4 Discourse and context II

4.1 Stylistics and pragmatics

Early work in stylistics focused primarily on the analysis of the formal linguistic elements of texts – for example, grammatical forms, phonological features and propositional meanings (see Chapter 2). It is no surprise that such work also focused mainly on the analysis of poems, since such texts are short (making it possible for the stylistician to analyse a complete text) and relatively straightforward in terms of discourse structure. That is, many poems have a single-tier discourse architecture in which the poet addresses the reader directly (Short 1996: 38). This makes a stylistic analysis of such texts relatively straightforward (at least in methodological terms), since it involves identifying stylistic effects at just one discourse level. This is considerably more straightforward than trying to identify the stylistic effects in a text with multiple discourse levels, such as a novel, which involves an address from the author to the reader, embedded in which is an address from a narrator to a narratee, embedded in which are the characters in the fictional world addressing each other. In texts composed of multiple discourse levels, the task for the analyst is considerably more difficult, since the analysis necessitates identifying and isolating stylistic effects at each of the text's constituent discourse levels. Also, such texts tend to be longer, making it unfeasible to produce analyses of complete texts (though corpus stylistics has alleviated this problem to a certain extent; we will discuss this fully in Chapter 7). These methodological issues also explain the comparative neglect of drama by early stylisticians. Dramatic texts, too, have complex discourse architectures, as well as the added complexity that comes if we try to take account of associated performances of the text. (The analysis of the multimodal elements of drama is something stylistics has only just begun to deal with; see, for example, McIntyre 2008.) Stylistics in its infancy was simply not equipped to deal with such text-types, and one reason for this was the lack of appropriate tools for analysing discursive interaction. Only with the advent of pragmatics were stylisticians able to deal with texts – both fiction and non-fiction – which involved (re)presented interaction. The eclectic nature of stylistics meant that it was quick to add pragmatic and sociolinguistic approaches to its analytical toolkit. Our aim in this chapter is to outline some of the key analytical frameworks from pragmatics that stylistics has utilised, and to show how such pragmatic approaches can shed interpretative light on texts that incorporate discursive interaction in the form of (re)presented conversation.

4.2 Interaction in discourse

During the 1970s and 1980s, advances in pragmatics (broadly speaking, the study of how context affects meaning) increased the scope of what stylistics was able to achieve. Previously, while the techniques of grammatical analysis had been available to help stylisticians uncover such aspects of text structure as viewpoint, the tools were not available to reveal the source of interpretative effects deriving from dialogue. For this reason, work on the stylistics of drama to some extent stalled, especially by comparison with work on the stylistics of prose and poetry. But prose, too, often incorporates dialogue (whether presented directly or otherwise), and so, in order to produce stylistic analyses of prose texts that take account of all aspects of discourse structure, the techniques of pragmatics (and other methods of dialogue analysis) are also needed. We will begin by examining the techniques available for the analysis of conversation structure before going on to consider pragmatic aspects of dialogue.

4.2.1 The structure of interactive talk

The notion of stylistics as something of a ‘magpie’ discipline is amply demonstrated in the extent to which it has borrowed techniques from Conversation Analysis (CA) and applied them in the analysis of dialogue in both fiction and non-fiction texts. The propensity of stylistics for adapting techniques from other disciplines is also demonstrated in the way in which it has taken CA methods and modified them for its own ends, as we shall see.

Conversation Analysis is an ethnomethodological approach to analysing spoken language, pioneered by Harvey Sacks in the 1970s (see Sacks 1995). Conceived as a method of describing and analysing interactive talk, it aims to demonstrate the structures that underlie all conversation. Stylisticians such as Burton (1980) and Mandala (2007) have since demonstrated how such structures are also apparent in fictional dialogue, showing CA to be an analytical method also appropriate for stylistic analysis. Key to CA is the notion of a turn. Turns are composed of one or more turn-constructional units, which are themselves defined as a complete grammatical unit, such as a sentence or a clause (or, in speech, an utterance; see Biber et al. 1999 for a discussion of the units of spoken discourse). In prototypical conversation, a turn-taking system is always in place, wherein participants take it in turns to hold the floor (defined as the right to speak and have other participants in the talk listen to you). The turn-taking system is composed of two components: (i) the turn allocational component and (ii) the turn constructional component.

The turn allocational component regulates turn change and assumes that only one speaker may speak at a time. The point in a conversation at which a change of speaker occurs is known as a transition relevance place (TRP) and at such places several options become available. The current speaker may select the next speaker (by, for example, asking a question of another participant), the next speaker may self-select (perhaps by rebutting the previous speaker's point) or the turn may lapse. If a lapse occurs, the current speaker may incorporate that lapse as a pause and continue until the next TRP. A turn-overlap at any other point than a TRP constitutes an interruption.

The turn constructional component deals with the length and linguistic complexity of the turn, and turn-length is locally managed. For instance, if you are telling a story and someone begins to talk over you, you might raise your voice as a paralinguistic signal that you need more time to complete your turn. Alternatively, you might signal this metalinguistically by saying, for instance, ‘Let me just finish.’ (Good examples of the need for management of the turn constructional component can easily be found in political discussion programmes and political interviews.)

But CA doesn't just concentrate on the production of talk. It examines how talk is interactive, and it does this by looking at how turns relate to each other. Turn-taking in conversation comes about in part because turns often call for another turn in response. Schegloff and Sacks (1974) describe such turns as adjacency pairs. An adjacency pair consists of a first part and a second part, and the second part of an adjacency pair may be preferred or dispreferred. An example will make this clear. A greeting prototypically consists of two parts. The first part involves a salutation from one speaker (for example, ‘Good morning’), and the prototypical response to this (the preferred response) would be a salutation in return. A dispreferred response would be to respond with, for example, an invective (‘fuck off!’) or to ignore the greeting and say nothing at all. Person (1999) summarises some typical adjacency pairs as shown in Table 4.1.

