Ed McDonald On There Being 'No Such Things As Words, Sounds Or Meanings'

Edward McDonald replied to David Kellogg on sys-func on 13 Dec 2022 at 10:50:

I would go even further than either you or Chris and say simply this: there are NO such things as "words" - or "sounds" or "meanings" for that matter. We're misled by the lexicogrammar of our metalanguage into seeing such things are "entities", whereas, as Chris in effect pointed out, they're really "intersections" of a whole array of overlapping features. This is how I've framed the issue for my editing clients:

There are no such things as “meanings”, only *contrasts* in meaning; there are no such things as “words”, only alternative *choices* of wording. This is the hard truth about using a language that fluent users know "instinctively” but many learners - and teachers! - seem to try and avoid: you can’t know what a word means unless you know the other possible alternative choices in that context. And because each contrast in meaning, through the appropriate choice in wording, derives from and leads on to further choices, the process of writing, like that of reading, is one of negotiating future choices in the light of past ones, the key at every point being to anticipate what your reader will be expecting.

Such a relational point of view is very hard to keep in mind - so much of modern linguistics and philosophy of language rejects it outright - but if we look at how language actually functions - text in context - then for me it's the only perspective that makes sense.

Blogger Comments:

[1] To be clear, from the perspective of SFL Theory, 'things' are construals of experience as meaning. 'Sounds' are construals as first-order (material) things, whereas 'words' and 'meanings' are construals as second-order (semiotic) things; that is: as meta-things. Semiotic things have no material existence.

[2] To be clear, the distinction here between "misled" (words, sounds and meanings) and "really" (intersections of overlapping features) is a theoretical distinction between a Token and a Value:

McDonald's view is that the higher level of abstraction (Value) of the identity is real, whereas the lower level abstraction (Token) is not real. Logically, McDonald's argument is:

P = Q 

∴ ~ P

[3] To be clear, 'contrasts in meaning' presupposes that there are meanings that can be contrasted.

[4] To be clear, 'choices in wording' presupposes that there are words that can be chosen.

[5] To be clear, 'what a word means' presupposes that there are words that mean.

[6] To be clear, the relational point of view that McDonald meant to express is Saussure's view that 

concepts... are defined not positively, in terms of their content, but negatively by contrast with other items in the same system. What characterises each most exactly is being whatever the others are not (Saussure 1983, 115; Saussure 1974, 117).

In other words:

But, as above, it can be seen that the nub of McDonald's misinterpretation of Saussure is to treat the Value (contrast) as 'real' and to dismiss the Token (meaning) as 'no such thing'.

See also The Thoughts Of Spinoza In Systemic Functional Linguistics.

David Rose Misrepresenting The Basis Of Systems

David Rose replied to David Kellogg on asflanet on 23/8/22 at 12:28:

Beyond Korean... good chance to sort out axis and strata

Re axis: Distinctions in mood options (features) must be realised by regular structural distinctions, that are dependent on other mood choices... e.g. modality options within the English mood system. Systems can only be drawn from structural proportionalities that are consistent across all instances.

Re strata: 

a) indicative-interrogative etc are features in grammatical mood systems. Speech functions are features in (discourse) semantic systems, that are realised by variations in mood.

b) Face-to-face meeting or newspaper reading is a choice at the stratum of tenor and mode (register), which is realised in interpersonal and textual language patterns. ...

 

Blogger Comments:

[1] This is misleading, because it is untrue. To be clear, systemic distinctions between features are not realised by structural distinctions between elements. For example, in the system of MODALITY, the systemic distinction between probability and obligation is not realised by the structural distinction between Finite and mood Adjunct, and either systemic feature may be realised by either structural element.

[2] This is misleading, because it is untrue. On the one hand, it misunderstands systemic proportionalities as structural. As Halliday (2008: 121) explains:

Proportionality means that the terms in the system stand in a constant relationship to one another; their significance will vary according to the context, but (for example) hats is to hat as hairs is to hair as silk is to silks, even though hats are more than hat, hairs are less than hair, and silks are kinds of silk.

On the other hand, contrary to Rose's formulation, it is the system that “gives value to” the elements of syntagmatic structure, as Halliday (2008: 5,6):

This is the “system” in the sense in which it was formulated and defined by JR Firth (Firth 1957a,b); the system is the paradigmatic relation that “gives value to” the elements of syntagmatic structure. […] It is the system that defines the set of options from which any feature derives its value. […] What characterises the system is the regular proportionality between its terms. The system is closed, so that its terms are mutually defining […]

[3] This is misleading, because it repeats Martin's confusion of context with register. To be clear, in SFL Theory, field, tenor and mode are systems of context, the culture as semiotic system, whereas register is a sub-potential of language. Halliday (2005 [1995]: 254):

David Rose Misrepresenting His Posts On 'Axis'

David Rose replied to Ed McDonald on sysfling on 9/8/22 at 9:07:

Rather than an argument, I’d call the macro-genre of this thread a jointly constructed factorial explanation, more informing for anyone interested, than arguing a position. Its factorial structure was given by the points you raised in your 3 August sysfunc post, and each Factor drew on others’ explanations. So explicitly heteroglossic (trying not to mansplain, pace Lexi). Factors were...

