Misrepresenting The Grammatical Realisation Of Discourse Semantic Conjunction

Martin (1992: 404):

CONJUNCTION was presented in Chapter 4 as the semantics of the clause complex and is oriented to activity sequences in field.  The way in which CONJUNCTION is realised however (between clause complexes, within clause complexes or within clauses) is very sensitive to mode.

Blogger Comments:

[1] As previously demonstrated, the system of conjunction does not model the semantics of the clause complex.  Instead, it confuses clause complex relations (logical metafunction) and cohesive conjunction (textual metafunction) and excludes several logical relations, the most important of which is projection.  This omission of projection follows from taking cohesive conjunction as the point of departure, since this system is the textual deployment of expansion.

[2] 'To be oriented' is to be positioned in a direction relative to something or someplace else.  The theoretical relation between two levels of symbolic abstraction, such as semantics and context, is realisation.

[3] The system of conjunction (Chapter 4) was not identified with regard to 'activity sequences in field' (Chapter 5).  Instead, temporal clause complex relations were said to realise the field system 'activity sequence', whose feature options were expectancy (realised by and (then) versus implication (realised by if/then).  No discourse semantic system was identified in the discussion (pp321-5).

[4] The notion of clause complex relations within clauses is a logical contradiction.  See the original critique here.

[5] In SFL theory, the realisation of the contextual system of mode in textual meaning and wording  — e.g. cohesive conjunction — varies according to register.

Misrepresenting The Grammatical Realisation Of Discourse Semantic Identification

Martin (1992: 403-4):

IDENTIFICATION is abstracted from nominal group DEIXIS, and alongside the lexicogrammatical systems of SUBSTITUTION and ELLIPSIS is a powerful measure of contextual dependency and thereby strongly associated with MODE.

Blogger Comments:

[1] The discourse semantic system of identification is presented as the semantics of reference.  However, rather than being theorised at a more abstract level than this cohesive system, a reworking of it has merely been moved to a different location within the theory.

[2] The nominal group system of deixis realises an orientation to the '"speaker-now", the temporal–modal complex that constitutes the point of reference of the speech event' (Halliday & Matthiessen (2004: 314).  The system of identification is a reworking of the non-structural cohesive system of reference, not the structural system of deixis.

[3] This discussion (p33) misconstrued context as co-text and material setting.  See original critique here.

[4] The "association" of these systems with the contextual system of mode is not due to their being 'a powerful measure of contextual dependency'.  Substitution–&–ellipsis and reference (the source of identification) are both systems of the textual metafunction, and mode is the contextual stratum system of the textual metafunction.

[5] In SFL theory, the realisation of the contextual system of mode in textual meaning and wording varies according to register.

Misrepresenting The Grammatical Realisation Of Discourse Semantic Ideation

Martin (1992: 403):

IDEATION, as was noted in Chapter 5, is realised through experiential grammar systems (including the LOGICO-SEMANTICS of the clause complex) and is strongly correlated with field.

Blogger Comments:

[1] Realisation statements specifying how ideation is realised in these grammatical systems are not provided.

[2] The realisation of the experiential discourse system of ideation in the logical relations of the clause complex was not discussed in Chapter 5. As the summarising Table 5.26 (p373) makes clear, the system of ideation is said to be realised in the lexicogrammar by transitivity, group rank experiential grammar, lexis as delicate grammar and collocation. The same table identifies the logical discourse semantic system of conjunction as that which is realised by the logico-semantics and interdependency of the clause complex.

The discussion of clause complex relations in Chapter 5 was restricted to temporal relations (p323-5), and these were said to realise the field system 'activity sequence', whose feature options were expectancy (realised by and (then) versus implication (realised by if/then).  (See the relevant critiques for an analysis of most of the inconsistencies involved.)  No experiential discourse semantic system was identified in the discussion.

[3] In SFL theory, the realisation of the contextual system of field in ideational meaning and wording varies according to register.

Misrepresenting Tenor As A 'Register Variable'

Martin (1992: 403):

NEGOTIATION is realised through the interpersonal grammatical systems of MOOD (alongside MODALIZATION, MODULATION and ATTITUDE); taking interpersonal metaphor into account, these systems are all strongly correlated with the register variable tenor.