Table 4.1 Adjacency pairs (Person 1999: 43)

In stylistic terms, a dispreferred response is foregrounded by external deviation and may be regarded as interpretatively loaded. Consider the following example from Dennis Potter's television drama Pennies from Heaven:

Here, Joan's question constitutes the first part of an adjacency pair, the preferred second part to which would be an answer. However, by repeating Joan's question and (as the stage directions explain) mimicking Joan, Arthur gives a dispreferred second part. The usefulness of CA to stylistics is in the replicability of its method: by the terms of CA, any analyst would identify Arthur's response as dispreferred and thus foregrounded.

The next stage in the analytical process is, of course, to interpret the significance of this element of foregrounding, and here is where stylistics differs from CA. Conversation analysts would claim that we should not speculate on what a speaker means by a particular utterance, since we can never know for certain whether our speculation is correct. We must, instead, concentrate our efforts simply on describing what speakers do within a particular conversation. In so doing, conversation analysts specifically exclude context in contributing to the meaning of utterances, and their failure to consider speaker-meaning seems to us an unremittingly reductive stance to take. Even natural scientists need to interpret the results obtained from experiments, and it may well be the case that other scientists might disagree with their interpretation. Nonetheless, the important thing is that the analytical procedure is such that the experiment can be replicated and the same results obtained. The same principle applies to text analysis. To stop short of interpreting the effects – or meanings – of a particular utterance or element of text structure is to provide an incomplete analysis. With regard to example 1, we might assume Arthur's utterance to betray his annoyance at Joan's question. (We will consider the means by which we might legitimately make such inferences in 4.2.2.)

The necessity of taking into account the implicatures that may be generated by the second part of an adjacency pair can be seen when we consider those that seem, on the surface, to be preferred responses but which in practice are not quite so straightforward. Imagine, for instance, a conversation between two friends:

In this example we have a question followed by an answer, and in terms of adjacency pairs we might, at a surface level, say that the answer constitutes a preferred second part. If we are to suggest that it does not, we have to do so on the basis that the question by speaker A requires an answer which is either affirmative or negative. However, to do this means speculating on the meaning of speaker A's utterance, something that conversation analysts would claim to avoid. It would seem, then, that it is impossible to analyse conversation accurately without some attempt to interpret speaker-meaning. For this reason, while stylistics utilises the descriptive procedures of CA, it disregards the methodological limitations of CA and prefers instead to incorporate analytical techniques from pragmatics, as we shall see in section 4.2.2.

Despite the problems with CA, then, it still has a part to play in the stylistic analysis of dialogue, especially when used in conjunction with pragmatics. For example, the relative power of speakers in a conversation is often reflected in the turn constructional components of the conversation. Short (1996: 219–20) provides a detailed account of the conversational behaviour of powerful and powerless speakers, explaining, for instance, that powerful speakers tend to dominate conversations and that this is reflected in their relative turn lengths. Of course, there are numerous contextual factors that it is necessary to take account of, as Short points out. First, the notion of power incorporates numerous different types. A speaker may be physically weaker than other conversational participants but may be imbued with substantial power as a result of having a particular institutional role (consider the example of a monarch or president's role as titular head of the armed forces). Second, the context of a conversation may place restrictions on the prototypical linguistic behaviour of participants. Job interviews, for example, usually involve the less powerful participants (the candidates) speaking substantially more than their interviewers. We will consider other structural elements of conversational behaviour in an extended analysis in section 4.3, after we have considered in more detail the pragmatic aspects of dialogue analysis.

4.2.2 From presupposition to implicature

We saw in the previous section the importance of taking into account contextual factors when analysing dialogue. In this section we will examine analytical procedures for uncovering contextual meaning.

Sometimes, meanings are inherent in the structure of what we say. For example, saying ‘Lee Harvey Oswald assassinated John F. Kennedy’ entails that ‘John F. Kennedy died’. An entailment is what follows from the propositions that are made in an utterance. Entailment, then, is a logical concept, not a pragmatic one. There cannot be any context in the real world where if the first assertion is true the second one is not. Entailment, therefore, relates to the utterance rather than to the speaker. Entailment is different from presupposition. The sentence ‘Lee Harvey Oswald assassinated John F. Kennedy’ presupposes the existence of a person called Lee Harvey Oswald and a person called John F. Kennedy. The difference between entailment and presupposition can be seen if we negate the sentence. If we say ‘Lee Harvey Oswald did not assassinate John F. Kennedy’, the existential presuppositions still hold but the entailment does not. If Oswald did not assassinate Kennedy then Kennedy cannot have died as a result. Presuppositions are also derived from the structural elements of utterances. (Levinson 1983 provides an extensive list of presupposition triggers which includes, for example, change-of-state verbs, comparatives, iteratives, etc.)

Sometimes, however, the full meaning of an utterance cannot be derived from its structural elements alone. For example, saying ‘The Vice-Chancellor of the University is a man of profound intellectual foresight’ presupposes the existence of a particular Vice-Chancellor but it does not indicate whether the speaker is sincere or sarcastic in what they say. To understand this we need to consider the notion of implicature.