1. MAKH’s view of axis (following Firth and Saussure)

2. Relations between features and structures in systems

3. Differences between stratal and axial realisation

4. Stratal relations between phonology/lexicogrammar and semantics in the ‘meaning/wording/sounding’ formula

5. Axis and strata in other modalities

6. Hjelmslev’s influence on MAKH’s stratal model

7. Axis and strata in protolanguage

So yes, very much theory internal (strong verticality in Jo Muller’s terms), but also highly appliable (strong grammaticality). That’s how axis is usually presented in SFL, as a theoretical tool for describing language and other modalities, and for theory building. But I think it is not simply a linguistic innovation by MAKH, but a discovery of how semiosis works. That was the inspiration for my diagram in Factor 2, centred on the claim ‘Structures are perceivable tokens of the abstract values of features’.

Blogger Comments:

[1] This is misleading, because it is untrue on several counts. 

• Firstly, the email thread in question was not jointly constructed. No other list members contributed to the points that Rose tried to make. 
• Secondly, it was not a 'factorial explanation'. According to Martin & Rose (2007: 345), a factorial explanation is a genre ('text type' in SFL) whose purpose ('rhetorical mode' in SFL) is explaining multiple causes, and whose stages ('semantic structure' in SFL) are Phenomenon^Explanation. The previous 20+ posts demonstrate that Rose was not concerned with multiple causes, but with multiple examples of theoretical uses of 'axis', inter alia.
• Thirdly, the posts in the thread were not informing, but misinforming, as demonstrated by the previous 20+ posts on this blog.
• Fourthly, the posts in the thread were indeed arguing a position, Martin's, which Rose explicitly stated as 'axis is key', axis is sufficient' etc.

[2] This is misleading, because it is untrue. The previous 20+ blog posts demonstrate that Rose's emails do not have the 'factorial' structure of Phenomenon^Explanation. Rose and his citations have been strong on assertion, but weak on explanation.

[3] This misunderstands heteroglossia. Heteroglossia involves the expressions of different points of view. In his posts, Rose has cited different authors that he believes support the same point of view.

[4] Mansplain verb. (of a man) explain (something) to someone, typically a woman, in a manner regarded as condescending or patronising.

[5] See, for example:

• David Rose Misrepresenting Axis As Key To Semantics-Grammar Relations
• David Rose Misrepresenting Halliday's Development Of SFL Theory
• David Rose Misrepresenting Axis
• David Rose Misrepresenting Halliday On Axis
• David Rose Misunderstanding Halliday's "Axial Breakthrough"
• David Rose Endorsing Martin's Misunderstanding Of Stratal Relations As Axial
• David Rose Endorsing Martin's Misunderstandings Of Protolanguage

[6] This is potentially misleading. On the one hand, the innovation, the distinction between paradigmatic and syntagmatic relations, was made in Course in General Linguistics (Saussure 1916), nine years before Halliday's birth. 

On the other hand, on the epistemological assumptions of SFL Theory, theories do not "discover truth" ('how semiosis works'). Instead, they reconstrue data validly, or otherwise, on the basis of the assumptions of the theory, valid or otherwise, and vary in their explanatory potential in different contexts of use. Rose here expresses a transcendent view of meaning — meanings are 'out there' to be discovered — whereas SFL assumes an immanent perspective, wherein meanings are construed in semiotic systems. See Halliday & Matthiessen (1999: 415-8).

[7] For the self-contradictions in Rose's diagram, see

• David Rose Misunderstanding Generalisation And Abstraction

David Rose Endorsing Martin's Misunderstandings Of Protolanguage

David Rose wrote to sys-func and sysfling on 7/8/22 at 16:52:

7. (final) Ed may be thinking of MAKH’s model of protolanguage consisting of content and expression planes (a la Hjelmslev), with grammar ‘emerging’ between them in the transition to mother tongue. Martin 2011 comments...

Again axis is sufficient. The axial model of protolanguage microfunctions is expanded by Painter 2003 as an affect system...

Blogger Comments:

[1] To be clear, Ed McDonald had questioned the view that axis is somehow more fundamental than the other dimensions that SFL uses to model language. Because this view is Martin's, Rose has bombarded the sys-func (and sysfling) list with what he falsely believes to be support for Martin's view; see below, and the previous 20+ posts.

[2] To be clear, the Martin (2011) extract seriously misunderstands protolanguage. As Halliday and Painter have demonstrated, protolanguage shows no evidence of structural realisations, so there are no system-structure cycles at all, whether protolanguage is modelled as two strata or one. In protolanguage, content choices are realised by expression choices.

On this basis, Martin's Figure 9 falsely includes structure, and his mono-stratal model misrepresents the stratal realisational relation between the content and expression planes as an axial realisational relation between system and structure.

It can be seen also that Martin fudges his argument for his mono-stratal model by misrepresenting the expression plane of the bi-stratal model as realisation statements instead of systems.

