Christian Matthiessen -- Language Evolving: Notes Towards A Semiotic History Of Humanity

Michael Halliday - Language Evolving: Some Systemic-Functional Reflections On The History Of Meaning

MAK Halliday At Connecting Paths 2010

MAK Halliday at Connecting Paths, Sun-Yat Sen University and City U HK, November 2010

Process Types As "Spectrum"

System Networks Construe A Continuous Semiotic Space

Halliday & Matthiessen (2004: 173):

Like all system networks, this [PROCESS TYPE] network construes a continuous semiotic space.

Terms In Systems Are Fuzzy Categories

Halliday & Matthiessen (2004: 174n):

Systemic terms are not Aristotelian categories. Rather they are fuzzy categories; they can be thought of as representing fuzzy sets rather than ‘crisp’ ones …

Grammatical Labels Reflect Core Category Signification

Halliday & Matthiessen (2004: 199):

… grammatical labels are very rarely appropriate for all instances of a category — they are chosen to reflect its central or ‘core’ signification ( … ‘prototypes’ …). These core areas are the central region for each process type … and the non-core areas lie on the borders between the different process types, shading into one another as the colours of a colour spectrum.

The Principle Of Systemic Indeterminacy

Halliday & Matthiessen (2004: 173):

The world of our experience is highly indeterminate; and this is precisely how the grammar construes it in the system of process type. Thus, one and the same text may offer alternative models of what would appear to be the same domain of experience, construing for example the domain of emotion both as a process in a ‘mental’ clause … and as a participant in a ‘relational’ one …

Process Types: Spherical Ordering

Halliday & Matthiessen (2004: 171-2):

There is no priority of one kind of process over another. But they are ordered; and what is important is that, in our concrete visual metaphor, they form a circle and not a line. (More accurately still … a sphere … .) That is to say, our model of experience, as interpreted through the grammatical system of transitivity, is one of regions within a continuous space; but the continuity is not between two poles, it is round in a loop.

Process Types As Continuous Regions With Core & Peripheral Areas

Halliday & Matthiessen (2004: 172):

The regions have core areas and these represent prototypical members of the process types; but the regions are continuous, shading into one another, and these border areas represent the fact that the process types are fuzzy categories.

Behavioural, Verbal & Existential Process Types As Categories At Boundaries

Halliday & Matthiessen (2004: 171):

Material, mental and relational are the main types of process in the English transitivity system. But we also find further categories at the three boundaries; not so clearly set apart, but nevertheless recognisable in the grammar as intermediate between the different pairs — sharing some features of each, and thus acquiring a character of their own.

Behavioural Processes At The Borderline

Halliday & Matthiessen (2004: 171):

On the borderline between ‘material’ and ‘mental’ are the behavioural processes: those that represent the outer manifestations of inner workings, the acting out of processes of consciousness and physiological states.

Verbal Processes At The Borderline

Halliday & Matthiessen (2004: 171):

On the borderline between ‘mental’ and ‘relational’ are the verbal processes: symbolic relationships constructed in human consciousness and enacted in the form of language, like saying and meaning …

Existential Processes At The Borderline

Halliday & Matthiessen (2004: 171):

And on the borderline between the ‘relational’ and the ‘material’ are the processes concerned with existence, the existential, by which phenomena of all kinds are simply recognised to ‘be’ — to exist or to happen …

Halliday & Matthiessen On Martin's Theme, Method Of Development And Field

Halliday & Matthiessen (1999: 407):

Thematic spaces in an ideational semantic network can be seen as a model of the systemic understanding of Theme and method of development articulated by Martin, where [Martin’s] “field” corresponds to what has been discussed here in terms of ideational semantic networks in the ideation base.

That is:

(1) Halliday & Matthiessen's thematic spaces in an ideational semantic network

is a systemic understanding of

Martin's Theme and method of development

(2) Halliday & Matthiessen's ideational semantics

corresponds to

Martin's field

Halliday & Matthiessen On Martin's Transitivity

Halliday & Matthiessen (1999: 504): 

As we have already noted, Martin, in his systemic treatment of processes in Tagalog, offers a different interpretation of nuclear transitivity: he defines it in terms of orientation, rather than configuration, and hence operates with a significantly different concept of participant function.

