Orientation Observation In-depth interviews Document analysis and semiology Conversation and discourse analysis Secondary Data Surveys Experiments Ethics Research outcomes



Social Research Glossary

About Researching the Real World



© Lee Harvey 2012–2020

Page updated 29 April, 2020

Citation reference: Harvey, L., 2012–2020, Researching the Real World, available at
All rights belong to author.


A Guide to Methodology

6. Conversation and discourse analysis

6.1 Introduction to discourse analysis
6.2 Conversation analysis
6.3 Genre analysis
6.4 Pragmatics

6.4.1 Introduction
6.4.2 Speech act theory
6.4.3 Context
6.4.4 Examples of pragmatic analysis
6.4.5 Example of pragmatic methodology

6.5 Discursive psychology
6.6 Interactional sociolinguistics
6.7 Ethnography of communication/speaking
6.8 Narrative analysis
6.9 Critical discourse analysis
6.10 Summary and conclusion

6.4 Pragmatics

6.4.1 Introduction
Pragmatics is the study of 'the factors that govern our choice of language in social interaction and the effects of our choice on others' (Crystal, 1997, p. 120). Put another way, pragmatics examines how people do things with words and how they interpret what others are doing when they speak.

Pragmatics uses whole grammatical phrases as its unit of analysis rather than words or sentence structure. As Margarida Puig (2003, p. 2) argued that discourse and pragmatics have much in common not least the concern with utterances rather than elements of a sentence. Pragmatics, like discourse analysis, goes beyond structural study of the phrase.

'Instead it focuses on speech acts and conversation turns, taking into account context and speaker intention, which it does through the establishment of implicit elements that the hearer has to access.'

NB. An intervention or turn refers to a chunk of speech within a conversation (one's turn to speak) and may consist of one or more speech acts. The notion derives from conversation analysis, but is used in many forms of discourse analysis.

Puig (2003, p. 2) described the difference between pragmatics and discourse analysis as follows:

But while discourse analysts explain the interpretation of the elements in question without going outside language, pragmatics resorts to other ambits of human activity (beliefs, feelings, knowledge, intentions...). Only in this way can one explain how utterances are interpreted and how successful interpretation of utterances is managed. It is only with the aid of considerations of a pragmatic nature that we can go beyond the question "What does this utterance mean?" and ask "Why was this utterance produced?".

In this explanation, though, Puig does not differentiate between different forms of discourse analysis and, the demarcation she draws would not apply, for example, to critical discourse analysis or discursive psychology.


6.4.2 Speech act theory
Pragmatics is usually seen as deriving from the work J.L. Austin undertook in the 1950s and that was published as How to do Things with Words (1962).

Austin (1962), in developing 'speech act theory', argued that all utterances have an intention of satisfying a goal that was the point of uttering a sentence in the first place. This requires that there is a mode of understanding of the utterance that is common to speaker and hearer. Austin asserted that there are three basic acts.

First, is the locutionary act, which refers to the speech act in its basic sense; the physical properties of speech sounds (phonetic), the communication of atmosphere or mood (phatic), and the intended or semantic or intended meaning (rhetic).

Second, is the illocutionary act, which describes the intended significance of the act (such as promising, warning, advising or threatening). The illocutionary force of an utterance is the speaker's intention in producing the utterance, which is culturally-dependent, and requires that the hearer understands the cultural context.

Third, is the perlocutionary act, which illustrates the actual effect on the listener.

Thus, in making an utterance the speaker makes a sound and usually adheres to grammar and other semantic elements. The statement, such as 'Can you open the door?', while a grammatically correct sentence may, in a certain context, not be asking about someone's ability to open a door but have an illocutionary intent of requesting that the door be opened. Nonetheless, the hearer may take the statement literally and agree that indeed it is possible to open the door, implying, perhaps that the 'you' is the more generic 'one'. The response is the perlocutionary act, indicating the effect on the listener.

(Of course, the sequence could be just as intended by the locutionary statement, with its intended illocutionary request to have the door opened with the perlocutionary effect being that the hearer opens the door.)

Searle (1979) identified five types of speech act:

1. Assertives: a statement of truth, for example, 'the swan is white'.

2. Directives: a statement that requires compliance or fulfilment of the utterance, for example, 'keep off the grass'.

3. Commissives: in which the speaker is commited to fulfil the content of the utterance, for example, 'I won't do it again'.

4. Expressives: in which the speaker displays emotion in context with the content of the utterance, for example, 'I'm sorry to hear that'.

5. Declarations: in which the speaker performs an action by producing the utterance, in effect changing reality by making the declaration, for example naming a baby in a baptismal ceremony, declaring a marriage union between partners or pronouncing someone guilty in a court of law.


