"The Fact That The Coverage Of Groups Is Far Fuller In The Cardiff Grammar"

Fawcett (2010: 270):

One of the major differences between the Sydney and the Cardiff frameworks is the fact that the coverage of groups is far fuller in the Cardiff Grammar than it is in the Sydney Grammar. See especially Tucker (1998) for a definitive description of the quality group and Fawcett (in press) for a fairly full account of all four of the classes of group that are recognised in the Cardiff Grammar. In large measure, it is the evidence from this mass of descriptive detail that has led us to replace the predictions of the 'rank scale' by probabilities as to what classes of unit fills what elements of structure.

Thus there is still considerable scope for the further development of the description of groups in the Sydney Grammar, and it may be that as this happens the over-narrow predictions set out in IFG will be replaced by a more wide-ranging statement — possibly expressed in terms of probabilities, as here.

Blogger Comments:

[1] This bare assertion, unsupported by evidence, is misleading. On the one hand, as previously demonstrated, every group structure of the Cardiff Grammar has an SFL counterpart. On the other hand, the Cardiff Grammar posits additional classes of group, quality and quantity, which, as previously demonstrated:

• are theoretically redundant,
• complicate the description unnecessarily, and
• introduce theoretical inconsistencies.

Moreover, as far as nominal groups are concerned, from the perspective of SFL Theory, the Cardiff Grammar fails to distinguish between embedding and submodification, as previously demonstrated.

[2] This is misleading, because it is not true. Fawcett's "over-narrow predictions" are the limitations on embedding that he falsely ascribes to SFL Theory, as a result of his failure to understand embedding in terms of rankshift and the principle of exhaustiveness, as previously demonstrated.

Seriously Misrepresenting Halliday (1993) And Halliday (1994) On The Rank Scale

Fawcett (2010: 270):

The overall picture that one gets from IFG, then, is that the concept of the 'rank scale' is just as central in IFG as it was in "Categories", but with far stricter conditions on 'rank shift'. The Sydney Grammar handles as 'hypotaxis' what other grammars treat as the embedding of clauses.

On the other hand, we should also recall the apparent diminution of the focus on 'rank' in Halliday's later descriptions of the theory, which we noted when surveying the basic concepts of "Systemic theory" and IFG in Chapters 5 and 6 respectively. As we noted in Section 6.2.2 of Chapter 6, Halliday chooses to make the point, when discussing the concept of 'rank' in IFG, that

the issue is whether, in a comprehensive interpretation of the system, it is worth maintaining the global generalisation, because of its explanatory power, even though it imposes local complications at certain places in the description" (Halliday 1994:12).

In view of the changes to the SF model of syntax set out here, it is tempting to see this statement as an expression of Halliday's willingness to reconsider the concept of the 'rank scale' — though in reality such a change of position seems improbable. Yet it is hard to think of any other interpretation of the omission of the concept of the 'rank scale' from "Systemic theory" (Halliday 1993).

Blogger Comments:

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

[2] To be clear, this is still misleading, because it is still untrue. As previously demonstrated, this misrepresentation arises from Fawcett's failure to understand embedding in terms of rankshift and the principle of exhaustiveness.

[3] To be clear, this is still misleading, because it is still untrue, as previously demonstrated here in the examination of Fawcett's Chapters 5 and 6.

[4] This is misleading, because it misrepresents Halliday (1994: 12). On the one hand, Halliday was specifically concerned with a rank scale of writing systems, not lexicogrammar; and on the other hand, he was more generally concerned with the theoretical trade-off between the explanatory power of a global generalisation and the cost of creating local complications, such as the question as to whether one or two sub-sentence levels are motivated in such a hierarchy:

[5] This is deeply and seriously misleading, because it is the exact opposite of what is actually true. Halliday (1995 [1993]: 273):

The Problem Of Halliday Excluding What Fawcett Recognises As Embedding

Fawcett (2010: 269-70):

However, there is a further problem with Halliday's specification of the permitted types of embedding. This is that it excludes very many of the types of embedding that are recognised in the present grammar. These are shown in Appendix B, where the symbols at the top of the diagram for each class of unit show the elements of structure that it can fill.

It is of course precisely Halliday's purpose to exclude many of these — especially the ways in which a clause may occur within a clause). The reason is that he has changed his mind since "Categories" (as we saw in Section 2.6.1 of Chapter 2), so that he now wishes to handle such cases as 'hypotaxis' (in the way to be described in Section 11.9). 

Secondly, however, his specification excludes many of the ways in which a clause fills an element of a group, as is also shown in Appendix B. Finally, the specification excludes the many ways in which a group may fill an element of a group (or cluster), again as shown in Appendix B. In the last two cases the omissions may in part be due to the great emphasis in IFG on the clause, so that groups are inadequately covered.

 

Blogger Comments:

[1] This is misleading. On the one hand, as previously demonstrated, Fawcett misrepresents Halliday's specifications of the types of embedding because of his failure to understand the rank scale and the principle of exhaustiveness. On the other hand, excluding the types of embedding recognised by the Cardiff Grammar is only a problem for SFL Theory if the types of embedding recognised by the Cardiff Grammar are valid in terms of SFL Theory.