Blogger Comments:

[1] Realisation statements specifying how negotiation is realised in these grammatical systems are not provided.

[2] This is irrelevant to the point being made.  Realisational relations across strata hold even when not "taking interpersonal metaphor into account".  Interpersonal metaphor is the incongruent varying of such interstratal relations.

[3] In SFL theory, the realisation of the contextual system of tenor in interpersonal meaning and wording varies according to register. 

[4] In SFL theory, tenor is the system that models the interpersonal potential of context.  It is not a "register variable" in the sense of being a system of register because register is a functional variety of language, whereas context is a semiotic system that is more abstract than language.

Misunderstanding The Textual Metafunction And Misrepresenting Context As Register

Martin (1992: 402-3):

And it follows from setting up a text-focussed discourse semantics of this kind (as opposed to a speech act-focussed pragmatics or a proposition-focussed semantics) that systems on this stratum will be concerned with text integrating relations. This does not mean that all discourse systems need therefore to be treated as metafunctionally textual. The grammatical systems that discourse systems are abstracted with respect to need to be taken into account. In addition, the relation of discourse systems to register variables is an important consideration.

Interpreted as an interface between context and grammar, discourse semantics can be seen to have its own metafunctional organisation, reflecting both the organisation of the lexicogrammatical resources realising its meanings as well as the organisation of context into the register variables tenor, mode and field.

Blogger Comments:

[1] In SFL theory, all systems "concerned with text integrating relations" are systems of the textual metafunction, by definition.  Ideational and interpersonal systems are not "concerned with text integrating relations". Any discourse semantic system that is "concerned with text integrating relations" is a system of the textual metafunction.  Any discourse semantic system that is not of the textual metafunction is not "concerned with text integrating relations".

[2] This would go without saying if the meaning of stratification and metafunction were properly understood.

[3] This continues the misrepresentation of context, a semiotic system that is more abstract than language, as register, a functional variety of language that realises a functional variety of context: a situation type.

Misunderstanding The Principles Of Metafunction And Stratification

Martin (1992: 402-3):

Considered from the point of view of an unstratified content plane, this allocation of cohesion to the textual metafunction makes sense; cohesion is about relating ideational and interpersonal meanings to each other, integrating them as text. … 

Considered from the perspective of a stratified content plane however, this metafunctional interpretation of cohesion is not appropriate.  Semantic systems in the model assumed here deal with meanings that are both more abstract and bigger in size than grammatical ones.  And it follows from setting up a text-focussed discourse semantics of this kind (as opposed to a speech act-focussed pragmatics or a proposition-focussed semantics) that systems on this stratum will be concerned with text integrating relations.

Blogger Comments:

[1] The notion of cohesion has, in fact, been theorised on the basis of a stratified content plane since its formulation.  Halliday & Hasan (1976: 304):

It has been pointed out that reference, while it is expressed by grammatical means, is actually a semantic relation, a relation between meanings of particular instances rather than between words or other items of linguistic form.  Substitution and ellipsis, on the other hand, are formal relations between elements at the lexicogrammatical level.

[2] The "allocation" of cohesion to the textual metafunction is consistent with the theoretical definitions of both cohesion and the textual metafunction.  Halliday & Matthiessen (1999: 398): 

The textual metafunction second–order in the sense that it is concerned with semiotic reality: that is, reality in the form of meaning. This dimension of reality is itself constructed by [the] other two metafunctions: the ideational, which construes a natural reality, and the interpersonal, which enacts an intersubjective reality. … The function of the textual metafunction is thus an enabling one with respect to the rest; it takes over the semiotic resources brought into being by the other two metafunctions and as it were operationalises them …

[3] Many of the discourse systems are not more abstract than lexicogrammar.  For example, as many previous posts demonstrate, identification is a misunderstanding of reference, ideation is merger of misunderstandings of lexical cohesion, lexicogrammatical delicacy and transitivity, and the system of conjunction is a merger of misunderstandings of logical relations between clauses and cohesive conjunction between messages.  Simply relocating reworkings of previous work by other theorists to a higher stratum does not make such systems more abstract.

[4] The notion of meanings being "bigger in size" — which is irrelevant to the principle (symbolic abstraction) on which stratification is based — suggests that meaning is conceived here structurally rather than systemically.  In SFL theory, as the name suggests, priority is given to system. With regard to syntagmatic extent, it is appropriate to point out that the textual and experiential discourse semantic units, participant and message part, are both less extensive than the clause, the unit of highest rank in lexicogrammar.