6.4.3 Context
Pragmatics, which studies language use and language users requires consideration of non-linguistic elements in order to fully interpret the utterance. Simply sticking to the analysis of the text, makes it impossible to understand the inferences in the utterances. Puig (2003) argued that pragmatics makes use of inference and therefore needs analysts to include non-linguistic knowledge to make sense of utterances. As she puts it, citing Reboul and Moeschler (1998, p. 35), interlocutors need 'to have knowledge of the world' and 'the study of language use has to explain how it is that sentences produced are successfully interpreted by interlocutors'. Puig recognises that context is not a novel contribution of pragmatics. Indeed, "context of situation" is important in Malinowski's (1923) cultural anthropology, which examined the general conditions under which a language is spoken. For Malinowski, situation and expression are inseparable.

Discourse analysis retains this concept and indeed makes it one of its central pillars. On the other hand, the concepts of speaker intention and inferences play a fundamental role in Speech Act theory and in formulations of Grice's principle of Cooperation. But every one of these terms has a different meaning in the different theoretical paradigms. (Puig, 2003, p. 3).

She argued that context tends to be fixed in discourse analysis but fluid in pragmatics. However, this ignores the intended fluidity in, discursive psychology, for example.

Gumperz (1982a, pp. 185–6) also referred to the need to go beyond analysis of the words to include 'cultural logic':

The fact that two speakers whose sentences are quite grammatical can differ radically in their interpretation of each other's verbal strategies indicates that conversational management does rest on linguistic knowledge. But to find out what that knowledge is we must abandon the existing views of communication which draw a basic distinction between cultural or social knowledge on the one hand and linguistic signaling processes on the other.

Context is thus important in understanding the illocutionary implications and the perlocutionary outcomes of an utterance. 'The lion is waking up' is a statement that is quite different if spoken to a zoo keeper inside a lion's cage than if spoken by a vet observing the progress of a sick lion.

So pragmatics takes context into account when analysing complete grammatical phrases as it is important to understand the illocutionary intent and perlocutionary outcome of speech acts, which, according to Serle, can be grouped into five types: assertions of truth; orders; commitments; emotional expressions; and declarations.


6.4.4 Examples of pragmatic analysis
Argumentation is one form of discursive activity that has attracted the interest of researchers, because of its prevalence in communicative activity, notably in legal, political and advertising settings.

Puig (2003) provided an example of conversation in which, during a political debate in the Catalan Parliament, speaker A urges speaker B to hurry up and undertake a task, to which speaker B replies (in Spanish) "Do you read the papers?". Puig (2003, p. 2) argued that:

While discourse analysis can only explain that this is a reply to the observation made by [Speaker A] or explain what type of sentences make up each of the utterances, pragmatics will explain what kind of reply it is.... For example, "if you read the newspapers you will know that I have done so many times", or "as I am sure that you read the newspapers, I think you know perfectly well that I have done so, therefore your observation is unnecessary". Taking a pragmatic approach, the linguist can successfully uncover the intention that [Speaker B] has in selecting "Do you read the papers?", and why he selected this utterance rather than another one.

Puig argued that pragmatics offers something new because it focuses on the speaker's interpretive strategy, in which qualities and moods such as rationality, desires and mental states are attributed to other speakers. The speaker's strategy is designed to predict the interpretive behaviour of other speakers. This requires taking into account context, intention and relevance; the latter is, in effect, the shared knowledge of speaker and hearer that make the intention intelligible.

Puig analysed political discussion using Searle's (1979) development of speech act theory. An argument is an illocutionary act associated with the perlocutionary act of persuasion; an act whose objective is to get interlocutors or audience to accept a series of ideas (the arguments) that involve the demonstration of a conclusion.

According to Puig (2003, pp. 3–4) van Eemeren (1984, pp. 43–45) characterised the illocutionary act of persuasion in the following way. It has:

Propositional content, the totality of propositions expressed.

An essential condition, the act of articulating a proposition constitutes an attempt by the speaker to justify an opinion to the hearer.

Preparatory conditions, the speaker believes:
a) that the hearer will not initially accept the opinion;
b) that the hearer will accept the totality of propositions expressed;
c) that the hearer will accept the constellation of propositions as a justification of the opinion.

Sincerity conditions, the speaker believes that:
a) the opinion is acceptable,
b) the propositions expressed in the utterances are acceptable,
c) that these propositions constitute a reasonable justification of the opinion.

To put it another way: speakers, who have an opinion that is not accepted at the outset by the interlocutors, employ a series of propositions, which are thought to be acceptable and that are thought to be a good justification of the opinion, in order to change interlocutors' initial opinion.