How To Distinguish Complement And Adjunct

(1) WHAT DOES THE TEXTBOOK SAY?

Complement

Halliday & Matthiessen (2004: 122-3):

A Complement is an element within the Residue that has the potential of being Subject but is not; in other words, it is an element that has the potential for being given the interpersonally elevated status of modal responsibility — something that can be the nub of the argument.

Adjunct

Halliday & Matthiessen (2004: 123):

An Adjunct is an element that has not got the potential of being Subject; that is, it cannot be elevated to the interpersonal status of modal responsibility.

(2) EXAMPLES

(a) Consider the following clause:

Maureen gave David the book.

Q1. Can ‘David’ be raised to Subject?

A. Yes, as follows: David was given the book by Maureen.

Conclusion: ‘David’ is Complement.

Q2. Can ‘the book’ be raised to Subject?

A. Yes, as follows: The book was given to David by Maureen.

Conclusion: ‘the book’ is Complement.

(b) Consider the following clause:

Maureen gave the book to David.

Q1. Can ‘to David’ be raised to Subject?

A. No. *To David was given the book by Maureen.

Conclusion: ‘to David’ is Adjunct.

(c) Consider the following clause:

The book was given to David by Maureen.

Q1. Can ‘to David’ be raised to Subject?

A. No. *To David was given the book by Maureen.

Conclusion: ‘to David’ is Adjunct.

Q2. Can ‘by Maureen’ be raised to Subject?

A. No. *By Maureen gave the book to David.

Conclusion: ‘by Maureen’ is Adjunct.

Material Vs Semiotic Abstractions

Halliday & Matthiessen (1999: 190ff) provide a taxonomy of simple things based on the participant roles they play in semantic figures — critically those of Senser, Sayer and Actor.

The most general distinction is between conscious and non-conscious.

Within non-conscious, the distinction is between material and semiotic.

Within material, the distinctions are animal, object, substance and abstraction.

Within semiotic, the distinctions are institution, object and abstraction.

Material abstractions — eg depth, colours — typically play the roles of Phenomenon, Carrier and Value. They have no extension in space and are unbounded, and are typically some parameter of a material quality or process.

Semiotic abstractions — eg information, truth — are typically realised by the Range of mental and verbal processes. They are unbounded semiotic substance with no material existence.

There are also intermediate categories in this taxonomy. For example:

Human collectives — eg family — are intermediate between conscious beings and institutions.

Discrete semiotic abstractions — eg thoughts and fears (mental entities) and questions and orders (speech functions) — are intermediate between semiotic objects and non-discrete semiotic abstractions.

A Useful Way To Visualise Instantiation

1. Think of a system network, such as that of TRANSITIVITY (IFG3 p302). Think of it as coloured black.

2. Now, for example, think of a clause.

3. Now colour green all the features and realisation statements that are selected for that clause.

The term 'system' refers to the entire TRANSITIVITY network.

The term 'instance' refers to just the green bits.

The term 'instantiation as process' refers to the process of applying the colour green.

The term 'instantiation as scale' — 'the cline of instantiation' — refers to the relation between the entire system and the green bits.

The green bits are both a subpotential of the system, and the "activation" of that subpotential.

The instance is the "activated" portion of the system. The relation of the instance to the system is the relation of the "activated" portion to the system as a whole.

Two Reasons Why Halliday’s And Martin’s Models Of Stratification Cannot Be Integrated

1) They do not mean the same thing by ‘register’.

For Halliday, ‘register’ means a functional variety of language (with roots back to the Prague School and Firth).

Martin, on the other hand, equates ‘register’ with Halliday’s ‘context’, which removes the notion of register as a functional variety of language from the model. This is because a higher stratum is not a functional variety of a lower stratum; eg lexicogrammar is not a functional variety of phonology.

2) They do not mean the same thing by ‘context’.

For Halliday, ‘context’ is

• what people do with language (H&M 1999: ix),
• the ‘semiotic environment’ of language (p375),
• the “culture”, considered as a semiotic potential (p606).

Halliday’s ‘context’ is a level of abstraction that is realised by language.

