3 Discourse and context I

3.1 Texts as discourse

In Chapter 2 we considered the origins of stylistics in Russian formalism, and the progress made in the analysis of literary texts in response to the developments of descriptive linguistics in the first half of the twentieth century. Many of the insights of the theory of foregrounding, and its various realisations in the analysis of form, are still relevant to stylistic analysis and have been refined and added to as linguistics has increasingly considered context and function as part of its scope. This chapter will introduce some of the main ways in which the consideration of function in language study has affected the way in which stylistics approaches the study of literary and other texts, and will begin by tracing the debates and controversies that accompanied some of these developments.

Stylistics has, on occasion, been the target of attacks from literary critics for what has been seen as an excessive concern with the linguistic form of (literary) texts at the expense of social, historical and other contextual factors that also play a role in a text's meaning. (Similarly, stylisticians have found themselves accused of failing to take adequate account of the important relationship between writer and reader which is mediated by the text. Recent advances in cognitive stylistics have addressed this criticism directly; see Chapters 5 and 6.) While no stylistician would accept that an analysis can incorporate too much linguistic detail, there is perhaps some truth to the point that stylistics has sometimes neglected contextual factors involved in meaning-making. The reason behind this is that stylistics has taken as its guiding light linguistics, which itself began with more formal concerns. Leech and Short (1981: 4) reflect on the changing nature of linguistics in the introduction to the first edition of their influential textbook on the style of prose literature:

linguistics itself has developed from a discipline with narrowly defined formal concerns to a more comprehensive, if more inchoate discipline, in which the role of language in relation to the conceptualization and communication of meaning has been fruitfully investigated.

The advances in linguistics noted by Leech and Short – incorporating a burgeoning of theories and related methodologies – are also reflected in the developments and scope of contemporary stylistics, as the remainder of this book will illustrate (see also the discussion of stylistics’ tendency towards eclecticism in section 1.6.3). Leech and Short (1981: 5) themselves comment upon the range of linguistic traditions which have influenced the development of stylistics and they attempt to characterise what they have in common as

a tendency to explore for pattern and system below the surface forms of language; to search for the principles of meaning and language use which activate and control the code.

This concern with form, then, does not disappear, just because there has been a rising interest in the functions of the forms described. Leech (2008) sees the link between form and function as vital to the definition of stylistics:

This interface between linguistic description and interpretation is precisely the sphere of stylistics as I see it: by undertaking a linguistic analysis as part of the interrelation between the two fields of study, we facilitate and anticipate an interpretative synthesis. Within stylistics, that is, linguistic and literary concerns are as inseparably associated as the two sides of a coin, or (in the context of linguistics) the formal and functional aspects of a textual study.

This integration of formal and functional aspects of language is probably not as new as it sounds and of course is not unique to stylistics. Even Saussure, who is often associated with formal analysis, explicitly concerned himself with language as a ‘social semiotic’ and the whole of systemic-functional grammar, particularly in the work of Halliday (1985), has been concerned to try to describe form and function within the same model. Some of the results of Halliday's work have proved particularly useful for the analysis of the style of literary and other texts, and others have added similar tools of analysis which combine form and function in their model of language. Note that the opposition between form and function is also a ‘convenient fiction’ both in the sense that Leech (2008) above sees it, since they are ‘two sides of a coin’ (and thus inseparable), and also in the sense that we could, and some analysts do, divide language up into more than two aspects. Thus, we could say that form is made up of the levels of language discussed in Chapter 2, and that function consists of at least semantics (at each of these levels) and pragmatics (and the grey area between them). In addition, there may be further layers of interpretation that we would want to associate particularly, but not uniquely, with literary texts. These latter are those layers of interpretation which are least closely tied to the text itself, though they may be deduced from the nature of the language.

Before we consider the use in stylistics of some of these functional tools of analysis, let us consider one more of the challenges that stylistics has faced, which is partly concerned with the question of the relationship between form and function. This is the charge of ‘interpretative positivism’ (see Simpson 1993: 111–16) which was most famously levelled against stylistics by Stanley Fish (1981) and which many stylistics books since then have felt compelled to answer. This charge will be explained below, but it is worth making the connection here with the related controversy about the objectivity of stylistics, which was discussed in section 1.6.2. Here, the attack made by Fish is concerned with the question of whether it is possible, or indeed desirable, to do as he claimed that stylistics did and to ‘read off’ meaning automatically from text:

For both critics operate with the same assumptions and nominate the same goal, the establishing of an inventory in which formal items will be linked in a fixed relationship to semantic and psychological values.

(Fish 1981: 75)

Whilst Fish is able to find some examples of this kind of apparent rigidity in seeing form and function as having a fixed relationship, in fact stylistics has long embraced the notion that form does not confer a particular meaning in any automatic sense. Thus, though there may be some conclusions one can draw from the form of a text, they are only the beginning of literary interpretation, which depends on the conjunction of form with particular content and also with particular contexts of production and reception. Thus, for example, we may wish to comment on the use of long and convoluted sentence structure in a novel to the extent that it draws attention to itself (i.e. is foregrounded). However, the significance of this tendency will depend on its year of production (since longer, more complex sentences have been declining for some centuries in English), on its genre (it would be unusual in a detective fiction novel and less so in a ‘literary’ one), on its topic (the more philosophical, the less unusual it would be) and on its year of reception (contemporary readers of eighteenth-century fiction would not react adversely to long and complex syntax, though the twenty-first-century reader may do so). In other words, the question of external deviation is not simply a qualitative linguistic calculation. Literary interpretation requires us to recognise which of these, and other, factors are influencing our view of a text, and included in the list of affective factors is the construction of the language itself.

This lack of automation in the move from language to interpretation (form to function) causes Fish (1981: 72–3) to attack stylistics from the opposite direction, for what he calls ‘a serious defect in the procedures of stylistics’:

the absence of any constraint on the way in which one moves from description to interpretation with the result that any interpretation one puts forward is arbitrary.

Many stylisticians have found fault with the apparent logic of Fish's attack on their field. Toolan, for example, acknowledges many of Fish's arguments in relation to the more automatic interpretative claims of stylistics, though he also takes up the question of whether stylistic analysis is still worth doing:

Fish rightly argues that the stylistician's focus on a particular phonological or syntactical pattern in a text is itself an interpretative act. As I hope to have suggested, this in itself does not constitute an overwhelming argument to stop doing stylistics unless (a) that interpretative act is shown to be incoherent or ill grounded, or (b) more coherent interpretative acts are presented, and preferably both.

(Toolan 1996: 131)

Though stylistics as a whole has never been either as rigid or conversely as loose as the worst cases represented by Fish, there is nevertheless some movement to be seen through the history of the subject from a more text-intrinsic approach towards an approach which recognises the reader as an integral part of the construction of textual meaning. Thus, most approaches to stylistics these days see the text (literary and other) as the centre of a communicative event which may take place in a range of places and timescales, and which includes the producer and the recipient. In other words, literary and other texts are considered as discourse events, and not just stable artefacts with stable meanings.