Grice (1975) distinguishes between what he calls conventional and conversational implicature. Conventional implicature is a type of pragmatic presupposition and is realised through the use of particular lexical items. The conjunction but is one such item, as in the following example:

Here the coordinating conjunction gives rise to the conventional implicature that old houses do not usually have nice atmospheres. In a real-life example (4) taken from a snooker commentary, the conventional implicature arising from the conjunction is unintentionally humorous, suggesting as it does that fine snooker players do not normally come from Wales:

The linguistic triggers of conventional implicatures are restricted to a small set of lexemes. Conversational implicature, on the other hand, is much more prevalent and arises out of the pragmatic behaviour of speakers. The theory which specifies how this occurs was originally proposed by the ‘ordinary language’ philosopher, Paul Grice (Grice 1975; see also Chapman 2005, 2006), and postulates a number of commonly held assumptions shared by participants in a conversation. Grice explains his so-called Cooperative Principle (CP) as being:

[…] a rough general principle which participants will be expected (ceteris paribus) to observe, namely: Make your conversational contribution such as is required at the stage at which it occurs, by the accepted purpose or direction of the talk exchange in which you are engaged.

(Grice 1975: 45)

Grice formulates his Cooperative Principle as a series of four sub-maxims, of quantity, quality, relation and manner. These may be summarised as follows:

• Make your contribution as informative as is required (for the current purpose of the exchange).
• Do not make your contribution more informative than is required.

• Do not say what you believe to be false.
• Do not say that for which you lack adequate evidence.

• Be relevant.

• Be perspicuous.
• Avoid obscurity of expression.
• Avoid ambiguity.
• Be brief (avoid unnecessary prolixity).
• Be orderly.

Although formulated as imperatives, Grice's maxims are not instructions to speakers but are best understood as encapsulating the assumptions that we prototypically hold when we engage in conversation. We expect, for instance, that if we ask a question of an interlocutor, we will receive a relevant answer that is truthful, clear and no more or less informative than is required. Nonetheless, in practice it is common for speakers not to observe these maxims, and there are a variety of means by which they may be broken. These can give rise to a number of different pragmatic effects, all of which may have stylistic import. We can summarise the means by which speakers can break the maxims as follows.

Flouting

When speakers intentionally fail to observe a maxim, and do so ostentatiously to the extent that it is clear to any interlocutors that the maxim is indeed being broken, then, according to Grice, he or she is flouting that maxim. Flouting maxims draws attention to an undercurrent of meaning – and this is what Grice refers to as conversational implicature. The following example from prose fiction demonstrates this:

Perhaps the most obvious instance of flouting in example 5 is Bertie's statement that his cousin is ‘the King of the Underworld’. Although a comic novel, the fictional world is one that is akin to the real world and so Bertie's claim that his cousin is the ‘King of the Underworld’ is clearly not the case, since this is a supernatural being that has no existence in the real world. Bertie's statement, then, is a flout of the maxim of quality. It is so obviously not true, that Bertie's interlocutor (and we as readers) are forced to look for an alternative meaning in order to make sense of Bertie's remark. This is likely to be that Bertie's cousin shares some of the characteristics of the fabled ‘King of the Underworld’; perhaps he is frightening, sly, cunning, unpleasant, etc. This, then, is the conversational implicature arising from Bertie's flout. We might also note that this piece of dialogue also constitutes a flout within the discourse world (see Chapter 6) of the novelist, P. G. Wodehouse. At this discourse level, flouting the maxim gives rise to a humorous effect that would not be present had Wodehouse simply observed the Cooperative Principle by having Bertie state clearly and concisely that his cousin was an unpleasant character.

Violating

‘Violating’ describes the situation wherein a speaker breaks a maxim on purpose and intends for his or her interlocutor not to notice this. Violations of the maxim of quality are notoriously difficult to spot in real life data, for the simple reason that an effective violation of this maxim is one that is not noticeable. One notable example, however, is the former U.S. President Bill Clinton's claim that he did not have sex with the White House intern, Monica Lewinsky. Following allegations that he did, Clinton said in a statement, ‘I did not have sexual relations with that woman.’ The status of this remark as a violation of the maxim of quality was revealed by Clinton's later admission that ‘I did have sexual relations with Miss Lewinsky, and I am deeply, deeply sorry for the pain I have caused my family and friends.’ Politicians can often be observed violating the maxims of manner, relation and quantity as a means of deflecting attention away from the fact that they are not directly answering the question posed to them. Successful violations do not generate conversational implicatures.

Infringing

‘Infringing’ refers to what happens in those situations where a speaker fails to observe a maxim, though with no intention to deceive or to generate an implicature. For example, infringement may occur if a speaker is tired, drunk or cognitively impaired in some other way. A fictional instance of the infringement of the maxim of quantity can be seen in the following extract from a courtroom scene in the film A Fish Called Wanda:

Here, the judge's last turn is clearly not intended to create an implicature, and cannot therefore constitute a flout. Nor is it covert and so it cannot be a violation. It would seem that the excitement of the situation causes the judge simply to infringe the maxim of quantity (by repeating the phrase ‘Clear the court!’) in his eagerness to have George removed from the courtroom.

Opting out

‘Opting out’ refers to the practice of ostentatiously refusing to be bound by the Cooperative Principle. For example, a politician faced with a potentially incriminating question may prefer to respond with ‘No comment’. Of course, one of the problems with such a strategy is that interlocutors will often interpret such an opt-out as a violation and assume that the speaker is deliberately withholding information and/or being unclear. This can lead to instances such as the one below, in which the former British Prime Minister goes to considerable lengths to emphasise that he is neither making implicatures nor withholding information as a result of making no comment:

A more light-hearted example, highlighting the fact that attempting to opt out of the CP can lead to unintentional implicatures, can be seen in the example below from Richard Curtis's film comedy, Notting Hill:

William's response, ‘No comment’, in turn 2 constitutes an attempt to opt out of the Cooperative Principle, though Max's statement in turn 3 exemplifies the fact that when speakers attempt to do this, it is difficult for their addressees to avoid making an inference on the basis of it.