Martin, on the other hand, in describing his model of stratification writes (1992: 496):

… the size of the circles also reflects the fact that the analysis tends to focus on larger units as one moves from phonology to ideology.  Thus the tendency at the level of phonology to focus on syllables and phonemes, at the level of lexicogrammar to focus on the clause, at the level of discourse semantics to focus on an exchange or "paragraph", at the level of register to focus on a stage in a transaction, at the level of genre to focus on whole texts …

As this quote makes clear, Martin’s ‘context’ refers to levels within language, since 'a stage in a transaction' and 'whole texts' are units of language.

Halliday & Matthiessen On Martin's Register

Halliday & Matthiessen (2004: 27n):

Here the term ‘register’ thus refers to a functional variety of language. It has also been used in a related, but different way, to refer to the contextual values associated with such a functional variety (see Martin, 1992; cf Matthiessen 1993).

Theme And Mood

Theme

Halliday & Matthiessen (2004: 85):

(i) Initial position in the English clause is meaningful in the construction of the clause as message; specifically, it has a thematic function.

(ii) Certain textual elements that orient the clause within the discourse, rhetorically and logically, are inherently thematic.

(iii) Certain other elements, textual and interpersonal, that set up a semantic relation with what precedes, or express the speaker’s angle or intended listener, are characteristically thematic; this includes finite operators, which signal one type of question.

(iv) These inherently and characteristically thematic elements lie outside the experiential structure of the clause; they have no status as participant, circumstance or process.

(v) Until one of these latter appears, the clause lacks an anchorage in the realm of experience; and this is what completes the thematic grounding of the message.

Unmarked Theme In Declarative Clauses

Halliday & Matthiessen (2004: 73):

In a declarative clause, the typical pattern is one in which Theme is conflated with Subject; … We shall refer to the mappaing of Theme on to Subject as the unmarked Theme of a declarative clause. The Subject is the element that is chosen as Theme unless there is good reason for choosing something else.

Marked Themes In Declarative Clauses

Halliday & Matthiessen (2004: 73, 74):

A Theme that is something other than the Subject, in a declarative clause, we shall refer to as a marked Theme. The most usual form of marked Theme is an adverbial group … or prepositional phrase … functioning as Adjunct in the clause. Least likely to be thematic is a Complement, which is a nominal group that is not functioning as Subject — something that could have been a Subject but is not … . Sometimes even the Complement from within a prepositional phrase functions as Theme … .

Theme In Exclamative Declarative Clauses

Halliday & Matthiessen (2004: 74):

There is one sub-category of declarative clause that has a special thematic structure, namely the exclamative. These typically have an exclamatory WH-element as Theme … .

Theme In Polar Interrogative Clauses

Halliday & Matthiessen (2004: 75, 76):

In a yes/no interrogative, which is a question about polarity, the element that functions as Theme is the element that embodies the expression of polarity, namely the Finite verbal operator. … but, since that is not an element in the experiential structure of the clause, the Theme extends over the following Subject as well.

Theme In WH- Interrogative Clauses

Halliday & Matthiessen (2004: 75):

In a WH- interrogative, which is a search for a missing piece of information, the element that functions as Theme is the element that requests this information, namely the WH- element … whether Subject, Adjunct or Complement.

Theme In ‘You-&-Me’ Imperative Clauses

Halliday & Matthiessen (2004: 76):

… here, let’s is clearly the unmarked choice of Theme.

Theme In ‘You’ Imperative Clauses

Halliday & Matthiessen (2004: 76):

… although the ‘you’ can be made explicit as a Theme … this is clearly a marked choice; the more typical form is … with the verb in thematic position. … here, therefore, it is the Predicator that is the unmarked Theme.

Theme In Negative Imperative Clauses

Halliday & Matthiessen (2004: 77):

… the principle is the same as with yes/no interrogatives: the unmarked Theme is don’t plus the following element, either Subject or Predicator. Again there is a marked form with you, … where the Theme is don’t you. There is also a marked contrastive form of the positive, … where the Theme is do plus the Predicator … .

Predicator As Theme

Halliday & Matthiessen (2004: 77):

The imperative is the only type of clause in which the Predicator (the verb) is regularly found as Theme. This is not impossible in other moods … but in such clauses it is the most highly marked choice of all.