[2] To be clear, Appendix B does not identify the classes of unit that are embedded, merely the elements that may be "filled" (realised) by them.

[3] This is misleading. It is not "Halliday's purpose" to exclude 'ways in which a clause may occur within a clause'. Instead, Halliday makes the distinction between (i) cases in which a rankshifted clause is embedded in a ranking clause, and (ii) cases in which a ranking clause is tactically related to another ranking clause.

[4] To be clear, on the one hand, it is true that IFG (Halliday 1994), in contradistinction to Fawcett's Cardiff Grammar, places more emphasis on the clause than the group. This is because the clause, unlike the group, is the semogenic powerhouse of the grammar. Halliday & Matthiessen (2014: 10):

The clause is the central processing unit in the lexicogrammar – in the specific sense that it is in the clause that meanings of different kinds are mapped into an integrated grammatical structure. For this reason the first half of this book is organised around the principal systems of the clause: theme, mood and transitivity. In Part II we move outward from the clause, to take account of what happens above and below it – systems of the clause complex, of groups and phrases, and of group and phrase complexes; and also beyond the clause, along other dimensions so to speak.

On the other hand, the question of whether or not groups are adequately covered in IFG (Halliday 1994) cannot be resolved by anyone demonstrably incapable of understanding Halliday's explication of the model.

Confusing Layers Of Structure With Layers Of The Rank Scale

Fawcett (2010: 268-9):

He then goes on to list the very small number of types of embedding that he does still allow (p. 242). These are (expressed here in both Cardiff Grammar and Sydney Grammar terms):

the occurrence of either a clause or a prepositional group/phrase (but no other class of unit, so not a nominal group) as:

1. the head in a nominal group or

2. a qualifier in a nominal group (also referred to as a "postmodifier" in some sections of IFG), or

3. the finisher in a quality group (in IFG a "postmodifier" in an "adverbial group" (but note that in IFG there is no provision for a similarly structured quality group with an adjective as its apex).

Halliday then goes on to state categorically that "there are no further types". The above specification of what types of 'rank shift' are permitted is therefore extremely narrow. It provides for cases such as what Jack built and for Jack to build a house as embedded clauses that fill the Subject, but only by filling the head of a nominal group that in turn fills the Subject. And it even provides for rare cases such as by the bridge as the Subject — but again only as the head of a nominal group that fills a Subject. Why, one wonders, should the clause or prepositional group not fill the Subject directly? Halliday simply states his position and gives no reason. Yet this approach introduces an additional layer of structure, which runs against his general approach of reducing the number of layers of structure in the representation to the minimum.²²

²² In a footnote (p. 242) Halliday re-affirms that the embedded clause or prepositional phrase does indeed function as the head of a nominal group — while at the same time stating that in such cases "we may leave out the intermediate (nominal group) step in the analysis and represent the embedded clause or phrase as functioning directly in the structure of the outer clause, as Subject or whatever." I welcome this small concession, and I suggest that there is, in fact, no reason why a clause should not be permitted to fill an element of another clause.

Blogger Comments:

[1] To be clear, in SFL Theory, the Qualifier of a nominal group is an element of experiential structure, whereas the Postmodifier is an element of logical structure. The two do not always conflate, as Halliday & Matthiessen (2014: 392) illustrate:

[2] This is misleading. While SFL Theory does not feature Fawcett's quality group, it does provide this type of structure as a nominal group, as Halliday (1994: 242) makes clear on the very page being cited by Fawcett:

[3] To be clear, in Fawcett's terms, a clause that is embedded in a clause does "fill" an element of clause structure; a phrase embedded in a nominal group does "fill" an element of group structure. The crucial thing that Fawcett does not understand is the basic principle that elements of structure at one rank are realised by units of the rank below. This means that if a clause is embedded in a clause, it is shifted to the rank of group where, like groups, it realises an element of clause structure.

[4] This is misleading, because it is untrue. On the one hand, here Fawcett misrepresents the rank scale — clause, group/phrase, word, morpheme — as layers of structure. In SFL Theory, structures are differentiated according to rank, so that 'layers of structure' are those of just one rank.

On the other hand, this "approach" is merely consistent with the SFL principle of exhaustiveness, which means that 'everything in the wording has some function at every rank' (Halliday & Matthiessen 2014: 84). 

[5] This is misleading, because it is untrue, since "reducing the number of layers of structure in the representation" is not Halliday's general approach. In fact, as previously demonstrated, Fawcett has spent considerable effort unsuccessfully arguing that Halliday's model has too many layers of structure.

Here again, Fawcett has confused layers of structure with the rank scale. It is the rank scale that has a fixed number of layers.  Halliday (2002 [1966]: 119):

By a rank grammar I mean one which specifies and labels a fixed number of layers in the hierarchy of constituents, such that any constituent, and any constitute, can be assigned to one or other of the specified layers, or ranks.