In addition to breaking the maxims, there are instances in which the Cooperative Principle may be suspended – i.e. occasions where there is no assumption that speakers will observe the Cooperative Principle at all. In interview situations, for instance, there is usually no assumption that candidates will observe the maxim of quantity, since one of the aims of this particular activity type is to convey as much about oneself as possible (see Culpeper and McIntyre 2010 for a full discussion of the Cooperative Principle in relation to the interview activity type). It is also the case that participants in a conversation will often have different conceptualisations of what it means to be truthful, relevant, concise, etc. Harris (1984) takes this truism as a starting point for her development of the notion of paradigms of reality, explaining that interlocutors often begin conversations from radically different perspectives on reality and that these different ‘reality paradigms’ sometimes clash. Archer (2002) develops Harris's work and suggests that the notion of reality paradigms can explain why an utterance may be interpreted differently by different participants in a conversation – i.e. why one utterance may cause different interlocutors to draw different inferences from it. The different reality paradigms from which characters interpret each other's utterances can give rise to a number of effects in drama, not least humour, as can be seen in the following extract from John Cleese and Connie Booth's celebrated sitcom, Fawlty Towers. In this extract, Alan is a guest staying at the hotel with his girlfriend. Basil, the hotel owner, has previously refused to give Alan a double-room because Alan and his partner are not married. Part of Basil's reality paradigm is that sex outside marriage is wrong. Because Basil has this world view, it causes him to draw an inference from what Alan says when, in fact, Alan is not attempting to create an implicature at all:

Basil's reality paradigm causes him to interpret Alan's first turn as flouting the maxim of quantity to generate an implicature along the lines of ‘Do you know where I can buy some condoms?’ In effect, Basil's reality paradigm causes him to see an implicature in Alan's first turn, where there is, in fact, none.

4.2.3 Politeness and impoliteness

As we have seen, there is strong evidence from pragmatics for some kind of Cooperative Principle being at work when we communicate with others. From a stylistic perspective, however, our concern is with the interpretative effects generated by the linguistic choices made by speakers (and writers), and we can gain a better understanding of these if we consider the reasons why participants in a conversation choose to observe or ignore the Cooperative Principle. One reason why participants often flout or violate the conversational maxims is for the sake of politeness (and its opposite, impoliteness), and this is an area that has received considerable attention from linguists.

There are a number of different conceptualisations of linguistic politeness, though of these the most widely adopted has been the model proposed by Brown and Levinson (1987). This is based on the concept of face and the notion that everyone has both positive face needs (the desire to be liked and approved of) and negative face needs (the desire to be able to go about one's business unimpeded). Any affront to these face needs is termed a face threatening act (FTA), and in order to mitigate FTAs, speakers may choose to use a variety of politeness strategies designed to appeal to their addressee's positive and/or negative face needs. The extent of the mitigating strategies used by speakers depends on the size of the FTA. A relatively innocuous FTA (asking if someone can spare the time to discuss something, for example) may not require extensive politeness strategies to be used, while a more significant FTA (a manager informing an employee that he or she is to be made redundant) will require substantial mitigation. Examples of these scenarios can be seen in the following extract from the screenplay of Sam Mendes's film American Beauty:

In turn 1, Brad's request for a minute of Lester's time constitutes a face threatening act, albeit a small one. In acknowledgement of the fact that his desire to speak to Lester is in some sense a threat to Lester's negative face (i.e. his desire to be unimpeded), Brad frames the FTA as an interrogative, a more indirect structure than, say, an imperative (‘I want to talk to you’). This is coupled with Brad's friendly greeting (constituting an appeal to Lester's positive face) and the informality suggested by the elision of the verb have in his second sentence. Such strategies serve to emphasise the insignificance of the FTA. Of course, given that this is not a substantial FTA, and that Brad has greater institutional power than Lester, Brad could perhaps legitimately have used fewer politeness strategies (e.g. ‘Lester, I want to talk to you’). Note, though, that this is a precursor to a much bigger FTA, namely Brad warning Lester that he is likely to be made redundant. Hence, in turn 3, Brad continues to use a number of politeness strategies to mitigate this significantly larger FTA. For instance, he appeals to Lester's intelligence (and hence, his positive face) by acknowledging that Lester is likely to understand the reasons behind the need ‘to cut corners’.

Brown and Levinson (1987: 103–227) provide an extensive summary of the variety of strategies open to speakers for mitigating both positive and negative face threats which, for reasons of space, we cannot consider here in much detail. Instead, we will simply note that Brown and Levinson divide these specific tactics among what they term five ‘super-strategies’: (i) Perform the FTA bald, on record; (ii) Perform the FTA on record using positive politeness; (iii) Perform the FTA on record using negative politeness; (iv) Perform the FTA off record; and (v) Don't perform the FTA. To begin with super-strategy (i), bald on-record politeness simply means that the speaker's utterance is in accordance with Grice's Cooperative Principle; that is, it is maximally efficient in communicative terms. Such an approach may be warranted by the context in which an utterance occurs. As an example, a Private may exclaim ‘Look out!’ to a superior ranking officer when advancing under fire if he or she notices that the officer in question is in immediate physical danger. The utterance is maximally efficient in terms of observing the maxims of quality, quantity, relation and manner, and the necessity of issuing a warning quickly allows the soldier to disregard the usual need to use the conventionalised polite address form.

