comments:

[1] This is misleading, because it is untrue.  The "subsequent" work, Halliday & Matthiessen (1999), was first conceived between 1980 and 1983 when Matthiessen was a research assistant at the Information Sciences Institute at the University of Southern California, more than a decade before Halliday (1993). Halliday & Matthiessen (1999: ix):

This book was conceived dialogically: it started as notes on discussions between the two authors when CM was working at the University of Southern California's Information Sciences Institute and MAKH was visiting there as a consultant.

 Moreover, Fawcett was the series editor for the publication.

[2] This is misleading.  Here Fawcett has falsely attributed his own view on the matter to Halliday.

[3] This is misleading, because it is untrue, as demonstrated in the previous post.

Misrepresenting Halliday (1993) On Semantics

Fawcett (2010: 89):

We have already noted several surprising omissions of "Categories" concepts from the basic concepts of "Systemic theory". I shall now identify another omission — though it is one of a different sort. You will recall, from Section 4.6 of Chapter 4, that in recent years Halliday has shown an increasing commitment the view that we should recognise an additional layer of 'meaning potential' — i.e., a 'semantics' above the level of 'meaning potential' that is represented in the system networks for TRANSITIVITY, MOOD and so on. We called this the 'two-level' model of meaning. Interestingly, Halliday does not include this concept in "Systemic theory". If it is as central to his view of language as some of his recent writings suggest (e.g., 1996:29), why, one wonders, has it been left out?

Blogger Comments:

[1] See the preceding posts for the invalidity of this claim.

[2] This is misleading, because it is untrue.  As previously demonstrated, the stratification of the content plane into semantics and lexicogrammar has been a feature of SFL theory at least since Halliday & Hasan (1976), if not before.

[3] As previously demonstrated, Fawcett confuses 'meaning potential' (language as system) with the stratum of meaning (semantics).  Here he compounds the misunderstanding by confusing it with the stratum of wording (lexicogrammar).

[4] To be clear, it is the content plane that is stratified, with one level of meaning and one level of wording.

[5] This is misleading.  Halliday (1995 [1993]: 273) writes:

The shift to a paradigmatic orientation led to the finding that the content plane of a language is organised in a small number of functionally defined components which Halliday labelled 'metafunctions.' … The stratal role of the lexicogrammar lies in mapping these semantic components into a unitary construct, one that is capable of being linearised.

On The Strange Omission Of The Vital Operation 'Insert Unit'

Fawcett (2010: 88):

Strangely, one vital operation appears to be missing from Halliday's list in "Systemic theory" — and it is also missing from the closely related lists given in Matthiessen & Bateman (1991) and Matthiessen (1995). This missing operation is 'Insert unit', and I shall comment on the possible reasons for its absence in Section 9.2 of Chapter 9.

Blogger Comments:

To be clear, Fawcett's operation 'Insert unit' corresponds to Halliday's features of the rank scale — clause, group/phrase, word, morpheme — as the entry conditions for more delicate systems at each rank, as will be seen in the examination of Section 9.2 of Chapter 9.  Fawcett's model lacks a rank scale, and so requires this extra 'operation' in his inventory of realisation rules. It will also be seen in that future discussion that Fawcett continues his previously identified confusion of structure with constituency.

Misrepresenting Halliday (1993) By Confusing Rank With Structure

Fawcett (2010: 88n):

You may have noticed that the term "rank" is used in (f) above, but this is not significant. This is because, strictly speaking, Halliday should have used here a term such as "layer of structure" or "unit", since the unit that is 'lower' in the structure is not necessarily of a lower 'rank' (e.g., a clause or a prepositional group/phrase frequently functions as a qualifier in a nominal group).

Blogger Comments:

Reminder:

(f) 'Preselect' some feature at a lower rank (e.g., preselect nominal group);

This is misleading, because it is untrue. Here Fawcett's wording "lower in the structure" confuses the rank scale of forms ("lower") with the syntagmatic axis ("structure").  The form that realises a function is not lower in structure, but lower in constituency.  In the case of a clause realising the Qualifier of a nominal group, the (higher rank) clause is a constituent of a (lower rank) nominal group, which is why the phenomenon is called 'rankshift'.