[3] This is misleading, because it is untrue. Axis is insufficient to model protolanguage or language in SFL Theory, because it is only one of several dimensions required. Others include stratification, instantiation, metafunction, and delicacy (Halliday & Matthiessen 2014: 32).

[4] This is misleading, because it is untrue. Painter's Figure 7 specifies how content plane features are realised on the expression plane. It does not relate system to structure. That is, the realisation relation is stratal, not axial.

David Rose Misrepresenting Halliday (1961) And Halliday (1972)

David Rose wrote to sys-func and sysfling on 7/8/22 at 16:28:

6. Re ‘semantics (content) and phonology (form), with lexicogrammar as the "interlevel" between them’

MAKH 1961 was strongly influenced by Hjelmslev, whose stratal model contrasted content and expression planes, and within each plane, form and substance...

The ‘it’ that Firth rejected was the formal/functional dichotomy, since their models included both, as system/structure cycles in phonology and lexicogrammar.

MAKH didn't explicitly equate semantics with substance, but via Hjelmslev in this 1972 quote...

The term «meaning» has traditionally been restricted to the input end of the language system: the «content plane», in Hjelmslev's terms, and more specifically to the relations of the semantic interface, Hjelmslev's «content substance».

As quoted in 1. below, he regarded phonology and grammar as ‘two strata of linguistic form’, and semantics as an ‘interlevel’ or ‘interface’ between grammar and context.

Blogger Comments:

[1] This is misleading, because it is untrue. Halliday (1961), which outlined Scale & Category Grammar, not Systemic Functional Grammar, did not use Hjelmslev's notions of content and expression. Halliday (2002 [1961]: 39):

The only mention of Hjelmslev is the following distinction in footnote 7 (of 113). Halliday (2002 [1961]: 73):

Hjelmslev’s (1953: 8) distinction between “hypothesis” and “theory”

[2] This is misleading, because it misrepresents Halliday (1961). It was Halliday that "rejected" the formal/functional dichotomy. Halliday (2002 [1961]: 51):

The “formal / functional” dichotomy is one of those which linguistics is better rid of;⁴⁵

In the footnote that Rose quotes, Halliday relates this dichotomy to others that Firth rejects.

[3] This is misleading, because it is not true. Halliday (2002 [1961]: 51) rejects the dichotomy on the following grounds:

The“formal / functional” dichotomy is one of those which linguistics is better rid of;⁴⁵ it is misleading to say even that classes are functionally determined, since they are set up with reference to the form of the unit next above – the whole description is both formal and functional at the same time, and “function” is merely an aspect of form.

Moreover, the notion of a system-structure cycle does not arise until the development of Systemic Functional Grammar, being first only foreshadowed in the description of 'realisational cycles', that relate content to expression, in Halliday (1979). Halliday (2002 [1979]: 204):

… the final output – the syntagm – that serves as input to the next realisational cycle.

[4] This is misleading. In this quote from 50 years ago, Halliday simply relates traditional uses of the term 'meaning' to Hjelmslev's term 'content substance'.

[5] This is misleading. While it is true that the quote (Halliday 2013) identifies 'wording or sound' as linguistic form, it does not identify "semantics as an ‘interlevel’ or ‘interface’ between grammar and context". This is because the latter describes Scale & Category Grammar (Halliday 1961) — see [1] above — not Systemic Functional Grammar. 

The Halliday (2013) quote in question, and Rose's misunderstandings of it, can be re-visited at:

• David Rose Misrepresenting Halliday On Axis

David Rose Endorsing Martin's Misunderstanding Of Stratal Relations As Axial

David Rose wrote to sys-func and sysfling on 7/8/22 at 8:14:

5. Re ‘ how this helps us with other semiotic systems’ 

Martin 2011 on Kress & van Leeuwen's Reading Images...

...the greater influence of Reading Images on the field of multimodal discourse analysis stems precisely from the provision of explicit system/structure cycles that give rise to a community of scholars who can confidently share this lingua franca as far as image analysis is concerned. Kress & van Leeuwen's relatively modest axially motivated proposals turn out to be far more powerful than alternatives not explicitly founded on system/structure complementarities.

Painter, Martin & Unsworth 2013 describe system/structure cycles in detail for visual images, for the register of picture books. Here their basic ambience system...

Axis is sufficient...

...to date, for these greyscale systems (and all our other picture book systems in fact) we have not been able to uncover a distinctive system of valeur underlying the systems in Fig. 10 and functioning as ambient semantics. As with SFL MDA work in general, we are proposing one system/structure cycle, with further specification of meaning at this stage dependent on its interaction with co-text in specific instances of a specific genre (Martin 2011).

 

Blogger Comments:

This is misleading, because it is untrue. To be clear, the realisation statements in Figure 10 specify systemic features on the expression plane (colour, black only, texture, lighting effects, outline drawing) not structures on the content plane. The system itself represents content plane distinctions: the meanings purported to be thus realised.