However, it is important to note that there remain some features of the text which can be identified and described irrespective of their intended and/or received effect. The interpretation of how such features may affect the writer's meaning and the reader's meaning is one of the more subjective aspects of stylistic analysis, though the link to textual features does at least achieve the scientific standard of explicitness that enables others to see how an interpretation is arrived at (see the discussions of methodology in section 1.6 and Chapter 7). Stylisticians rightly claim that their procedures in analysing literary and other texts are explicit, logically argued and do not assume that language is transparent in its meaning. Together, these aspects of the stylistics methodology make sure that those encountering a particular stylistic description of a text will know how the interpretation is arrived at, and will be able to engage with it on its own terms.

The remainder of this chapter will introduce some of the functional categories that have been applied to stylistic analysis. Where these tools of analysis depend on other linguistic knowledge this will be explained in the relevant section.

3.2 Functional categories and style

Although the analytical tools described in the following sections are often labelled as ‘functional categories’ in Hallidayan and other systems of description, they are not really categories at all. What we have in most of the cases is a system of related concepts (i.e. meanings) which are realised linguistically in a range of ways, often with a prototypical realisation in one of the systems of the language, but usually including a set of other, less central possibilities for producing the same effect. We will see below how this works for a number of functions, but it is convenient to illustrate the point about how these systems work using negation.

If we take a very strict, grammatical, view of negation, we would probably say that it is realised by a negative particle which is attached to the verbal element of the clause. Thus in ‘I didn't know you were coming’, the negator, ‘not’, is attached to the first auxiliary, the operator in the verb phrase, and here occurs in a reduced form (‘n’t’). A less central, but still clearly negative, form is the use of the adjectival ‘no’ before nouns in noun phrases as in ‘no time’ or ‘no food’. A still clearly negative meaning is attached to a range of prefixes in English, which have a variety of precise meanings, depending on their root word, but all denote some kind of negation, as in ‘incredible’, ‘unsuccessful’ and ‘asocial’. Whilst their meaning may be clearly negative, these morphemes are not able to be added to all equivalent root forms (note ^*asincere, ^*insubtle, ^*unsecure). This, then, is a less central form than the verbal and nominal negation, which are not similarly restricted. Still less formal is the existence of a range of lexical items in English and other languages where the semantics indicate a negated denotation. These include ‘lack’, ‘loss’ and ‘absence’ as clear members of the group, though the parameters for identifying a lexical item as negative are not clear-cut and result in items being more peripheral to the group, including for example those which have negative evaluation as part of their meaning, such as ‘feeble’ or ‘cruel’. In the case of the clear examples, though it is difficult to establish a straightforward methodology for deciding whether they are performing some kind of negation, we can nevertheless say that their denotation is most easily paraphrased using a negated phrase or clause. So ‘lack of x’, for example, is clearly denoting the situation where there is ‘no x’, and ‘loss of x’ clearly denotes ‘no longer having x’. This test is less clear-cut in the latter cases, where we could use a negative definition, or a positive one. Thus, being ‘feeble’ could be defined as ‘having no strength’, but it is arguable that most pairs of gradable adjectives have one partner which could be thus defined: cold (absence of heat), calm (absence of strong feelings), short (lack of length).

We will see below that the strength of this arrangement of form–function relationships is in its very lack of one-to-one conformity. Thus, it is the fact that a set of related meanings, such as those we can define under the term negation, can be produced by an array of different forms in a language like English that makes all sorts of meanings possible. If there was one, and only one, way to negate concepts in English, the language, and literature as a result, would be thereby impoverished in its potential effects, as we shall see in the sections below. This point, that the lack of one-to-one form–function relationship is at the heart of human language, takes us back to one of the founding principles of stylistics itself, the assumption that style is made up of a series of choices among options provided by the language. We considered this idea in section 1.6.4. Here, we will see how choices made by text producers have the capacity to influence the meaning created by text recipients.

3.2.1 Transitivity

The syntactic system of transitivity is the central component in Halliday's functional grammar, which was conceived in relation to English, and is thus perhaps peculiarly suited to that language, although the development of systemic-functional grammar theory and models is not so restricted.

This system acknowledges that the verbal element is the core of the clause, and that the choice of lexical verb itself is somehow crucial to the rest of the choices in the clause. So, for example, if a clause is being constructed with the verb ‘eat’ as the lexical verb, the choice of this verb, which is labelled a ‘Material Action – Intention’ process, will dictate that the grammatical subject of that verb will be an Actor (an animate being) and there will be a grammatical object following the verb which will be the Goal. Transitivity analysis, then, takes the traditional idea that some verbs require objects (transitive) and some do not require objects (intransitive) one step further, and gives groups of verbs semi-semantic labels, describing them as occurring in the context of a number of participants, also with semantic labels. There are a number of versions of these categories of verb, but here we will draw upon that used by Simpson (1993), as he was using the model specifically for stylistic purposes (see Table 3.1).

Table 3.1 Transitivity categories from Simpson (1993)

Before we consider the problems with this model as well as the uses to which it can be put in stylistics, let us look at some simple examples of the way in which transitivity analysis characterises clause structure:

As we will see below, the basic idea that the use of different process types can make a difference to the effect of a text is of use to the stylistician. However, the model is not without its problems. For one thing, there is no single agreed set of transitivity types, as a result of their being based not solely on formal properties. The decision as to how many different kinds of Mental Cognition categories there should be in English, for example, is not a clear-cut one. Similarly, we cannot assign verbs to transitivity categories out of context, because they may have a range of possible transitivity behaviours. Thus, the verb ‘drop’, for example, may be a Material Action Supervention verb on one occasion (She dropped the glass in her fright); a Material Action Intention in some cases (She dropped the glass to see it shatter) and a Material Action Event in other cases (The glass dropped from her hand). In addition, of course, there may not be enough information in a particular context for us to be sure which transitivity process is intended, though there is often a preferred interpretation, which in the case of ‘drop’ would be the Supervention one (She dropped the glass).

Of course, very many verbs in English and in other languages are used metaphorically in everyday contexts and much of the time. Whilst this means that we may in some cases distinguish between the core use and the creative use of an item, the normal use of an item may in some cases be the metaphorical one, and this would make the metaphorical meaning more basic. In the case of ‘drop’, the Goal is the key to its meaning, so that instead of ‘She dropped the glass’, the utterance ‘She dropped the case against him’ would bring the metaphorical meaning to the fore. This may still be classified as a Material Action, though it is clearly intentional, whereas the default interpretation in literal cases would be Supervention. The question of how to categorise other metaphorical uses of a verb may be more difficult, so that one could argue about the transitivity of ‘She dropped her voice’, given that it describes a Verbalisation process as though it were a Material Action.

Despite these problems, which really all stem from the system being on the form–function borderline, transitivity analysis can be helpful in characterising aspects of style that no other tool has yet captured. The key to dealing with the problems of categorisation is not to treat transitivity types as categories at all, but as points of reference on a continuous plane of meaning, which is probably multi-dimensional, but may be likened to the vowel chart of phonetics which represents the mouth in diagrammatic form and in two dimensions. Thus, the different types of transitivity are ‘idealised’ types of meaning, with concrete actions being the most typical Material Actions, and more abstract ‘actions’, such as deciding or condemning, having dual identity as material actions and mental cognition or verbalisation respectively.