Misrepresenting Halliday (1961, 1993) And Matthiessen (1995) On Element

Fawcett (2010: 88):

In the Sydney Grammar, it is the word "function" that should, strictly speaking, be used to refer to concepts such as 'Subject' and 'Theme', e.g., in (a), (b) and (c) of Halliday's realisation statements. The term "element" is typically used for the component of the clause into which such "functions" combine. Interestingly, Matthiessen makes no use at all of the term "element", using instead the informal term "bundle of functions". Thus in "Systemic theory" Halliday uses "element" in Matthiessen's sense of "function" — such that the "conflation" (or 'fusion') of two or more such "functions" combines to constitute a single element of the clause, in the way to be described in Section 7.2 of Chapter 7. It is this unified sense of "element" that corresponds most closely to the meaning of the term "element" in "Categories". It may be thought that this is not a major difference, but it is nonetheless a significant one, because it reflects the addition to the theory of the concept that an element may carry several meanings at the same time — this being the third of the major developments in the theory that we noted in Chapter 4.

Blogger Comments:

[1] This is misleading, because it is untrue.  Subject and Theme are each elements of function structure.  In theoretical terms, this specifies them as located on the syntagmatic axis, whereas the unqualified term 'function' does not.  For example, as Halliday & Matthiessen (2014: 83) point out: 'Theme functions in the structure of the clause as message'.

[2] This is misleading, because it is untrue.  Elements of function structure, such as Subject and Location, map onto clause constituents, such as nominal groups and prepositional phrases.  That is to say, here Fawcett has merely confused the term 'element' (of function structure) with the term 'constituent' (as modelled by the rank scale of form).  Halliday & Matthiessen (2014: 9):

Grammatically, however, the constituent of a clause is not, in fact, a word; it is either a phrase or a word group…

[3] To be clear, the terms 'element' and 'bundle of functions' are not synonymous.  While an element is a structural function such as Theme or Subject or Senser, a bundle of functions are all the structural functions that map onto a clause constituent, such as Theme and Subject and Senser all mapping onto a nominal group.

[4] This is doubly misleading, because it is doubly untrue. On the one hand, in Halliday's Scale and Category Grammar (1961), before the theorising of metafunctions, there are no functions to be conflated on a constituent, and on the other hand, Halliday (1961) uses the term in precisely the same way as in Systemic Theory (1993).  Halliday (2002 [1961]: 47):

In the statement of English clause structure, for example, four elements are needed, for which the widely accepted terms subject, predicator, complement and adjunct are appropriate.

Misrepresenting Halliday (1985, 1993, 1994) And Matthiessen (1995) On Element

Fawcett (2010: 87-8):

Halliday's seven types of 'realisation statement' are, in his words:

(a) 'Insert' an element (e.g., insert subject); 

(b) 'Conflate' one element with another (e.g., conflate subject with theme); 

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

(d) 'Classify' an element (e.g., classify process as mental: cognition); 

(e) "Split' an element into a further configuration (e.g., split mood into subject + finite); 

(f) 'Preselect' some feature at a lower rank (e.g., preselect nominal group); 

(g) 'Lexify' an element (e.g., lexify subject : it). 

(Halliday 1993:4505) 

… Earlier, we were considering the fact that the concept of 'element' was not presented as a "basic concept" in "Systemic theory". However, as you can see from the number of instances of the word "element" in boldface in Halliday's realisation statements, this term certainly plays a central role in the process of building 'structure' in his theory. Notice, however, that the term "element" is being used here in a different sense from that in "Categories" — and also from that in which it is typically used in IFG and in Matthiessen (1995:23-5).

Blogger Comments:

[1] This is misleading, because it is untrue.  Despite the fact that the two publications outline different theories,  the use of the term 'element' in Halliday (1993) is the same as its use in Halliday (1961), as the following quote from "Categories" (Halliday 2002 [1961]: 46-7) makes plain:

A structure is made up of elements which are graphically represented as being in linear progression; but the theoretical relation among them is one of order. … A structure is thus an arrangement of elements ordered in places. Places are distinguished by order alone: a structure XXX consists of three places. … In the statement of English clause structure, for example, four elements are needed, for which the widely accepted terms subject, predicator, complement and adjunct are appropriate. These yield four distinct symbols, so that S, P, C, A would be the inventory of elements of English clause structure.

What differs in the two theories is that "Categories" (Halliday 1961) does not yet differentiate between metafunctions, and so the elements of structure are restricted to what later became those of the interpersonal metafunction in Systemic Theory.