[5] This continues the misunderstanding of stratification.  In SFL, the content plane is stratified into meaning (semantics) and wording (lexicogrammar).  "Grammatical meanings" are the meanings of the semantic stratum that are realised in the wordings of the lexicogrammatical stratum.

[6] This does not follow at all.  A semantics that is "text-focussed" is not limited to "text integrating relations". In SFL theory, the text is a semantic unit and semantics involves all metafunctions.  The text forming systems, on the other hand, are those of the (aptly named) textual metafunction.  The ideational and interpersonal metafunctions are not concerned with integrating text, but with construing experience and enacting intersubjective relations.

Misrepresenting Halliday On The Stratification Of Content

Martin (1992: 401):

With notable exceptions (e.g. Halliday 1984) Halliday's work on English content form has generally assumed an unstratified system/structure cycle organised by rank and metafunction.

Blogger Comment:

This is a very serious misrepresentation.  The stratification of the content plane into two strata, semantics and lexicogrammar, has long been at the very heart of SFL theory — not least because the notion of grammatical metaphor depends on it — and long precedes the work of Halliday's students, such as Martin (1992).  For example, in the work most cited in Martin (1992), Halliday & Hasan (1976: 3) write:

Language can be explained as a multiple coding system comprising three levels of coding, or 'strata': the semantic (meanings), the lexicogrammatical (forms) and the phonological and orthographic (expressions).  Meanings are realised (coded) as forms, and forms are realised in turn (recoded) as expressions.  To put it in everyday terminology, meaning is put into wording, and wording into sound or writing…

The stratification of content, cross-coupled with the distinction of paradigmatic and syntagmatic axes, produces two system–structure cycles, one on each stratum.  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.

It is another of Halliday's students who proposes a single system–structure cycle, though with a stratified model of content.  Halliday & Matthiessen (1999: 429):

In Fawcett’s model, there is only one system–structure cycle within the content plane: systems are interpreted as the semantics, linked through a “realisational component” to [content] form, which includes items and syntax, the latter being modelled structurally but not systemically… 

The importance of modelling content as stratified in SFL theory cannot be overstated.  For example, Halliday & Matthiessen (2004: 25):

The stratification of the content plane had immense significance in the evolution of the human species — it is not an exaggeration to say that it turned Homo … into Homo sapiens. It opened up the power of language and in so doing created the modern human brain. …

Halliday & Matthiessen (1999: 604):

This deconstrual of the content plane into two strata … is a unique feature of the post-infancy semiotic, corresponding to Edelman’s (1992) “higher–order consciousness” as the distinguishing characteristic of Homo sapiens.

Misrepresenting Grammatical Metaphor

Martin (1992: 401):

Grammatical metaphor, like interaction patterns, will be interpreted as a process here, rather than as a synoptic system taking its place alongside the text forming systems proposed for English above.  Metaphorically speaking, it is part of the conversation that must go on among text forming systems across strata if meanings are to be integrated in contextually effective ways. … The gatekeeping function of grammatical metaphor is incorporated as part of English Text's classification of text forming resources in Table 6.12.

Table 6.12. Grammatical metaphor as a texturing interface
Discourse Semantics		Lexicogrammar	Phonology/Graphology
negotiation	grammatical metaphor	substitution & ellipsis
			information
identification		theme
conjunction & continuity			tone concord & tone sequence
ideation		collocation

Blogger Comments:

[1] This confuses realisation with instantiation.  Grammatical metaphor is an incongruent relation of realisation between semantics (meaning) and lexicogrammar (wording).  The deployment of grammatical metaphor during logogenesis is effected by the process of instantiation: the selection of systemic features and the activation of realisation statements.

[2] This is misleading.  Grammatical metaphor is not a relation between the text forming systems across strata.  In SFL theory, the text forming systems are those of the textual metafunction.  Grammatical metaphor involves incongruent realisations between ideational meaning and wording or between interpersonal meaning and wording.

[3] It is theoretically inaccurate to describe the function of grammatical metaphor as a "gatekeeping" interface between strata. Where gatekeeping is a filtering process, grammatical metaphor is an incongruent relation between two levels of symbolic abstraction: meaning and wording.

