Realisations Of Expansion Relations

Halliday & Matthiessen (1999: 118-9):

Relations of expansion are typically realised in the grammar by conjunctions or conjunctive expressions linking a pair of clauses, either paratactically or hypotactically (e.g. that is, in other words; and, but, or, also, besides; so, yet, then, when, if, because, unless). Some of these may realise more than one category; for example, 

but may be adversative 'and yet' (extending), or concessive 'and in spite of this' (enhancing);  

while may be additive 'and in addition' (extending), or temporal 'and at the same time' (enhancing);  

or may be alternative 'or else' (extending: alternatives in the external world, like take it or leave it), or restating 'in other words' (elaborating: alternatives in the world 'internal' to the discourse, like they are reduced to the smallest size, or microminiaturised.). 

Overlaps in realisation of this kind show that the primary categories we have set up do in fact shade into one another; in particular, extending in some sense occupies a space intermediate between elaborating and enhancing, and shares a fuzzy borderline with each.

Examples Of The Principal Categories Of Expansion

Halliday & Matthiessen (1999: 118):

The following passage of spontaneous conversation exemplifies the principal categories of expansion (note than it also contains one example of projection):

KEY: parataxis [equal]: 1, 2; hypotaxis [unequal]: 𝛂, 𝛃); elaboration: =; extension: +; enhancement: x. (Taken from Svartvik & Quirk, 1980:430.)

The Cline From Sequence To Figure

Halliday & Matthiessen (1999: 118):

There is no sharp line between a figure and a sequence of figures: a quantum of experience is not defined before it is construed, and the grammar rather sets up a cline from sequence to simple figure. (This is, in fact, a variable across languages; we shall return to this point from a cross-linguistic point of view in Chapter 6.) For example:

Expansion: Limiting Cases

Halliday & Matthiessen (1999: 117-8):

From one standpoint the limiting case of expansion would be accumulation in a temporal sequence (hence our general term "sequence" for the product of this construal): 'a happens, then x happens'. This gives value to a as the temporal circumstance of x. From this we could derive a wide range of more complex enhancing relations: variations on the simple temporal sequence (after, before, at the same time, immediately after, &c.) and further circumstances such as cause, condition, concession, and their subcategories.

We shall not try to enumerate these here; they are familiar as categories at the level of lexicogrammar (see Halliday, 1985: Ch. 7; Matthiessen, I995b: Ch. 3). These are 'enhancements', multiplying one figure by another, as it were. But figures may also be added to one another, making them part of the same story without assigning any kind of logical priority to either: 'x as well as/instead of/in contradistinction to a'. We have referred to these as extensions. And there is the third type of accumulation where the logical relation is that of 'equals': 'x is the same figure as a'. Here at this end the limiting case is a simple repetition; this may be further elaborated in such a way that one figure is reworded as another, or else further clarified or brought out by an example.

Expansion

Halliday & Matthiessen (1999: 117):

Expansion can be thought of as construing another dimension of experience, so that superimposed, as it were, on the construal of a figure — a basic fragment of experience in the form of a quantum of change (event, action, behaviour &c.) — is the construal of a nexus between two figures, such that one such fragment is non-randomly (i.e. meaningfully) cumulated with another.

Sequences: Expansion Relations Between Figures

Halliday & Matthiessen (1999: 117):

Expansion is a highly generalised type of relation, whereby one figure is joined logico-semantically to another figure by a relator to form a sequence of the same order of reality. It will be easier to characterise expansion at the next step in delicacy, in terms of its immediate subtypes: elaborate (reiterate), extend (add), and enhance (qualify):

(i) elaboration is a (partial) identity relation between figures: one is identified with another with a difference in perspective (it matters a lot; it plays an important role) or one is included under another as an example (it plays an important role; e.g., it provides the infrastructure). These are clearly related to one another: identity is the limiting case of inclusion and inclusion is partial identity.

(ii) extension is an additive relation between figures: a sequence is made bigger by the addition of another figure. This may involve pure addition ('and': he is too young and he doesn't speak the language) or addition with an adversative feature ('and yet': he speaks the language but he is too young ). As a variant of addition, we also have alternation (he is too young or else he is just immature).

(iii) enhancement is a circumstantial or qualifying relation between figures: it is, in a sense, extension plus a circumstantial feature — 'and' + time ('and then', 'and at the same time', etc.), 'and' + manner ('and in the same way', 'and likewise'), 'and' + cause ('and therefore', etc.), etc.: it is autumn, so the leaves are turning brown.

The Possibilities Of Projecting Sequences Summarised

Halliday & Matthiessen (1999: 116):

To sum up the distinctions discussed so far. In a projecting sequence, a figure of thinking or saying projects another figure as an idea or a locution, either of which may be a proposition or a proposal: see Figure 3-8.

Proposals Metaphorically Realised By Modulated Indicative Clauses

Halliday & Matthiessen (1999: 115-6):

We saw that quoted offers may be realised by modulated indicative clauses. This would seem to be an exception to the generalisation that indicative clauses realise propositions. However, it is a principled one: the type of indicative clause involved is precisely the type that lies closer in the interpersonal clause grammar to imperative clauses — modulated indicative clauses, i.e. those with an imperative modality. As can be expected, this is then also a possible realisation of commands; it is a metaphorical strategy for expanding the meaning potential, typically to vary the tenor between speaker and listener. For example:

she told him: "Do the laundry!"; "You should do the laundry."

she asked him: "Could you do the laundry?"

Commands can be reported in the same way:

she told him —> to do the laundry.

she told him —> he should do the laundry.

she asked him —> whether he could do the laundry.

Quoted vs Reported Offers

Halliday & Matthiessen (1999: 115):

Quoted offers naturally retain the property of being variously realised in the mood system:

declarative: She said: "I can do the laundry."

interrogative: She said: "Shall I do the laundry?"

imperative: She said: "Let me do the laundry!"

However, reported offers can always be realised in the same way as reported commands — as in the case of reported propositions, the distinction in orientation between giving and demanding is realised by the projecting clause; for example:

command: she told him —> to do the laundry

offer: she offered (promised; threatened) —> to do the laundry

The category of reported projected proposal is thus realised generally as a perfective non-finite clause when it is reported.

Offers: Grammatical Realisations

Halliday & Matthiessen (1999: 115):

Offers, in English, are not grammaticalised in the mood system; that is, while the other categories, statements, questions and commands, have corresponding mood categories in the grammar (declarative, interrogative and imperative), offers do not. They may be realised by any of the mood categories; for example:

declarative: I can do the laundry.

interrogative: Shall I do the laundry?

imperative: Let me do the laundry!