[2] This is misleading, because it is untrue.  The meaning of 'element' in both IFG and Matthiessen (1995) is 'structural function'.  For example, Halliday (1985: 32; 1994: 30) writes:

…each clause contains one element which can be identified as its Subject…

It will be seen in the following post that Fawcett confuses (functional) 'element' with (formal) 'constituent'.

On There Being No Important Difference Between The Terms 'Realisation Statements' And 'Realisation Rules'

Fawcett (2010: 86n-7n):

Halliday prefers the term 'statements' to rules', but there is no difference of substance here. Following Hudson's pioneering work on realisation in Hudson (1971), I use the term "realisation rules". Strictly speaking, what Halliday refers to here as "realisation statements" are 'realisation operations', because it is possible for the realisation statement for a given feature to include two or more such 'operations'.

Blogger Comments:

[1] This is misleading, because it is untrue.  On the one hand, Halliday's term 'realisation statement' is consistent with the notion of systems as probabilistic, since probability is a modality of statements. 

On the other hand, Fawcett's term 'realisation rules' construes commands, the agnate modality of which is obligation. This is inconsistent with both modelling systems as probabilistic and with the fundamental view of language taken by SFL theory: meaning as choice.

[2] To be clear, this is a non-sequitur.  Fawcett's argument is as follows:

Premiss: a realisation statement for a feature can include two or more operations,
Conclusion: realisation statements should be called realisation operations.

On the one hand, the Premiss is false, because, in Halliday's model, each individual realisation statement constitutes only one 'operation', not two or more.  On the other hand, the reasoning of the argument is invalid because it deploys the type of circular reasoning known as 'begging the question' (petitio principii), wherein the truth of the conclusion is assumed in the premiss, namely: that realisation statements are operations.

On The Location Of Halliday's Realisation Statements In Fawcett's Model

Fawcett (2010: 86-7):

The selection expression is the input to the realisation statements. The function of each of these is to specify, for a given feature in the system network, the operation through which that feature contributes to "the structural configuration" that is being generated. In "Systemic theory" Halliday specifies seven types of realisation statement, his claim being that every such statement conforms to one of the seven types. As we saw in Chapter 3, a theory of syntax must be concerned with how the grammar specifies both (1) the syntax potential and (2) the instances of syntax, i.e., the outputs from the grammar. In terms of Figure 4 in Chapter 3, then, Halliday's 'realisation statements' belong in the box labelled "realisation rules / statements".

I shall now list the set of types of 'realisation operation' given in "Systemic theory". However, I shall leave the full explanation and evaluation of each to Chapter 9 of Part 2, because they are relatively close to the set that is required for this component of a modern theory of syntax for a SF grammar — though the set to be introduced in Chapter 9 set is slightly fuller. They will therefore be explained and evaluated at that point, i.e., in Section 9.2.1 of Chapter 9.

Blogger Comments:

[1] To be clear, here Fawcett has momentarily switched from his red herring argument, of assessing Halliday (1993) in terms of Halliday (1961), to merely restating where Halliday's realisation statements are located in his own model (Figure 4), which, as previously explained over and over (e.g here), is riddled with internal inconsistencies that result from his theoretical misunderstandings.

[2] Here, once again, Fawcett primes the uncritical reader by promising a critique ('evaluation') elsewhere.

Misconstruing Selection Expressions As Instances Of Meaning

Fawcett (2010: 86):

Let us begin with instantiation. In Halliday's words: 

'Instantiation' is the relation between the semiotic system and the observable events, or 'acts of meaning'. (Halliday 1993:4505)

Even a selection expression, which is strictly speaking not "observable", is an 'instance', i.e., an 'instance of meaning', in that it is the set of features that have been chosen on one traversal of the system network. Thus the instance of meaning' chosen in Section 3 of the worked example in Appendix A is: 

[thing, count, plural, student, nearness to performer, un-near]

Blogger Comments:

[1] This is misleading, because it is not true.  A selection expression is the systemic classification of a given unit.  Halliday (1993: 273):

The selection expression constitutes the grammar's description of the item (e.g., the particular clause so specified); it is also, by reference to the network, the representation of its systemic relationship to other items in the language — since the grammar is paradigmatic, describing something 'consists in' locating it with respect to the rest (showing its total lineage of agnate forms).

Importantly, this may be the unit as potential, unit as instance, or unit as somewhere between (as sub-potential or instance type).  For example, the selection expression of a phoneme, such as [voiced, velar, stop], may describe it as phonological potential or as an actual instance when spoken.