David Rose Misrepresenting 'Meaning, Wording, Sounding' As An "Iconising Tricolon"

David Rose wrote to sys-func and sysfling on 6/8/22 at 13:56:

Re the tricolon ‘meaning, wording, sounding’

In the immortal words of Sesame St...

One of these things is not like the others,

One of these things just doesn't belong,

Can you tell which thing is not like the others

By the time I finish my song?

‘Commonsensical’ because it replaces the technicality of semantics/lexicogrammar/phonology, with a rhetorical device designed to appeal to those not (yet) trained in SFL (which is fine).

The logical connexion between units of a tricolon is internal similarity, intensified by the repetition. Its rhetorical function is to iconise the class of items construed by the tricolon, ‘charging’ it interpersonally, while ‘discharging’ it ideationally. In this instance it iconises the stratal hierarchy, while eliding the categorical difference between wording/sounding as ‘strata of linguistic form’ and semantics as a stratum of ‘meaning’.

That is why I was struck by MAKH’s 1972 explanation that the latter simply followed linguistic tradition, rather than his own theoretical position of ‘language as a whole’ (which he adds here in brackets)...

The term «meaning» has traditionally been restricted to the input end of the language system: the «content plane», in Hjelmslev's terms, and more specifically to the relations of the semantic interface, Hjelmslev's «content substance». We will therefore use «meaning potential» just to refer to the semantic options (although we would regard it as an adequate designation for language as a whole).

More technically, it conflates the metaredundant lexicogrammar/phonology relation of patterns-of-patterns of axial system/structure cycles, with the direct realisation between features proposed in the message/clause semantic model (if that makes sense:).

Blogger Comments:

[1] The term for a division into three categories is a 'trichotomy'. Rose has chosen the inappropriate word 'tricolon' as a rhetorical device; see further below.

[2] This is misleading, because it is untrue. The use of 'meaning, wording, sounding' for 'semantics lexicogrammar phonology' is not rhetorical. These three are used to clarify technical theoretical terms, not to persuade or appeal to readers.

[3] This misrepresents both what constitutes a tricolon and its rhetorical function. To be clear:

Tricolon is a rhetorical term that consists of three parallel clauses, phrases, or words, which happen to come in quick succession without any interruption. The origin of this rhetorical device is traced to the Greek word tricolon, meaning “section of a sentence.” These three parallel words, phrases, or clauses have almost the same length, though this condition is not strictly followed.

It also refers to a collection of three lines, paragraphs, chapters, or stanzas. For instance, William Shakespeare’s Julius Caesar uses it in his famous speech in ascending order as, “Veni, vidi, vici (I came, I saw, [and] I conquered.“) The purpose of tricolon is to give a greater sense of roundness, completeness, and wholeness, whereas the third part brings in a surprising effect in the sentence.

[4] This is misleading, because it is untrue. The terms 'meaning, wording, sounding' merely relabel 'semantics, lexicogrammar, phonology' in less technical language. There is no "iconisation".

[5] This is misleading, because it is untrue. On the one hand, these less technical terms make the exact same categorical distinctions as their more technical counterparts. On the other hand, in SFL Theory, wording/lexicogrammar is form interpreted in terms of the meaning it realises. To be clear, Rose himself follows Martin in misconstruing all strata as linguistic meaning, even those of phonology and context.

[6] This is misleading, because it is untrue. The quote from Halliday (1972) merely specifies how the term 'meaning potential' will be used in the paper, relative to traditional usages of 'meaning' (interpreted in terms of Hjelmslev's model).

[7] To be clear, to the extent that this makes any sense, it is nonsense. Firstly, it misunderstands 'metaredundancy'. As the term implies, 'metaredundancy' is a redundancy on a redundancy. Applied to the stratification hierarchy, it means that semantics is redundant on the redundancy of lexicogrammar and phonology, or alternatively, that the redundancy of semantics and lexicogrammar is redundant on phonology. Applying the term to just two strata, lexicogrammar and phonology, as Rose does, is therefore nonsensical.

Secondly, the 'direct realisation between features proposed in the message/clause semantic model' is the specification of how semantic features in Hasan (1983) are realised in lexicogrammar, both systemically and structurally: 

Rose's claim, then, is the terms 'meaning, wording, sounding' conflate a lexicogrammar-phonology relation, misunderstood as metaredundant, with a semantics-lexicogrammar relation. This again, therefore, is nonsensical.

David Rose Misunderstanding Lexicogrammatical Form (inter alia)

David Rose wrote to sys-func and sysfling on 5/8/22 at 14:17:

3. Re linguistic ‘progress’...

The term realisation is also used for relations between strata. Like axis, it is a relation of abstraction, but of a different order.

In his foreword below, MAKH refers to phonology and grammar as ‘these two strata of linguistic form’. In the terms developed below, form is associated with perceivability of structures. The form of phonological structures is ‘sounding’ (i.e. patterns of sounds); the form of grammatical structures is ‘wording’ (patterns of words). The axial relation resolves the dualism of form and meaning, as patterns of forms at each stratum are at once patterns of meanings.