Significantly, the indicative clauses realising offers are modulated; they select for an imperative modality of readiness or obligation (see e.g. Halliday, 1985: Section 10.4).

Reported Projections: Speech Function And Mood

Halliday & Matthiessen (1999: 114-5):

As was said above, in quoting, where the projected clause retains its mood, the general verb say can be used whatever speech function is being projected. In reporting, on the other hand, the projected clause is no longer specified for mood; its speech function is signalled by the verb in the projecting clause (asked, ordered, etc.). This gives the projected element more of an ideational status (cf. its treatment in traditional grammar as "object" of the projecting verb), and opens up the way to a series of agnate expressions such as (the king ordered) "Execute him!" / that he should be executed / him to be executed/him executed/his execution.

Grammatical Realisations Of Projected Propositions vs Proposals

Halliday & Matthiessen (1999: 114):

Grammatically the distinction between propositions and proposals is constructed as follows.

When the projection is reported, propositions are realised by finite bound clauses, i.e. clauses that select for primary tense or modality; and proposals are realised by irrealis (infinitival) non-finite bound clauses — taking the contrasting examples from above: 

(he said) that he had done the laundry : (she told him) to do the laundry.

When the projection is quoted, propositions are realised by indicative clauses, i.e. clauses that select for primary tense or modality; and proposals are realised by imperative clauses — taking the contrasting examples from above: 

(he said) "I have done the laundry" : (she said to him) "do the laundry". 

These realisational patterns are summarised below:

Projected Proposals (Non-Actual) vs Projected Propositions (Actual)

Halliday & Matthiessen (1999: 114):

Projected proposals are non-actual, or uninstantiated: that is, the occurrence of a projected proposal is always future in relation to the figure that projects it. In contrast, propositions are actual, or instantiated: that is, the occurrence of a proposition is located in actual time (which may be past, present or future).

Projected Propositions vs Proposals

Halliday & Matthiessen (1999: 114):

The interpersonal system for dialogic interaction thus creates a fundamental distinction between 'content' as proposition and 'content' as proposal. This distinction is then reflected in the ideational system of projection: a figure is projected in one or other of these two modes, as a proposition or as a proposal. The two categories combine freely with ideas and locutions:

Blogger Comments:

The table confusingly presents verbally reported locutions as ideas. Instances of genuine reported ideas are:

• he thought —> that he had done the laundry (proposition)
• she wanted him  —> to do the laundry (proposal).

'Proposition ideas' are projected by cognitive mental Processes, whereas 'proposal ideas' are projected by desiderative mental Processes.

Proposition (Information) vs Proposal (Goods-&-Services)

Halliday & Matthiessen (1999: 113-4):

Information is either given or demanded; in either case it is encoded as a proposition. Similarly, goods & services are either given or demanded, but the linguistic act is a mediating one, specifying the (typically non-linguistic) action that embodies the exchange, as an offer to do or a demand that something should be done. This is a proposal for a deed, one that commits either the speaker (giving) or the addressee (demanding); together with a third type, which is a combination of these two, viz. suggestion: Let's do the laundry.

Interpersonal Semantics: Semiotic vs Material Commodities of Exchange

Halliday & Matthiessen (1999: 113):

Interpersonal semantics is centrally concerned with varieties of symbolic exchange. Here, as in other places in the meaning base, the system is organised in such a way that it creates a difference between non-symbolic reality and symbolic reality, between phenomena and metaphenomena. The "commodity" that is being exchanged in interpersonal dialogue is either semiotic or material: it is either one that is construed by language itself — information — or it is one that exists independently of language — goods & services. In the first case, language constitutes the exchange; in the second, it facilitates the exchange of a nonlinguistic commodity. For example:

No Temporal Constraints On Projection

Halliday & Matthiessen (1999: 112):

In general, there is no temporal constraint between the projected and the projecting figures: the projected figure may be past, present or future relative to the projecting one. The dimension affected by projection [is] order of reality, not time. However, there is a major dichotomy between two types of projection which does have temporal implications. To explore it, we have to look briefly at a fundamental interpersonal category, that of mood.

Default Alignments: Quoted Locutions And Reported Ideas

Halliday & Matthiessen (1999: 111-2):

The second consequence is that although the two variables, locution/idea and quoting/reporting, can combine in either of the two possible alignments, there is a natural default condition, which is that locution goes with quoting and idea goes with reporting, as shown in Figure 3-6. 

The reason for this is clear where the first-order phenomenon is one of saying (prototypically shared semiosis), the projected figure can be presented as if it was also of the same order: Harriet said + "Shall I feed the cat?" (reversible as "Shall I feed the cat?" + said Harriet). 

Where the first-order phenomenon is one of sensing (unshared semiosis), the projected figure has no counterpart on the first-order plane of experience and cannot be naturally presented as if it had; so, Harriet wondered + whether she should feed the cat. 

But there is always the possibility of semogenic extension by cross-coupling; so we also find the corresponding marked alignments: locution/report Harriet asked + whether she should feed the cat, and idea/quote Harriet wondered + "Shall I feed the cat?" — the latter being again reversible. See Halliday (1985: Chapter 7) and Nesbitt & Plum (1988).

Quoting vs Reporting: Wording

Halliday & Matthiessen (1999: 111):

There are two consequences of this. The first is that, in quoting, the form of wording is still that of the first-order realm of experience: "Shall I feed the cat ?" — taken by itself, there is nothing about this to show it is a projection. In reporting, on the other hand, the form of wording is clearly marked as being of the second order: whether she should feed the cat has lost its first-order semantic features — and hence has to be projected by words that specify its particular speech function: asked/wondered, not said/thought.

Equal vs Unequal Semantic Weight: Quoting vs Reporting

Halliday & Matthiessen (1999: 111):

But there is another variable intersecting with the above. We pointed out in Chapter 2, Section 2.2 that whenever two figures are related in a sequence, they may be either equal or unequal in semantic weight. We illustrated this interdependency system with reference to expansion; but it applies also to projection. The projecting and the projected figures may have equal status in the sequence: this relation is that of quoting, as in Harriet said/thought "Shall I feed the cat?". Or they may have unequal status: this relation is that of reporting, as in (Harriet asked/wondered) whether she should feed the cat. 