To be clear, this misunderstanding is fundamental to Fawcett's model (Figure 4), and, as a misunderstanding, constitutes one of the lines of evidence that invalidates it.

[2] To be clear, here Fawcett confuses language ("observable") with the linguistic description of it (selection expression).  That is, he confuses the data with the model.

[3] To be clear, selection expressions are relevant in Systemic theory wherever there are systems, and thus are not limited to the level of meaning, semantics.  This is only the case in Fawcett's model, where they are misunderstood as instances.

Misrepresenting Halliday (1993) On Realisation

Fawcett (2010: 85-6):

Finally, let us look at what has happened to the term realisation (Halliday's 1966 replacement for the original "Categories" term "exponence"). Halliday originally brought the concept of 'realisation' into use as a result of the elevation of 'system' to model 'meaning potential', as we saw in Figure 4 (in Section 3.2 of Chapter 3). However, the original "Categories" concept of 'exponence / realisation' has now become the concept that denotes the relationship between two levels of language. In Halliday's words:

'Realisation' is the relationship between the 'strata' (or levels) of a [...] semiotic system" (Halliday (1993:4505).

Thus the term has significantly changed its meaning as a result of the elevation of 'system' to model 'meaning potential', just as 'system' itself and 'delicacy' have — but in this case the change of meaning has been marked by a change of name.

Blogger Comments:

[1] To be clear, here Fawcett claims the meaning (Value) of a term 'realisation' has changed despite the fact that it is actually the term (Token) that has changed (from 'exponence'), not the meaning.

[2] This is misleading, because it is untrue.  Halliday adopted the term 'realisation' as the name for the relation between levels of symbolic abstraction, in the first instance, between strata.  As previously demonstrated, Fawcett's notion of 'the elevation of system to model meaning potential' confuses language as potential (system) with the semantic stratum (meaning).  Moreover, the fact that this is irrelevant to the notion of 'realisation' demonstrates that Fawcett does not understand the meaning of this theoretical term.

As the grammar makes plain, 'realisation' is a nominalisation of the intensive identifying Process 'realise' which relates a less abstract (lower level) Token to a more abstract (higher level) Value.

[3] This is misleading, because it is untrue.  Fawcett's own model (Figure 4) says nothing about changes to Halliday's model, actual or imaginary, even when the latter is understood.

[4] This is misleading, because it is untrue.  As demonstrated in previous posts, the terms 'system' (evidence here) and 'delicacy' (evidence here) have not changed their meaning.

More importantly, the meanings of terms in one theory, Scale & Category Grammar (Halliday 1961), are irrelevant to an examination of another theory, Systemic Theory (Halliday 1993).  Fawcett's entire enterprise in comparing terms across theories is an instance of the Red Herring logical fallacy.

Misrepresenting Halliday (1993) On Rank

Fawcett (2010: 85):

Notice that while Halliday introduces the word "rank" in "Systemic theory", the idea that it means a 'rank scale' of units with the associated concept of 'accountability at all ranks' is simply omitted from what is presented as a summary of the "basic concepts" of the theory. One reason why this is particularly surprising is that the concept of the 'rank scale' is still reflected quite strongly in IFG, as we shall see in the next chapter. Nor is there any help on this matter in Halliday (1996). On the other hand, Matthiessen uses the concepts of the 'rank scale' and 'rank shift' quite freely in his Lexico-grammatical Cartography, and he defines 'rank' in the standard "Categories" manner in the book's useful Glossary section (Matthiessen 1995:790). Thus it would be premature to interpret Halliday's failure to foreground the concept of the 'rank scale' in "Systemic theory" as a weakening of his commitment to the 'rank scale'. It is nonetheless a curious omission.

Whatever the reason for this omission …

Blogger Comments:

This is yet another attempt to mislead.  What Halliday (1993: 273) actually says is:

3. Other Basic Concepts

Systemic theory retains the concepts of 'rank,' 'realisation,' and 'delicacy' from scale and category grammar. 'Rank' is constituency based on function, and hence 'flat,' with minimal layering;

Here Fawcett is attempting to cast doubt on the notion of rank, on the basis of the brevity of an entry in an encyclopædia entry.  The reason Fawcett wishes to do so is that his own model does not include a rank scale, and so, by casting doubt, Fawcett primes the uncritical reader for his later dismissal of the theoretical value of a rank scale.