[4] In SFL theory, information is a system of the lexicogrammatical stratum.  Here it is again misconstrued as a system of the phonological/graphological stratum.  That is, it is misconstrued as expression rather than content.  The phonological systems that realise the grammatical system of information are tonicity (the placement of tonic prominence) and tonality (the placement of tone group boundaries).

Misrepresenting Intrastratal Studies As Interstratal

Martin (1992: 392):

As far as lexical strings and reference chains are concerned, this interaction across strata is better understood than that among discourse systems themselves.  Fries (1981), Hasan (1984), Halliday (1985) and Plum (1988) have all done pioneering work in this area (see also Martin 1991).

Blogger Comment:

To be clear, in SFL theory, the cohesive relations effected by lexical cohesion and reference are located at the level of lexicogrammar and within the textual metafunction.  Studies that examine how the non-structural systems of the lexicogrammar pattern with the structural systems of the lexicogrammar are thus not concerned with "interaction across strata".

To be clear, in discourse semantic theory, lexical strings are construed as experiential structures at the level of discourse semantics, and reference chains are construed as textual structures at the level of discourse semantics.

Accordingly, the work of Fries (1981), Hasan (1984) and Halliday (1985) does not support the relocation of non-structural textual systems at the level of lexicogrammar to structural experiential and textual systems at the level of discourse semantics.

Blurring The Distinction Between Realisation And Instantiation

Martin (1992: 392):

The modularity imposed by stratification is also an important issue.  Discourse systems generate structures which in principle cut across grammatical and phonological ones.

Blogger Comments:

[1] This continues the misconstrual of strata as modules instead of complementary levels of symbolic abstraction.

[2] To be clear, in SFL theory, the relation between system and structure is realisation.  This axial relation is distinct from the process of instantiation — the selection of systemic features and the activation of realisation statements — during logogenesis.

Misunderstanding Stratification

Martin (1992: 391):

Within discourse semantics, the ways in which systems co-operate in the process of making text is much less well understood. … A more explicit account of this co-operation is clearly an urgent research goal; English Text has been concerned not so much with addressing this goal as with making it addressable by proposing four relatively independent discourse modules to beg the question… .  The point is that integrating meanings deriving from different metafunctions is not a task that can be left to lexicogrammar alone.

Blogger Comments:

[1] This continues the misconstrual of the metafunctions as modular units, rather than as complementary perspectives.

[2] This misunderstands the meaning of 'to beg the question'.  Begging the question is a logical fallacy identified by Aristotle in which the premises include the claim that the conclusion is true or (directly or indirectly) assume that the conclusion is true.  See further here or here.

[3] This misunderstands stratification.  The wordings of lexicogrammar realise the meanings of semantics; the meanings of semantics are realised by the wordings of lexicogrammar.  Further, it ignores the trinocular perspective on which linguistic analysis is based in SFL.  Each stratum is viewed 'from above', 'from roundabout' and 'from below'.  This includes looking at lexicogrammar 'from above' (in terms of the meaning being realised) and looking at semantics 'from below' (in terms of the wording that realises it).

Misrepresenting Realisation And Preselection

Martin (1992: 390):

Within grammar, the problem of mapping different systems onto each other is handled by realisation.  Structures deriving from different metafunctional components are conflated and preselect options from constituent ranks until lexicogrammatical options are exhausted.

Blogger Comments:

[1] This is untrue. Conflation is not realisation. The relation between the different metafunctional systems on the lexicogrammatical stratum is not one of realisation.  Realisation is the relation between different levels of symbolic abstraction, as between strata, between function and form and between system and structure.  The metafunctional systems on the lexicogrammatical stratum are of the same level of symbolic abstraction.  The metafunctions on the lexicogrammatical stratum are different perspectives on the same phenomenon: wording.

[2] This is untrue.  Structures do not preselect options.  The selection of a feature of a paradigmatic system can preselect a feature of a system on a lower stratum (Halliday & Matthiessen 1999: 378-9), and, across axes, paradigmatic specifications can select syntagmatic specifications (Halliday & Matthiessen 1999: 94).  The latter has been renamed 'eco-functional selection' (ibid.).