Seriously Misrepresenting Halliday On Embedding

Fawcett (2010: 268):

What, then, is the place of embedding in Halliday's current theory? As we saw in Section 5.3 of Chapter 5, the concept is not mentioned at all in "Systemic theory". However, it is present in IFG — and in a version that is far more restrictive than in "Categories". Halliday's definition is as follows:

Embedding is a mechanism whereby a clause or phrase comes to function as a constituent within the structure of a group [my emphasis], which is itself a constituent of a clause. Hence there is no direct relationship between an embedded clause and the clause within which it is embedded; the relationship [...] is an indirect one, with a group as intermediary. The embedded clause functions in the structure of the group, and the group functions in the structure of the clause. (Halliday 1994:242)

Unfortunately, Halliday does not explain why he thinks it desirable that a clause should not be permitted to fill an element of another clause. This omission is especially surprising, given that this was permitted in his earlier S&C grammar.

Blogger Comments:

[1] This is misleading, because it is untrue. The version of embedding (rankshift) in SFL Theory (IFG) is not at all more "restrictive" than the version in Scale & Category Grammar ("Categories"). This false claim arises from Fawcett's misunderstanding of embedding, as demonstrated below.

[2] This is very misleading indeed, and a very serious misunderstanding of Halliday (1994). In Fawcett's terms, an embedded clause does "fill" an element of clause structure. In terms of SFL Theory, a clause that is embedded in a clause is shifted to the rank of group, and it is units at the rank of group that realise elements of clause structure, such as Senser and Phenomenon in the following example:

Misrepresenting Halliday On Hypotaxis In A Footnote

Fawcett (2010: 268n):

²¹ It sometimes seems as if Halliday introduced the concept of 'hypotaxis' precisely to avoid having to embed one clause inside another — though Halliday nowhere explains why this phenomenon, which most other grammarians recognise as occurring with great frequency in many types of text, should be regarded as a Bad Thing. It is the fact that Halliday's grammar minimises embedding that forces him — or enables him, depending on your viewpoint — to interpret all of the types of direct 'clause within a clause' embedding recognised here as cases of a 'hypotactic' relationship.

 

Blogger Comments:

[1] This is misleading. To be clear, the theoretical value of the concept of 'hypotaxis' is its explanatory potential relative to other approaches. For example, it provides a means of recognising the commonality and difference of reports (hypotaxis) and quotes (parataxis), as well as the commonality and difference of reports (hypotaxis) and facts (embeddings), and non-defining relative clauses (hypotaxis) and defining relative clauses (embeddings).

[2] This is misleading. Halliday nowhere regards embedding as a "Bad Thing", which is why he nowhere explains why he regards embedding as a "Bad Thing". Instead, Halliday distinguishes embedding from hypotaxis because of the explanatory potential of doing so. Fawcett's Cardiff Grammar does not distinguish embedding from hypotaxis, and so falls short in this regard. This is why Fawcett is going to such great lengths to misrepresent Halliday on the subject.

[3] This is misleading, because it confuses reason and result. It is because Halliday distinguishes embedding from hypotaxis that an analysis reveals fewer instances of embedding (relative to the number of embeddings if the distinction is not made).

Misrepresenting Halliday On Rankshift

Fawcett (2010: 267-8):

In "Categories" Halliday referred to the phenomenon of embedding as "rank shift". The concept of 'rank shift' is, of course, directly dependent on the concept of the 'rank scale', i.e., the general expectation that a unit will fill an element of structure of a unit higher on the 'rank scale' than itself. According to "Categories", 'rank shift' down the 'rank scale' is allowed (e.g., a clause can fill an element of another clause or of a 'lower' unit, but 'upward rank shift' is not allowed (i.e., a word cannot operate as an element of a clause, and a morpheme cannot operate as an element of a group or a clause).

Halliday later replaced the term "rank shift" by the more widely used term "embedding", and in my view this second term is indeed greatly preferable. This is because the term 'rankshift' can be interpreted in two ways that are misleading. The first is the implication that something has been "shifted", perhaps from its typical place in a structure to some other place, whereas in fact the unit has not been moved at all. The second possible misleading inference is that there is something unusual about a 'higher' unit filling an element of a 'lower' unit. But for virtually all linguists other than those who work in the IFG framework, embedding is a natural and frequently occurring phenomenon.²¹

Blogger Comments:

[1] This is misleading, because it confuses form (a rank scale of units) with function (element of structure). The rank scale is a hierarchy of formal constituency — clause, group/phrase, word, morpheme — whereas as elements are the functions realised by forms in the structure of a unit of a higher rank.

[2] To be clear, it is not a matter of whether types of rankshift are allowed or not, but of what types of rankshift are consistent with the rank scale model.

[3] This is misleading, because it is the exact opposite of what is true. Halliday explicitly prefers the term 'rankshift' because the term 'embedding' elsewhere includes hypotaxis, which is not a type of rankshift. In the publication most cited by Fawcett, Halliday (1994: 188) explains:

[4] To be clear, the reason why Fawcett prefers the term 'embedding' to 'rankshift' is because (a) he treats hypotactic clauses as embedded (see [3]) and (b) he claims not to use a rank scale along which units can be shifted.

[5] This is misleading, because it is Fawcett's two interpretations that are misleading; see [6] and [7].

[6] This is misleading, because it is untrue. A rankshifted clause is shifted from clause rank to a lower rank; a rankshifted group is shifted from group rank to a lower rank. For example, the embedded clauses in [[what you see]] is [[what you get]] are shifted from clause rank to group rank, where they serve as elements, Token and Value, of the experiential structure of a ranking clause.