Not Understanding Halliday (1993) On Rank

Fawcett (2010: 84):

We turn now to the concept of 'rank' — the 'scale' which, with 'unit', provides the major organising principle of the grammar in "Categories". The definition that Halliday gives of it in "Systemic theory" is one of the more puzzling sections of the paper. Here he defines 'rank' in a somewhat opaque manner, writing: 

'Rank' is constituency based on function, and hence 'flat', with minimal layering. (Halliday 1993:4505) 

The meaning of "and hence flat" can be clarified by expanding the last part of Halliday's definition to "and hence represented by diagrams that resemble 'flat trees' rather than 'trees with multiple branching'". And such trees naturally have "minimal layering".

Blogger Comments:

To be clear, Fawcett provides no evidence that he understands Halliday here.  By 'constituency based on function, and hence 'flat', with minimal layering' Halliday means constituency based on minimal bracketing (ranked constituents) rather than maximal bracketing (immediate constituents).  Halliday (1985: 23):

 

 

Minimal bracketing at three ranks yields (Halliday 1985: 21):

Misrepresenting Halliday (1993) On Delicacy

Fawcett (2010: 84):

Let us begin with delicacy. I suggested in Section 2.4 of Chapter 2 that this concept belongs essentially with that of 'system' (and so with 'system network'), such that one system is more "delicate" than another if its entry condition is a feature in that other system — and so on, across a chain of such dependencies (as illustrated on a small scale in Figure 1 of Appendix A). It follows, then, that the elevation of the 'system' to model contrasts in 'meaning potential' rather than contrasts between forms brings with it a similar change in the meaning of 'delicacy' — so that 'delicacy' has similarly become a partly different concept. It is therefore not simply "retained", as Halliday says (p. 4505); it has been adapted, just as the concept of 'system' has been adapted. However, there is what we might term a "second order" manifestation of the concept, i.e., its use to describe the relationship between 'primary' and 'secondary structures'. (See Figure 7 in Section 7.2 of Chapter 7 for the presentation of MOOD as having both a 'primary' and a 'secondary' structure, and see Section 10.3.4 of Chapter 10 for a critical discussion of this concept.) With these substantial provisos, then, we may say that the term "delicacy" is used in "Systemic theory" in a roughly similar sense to that which it had in "Categories". 

Blogger Comments:

[1] This is misleading. To be clear, the theoretical meaning of systemic 'delicacy' is Halliday's formulation, not Fawcett's suggestion.

[2] This is misleading because it seriously misunderstands the notion of delicacy.  Because the scale of delicacy is a dimension of systems — semantic, lexicogrammatical, phonological — the stratal location of systems does not change the meaning of systemic delicacy.

[3] Here again Fawcett confuses language as system (meaning potential) with semantics as stratum (meaning).

[4] This is misleading.  To be clear, delicacy is a dimension of system, not structure.  To anticipate the examination of Chapter 7, compare Fawcett's Figure 7 with the delicacy of the MOOD system of Halliday & Matthiessen (2014: 162):

 

Using Halliday (1961) As The Benchmark Of Halliday (1993)

Fawcett (2010: 83-4):

Thus, only one of the four "fundamental" categories of "Categories" appears in the list of "basic concepts" in "Systemic theory" (i.e., 'system') — and even then it has a significantly modified sense. Two are either used informally or presupposed (i.e.,  'element' and 'class (of unit)'), but it is really surprising to find that they are not presented as "basic concepts". And one of the four original "fundamental concepts" ('unit') is completely missing.

How far, then, are the 'scales' of "Categories" still treated as "basic concepts" in "Systemic theory"? In the section headed "Other basic concepts" Halliday makes the claim that "systemic theory retains [from "Categories"] the concepts of 'rank', 'realisation' [his 1966 replacement for 'exponence'] and 'delicacy'" (Halliday 1993:4505-6). But are these the same concepts that we met in the summary of "Categories" in Chapter 2? 

Blogger Comments:

[1] See the preceding 6 posts for the validity of these claims.

[2] See the next 4 posts for the validity of these claims.