Relations between the strata are patterns-of-patterns... patterns of lexicogrammatical structures are realised by patterns of phonological structures. Discourse semantics proposes that the semantic stratum is also organised axially (as MAKH predicted in 1972). Its structures are discourse structures... and patterns of discourse structures are realised by patterns of both lexicogrammatical structures and phonological structures.

Blogger Comments:

[1] This is potentially misleading. Realisation is precisely the same relation (Token-Value) in each case. The difference lies in what is related: lower stratum to higher stratum (stratification) vs structure to system (axis).

[2] This again mistakes a lower level of abstraction for perceivability (see previous post).

[3] To be clear, 'sounding' refers to the stratum of phonology. A phonological system is not a 'pattern of sounds'; a phonological system models 'sounding' choices.

[4] This is a very serious misunderstanding. SFL Theory models lexicogrammatical form as a rank scale. The term 'wording' refers to lexicogrammar as a stratum, in which lexicogrammatical form (e.g. verbal group) is interpreted in terms of its function of realising meaning (e.g. process). A lexicogrammatical system is not a 'pattern of words'; a lexicogrammatical system models 'wording' choices.

[5] This is a very serious misunderstanding. The axial relation does not 'resolve the dualism of form and meaning'. Form and meaning are located on both axes in the lexicogrammar. (There is meaning, but not form, on the semantic stratum, and form, but not meaning, on the expression plane.)  Paradigmatically, each formal rank unit is the entry condition to systems of functions, and syntagmatically, each element of function structure is realised by a formal unit of the rank below.

[6] This misunderstands stratification. There are no 'patterns of form' on the semantic stratum, and no 'patterns of meaning' on the expression plane. On the lexicogrammatical stratum, form is interpreted in terms of its function in realising meaning.

[7] This is misleading. SFL Theory, as the name implies, gives priority to system over structure. Systemically, a pattern is a pattern of instantiation, that is: a pattern of selecting features and activating realisation statements. The notion that 'patterns of lexicogrammatical structures are realised by patterns of phonological structures' not only gives priority to structure, but also confuses instantiation ('patterns') with stratification ('are realised by').

[8] This is misleading by implication. Rose has previously falsely claimed that Halliday does not model semantics in terms of both system and structure, despite the role semantic structure plays in grammatical metaphor, and despite explicit contradictions such as the following provided by Halliday & Matthiessen (1999: 429):

… in our model there are two system-structure cycles, one in the semantics and one in the lexicogrammar. Terms in semantic systems are realised in semantic structures; and semantic systems and structures are in turn realised in lexicogrammatical ones.

[9] This is misleading. Halliday (1972) weighs up the theoretical pros and cons of proposing structure for the semantic stratum, as demonstrated in previous posts. No "predictions" were made.

[10] This again gives priority to structure over system and confuses instantiation ('patterns') with stratification ('are realised by'); see [7] above.

David Rose Misunderstanding Halliday's "Axial Breakthrough"

David Rose replied to Ed McDonald to sys-func and sysfling on 5/8/22 at 8:34:

2.

Thanks to Ed, for the challenge presented by your reading of my diagram below as ‘(paradigmatic) "features" and (syntagmatic) "structures"’.

The diagram is meant to show that structures are part of systems, not opposed to them as the system/structure and paradigmatic/syntagmatic terms might suggest. Structures don’t realise systems, but features in systems. Structures are perceivable tokens of the abstract values of features. Each instance of structure is recognisable by its similarity to other instances of the same structure, and its difference from other types of structure. Similarity is represented in the diagram by hexagons, and difference by the system. The entry condition to a system is a more general similarity that is shared by its structures, e.g. the structure +Subject;+Finite is shared by the structures Subject^Finite and Finite^Subject. These similarities and differences in structure realise similarities and differences in features, such as indicative: declarative/ interrogative. Hence the system of features is a mirror of the system of structures. They are two faces of a system. This is my understanding of MAKH’s axial breakthrough.

 Blogger Comments:

[1] Some of the problems with this diagram were identified in the earlier post David Rose Misunderstanding Generalisation And Abstraction.

[2] This is a very serious misunderstanding. Systems and structures are distinct levels of symbolic abstraction (elaboration + identity), whereas the relation that obtains between a whole and its parts is composition (extension). A part of a system is a subsystem, whereas a structure (e.g. of a group) is part of a larger structure (e.g. of a clause).

[3] This is a very serious misunderstanding. It is the entire system of a rank unit that specifies how its entire structure is realised, and this is achieved through realisation statements, not features.

[4] This confuses lower abstraction with perceivability. The structure of a nominal group cannot be perceived, unless, like its system, it is represented theoretically.

[5] This confuses the realisation (token-value) relation between structure and system with the instantiation (token-type) relation between instances and instance type.

[6] This confuses instance type (similarity) with paradigmatic order (difference).

[7] This confuses instance type (similarity) with the entry condition of a system. In the grammar, the first entry condition of a system is a formal unit on the rank scale.

[8] This is misleading, because it is untrue. Difference in system is difference between paradigmatic features, such as declarative vs interrogative, whereas difference in structure is difference in elements, such as Subject vs Complement.