[7] This is misleading, because it is untrue. In SFL Theory, there is no suggestion that rankshift is unusual. On the contrary, rankshift is explicitly recognised as a common feature of written MODE where it is a major contributor to the increased lexical density that distinguishes written language from spoken language. Here Fawcett has not only misrepresented Halliday's theory, but additionally misrepresented the linguists who use the theory.

Misrepresenting The Rank Scale And Rankshift

Fawcett (2010: 266, 267):

The concept of the 'rank scale' expresses a hypothesis about the 'natural' depth of layering in a structure (although Halliday himself has never suggested that this was a motivation for introducing it). Thus it predicts that, in the unmarked case, groups will occur at a depth of one, words at a depth of two, and morphemes at a depth of three. However, few natural texts conform to this pattern. …

The crucial notion, then, is not that of embedding in the strict sense of the term, but the amazing ability of human language to construct units that contain other units within them. This ability — of which the embedding of clauses within other clauses and within various classes of group is simply the most salient characteristic — is one of the chief glories of human language. Yet the use of terms such as 'rank scale' and 'rank shift' suggests a view of language in which embedding is regarded as an aberration from the right ordering of language — rather than being, as I believe it to be, a vital contribution to its elegant power to model our physical and mental worlds.

 

Blogger Comments:

[1] This is misleading, because it is untrue. Like any theoretical approach, the rank scale is concerned with explanatory potential, not with what is 'natural'. 

[2] This is misleading, because it is untrue. On the rank scale model, all units are assigned to a specific number of ranks. Halliday (2002 [1966]: 119):

By a rank grammar I mean one which specifies and labels a fixed number of layers in the hierarchy of constituents, such that any constituent, and any constitute, can be assigned to one or other of the specified layers, or ranks.

However, in the case of rankshift, an embedded unit operates ("occurs") at a lower rank. For example, a clause embedded in a clause operates at group/phrase rank; a group embedded in a group operates at word rank; etc.

[3] This is misleading, because it is untrue. On the rank scale model, in all texts, all units are either ranking or rankshifted. For example, a ranking nominal group operates at group rank, whereas a rankshifted nominal group operates at word rank.

[4] To be clear, as previously demonstrated, from the perspective of SFL Theory, all the cases that Fawcett claimed were not instances of embedding were, in fact, either cases of embedding or submodification.

[5] This bare assertion, unsupported by evidence, is misleading, because it is untrue. The terms 'rank scale' and 'rankshift' say nothing about "the right ordering" of language, and do not construe embedding as an aberration. On the contrary, together they provide a highly systematic means of modelling embedding as an integral feature of language.

[6] To be clear, from the perspective of SFL Theory, the increased semiotic potential of rankshift (embedding) arises, for example, from making the meaning potential of the clause available at group rank, and the meaning potential of the group or phrase available at word rank.

Failing To Recognise Embedding

Fawcett (2010: 265-6):

Let us consider an example of this crucial point. Figure 24, which is taken from Fawcett (in press), illustrates the important fact that nominal groups frequently contain other classes of groups within them. In this invented example, I have chosen to illustrate the use of units other than nominal groups, so that the only cases of 'embedding' are the two nominal groups that occur inside (1) the genitive cluster and (2) the prepositional group. It is well known, of course, that nominal groups frequently contain clauses that function as a qualifier ('relative clauses'), and Figure 24 complements this by illustrating the less widely-recognised fact that the richness in the layering of structure in the nominal group comes equally often from the occurrence of groups other than nominal groups.

Thus much of the very great richness in the syntax of English results from the fact that, when we come to fill the elements of the units of English, we very frequently do not take a step down the supposed 'rank scale', and so do not use embedding, in the proper sense of the term.

 

Blogger Comments:

[1] To be clear, from the perspective of SFL Theory, the complexity of the nominal group of Figure 24 arises from three levels of embedding (rankshift) and three instances of submodification. The layers of embedding can be represented as follows: 

| about ten more of [the most expensive of [ [my supplier's] amazingly sweet  apricots [from Iran] ] ] |

That is, 

(a) the nominal group [my supplier's] and prepositional phrase [from Iran] are embedded in the nominal group 

[my supplier's amazingly sweet apricots from Iran]

(b) this nominal group is embedded in the nominal group 

[the most expensive of my supplier's amazingly sweet apricots from Iran]

(c) this nominal group is embedded in the nominal group 

| about ten more of the most expensive of my supplier's amazingly sweet apricots from Iran |

The three instances of submodification, the first with two layers, can be represented as follows:

[2] As previously observed, Fawcett (in press) is still unpublished, 21 years after the first edition of this publication.

[3] To be clear, from the perspective of SFL Theory, the only embedded unit here that is not a nominal group is the prepositional phrase from Iran that serves as the Qualifier in the nominal group my supplier's amazingly sweet apricots from Iran.

[4] To be clear, these other units, quantity group, quality group and genitive cluster, are not units in SFL Theory.