Misconstruing Strata And Metafunctions As Modules

Martin (1992: 390):

Each of the presentations of linguistic text forming resources considered above adopted a modular perspective.  As far as English Text is concerned this has two main dimensions: stratification and, within strata, metafunction.

Blogger Comments:

[1] Taking a modular perspective on strata and metafunctions is a source of serious theoretical errors, as will become apparent as this chapter is examined.  Modules are distinct units that can be combined with others to make more complex structures. This is at odds with the status of strata and metafunctions in SFL theory as complementary perspectives.

[2] This is misleading.  Stratification and metafunction are dimensions of SFL theory; the observation is not an insight of English Text.

Misrepresenting Interstratal Realisation, Grammatical Metaphor & Register

Martin (1992: 389-90):

The relationship of SPEECH FUNCTION to MOOD mediated by interpersonal metaphor is precisely parallel to that between CONJUNCTION and the clause complex as mediated by ideational metaphor.  Because of this it was possible to present a register neutral description of the semantics of dialogue in Chapter 2, just as it was possible to produce a register neutral description of conjunctive relations in Chapter 4.  Both these types of organic relation are essential components of English text forming resources and need to be interpreted systematically as semantic systems in language, not as register specific features of context.

Blogger Comments:

[1] This is untrue.  Speech function and mood are both interpersonal systems — at the level of semantics and lexicogrammar respectively — and the relationship between them is realisation (congruent or metaphorical).  However, whereas clause complexing is a manifestation of the logical metafunction in lexicogrammar, Martin's discourse semantic system of conjunction takes as its point of departure the textual system of cohesive conjunction.  Because this cohesive system is the textual deployment of expansion relations, the other major logico-semantic type, projection, is absent from the discourse semantic model.  This is a major theoretical shortcoming, since clause complexes involving projection are not accounted for semantically.  Crucially, this in turn removes the means of distinguishing congruent vs metaphorical realisations in the grammar.

[2] The realisation relation between strata is not "mediated" by grammatical metaphor.  Realisations are either congruent or metaphorical, the latter being a manifestation of the textual metafunction as a second-order resource (Halliday & Matthiessen 1999: 398-9).

[3] The realisation relation between strata applies to all systems on all strata.  It does not provide a special means of presenting "register neutral" descriptions for two of the four discourse semantic systems.

[4] In SFL theory, the relation between the general system of semantic potential and the semantic systems of specific registers is theorised as instantiation.  The interpersonal semantic systems of specific registers are termed exchange relationships.  Halliday & Matthiessen (1999: 377-8): 

Midway between potential and instance, sets of such strategies cluster within ranges of tenor values. Such a cluster is the interpersonal analogue of a domain in the ideation base: it is a region within the overall interpersonal space of meaning, selected according to tenor, just as a domain is a region within the overall ideational space of meaning, selected according to field. The options in interpersonal meaning that make up the cluster together enact a tenor relationship … We might call such a cluster an exchange relationship to foreground that it is semantic (i.e. constituted in meaning through exchanges of meaning) and that it is interpersonal (rather than one-sidedly personal). To indicate that it is analogous to a domain model, we might have called it an exchange or interaction “model”; but we have avoided that term because it suggests a construal of something and construal is the ideational mode of meaning — it is more like a protocol than a model.

[5] To be clear, the term 'organic relation' is used by Hasan (1985: 81) with respect to the textual metafunction: to differentiate conjunction from the other ('componential') types of cohesion:

These devices are ORGANIC; the terms in the tie are whole message(s) rather than message components…

[6] In SFL theory, the text forming resources are the systems of the textual metafunction.  In discourse semantics, the systems of all metafunctions are said to be text forming.

[7] To be clear, in SFL theory, 'register specific features of context' are the features of specific situation types (field, tenor and mode) that are realised by specific registers of language.  This is not what Martin means.  See the critiques of field in Chapter 5.

Misrepresenting Halliday & Hasan And Confusing Metafunctions

Martin (1992: 389):

The other main difference between English Text and Cohesion in English has to do with the recognition of NEGOTIATION as a linguistic text forming resource.  Halliday and Hasan treat this system as an aspect of the "macro-structure" of text, alongside generic structure — as a feature of register rather than of language (1976: 324 & 327). Halliday's (1984) interpretation of SPEECH FUNCTION as the semantics of MOOD however is preferred here.