We can relate this to the fact that projection is the creating of a second-order reality. In reporting, the status of the two parts is unequal — the projected figure is dependent on the projecting: hence the projected figure is clearly construed as belonging to a different, second-order plane of reality — a reality that is made of meaning, as it were. In quoting, on the other hand, the two have equal status as independent figures; the projected figure is thus projected as if it was still part of the same first-order reality.

Symbolic Processing: Sensing vs Saying

Halliday & Matthiessen (1999: 110-1):

Let us look more closely at these two modes of projecting, in both of which the projecting figure represents symbolic processing, the processing that brings the other figure into symbolic existence. Either the projection takes the prototypical form of semiosis: it is presented as verbal, shared, an exchange or joint construction of meanings (e.g. Harriet said 'me feed cat?'); or it is fashioned into a derived semiotic form, unshared, interiorised, and without any meaning being exchanged (e.g. Harriet thought 'me feed cat?'). In the first, the projecting figure is one of saying; the projected is referred to as locution. In the second, the projecting figure is one of sensing; the projected is referred to as idea. Ideas are projections which are sensed, locutions are projections which are said.

Thinking Ideas (Meanings) vs Saying Locutions (Wordings)

Halliday & Matthiessen (1999: 109, 110):

In (i) the 'content' of his thinking is the meaning. It remains internal to his consciousness and unrealised, i.e. unworded. In (ii) the 'content' of his saying is the wording. It has been externalised, realised as a wording. We will refer to these in the context of projection as (i) ideas and (ii) locutions. The contrast between these two statuses of projected figures is typically reflected in the shape of balloons in comic strip projections: see Figure 3-5.

Projected Content: Meaning (Idea) vs Wording (Locution)

Halliday & Matthiessen (1999: 109, 110):

We have seen that the content plane is stratified into two levels — semantics (the level of meanings) and lexicogrammar (the level of wordings). Consequently, we would expect projections to be located at either or both of these levels, and this is indeed what happens: a projected figure is either a meaning or a wording: see Figure 3-4.

Projected Content

Halliday & Matthiessen (1999: 108-9):

We have suggested that the relation of projection sets up one figure on a different plane of reality — we refer to this as the second-order or semiotic level. This second-order level of reality is the content plane of a semiotic system. That is to say, the projected figure is projected in the form of 'content': see Figure 3-3.

Conscious And Non-Conscious Projection

Halliday & Matthiessen (1999: 106-7, 108):

We could say that the world view that is constructed in this way is (quite reasonably, from our point of view) focussed on humans, with human consciousness occupying a privileged place. Thus typically only humans can project into second-order reality. However, since human consciousness is the locus of semiotic activity, it has the power of interpreting as metaphenomenon that which is manifested by some other, non-conscious symbolic source. Thus while "sensing" (that is, semiotic activity that is unmanifested, like thinking) does require a human senser, saying can be ascribed to a nonhuman as well as to a human sayer (cf. he thinks —> the moon is a balloon vs he says/the book says —> the moon is a balloon. Figure 3-2 shows the relationship between the two orders of metaphenomena and the presence of consciousness in the act of projection.

Expansion vs Projection In Comic Strip Conventions

 Halliday & Matthiessen (1999: 106, 107):

We can explore this distinction through the conventions of comic strips. Expansions are typically represented pictorially by means of consecutive frames — 'x then y', 'x so y', 'x meanwhile y', and so on (the relationship is sometimes spelt out in writing, e.g. Meanwhile... ). 

We can even identify the three different kinds of expansion, those of elaborating, extending, and enhancing. When further detail is given in a picture of a magnifying glass, this is elaboration; where two frames are joined to form a continuous picture, this suggests extension; otherwise, consecutive frames display a relation of enhancement But in all cases the representation is pictorial throughout. 

In contrast, projections involve two modes: the projecting figure is represented pictorially as the symbol source, but the projected figure is represented linguistically and this second-order reality is typically named within a 'balloon'. … The two types of relation can be shown as in Figure 3-1.

Expansion vs Projection In Sequences

Halliday & Matthiessen (1999: 106):

This differentiation is embodied in relations of sequence in the following way. Either a sequential relation expands one figure by adding another one to it, the two still remaining at the same phenomenal level; or the sequential relation projects one of the two figures onto the plane of second-older, semiotic phenomena, so that it enters the realm of metaphenomena (meanings or wordings). This is the distinction between expansion and projection. For example, 

expansion: highs will be mid-8Os to mid-90s —> but parts of Texas could reach the 100s; 

projection: the forecast predicts —> "parts of Texas could reach the 100s".

Two Orders Of Reality

Halliday & Matthiessen (1999: 106):

Instead of referring to formal logic, let us ask what kind of construal of experience is embodied in semantic sequences. 

Throughout the semantic construal of human experience, there is a differentiation between two orders of reality: between the everyday reality of our material existence on the one hand and on the other hand the second-order reality that is brought into existence only by the system of language. This is a contrast between semiotic phenomena, those of meanings and wordings, and the first-order phenomena that constitute our material environment. (Note that the linguistic processes themselves, as apprehended by our senses, are part of the first-order reality; second-order reality is formed of the meanings and the wordings that these processes bring into being.)

 

Blogger Comments:

To be clear, if meaning is immanent, then both first-order phenomena/reality and second-order phenomena/reality are construals of experience as meaning.

Characterising Sequences By Reference To Propositional Logic?

Halliday & Matthiessen (1999: 105-6):

In a restricted register, such as those of weather forecasts and recipes, it may be possible to define sequential relations more precisely; but that is only because these registers are special cases, rather in the same way that the applications of propositional logic are special cases.

One might try to characterise sequences by reference to propositional logic; this would give some indication of what type of knowledge representation system is embodied in this domain of ideational semantics. However, such a characterisation would be likely to distort our understanding of sequences, because sequences cover and organise a considerably larger domain of the ideational semantic space than propositional logic attempts to do.

Why Natural Logic Differs From Propositional Logic

Halliday & Matthiessen (1999: 105):

Both the difference in scope and the difference in 'definability' can be explained in functional terms. 

Sequences have evolved in the interpretation of human experience in general; consequently, they have to be flexible and powerful enough to cope with a large amount of variation, and the implicit 'definition' of each relation (i.e., its location in the semantic system along various dimensions) is the evolving distillation of innumerable instances where it is invoked (for a revealing account of how sequences may construe rationality in everyday talk, see Hasan, 1992). 