The question of why and how we would use this tool to address questions of style is answered when we see how well it can explain certain perceived differences in style that were previously difficult to pin down. Simpson (1993: 110) comments on Halliday's own use of transitivity analysis to explain how William Golding manages to convey the mind-style of the Neanderthal people in The Inheritors:

Where the world depicted from the perspective of ‘new’ people is very much like our own, the world seen by Lok and his tribe is distinctly unfamiliar. Within the limits of Lok's understanding, people appear to move aimlessly, seldom acting directly on objects in their physical environment. This sense of discontinuity, Halliday argues, is created through particular selections from the system of transitivity.

The analysis by Halliday is one of those which Fish (1981) takes exception to, and Simpson (1993) defends. Simpson's description of the problem which Fish identifies is as follows:

The criticism rests primarily on what might be termed the interpretative positivism shown by stylisticians who simply invoke linguistic descriptions as a way of confirming the decisions they have already taken about a text's meaning.

(Simpson 1993: 111)

The way in which Simpson defends stylistic analysis – in particular transitivity analysis – from Fish's attack is to take another Golding novel, Pincher Martin, which is about a drowning man, and subject it to the same kind of analysis that Halliday used on The Inheritors. His conclusions are that the two texts, despite their rather different storylines, characters and settings, ‘display uncannily similar patterns of language’. Thus, he notes that both texts use an abundance of body parts as Actors (e.g. ‘his hand let the knife go’) and frequent uses of event processes with inanimate actors where we would normally attribute the actions to human agency (e.g. ‘the lumps of hard water jerked in the gullet’). Simpson's point is that stylistic analysis can distinguish between the linguistic choices in a text and their literary (or other) effect, and that in practice ‘avoiding interpretative positivism often requires no more than a modicum of caution’ (Simpson 1993: 113). The problem, as Simpson points out, is when ‘a direct connection is made between the world-view expounded by a text and its linguistic structure’. In comparing his own analysis of Pincher Martin with Halliday's analysis of The Inheritors, he comes to the conclusion that although the specific interpretation of each novel is different, there is ‘an interpretative “lowest common denominator”’ which demonstrates that the texts ‘despite markedly divergent story-lines, are at one level of analysis stylistically very similar’ (Simpson 1993: 113).

Having considered the model of transitivity and the issues arising from its use in stylistics, let us look at some examples of analysis of a prose passage, and consider the contribution of transitivity choices to the style of the text. Here, then, is a passage from the opening pages of Ever After by Graham Swift (1992: 3), which is narrated by a university professor who has recently attempted (unsuccessfully) to commit suicide:

This short passage uses transitivity choices which might not be normal for first person narration, where it is to be expected that the narrator will often be the Actor in any material actions that take place. Instead, in this passage at least, the narrator is the Goal of a range of material actions (rush away, pump, thump, jump-start and wire) because at that point he is unconscious through his efforts to commit suicide. Before this, the passage has another slightly surprising transitivity choice, where he is admitting that he has failed in his (material action) of killing himself, but chooses to use a Relational Intensive process instead: ‘My recipe, you see, was at fault.’ Here, the cocktail of drugs that he has taken is blamed for the failure, rather than the producer of that cocktail, the narrator himself. This effect is strengthened in the next sentence where the passive constructions (‘can be well-executed or hopelessly botched’) do not include any agents, and the generic nature of the statements themselves add to the effect of the narrator not taking responsibility for the failed suicide attempt. Once the narrator is conscious again, he becomes the Actor in material processes, though the choice of ‘breathing’ is interesting as it would probably normally be seen as almost unintentional (supervention?) because we don't have to breathe consciously, but in this circumstance of course, the laboured breathing of one back from the brink of death is almost a conscious choice, as indicated by ‘breathing and thinking for myself’.

The following passage comes from A Spot of Bother, by Mark Haddon (2007: 13), in which another man in his 50s is coping with a breakdown and in this passage is suffering from extreme phobia of flying:

The enforced immobility of sitting in an aircraft is underlined here by the transitivity choices where the most active that the protagonist can be is to ‘imagine’ – a mental cognition process, and to ‘see’ and ‘hear’ – mental perception processes. This man, it turns out later, is increasingly at the mercy of his mental breakdown, and is swept along by the tide of his emotions. The fear-of-flying episode, then, is a microcosm of his life experience as told in the novel, and prepares the reader for his frequent experiences of this kind, one of which occurs on the following page when his daughter announces that she is planning to marry Ray:

The transitivity choices here, then, are varied but on the whole are not intentional material actions, except the two actions he watches himself perform (‘kiss’ and ‘shake’). The others are mental perception (‘look’ and ‘watch’), relational possessive (‘had’) and material action supervention (‘fall’). The final two sentences remove George from the Actor role entirely and in the final sentence we have an example similar to those analysed by Halliday and by Simpson, in which the body seems to take control of the person rather than vice versa.

Whereas in the case of Neanderthal people, this transitivity use reflects a lack of understanding of cause and effect on the part of Lok and his relations, and in the case of Pincher Martin, it reflects the lack of control of someone who is in physical distress as he drowns, in the case we have here, the lack of control over his body is one of the symptoms of George's mental breakdown. So, it seems that, as Simpson claims, it is entirely possible to see different interpretative points in the use of this particular transitivity pattern, and also to trace the ‘lowest common denominator’ that he speaks of in terms that can include all of these examples. Thus, the loss of bodily control, or perceived lack of such control, is at the root of all of these examples.

Transitivity analysis is not to be seen as a rigid tool of stylistics, then, but an indicator of the ways in which certain effects can be achieved. Because verbs occur in every clause in English (with only a few exceptions), there is clearly a great deal of transitivity that can be analysed in even a short text. The student of stylistics will need to consider carefully the judicious use of such analysis and the extent to which it can be carried out in any one project. The reader is referred to section 7.1.3 and other sections of Chapter 7 for discussion on these kinds of question.

3.2.2 Modality

Like transitivity, modality is also one of the systems discussed in Hallidayan approaches to linguistic description. This is another system where there is a set of prototypical forms reflecting the meanings of modality, but where there are also more peripheral forms which can produce similar meanings. Let us consider first, then, the general functions of modality, and its various forms, before considering its role in the production of literary meaning.

The general system of modality reflects what Halliday (1994) calls the ‘interpersonal’ metafunction of language, which is one of three metafunctions he sees as co-existing in linguistic usage. Thus, if transitivity reflects the ideational metafunction, in which language represents the world in certain ways, then modality is intended to reflect the interpersonal function, in which language mediates between people. However, as we will see, modality may be seen as ideational in literary and other texts too. Its main contribution to textual meaning is to reflect the producer's opinion about what s/he is saying or writing. Thus, the modal auxiliary verb might, for example, indicates that the speaker is not completely sure that the process following is going to happen. Take the sentence ‘Joanne might be home by nightfall’, for example. The question of whether Joanne will indeed be home before dark is left uncertain as a result of the use of ‘might’, and it is the speaker's uncertainty which is communicated.