[5] To be clear, from the perspective of SFL Theory, on the one hand, these are not the only cases of embedding — see [1] above — and on the hand, the nominal group in the prepositional phrase ("prepositional group") from Iran is not embedded. Moreover this analysis flatly contradicts Fawcett's own model (p264):

So we shall not say that we have a case of embedding in on the table, where the nominal group the table fills the completive of the prepositional group on the table…

[6] This is misleading because it is untrue. For a theory that models formal constituency as a rank scale, embedding, in the sense of rankshift, is precisely what Fawcett's example illustrates — together with the fact that Fawcett's Cardiff Grammar only sometimes manages to recognise embedding, thereby creating theoretical inconsistency with the more usual occasions when it does not recognise it.

Misrepresenting Halliday (1994) On Embedding (Rankshift)

Fawcett (2010: 265):

By far the most important type of embedding is the embedding of a clause as an element of a higher clause, or a clause as an element of a nominal, quality or quantity group — and so indirectly as an element of a higher clause. …

However, in the present grammar it is not the concept of embedding per se that leads to the richness of syntax, but the grammar's ability to fill many elements of many units by many other units. Sometimes this results in the embedding of a unit inside another of the same class (e.g., a clause within a clause, as described above), but often it is one class of group within another — and so one that would, in IFG terms, be at the same 'rank' as the higher unit.

Blogger Comments:

[1] To be clear, in SFL Theory, when an embedded clause serves as an element of a ranking clause, it is shifted to the rank of group/phrase, the rank whose units serve as the elements of clause structure. 

[2] To be clear, in SFL Theory, when an embedded clause serves as an element of a ranking nominal group, it is shifted to the rank of word, the rank whose units serve as the elements of group structure. The notion of quality or quantity group is unnecessary and inconsistent with  SFL Theory, as previously demonstrated.

[3] To be clear, in SFL Theory, a clause embedded in a nominal group serves as an element of a nominal group, not as an element of a clause — directly or indirectly. For example, in the following instance, the embedded clause that time forgot serves only as the Qualifier element of a nominal group, not as any of the elements of the clause:

[4] To be clear, from the perspective of SFL Theory, this "filling" includes both embedding and submodification; see the following post.

[5] Trivially, an adverbial clause is a class of unit, whereas a clause is just a unit.

[6] This is misleading, because it is untrue. In IFG terms, a (rankshifted) group that is embedded in another group is not at the same rank as the (ranking) group in which it is embedded. The embedded group is shifted to word rank, as explained in [2] above. This is one reason why the term 'rankshift' is more informative than the term 'embedding'.

Misrepresenting Halliday (1994) On Embedding

 Fawcett (2010: 264-5):

The second type of recursion is embedding. This occurs when a unit fills an element of the same class of unit — and also, in a looser sense, when a unit of the same class occurs above it in the tree structure. So we shall not say that we have a case of embedding in on the table, where the nominal group the table fills the completive of the prepositional group on the table (as Halliday would; see p. 242 of IFG). However, if the table occurred in the box on the table, this is embedding in a looser sense of the term, because the nominal group the table fills the completive of the prepositional group on the table, and this in turn fills the qualifier of the higher nominal group the box on the table.

And, in an even looser use of the term, one could refer to any case in which a unit appears lower in the tree than the second layer as 'embedding'. Here, however, I shall normally use the term "embedding" in the sense of the occurrence (direct or indirect) of a class of unit within the same class of unit.

Blogger Comments:

[1] This is misleading, because it is the direct opposite of what is true. That is, Halliday (1994: 242) does not say that a nominal group serving as the Range of a prepositional phrase is embedded; instead, he says that a prepositional phrase (or clause) serving as the Postmodifier of a nominal group (or adverbial group) is embedded:

[2] To be clear, from the perspective of SFL theory, this is an instance of embedding in the strict sense, since the prepositional phrase on the table is shifted to word rank where it serves as the Postmodifier/Qualifier element of the nominal group the box on the table:

[3] To be clear, from the perspective of SFL theory, rankshift is not limited to embedding with the same class of unit. Halliday & Matthiessen (2014: 492) provide examples of rankshifted clauses and prepositional phrases embedded in nominal groups and adverbial groups:

Confusing Word Complexing With Group Embedding

Fawcett (2010: 264):

Occasionally we need to allow for embedding in co-ordinated structures, as in examples such as ten boys and girls (where ten quantifies both the boys and the girls). The linker and is attached to girls (for the reasons given above) and the two nominal groups of (1) boys and (2) and girls jointly fill the head of a higher nominal group whose quantifying determiner is ten.

One notable characteristic of co-ordination is that the semantic and syntactic similarities between two units often result in a partial syntactic parallelism — and that this in turn often leads to ellipsis. Thus, in The thieves have stolen our TV and drunk all my whisky, the two elements of they (and not, it should be noted, the thieves) and have have been ellipted from the second clause. Ellipsis in co-ordinated clauses can become quite complex, as in Ivy is going out with Paul and not Fred. Here, to provide an adequate analysis, we need to reconstruct the ellipted elements, as follows: Ivy is going out with Paul and (she is) not (going out with) Fred. It is often the presence of the Negator not or an Adjunct that alerts the analyst to the presence of an ellipted clause.