Blogger Comments:

[1] In SFL theory, the text forming resources are those of the textual metafunction.  Negotiation, on the other hand, is presented as a system of the interpersonal metafunction.

[2] Halliday & Hasan (1976: 324, 326-7) are concerned with the textual metafunction — not an interpersonal system of negotiation.  They cite the work of Sacks and Schegloff on adjacency pairs to illustrate their point that 'every genre has its own discourse structure', even casual conversation.  Such a macrostructure is the one of three components of texture, the other two being non-structural cohesion and structural theme and information.

[3] This is manifestly untrue in two respects.  Firstly, Halliday & Hasan (1976: 324, 326-7) do not describe the macrostructure of a text 'as a feature of register rather than of language'.  Secondly, the opposition of register vs language is a nonsense, given that register is defined as a functional variety of language.

[4] These are not two alternative views of the same phenomenon.  The macrostructure of the text is a manifestation of the textual metafunction, and varies according to the mode being realised.  Speech function, on the other hand, is a system of the interpersonal metafunction, at the level of semantics.

Misconstruing Textual Grammar As "Redounding With" Interpersonal And Experiential Semantics

Martin (1992: 389):

As far as realisation across strata is concerned, at clause rank SUBSTITUTION and ELLIPSIS redound with NEGOTIATION while at group rank they redound with IDEATION (cf. Hasan's category of co-classification above).

Blogger Comments:

[1] No realisation statements are provided that specify how these interpersonal and experiential discourse semantic systems are realised in the textual lexicogrammatical system of substitution–&–ellipsis.

[2] The proposal here is that the non-structural textual system of substitution–&–ellipsis in the domain of the clause on the lexicogrammatical stratum "construes and is construed by" the (structural) interpersonal system of negotiation on the discourse semantics stratum.  In this grammatical domain, textual wording "construes and is construed by" interpersonal meaning.

[3] The proposal here is that the non-structural textual system of substitution–&–ellipsis in the domain of the group on the lexicogrammatical stratum "construes and is construed by" the (structural) experiential system of ideation on the discourse semantics stratum.  In this grammatical domain, textual wording "construes and is construed by" experiential meaning.

[4] Hasan's (1985: 82) category of co-classification refers to the type of cohesive tie in substitution–&–ellipsis (and lexical cohesion).  It is irrelevant to claims that substitution–&–ellipsis realises either experiential or interpersonal meanings, according to the grammatical domain of the textual cohesion.

Misconstruing Stratification

Martin (1992: 389):

What this argument amounts to saying is that it is more economical to leave SUBSTITUTION and ELLIPSIS in the grammar, generating cohesive ties with respect to grammatical functions defined on that stratum.  Since text forming resources are distributed across all three strata in the meaning making model assumed by English Text, there is nothing remarkable in this.

Blogger Comments:

[1] This betrays Martin's modus operandi of theorising a discourse semantics — taking Halliday & Hasan's ideas from where they fit consistently within the overall theory, and moving them to a new ill-considered theoretical location.

[2] To be clear, in SFL theory, the text forming resources across all strata are those of the textual metafunction.

[3] This continues the confusion of stratification and semogenesis.

[4] There is much that is remarkable in this, not least because the meanings realised this grammatical system need to be modelled by a semantic theory.

Misconstruing Cataphoric Reference

Martin (1992: 388-9):

The difference between presuming grammatical items and presuming meanings is illustrated in [6:4] and [6:5] below.  The lengthy monograph in [6:4bi] does not refer to any particular grammatical unit in [6:4a]; it simply presumes meaning that was there implied:

[6:4]	a.		It took several months of writing
	b.	i.	but in the end the lengthy monograph was complete
		ii	?? but in the end the lengthy one was complete

Blogger Comment:

This has the reference relation the wrong way around.  The reference in this instance is cataphoric, not anaphoric: it refers to the lengthy monograph.  In SFL theory, the nominal group the lengthy monograph is not a reference item.

To be clear, reference is a semantic relation between elements, whereas substitution–&-ellipsis is a lexicogrammatical relation (Halliday & Matthiessen 2004: 538, 561-2).