[9] To be clear, the claim that "Halliday's axial breakthrough" is that system and structure are "two faces of a system" derives from the multiple misunderstandings identified above.

David Rose Misrepresenting Halliday On Axis

David Rose replied to Ed McDonald on sys-func on 4/8/22 at 8:45:

These are really important points Ed, that everyone engaging with SFL needs to grapple with (as we are).

That is why I posted MAKH’s own account of his engagement with axis, in the foreword to Martin, Wang & Zhu (reposted below).

1.

His starting point here is not with the paradigmatic or syntagmatic axis, but with the relation between them, as recognised by Saussure. His second point is that most linguists since have ignored this relation. His third point is that his teacher Firth foregrounded it. The fourth is his own effort to describe language ‘biaxially’, and coming to see it as sets of choices (systems). His fifth point is Firth’s very specific conception of the axial relation ‘as the paradigm of options that were available at a given location in the structure’. He then devotes more detail to this point, because it is central to systemic theory. Every system in SFL starts from ‘a given location in the structure’, such as syllable or nom gp. Every feature in a system is realised by a distinct structure, such as Onset+Rhyme or Deictic+Thing. His last point here is that this representation enables us to see the interdependence of diverse system/structure options, which is what ‘opened up metafunctions and rank scale system/structure cycles’ (as you quote me below) as well as strata.

Blogger Comments:

[1] This is misleading, because it is untrue. Halliday begins by explaining the difference between the paradigmatic and the syntagmatic axes, not the relation between them. In SFL Theory, the relation between the axes is realisation (symbolic identity).

[2] This is misleading, because it is untrue. Halliday claims that most linguists in the 20th Century focused on the syntagmatic axis rather than the paradigmatic axis, not that they ignored the realisation relation between the axes.

[3] This is misleading, because it is untrue. Halliday claims that Firth was one linguist who did not focus on the syntagmatic axis at the expense of the paradigmatic, not that Firth foregrounded the realisation relation between the axes.

[4] This is misleading, because it is untrue. Halliday is here describing Firth's approach to systems, not the SFL approach that Halliday instigated. For Firth, each element of function structure (e.g. Deictic) is an entry point to a network of choices. For Halliday, it is the formal rank unit (e.g. nominal group) that is the entry condition for the network of choices that specify how the unit is structured in a given instance. Rose fails to notice this important distinction between Firthian linguistics and SFL Theory.

[5] This is misleading, because it is untrue. No single feature specifies the structure of a unit. Instead, a realisation statement associated with a feature specifies the insertion, conflation etc. of elements. Halliday (1995 [1993]: 272) identifies seven types of realisation statement:

(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).

[6] This is misleading, because it is untrue. Halliday is concerned with explaining how systems can be elaborated in delicacy and combined into networks. He does not claim that this is what 'opened up metafunctions and rank scale system/structure cycles as well as strata'.

Importantly, this last bare assertion is the crucial point Rose wants to make, and all the misrepresentations that precede it are falsely presented as evidence for what is, logically, a non-sequitur.

David Rose Endorsing Martin's Misunderstandings Of Individuation

David Rose responded to a Halliday quote on Bernstein's code posted by ChRIS CLÉiRIGh on sys-func on 3/8/22 at 13:49:

The complexity of relations between code, register and language seems to be resolving with studies in individuation, such as this paper of yours...

Martin, J. R., Zappavigna, M., Dwyer, P., & Cléirigh, C. (2013). Users in uses of language: embodied identity in Youth Justice Conferencing. Text & Talk, 33(4-5), 467-496.

Blogger Comments:

[1] This is not the case, as demonstrated below.

[2] To be clear, Martin is responsible for the misunderstandings (see below) in this paper. Cléirigh's contribution was to provide the model of body language that Martin & Zappavigna (2019) rebranded as their model of paralanguage. Evidence here.

[3] To be clear, Martin's model of individuation confuses an elaboration relation within language with an extension relation between language users. To explain:

Firstly, in this context, Bernstein's notion of reservoir and repertoire refer to the language of users, not to the language users themselves. The two differ in terms of order of experience: a language user is a first-order phenomenon, whereas the language that a user projects is a second-order phenomenon (metaphenomenon).

Secondly, the relation between reservoir and repertoire is elaboration, since varying repertoires are different subtypes of a reservoir, whereas an affiliation relation between language users, as association, is a subtype of extension (Halliday & Matthiessen 1999: 146).

[4] To be clear, Martin's Figure 4 takes the fundamental confusions outlined above and adds to them. On the one hand, if Bernstein's distinction is viewed as a cline of individuation, then in SFL terms, the cline is

reservoir — reservoir subpotential/repertoire type — repertoire

where each point on the cline is the stratal hierarchy of systems

context — semantics — lexicogrammar — phonology.

On the other hand, Figure 4

• treats a persona (a language user) as an individuation of culture (context potential);
• inserts a subtype (master identity) between culture and its named subtype (subculture);
• models affiliation as an elaborating relation between persona and culture, instead of an extending relation perpendicular to the cline (between personas, between subcultures etc.)

