Misrepresenting 'Field' And Mistaking Range For Medium And Medium For Agent

Martin & Rose (2007: 92):

In the grounds that Tutu gives for his second Argument, he names its field as the granting of amnesty. This field is expanded in the following clauses as processes of ‘giving’, ‘not giving’, ‘refusing’ and ‘applying for’ (in italics below), of which amnesty is the Medium (in bold), with various Agents and Beneficiaries (underlined):

It is also not true that THE GRANTING OF AMNESTY encourages impunity ...
because amnesty is only given to those who plead guilty ...
Amnesty is not given to innocent people or to those who claim to be innocent.
It was on precisely this point that amnesty was refused to the police officers
who applied for [amnesty] for their part in the death of Steve Biko.

Blogger Comments:

[1] To be clear, the ideational meaning instantiated in Tutu's text is semantics, not context.  In SFL theory, 'field' refers to the ideational dimension of the context that is realised by language — along the cline from the culture that is realised by language as system to the situation that is realised by an instance of language (a text).  For Martin's misunderstandings of the SFL notion of field, see the clarifying critiques here.

[2] This is potentially misleading, since field (context) and clause (lexicogrammar) are different levels of symbolic abstraction. To be clear, taking metaredundancy into account, context (e.g. field) is realised in the realisation of semantics in lexicogrammar (e.g. clauses).

[3] To be clear, amnesty serves as the Range of these four Processes, not the Medium.  The processes of ‘giving’, ‘not giving’, ‘refusing’ and ‘applying for’ are not actualised by amnesty, but by the Commission, which is omitted from these medio-receptive clauses.

[4] To be clear, no Agents are underlined.  Of the underlined, the only participant that is not a Beneficiary, who, serves as the Medium through which the Process applied (for) is actualised — in a rankshifted clause serving as the Qualifier of a nominal group.

because	amnesty	is	only	given	to those who plead guilty
	Range	Pro-		-cess	Beneficiary

amnesty	is not given	to innocent people or to those who claim to be innocent
Range	Process	Beneficiary

[[amnesty	was refused	to the police officers [[who applied for (amnesty) for their part in the death of Steve Biko]] ]]
Range	Process	Beneficiary

[[who	applied for	(amnesty)	for their part in the death of Steve Biko]]
Medium	Process	Range	Cause

Misunderstanding The Basis Of Nuclearity

Martin & Rose (2007: 92):

The Medium may be affected by the process, but the Agent is left implicit, as in I’m going to be haunted, amnesty was refused. As Agent and Beneficiary may be left out of the clause, they are relatively marginal in terms of nuclear relations. 

How do these grammatical functions interact with the lexical elements that instantiate them in particular texts?

Blogger Comments:

[1] To be clear, on the ergative model, the Medium is participant through which the Process is actualised; on the transitive model, in contrast, the Goal is the participant that is impacted by the Process.

[2] To be clear, the clause amnesty was refused is middle (medio-receptive), not effective.  That is, the clause lacks the feature [agency], so it does not feature an Agent, implicit or otherwise.

amnesty	was refused	by the Commission
Range	Process	Medium

[3] This seriously misunderstands nuclearity.  Agent and Beneficiary, as participants, are "marginal" in the sense of their participation in the Process, not because they can be left out.  The problems with 'omittability' as a criterion include:

• the omission of a participant serves a textual function, not an experiential function; and
• even the most nuclear participant, the Medium, can be omitted, as demonstrated by the medio-receptive clauses such as amnesty was refused, you were seen!  What was said?

It will be seen later that Martin & Rose contradict their own principle by interpreting 'Range: process' — e.g. a song in he sang (a song) — as central, despite the fact that it can be omitted.

[4] The notion of grammatical functions interacting with lexical items betrays the Martin's (1992) misunderstanding of the dimensions of SFL theory as the interaction of modules; see the clarifying critiques here.