Resorting To A Misconstrual of Stratification

Martin (1992: 388):

In spite of the fact that SUBSTITUTION and ELLIPSIS are used to link items cohesively between clause complexes, English Text has not treated them as a discourse semantic system.  There are two reasons for this. … The second reason is that SUBSTITUTION and ELLIPSIS presume grammatical functions, not meanings.  At clause rank, they have to be defined interpersonally, with respect to Mood and Residue; at group rank specific nominal and verbal group experiential functions are similarly presumed.

Blogger Comment:

To be clear, in SFL theory, substitution–&–ellipsis is a grammatical system that sets up textually cohesive (non-structural) grammatical relations (Halliday & Matthiessen 2004: 561-2).  All lexicogrammatical (wording) selections realise semantic (meaning) selections.  A comprehensive theory of semantics accounts for the meanings realised in wordings.

Misconstruing Stratification And Grammatical Metaphor

Martin (1992: 388):

In spite of the fact that SUBSTITUTION and ELLIPSIS are used to link items cohesively between clause complexes, English Text has not treated them as a discourse semantic system.  There are two reasons for this.  One is that there is nothing in the grammar to stratify them with respect to.  Unlike NEGOTIATION, IDENTIFICATION, CONJUNCTION and IDEATION they are not a semantic resource with diversified lexicogrammaticalisations — there is no such thing as incongruent (i.e. metaphorical) SUBSTITUTION and ELLIPSIS, nor is there a semantic motif running through the grammar which disperses their realisation.

Blogger Comments:

[1] This misconstrues stratification.  Substitution–&–ellipsis is lexicogrammatical system that realises semantic choices.

[2] The reason there is no metaphorical substitution–&–ellipsis is that it realises textual meanings, not ideational or interpersonal meanings.  There is ideational and interpersonal metaphor, but no textual metaphor.  Metaphor itself is a resource of the textual metafunction.  Halliday & Matthiessen (1999: 398-9):

One manifestation of the second–order nature of the textual metafunction that is important for our purposes is grammatical metaphor. Grammatical metaphor is a ‘second-order’ use of grammatical resources: one grammatical feature or set of features is used as a metaphor for another feature or set of features; and since features are realised by structures, one grammatical structure comes to stand for another.

Halliday & Matthiessen (1999: 401):

But ideational grammatical metaphors typically have a discourse function of this kind; they are as it were pressed into service by the textual metafunction, to provide alternative groupings of quanta of information.

Misrepresenting Halliday & Hasan

Martin (1992: 388):

As Halliday and Hasan point out, this redundancy is typically exploited in the context of repudiation.  The second response in [6:3] illustrates this function.  The Thing drink is presumed from the offer by the substitute one; this allows the unmarked tonic to fall on small, which is contrastive in this environment.

[6:3]	Would you like a drink?		a drink
	— //13 A small gin and tonic thanks //		#
	— //13 Just a small one thanks //		one
	— //13 Just the smaller one thanks //

Note the absence of repudiation in the first response, where gin and tonic is realised explicitly, proposed as a hyponym of drink.  The third response illustrates the difference between repudiation and comparison; one presumes the Thing drink as in the second response, but in addition smaller presumes a set of bigger drinks which are not identified in the second.

Blogger Comments:

[1] Contrary to what is implied here, the discussion in Halliday & Hasan (1976: 93-5) of repudiation in substitution–&–ellipsis does not conceive it in terms of redundancy, and so does not support Martin's interpretation.  [Note the omission of page references.]

[2] This discussion completely misunderstands the notion of repudiation, and the text does not illustrate it.  Cf. Halliday & Hasan (1976: 93):

The notion of repudiation is explained as follows.  In any anaphoric context, something is carried over from a previous instance.  What is carried over may be the whole of what there was, or it may be only a part of it; and if it is only a part of it, then the remainder, that which is not carried over, has to be REPUDIATED.  For example, in

[3:10] We have no coal fires; only wood ones.

fires is carried over anaphorically, but coal is repudiated.

[3] It is not true that using the substitute one 'allows the unmarked tonic to fall on small.  The tonic falls on small to mark contrastive prominence.  If the substitute had not been used, the tonic would still fall on small, in the unmarked case.

[4] There is no repudiation in any of the responses, because there is nothing to repudiate in the offer.  See [2] above.

[5] Since there is no repudiation in the third response, it does not illustrate the difference between repudiation (substitution–&–ellipsis) and comparison (reference).