The auxiliary modal verbs are the prototypical carriers of modal meaning. The set of modal verbs includes at least will, would, may, might, shall, should, can, could and ought (to) and for some speakers, other verbs, which have become lexical for some dialects of English, remain modal too. These include need and dare, which in their modal form tend to allow the same kind of reduction of form in casual usage that is common for more central modal verbs (needn’t, daren’t). In addition to the modal auxiliaries, there are modal adverbs and adjectives which carry very similar meanings, and can be used to paraphrase clauses with modal verbs:

Notice that these two sets of examples demonstrate two of the main types of meaning that are normally recognised in relation to modality. Thus, the question of whether something is likely/certain or unlikely is covered by the broad concept of epistemic modality. It reflects the speaker's level of confidence in the truth of their utterances. Related to epistemic modality, and often thought to be a sub-category of it, is perception modality which is usually delivered by verbs of perceiving, such as ‘see’ and ‘hear’, though with the meaning of ‘understand to be true’ as in:

The other way that perception modality is delivered is through adjectives and adverbs which relate literally to perception (apparently, clearly) but when used modally are used to draw conclusions about the certainty or otherwise of the proposition.

So, on the one hand we have epistemic modality (including perception modality), which gives information about the speaker's confidence in his/her utterances. On the other hand, however, we have the two related types of deontic and boulomaic modality, which indicate the necessity and desirability respectively of the proposition in the utterance. The following are examples of these modality types (example 8 indicates deontic modality and example 9 boulomaic):

Note that most of the modal auxiliary verbs have the capacity to deliver more than one of these types of modal meaning, as can be seen in the following groups of examples:

It is important to note that there is no comprehensive list available of all the modal forms in English, or any other language. This is partly because the work has not been done, but this in turn may be partly because it is possible that such a task can never be complete. Not only adjectives (definite, certain, unsure) and adverbs (possibly, certainly, hopefully) can be modal, but main verbs also carry modal meaning in some contexts:

Note that these verbs (think, want, understand) are not always unambiguously modal in their meaning. For example, with third person subjects, or with different polysemous senses, they may not reflect the speaker's view of how things are:

Examples 18 and 19 report the thoughts and desires of a third party and are therefore not reflecting the speaker's viewpoint. Example 20 uses a non-modal sense of ‘understand’ to make a categorical assertion.

In addition to these modal uses of main verbs and the adverbs and adjectives already mentioned, it is possible for modality to be conveyed by more peripheral forms still. For example, the epistemic uncertainty of ‘she might get better’ can also be conveyed by a shrug of the shoulders or the rising intonation of a question imposed over the less modal form of ‘She’ll get better?’. Note the use in the last sentence of ‘less modal’, rather than ‘unmodalised’ because although this sentence is as certain as English can be about its content, it is nevertheless marginally modal because of the presence of the modal auxiliary will (in this case reduced to ‘’ll’. This is language-specific because some languages do not create future reference in their verbs by use of a modal auxiliary, using instead a future tense form, morphologically created. In English, however, the verb element of clauses does not allow future reference through tense forms, but makes a future form by using the will auxiliary to predict the future as accurately as possible. It is quite sensible in some ways for the language to reflect what we know to be true – that the future is not certain – by using a modal auxiliary.

There are a range of possible effects of the different kinds of modality when used in texts, both literary and non-literary. Before we take a look at some examples, there is one more important point to be made in relation to modality, and that is to consider what happens when it is not there. Unmodalised utterances are normally labelled ‘categorical’ and they assert their propositions with none of the undermining of certainty which would be the case in a modalised utterance. The first of the following examples, which uses weak epistemic modality, then, is very much less certain than the categorical version which follows it:

What is perhaps surprising, but less so on reflection, is that the third version here, which uses strong epistemic modality, is also less convincing than the categorical version. There is something about modality, even strong epistemic modality, which undermines the certainty of the proposition, even while it asserts it emphatically. It is precisely this emphatic nature of the strong modality which seems to undermine it, probably by drawing attention to the very question of its certainty. The categorical utterance, by contrast, is more confident in its assertions and appears to not even raise the question of reliability.

Let us see how this simple division of clauses into categorical and two main types of modality (epistemic and deontic/boulomaic) may affect a poem's potential for meaning. The following opening of a sonnet by Douglas Dunn recalls the experience of bringing food up to his terminally ill wife:

The first eight lines of the poem, rather surprisingly, use epistemic modality to describe what the husband ‘might’ find when he climbs the stairs, though it is clear that the addressee of the poem is no longer there, and this is confirmed later (‘at where you died’). The more obvious modal forms to use in a situation where the bereaved person is wishing his partner was still there would have been the boulomaic forms such as ‘wish’ or ‘want’. Instead, this poem presents the terrible truth of bereavement, which is that the bereaved hope and almost expect to see their loved ones again, which is reflected in the epistemic modals here. This is incongruous for the reality which is boulomaic, since they are desiring or wishing for the return of the person who has died. The use of epistemic modality here is thus foregrounded.

The question of how modality affects literary and other textual meaning has been addressed at some length by Simpson (1993), particularly in relation to narrative fiction. Simpson bases what he terms a ‘modal grammar’ on the types of modality introduced above, and demonstrates that certain genres of fiction tend to use modality in recognisably different ways, reflecting the viewing position or ‘point of view’ (POV) of the narrative. His modal grammar of POV is a development from Fowler's (1986) earlier four-way classification of narrative types:

• Internal type A: first person narration, from POV of participating character, foregrounded modality and verba sentiendi (words denoting thoughts, feelings and perceptions)
• Internal type B: third person narration; omniscient narrator, privileged access to participants’ thoughts and feelings, though no authorial presence, verba sentiendi
• External type C: no account of participants’ thoughts and feelings, third person narration, no modality or verba sentiendi
• External type D: explicit modality, words of estrangement, some first person use, but no privileged access to participants’ thoughts and feelings

Simpson introduces the model of modality sketched above into this framework, in order to reflect the more complicated range of narrative types that he observes. Thus, the division of narrative types into first person narration (Category A) and third person narration (Category B) is further sub-divided as shown in figure 3.1.