In contrast, the truth-functional connectives of propositional logic have been designed for a very restricted purpose — the kind of deductive reasoning western philosophy came to focus on — and their definitions are fixed by reference to values of "true" and "false" (by means of truth tables).

Types Of Natural vs Propositional Logic

Halliday & Matthiessen (1999: 104-5, 105n):

There is another important difference between propositional logic and the natural logic of sequences. While there is only a very small handful of truth-functional connectives in propositional logic (conjunction, disjunction [exclusive or non-exclusive], implication), there is a very wide range of sequential relations in language — all the more specific varieties of projecting and expanding. A summary of these is given in Table 3(1).

² Projecting figures of the 'think' type — 'know', 'believe', 'want — have been represented outside standard logic within intensional logic.

Natural Logic vs Propositional Logic

Halliday & Matthiessen (1999: 104, 104n):

Sequences might be said to constitute the 'natural logic' equivalent of propositional logic — that is, the evolved system for reasoning about relations of cause, conditionality, etc. from which propositional logic has been derived by design. Thus we have parallel series such as:

But, as has often been pointed out, the two are not translation equivalents; for example, material implication (p –> q) applies even when the rendering in ordinary language seems odd, and disjunction in logic is either inclusive or exclusive whereas natural disjunction is non-committal. Since propositional logic is a designed system, its relations are codified and defined (typically in truth-functional terms¹). In contrast, sequential relations have evolved. A certain type of relation will have a core — the prototypical representatives of that type; but there will also be more peripheral representatives and 'grey areas' where one type shades into another.

¹ Propositional logic is interpersonally invariable. Unlike natural logic, it is only concerned with statements, or rather — since language is concerned with validity rather than truth — with the philosophical version of what are statements in natural language.

Sequence [Defined]

Halliday & Matthiessen (1999: 104):

A sequence constitutes a model of how figures can be related. One prominent form of this relationship, which has been foregrounded in various guises in science and logic, is that of cause & effect, whereby experience is given a causal interpretation. But that is only one among many such possible relationships, which taken together can be said to constitute the logic of natural language.

Reference As A Means Of Differentiating Macrophenomena And Metaphenomena

Halliday & Matthiessen (1999: 102-3):

These special subcategories have the effect of construing other elements as "referable" — that is, of enabling them to retain their semiotic identity for subsequent access and hence as it were authenticating them (e.g. Don't give me any more of that peanut butter! I can't stand the stuff.). This applies primarily, though not exclusively, to participants (cf. Webber, 1987, on reference to phenomena other than things). At the same time it allows us to recognise not only participants of the 'simple' type but also the 'larger' elements known as macro- and meta-phenomena. Macro-phenomena are figures downranked to function as ordinary elements; meta-phenomena are figures projected as elements of a second order. Halliday & Hasan (1976), where reference to macro-phenomena is called extended reference, and reference to meta-phenomena is called reference to facts, cite ambiguities which bring out the difference among elements of these various kinds:

(i) extended reference — to macro-phenomenon:

They broke a Chinese vase.

(i) That was valuable. (phenomenon: thing — the vase)

(ii) That was careless, (macro-phenomenon— the act of their breaking of the vase)

(ii) reference to 'fact' — to meta-phenomenon

It rained day and night for two weeks. The basement flooded and everything was under water.

(i) It spoilt our calculations. (meta-phenomenon (fact)—the fact that it rained so much upset our predictions)

(ii) It spoilt our calculations. (macro-phenomenon —the act of raining destroyed physical records)

Note how this relates to the shift of perspective from figure to participant: a figure such as 'catch + mouse' becomes referable as a macro-phenomenon (Your cat's caught a mouse — It's never done that before), and by the same token can enter into participant-based construal through the system of modification (Table 2(11)): (cats are) creatures that catch mice.

Grammatical Evidence For The Differentiation Of Participants And Circumstances

Halliday & Matthiessen (1999: 99, 102):

The view from grammar brings into relief a number of relevant factors relating to participants and circumstances. There are certain special subcategories of these, in the grammar, which are distinguished by the fact that they embody features of interpersonal or textual meaning:

(1) interpersonal: questioning

interrogative: who, what, when, where, how far, how long, how, why

(2) textual: cohesive

(i) referring, personal: he, she, it, they; demonstrative: this, that, now, then,, here, there, thus

(ii) generalising: lexical items such as person, creature, thing, stuff, affair

These are significant in their own right because they are critical to the construction of discourse: the textual ones provide internal cohesion, while the interpersonal ones construe dialogic speech roles. They have a further significance in that they reveal by reactance the major subclasses within the general classes of participant and circumstance, as shown in Table 2(12). 

The Three Types Of Element

Halliday & Matthiessen (1999: 99, 101):

There are three types of semantic elements serving in figures — processes, participants and circumstances. Broadly speaking, these are constructed in the grammar as follows.

(i) Within the ideational component: they are realised by different classes of units:

process ➘ verbal group

participant ➘ nominal group

circumstance ➘ adverbial group, prepositional phrase

Participants tend to be inherent elements of a figure; circumstances are typically optional.

(ii) Relating to the interpersonal metafunction: participants can serve as Subject; circumstances and processes cannot. Furthermore, participants and circumstances can serve as WH-elements, but processes cannot (if the process is being questioned, a participant element has to be construed as a Range: what...do?)

(iii) Relating to the textual metafunction: participants and circumstances can both readily serve as Theme (though their potentials differ); processes only rarely, other than in imperative clauses. Participants and (more restrictedly) circumstances can serve as referables identified by referring expressions, but processes cannot — as with WH-interrogation, they have to be construed as a Range together with the pro-verb do as Process: do it/that (see Halliday & Hasan, 1976: 125).

These properties are summarised in Table 2(11).

The Four Types Of Figure

Halliday & Matthiessen (1999: 98, 100):

There are four types of figure — doing & happening, sensing, saying, being & having. Broadly speaking, these are constructed in the grammar as follows:

(i) Within the ideational metafunction, each is realised congruently by one particular transitivity type: doing & happening ➘ material, sensing ➘ mental, saying ➘ verbal, and being & having ➘ relational. These have various reactances, such as the number and nature of participants and the unmarked present tense selection (see Table 2(10)).

(ii) Relating to the interpersonal metafunction: in any given register there may be typical correspondences between the type of figure and the speech function; e.g. in procedural registers, material clauses are typically imperative, relational ones indicative.