In drama, politeness strategies can work as elements of characterisation. In Alan Bennett's The Lady in the Van, for example, the character of Miss Shepherd frequently uses the bald on-record strategy, even in circumstances where politeness would be the norm. Inevitably this characterises her as someone who is rude and cares little about social niceties:

Miss Shepherd's answer is potentially damaging to the social worker's positive face, though she does nothing to mitigate it, preferring instead to answer in accordance with the Gricean maxims.

In cases where mitigation of the FTA is necessary, super-strategies (ii) to (iv) may be employed.

Super-strategies (ii) and (iii) call for the FTA to be performed on record. This simply means that the FTA should be acknowledged and appropriate tactics used to mitigate it. Performing an FTA on record using positive politeness means using strategies designed to redress the positive face threat to the hearer. Examples might include intensifying one's interest in the addressee, seeking agreement with him or her and avoiding disagreement, using in-group identity markers, joking and making small talk. In the following extract from The Lady in the Van, the social worker is responding to an irritable comment from Alan Bennett:

The social worker's choice of words appears to be governed by an awareness that simply telling Alan that he is behaving in a hostile manner would be threatening to his positive face. To mitigate the FTA the social worker uses a number of strategies: (1) She uses Alan's first name to indicate a close relationship to him; (2) she refers not to what Alan is doing but to what she is feeling; (3) she ‘distances’ the hostility from Alan and herself by referring to it using the distal empathetic deictic term ‘that’ (see Chapter 6 for an explanation of deixis).

Performing an FTA on record using negative politeness involves using strategies designed to reduce the negative face threat to the addressee. Here, speakers may be indirect, they might hedge, be pessimistic about the outcome of whatever request/statement they are making, apologise, or impersonalise the FTA through the use of passivisation or the use of impersonal pronouns to refer to the speaker and hearer. The next example, again from The Lady in the Van, is a rare instance of Miss Shepherd being polite. Here, she is dictating a request to the College of Cardinals in Rome:

Disregarding Miss Shepherd's tenuous grip on reality, her request is one which, if acted upon, would create difficulties for the College of Cardinals. Recognising this, she mitigates the FTA by using indirectness (framing her suggestion as an interrogative), by downplaying her own status (via the adverb ‘humbly’) and by hedging (using ‘possibly’ and ‘for instance’).

Super-strategy (iv), ‘Perform the FTA off record’, means performing the FTA using an indirect illocutionary act (illocutionary force is discussed in more detail in section 4.2.4). Put simply, this means performing the FTA in such a way that the speaker's utterance might be interpreted in more than one way, allowing the speaker the get-out clause of denying having performed an FTA at all if offence is taken by the hearer. For instance, a speaker wanting a favour from a colleague might say, ‘It's really inconvenient that I have to work an evening shift today.’ If the colleague replies along the lines of, ‘Well I can't help you, I’m busy myself you know!’, the indirectness of the first speaker's utterance allows him or her to deny the initial negative face threat by claiming that he or she was simply making a statement as opposed to a request.

Finally, with regard to super-strategy (v), Brown and Levinson note that in some cases an action may be deemed so potentially face-threatening that it is simply not worth performing the FTA. This evaluation of the size of the FTA is what determines the choice between the five super-strategies, along with other contextual factors such as the relative power differential between the speaker and hearer. ‘Small’ FTAs may be performed bald on record, while weightier ones may require on-record positive or negative politeness, or even an off-record strategy.

It should be apparent from the above explanation of Brown and Levinson's theory of politeness that utilising many of their proposed super-strategies involves breaking the Gricean Cooperative Principle in some way. A clear connection can be seen, then, between these two theories of conversational exchange. What is also the case is that the Cooperative Principle can be broken in the pursuit of less harmonious exchanges: that is, speakers and writers may choose to flout or violate the Gricean maxims in order to be impolite.

It is generally agreed by most scholars working in this area that the Brown and Levinson theory of politeness cannot adequately account for instances of impolite conversation. Nonetheless, a number of these scholars have proposed that impoliteness may be explained using a face-based model similar to that proposed by Brown and Levinson to account for polite verbal behaviour. Culpeper (1996), for example, suggests that if impoliteness is viewed as being the polar opposite of politeness, then strategies for achieving impolite exchanges must be in opposition to the strategies typically used to maintain conversational harmony. In his view (and in that of Lachenicht 1980) impoliteness may be defined as an attack on the positive and/or negative face needs of an addressee. In explaining how this might work, Culpeper identifies five super-strategies for attacking face. These are: (i) Perform the FTA using bald on-record impoliteness; (ii) Perform the FTA using positive impoliteness; (iii) Perform the FTA using negative impoliteness; (iv) Perform the FTA using sarcasm or mock politeness; and (v) Withhold politeness.

With regard to super-strategy (i), Culpeper (1996) points out that while the bald on-record politeness super-strategy is usually employed in situations where the context is such that little face damage is likely to result, the bald on-record impoliteness super-strategy tends to be used in instances where the converse is true. Super-strategy (ii) governs the use of tactics designed to attack the positive face needs of the addressee (for example, ignoring the addressee, using inappropriate address forms, using taboo language, avoiding agreement and seeking disagreement), as in this example from Quentin Tarantino's screenplay Reservoir Dogs:

Super-strategy (iii), obviously enough, covers the use of tactics designed to attack negative face (for instance, hindering, being condescending of the addressee's position, invading his or her space either literally or metaphorically), as in this example from later on in the Reservoir Dogs screenplay:

Mr Blonde's reference to Mr White as ‘little doggie’ is condescending in the extreme, especially considering the implicature (via a flout of the maxim of relation) that Mr White is all talk and lacks the courage to attack Mr Blonde physically.