To be clear, the relation between grammatical functions and lexical items involves both delicacy and realisation, since each lexical item is the synthetic realisation of the most delicate features of lexicogrammatical systems.

[5] The notion of lexical items instantiating grammatical function confuses instantiation — the relation between potential and instance — with delicacy (and realisation), as explained above in [4].

∞

Importantly, none of this is

1. discourse semantic,
2. beyond the clause, or 
3. the authors' original theorising.

Misconstruing Middle Clauses As Effective

Martin & Rose (2007: 91-2):

Some effective processes can also be extended to a third participant, known as a Beneficiary:

 

Blogger Comments:

[1] To be clear, the acceptance of a Beneficiary applies to middle clauses as well as effective clauses, as demonstrated by the three clauses that Martin & Rose present above, all of which are actually middle, not effective:

The Commission	may grant	amnesty	to those who plead guilty
Medium	Process	Range	Beneficiary

amnesty	is not granted	by the Commission	to innocent people
Range	Process	Medium	Beneficiary

the police officers	were refused	amnesty	by the Commission
Beneficiary	Process	Range	Medium

That is, the process of granting or refusing is actualised through the Commission, with amnesty being the range of their granting or refusing.  It will be seen that the authors subsequently build their rebranding of Halliday's clause nuclearity on such misunderstandings.

[2] To be clear, it is the clause that is effective (+agency) or middle (–agency), not the Process.

[3] To be clear, in SFL theory, the variable of extension applies to the transitive model, not the ergative model of nuclearity, where the variable is instead one of causation (Halliday & Matthiessen 2014: 339-40). Moreover, the relation of the Beneficiary is to the clause Nucleus (Process/Medium), and the relation is one of enhancement (cause), not extension; see Halliday & Matthiessen (2014: 348).

Agent And Voice

Martin & Rose (2007: 91):

In addition to the Medium, one or two other participants may be involved in the process, including Agent, Beneficiary and various types of Range. An Agent instigates the process, which affects the Medium in some way:

These effective clauses can be reversed in passive form, with the Agent as a ‘by phrase’:

Blogger Comments:

[1] Trivially, Agent, Beneficiary and Range are three types of participant, but less trivially, all three may be involved in a single Process, as in:

the minister	made	Bach	play	organ	for the choir
Agent	Pro-	Medium	-cess	Range	Beneficiary

[2] To be clear, Agent is the external cause of a Process, construed as a participant (Halliday & Matthiessen 2014: 336).  For example:

meaning	is realised	by wording
Medium Value	Process	Agent Token

Having here identified the Agent as the participant that instigates a Process, the authors later (p94) present the Medium as instigating a Process.

[3] To be clear, an Agent causes the actualisation of a Process through a Medium.

[4] Trivially, in SFL theory, this is termed 'receptive' at clause rank, with passive' being reserved for the verbal group.  Non-trivially, the choice of voice serves a textual function, not an experiential function; see Halliday & Matthiessen (2014: 348). That is, the two variants construe the same experience, but differ in which experiential function is made Theme and (unmarked) New.

Importantly, the reason Martin & Rose introduce this spurious variable here, is that they are preparing to misinterpret Halliday's ergative model of clause nuclearity in terms of what can and cannot be omitted for textual reasons.  That is, as will be seen, their model of nuclear relations, as experiential discourse semantics, is based on a confusion of the experiential metafunction with the textual.

Medium

Martin & Rose (2007: 91):

Nuclear relations within the clause

To set the scene for exploring nuclear relations, we first need to discuss a few of the semantic patterns within the clause described by Halliday (1994/2004). The essential experiential pattern is that people and things participate in a process. In Halliday’s terms the core participant in the process is known as its Medium, ‘without which there would be no process’. Here are some familiar examples:

Blogger Comments:

[1] To be clear, Martin's (1992) model of 'nuclear relations' is Halliday's ergative model of clause transitivity, misunderstood, and relocated from lexicogrammar to discourse semantics.  See the clarifying critiques here.

