David Rose Misrepresenting Michæl Halliday On Meaning

David Rose replied to Ed McDonald on SYSFLING on 27 Aug 2020 at 10:01:

I wonder where you might locate ‘concepts’ in our stratal hierarchy. I believe MAKH followed Firth’s distributed view of meaning, where "The central proposal of the theory is to split up meaning or function into a series of component functions. Each function will be defined as the use of some language form or element in relation to some context. Meaning, that is to say, is to be regarded as a complex of contextual relations, and phonetics, grammar, lexicography, and semantics each handles its own components of the complex in its appropriate context.” 1957 p5-6

Blogger Comments:

[1] To be clear, 'concepts', as ideational meanings, are located on Halliday's semantic stratum.

[2] This is misleading. In Halliday's model of language as meaning potential, meaning is construed as the highest level of symbolic abstraction: the stratum of semantics. This is distinguished from wording, a lower level of symbolic abstraction: the stratum of lexicogrammar, and from sounding, a still lower level of symbolic abstraction: the stratum of phonology.

Rose uses the wording 'distributed view of meaning' in defence of Martin (e.g. 1992), where all strata are misunderstood as strata of linguistic meaning — even context and phonology. The misunderstanding is encapsulated in Martin's 'all strata make meaning', which, though a statement about semogenesis (making meaning), is misinterpreted as a statement about stratification (levels of symbolic abstraction): all strata are strata of meaning.

[3] To be clear, although this 1957 quote from Firth is largely consistent with Halliday's later stratification of language into semantics, lexicogrammar and phonology, Firth's use of 'context' is not the same as Halliday's later construal of context as the culture as a semiotic system. This can be seen in Halliday (1961) where Halliday's earliest use of 'context' is closer to Firth's use of the term:

David Rose On Jim Martin's Individuation

David Rose replied to Ed McDonald on SYSFLING on 24 Aug 2020 at 13:36:

Ed McDonald: 

3. I’m not exactly sure what you mean by individuation, but would you recognise a greater degree of “individuation” from phonology through lexiocgrammar and semantics to the “material and social worlds of human communities”

No, in a word. Individuation varies at each stratum, from culture to persona… figs [4 and 5] from 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), 467-96

Blogger Comments:

[1] To be clear, the process of individuation is the differential ontogenesis of potential — from culture to phonology — across individuals, and the cline of individuation is the relation between the different potentials of different individuals and the collective potential of all meaners.

[2]  To be clear, Martin's Figure 4 confuses points on the cline of instantiation of context (culture and sub-culture) with categories of language users (master identity and persona). Moreover, it proposes that context (culture) is individuated as a language user (persona).

Strictly speaking, the individuation of culture is the differential ontogenesis of the context potential across individuals, and if 'persona' is interpreted as individuated potential, then it is the individuation of both context and language as potential.

[3] To be clear, there are several theoretical inconsistencies in Martin's Figure 5.

Firstly, it presents instantiation and individuation as if they were internal dimensions of meaning potential like stratification. However, the clines of instantiation and individuation model different perspectives on meaning potential as a whole, rather than scales within meaning potential.

Secondly, it misrepresents text as an instance of context (culture) as well as language. In SFL Theory, it is situation that is an instance of context. This relates to Martin's misunderstanding of context as varieties of language, genre and register (which accounts for the superfluous stratum in the figure).

Thirdly, by its vertical dimension, it (unintentionally?) misrepresents the cline of individuation as applying to text as well as potential. This makes the nonsensical claim that every text is common to all language users, varying according to user.

[4] To be clear, the Martin et al paper was written by Zappavigna, under the control of Martin, who is responsible for Figures 4 and 5. The blogger's name was included as author because he, as part-time Research Assistant, devised the model of gestural and postural semiosis that forms the intellectual backbone of the paper. This model has since been misunderstood and rebranded as Martin's model of paralanguage, as documented here.

David Rose On Jim Martin's Coupling/Co-instantiation

David Rose replied to Ed McDonald on SYSFLING on 24 Aug 2020 at 13:36:

Ed McDonald:

2. The idea of co-instantiaton is also an interesting one:

My gloss for coupling, defined by Martin et al. (2013, p. 469) as ‘the co-selection of linguistic resources across ranks, metafunctions, strata, and modalities which are not specified by system/structure cycles’.