Adjunct As Theme In Imperative Clauses

Halliday & Matthiessen (2004: 78):

Imperative clauses may have a marked Theme, as when a locative Adjunct is thematic in a clause giving directions … The adjunct part of a phrasal verb may serve as marked Theme in an imperative clause with an explicit Subject, as in Up you get! … .

System Networks

System networks are not flowcharts.

Nothing flows through the network.

A speaker does not choose how far to go in delicacy.

System networks map how features relate to each other.  The architecture of the network can be understood in terms of the logical relations set out in the theory: specifically, the three types of expansion.

elaboration: delicacy

extension: conjunction, disjunction

enhancement: entry conditions

The arrows in a system network represent the relation of condition, not temporality.
A "traversal" of the network is a specific map of logically related features.

Why ‘Realisation’ Applies To Both Strata And Rank And Where The Difference Lies

The overall architecture of SFL theory can be understood in terms of theoretical constructs within the model, namely: relational processes (identifying and attributive) and logico-semantic relations (expansion and projection).

‘Realise’ is an intensive identifying process, which means that it combines ‘identifying’ with ‘elaboration’.

In the case of strata, where a lower stratum realises a higher stratum, the relation between them is thus identifying + elaboration.

The rank scale, on the other hand, is organised in terms of composition, which is a subtype of extension. ‘Realise’ is used on the rank scale to relate the function of a higher rank to the form of a lower rank (eg Process is realised by verbal group). This relation between function and form is thus also identifying + elaboration.

The similarity thus lies in elaboration + identifying being an organising principle for both stratification and the rank scale.

The difference lies in the fact that, whereas stratification involves only elaboration + identifying, the rank scale combines extension (relation between forms) with elaboration + identifying (function–form relations).

(The term ‘realisation’ is used wherever there is an intensive identifying (token-value) relation in the theory. eg between system and structure on a given stratum, between strata, between function and form in the rank scale.)

—∞—

Halliday & Matthiessen (1999: 4n):

The ambiguity [of ‘level’ as either stratum or rank] resides in the overlap of two grammatical relations, those of elaboration (‘be’) and of extension (‘have’) …

Distinguishing The Instance From The System

The Difference Between Context And Register

Context is realised by language.

Registers are types of language (that realise types of context).

The context-language relation is value-token.

The register-language relation is carrier-attribute.

Context is more abstract than language.

Registers (of language) are more specific than language (in general).

Halliday & Matthiessen On Fawcett's Relational Processes

Halliday & Matthiessen (1999: 504): 

His abandonment of the distinction between attributive and identifying seems harder to motivate, since this cannot in fact be explained as a textual (thematic) system in the way that Fawcett proposes.

Halliday & Matthiessen (2004: 229):

Note therefore that Identified–Identifier cannot simply be explained as Given–New in an ‘identifying’ clause; not surprisingly, since the former are experiential functions whereas the latter are textual.

Halliday & Matthiessen On Fawcett's Relational Processes

Halliday & Matthiessen (1999: 504): 

Fawcett incorporates into the “relational: possessive” category, processes of giving and acquiring; reduces the circumstantial to locational processes only; and includes within these, processes of going and sending. As is to be expected, this alternative analysis embodies certain generalisations that are not made in our account of figures, and ignores certain others which are.

Halliday & Matthiessen (1999: 505):

Fawcett’s alternative model for relational processes, with its particular features such as treating ‘giving’ and ‘placing’ as agentive possessives and locatives (‘make…have’, ‘make…be at’) rather than as material dispositives, has to be understood in its total explanatory context: 

(i) in relation to its repercussions within the transitivity system, both the trinocular perspective on transitivity itself (from above, as generalisations about meaning; from roundabout, its consequences for agnateness, delicacy and the move towards lexis; from below, as regularities in the realisation) and the overall topology of content — transitivity in relation to the semantic construal of causality, agency, disposal, and so on; 

(ii) in relation to Fawcett’s architectural design, which differs from ours in having a single system–structure cycle for the two strata of semantics and lexicogrammar (his “syntax”) and then adding a further level of description that is expressed in cognitive terms.