Blogger Comments:

[1] To be clear, from the perspective of SFL Theory, in this instance, boys and girls is a ranking — not embedded — extending paratactic word complex realising the Thing/Head of a nominal group.

[2] To be clear, since this is clause rank ellipsis, it is an element of clause structure, the Mood element comprising Subject and Finite, that has been ellipsed in this instance:

On the basis of the first clause, the ellipsed Subject is the thieves, not they. Since the speaker did not select this personal reference item as the Subject of the first clause, it cannot be understood as the Subject of the second clause, without presuming a reference that was not actually made.

[3] The ellipsed Subject here is Ivy, not she; see [2] above.

[4] To be clear, a more reliable diagnostic of clause rank ellipsis is the absence of one or more Mood or Residue elements from clause structure.

Fawcett's 'Co-ordination' Viewed Through The Lens Of SFL Theory

Fawcett (2010: 263-4):

The first type of recursion is co-ordination. Here two or more units fill a single element of structure. It occurs between all units: clauses and all classes of groups and occasionally clusters. Typically the units are (or in the case of the genitive cluster, contain) potential referring expressions, because co-ordination is ultimately not between syntactic units but between mental referents.

Co-ordination is typically marked by an overt Linker, such as and or or, and these two Linkers can occur with all units. (The linker has a lower case "l" when it co-ordinates groups.) A Linker or linker may co-occur with intonational marking or a punctuation mark (e.g., the two commas in Peter, his brother, his brother's wife and their children).

While some co-ordinators (e.g., and and or) occur with all units, each class of unit has its own set, with its own probabilities. Thus there is not a single set of choices in co-ordination for all units, as some grammars imply.

Sometimes the meaning of 'co-ordination' is re-enforced by another linker that introduces the first unit, as in both my wife and myself. This suggests that the L/linker should be treated as an element of the unit that it introduces, rather than as a 'structural signal' that is not part of either unit (cp. Halliday 1966).

Blogger Comments:

[1] To be clear, from the perspective of SFL Theory, Fawcett's co-ordination largely corresponds to paratactic extension (but see a later post for anomalies in this regard).

[2] To be clear, from the perspective of SFL Theory, Fawcett's "mental referents" correspond to ideational meanings — prototypically participants (Halliday & Matthiessen 1999) — and, grammatically, paratactic extension ("co-ordination") obtains between formal units of the rank scale. Fawcett's notion of "mental referents" is inconsistent with the epistemological assumptions of SFL Theory (see, e.g., Halliday & Matthiessen 1999: 416, 426, 442, 603).

[3] To be clear, in SFL Theory, grammar is linguistic content, whereas graphology and phonology are linguistic expression. That is, importantly, they are different levels of symbolic abstraction.

[4] To be clear, in SFL Theory, system choices are functional rather than formal. Here again, contrā SFL Theory, Fawcett takes the view 'from below', identifying systemic features with the expressions of meanings rather than the meanings they express.

[5] This is potentially misleading. In his reply to Matthews' criticisms of the concept of rank, Halliday (2002 [1966]: 125) wrote:

Given an adequate representation of the underlying grammar, there is no need to insist that every element should be assigned constituent status at all; it is quite usual not to recognise intonation features as constituents, and the same considerations could apply, as Matthews points out, provided limitations were stated, to markers such as and and or. I do not know how to specify in a general formation the conditions under which accountability in constituent terms would not be required.

In SFL Theory, the equivalents of Fawcett's L/linkers are analysed as constituents at all ranks; eg clause (Halliday & Matthiessen 2014: 457):

group (op cit: 566):

and word (op cit: 564):

and, at clause rank, they serve as structural Theme (op cit: 107-8).

Fawcett's 'Recursion' Viewed Through The Lens Of SFL Theory

Fawcett (2010: 263):

There are three types of recursive relationship in English: co-ordination, embedding and reiteration. Reiteration is much less central to the grammar of English than the other two, and it is used more in other languages. But do 'recursive structures' in fact occur in language? Strictly speaking, they do not.

Recursion occurs when a choice in the system network leads to a realisation rule which specifies a re-entry to the system network and the choice of the same feature again. What we find at the level of syntax is two types of the 'repetition' of a class of unit and, in reiteration, the repetition of an item (as in He's very very happy.). All three of these are cases of the realisation at the level of form of recursively selecting the same feature in the system networks. The effect of choosing such a feature is to generate a unit alongside an existing unit (this being is co-ordination) or inside another unit (this being second embedding [sic]. In the strict sense of embedding, a unit fills an element of the same class (most frequently a clause filling an element of a clause, as illustrated in Appendix B). In co-ordination the two or more co-ordinated units are typically of the same class — but not necessarily, as we will see in the next section. (We shall recognise a looser sense of 'embedding' in Section 11.8.3.)

It is the relationship of filling that makes possible the first two types of recursion, and the relationship of exponence that enables reiteration to occur.

Blogger Comments:

[1] To be clear, these claims are theory-dependent and represent Fawcett's position on the phenomena.

[2] In SFL Theory, RECURSION is a system that specifies complexing at all ranks. Halliday & Matthiessen (2014: 438) provide a clause rank example:

Importantly, recursion does not necessitate "the choice of the same feature again".