Systems at each stratum make distinct contributions…eg variations in tone and mood couple to instantiate variations in speech function

Blogger Comments:

[1] To be clear, Martin's definition of 'coupling' is invalid, because system/structure cycles do not specify any co-selection of paradigmatic features ("linguistic resources"). To explain:-

The process of instantiation is the selection of paradigmatic features and the activation of realisation statements which specify how systemic options are realised along the syntagmatic axis, whether structural or cohesive. System/structure cycles are the iteration of this realisation relation between axes during logogenesis, the unfolding of text. On this basis, system/structure cycles do not specify which features are (co-)selected; feature (co-)selection is the paradigmatic dimension of system/structure cycles.

[2] To be clear, this confuses instantiation with realisation. Instantiations of "variations in speech function" are the speech function features that are selected during logogenesis. The relation between speech function and mood, and between mood and tone, is realisation, since these three systems are located on three different strata: semantics, lexicogrammar and phonology.

However, strictly speaking, the system of TONE does not realise the system of MOOD, but the system of KEY, which is associated with the system of MOOD (Halliday & Matthiessen 2014: 168). It is the combination of KEY and MOOD that realises SPEECH FUNCTION.

David Rose On Stratal Relations

David Rose replied to Ed McDonald on SYSFLING on 24 Aug 2020 at 13:36:

Ed McDonald: 

[1.] The “variability of relations” in stratification sounds interesting: could you elaborate a bit on what other kinds of relations you would recognise here?

OK…'natural/conventional’ debates pre-date development of phono rank scale. They assumed ‘arbitrary’ sound/word relations, to then debate grammar/semantic relations. But phono/grammar relations actually vary by phono rank and system. So let’s allow for such variability between other strata.

Blogger Comments:

[1] The terms 'debates' and 'assumed' are misleading here. The conventional (arbitrary) relation between a word (e.g. the) and the sounds that realise it (e.g. [ꝺə]) is evident from observation, and so does not need to be assumed. The natural relation between semantics and grammar, in SFL Theory, refers to such non-arbitrary relations as those between participant and nominal group and between process and verbal group, in the absence of grammatical metaphor.

[2] To be clear, the stratal relation between grammatical forms (e.g. nominal groups and verbal groups) and their phonological realisations is invariably conventional (arbitrary), and does not vary by phonological rank. That is, the semantic distinction between participant and process, which is realised by a grammatical distinction between nominal group and verbal group, is not realised by a phonological distinction in intonation, rhythm or articulation.

What Rose could have in mind here is the relation between speech function and tone, and perhaps the relation between information focus and tonic prominence, neither of which is a relation between lexicogrammatical form and phonology.

[3] To be clear, the relation between other linguistic strata, semantics and lexicogrammar, in the absence of grammatical metaphor, is invariably natural (non-arbitrary) in the sense specified by Halliday. Moreover, the relation between adjacent strata is invariably realisation (intensive identification). Any model in which strata are not related by realisation is inconsistent with the  ordering principle of stratification; see Halliday & Matthiessen (2014: 20).

David Rose On Stratification, Instantiation And Individuation

David Rose replied to Ed McDonald on SYSFLING on 23 Aug 2020 at 08:31:

That is, to start answering your questions, could we….

1. Take our long work on stratification as a starting point, and admit variability of relations between all strata, not just realisation hierarchies… moving beyond old dichotomies like form/content, language/context, natural/conventional 

2. Reconceptualise system/text relations as co-instantiation or coupling of distinct contributions from each mode and stratum, as ‘all strata instantiate' 

3. Model language typology in terms of individuation, as ‘all strata individuate’, from phoneme systems to material and social worlds of human communities (always already semiotic)

?

Blogger Comments:

[1] To be clear, "our long work on stratification" is Martin's model, which replaces

• semantics with discourse semantics, and
• culture as context with genre and register as two strata of context.

As demonstrated in detail here, Martin's model is based on several theoretical misunderstandings and is internally inconsistent. For example, Martin's discourse semantics is a mixture of Halliday's semantics: speech function (rebranded negotiation) and Halliday & Hasan's lexicogrammatical cohesion (reference and ellipsis rebranded as identification, conjunction now rebranded as connexion, and lexical cohesion rebranded as ideation). That is, Martin's discourse semantics confuses two distinct levels of symbolic abstraction: semantics and lexicogrammar.