‡

The model of Halliday & Matthiessen (2004: 212) has the explanatory advantage of revealing the following consistent systemic proportionalities:

Thus static location in space [enhancement] is construed relationally …

but dynamic motion is construed materially … 

static possession [extension] is construed relationally …

but dynamic transfer of possession is construed materially …

and static quality [elaboration] is construed relationally … 

but dynamic change in quality is construed materially …

which reflects the more general contrast in the ideational semantics (Halliday & Matthiessen 1999: 131):

The domain outside this conscious-semiotic [sensing, saying] centre of the ideational universe is then quintessentially either active (doing) or inert (being) …

Halliday & Matthiessen On Fawcett's Extra-Linguistic Knowledge

Halliday & Matthiessen (1999: 429):

It follows then … that for us [Fawcett’s (extra-linguistic)] “knowledge of the universe” is construed as meaning rather than as knowledge. This meaning is in the first instance created in language; but we have noted that meaning is created in other semiotic systems as well, both other social-semiotic systems and other semiotic systems such as perception. Our account gives language more of a central integrative rôle in the overall system. It is the one semiotic system which is able to construe meanings from semiotic systems in general.

Halliday & Matthiessen (1999: 3):

We contend that the conception of ‘knowledge’ as something that exists independently of language, and may then be coded or made manifest in language, is illusory. All knowledge is constituted in semiotic systems, with language as the most central; and all such representations of knowledge are constructed from language in the first place.

Halliday & Matthiessen On Fawcett's System–Structure Cycle

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; […] 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.

… 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. Since we [unlike Fawcett] allow for a stratification of content systems into semantics and lexicogrammar, we are in a stronger position to construe knowledge in terms of meaning. That is, the semantics can become more powerful and extensive if the lexicogrammar includes systems.

—∞—

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.

Martin's Claim That All Strata Make Meaning

Grammatically, this is an abstract creative material clause.

All strata : Actor
make : Process: material: creative
meaning : Goal

It is consistent with the notion of 'semogenesis': the creation of meaning.

On the other hand, an attributive reading, such as 'All strata have meaning' is not consistent with either 'semogenesis' or the dimension of stratification.

Meaning is not an attribute of every stratum.

Meaning is but one stratum in the levels of symbolic abstraction.

Wording [lexicogrammar] realises meaning [semantics].

Sounding [phonology] realises the realisation of meaning [semantics] in wording [lexicogrammar].

SFL's Architecture Of Language Classified By SFL Relations

A. system networks
i. delicacy = elaboration + attributive
ii. disjunction = extension: alternative
iii. conjunction = extension: addition
iv. entry condition = enhancement: condition
v. realisation statement = elaboration + identifying

B. rank scale = extension: composition

C. realisation = elaboration + identifying
1. higher stratum realised by lower stratum [stratification]
2. system realised by structure [axis]
3. function realised by form [rank scale]

D. instantiation cline = elaboration + attributive

Nominal Groups Inside Prepositional Phrases: Indirect Participants

Q1. WHAT ARE THE EXPERIENTIAL IMPLICATIONS FOR THE FACT THAT THE COMPLEMENT OF A PREPOSITION (MINOR PREDICATOR) CAN OFTEN BE RAISED TO SUBJECT?

Halliday & Matthiessen (2004: 296-7):

… the Complement of a preposition can often emerge to function as Subject … This pattern suggests that Complements of prepositions, despite being embedded in an element that has a circumstantial function, are still felt to be participating, even if at a distance, in the process expressed by the clause.

Q2. HOW DO WE DISTINGUISH THESE TWO TYPES OF PARTICIPATION?

Halliday & Matthiessen (2004: 261):

We can make a contrast, then, between direct and indirect participants, using ‘indirect participant’ to refer to the status of a nominal group that is inside a prepositional phrase …

Q3. HOW DO WE DISTINGUISH INDIRECT PARTICIPANTS FROM CIRCUMSTANCES?

Halliday & Matthiessen (2004: 278):

Wherever there is systematic alternation between a prepositional phrase and a nominal group, as in all the instances in Participant functions realised by prepositional phrases, the element in question is interpreted as a participant.

Q4. WHAT IS THE FUNCTIONAL DIFFERENCE BETWEEN DIRECT AND INDIRECT PARTICIPANTS?