[3] To be clear, from the perspective of SFL Theory, these two types of repetition are both cases of complexing: the first at ranks above the word, the second at word rank.

[4] To be clear, SFL Theory makes a very important distinction between taxis ("co-ordination") and rankshift (embedding). This distinguishes, for example, non-defining relative clauses (taxis) from defining relative clauses (rankshift), and projected ideas (taxis) from pre-projected facts (rankshift).

[5] To be clear, contrary to this strict sense of embedding, Appendix B (p306) presents both a preposition group in Kew and a clause we've seen as elements (qualifiers) of a nominal group.

[6] To be clear, in this later discussion, it is the nominal group of a preposition group that is said to be embedded in a nominal group, thereby allowing him to overlook the prepositional group as an embedded unit. Fawcett (p264-5):

The second type of recursion is embedding. This occurs when a unit fills an element of the same class of unit — and also, in a looser sense, when a unit of the same class occurs above it in the tree structure. So we shall not say that we have a case of embedding in on the table, where the nominal group the table fills the completive of the prepositional group on the table … . However, if the table occurred in the box on the table, this is embedding in a looser sense of the term, because the nominal group the table fills the completive of the prepositional group on the table, and this in turn fills the qualifier of the higher nominal group the box on the table.

And, in an even looser use of the term, one could refer to any case in which a unit appears lower in the tree than the second layer as 'embedding'. Here, however, I shall normally use the term "embedding" in the sense of the occurrence (direct or indirect) of a class of unit within the same class of unit.

[7] To be clear, from the perspective of SFL Theory, the "first two types of recursion", co-ordination and embedding, correspond to taxis and rankshift. However, clause complexes (taxis) do not realise ("fill") and element of a higher structure, though rankshifted units do. Fawcett's third type of recursion, reiteration, corresponds to a paratactic elaborating word complex realising ("expounding") an element of group structure.

Unit Discontinuity As A Result Of Thematisation

Fawcett (2010: 262):

However, discontinuity is most frequently the result of various types of thematisation. Thus in Who were you seen by? the completive (cv) who is an element of the discontinuous prepositional group by [...] who. …

Thus the present theory of syntax provides for all types of discontinuity. See Fawcett (in press) for a fuller account of the various types of discontinuity that occur in English.

Blogger Comments:

[1] To be clear, as previously observed, in Fawcett's model, Subject is always conflated with Theme (p149), and the only structural representation that includes Theme (p148) only presents this conflation:

In SFL Theory, the Theme of the clause Who were you seen by? is the Complement of a prepositional phrase (Fawcett's cv), but the Subject is not Theme:

cf.

[2] To be clear, another type of discontinuity, not discussed here, but like all types, motivated by the textual metafunction, occurs in medio-receptive clauses:

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

Discontinuous Quality Groups Viewed Through The Lens Of SFL Theory

Fawcett (2010: 262):

Within the structure of the nominal group, the same two principles lead to discontinuous quality groups such as the one indicated by the underlined portion of a more important person than he is. Here, the words than he is constitute a finisher in the discontinuous quality group more (t) important (a) [...] than he is (f). Figure 23 (taken from Fawcett in press) illustrates the notation for showing this type of discontinuity. Thus, whenever a line showing the relationship of componence crosses another representation of a relationship (here exponence) we have a case of discontinuity.

Blogger Comments:

To be clear, in SFL Theory, which does not feature quality groups, there is no discontinuity here:

But see also Halliday & Matthiessen (2014: 422-3).

Explaining The Discontinuous Nominal Group

Fawcett (2010: 261):

Typically, the elements of a unit occur in an uninterrupted sequence. Such a unit can be said to be a continuous unit, and its elements are in the simple 'segmental' relationship of adjacency to each other. This conforms to the general principle that, once a performer has begun on the production of a semantic unit — and so on the production of the syntactic unit in which it is realised — it is helpful to the addressee to finish it before starting on another unit.

But sometimes this principle of "Finish the current unit" is in conflict with one or more other principles. Two of these are what we may call the "Get the pivotal element in soon" principle and the "End weight" principle. For a fuller presentation of these principles, which can be compared with those in Leech (1983), see Fawcett (in press).

In an example such as The time has come when you must leave, the nominal group filling the Subject is the time [...] when you must leave. But it is a discontinuous nominal group, with the qualifier when you must leave following the Main Verb leave come. In such examples the combined influence of the "Get the pivotal element in soon" principle and the "End weight" principle is more powerful than the everyday "Finish the current unit" principle, and the clause moves to the pivotal element of the Main Verb come as quickly as it can, letting the semantically 'heavy' qualifier from the nominal group when you must leave follow, probably with its own information unit, i.e., in its spoken form:

|| the time has come | when you must leave ||

Blogger Comments:

[1] To be clear, a functional grammar explains this type of variation in terms of function, rather than formal principles. In SFL Theory, the variation is explained as motivated by the textual metafunction. The nominal group in Fawcett's example is discontinuous in order achieve the desired texture in terms of Theme-Rheme and Given-New.