Martin's model of context locates varieties of language, register and genre, outside language: as context, while at the same time claiming that their instances are instances of language. In Hjelmslevian terms, Martin's model takes varieties of the expression plane of a connotative semiotic and relocates them as the content plane of a connotative semiotic. In terms of SFL Theory, Martin's model takes the midway point on the cline of instantiation of language and reconstrues the two perspectives on that point, register and text type (genre), as the system pole of the cline of instantiation of context. That is, Martin's model is inconsistent in terms of two dimensions of SFL Theory: stratification and instantiation. Compare the theoretically consistent model of Halliday:

[2] This is misleading. The theoretical dimension of stratification is the layering of different levels of symbolic abstraction. The relation between levels of symbolic abstraction is realisation. While the relation between levels may be natural or conventional, the invariable relation between levels is realisation.

[3] As previously explained, Martin's 'co-instantiation or coupling' is theoretically superfluous, since, because instantiation is the selection of features (and activation of realisation statements), the relation between instantiated features is already given by the systemic and stratificational architecture.

[4] To be clear, with regard to language, the theoretical dimension of individuation is the relation between a language, as a whole, and the varieties of that language that develop in its individual speakers, whereas language typology is the classification of different languages according to theoretical criteria.

[5] To be clear, Martin's mantras 'all strata instantiate' and 'all strata individuate' say nothing about either instantiation or individuation; cf all X instantiate, all X individuate. The process of instantiation is the selection of potential during logogenesis; the process of individuation is the differential ontogenesis of potential across individuals.

David Rose On "Couplings In Instantiation"

David Rose replied to Ed McDonald on SYSFLING on 21 Aug 2020 at 10:24:

Thanks for this teaser. There's an impossibly large number of threads picked up in these few little questions… variability in relations between strata in semiotic systems… their contributions and couplings in instantiation… phylogenesis of language in general and languages in particular… and histories of linguists trying to theorise (what they can see of) these problems

Blogger Comments:

[1] This is misleading, because it is untrue. McDonald's questions concerned:

• the natural relation between meaning (e.g. process) and linguistic form (e.g. verbal group),
• the construal of experience as meaning, and
• the co-ordination of language with other social semiotic systems made possible by language.

See the examination of McDonald's original post here.

[2] To be clear, 'coupling' or 'co-instantiation' is one of Martin's many misunderstandings of SFL Theory. As previously noted, because the process of instantiation is the selection of features and the activation of realisation statements (Halliday & Matthiessen 1999: 45), it necessarily entails the co-selection of features that are related along the dimensions of the theory, including delicacy, rank and strata. Martin's superfluous notion of "co-instantiation" is merely an acknowledgement that features are co-selected.

John Bateman On The Natural Relation Between Semantics And Grammar

John Bateman replied to Ed McDonald On SYSFLING on 20 Aug 2020 at 17:25:

Edward McDonald:

So “natural” presumably can’t mean “given by nature”, because in that case the semantics of all languages would be the same; but if this “naturalness” is “conventional”, to what degree?

strikes me the most straightforward way to see this as a case of iconic diagrammaticity (which would be a bit more general than Hjelmslev's more direct linking): so I would expect lexicogrammar and semantics to stand in an iconic relationship in this sense. The details of each are conventional, but the more fundamental relationship of diagrammicity holds indeed by their nature rather than by convention.

 Edward McDonald:

And if all human communication is multimodal, and all semiotic systems are “natural” in their different ways, how do these different “natures” get to be coordinated or mutually-enforcing in communicative contexts?

discourse semantics discourse semantics (some embodiment) and some more discourse semantics.... (see our work on this...).

Blogger Comments:

[1] For an examination of the McDonald post that Bateman is responding to here, see the immediately preceding post.

[2] To be clear, Halliday's notion of a natural relation between semantics and lexicogrammar is the non-arbitrary (non-conventional) relation between meanings and grammatical forms that realise them, such as between process and verbal group and participant and nominal group; see Halliday (1985: xvii-xix).

[3] To be clear, here Bateman is reinventing Halliday's wheel. Halliday (2002 [1984]: 293):

Grammars are ‘natural’, in the sense that wordings are related iconically to meanings;

[4] To be clear, for Halliday, it is the cultural context — that language and other parallel social semiotic systems realise — that makes this possible. Halliday & Matthiessen (1999: 602):

In other words, language has evolved as part of our own evolution. It is not arbitrary; on the contrary, it is the semiotic refraction of our own existence in the physical, biological, social and semiotic modes. It is not autonomous; it is itself part of a more complex semiotic construct — which, as we have tried to show, can be modelled in stratal terms such that language as a whole is related by realisation to a higher level of context (context of situation and of culture). This contextualisation of language, we suggested, was the critical factor which made it possible to relate language to other systems-&-processes, both other semiotic systems and systems of other kinds.