Halliday & Matthiessen (2004: 295-6):

… the choice of ‘plus or minus preposition’ with Agent, Beneficiary and Range … serves a textual function. … The principle is as follows. If a participant other than the Medium is in a place of prominence in the message, it tends to take a preposition (ie to be construed as ‘indirect’ participant); otherwise it does not. Prominence in the message means functioning either (i) as marked Theme (ie Theme but not Subject) or (ii) as ‘late news’ — that is, occurring after some other participant, or circumstance, that already follows the Process. In other words, prominence comes from occurring either earlier or later than expected in the clause; and it is this that is being reinforced by the presence of the preposition. The preposition has become a signal of special status in the message.

Lexis As Most Delicate Grammar

The most delicate features on the lexicogrammatical stratum are not lexical items.  The cline of lexicogrammatical delicacy is not a scale of increasingly delicate lexical items.

Lexical items are the "output" of the network.  Each lexical item is a "bundle" of co-selected features, just as the phoneme /b/ is a bundle of the articulatory features {bilabial, voiced, stop}.

Halliday & Matthiessen (1999: 198-9):

The paradigmatic strategy … is typically associated with feature networks: that is, networks made up of systems of features, such that each lexical item (as the name of a thing) realises a certain combination of these features selected from different systems within the network — a particular clustering of systemic variables. … This resource, the construal of systematically related lexico-semantic sets, illustrates well the principle of “lexis as most delicate grammar”. …

Note that it is not (usually) the lexical items themselves that figure as terms of the systems in the network.  Rather, the systems are systems of features, and the lexical items come in as the synthetic realisation of particular feature combinations.  Thus lexis (vocabulary) is part of a unified lexicogrammar; there is no need to postulate a separate “lexicon” as a pre-existing entity on which the grammar is made to operate.

Semantics Is Not Just A Relabelling Of Lexicogrammar

Halliday & Matthiessen (1999: 26): 

Thus when we move from the lexicogrammar into the semantics, as we are doing here, we are not simply relabelling everything in a new terminological guise. We shall stress the fundamental relationship between (say) clause complex in the grammar and sequence in the semantics, precisely because the two originate as one: a theory of the logical relationships between processes.

Halliday & Matthiessen (1999: 604)

But in modelling the semantic system we face a choice: namely, how far “above” the grammar we should try to push it. Since the decision has to be made with reference to the grammar, this is equivalent to asking how abstract the theoretical constructs are going to be. We have chosen to locate ourselves at a low point on the scale of abstraction, keeping the semantics and the grammar always within hailing distance. There were various reasons for this. First, we wanted to show the grammar at work in construing experience; since we are proposing this as an alternative to cognitive theories, with an “ideation base” rather than a “knowledge base”, we need to posit categories such that their construal in the lexicogrammar is explicit. Secondly, we wanted to present the grammar as “natural”, not arbitrary; this is an essential aspect of the evolution of language from a primary semiotic such as that of human infants. Thirdly, we wanted to explain the vast expansion of the meaning potential that takes place through grammatical metaphor; this depends on the initial congruence between grammatical and semantic categories. 

But in any case, it is not really possible to produce a more abstract model of semantics until the less abstract model has been developed first. One has to be able to renew connection with the grammar. 

David Rose On Beyond the Clause (Complex)

David Rose wrote at 11:07 on 2/3/14 to sysfling:

Beyond the clause (complex) is the text. Here is a relevant quote from Ch 1 of Working with Discourse: meaning beyond the clause…

Blogger Comments:

[1] Beyond the clause (complex) is the cotext at the level of wording (lexicogrammar).

In terms of stratification, the clause is the uppermost rank unit in stratum of wording (lexicogrammar) — and the clause complex is an expansion of that unit — whereas text is the uppermost 'unit' in the stratum of meaning (semantics).   In terms of instantiation, text is an instance of the system, and so can be viewed at the level of meaning (semantics), wording (lexicogrammar) and sounding (phonology).

[2] Meaning 'beyond the clause' is the meaning (semantic stratum) that is realised on the grammatical stratum as the textual systems of cohesion: CONJUNCTION, REFERENCE, SUBSTITUTION & ELLIPSIS and LEXICAL COHESION.  See Halliday & Matthiessen (2004: 524-85).