Misunderstanding Halliday (1993) On Class

Fawcett (2010: 83):

The fourth "fundamental category" in "Categories" is 'class' — and this concept too is missing from the list of "basic concepts" in "Systemic theory". This absence is equally surprising, because the concept of 'class of unit' is essential to any generative SF grammar. The reason is that the way in which such grammars work is that each pass through the system network builds the structure of a given class of unit. For example, the class of unit that the simple system network in Figure 1 of Appendix A generates is the nominal group, and it cannot generate any other class of unit. The class of unit that is generated in all of the illustrative generative grammars that Halliday presents is the clause (e.g., Halliday 1969/81). "Systemic theory" is oriented to the use of the theory in generation so that the concept of 'class (of unit)' is, in a sense, presupposed throughout, yet the fact is that it is not presented as a "basic concept".

Blogger Comments:

[1] To be clear, the reason why 'class' is not presented as a "basic concept" in Halliday (1993) is because Systemic Theory is "basically" a functional theory, rather than a formal theory, and 'class' is the categorisation of form.

This, of course, does not mean that 'class' is not accounted for by the theory.  SFL models grammatical form as the rank scale, where it distinguishes classes, most explicitly at group and word rank; e.g. nominal vs verbal vs adverbial vs preposition vs conjunction group.

[2] To be clear, as already demonstrated, Fawcett's system network is inconsistent with SFL theory, not least because it construes lexical items as grammatical features of the nominal group.

[3] To be clear, here Fawcett mistakes a rank unit ('clause') for a class of rank unit (e.g. 'adverbial clause').

Deploying A Logical Fallacy In Discussing Halliday (1993) On Unit

 Fawcett (2010: 83):

How, then, have the two other "fundamental" categories from "Categories" fared, i.e., unit and class (of unit)? Amazingly, these two terms are also missing from "Systemic theory". One reason why it is surprising that the concept of 'unit' is not listed as a "basic concept" is that it is a vital part of the 'rank scale' concept — and it is this concept which provides the main organising principle in "Categories" (as we saw in Section 2.3 of Chapter 2). We shall discuss the curious treatment of the related concept of 'rank' shortly, when we ask what status is given in "Systemic theory" to the 'scales' that were introduced in "Categories".

Blogger Comments:

[1] To be clear, here Fawcett is amazed because two terms from an earlier theory, Scale and Category Grammar, are missing from a brief summary of a later theory, Systemic Functional Grammar. This is equivalent to being amazed because the term 'epicycle' from the earlier planetary model of Ptolemy is missing from the later planetary model of Copernicus. Clearly, Fawcett's presumption that the terms of an earlier theory should be included in a later theory is an instance of the 'red herring' logical fallacy. But see [2].

[2] This is very misleading indeed.  In this instance, Fawcett switches from 'term' to 'concept'. Contrary to Fawcett's claim, the concept of 'unit' is included in Systemic Theory, because the concept of 'rank', which is included (Halliday 1993: 273), entails the concept of 'unit', since a rank scale is a scale of units. Readers can judge for themselves whether Fawcett is deliberately trying to deceive his readers.

[3] Here again Fawcett uses the rhetorical strategy of implying a critique ("curious") while promising an argument elsewhere.  "We shall discuss" the misunderstandings that Fawcett deploys "shortly".

Misrepresenting Halliday (1993) On Element

Fawcett (2010: 83):

Perhaps, you might think, the concept of 'structure' in the "Categories" sense would still be reflected in "Systemic theory" through the presence of the term element, with which it is mutually defining? The term "element" does indeed appear in "Systemic theory", but once again it is used in a somewhat different sense (as I shall explain shortly). But the most surprising fact of all about the concept of 'element' in "Systemic theory" is that it is not presented as one of the "basic concepts". It is simply employed in defining other concepts — rather as if it was considered to be a non-theoretical concept whose meaning the reader is expected to interpret without guidance. In "Systemic theory", then, the concepts of both 'structure' and 'element' are used in different senses from those with which they were used in "Categories".

Blogger Comments:

[1] This is misleading, because it is untrue.  Halliday (1993: 272-3) uses 'element' in the same sense: as an element of structure, such as Subject, Finite, Theme, Process etc.

[2] This is misleading in two ways.  On the one hand, Halliday (1993: 272-3) presents 'element' in his discussion of the "basic concept" of Systemic Theory, systems, where it features in realisation statements.  On the other hand, it is not surprising that the concept should be presented this way, especially in a brief encyclopædia article, because what distinguishes Systemic Theory from Scale and Category Grammar (1961) is the 'shift to a paradigmatic orientation' (p273).  (Fawcett's orientation is to structure and form, rather than system and function.)