While we construct the discourse we impose on it a precise process of interpretation, offering guidance on how to attribute meaning to our utterance. In this way we guide listeners along the interpretative path which will lead them to understand what we say and the intention with which we say it. And we do so by devising a strategy, applying effort to the selection of words and discourse movement, with a view to achieving certain specific communicative objectives. If we apply this concept to argumentative discourse, we can tease out three basic argumentative orientations: the concessive, the consecutive and the conclusive. (Puig, 2003, p. 4)

a) Concessive orientation operates with two speech acts or two interventions. This process is such that the implications of the first half of the speech act (1) is negated by the second half (2).

For example, the statement:

'(1) Even though you come (2) you won't see her',

operates as follows. The initial implicit conclusion is that,

'(1) If you come (2) you'll see her' (or else why would you be coming?).

However, in reality,

'(1) You will come but (2) you won't see her',

thus (2) makes it clear that the first act is not relevant, argumentatively speaking, because its implicit conclusion is not acceptable.

Put another way, 'there is no use in your coming if what you want is to see her'.

b) Consecutive orientation also contains only two speech acts or interventions. One is an argument in favour of a proposition and the other is the argumentative conclusion of the proposition: a form of reinforcement. For example,

'(1) You don't want to come, (2) so don't come.'

The initial part (1) serves as the argument for the second part (2).

'As you don't want to come, don't do so'.

c) Conclusive orientation operates with three constituents, instead of two. One in favour of the implicit conclusion of the proposition, another in favour of its negation and a third that acts as a conclusion, which is orientated towards the negation. For example,

'(1) The weather man said it would rain, (2) but she didn't believe him. (3) As it turned out, she didn't get wet.'

The first part (1) orientates towards the implicit conclusion 'she will get wet'; the second (2) orientates towards the negation 'she won't get wet'. The third act, however, goes in the same direction as (2); in the end she doesn't get wet.

On the basis of analysing a debate in the Catalonian parliament, Puig concluded that that the concessive orientation is the most usual type of argumentation in political debate, followed by the consecutive orientation. The conclusive orientation is rare in discussion and tends to occur in pre-prepared monologue (that usually open and close debates).

Further, she stated, that comparative analysis of the argumentative orientations in the speech of the different politicians made it considerably clearer as to why some were more, or less, effective than others.

Another study has explored how context overrides intention in situations where a dominant group negates the utterances of someone from a non-dominant group.

Rebecca Kukla (2012) argued that a speaker's membership in a disadvantaged social group can make it difficult or impossible to deploy discursive conventions in the normal way. The outcome is that the performative force of some utterances are distorted in ways that reinforce disadvantage. This distortion is, in fact, the production of a different kind of speech act, despite using the same words that someone in dominant social group would use. Kukla referred to this as discursive injustice.

Kukla, citing Langton (1993), stated that locutionarily silence occurs when a person is prevented from speaking. Perlocutionarily silence occurs when a person's speech cannot have its intended causal effects. Illocutionarily silence occurs if the person is unable to perform the intended speech act because the speech act cannot receive the right uptake. She thus focused on how 'a speech act can, in virtue of its uptake, become a different speech act than it would typically be, given its social context and standard discursive conventions' (Kuka, 2012, p. 3 ).

Uptake is important because that determines the perlocutionary force: that is, the hearer responds appropriately to the speech act. If there is no shared understanding or the hearer is mistaken in what is intended, then the result is not as intended. However, in some cases, Kukla argued, there is a systematic failure of uptake.

Kukla used the example of a female in an authority position in a male-dominated factory who is entitled to issue orders but gets little compliance. Kukla argued that this could be regarded as a case of general insubordination but is more likely to be a more subtle problem. The men do not see the woman in an authority position despite her formal role. In which case they treat her orders as requests. Rather than an order that places an obligation, the order is treated as a request, compliance with which is voluntary. The woman supervisor is also perceived by the men as unpleasant because she does not ask nicely.

The female's authority is undermined by the lack of compliance with orders and it would also be undermined if she switched to issue appropriately worded requests. Kukla (2012, p. 8) concluded the example:

[Her] gender queers her ability to exercise her agency through speech, precluding her from effectively deploying the discursive conventions suited to her social position and needs. She is in a discursively unmanageable position where each new attempted order and each new response chips away yet more at her ability to use speech to control action in the workplace.

In this case despite her ability to speak and the clear intention, the context and her disadvantaged status, vis vis the hearers, means that there is no perlocutionary force to the utterances. 'Discursive injustice occurs when our loss of control over our speech comes from our inability to mobilize conventions in the standard way, resulting in a failure of agency that tracks and enhances social disadvantage. Thus discursive injustice is a risk faced by the relatively disempowered' (Kukla, 2012, p. 15).