(iii) Relating to the textual metafunction: different types of figure are presumed in different ways and have different potential for textual prominence; e.g. only material clauses are substituted by the pro-verb do (to/with).

Blogger Comments:

With regard to Table 2(10), the Medium of an identifying clause is the Identified, which coincides with the Value only when the relation is encoding. In decoding clauses, it is the Token that is the Medium/Identified.

Semantic Categories Are Construed By Means Of Realisation In The Grammar

Halliday & Matthiessen (1999: 97-8):

The semantic categories are themselves construed by means of realisation; they are constructed within the grammar and the lexis of a language. If we model the ideation base as a semantic space, we are foregrounding one aspect of the construction of meaning in language, namely the way in which lexicogrammar construes our experience of the world in the guise of multidimensional matrices or grids. This is an important feature of language as a semiotic system, an inevitable consequence of the principle of arbitrariness: since the forms of expression are arbitrary, they impose discontinuity on the content. We have to decide about any given instance whether it is singular or plural, temporal or causal, possible, probable or certain; whether it is a bus or a van, smiling or grinning, cloudy, misty or foggy. But the semantic categories themselves (seen from above, as it were) are much more fluid and indeterminate than their realisations in wording imply. The notion of semantic space allows us to adopt a complementary standpoint from which we can view these phenomena topologically, bringing out the inherently elastic quality of the dimensions involved, and gaining a deeper insight into the semogenic processes by which the meaning potential is ongoingly remoulded in the history of the system, of the user, and of the text.

The Ontogenetic Move From Concrete To Abstract Phenomena

Halliday & Matthiessen (1999: 97):

Children are able to construe the semantic system because they start in situations with a material setting that is shared — cf. above, and see Hasan (1985b: 25 ff.) on Malinowski's contribution to this insight. As they build up their ideation base, they can begin to construe categories internal to the system out of existing ideational values; and they can begin to move into the abstract domains of a purely symbolic world, where the significations of semantic categories are abstract categories in social and socio-semiotic systems. But precisely because these abstract categories are construed as meanings, they can still be built up, negotiated and validated in collaboration with other members of the meaning group. The move from the realm of concrete phenomena to the various realms of abstract phenomena is made possible through the homogenising power of meaning.

Extra-Linguistic Categories Are Construed As Semantic Categories

Halliday & Matthiessen (1999: 96-7):

We saw in the examples of how categories are first developed that young children will typically construe concrete phenomena that are part of the field of visual perception they share with their interactants. In other words, they are construing into linguistic meanings their experience of the material world as it is construed in the categories from another semiotic system, viz. (visual) perception. These extra-linguistic categories are construed as the signification of the semantic categories of the ideation base — always in some particular situation when the child first engages with them. To construe experience of concrete phenomena as meaning is thus to construe some signification which lies outside the ideation base as a value which is internal to the ideation base system. Part of the power of categorisation is that extra-linguistic phenomena that are quite varied in signification can be construed as alike in value.

Blogger Comments:

[1] To be clear, the 'material world' is meaning construed by semiotic systems.

[2] To be clear, concrete phenomena are meaning, in this instance: meanings of perceptual systems.

[3] To be clear, on Edelman's model of brain function, relations between perceptual meanings become organised into systems that he (unfortunately) terms 'conceptual'. That is, some of the work of relating perceptions is already done before language reconstrues them.

Stratal Angles On Construal

Halliday & Matthiessen (1999: 96, 97):

A phenomenon of experience is construed as a category in the ideation base by being given a location in the three semantic networks described in the preceding section. This taxonomic, meronymic and eco-functional location is the category's value (valeur) in relation to the other categories within the ideation base. By being assigned a value internal to the ideational system, a semantic category is also being related to categories that lie beyond semantics itself. On the one hand it is being related to categories within systems that lie outside language but which the semantic system interfaces with; on the other hand it is being related to the grammatical categories in terms of which it is realised. Figure 2-18 shows these three strata! angles on the process of construal.

Blogger Comments:

Importantly, these are categories of perceptual semiotic systems. SFL Theory makes the assumption of immanence: all meaning is within semiotic systems.

A General Semogenic Strategy

Halliday & Matthiessen (1999: 95):

Construing a category thus includes locating it not only taxonomically and meronymically but also eco-functionally, as in the case of 'conscious being' just referred to. We noted above in connection with the cat example how the fact that a participant serves a participant role in a figure can be viewed either from the perspective of the figure or from the perspective of the participant (i.e. as a property of the participant). If cats kill mice, the ideation base accommodates the view from the angle of the figure: 'cats kill mice'; but it also accommodates the view from the angle of the participant: 'animals that kill mice'. Here the figure in which cats participate as actors has been construed as if it was a property, so that the category of cats might be construed within the system of the ideation base by means of the definition 'cats are animals that kill mice'. (In formal systems of representation, this can be expressed by means of lambda abstraction.)

There is a general semogenic strategy at work here. We referred to 'sequences', 'figures' and 'elements' as three orders of complexity in Section 2.1. The shift in perspective means that configurations of meanings that are of a particular order of complexity can be accessed through selection not only in their normal environment (within phenomena of the next higher order of complexity) but also within phenomena of lower orders of complexity. Through this semogenic strategy of opening up the possible domains of selection, a great deal of experiential complexity can be imported into the construal of a participant.

The Network Of Selection Is Inter-Axial

Halliday & Matthiessen (1999: 94-5, 96):

The network of selection is inter-axial in that it relates syntagmatic specifications to paradigmatic ones; it 'cuts across' the paradigmatic organisation of the ideation base, establishing correspondences between paradigmatic types through the syntagmatic functions associated with them, as we have just illustrated by reference to Figures 1-12 and 2-6. We can trace one further example from sequence via figure to element: see Figure 2-17.

Selections Are Bidirectional

Halliday & Matthiessen (1999: 94, 44):

Selections are bidirectional. We have already seen that functions associated with paradigmatic types constitute 'structural properties' by which these types are distinguished and that selections provide further information about these 'properties'.

Thus Figure 1-12 above illustrates how a figure of spearing is characterised by reference to the function of Means, which selects for the thing 'spear'. Conversely, semantic types which are selected are characterised by the functional environment within which they are selected. Thus the weapon 'spear' is characterised by its potential for serving as Means in a figure of spearing. Similarly, part of the meaning of a 'conscious being' is that it has the potential for serving in the Senser role of a figure of sensing.