[2] To be clear, the subtitle of this publication is Meaning Beyond The Clause.  Martin's model of nuclear relations is concerned with the realisation of meaning as wording — within the clause.

[3] To be clear, 'people and things' are realised at the rank of group, not clause.

[4] To be clear, the Medium is the participant through which the Process is actualised (Halliday & Matthiessen 2014: 336).  The Medium is not obligatory in medio-receptive clauses like you were seen, they were told etc., or location-receptive clauses like this bed hasn't been slept in.

[5] To be clear, the mood Adjuncts even and never do not serve any experiential function, let alone Process.

Collocation

Martin & Rose (2007: 91):

As we flagged in the introduction to this chapter, the clause construes experience in terms of a process involving people and things, places and qualities. We have explored taxonomic relations between these elements, from one clause to the next as a text unfolds. In this section we will examine lexical relations between these elements within clauses. As they are more or less centrally involved in the process, lexical relations within the clause are known as nuclear relations.

Traditionally these kinds of lexical relations have been regarded as collocations, that is words that are commonly found together in the same structure, such as tennis-ball or play-tennis. What we will show in this section is how such collocations are dependent on the nuclear patterns of the clause, and again we will link lexical relations to the field construed in texts. The categories of nuclear relations presented in this section will then be applied to text analyses in the following section on activity sequences.

Blogger Comments:

[1] See the immediately preceding post for the misunderstandings here. 

[2] To be clear, "traditionally", collocation refers to the habitual juxtaposition of a particular word with another word or words with a frequency greater than chance.  Importantly, this is distinct from Halliday's model of collocation as a form of textual cohesion and from Halliday's model of clause ergativity in terms of nuclearity.  That is, Martin & Rose here again fail to acknowledge Halliday as their source and confuse two distinct regions of the grammar and rebrand the confusion as their model of discourse semantics.

[3] To be clear, collocation is not "dependent on the nuclear pattern of the clause" because

• collocation is not a feature of clause structure, and 
• collocation relations also obtain within in other grammatical domains, such as the nominal group, adverbial group, and prepositional phrase, and
• collocation is distinct from nuclearity, since nuclearity is the perspective on experiential meaning that is provided by the ergative model of the clause (Halliday 1985: 147).

What will be shown here is that Martin & Rose misunderstand Halliday's notion of clause nuclearity, and additionally misconstrue the expansion relations that obtain between the grammatical elements (which they misconstrue as lexical items).

[4] To be clear, in SFL theory, the function of collocation is to create texture by lexical means at the level of lexicogrammar. As such, the relation of collocation does not, in itself, build "the field construed in texts", no matter how "field" is understood.

(As previously explained, Martin & Rose misunderstand field as the ideational dimension of register, which they, in turn, misunderstand as a stratal system of context.  In terms of SFL theory, what Martin & Rose refer to as "field" here actually correlates with the ideational semantics instantiated in text.)

Nuclear Relations

Martin & Rose (2007: 90-1):

As we flagged in the introduction to this chapter, the clause construes experience in terms of a process involving people and things, places and qualities. We have explored taxonomic relations between these elements, from one clause to the next as a text unfolds. In this section we will examine lexical relations between these elements within clauses. As they are more or less centrally involved in the process, lexical relations within the clause are known as nuclear relations.

Blogger Comments:

[1] Here Martin & Rose misunderstand the SFL model of the clause. The clause construes experience as a process, participants in the process and circumstances attendant on the process. Importantly, the elements are defined relative to each other.

[2] Here Martin & Rose confuse grammatical elements with lexical items. Their model of taxonomic relations is claimed to be concerned with lexical items, not elements of grammatical structure.

[3] To be clear, this encapsulates a fundamental misunderstanding that invalidates Martin & Rose's model of nuclear relations: the mistaking of elements of grammatical structure for lexical items.