∞

For Halliday's distinction between socio-semiotic meaner and social persona, and Lemke's (1995) examination of the 'notion of the individual human subject', see Halliday & Matthiessen (1999: 610-1).

See also David Rose On Jim Martin's Individuation.

David Rose Misrepresenting The Appliability Of Models

David Rose wrote to sys-func on 2/8/22 at 14:35:

A third difference is in appliability of the models. As classification of types is the focus of the message/clause semantic (MS) model, a major application has been classifying types of discourse, by their frequencies of message/clause types. … 

As structuring of texts is the focus of the discourse semantic (DS) model, applications have been in various fields requiring text analysis. ...

In terms of which model is more ‘correct’ (theory as rule)...

MS expects direct grammatical realisations for each semantic feature

DS expects discourse structures realising each discourse semantic feature.

In terms of which model is more appliable (theory as resource)...

MS can classify discourse types statistically from large corpora

DS can explain discourse varieties by contrastive analysis of texts.

Blogger Comments:

[1] To be clear, 'classifying types' is the systemic approach to theorising language as meaning potential.

[2] To be clear, 'classifying types of discourse' is the identification of different sub-potentials — registers — of the overall potential, each of which is distinguished by the probabilities of feature selection, with these probabilities being instantiated as frequencies of feature selections in texts. The explanatory potential of this approach is not available in Martin's model, because he misconstrues register as the systems of context that language realises.

[3] To be clear, if the structuring of texts is the focus of Martin's discourse semantic model, then it is inconsistent with the fundamental functional principle of SFL Theory of giving priority to the view 'from above'. As Halliday & Matthiessen (2014: 49) explain for the lexicogrammatical stratum:

Giving priority to the view ‘from above’ means that the organising principle adopted is that of system: the grammar is seen as a network of interrelated meaningful choices. In other words, the dominant axis is the paradigmatic one: the fundamental components of the grammar are sets of mutually defining contrastive features (for an early statement, see Halliday, 1966a). Explaining something consists not in stating how it is structured but in showing how it is related to other things: its pattern of systemic relationships, or agnateness (agnation…).

[4] This is misleading. Text analysis can be carried out regardless of whether the focus is on structure (Martin) or on system (SFL). Focusing on structure provides writing templates for teachers; focusing on system provides a systematic way of relating functional varieties of language in terms of specific speaker choices.

[5] These confuse assessments of models (correct, appliable) with epistemological stances (theory as rule/resource). To be clear, Formal Linguistics models language as 'rule', whereas SFL models language as 'resource'.

[6] This is misleading. Rose is again using Hasan (1983) to falsely claim that Halliday's model does not feature semantic structures. See

• David Rose On The Absence Of Semantic Structures In Halliday's Model

[7] To be clear, discourse semantics may "expect" discourse structures realising each discourse semantic feature, but it certainly does not specify them; see Martin (1992) or Martin & Rose (2007). The one apparent exception to this, NEGOTIATION, Martin's rebranding of Halliday's SPEECH FUNCTION, does not coherently specify the range of structures it claims, See, for example:

• Problems With The Exchange Structure NEGOTIATION System
• Problems With The Authors' Negotiation System Network

David Rose Misrepresenting Axis

David Rose replied to Ed McDonald on 1/8/22 at 20:42:

Axis pairs perceivable forms with conceivable meanings, which makes semiosis possible. Turning your question around, what the others all have is axis. It’s also what makes SFL systemic.

and on 2/8/22 at 12:41:

I liked my bon mot ‘Axis pairs perceivable forms with conceivable meanings, which makes semiosis possible.’

Blogger Comments:

[1] This is misleading, because it is untrue. In SFL Theory, axis is the local dimension whose orders are paradigmatic and syntagmatic, with system as the dimension of paradigmatic order and structure as the dimension of syntagmatic order (Halliday & Matthiessen 2014: 20, 32).

Importantly, axis does not "pair forms with meanings". Form and meaning are located on both axes in the lexicogrammar. (There is meaning, but not form, on the semantic stratum, and form, but not meaning, on the expression plane.) Lexicogrammatical form is modelled as a rank scale, and rank units are interpreted in terms of their function in realising meaning. Paradigmatically, each formal rank unit is the entry condition to systems of such functions, and syntagmatically, each element of function structure is realised by a formal unit of the rank below.

With regard to "perceivabilty", lexicogrammar is 'a purely abstract level of representation "in between" the two faces of the sign' (Halliday & Matthiessen 1999: 613).

[2] This is misleading. To be clear, the 'others' here are those identified by Ed McDonald:

I noted the reference to axis, and I'm a bit puzzled as to how and why this seems to have become the theoretical fons et origo. Do we really to trace everything back to that single distinction? And if so, why paradigmatic/syntagmatic (and not and / or system/text, content/form, synchronic/diachronic...)? Does it have a privileged relation to realisation and / or instantiation?