Eco-functional Selection

Halliday & Matthiessen (1999: 94, 58):

Taxonomic elaboration and meronymic extension form complementary networks of relations that together make up the paradigmatic organisation of the ideation base. They define, as we have noted, options in delicacy of categorisation and in delicacy of focus. The third type of network involved in the organisation of the ideation base which we identified above serves to relate paradigmatic organisation and syntagmatic organisation. Specifically, it relates syntagmatic functions or roles associated with paradigmatic types to the paradigmatic types that can serve in these functions. For example, the syntagmatic function 'Senser' is associated with the paradigmatic type 'sensing', and it is related to the paradigmatic type 'conscious being', since only participants of this subtype can serve as Sensers.

Figure 2-6 above shows this relation, together with a set of other such relations. We shall refer to such network relations as eco-functional selection in order to indicate that they specify the syntagmatic environment of semantic types by showing them as selections for syntagmatic functions. Such selections have been referred to as "pre-selections", but in order to avoid any connotations of temporal sequence, we prefer the term "selection" for such relations in the ideation base.

Delicacy Of Categorisation (Hyponymy) vs Delicacy Of Focus (Meronymy)

Halliday & Matthiessen (1999: 92):

While the choice of 'level' in hyponymic elaboration is the choice in delicacy of categorisation, the choice of level in a meronymic taxonomy is the choice in delicacy of focus. The focus is typically on the whole (i.e., the most inclusive region within the meronymy) even if a specific part is particularly important (cf. Langacker, 1984).

Process Meronymies?

Halliday & Matthiessen (1999: 91):

The examples of meronymies given above come from the region of 'participant' within the ideation base. It is not clear to what extent, or in what sense, meronymic taxonomies extend beyond that region. We can certainly recognise that processes have phases — 'begin to do, keep doing, stop doing'; but it is not immediately clear that these form a process meronymy analogous to the parts of a participant. Although we might reconstrue he began to dance metaphorically as the beginning of his dance on the model of 'the beginning of the book', this is a metaphorical reification of the process 'dance' and we have to be cautious in interpreting the implications for the congruent process 'dance'. If we probe a little further, we can see that process phase is concerned with the occurrence of a process in time — its temporal unfolding: 'begin to do' means 'begin to be actualised (to occur) as doing in time'. In contrast, participant meronymy is not tied to the existence of a participant in referential space.

Why Meronymic Taxonomies Are Localised Within Hyponymic Taxonomies In SFL

Halliday & Matthiessen (1999: 90-1):

And this is in fact how we shall model them — as local meronymic taxonomies, applicable to some particular region of the hyponymic taxonomy, and not as one global meronymy superimposed on our taxonomy. There would seem to be far too many discontinuities to create a global meronymy; meronymies tend to occur only where there are contiguous parts of an independent whole. Thus while concrete objects are regularly construed meronymically (with the human body as both a representative example and a model for other meronymies), substances are not; substances are extended through measure ('unit of') rather than through part ('part of'). Similarly, taxonomies tend to be more extended meronymically for concrete regions than for abstract ones (although even things in an abstract region can have parts, e.g. aspect of an idea). Thus there is a generalised set of categories such as part, element, component, aspect; and also a generalised set 'facets' of spatial or temporal orientation, top, bottom, side, front, back, middle, centre; beginning, middle, end. There are more specific variants for parts of particular concrete things, such as facade, roof, wall [of a building]; ceiling, floor, wall [of a room]; slope, peak [of a mountain]; limb, trunk, root, bark [of a tree]; skin, core, pips [of an apple]; crust, crumb [of bread]; preface, epilogue [of a book].

Meronymic Extension (vs Hyponymic Elaboration)

Halliday & Matthiessen (1999: 89, 90):

The ideation base construes phenomena as organic wholes that may take on roles in other kinds of phenomena; but it also deconstrues many such organic wholes into their component parts. When these component parts are phenomena of the same type — participants (e.g., chair: legs, seat, back), figures (e.g., baking a cake: stages in the procedure), this is known as meronymy (or meronymic taxonomy; cf. meros 'part'). We find local taxonomies of this kind, often interlocked with hyponymies; for example, see Figure 2-16. Taxonomies thus embody the two types of expanding relationships we mentioned at the beginning of this section — extension and elaboration. The meronymic type of taxonomy is extension, whereas the hyponymic type is elaboration.

Degrees Of Delicacy In The Lexical Zone

Halliday & Matthiessen (1999: 89, 90):

Within the lexical zone, we have already referred to the differentiation according to the degrees in delicacy; in particular, the 'basic' degree of delicacy stood out as most highly elaborated. There are also grammatical reactances indicating the taxonomic differentiation between the "basic" degree of delicacy and lower degrees of delicacy, at least in cases where the taxonomic relationship is of a particular kind. For example, the less delicate category may be construed by a mass noun whereas the more delicate, basic degree categories may be construed as count nouns: see Table 2(8).

Wierzbicka (1985: 321-2) identifies this phenomenon and shows that it is not arbitrary. Her explanation is in fact that the relationship between e.g. 'furniture' and 'chair, table' as super-category and subcategory is not the normal 'kind of relationship between e.g. 'bird' and 'swallow, magpie', but rather a grouping of different kinds according to similarity in use:

Thus, supercategones such as bird or tree are 'taxonomic', i.e. they belong to hierarchies of kinds (where each 'kind' is identified on the basis of similarity between its members); supercategories such as crockery, cutlery or kitchenware are not taxonomic — they include things of different kinds, grouped on the basis of contiguity and/ or similarity of function, not on the basis of similarity of form.

We take delicacy to include both types of the relationship of supercategory to subcategory — both the "taxonomic" and the "non-taxonomic" ones. However, it would seem that with the relationship between 'furniture' and 'chair', 'table' &c. we are on the borderline between elaboration and extension.

Semantic Types Are Classified By Grammatically-Construed Criteria And Structural Roles

Halliday & Matthiessen (1999: 87-8):

Semantic types in different taxonomic regions are distinguished according to different criteria; and they have different sets of roles. But the different criteria and sets of roles are construed within the grammar. …

(i) Phenomena within different taxonomic regions are classified according to different kinds of grammatically construed criteria. For example:

(a) perception (figures: subtype of sensing): according to means of perceiving: see (means: eyes)/ hear (means: ears)/ smell (means: nose)/...

(b) transformation (figures: subtype of doing): according to result, e.g. break (result: into pieces)/ melt (result: into liquid state)/ shrink (result: smaller)/ pulverise (result: powder, dust)/...