[4] This is very misleading indeed. Here Martin & Rose introduce Halliday's (1985: 147) model of ergativity without acknowledging its source, thereby once again inviting the reader to credit them with the work of others. Halliday & Matthiessen (2014: 348):

Moreover, the model is grammatical, not discourse semantic, and concerned with elements of grammatical structure, not lexical items and, as will be seen, Martin & Rose thoroughly misunderstand Halliday's model, and confuse it with Halliday's textual system of collocation, a type of lexical cohesion.

Importantly, for the semantics of clause nuclearity, see Halliday & Mathiessen (1999: 165-76) on degree of participation (of participants) and degree of involvement (of circumstances).

The Notion Of Lexical Relations "Building A Field"

Martin & Rose (2007: 90):

In contrast the discourse semantic perspective we are taking here foregrounds the ideational function of lexical relations in building a field, so our starting point is with class and part relations: a young man - my first love. Synonymy draws on common class membership to identify items with each other, with repetition as the limiting case. Contrasts then function to distinguish categories. This is a metafunctional view on discourse semantics, in which taxonomic relations complement reference relations to build the field and maintain cohesion as a text unfolds.

Blogger Comments:

[1] To be clear, to take a discourse semantic perspective on lexical relations is to take a discourse semantic perspective on lexicogrammar, since lexical items are the synthetic realisation of lexicogrammatical features.  And to take any perspective on lexicogrammar is to model lexicogrammar, not discourse semantics.

Moreover, as previously demonstrated, Martin & Rose do not understand the notion of 'lexical item', having previously (p84) regarded instances like each of the following as single lexical items:

• punishment with little consideration for victims and hardly any for the perpetrator
• opportunity to be reintegrated into the community (the perpetrator) has injured by his or her offence

[2] There are multiple confusions here.  On the one hand, lexical relations do not "build a field", and on the other hand, Martin & Rose, like Martin (1992), do not understand the SFL notion of field, and unwittingly use the term for quite distinct theoretical dimensions.  These will be explained in turn.

Firstly, as Halliday & Matthiessen (2014: 643) point out, the relations between lexical items (synonymy, antonymy etc.) are inherent to the organisation of lexis itself, and so do not "build a field".  For example:

• how does the inherent lexical relation of synonymy between big and large "build a field"?
• how does the inherent lexical relation of antonymy between good and bad "build a field"?
• how does the inherent lexical relation of hyponymy between colour and purple "build a field"?
• how does the inherent lexical relation of meronymy between toe and foot "build a field"?

Secondly, the notion of "building a field" (Martin 1992) is itself a tangle of multiple confusions. To take field first, in SFL theory, field is the ideational dimension of context; that is: the culture construed as a semiotic system. However, Martin misconstrues field as the ideational dimension of register, a functional variety of language, which, in turn, he misconstrues as context.

However, again, from the perspective of SFL theory, the notion of field in build a field does not correlate with context, but with the ideational semantics of a text. Martin's notion of "building a field" corresponds to the instantiation of ideational meaning during logogenesis.

[3] To be clear, these two instances are nominal groups, not single lexical items, and there are no hyponymic or meronymic relations between the lexical items young, man, first, love.

[4] This misunderstands synonymy and repetition.  As Halliday & Matthiessen (2014: 644) point out, both relations involve identity, not class membership.  It is hyponymy that is based on class membership.

[5] To be clear, even ignoring the previously identified problems with Martin & Rose's notion of 'contrast' — e.g. here, here and here — this is nonsensical.  Contrast doesn't "function to distinguish categories" any more than meronymy or hyponymy do.  Contrast is proposed as a taxonomic relation between lexical items.