Figure 3.1 Simpson's modal grammar

Simpson's Category B, then, is divided into those narratives which reflect the viewpoint of one (or more) of the characters (Reflector Mode) or no particular viewpoint from within the story (Narratorial Mode). Then each of these categories is further sub-divided into positive, negative and neutral ‘shading’ which are characterised as follows:

• Positive shading: deontic and boulomaic modality foregrounded; generic sentences and verba sentiendi present
• Negative shading: epistemic and perception modality foregrounded; supplemented with ‘words of estrangement’
• Neutral shading: dominated by unmodalised categorical assertions; few verba sentiendi or evaluative adjectives and adverbs

The resulting classification of narrative types into nine categories allows the analyst to demonstrate similarities and differences of narrative style across a range of genres and text-types. In order to give a flavour of these categories, we will consider a number of passages from narrative fiction here, though there will not be room for nine examples, and as it happens, many narratives are made up of passages from different categories. The following passage comes from McCall Smith (2007: 2–3):

The transitivity choices in this passage reflect the different characters of the two protagonists in this detective novel. This is unlike the so-called ‘hard-boiled’ detective novels of tradition, which Simpson claims use mostly Category A (i.e. first person) narrative style, with neutral shading, so that there is no modality, few verba sentiendi and few evaluative adjectives or adverbs. This novel is one of a series about the ‘No. 1 Ladies’ Detective Agency’, set in Botswana, so it is different from the norm of this genre in being about female detectives in an African setting who are not trying to emulate the habits of the male detectives in all details. These novels are written in third person (Category B), but through a combination of modality and free indirect style (see section 3.2.4) give the reader a strong sense of the opinions and views of the two main characters, Mma Ramotswe and Mma Makutsi.

As seen in the passage above, the narrative tends to be in Reflector Mode, though alternating between the viewpoints of the characters. The first section, then, reflects the thoughts of Mma Ramotswe, the senior of the two detectives, and we see the evidence of her opinions in the deontic modality (‘behaved themselves as they should’) and the use of generic sentences (‘human beings can't really help themselves’). This, then, gives a positive shading, where the action is ‘located within [the] viewing position of [a] character, offering their opinions and judgements’ (Simpson 1993: 75). In the second part of the passage, Mma Makutsi thinks about her employer's views, and though it remains in Reflector Mode, there is a more negative shading with the use of epistemic modality (‘she thought Mma Ramotswe was right’; ‘she knew that’) so that Mma Makutsi's lower self-confidence is reflected in this slightly more ‘estranged’ style, though the presence of one of the verba sentiendi (‘she felt that’) is more indicative of positive shading.

As the reader will see from this analysis, even in a short extract from a literary work there is no single consistent style that can be categorised according to Simpson's modal grammar of point of view. This demonstrates that just as with the transitivity analysis, what we have are not categories so much as idealised reference points which can be used to characterise the fluctuating style of any narrative. The following is an extract from a rather different kind of novel:

This is from A Star Called Henry, which relates the experiences of a young Irish man caught up in the Easter uprising in 1916 through his voice and in first person (Category A). In this passage, having been captured and badly beaten by the British soldiers, he shows signs of detachment from his surroundings, with a mixture of categorical assertions (‘There was a light this time’), demonstrating neutral shading, and of estrangement through the negative shading of epistemic modality (‘I knew’) and perception modality (‘I heard’, ‘I saw’). Note that these effects of estrangement and detachment are reinforced by the transitivity choices (see section 3.2.1), which include only the most basic of intentional material actions on the part of the narrator (‘I lay on the floor’; ‘I sat up’) and some which are barely actions (‘I stayed still’). Mostly, there are descriptions using intensive relations (‘It was dark again’) and events (‘A slamming door had woken me’) as well as one or two of the kind of involuntary movements of body parts – and bodies – that we saw in Halliday's and Simpson's analysis of Golding's novels (‘a man who hit the ground hard’; ‘the breath rattled’). Though other parts of this novel demonstrate different aspects of Simpson's modality types, so that for example, there are times when the narrator expresses his wishes (deontic modality, positive shading), the life of the young Henry Smart is so traumatic that he clearly learns the skills of staying detached when his body and mind are under attack, and these are experiences that allow him to survive the brutal treatment he gets later.

Modality, as mentioned at the beginning of this section, is often seen by Hallidayan scholars as forming part of the interpersonal metafunction of language, rather than, say, the ideational metafunction. Its use in fiction, however, is not quite as simple as its use in face-to-face interaction. For example, deontic ‘should’ would be interpersonal if you said to a friend ‘You should phone John’, because it informs the hearer of the opinion of the speaker. However, in a novel, the use of deontic ‘should’, though remaining the opinion of the speaker, is no longer directed at the recipient, at least not if that recipient is seen as the reader.

3.2.3 Cohesion

The third functional category we will consider here, as being a useful tool of analysis for style, is cohesion. Halliday was again instrumental in bringing the concept of cohesion to the attention of linguistics, and Halliday and Hasan (1976) describes a model of cohesion that can be used to assess certain distinguishing features of literary and other texts.

The basis of cohesion is the idea that texts are not random sequences of sentences and that there must therefore be some structuring devices that link adjacent sentences in a text, which help the reader to make sense of their relation to each other. Whilst not as strictly rule-governed as the internal structure of sentences, the cohesion between them is nevertheless traceable to certain structural and semantic features which we will investigate in this section as they affect style.

The basic concept of cohesion is the idea of a textual ‘tie’ between units in different sentences which helps the reader to perceive the referential identity or topical consistency of different parts of a text. The simplest, and most prototypical, tie is that between a pronoun and its antecedent, as in:

Note that the mysterious man is introduced by an indefinite article (‘a man’) and in the following sentence the same person is referenced as ‘he’ twice. This is the normal order of things: a person, animal or thing is introduced in one sentence and is referred to by a pronoun in ensuing sentences until there is a danger of ambiguity, when some further unique referring phrase will be used. This kind of reference is known as ‘anaphoric reference’, which means that the pronoun refers backwards to a fuller phrase in an earlier sentence. It is possible, though rarer, for the pronoun to precede the phrase explaining its referent, as in ‘He was good-looking. The man lit a cigarette and smiled.’ The main mechanisms of cohesion are:¹

• Repetition – of similar or identical words, phrases or clauses.
• Reference – different ways of identifying the same referent.
• Substitution – the use of pro-forms (pronouns and other forms) to replace full phrases.
• Ellipsis – the missing out of highly predictable words and phrases.
• Conjunction – the conjoining of sentences by conjunctions (and, but, or).
• Lexical cohesion – the inclusion of words with semantic relations (e.g. synonymy) between them.

Whilst it is true to say that all texts are cohesive, the variability in the extent of cohesion of a text is quite significant in literary terms. Thus, although some cohesion is necessary for a reader to make sense of the text, some genres and text-types are more likely than others to minimise – or maximise – the concentration of cohesive ties. Children's literature, for example, is highly likely to include a large intensity of cohesive ties, as clues to make sure that the child is following the narrative. The following example is the text of a picture book intended for young children:

Here, there is a considerable amount of exact repetition (e.g. ‘Oh NO!’ said Mr. Bear, ‘I can't stand THIS.’) as well as some partial repetition of frames with different fillers (e.g. ‘So he got up and went to sleep in Baby Bear's room’ / ‘the living-room’). This short passage also has related lexical items forming a semantic field, and thus giving a semantic cohesion to the passage (‘late’ – ‘tired’ – ‘bed’ – ‘asleep’ – ‘snore’). There is a small amount of substitution of pronouns for names (‘he’ for ‘Mr. Bear’), but less than one might expect in literature for older age groups, perhaps. There is also one example of ellipsis, where ‘Mr. Bear didn’t’ implies ‘Mr. Bear didn't fall asleep’. It is one of the features that marks this story out as being suitable for children, rather than babies, because they need to be able to follow the cohesive ties, including the ones where something is actually missed out.