(c) motion (figures: subtype of doing): according to manner, location, purpose, e.g. fall (location: downwards)/ stroll (purpose: for pleasure)/ flee (purpose: to escape)/ walk (manner: with legs, fairly slowly)/...

(d) higher animals (elements: subtype of conscious thing, at the taxonomic depth of species): according to epithets of age and sex, e.g. cow (sex: female, age: adult), bull (sex: male, age: adult), calf (age: nonadult).

(e) artefacts (elements: a subtype of object): according to material, purpose, e.g. containers (purpose: to contain substances) — barrel (material: wood), basket (material: cane or other woven material), basin (material: metal), bowl (material: earthenware or glass).

(ii) Phenomena within different domains have different grammatically construed structural roles associated with them. For example:

(a) perception: perceiver & phenomenon being perceived;

(b) concrete thing (elements: subtype of object): various epithets, specifically of physical dimensions such as size, shape, weight, colour and age.

(c) weight (elements: subtype of quality): tensor, showing degree of intensity.

The Construal Of Delicacy In Ideational Semantics

Halliday & Matthiessen (1999: 87, 88):

In general, then, we can say that the move in delicacy in the ideation base from 'most general' to 'most delicate' is construed lexicogrammatically as the move from 'grammar' to 'lexis': see Figure 2-15. 

This is of fundamental significance in the construal of semantic categories. The early part of the scale of delicacy is construed in the grammatical 'zone'. This zone provides the resources of grammatical schematisation for construing more delicate categories: those categories are realised lexically but construed according to the systemic parameters of the grammar. For example, the grammar of the nominal group provides a schema for construing various delicate categories of things, by classifying, describing, ordering and other such strategies.

Lexically- vs Grammatically-Construed Taxonomies

Halliday & Matthiessen (1999: 86):

So far, our discussion of taxonomic elaboration has focussed on those steps in delicacy that tend to be construed lexically in the lexicogrammar. But in the overall meaning base, lexically construed folk and scientific taxonomies do not start at the highest degree of generality in delicacy; they are ordered in delicacy after those systems that are construed grammatically. For example, while 'plant' is the "unique beginner" of a particular folk taxonomy, there are several steps in the construal of things that are more general than this — steps that are constructed in the grammar, in categories such as 'thing', 'conscious/ non-conscious thing', 'countable/ non-countable thing' and the like.

The Move From Folk To Scientific Taxonomies

Halliday & Matthiessen (1999: 85-6, 7):

The ideation base embodies not only folk taxonomies but also a range of taxonomic models such as those used by experts and by scientists. These are all variants within the overall ideational system. …

The move from folk taxonomies towards scientific ones involves both an increase in steps in delicacy and a change in the criteria used for classification. Wignell, Martin & Eggins (1990) give examples from the classification of roses and of birds of prey. … For birds of prey, they contrast a folk taxonomy with a "birdwatchers' vernacular taxonomy". We superimpose these to give a sense of the difference; see Figure 2-14. 

These examples show the difference in degree of delicacy quite clearly; an increase in delicacy reflects the move in the direction of scientific knowledge. At the same time, the criteria for categorisation also change, from overt criteria that are accessible to the naked eye to covert criteria available only through the application of scientific techniques. The change in taxonomic criteria is very clear from a review of the early taxonomic stage of science in Europe in the 16th and 17th centuries: see Slaughter (1986).

The Most Favoured Degree In Delicacy

Halliday & Matthiessen (1999: 84):

The most favoured 'band' in delicacy is that of the basic level: it is taxonomically most highly elaborated, it tends to be learnt first by children, and it construes categories with syndromes of usually salient functional and perceptual properties. … In addition, the "basic' degree in delicacy may have a special status in the instantiation of categories in text.In her study of lexicalisation in the pear stories, Downing (1980) found that the basic level was by far the most preferred one; these are the figures for concrete, nonhuman things (p. 106):

Such findings are interesting because they give some indication of what factors to consider in lexicalisation in text generation: a given phenomenon can, in principle, be construed at any point along the scale of delicacy. They are also interesting because they point to the relationship between the system and the instance: from the perspective of the system, we can observe that the basic degree of delicacy is most highly elaborated; from the perspective of the instance, we can observe that it is most frequent. However, we have to allow for considerable variation within a language — in particular, functional or registerial variation within the system according to the context of use, as in the variation between the spoken system of everyday life and the written system of science to which we turn next.

Berlin On The Elaboration Of Labelled Taxonomies

Halliday & Matthiessen (1999: 83-4):

Berlin (1972: 53) characterises the elaboration of labelled taxonomies as follows:

Generic names are fundamental and will occur first. These will be followed by the major life-form names and specific names. At yet a later period, intermediate taxa and varietal taxa will be labelled. Finally the last category to be lexically designated ... of any ethnobotanical lexicon will be the unique beginner. The suggested sequence can be seen diagrammatically as follows:

... no temporal ordering is implied for some categories. Thus no claim is made as to the priority in time of specific names over major life form names. On the other band a claim is made that a language must have encoded at least one major life-form name and one specific name before the appearance of intermediate and varietal named taxa.

The Nature And Degree Of Taxonomic Elaboration

Halliday & Matthiessen (1999: 83, 85):

Delicacy is a uniform ordering from most general to most delicate; but along this scale, semantic systems differ both in the number of distinctions at any one degree of delicacy and in the overall delicacy that is achieved. There are specific differences associated with particular taxonomic regions. For instance, humans and higher animals are much more highly elaborated than lower animals. But there are also general differences in the nature and degree of taxonomic elaboration associated with different 'bands' in delicacy. Such general taxonomic principles are probably best known for folk-taxonomies in the domains of plants and animals, diseases, and the like (e.g., Berlin, Breedlove & Raven, 1973; Conklin, 1962; Frake, 1962; Slaughter, 1986); these fall within 'element: participant: thing' in the ideation base. The maximum steps in delicacy in a folk taxonomy are kingdom (unique beginner), life form, basic (generic) level, specific level, and varietal level. These steps are by no means always present in every particular taxonomy; and they have different characteristics, summarised in Table 2(6) (for taxonomic examples, see Leech, 1974).

Taxonomic Elaboration

Halliday & Matthiessen (1999: 83):

Any option (category, semantic type) is thus located in the network relative to other options, first within its own system, and secondly in terms of the location of that system relative to others. For example, 'participant' is one option in the system 'participant/process/circumstance'; and that system is a more delicate elaboration of 'element'. We noted above in reference to Painter's work that the resources of figures of being are deployed in construing taxonomic relations. There is also a generalised set of nominal categories for construing steps in the delicacy hierarchy — e.g. type, kind, class; and some taxonomic regions have specific categories of their own, for example: brand, model, make, issue; genus, species, family.