[6] To be clear, SFL theory is a "metafunctional view" on language.  In terms of metafunction, Martin's (1992) system of IDEATION, including taxonomic relations, is a rebranding of Halliday & Hasan's (1976) textual system of lexical cohesion as a system of the experiential metafunction, misunderstood, and relocated from lexicogrammar to discourse semantics, despite being concerned with lexicogrammar — relations between (misunderstood) lexical items — rather than semantics.  For a thorough examination of Martin's (1992) system of IDEATION, see the clarifying critiques here.

Misrepresenting And Misunderstanding Lexical Cohesion

Martin & Rose (2007: 90):

In the past, studies of taxonomic relations have tended to focus on their roles in maintaining the cohesion of a text, through lexical ties between clauses (e.g. Halliday and Hasan 1976). The starting point in such cohesive models is with repetition, since the most explicit possible way of tying one item to the next is by repeating it. Next come synonymy and antonymy, which tie items to each other by similarity and contrast, with hyponymy and meronymy considered last. This is a grammar-based perspective, in which lexical relations are seen as serving textual functions, linking grammatical elements to each other in strings, similar to cohesive relations between reference items such as pronouns and articles: a young man - this man - he (see Chapter 5 below).

Blogger Comments:

[1] This is misleading.  Halliday & Hasan (1976) provides the original theory, lexical cohesion, a textual system of lexicogrammar, which Martin & Rose, following Martin (1992), have merely reinterpreted as an experiential system, thereby creating multiple theoretical inconsistencies, rebranded as IDEATION, and relocated to Martin's stratum of discourse semantics.  For a thorough examination of Martin's "original" theorising, see the clarifying critiques here.

[2] This misunderstands lexical cohesion. Lexical cohesion actually operates independently of grammatical structure, with relations also potentially obtaining between lexical items within the same clause.

[3] To be clear, the sequencing in this description ('starting point', 'next', 'last') is irrelevant. Unknown to the authors, the different types of lexical cohesion align with different types of expansion.  Halliday & Matthiessen (2014: 644):

[4] This is another misunderstanding of lexical cohesion.  As the name makes crystal clear, lexical cohesion involves relations between lexical items, not grammatical elements.

[5] Here the authors follow Martin (1992) in mistaking the 'indefinite article' for a reference item.  Martin's misinterpretation of Halliday & Hasan's reference, rebranded as his system of IDENTIFICATION, confuses DEIXIS, a system of nominal group structure, with the non-structural system of REFERENCE, and confuses textual reference with ideational denotation, as demonstrated in great detail here.

Misunderstanding Meronymy

Martin & Rose (2007: 90):

Wholes to parts

Likewise relationships of wholes to parts are also given various names in English, depending on the field, e.g. part, content, ingredient, constituent, stratum, rank, plane, element, factor, fitting, member, component, faction, excerpt, extract, episode, chapter, selection, piece, segment, section, portion, measure. 

In addition, facets name parts that are locations of wholes, e.g. the bottom of his soul, top, inside, outside, side, edge, middle, perimeter, environs, start, finish, beginning, rest. 

Measures name some portion of the whole, e.g. a cup of coffee, glass, bottle, jug, can, barrel, loaf mouthful, spoonful, ounce, pound, kilo, metre, acre. Again part-whole relations can be used cohesively between messages:

parts The chair's broken, – Which part?
facets Was it a good marriage? – Only at the start.
measures How much is petrol today? – More than a dollar a litre.

Technically whole-part relations are known as meronymy (mero- from Greek ‘part’).

Blogger Comments:

[1] To be clear, here Martin & Rose misconstrue the names of parts as names of part-whole relations.

[2] To be clear, the notion that a part is the location of a whole is nonsensical, as demonstrated by the simple fact that the side of a triangle is not the location of the whole triangle.

[3] To be clear, facets and portions serve as extended Numeratives in nominal groups such as the bottom of his soul, a cup of coffee; Halliday & Matthiessen (2014: 395):

 

[4] More importantly, in the authors' terms, the concern here is meant to be the taxonomic relation of meronymy, as a feature within the experiential discourse semantic system of IDEATION.  Instead, however, the authors have again unwittingly confused two distinct metafunctional systems of lexicogrammar: nominal group (experiential) and lexical cohesion (textual).