Discourse semantics, on the other hand, is Martin's rebranding of (his misunderstandings of) the prior work of Halliday and Hasan, with

• speech function (Halliday's interpersonal semantics) rebranded as negotiation (Martin's interpersonal discourse semantics),
• reference (Halliday & Hasan's textual lexicogrammar) rebranded as identification (Martin's textual discourse semantics),
• conjunction & continuity (Halliday & Hasan's textual lexicogrammar) now rebranded as connexion (Martin's logical discourse semantics), and
• lexical cohesion (Halliday & Hasan's textual lexicogrammar) rebranded as ideation (Martin's experiential discourse semantics).

For the evidence on which these claims rest, see the examination of Martin's English Text (1992) here, and the examination of Martin & Rose's Working With Discourse (2007) here. For evidence that Bateman is oblivious to the theoretical inconsistencies in Martin's model of discourse semantics, see the examination of his review of English Text here.

Edward McDonald On The Natural Relation Between Semantics And Lexicogrammar

 Edward McDonald wrote to SYSFLING on 20 Aug 2020 at 17:11:

I remember when I read that passage [Halliday (1985: xvii-xix) on the natural relation between meaning and wording] way back in the day I thought it made perfect sense, but more recently I’ve come to wonder what exactly “natural” implies. In a multilingual context, it must presumably mean “ ‘natural’ as defined within each language”, while in a multimodal context it would refer to the intersection of the range of different semiotic systems involved, or some common system lying behind all of them (this latter possibility I personally would be rather resistant to because I’m against semiotic universalism as I am all other kinds).

So “natural” presumably can’t mean “given by nature”, because in that case the semantics of all languages would be the same; but if this “naturalness” is “conventional”, to what degree? Or to put it another way, how do the many commonalities across the material and social worlds of human communities get “translated” or “semioticised” into individual human languages. And if all human communication is multimodal, and all semiotic systems are “natural” in their different ways, how do these different “natures” get to be coordinated or mutually-enforcing in communicative contexts?

Blogger Comments:

[1] To be clear, by 'natural' Halliday means a non-arbitrary relation between lexicogrammatical form and the meaning it realises. Halliday (1985: xvii, xviii, xix):

The relation between the meaning and the wording is not, however, an arbitrary one: the form of the grammar relates naturally to the meanings that are being encoded. A functional grammar is designed to bring this out; it is a study of wording, but one that interprets the wording by reference to what it means. …

What this means is that both the general kinds of grammatical pattern that have evolved in language, and the specific manifestations of each kind, bear a natural relation to the meanings they have evolved to express. … the distinction into word classes of verb and noun reflects the analysis of experience into goings-on, expressed as verbs, and participants in the goings-on, expressed as nouns; and so on. …

Since the relation of grammar to semantics is in this sense natural, not arbitrary … 

[2] To be clear, according to Halliday, the non-arbitrary relation between meaning and wording evolved naturally in language; 'natural' is thus not defined differently for different languages.

[3] To be clear, the natural relation between meaning and wording only concerns language, because language is the only semiotic system with a content plane stratified into meaning and wording. That is, language is the only semiotic system that affords the verbal projection of locutions (wordings).

[4] To be clear, according to Halliday, language — evolved and developed from protolanguage — is the social semiotic system that makes other such social semiotic systems possible; see Halliday & Matthiessen (1999: 602-6).

[5] To be clear, the natural relation between semantics and lexicogrammatical form does not entail that the semantics of all languages would be the same, since it is the relation that is shared by languages, not (necessarily) the semantics.

[6] To be clear, in this context, 'natural' (non-arbitrary) is the opposite of 'conventional' (arbitrary).

[7] To be clear, on the one hand, this topic, the construal of experience as meaning, is a different issue from the foregoing discussion on the natural relation between meaning and wording. On the other hand, it assumes that experience is already categorised (as "commonalities across the material and social worlds of human communities") before being construed as meaning by language. SFL Theory assumes the opposite: that experience only becomes meaning when construed as such by semiotic systems. The former is known as the transcendent view of meaning, the latter is known as the immanent view of meaning. See Halliday & Matthiessen (1999: 416, 426, 439, 441, 603, 605).

[8] To be clear, in this context, it is not semiotic systems that are "natural", but the relation between meaning and wording in the only semiotic system with wording: language.

[9] As can be seen from the foregoing, this question is distinct from what Halliday means by a natural relation between meaning and wording. For Halliday (Halliday & Matthiessen 1999: 602), it is context that makes the co-ordination of language with parallel social semiotic systems possible.