Super-strategy (iv) refers to the performance of an FTA using insincere politeness forms, as in this response from Mr Pink to Mr White's revelation that he has told a fellow criminal his real name (despite being expressly instructed not to do this) because the man in question had been shot and was scared:

Finally, super-strategy (v) refers to the withholding of politeness when it is expected, by keeping silent.

There are, of course, issues with Culpeper's approach to impoliteness, as, indeed, there are issues with any theory, model or analytical framework. Bousfield (2007a: 63), for instance, questions the necessity of the bald on-record impoliteness strategy, arguing that any bald on-record utterance is threatening to face to some extent, and that bald on-record tactics might be better thought of as being subsumed within both the positive and negative impoliteness super-strategies. Bousfield (2007a: 91) also raises the potential problem of the open-ended nature of Culpeper's list of the tactics that might be employed in pursuit of any of the five super-strategies for aggravating face. If there is no theoretical restriction on the number or type of tactics that might be subsumed within each of the super-strategies, then the approach itself is in danger of being unfalsifiable. Culpeper counters this problem to some extent in his more recent work (Culpeper 2005) by integrating Spencer-Oatey's (2002) notion of rapport-management into his approach in order to take account of the culturally and contextually managed aspects of face. Despite the issues, the approach is useful for its capacity for explaining interpretative significance in the analysis of dialogue, and the wealth of research currently being undertaken in this area is likely to result in further refinements and improvements in approaches to analysing impolite exchanges.

4.2.4 From the performative hypothesis to speech act theory

A further pragmatic concept which has been borrowed into stylistics (and particularly the stylistics of drama) is that of the speech act. The origins of speech act theory may be traced to the work of J. L. Austin, who first noted that language is not just used to make statements or ask questions, but might also be used to perform actions (see Austin 1962).

Austin distinguished between what he called constatives (e.g. ‘I like Guinness’) and performatives (e.g. ‘I apologise’), on the basis that performatives can never be said to be true or false while constatives can. There is no sense, for example, in which ‘I apologise’ can be said to be either true or false. We might discuss the sincerity of the apology but this is a different matter to proclaiming truthfulness or falsity. In further describing the linguistic notion of a performative, Austin points out that as well as being unfalsifiable, all performatives are self-referential; that is, they refer to what the speaker is doing. Austin also distinguished a number of different sub-varieties of performatives. Metalinguistic performatives, for instance, refer to some linguistic act (e.g. ‘I say that he is a bounder and a crook!’ or ‘I propose that capital punishment be reinstated’) and by their nature are always successful (felicitous in Austin's terms). That is, by uttering a metalinguistic performative no-one can accuse you of not having performed the act specified by the verb. Not all performatives are intrinsically felicitous. Ritual performatives, for example, rely on external factors for their success. Consider, for instance, ‘I sentence you to 25 years’ hard labour.’ While anyone might make such an utterance, only a small minority of people within the criminal law system have the institutional authority to utter this and have it acted upon (i.e. the defendant taken away to prison). The external factors that need to be in place for a performative to be felicitous are referred to by Austin as felicity conditions and he outlines these as follows:

A:

i) there must be a conventional procedure having a conventional effect.

ii) the circumstances and persons must be appropriate.

B: The procedure must be executed (i) correctly, (ii) completely.

C: Often

i) the persons must have the requisite thoughts, feelings and intentions and

ii) if consequent conduct is specified, then the relevant parties must do it.

(Austin 1962: 14–15)

Sometimes a speaker may make explicit reference to felicity conditions when uttering a performative, as can be seen in example 17 (the underlining denotes this explicit reference):

Some performatives are collaborative in nature and require the consent of others to be felicitous. It is impossible, for example, to make a bet unless that bet is taken up by a second party. The effects that can arise from infelicitous performatives can, unsurprisingly, be exploited for dramatic purposes, as in example 18:

The metalinguistic performative uttered by Camembert in turn 3 (‘I demand immediate satisfaction!’) is an indirect challenge to Sir Rodney, which fails because it is not taken up by him; indeed, Sir Rodney deliberately avoids responding directly to the performative, thereby giving rise to a number of humorous effects. (Note that although the status of ‘demand’ as a metalinguistic performative means it is intrinsically felicitous, it is functionally infelicitous when viewed as an indirect challenge.)

Austin's work on what he called his performative hypothesis raises interesting issues about the nature of language, but it is inherently flawed. Thomas (1995: 44) summarises the reasons why the performative hypothesis does not work by making three major points. The first is that there is no formal (i.e. grammatical) way of distinguishing performative verbs from other sorts of verbs. Thomas points out that performatives do not necessarily have to be first-person present tense forms (for example, ‘Your contract is hereby terminated’ and ‘The student asserted that he had not plagiarised’ are both performatives despite being second- and third-person respectively). Second, the presence of a performative verb does not guarantee that the specified action is carried out (consider if ‘I promise to help you paint your kitchen’ is indeed a promise if the speaker does not then go through with his or her intended action). Finally, not all performative acts in English have an associated performative verb. Hinting, boasting, inviting, etc. are all carried out without using a performative, therefore it is clear that performatives are not necessary in order to perform actions in language. Indeed, the fact that there is no direct correlation between the form of an utterance and its force was eventually recognised by Austin, who then proposed a three-way distinction between locution (what is said), illocution (what is meant) and perlocution (the effect on the hearer of what is said). For example, saying ‘I’m not bad with a paintbrush’ has the form of a statement but may, in the right circumstances, have the illocutionary force of an offer (‘Would you like me to help you with your decorating?’). The perlocutionary effect of this utterance may be that the hearer accepts the assistance offered indirectly. Austin's term for an utterance and the situation in which it is used was speech act.