[3] This is misleading, because it is untrue, as demonstrated above for 'element', and in the previous post for 'structure'.  That is, Fawcett has misrepresented Halliday (1993) on both matters.

Misrepresenting Halliday (1993) On Structure

Fawcett (2010: 82-3):

However, Halliday also uses the term system in a more general sense. In this second sense it denotes 'paradigmatic relations' (or 'options in meaning potential'), and it is in contrast with the term structure, which is itself to be understood in a broad sense that embraces all aspects of syntagmatic organisation at the level of form. But we should notice this is not the sense in which the term "structure" was used in "Categories". There it was used in a quite specific sense, i.e., one in which it was mutually defining with the concept of 'element'. In other words, a second "fundamental category" is now being used in a different sense.

Blogger Comments:

[1] To be clear, this is the same construal of system.  In this case, system is being theoretically located in terms of the axial relation between paradigmatic system and syntagmatic structure, rather than in terms of the instantiation relation between system and instance.

[2] This is misleading, because 'form' is a level in Fawcett's model of syntax, not in Halliday's functional linguistics.  Halliday models form as a rank scale at the level (stratum) of lexicogrammar.

[3] This is misleading, because it is untrue.  In Categories, Halliday  (2002 [1961]: 46) writes

A structure is made up of elements which are graphically represented as being in linear progression; but the theoretical relation among them is one of order.

This is the same sense of 'structure' in Systemic Theory; e.g. the experiential structure of a clause is a configuration of elements, such as Senser ^ Process ^ Phenomenon.  What is true is that the notion of structure is backgrounded in Halliday (1993), since the priority of Systemic Functional Theory is system, not structure (and function, not form).

[4] This is misleading, because it misrepresents a fundamental category in a superseded theory (Scale & Category Grammar) as a fundamental category in the theory that replaced it (Systemic Linguistics).

Misrepresenting Halliday (1993) On System

Fawcett (2010: 82):

Let us begin by looking for the four concepts from "Categories" of 'unit','class', 'structure' and 'system'. We shall find, as we would expect in a theory that has been re-named "systemic (functional) theory", that system is strongly fore-grounded, so let us begin with this concept. In Halliday's words (1993:4505): "the system takes priority," and he defines a system as a set of "options in meaning potential [my emphasis]". He then goes on to show how, because an option in one system can serve as an entry condition to another, large numbers of such systems combine to form a system network. … Thus the first two "basic concepts" of "Systemic Theory' are 'system' and 'system network', where they model the "meaning potential" of a language.

One might be tempted to say that the concept of 'system' has been expanded from what it was in "Categories" rather than changed — if it were not for the vital fact that the system networks of TRANSITIVITY, MOOD, THEME and so on now model choices between meanings, as Halliday's use of the term "meaning potential" indicates.

Blogger Comments:

[1] This is misleading in a way that serves Fawcett's argument.  Systemic Theory is not a renaming of another theory.  It is a different theory.

[2] Here Fawcett continues his confusion of language as meaning potential with meaning as the semantic stratum of language.  To be clear, systems on the stratum of phonology represent the meaning potential of language, but they do not represent "choices between meanings" (semantics).  That is, Fawcett confuses instantiation (potential–instance) with the stratification of content (meaning–wording).

[3] This is very misleading. By 1993, the system networks of TRANSITIVITY, MOOD, and THEME modelled choices between wordings (lexicogrammar); see [2].

Examining Systemic Theory (Halliday 1993) By Comparing It With Scale & Category Grammar (Halliday 1961)

Fawcett (2010: 81-2):

With these issues in mind, we shall turn now to an examination of the concepts presented in "Systemic theory". Then in Chapters 6 and 7 we shall examine those in IFG. …

Halliday begins "Systemic theory" by reminding us that "systemic theory" is short for "systemic-functional theory", and this establishes from the start that the "systemic" approach to explaining language is one that is inherently oriented to 'function' and 'meaning'. The fact that SFG is a different theory of language from S&C is underlined when Halliday describes it (1993:4505) as "a development of Scale and Category Grammar"; indeed, the encyclopaedia in which "Systemic theory" appears contains a separate article on S&C grammar (Butler 1993a).

The question that we need to ask is: "How far do the seven 'fundamental' concepts of "Categories" (as described in Chapter 2 of this book) coincide with the 'basic concepts' presented in "Systemic theory", thirty-two years later?" 