Whilst there are clearly practical and pedagogical reasons for the super-cohesive style of this particular narrative, similar features may be used by other writers wishing to evoke the innocence of children's stories in their own. One writer who does this is McCall Smith, whose series of books about the No. 1 Ladies’ Detective Agency are known for their ‘simple’ style, reflecting a down-to-earth morality from Botswana which the protagonist, Mma Ramotswe, tries to follow. One of the features of this style is the apparent ‘overuse’ of full names, rather than shorter versions of the names or pronouns, some repetition with variation as we saw in the children's story and a reasonably dense lexical cohesion:

The obvious stylistic features to notice in the above extract are:

• Use of full names: ‘Mma Ramotswe’ and ‘Mr J. L. B. Matekoni’ repeatedly used rather than shorter variants.
• Repetition: ‘Mma Ramotswe liked to lie down’; ‘Mr Matekoni would not lie down’
• Semantic cohesion: ‘Lunch’ – ‘midday meal’ – ‘vegetables’; ‘lie down’ – ‘sleep’

At the other extreme from these highly cohesive texts are those poems which are often seen as ‘difficult’ to read, partly because they are lacking those cohesive ties that would make the meaning plain and the links between sentences clear. Here is the opening of a poem by McGuckian, to demonstrate (N.B. we began to discuss this poem in Chapter 2; see example 34 in that chapter):

The subject-matter of this poem is very difficult to work out from the opening lines, though it does become clearer later on. The semantic (lexical) cohesion is almost entirely lacking, if you look at the lexical items in this stanza:

• Crocus, night, life, live, rain, sun, red, gold, sickness, hands, face, summer, ruin, crumpled, paper, dry, snap, suitcase, closing

Each time a semantic link appears (life – live; rain – sun; red – gold) it is followed by another semantic field with no clear link to the ones already established. In addition, though there is a consistent second person pronoun (‘you’) used throughout, it is not clear who it refers to, nor whether it is intended to be interpreted as second person (i.e. the reader or a particular addressee) or as the equivalent of ‘one’ (generic third person), which may in turn be interpreted as ‘I’.² As a text, then, it barely manages the basic cohesion that would make it comprehensible, as there is no clear narrative, the various processes being described as hypothetical, rather than occurring in some kind of time-frame. This will lead the practised reader of contemporary (and particularly of McGuckian's) poetry to search for other links, such as the interpretation of ‘rain time’ and ‘sun time’ as seasons, which then link to summer, and possibly also to the red and gold sickness. Crocuses can then be seen as connected to spring, though the connection of all of this with the ‘dry snap of a suitcase closing’ remains to be inferred retrospectively from other parts of the poem where it becomes clearer that it is a partner that is leaving the narrator.

Cohesion, then, can have an effect on the meaning of a text by making the style either very plain and childlike or difficult to work out, with the effect that the reader has to work extra hard to interpret the text.

3.2.4 Discourse presentation

This chapter has been considering textual analysis beyond the micro-structural concerns of the levels model of language, and this has so far led us to consider those systems identified by Halliday and his followers as simultaneously formal and functional. On a similar level of analysis to these Hallidayan systems, there are some other ‘local textual functions’ (Mahlberg 2007) that are just as useful for stylistic analysis, and which also seem to combine form and function in similar ways. In this section, we will consider one aspect of narratives which has received a great deal of attention from stylisticians in its own right: the presentation by a narrator of others’ words (written or spoken) or of their thoughts. This system has frequently been called ‘speech and thought presentation’ but since writing started to be included, it has become simpler to label it ‘discourse presentation’ to cover both written and spoken language and also the thoughts that are sometimes presented by narrators as though they were witnessed directly.

The model we will use here is from Leech and Short (2007), which reproduces the model in the first edition of their book (Leech and Short 1981), though there have been changes to the model in other places (e.g. Semino and Short 2004) and this process is continuing in more recent work. The original model, then, includes the following categories of speech presentation:

The formal differences between indirect and direct speech presentation include the following markers, assuming that indirect speech is converted from direct speech:

• The inverted commas are removed
• The speech is marked by a subordinator (e.g. that)
• Any first and second person pronouns change to third person
• Any present tense verbs change to past (including auxiliaries)
• Any proximal deixis (see section 6.3) changes to distal (e.g. this → that)
  (adapted from Leech and Short 2007: 256)

If we look at the following related sentences, all of these changes have taken place:

The other categories of speech presentation in Leech and Short's model have a range of formal features which distinguish them from the two traditional categories of direct speech and indirect speech presentation. Thus, NRSA does not include any reference to the actual speech itself, though it gives an indication of the speech act that was used. (e.g. ‘He agreed’). Note that in later models (e.g. Semino and Short 2004), an even more minimal category has been added which does not even relate which kind of speech act has occurred, and simply states that speech has taken place (e.g. ‘He spoke’). This category is labelled ‘narrative report of voice’ (NV). In many narrative contexts, the use of such minimal reports of speech are useful to progress the narrative, though most writers use them sparingly. Here is an example of NRSA used to avoid reporting a conversation where the detail would not be of great interest to the reader:

Here, the companion of Sherlock Holmes, Doctor Watson, narrates his actions in excusing himself from home and taking the latest ‘case’ to the great detective. However ‘shortly’ he explained where he was going to his wife, the reader would have been frustrated in having to read his words verbatim as it would delay the narrative proper.

The categories of speech presentation vary in their functional aspects in two related ways. At one end of the cline are those categories (NV and NRSA) that indicate the most intrusion by the narrator and, consequently, the most distance from the actual speech referred to. At the other end of the cline, free direct speech (FDS) is speech that is unmediated by a narrator. The examples given above are not unambiguous in their allocation to categories because for some of the categories, context is required in order to be clear which type of speech presentation is involved.

In particular, it is often difficult to identify text extracts as unambiguously FIS, particularly out of context. Thus, the following rhetorical question, out of context, may seem to the reader like the opinion of the narrator (or the text's producer):

There is, in fact, a clue here that this is free indirect speech and this is the use of the past tense form of the auxiliary verb, have, which if it were original direct speech or writing, would be in the present (‘Haven't people’ etc.). The context, as often with FIS, helps the reader to conclude that the ‘voice’ of this sentence is the character Isabel, as we can see in the following:

This short passage from The Sunday Philosophy Club (McCall Smith 2005: 73) includes a number of examples of speech presentation as we can see:

• IS: People who had metal piercing in their heads were asking for trouble, Grace had once said.
• NRSA: Isabel had asked why this should be so.
• FIS: Hadn't people always worn earrings, and got away with it?
• IS: Grace had replied that metal piercings attracted lightning, and that she had read of a heavily pierced man who had been struck dead in an electric storm while those around him, unpierced, had survived.