For a detailed account of a pragmatics analysis and comparison to other discourse analysis approaches see the study by Maria Stube, et al. (2003) in CASE STUDY Discourse analysis comparative study.


6.4.5 Example of pragmatic methodology
Sara Lynch (2013) explored the variations of the word 'sorry' as part of Irish English discourse. She used data taken from ICE-Ireland corpus, which contains about a million words and is one of 18 national parts of the International Corpus of English (ICE) project.

A corpus is a collection of language data (in the case of ICE it is natural conversation and text) brought together for linguistic analysis. The ICE corpus is large, diverse with informal and formal discourse styles, a good gender and age mix and group sizes ranging from two to ten conversational participants. ICE has a uniform method of transcribing and analysing. The transcription codes provide additional information on pauses, overlapping speech, incomplete words, as well as editorial comment, information about the speakers and context. What the data does not provide is any information on social distance and power relationships among the speakers, which Olshtain & Cohen (1983) described as important social factors when surveying politeness. However, that reservation apart, Lynch argued that using ICE data would be the most effective approach for her enquiry.

Lynch only used the speech samples, not text samples, and selected the 15 Southern Ireland speech samples from the 300 available. She looked at the subsections of these 15 that included apologies, which was the subject of her study. She ended up with speech samples from 80 males and 79 females. (Lynch, 2013, pp. 24–5). In addition to qualitative analysis she undertook quantitative analyses.

Lynch's study identifies a set of seven investigative questions as well as clearly-stated procedural steps. The questions included:

  • Is sorry used as a discourse marker or something else?
  • Do certain social groups use the speech act of apologising more frequently than others?
  • Do social factors such as age, sex or religion influence frequency of use of sorry?
  • What do the findings say about the token sorry and its uses in Irish English?

The procedure was as follows (adapted from Lynch (2013, pp. 19–26):

1. Locate all instances of the sorry token within the ICE-Ireland Corpus.

2. Count up the total number of words in the selected sample from the corpus.

3. Count how many different speakers are in the sample.

4 Locate 'sorry' in each transcription using search facility (in Word).

5. Copy each extract, limited by the start and end of that speaker's utterance.

6. Add speaker details.

7 Tokenise each instance of sorry based on function (modified from Deutschmann's interpretation of apologies in British English); that is identify what kind of 'sorry' is being stated and what is the intention of the statement.

8. Examine each sorry as a discourse marker. "I did this by removing it from the utterance and checking if the sentence was altered by its absence".

9. Compare each token of sorry with its annotated illocutionary act (as listed by Searle).

10 Counting instances by independent variables and graph the results ready for analysis.

11. In pragmatics, quantitative analysis involves counting the number of occurrences of the target phrase (dependent variable), in Lynch's study 'units of apologies in British English', and then analysing the extent to which frequency varies for different sub groups or in different contexts (independent variables).

To do this requires a considerable database of spoken material from which to work to avoid the problems of bias that may occur from a small selective sample, as it would be difficult to identify what constituted a 'random sample' of British English speech. The use of established large data sets such as ICE provide reasonable assurances for quantitative analysis of speech utterances.

Lynch (2012, p. 50) concluded:

In this study, it has been demonstrated that sorry plays an important role in Irish English discourse and performs a variety of functions. It has been shown that sorry occurs frequently in Irish English and generally is realised as a positive politeness strategy.

Referring to Searle's (1979) categorisation, she added:

We have seen that as a speech act, sorry correlates with expressives, directives and representatives but there is no evidence to support that it is commonly found with commissives or declaratives. We have found that sociological factors to influence the frequency of sorry in Irish English discourse. It has been presented that females are more inclined to use this form. I have suggested that sorry does not only function as an apology but is also used as a hedging device and also performs the role of a request and a conversation entrance marker. Correlating with prior studies, I have put forward that younger and older speakers tend to favour the use of sorry, while middle aged speakers are more likely to choose other lexical forms. The more formal the conversation, the less likely sorry will appear, while informal conversationalists tend to use it in the form of a positive politeness strategy. Religious identification appears to influence frequency of use whereby those who identified as Christians tended to favour the form over those who did not identify as Christians with positive politeness strategy-sorry being the preference. Overall we can assume that Irish-English, at least in conversation, do not appear to use sorry in a hugely different way to their British English counterparts despite prior suggestion though frequency would still need to be further examined. Other cases such as sorry uttered on the street if someone brushes off another cannot be examined in a Corpus based study. Cases such as these would possibly be better examined in a questionnaire or DCT [discourse completion test]. (Lynch, 2013, pp. 50–51)


Next 6.4.5 Discursive psychology