To explain, the most probable focus of New information in Fawcett's example is leave, and the discontinuous nominal group positions leave in the unmarked location of New information: 

If this option is not taken, then either the unmarked focus of New information falls on come, which is unlikely to be the focus in this context:

or else leave is presented as marked information focus, which carries the additional semantic feature of contrast (Halliday & Matthiessen (2014: 282):

which may have unintended interpersonal implications, depending on tenor.

[2] As previously observed, Fawcett (in press) is still unpublished, 21 years after the first edition of this publication.

[3] To be clear, in the unmarked case, this rendering assigns an additional tonic prominence to come, making it the first focus of information. A more probable rendering would be a single tone group uttered with tone 1:

// 1 ‸ the / time has / come when / you must / leave //

A Complete Representation Of A Text-Sentence At The Level Of Form

Fawcett (2010: 259-61):

Let us now take the example analysed in Figure 18, and add to it the one relationship of exponence that would be generated on the first traversal of the system network, i.e., the exponent of the Main Verb. This is the representation shown in Figure 21.

Let us now imagine that the system network has been re-entered three times more, to generate the three nominal groups shown in Figure 21, i.e. Ike, his shirt and the co-ordinated nominal group and his jeans. In Figure 22 I have also added the symbol for 'filling' to show the relationship between the clause and the topmost element in the structure, i.e., 'sentence' (Σ), and also the clause element Ender. … The result of these additions to Figure 21 is a complete representation of a text-sentence at the level of form.

Figure 22 can be taken as a summary of the four 'core' relationships in syntax, i.e., the relationships between syntactic categories that are the direct result of the application of realisation operations.

Blogger Comments:

[1] To be clear, this system network can only be imagined, since it is not provided anywhere in this publication.

[2] For contrast, compare an analysis of this clause using SFL Theory:

with the Goal/Complement realised by an extending paratactic nominal group complex:

[3] To be clear, the precise realisation operations are not provided by Fawcett, and so must be taken on trust.

Inconsistencies Caused By The Notion Of Exponence

Fawcett (2010: 259):

How far should this useful principle be extended? For example, should we say that the Subject of a clause that is ultimately expounded by a single pronoun such as he is to be directly expounded by it? Clearly, few linguists would wish to do this, but the criteria for not doing so are rarely stated clearly. 

There are in fact several reasons for not extending the principle to such cases. The first is the sheer centrality of the choices in the nominal group in the grammar as a whole. It is the unit that is used for referring to 'objects', and within it we need to be able to choose between the three major ways of referring to objects that are exemplified by my friend, she and Ivy respectively. The second reason is the sheer frequency of all three types. And the third is the need to be able to co-ordinate different types (as in my friend and I and Ivy and I).

A more problematical case is the treatment of tall in a tall man. Since modifiers are quite frequently filled by quality groups, as in the underlined portion of a very tall man, we treat examples such as a tall man as cases where the modifier is filled by a quality group that has only an apex.

Ultimately, then, the criterion is a matter of economy. In other words, when a further layer of structure unit is required relatively frequently — as in the case of the nominal group that fills the Subject and the quality group that fills the modifier — we always introduce the additional unit. However, when the lower unit occurs only relatively infrequently — as in the case of the quantifying determiner cited above, we introduce a system to handle the choice that is manifested ultimately as variation in depth of exponence.

Blogger Comments:

[1] To be clear, in SFL Theory, the rank scale makes this principle — and the problems it generates — unnecessary, as demonstrated previously and below.

[2] To be clear, in SFL Theory, the pronoun he simply serves as the Thing/Head of a nominal group that serves as the Subject of a clause.

[3] To be clear, in SFL Theory, nominal groups do not "refer" to 'objects'. Instead, congruently (and ideationally) they realise participants; see Halliday & Matthiessen (1999: 177). Clearly, the nominal groups rare privilege and insatiable greed do not refer to 'objects'.

[4] To be clear, in SFL Theory, these nominal groups are structured as follows:

The participant that each realises depends on the clause in which each functions.

[5] To be clear, in SFL Theory, these are instances in which the Subject is realised by an extending paratactic nominal group complex:

[6] To be clear, in SFL Theory, Fawcett's quality group corresponds to the (± submodified) Epithet of a nominal group:

Fawcett's Genitive Cluster And Main Verb Extension Viewed Through The Lens Of SFL Theory

Fawcett (2010: 259):

A similar case arises with deictic determiners that express 'possession'. In examples such as my books and his books the items my and his are modelled as directly expounding the deictic determiner (dd). But in the case of my friend's books a genitive cluster fills the dd (its possessor being filled by the nominal group my friend). 

Another such case is the Main Verb Extension (Mex in Appendix B) — as in the example of the word out in She threw it out into the yard, because the Mex can be 'expanded' to right out. (Here the unit of the quantity group is borrowed to handle the structure.)

Blogger Comments:

[1] To be clear, in SFL Theory, my friend's books is a nominal group whose Deictic element is realised by a rankshifted (embedded) nominal group:

[2] To be clear, in SFL Theory, this instance of Fawcett's Main Verb Extension corresponds to a circumstantial Adjunct:

which is realised by an adverbial group:

However, not all Fawcett's Main Verb Extensions correspond to circumstantial Adjuncts, as demonstrated, for example, by (gone) walking (p305).