Note that the FIS here is within a longer passage of reported speech and this sets up the expectation that the FIS sentence is an elaboration of the speech act referred to in the previous sentence (‘Isabel asked why this should be so’). The labelling of NRSA assumes that the exact words used are not reflected in ‘why this should be so’, so it is possible that this is actually IS, if these were the words used. This whole passage in turn is part of an internal thought sequence as the character Isabel remembers the conversation being reported. The section begins as follows:

This demonstrates the importance in stylistic analysis of using more than one tool to describe the potential effect of textual features. Here, the use of a verb of perception, ‘felt’ (see section 3.2.2), with the reported thought make this a Category B(R) narrative with positive shading, in Simpson's modal grammar of point of view. We are seeing the narrative from Isabel's viewpoint, and being presented with her opinions and thoughts. At this point, the section analysed above occurs, as a kind of daydream as Isabel remembers the conversation with Grace which is followed by ‘The students exchanged glances, and Isabel turned away.’ The narrative then returns to the ‘present’.

Though often labelled ‘categories’, speech presentation types are similar to the so-called categories of Simpson's modal grammar or Halliday's transitivity system in that they are often not clear-cut at all but rather represent points of reference on a continuous, and often multi-dimensional, plane of meaning or function. There is, for instance, no precise formal way of identifying FIS or NRSA, and many analytical decisions rely on context. However, a category-based model of speech and thought presentation remains useful because it allows us to identify prototypical points on a cline. Furthermore, the Leech and Short model of discourse presentation has been developed most recently through large-scale corpus studies, which have necessitated clear decisions about the labelling of items in the corpus. In this respect, a category-based model is essential. We will return to this topic later in this section, but for now we will consider the equivalent types of thought presentation. There exists a similar range of ways of presenting others’ thoughts in narratives:

Like the speech presentation system, this range indicates the closest to the ‘actual’ thoughts at the bottom of the list, and the most narrator intervention at the top. However, the one important difference between the two lists is that speech can be heard by others, and thoughts cannot. The fictional narrator, even if s/he is a non-participating omniscient narrator, can in theory still tell the reader what was said verbatim, so that the reader has some confidence in the faithfulness of the words. The reporting of thoughts is always hypothetical, whoever is narrating, and we therefore have a different semantic ‘norm’ in the two systems, as Leech and Short point out, using figure 3.2 to help them.

Figure 3.2 Norms for speech and thought presentation (from Leech and Short 2007: 276)

The way Leech and Short explain the relevance of the different norms is as follows:

Thoughts, in general, are not verbally formulated, and so cannot be reported verbatim. Given that the norms for speech and thought presentation are at different points on the continuum, the different values of FIS and FIT can be naturally explained. FIS is a movement leftwards from the norm in figure [3.2] and is therefore interpreted as a movement towards authorial intervention, whereas FIT is seen as a move to the right and hence away from the author's most directly interpretative control and into the active mind of the character. Because the direct perception of someone else's thought is not possible, DT is perceived as more artificial than more indirect forms.

It has, of course, been a standard device of novelists to give the reader insights into the minds of the characters they create. The reporting of thoughts is also tied up with the features that Simpson's modal grammar discusses, so that one of the differences in style between the ‘hard-boiled’ detective fiction and the romantic novel is that the thoughts of the detective are less readily exposed to the reader, whereas the feelings and thoughts of the romantic heroine are typically on show for all to see. We saw earlier that McCall Smith's lady detective in Botswana does not conform to this stereotype, and neither does his Scottish philosopher-detective, Isabel Dalhousie, whose thoughts regularly bring his style into Simpson's Category B(R) with positive shading as we saw earlier. Here is another passage, where Isabel is thinking about her niece, Cat, and the boyfriend she has spurned:

Here, the thoughts are presented as follows:

• DT: What more could Cat want? she thought.
• FDT: Really! What else could a girl possibly require than a Scotsman who looked Mediterranean and could sing?
• NRTA: The answer came to her unbidden, like an awkward truth that nudges one at the wrong moment.
• FIT: Jamie was too nice. He had given Cat his whole attention – had fawned on her perhaps – and she had grown tired of that. We do not like those who are completely available, who make themselves over to us entirely. They crowd us out. They make us feel uneasy.

The whole passage, then, is an internal monologue with only the first sentence needing to indicate that this is thought, leaving the reader in no doubt that the thinking continues in the other sentences. The FDT section is labelled as such because of the exclamation (‘Really!’) which seems to indicate the ‘verbatim’ thought, even though, as Leech and Short point out, thoughts are not usually couched in language. The NRTA sentence requires us to consider the transitivity choice that has been made here, as the Actor is inanimate (‘the answer’) and the result is a material event, with Isabel at the mercy of the inanimate – in this case abstract – world.

The final section, labelled FIT, is Isabel's viewpoint, since it follows the other thought processes with no intervening narrative. The generic sentences (‘We do not like’ etc.) confirm the Category B status of this passage, and the use of past tense verbs (‘was’ and ‘had’) makes this indirect, as they would be present tense if the thoughts were being presented ‘verbatim’. McCall Smith presents his characters’ viewpoints using all the available speech and thought presentation techniques. As we can see here, the apparently unmediated presentation of thought, by the use of FIT, is one way he may cause readers to feel that they are accessing the characters’ thoughts directly.

3.2.5 Other functional categories

In this section, we will introduce two other ways in which functional concepts, realised by a range of forms, may influence the meaning likely to be derived from a text. What these categories share with transitivity, modality and discourse presentation is that they are predominantly conceptual and not formal, though each has a prototypical realisation, like the categories introduced in previous sections.

The first of these local textual functions is the creation of opposites in context. This phenomenon has only recently been fully explored (see Davies 2008 and Jeffries 2010a), though it has been mentioned incidentally in earlier work (Jeffries 1998) and was used as a tool of analysis in Jeffries (2007b):

This textual creation of what are normally seen as context-free lexical semantic meanings is just one of the ways in which we can see the symbiotic relationship between code and usage or between langue and parole.

(Jeffries 2007b: 102)

The textually-created opposites, then, draw on what the reader already knows about relationships of opposition, including the fact that there are different types of oppositional relationship. These include as a minimum³ the mutually exclusive complementaries (alive/dead), the mutually dependent converses (buy/sell), the gradable antonyms (dark/light) and the directional or reversive opposites (pack/unpack). The textually created opposites are usually triggered by one of a number of structural features, most prototypically the X-not-Y structure, using negation as a way of setting two concepts in opposition to each other. Other triggers include those which depend on parallel structures to contrast items (e.g. He wanted X. She wanted Y) and those that are explicit about the fact that they are being contrasted (X in contrast to Y). A very much more comprehensive list of potential triggers of opposition can be found in Davies (2008), though it should be pointed out that, as with modality, it may be impossible to compile a complete list, as this is a fuzzy category with changing boundaries.