As McDonald implies, axis is but one dimension of language in the SFL model. Halliday & Matthiessen (2014: 32):

[3] This is misleading. What makes SFL systemic is the priority it gives to system over structure, just as what makes SFL functional is the priority it gives to function over form.

David Rose On The Absence Of Semantic Structures In Halliday's Model

David Rose wrote to sys-func on 1/8/22 at 17:09:

MAKH’s procedure for semantic networks was taken up by RH, who generalised their application, specifying the entry condition as messages, realised by clauses (Hasan, Cloran, Williams, & Lukin 2005)...

Naturally, the focus had to be not on a context specific semantic network, but on a language-exhaustive one, or contextually open...

What remained the same was classification of sub-types, still realised as features in grammar systems, rather than semantic structures. In RH’s 1983 system below, features classify sub-types of questions, by more general grammatical criteria than MAKH 1972.

Blogger Comments:

[1] To be clear, the procedure that Hasan (1983) took up from Halliday (1972) was simply the use of system networks to model semantics paradigmatically.

[2] This is misleading, because it is untrue. The networks in Halliday (1972) do not specify how semantic features are realised lexicogrammatically; see previous post. 

[3] This is misleading. As can be seen by examining the semantic network in Hasan (1983), the features are distinguished on the basis of meaning, not on grammatical criteria.

∞

It is important to understand the nature of Rose's argument in these posts. His claim is that Halliday's model of semantics does not include structures, and his method of arguing is to seek out old papers (1972, 1983) that were written before semantic structures were discussed in the modern framework. That is, Rose studiously ignores later evidence, such as Halliday & Matthiessen (1999), in which, for example, grammatical metaphor is modelled as a Token-Value relation between semantic structures. Halliday & Matthiessen (1999: 208-9):

It is the fact that metaphor multiplies meanings within the semantic system that opens up the possibility of metaphorical chains, with one congruent starting-point and another highly metaphorical end-point (A"' stands for A" stands for A' stands for A; e.g. 'engine failure' stands for 'the failing of an engine' stands for 'an engine failed'). The semantic system is being expanded along the dimension of the metaphorical token-value relation; but the expansion is still within the semantic system itself. 

 

As Halliday & Matthiessen (1999: 429) explain:

… in our model there are two system-structure cycles, one in the semantics and one in the lexicogrammar. Terms in semantic systems are realised in semantic structures; and semantic systems and structures are in turn realised in lexicogrammatical ones. As we saw in Chapter 6 in particular, grammatical metaphor is a central reason in our account for treating axis and stratification as independent dimensions, so that we have both semantic systems and structures and lexicogrammatical systems and structures.

∞

As Peter Medawar once wrote of another author:

Its author can be excused of dishonesty only on the grounds that before deceiving others he has taken great pains to deceive himself. 

David Rose Misrepresenting Halliday's Semantic Networks

David Rose wrote to sys-func on 1/8/22 at 16:25:

A related difference in the models is in the nature of the features. In the negotiation system, features are motivated from above by exchange type: knowledge/action, and within the same stratum, by sequencing of interactant roles. So the entry condition is [exchange], and the features are alternating exchange roles.

In contrast, semantic networks are not concerned with sequencing of roles, but with classifying subtypes of what MAKH calls ‘verbal behaviour’. They are motivated from above by ‘behaviour potential’, and from below by grammatical realisations, but not axially by exchange structure. Their entry conditions are behaviour types, and their features specify behaviours realised by instantiating clauses, below...

Blogger Comments:

 [1] This is very misleading indeed. Rose is here misrepresenting a network from Halliday (1972) as a network of Halliday's current model, and comparing it with Martin's negotiation network (Martin & Rose 2007):

[2] This is misleading, because it is untrue. The features 'knowledge' and 'action' are located within the negotiation network, and so are not 'above' the network. Rose here confuses delicacy with stratification.

[3] This is misleading, because it is untrue. The 'sequencing of interactant roles' is how features are realised structurally, and so are 'below' the system.

[4] For the serious problems with the negotiation network, see

• Problems With The Exchange Structure NEGOTIATION System
• Problems With The Authors' Negotiation System Network

[5] To be clear, in this early paper, Halliday (1972) was exploring the possibility of a sociological semantics, where context was conceived as behavioural potential, with language as one means of realising behavioural potential. This is very different from the semantics of Halliday & Matthiessen (1999) where context is conceived metafunctionally in terms of field, tenor and mode, and semantic systems are the meanings construed by lexicogrammatical systems.

[6] This is very misleading. The reason why the networks in Halliday (1972) are not concerned with the sequencing of roles or exchange structure is that they predate Halliday's system of SPEECH FUNCTION, which Martin misunderstood and rebranded as his system of NEGOTIATION. Evidence here. It is Halliday's SPEECH FUNCTION network that would be the appropriate model for comparison.

[7] This is misleading. In Halliday (1972), the entry conditions to semantic systems are not behaviour types. Instead, behavioural potential is the higher-level context that semantic systems realise. Halliday (2003 [1972]: 341):

[8] This confuses the relation between strata (realisation) with the relation between potential and instance (instantiation).