Blogger Comments:

To be clear, Fawcett intends to examine one theory, Systemic Theory (Halliday 1993), not in its own terms, but by comparing it with a different theory, Scale & Category Grammar (Halliday 1961).  It will be seen that Fawcett frames the differences between the two theories as shortcomings of the later theory.

Promising A Critique Of An Imaginary Theoretical Inconsistency

Fawcett (2010: 81):

However, as we shall see in Chapter 7, it seems that two different positions on how a grammar works have been allowed to develop alongside each other in the Sydney version of SFL — and that in this case the two positions are not compatible with each other. This then raises the difficult question of which one is the 'correct' version of the theory — and, in relation to the other version, why it is different and what status it has in the theory.

Blogger Comments:

[1] It is worth mentioning, given how frequently Fawcett uses it as a rhetorical strategy, that Fawcett  primes the uncritical reader by repeatedly making substantial claims for which the evidence is said to lie elsewhere (in an upcoming chapter, in a past or forthcoming publication, etc.).

[2] To be clear, Fawcett's Chapter 7 is concerned with the 'box diagram' representations of metafunctional structures at the rank of clause in IFG. "As we shall see in the examination of Chapter 7", Fawcett's misunderstandings in this regard are multi-dimensional and, as such, very instructive.

Misrepresenting Halliday's Presentations Of Theory

Fawcett (2010: 80-1):

Thus, when Halliday was asked to write an account of 'systemic theory' for the Encyclopaedia of Languages and Linguistics, it was natural that he should draw primarily on the theoretical-generative strand of the theory. However, this leaves unanswered the question of why the aspects of the theory that are relevant to the 'text-descriptive' work that is presented in IFG are so different from the aspects presented in "Systemic theory".

Blogger Comments:

[1] To be clear, what Fawcett previously (p79) described as "most striking" is now described as 'natural'.

[2] This is misleading. On the one hand, all "aspects of the theory" are potentially "relevant to the 'text-descriptive' work", whether theory is being used to describe a language or to analyse text.

On the other hand, the reason for the different presentation of theory in Halliday (1993) and in IFG (1985, 1994) is that these texts are instances of different registers, each realising a different context.  For example, in terms of tenor, the addressee of Halliday (1993) is the general reader, whereas the addressee of IFG is an undergraduate linguistics student taking a course in Systemic Functional Grammar.

Misrepresenting Halliday's IFG

Fawcett (2010: 79-80):

The second strand of work in SFL is the one that is more widely known: this is the text-descriptive strand of work. Like the term 'theoretical-generative', the term 'text-descriptive' is intended to evoke two types of work in linguistics that do not necessarily occur together, but which in SFL tend to be mutually supportive. The first is the description of languages, where the goal is to achieve a functional description of a language with a broad coverage (rather than, for example, the identification of particular syntactic phenomena as part of a search for syntactic 'universals', as in some other theories). … Works such as Halliday's IFG and Fawcett (in press) are also in the text-descriptive tradition.

It is a descriptive linguistics of this type that is needed by many researchers in many fields of 'applied linguistics' for use in the description of texts. … The term "text-descriptive" is therefore intended to cover both of these aspects of what is perceived here as one general strand of work: the text-based description of languages and the description of texts. …

The two editions of IFG (1985 and 1994) and the many spin-off publications are the main testimony to this strand of work in the Sydney framework.

Blogger Comments:

[1] As previously explained, Fawcett's notion of a 'text-descriptive aspect of theory' conflates the focus on instances of language ('text') with the practice of describing particular languages ('descriptive').

	theory	description
linguist as grammarian	language as potential	a language as potential
linguist as discourse analyst	language as instance	a language as instance

The failure to distinguish these "strands" leads to misunderstanding and misrepresentation, as demonstrated below.

[2] This is misleading.  Halliday's IFG uses a theoretical description of one language, English, to demonstrate his theory of language, Systemic Functional Grammar.  That is, in terms of Fawcett's dichotomy of 'theoretical-generative' vs 'text-descriptive', it is both 'theoretical' and 'descriptive', but not 'text', since it is concerned with language-as potential, not with language-as-instance.

[3] To be clear, this confuses the theory–description dimension with the potential–instance dimension.  The description of language-as-instance (text) requires a description of language-as-potential.  This is why linguists working on an undescribed language need to describe the language in terms of theory in order to describe (analyse) texts of the language.