Misunderstanding Hyponymy

Martin & Rose (2007: 89)

Relations of class to member are given various names in English, depending on the field, e.g. a class of words, a make of car, a breed of dogs. Common examples include class, kind, type, category, sort, variety, genre, style, form, make, breed, species, order, family, grade, brand, caste. These can be used cohesively between messages, e.g. Like my new car? Yes, what make is it? Technically, class-member relations are known as hyponymy (hypo- from Greek ‘under’).

Blogger Comments:

Here Martin & Rose first claim that terms like class, make, breed etc. are names of hyponymic relations, and immediately contradict this by misinterpreting them as the superordinate terms in hyponymic relations ('car-make'); cf genuine hyponymic relations such as verb-word, Bentley-car, kelpie-dog.

To be clear, such terms are all synonyms for 'variety', which is the function they serve as extended Numerative in nominal groups such as a class of words, a make of car, a breed of dogs; Halliday & Matthiessen (2014: 395):

 

Moreover, in the authors' terms, the concern here is meant to be the taxonomic relation of hyponymy, as a feature within the experiential discourse semantic system of IDEATION.  Instead, however, the authors unwittingly confuse two distinct metafunctional systems of lexicogrammar: nominal group (experiential) and lexical cohesion (textual).

Misconstruing Temporal Circumstances As Lexical Items

Martin & Rose (2007: 88-9):

In other genres, series are an important resource for interpreting things and events. Newspaper stories for example jump around in time, so that readers must be able to recover relations between times in order to construe the sequence of events. The following extract recounts the events surrounding the 2001 rescue of shipwrecked refugees trying to reach Australia by the Norwegian freighter Tampa, and the Australian government’s shameful refusal to help them:

DRIFTING 22km off Christmas Island and with food and supplies running low, Captain Arne Rinnan was last night trying to maintain order on his besieged ship after being turned away by Australia and warned off by Indonesia. The Norwegian captain of the MS Tampa last night told The Daily Telegraph by satellite phone many of the 438 men, women and children on his ship were ill after their 11th day at sea ... 

But Prime Minister John Howard said after a cabinet meeting yesterday afternoon that the ship would not be allowed to enter Australian waters ... Hours later, the Indonesian Government responded by saying the boat people - who are believed to be from Pakistan, Sri Lanka, Afghanistan and Indonesia - could not return to Indonesia. 

Capt Rinnan told The Daily Telegraph he had not yet informed the boat people last night that Australia had refused them permission to land at Christmas Island. Asked if he was afraid of violence, he said: ‘Not at the moment, but we were and we will be if they are turned away. They are starting to get frustrated.' .., 

When he picked up the distress call 24 hours earlier, he believed he would be carrying out a rescue operation, delivering the boat people to the nearest Indonesian port. After reaching the stricken 20m wooden vessel, KM Palapa 1, the crew helped the boat people on board. With the strong south-easterly winds which buffet the area at this time of year, it took the Tampa crew three hours to get them all on board ... 

Capt Rinnan said the boat people had become distressed when told they might have to return to Indonesia earlier in the day, with some threatening to jump overboard. 'I said we are heading towards Indonesia and they said "No, you must head to Australia".' Capt Rinnan said they were 'just hanging around' late yesterday, waiting for Australian officials to come on board. (Tsavdaridis 2001: 1).

The potential complexity of tracking the events through the story is evident in the following list of times as they appear in the text:

last night
their 11th day at sea
yesterday afternoon
hours later
last night
at the moment
24 hours earlier
three hours
earlier in the day
late yesterday

As these times are out of sequence in such genres, time cycles are an essential lexical resource for recognising a sequence of events.