In an effort to systematise Austin's work, one of Austin's pupils, J. R. Searle, attempted to formulate a series of rules to explain how speech acts work. Searle proposed that all speech acts could be broken down into four components: the propositional act, the preparatory condition, the sincerity condition and the essential condition. We can see how this works if we consider the prototypical speech act of ‘offering’, which we might describe as follows:

Following Searle's scheme, the example below constitutes a prototypical offer:

Conversely, while example 20 may appear on the surface to have the qualities of an offer, it does not fulfil the relevant preparatory condition (i.e. that the speaker believes the offer is in the Hearer's interest) and so, according to Searle, we would look for an alternative explanation of the speaker's meaning. We might, for instance, say that example 20 has the locutionary form of an offer but the illocutionary force of a threat; that is, it is an indirect speech act, which may be defined as one speech act performed by means of another.

In a sense, then, Searle's improvement on Austin's work was a means of explaining how non-literal meaning is conveyed – a similar goal to that of Grice (1975). However, there are problems with speech act theory as a concept that makes Grice's approach to the analysis of non-literal meaning more attractive. For instance, there are certain speech acts that it is difficult to distinguish between. Orders, commands and requests all have the same felicity conditions, for example (Thomas 1995), so that the only differences between these speech acts are contextual rather than linguistic differences (e.g. that one speaker must be in authority over another in order to command). The other major problem with Searle's approach to meaning is that it is rule-based – and language does not lend itself to being tied down by rules (Jeffries 2007a shows how this is the case in an analysis of the speech act of apology). Thomas (1995) provides a very succinct discussion of the differences between rules and principles in pragmatics, and notes that the advantage of Grice's work over Searle's is that it is principle-based rather than rule-governed. Nonetheless, speech act theory retains some usefulness, especially when applied in the stylistic analysis of dialogue (either fictional or non-fictional), since it provides a means of explaining how foregrounding effects can be achieved in interactive talk. For instance, example 20 has a foregrounding effect that is achieved by deviating from the conventional speech act rules for making an offer, which gives rise to a very different illocutionary force. We should be careful, then, not to throw out the baby with the bathwater when criticising Austin's and Searle's work, not least because of its value for stylistic analysis.

4.3 An extended analysis of an extract from a dramatic text

The analytical approaches described in this chapter will be better understood in the context of a full analysis of a text. We have chosen a light-hearted example to discuss, as it is important to emphasise that stylistic frameworks are appropriate for analysing a wide variety of texts – non-fiction and fiction, literary and non-literary. The extract below is taken from the screenplay of the film Monty Python and the Holy Grail (Chapman et al. 1974):

We might start by observing that the situation and dialogue presented in the extract are absurd, especially when we consider our prototypical expectations for a conversation between a medieval king and his subjects. We would expect, for example, a high degree of subservience from those whom the king is addressing. We might expect them to use respectful forms of address when responding to the king. In terms of conversational structure, we might guess that the institutional power of the king would result in his linguistic dominance. None of these prototypical expectations are met in the above extract, resulting in foregrounding effects which we might point to as the linguistic locus of much of the humour in the text.

If we start with a simple CA consideration of the text, we can note that, of the three characters, Arthur takes twenty-six turns, Dennis takes twenty and the woman ten. On the surface, this would seem to bear out our expectation that the relative power of the king would lead him to dominate the conversation. We might also note that the fact that the woman takes fewest turns is perhaps representative of the subservient social role of women in this period. However, once we start to examine the mechanics of turn-taking, it becomes apparent that such assumptions will not hold for this text. For example, despite being the king, Arthur speaks the smallest number of words – an average of just 6.8 words per turn, which contrasts significantly with Dennis's 14.4. Even the woman speaks an average of 7.3 words per turn, which, although not a significant difference from Arthur's figure, is surprising nonetheless for the fact that it is a higher average per turn than that of the king.

Arthur's failure to establish his authority is further seen in the mechanics of turn allocation among the characters. Typically, we expect powerful characters to allocate turns to less powerful characters. Arthur does attempt this by asking six questions of Dennis and the woman, in turns 3, 16, 25, 27, 29 and 49 (albeit a non-genuine tag question). However, none of these questions is answered, and the peasant characters further defy the stereotype by allocating turns to Arthur through their own questions (see turns 10, 14, 15, 17, 19, 38 and 42). Arthur's relative weakness in the face of their conversational assault is made greater by the fact that he does attempt to answer their questions.

In general, it is not the fact that Arthur is not behaving like a king that gives rise to foregrounding effects, but the fact that Dennis and the woman are subverting our expectations for their characters. We see this if we examine the initiation of new topics in the text. Arthur initiates on four occasions (turns 1, 16, 25 and 27) but so too does Dennis (turns 4, 12, 30 and 50) and so too does the woman (turns 15, 19, 21 and 42). Tellingly, the only character to respond to the initiation of another is Arthur (see turns 5, 7, 9, 11, 13, 20, 39, 41 and 43). Clearly, Dennis and the woman are controlling the conversation. We also see this in the fact that Dennis and the woman have no compunction about interrupting Arthur – Dennis in turn 12 and the woman indirectly, as a result of interrupting Dennis in turns 15 and 23. This relatively straightforward analysis of conversation structure already begins to account for why we perceive the behaviour of the characters to be foregrounded.

We can also note a number of instances of dispreferred responses to Arthur's questions. In turn 4, for instance, Dennis gives a dispreferred response (‘I’m thirty-seven’) to Arthur's question concerning the inhabitant of the distant castle. The same is true of the woman's response in turn 17. This is arguably even more marked than Dennis's response in turn 4, since at this stage in the conversation, the woman is fully aware that she is addressing the king. These turns are thus foregrounded as a result of being conversationally deviant. Our expectations about the kind of encounter dramatised in the extract are also likely to lead us to interpret as deviant the fact that Dennis fails to give the preferred response to Arthur's orders in turns 35, 37, 45 and 47. Arthur even uses a metalinguistic performative in turn 37 (‘I order you to be quiet!’) but this still fails to establish his authority. Here again we see the adjacency pairs completed, but with dispreferred second parts.