Let us consider some of the possibilities for literary effect in the creation of opposites by looking at a poem by Philip Larkin which is built upon created opposition:

The poem opens with a negated opposite, where ‘Ireland’ is opposed to ‘home’, and this is followed by the juxtaposition of ‘rebuff’ with ‘welcome’, which are semantically opposed, though not conventional opposites. Here, though, it is the very rebuff which creates the welcome, since the whole poem's argument is predicated upon the notion that we only feel that we belong somewhere if other places are strange to us. So, the narrator's feeling of strangeness is important to his awareness of home. In expanding upon this idea, Larkin describes the range of scenes that seemed strange to him in Ireland, adding that they ‘went to prove me separate, not unworkable’. Here, then, there is another negated opposition created between ‘separate’ and ‘unworkable’, where the feeling of being a foreigner is seen as evidence of difference, but not a chasm that cannot be bridged. Negated opposites are usually complementary, so that there is no conceptual intermediate position between ‘separate’ and ‘unworkable’: you have to be one or the other in Larkin's world. Nevertheless, Larkin has played with the conventional idea that strangeness creates unbridgeable distance between people. Here, instead, we see that you can be separate from people without being alien to them. This, in effect, introduces a middle point between unworkable and close, which is separateness, and this intermediate concept in human relations is the one that allows us to feel a sense of belonging at home. In the final line, Larkin's ‘deictic centre’ (see section 6.3) has returned to England, where he cannot refuse to join in with the customs, as it is his own culture. He ends the poem by reiterating his main theme, using here and elsewhere, rather than the conventional here/there, to emphasise his rootedness in England. The essential link between home and elsewhere is maintained, however, by his existence being dependent on (‘underwrites’) the other place(s).

The creation of local textual opposites is a phenomenon that, once noticed, is very common in literary, but also in other, texts. Volvo car adverts, for example, have recently been using the line ‘less emissions,⁴ not less style’, which implies that there are some circumstances in which style and low emissions are incompatible, in other words, are complementary opposites, where you can only have one or the other. This ‘trick’ of advertising is ubiquitous; the argument that you can have value and quality for example is one that many adverts try to make.

The prototypical created opposition is the negated opposition, where a positive and a negative are placed next to each other to create contrast. This means that created opposites are often linked to the final textual function to be discussed in this chapter: negation. Like the other systems introduced here, negation has a prototypical form, which is the negator, not, modifying the verb (e.g. hasn’t, didn’t), although the premodification of nouns by no is also reasonably prototypical (e.g. no money, no food). However, like the other systems in this chapter, negation also takes a number of other forms, including morphological ones (e.g. de-humidify, unreasonable), semantic ones (e.g. lack, fail) and like modality (the shrug) negation can be delivered paralinguistically by shaking the head or waving the hands in certain (probably culturally-specific) ways.

The discoursal significance of negation, and the reason that it is closely tied in to opposition, is that it simultaneously evokes the negated scene, at the same time as denying its existence. So, the opening comment in Larkin's poem above, ‘since it was not home’, both informs the reader that Ireland is not the narrator's home and also conjures up another place that is indeed his home. Similarly, the following sentence from Kate Atkinson's novel, Case Histories (Atkinson 2004: 35), conjures up the very scenes that Rosemary is rejecting in relation to her academic husband:

This sentence negates two possible scenarios in which Rosemary's own father was habitually to be found, and her comparison of her earthy father with her distant intellectual husband is made more effective by the technique of negation, so that both Rosemary and the reader are invited to envisage the very scenes she is supposed not to be able to imagine, and reject them as unreal. This is taken literally in some advertising where the very thing that is negated is shown happening, or shown not happening. Thus, an advertisement for Drench spring water⁵ has the slogan Dehydrated Brains Don't Perform Well, which appears after a short video clip of a puppet called Brains (from the popular 1960s marionette series Thunderbirds) singing (well) on stage and then falling over, presumably when his brain dehydrates. The negated proposition, then, helps us to envisage both scenarios. The converse is not normally true of positive propositions, such as ‘Jackson had arrested a jeweller once’ (Atkinson 2004: 74), where the scenario in which Jackson had not arrested a jeweller is not clearly relevant to the narrative. ‘Nicola was not a great cook, apparently’ (Atkinson 2004: 77), by contrast, does conjure up the kinds of people who are good at cooking, and Nicola is rejected as one of them.

3.3 Summary and conclusions

This chapter introduced a set of stylistic tools of analysis which go beyond the early stylistic analysis in two ways. First, the systems of functional meaning which operate at levels of analysis higher than the sentence have been discussed. These include transitivity analysis and modal grammar, which can be used to comment on individual clauses and also whole texts or text extracts. Similarly, the analysis of cohesion is predicated on the idea that there are structures and functions which work at text level. The final two analytical systems, opposition and negation, work both locally in texts and also more extensively throughout a text.

The second way in which these tools of analysis work differently to those studied in Chapter 2 is that they are all concept-driven. That is, they each address large-scale concepts, such as certainty (modality), representation of events and processes (transitivity) and the more transparent concepts of cohesion, negation and opposition. In each case, there is a prototypical form which delivers the concept most recognisably, and further ways of realising the same kinds of idea, which shade out towards the paralinguistic, the pragmatic and the contextual at the edges of what seem to be ‘fuzzy’ categories at best, if they are categories at all.

It has been suggested that, perhaps, one way to conceptualise these systems is as a set of reference points to which actual textual examples will approximate, and which help us to understand texts as discourse without constraining us to label (except for the sake of computer analysis) unambiguously in every case.

Further reading

Readers wishing to pursue the debate about the value and integrity of stylistics may trace the argument through from Fish (1981), who claims that it is ill-founded and illogical, to Simpson (1993: 111–16), whose defence of stylistics from charges of ‘interpretative positivism’ depends on the notion of ‘lowest common denominator’ coupled with contextual variation, and to Toolan, who responds from the point of view of ‘integrationalism’, which attempts to recognise the fluidity of textual meaning throughout stylistic analytical practice and detects at least some literary critical value in the techniques and methods of stylistics.

General discussions of the kind of combined formal/functional tools of analysis described in this chapter can be found in Jeffries (2007b: 12–15) and Jeffries (2010a). Halliday (1994) is the source of the transitivity and modality systems, though readers may find Butt et al. (2000) or Halliday and Matthiessen (2004) an easier introduction to Halliday's ideas. Similarly, Simpson's (1993) book on ideology and point of view in literature is accessible and the models of transitivity and modality given there are those used in Jeffries (2007b and 2010a) and the basis of the discussion in this chapter too.

Cohesion is best explained by Halliday and Hasan (1976), though many introductory linguistics books have introductory explanations of the kind also found here. See, for example, Jeffries (2006: 183–7).

Discourse presentation has been developed as a model and analytical tool primarily by Leech and Short (2007: 255–81), Short (1996: 288–324) and Semino and Short (2004), though accessible accounts can also be found in Simpson (1993: Chapter 2) and Toolan 1988: Chapter 4). Applications of the speech and thought model to spoken data can be found in McIntyre et al. (2004).

The other tools of analysis presented here, opposition-creation and negation, are less widely covered in stylistic literature at the moment, though opposition is the subject of a monograph (Jeffries 2010a) and a PhD thesis (Davies 2008) and can be seen in practice in Davies (2007) and Jeffries (2007b: 109–19). Negation as a stylistic tool is developed by Hidalgo-Downing (2000), Nørgaard (2007) and Nahajec (2009), and is explained as a tool of critical discourse analysis in Jeffries (2010a).