Blogger Comments:

To be clear, the wordings that Martin & Rose misrepresent as instances of taxonomically related lexical items are actually grammatical functions: temporal circumstances of Extent (three hours) or Location (all the rest).  In addition, two of these are only partially identified:

• their 11th day at sea features in the Location after their 11th day at sea
• at the moment features in the Location not at the moment.

In providing the location and extent of processes in time, such circumstances "are an essential grammatical resource for recognising a sequence of events".

Misrepresenting Extending Grammatical Relations As Elaborating Lexical Relations

Martin & Rose (2007: 88):

Many such antonyms are construed in the principles motivating the Reconciliation Act, with the contrast emphasised by negative polarity not, and the contrastive conjunction but:

SINCE the Constitution states that there is a need for understanding but not for vengeance, a need for reparation but not for retaliation, a need for ubuntu but not for victimisation;

Blogger Comments:

In terms of SFL theory, Martin & Rose here misconstrue extending relations between grammatical units as elaborating (antonymic) relations between lexical items (Halliday & Matthiessen 2014: 644).  In each of these instances, a variation: replacive relation (op. cit.: 471) obtains between two prepositional phrases in an embedded complex:

for understanding	but not for vengeance
for reparation	but not for retaliation
for ubuntu	but not for victimisation
1	+ 2

Misunderstanding The Text

Martin & Rose (2007: 88):

Finally Tutu rests his case on the contrast between retributive and restorative justice. Interestingly he argues that both types treat the converse roles of victim and perpetrator in some ways similarly. Retributive justice gives little consideration to either, whereas restorative justice classifies both as people:

Further, retributive justice - in which an impersonal state hands down punishment with little consideration for victims and hardly any for the perpetrator - is not the only form of justice. I contend that there is another kind of justice, restorative justice, which is characteristic of traditional African jurisprudence. Here the central concern is not retribution or punishment but, in the spirit of ubuntu, the healing of breaches, the redressing of imbalances, the restoration of broken relationships. This kind of justice seeks to rehabilitate both the victim and the perpetrator, who should be given the opportunity to be reintegrated into the community he or she has injured by his or her offence. This is a far more personal approach, which sees the offence as something that has happened to people and whose consequence is a rupture in relationships.

Blogger Comments:

[1] To be clear, retributive justice and restorative justice are nominal groups, not lexical items, and Tutu's argument is made by the grammar, not by the relation between the lexical items retributive and restorative.

[2] This misunderstands the text.  The two types of justice treat the victim and perpetrator in significantly different ways:

retributive justice … hands down punishment with little consideration for victims and hardly any for the perpetrator

whereas

restorative justice … seeks to rehabilitate both the victim and the perpetrator … 

Misreading The Text

Martin & Rose (2007: 87-8):

And the contrast between innocence and guilt underlies his second Argument:

It is also not true that the granting of amnesty encourages impunity in the sense that perpetrators can escape completely the consequences of their actions, because amnesty is only given to those who plead guilty, who accept responsibility for what they have done. Amnesty is not given to innocent people or to those who claim to be innocent. It was on precisely this point that amnesty was refused to the police officers who applied for it for their part in the death of Steve Biko. They denied that they had committed a crime, claiming that they had assaulted him only in retaliation for his inexplicable conduct in attacking them. 

Here there is a double contrast implied, between the innocent and the guilty, and between those who confess their guilt and those who falsely claim innocence, thus compounding their crimes.

Blogger Comments:

[1] To be clear, the argument is made through the grammar, not merely by the antonymic relation between the lexical items guilty and innocent, which is textual (cohesive) in function, not experiential.

[2] To be clear, the primary contrast here is between those who admit guilt and everyone else.  That is, the discourse analysts Martin & Rose have missed the main point of Tutu's argument: that amnesty is not primarily concerned with the contrast guilt vs innocence; it is given to those who admit their guilt, thereby taking responsibility for their actions.