If we now turn to the pragmatic behaviour of the characters, again we see that Dennis and the woman subvert our expectations for this kind of interaction. This is apparent both in terms of their non-observance of the Cooperative Principle and their lack of regard for politeness when addressing Arthur. Arthur, by contrast, makes substantial efforts to take account of Dennis and the woman's negative face wants by mitigating his FTAs. Despite beginning with a bald on-record request (‘What knight lives in that castle over there?’), which may be viewed as appropriate given Arthur's level of institutional power, Arthur does attempt to mitigate the damage to Dennis's positive face caused in his first turn by prefacing his request with a conventional apology. Similarly, in turn 16, Arthur prefaces his direct question to the woman with a conventionalised enquiry as to her health, and also uses a respectful address term (‘good lady’) to enhance her positive face. In contrast, Dennis and the woman repeatedly violate the maxim of relation in their failure to answer Arthur's questions. This in itself constitutes a threat to Arthur's positive face and may be construed as impolite behaviour conveyed via positive impoliteness. We can also observe Dennis using sarcasm in turn 14 (‘Oh, King, eh, very nice.’), which is clearly insincere politeness given what he goes on to say in the rest of the sketch. Nonetheless, Arthur's behaviour towards them remains fairly reasonable. His orders in turns 35 and 37 for Dennis to be quiet are bald on-record as opposed to positively or negatively impolite, and even when he increases the force of his command by using a metalinguistic performative (‘I order you to be quiet!’) this is not marked by further impolite strategies. Indeed, Arthur has been remarkably restrained so far, even addressing Dennis's positive face needs in 31 and 33 by communicating acceptance of the political points he is making. Dennis and the woman, meanwhile, fail to acknowledge Arthur's face needs. Their lack of concern for these also escalates as the scene progresses. In turn 44, Dennis flouts the maxim of manner when he says ‘Listen, strange women lying in ponds distributing swords is no basis for a system of government. Supreme executive power derives from a mandate from the masses, not from some farcical aquatic ceremony.’ The implicature created here is that Arthur's power has no democratic validity, and it seems that Dennis's choice of linguistic strategy here is designed to increase the threat to Arthur's positive face via positive impoliteness. In response, Arthur attempts another bald-on-record strategy by saying ‘Be quiet!’ (turn 45) and ‘Shut up!’ (turn 47), though the latter command is even mitigated by the application of a question tag (‘will you?’). Only towards the very end of the scene does Arthur break type and use a deliberately impolite strategy when he calls Dennis a ‘bloody peasant!’, the impolite address form constituting a tactic for aggravating the harm to Dennis's positive face needs.

In addition to the various linguistic elements we have pointed out so far, we might also note that the humour of the scene is accentuated through the characters’ use of anachronistic vocabulary and political ideas, giving rise to the incongruity necessary for the creation of humour (see, for example, Nash 1985). Nonetheless, an analysis of the dramatic dialogue enables us to identify the sources of foregrounding in the text, which arguably constitute the ultimate source of the humour in the scene. A stylistic analysis of the dramatic text, then, provides a means of accounting for our initial reactions to the scene and its effect on audiences.

4.4 Summary and conclusions

In this chapter we have outlined key theories and analytical frameworks for use in the analysis of dialogue. In applying these we are able to account for our initial interpretative responses to texts that contain such dialogue and (re)presented conversation. We have concentrated particularly on the analysis of dramatic texts, though such techniques are equally applicable to conversation in prose fiction and (re)presented conversation in non-literary texts. (In the exercises below we have used a range of different texts to illustrate this.) The analytical frameworks for the stylistic analysis of dialogue enable us to uncover the pragmatic motives of characters, narrators and writers, but in addition, we might also note that the application of such frameworks uncovers the discoursal equivalent of the foregrounding that arises from deviating from linguistic norms (see Chapter 2). In effect, analysing dialogue using the techniques described in this chapter allows us to identify conversational foregrounding, the upshot of which is to enable the identification of areas of the text which may prove likely to be of particular interpretative consequence.

Further reading

For an application of conversation-analytic and sociolinguistic approaches to the stylistic analysis of dialogue see Burton (1980), Short (1981) and Mandala (2007). Jenny Thomas's Meaning in Interaction (1995) provides an excellent introduction to key aspects of pragmatics, while such approaches are used in the analysis of drama by Bennison (1993), Cooper (1998), Culpeper (1998) and Bousfield (2007b), and in the analysis of prose fiction by Short (1995). Culpeper and McIntyre (2010) demonstrate a holistic approach to the analysis of drama by integrating a number of different analytical frameworks using activity type theory. Grice (1975) is the classic article on the Cooperative Principle and implicature. Austin (1962) outlines his notion of performatives and speech acts in a very readable collection of lectures, while Searle (1969, 1979) extends and develops speech act theory in considerable detail. Jeffries (2007a) and Short (2007) provide applications of speech act theory in non-literary and literary contexts respectively. Brown and Levinson (1987) is the standard work on politeness, while Culpeper (1996, 2005) demonstrates a face-management model of impoliteness. Bousfield (2007a) tests this model and provides a thorough overview of other work in this area. Levinson's Pragmatics (1983) is a higher-level textbook covering many of the pragmatic issues raised in this chapter and is highly recommended.