Construing As Locating In A Network Of 3 Types Of Relations

Halliday & Matthiessen (1999: 82-3):

From the typological point of view, construing experience in terms of categories means locating them somewhere in this network of relations. When we examine this network more closely, we find that there are actually three types of network involved: 

(1) taxonomic in the strict sense (i.e. based on hyponymy, 'a is a kind of x/ x subsumes a, b, c'), 

(2) taxonomic in the extended sense (i.e. based on meronymy, 'd is a part of y/ y has parts d, e, f'), and 

(3) eco-functional (i.e. based on selection, 'g has function m in environment z/ environment z comprises functions m, n, p, and function m may be filled by g, h, j'). 

Of these three, the first provides the global organising principle of elaboration in delicacy; while the third relates paradigmatic organisation to organisation on the syntagmatic axis.

Summary Of The Ontogenetic Perspective On Construing Experience

Halliday & Matthiessen (1999: 81-2):

We can now sum up what can be learnt about construal from the ontogenetic perspective:

(i) Initially the child construes experience of phenomena that are in, or are brought into, a shared visual field; once constituted into meaning, the experience can be shared, validated and scaffolded dialogically in collaboration with other members of his/her meaning group.

(ii) The earlier experiences are clearly situated by virtue of the child's perception. But once the process of construal has been established, experience can be generalised in the form of semantic classes, and classes of classes; it can be further explored in terms of the semantic system itself; it can be developed vicariously through discourse, and extended to include purely abstract categories.

(iii) Categories are construed in the network of the ideation base in terms of different kinds of relations they enter into:

[1] globally, they are construed in terms of taxonomic elaboration: they form part of a system that is located somewhere in delicacy within the meaning base; 

[2] locally, they may be construed in terms of meronymic extension: they may form part of a local meronymic taxonomy, such as the cat and its body parts; 

[3] transcategorially, they may be construed in terms of the roles they play in some other type of category, such as the cat serving in certain participant roles of figures.

(iv) Categories are thus located within the ideational network. The network defines a multidimensional, elastic space; and locations within this space are not fixed, clearly bounded regions but rather regions with core or focal areas and more peripheral areas that shade into one another.

(v) The ideation base is a resource for construing extra-linguistic experience (such as visual experience). But it is also a semogenic resource for construing itself, since it is built up out of the kinds of relations it itself construes — relations such as intensive ascription. Once critical semantic mass has been built up, new categories can be construed internally within the system of the ideation base.

SFL Theory vs Semantic Primitives

Halliday & Matthiessen (1999: 76n):

² The ideational potential for construing such relations of restatement within the ideation base does not at all imply that meaning is analysed by means of decomposition into semantic primitives, 

(i) Semantic types (such as 'balance', 'cat') are not decomposed into their defining glosses within the ideation base; rather, they are construed as standing in an intensive token-value relationship to these glosses, 

(ii) The semantic types are construed internally to the ideation base according to their location in the elaborating taxonomy and can be restated in various ways, 

(iii) The semantic types may also construe extra-linguistic categories of experience; that is, they may have signification outside the ideation base.

Ontogenesis: Construing A Lexicalised Token As A Grammaticalised Value

Halliday & Matthiessen (1999: 76-7):

Stephen's definition of 'balance' is on a developmental path towards the kinds of definition used in educational and scientific discourse — definitions such as the dictionary definition of 'cat': a cat is a carnivorous mammal long domesticated and kept by man as a pet or for catching rats and mice… . The definition of 'cat' involves a move to a less delicate category ('carnivore' or 'carnivorous mammal', which can be reconstrued as 'mammal that eats meat' …) plus a qualification by a downranked sequence of figures of doing: see Figure 2-12.

In both examples, the definition construes a token-value relation between a fairly delicate semantic type that is lexicalised within the lexicogrammar and a restatement of this type by means of other resources in the ideation base. The restatement draws more on the resources towards the grammatical end of the scale, so that in the definition a lexicalised token is construed as a grammaticalised value.

This entails a shift from construing experience through depth in the experiential taxonomy towards construing experience through expansion in a logical sequence. In the first example, this only involves a downranked sequence of figures; in the second example, this involves a move towards a less delicate category plus a downranked sequence of figures. The first example can actually be interpreted as meaning 'the condition when you hold it on your fingers and it doesn't fall off', showing that the downranked sequence has the class meaning of 'condition'.

Ontogenesis: Construing New Meanings In Terms Of The Semantic System Itself

Halliday & Matthiessen (1999: 75-6):

Also in his fourth year, Stephen begins to extend his deployment of figures of being to include not only the ascriptive mode ('a is a member of x') but also the identifying mode ('y equals x') so that he can explicitly relate meanings in his ideation base in the form of definitions.

For example, in Balance means [[you hold it on your fingers and it doesn't go]], he construes a single class of abstract participant, 'balance', as something defined in terms of a sequence of figures, 'you hold it on your fingers and it doesn't fall off: see Figure 2-11. 

 

He is now in a position to construe new meanings in terms of the semantic system itself; that is, they do not necessarily have to be 'imported' from direct extra-linguistic experience. As Painter points out, this is yet another expansion of his semantic resources preparing him for educational learning in later life where experience is very often vicarious experience, constructed entirely in educational discourse.

Ontogenesis: Providing Causal Evidence For Categorisation

Halliday & Matthiessen (1999: 75):

In his fourth year, Stephen begins to provide causal evidence for categorisation, by adding an enhancing figure. The figure that is linked causally specifies attributes that are critical to the construal. For example: Mother refers to Bond airship as 'spaceship balloon' — Stephen: Not a spaceship — an airship — cause a spaceship has bits like this to stand it up; Stephen (pointing at page numbers): That's fifteen because it's got a five; that's fourteen because it's got a four. 

These causal relations are internal (Halliday & Hasan, 1976), i.e. oriented towards the interpersonal act of communication itself ('I know it is an x, because it has feature y') rather than external ('it is an x, because it has feature y'). 

In other words, Stephen is attending to the act of construal itself — 'I construe it as x because of y'. At this stage in the development of Stephen's construal of experience, the construal has become something that is not only shared, but can also be explicitly negotiated and argued about.