The method of discourse analysis is complex and cannot be properly understood without extensive reading. The aim of this web page is to provide you with an outline of the approach so that if you haven't read much about it you can, among other things, decide whether you would like to learn more about the method or whether you would like to carry out research using this approach. An example is provided further down the page to illustrate a popular kind of discourse analysis, which is known as thematic analysis. It includes some examples of transcription symbols.
Discourse analysis is a qualitative method that has been adopted and developed by social constructionists. Although discourse analysis can and is used by a handful of cognitive psychologists, it is based on a view that is largely anti-scientific, though not anti-research. Social constructionism is not easy to define, but it is possible to outline some basic assumptions of the approach:
• Psychologists cannot be objective when studying human behaviour. In the scientific approach there is the belief that knowledge can be gained by objectivity (observations made without bias or preconceptions, rather as though the investigator is an alien from another planet and has no preconceived notion of what is being observed). However, this belief has been disputed – people, including researchers, cannot be objective. A researcher is very likely to hold some position (expectation, bias, belief, or set of cultural values) when they are conducting their research. The result is that people can construct their own versions of reality and not necessarily objective versions either.
• Reality is socially constructed. In the scientific approach it is assumed that it is possible to categorise reality, and that ideas or constructs that psychologists use, such as personality and intelligence, are naturally occurring things or categories. However, this ignores the fact that language shapes the categories and constructs we use. Since language is a social and cultural thing, our sense of reality is socially and culturally constructed.
• People are the products of social interaction. In the scientific approach it is assumed that many of the constructs used are ‘inner essences'. That is to say that personality, anxiety, drives, and so on, exist somewhere within. However, it may be the case that many of these so-called essences are actually the products of social interaction.
In order to understand these assumptions, let's look at the example provided by Burr (1995) on the issue of personality.
The traditional view of personality
Personality consists of a number of traits such as generosity, shyness, charm, and so on. What makes people different is that they can be high on some traits and low on others. A further assumption is that, by and large, personality is stable over time – although a generous person may have one or two lapses, they are generous most of the time. Another assumption is that personality influences our behaviour – we ascribe the inner essence ‘generosity' to someone acting very generously. This view of personality seems just common sense and it seems to make sense in our everyday understanding of people.
The social constructionist critique of the traditional view of personality
Personality, it is argued, is a socially constructed concept and traits do not exist as inner essences at all. The way people are arises from their interactions with other people. When you think about the traits we use to describe people, virtually all of them involve actions that can only take place with reference to other people. For example, a shy person is only shy in the presence of others, a generous person can only be generous to other people, and so on. One could ask whether such personality traits would still exist in someone stranded on a desert island!
Social constructionists remind us that since personality CANNOT be observed directly its existence has to be inferred, and it is inferred from behaviour. However, someone's behaviour can be very different depending on the context or situation they are in. Furthermore, people can be both nice and nasty, that is, behave in opposite ways to the traits they are commonly described as having.
So, who am I?
By now you may be completely confused, as I was when I first encountered this viewpoint. If personality and inner essences do not exist then we must ask ourselves who we are and what makes us who we are? According to social constructionism each version of ‘you' is a product of your relationships with others. The word ‘identity' is used to refer to a person's purpose within a social relationship. In other words, we have different identities based on who we are with, where we are, the situation we our in, and so on. The creation and use of such identities can be understood by psychologists by trying to study the language that people use. Furthermore, by studying conversations and all forms of communication we can understand how people and society ‘construct' their own versions of reality.
Why discourse analysis?
Discourse analysis is a way of understanding social interactions. The researcher acknowledges their own bias and position on the issue, known as reflexivity. The aims of research vary: The aim of one investigator might be to understand power relationships in society in order to bring about change; another may be interested in appearance and how it can shape identify; and another investigator may be interested in an interaction or conversation simply for its own sake (in terms of not knowing what the study might uncover). The research begins with a research question (and not a hypothesis in the formal sense) that is aimed at a theoretical position. A conversation or piece of text is transcribed and then deconstructed. This involves attempting to identify features in the text, such as discourses. A discourse is a particular theme in the text, especially those that relate to identities, for example such as a statement that reiterates a view or claim that men find weddings dull, and so on. Topics that have been studied include men's friendships, family conversations of the royal family, an interview with Princess Diana, media constructions of racism, gender categories in discourse, lesbian motherhood, conversations about marriage, men's talk about fatherhood, and so on.
How to do a discourse analysis
The first point to note is that in order to do a discourse analysis you need to have read a handful yourself first. By reading published articles that use the method, you will have a better understanding of (1) how to do an analysis and (2) some of the theoretical orientations that you will need to know to do your own analysis. Having identified a theory and a chosen item (text or recorded conversation) to analyse, you need to transcribe it in one of the accepted/published ways. The transcript must always appear in the appendices. There are many different forms of discourse analysis, so here we will focus on thematic analysis as an example.
What is thematic analysis?
Thematic analysis is about trying to identify meaningful categories or themes in a body of data. By looking at the text, the researcher asks whether a number of recurring themes can be abstracted about what is being said. For example, on one level you might find an inconsistency, an attempt to assign blame, an attempt to cite others to support one's views, a regular interruption of other people, an attempt to make one's account of some event sound more authentic, and so on. On another level, you might idenitify a regulalry occurring attribution of blame or the repeated reference to some specific cause of an event. The reference might take slightly different forms but refers to the same cause. An example might be football fans blaming various aspects of a player's motivation for the failure of their team (e.g., "he gets so much money, doesn't need to try", "he looked as though he wasn't bothered", "he didn't want the ball", and so on).
In the results section of the report, the themes abstracted are collated and reported on. In doing so, it is usual to cite from the transcription examples of the points you are trying to make. A summary of the findings can be offered but also a critique of the author's own interpretations – this refers to the concept of 'reflexivity', that the author's is only one interpretation of the text.
When transcribing text, a conventional system of symbols is used. A table of the symbols used in the transcription can often appear as an appendix, such as the following (some of which are taken from Potter and Wetherell, 1987):
For further reading on this method you could read Chapter 3 and Chapter 5 of Hayes (1997). Each chapter is a description of how a particular analysis was carried out. One is an analysis of conversations about the royal family and the other is the analysis of an interview with a student about social class.
The following is loosely based on Barker (2003), which is a discourse analytic study of conversations with ‘Goths and Pagans’ – youths who belong to specific subcultures. The paper has been rewritten as an example item of coursework (with the author’s permission). When you read the report you may notice several features that are very different from how a lab experiment is reported.
Firstly, there is no attempt to identify causes from an objective perspective – why people choose to wear Goth clothing, but rather the focus is on how Goths themselves perceive their own subculture, their identities within it, and their experiences of how non-Goths perceive them.
Secondly, the shape of the report is quite different – there are no graphs or tables, and no statistics to report. The main source of data is the transcriptions of the interviews and much of the text refers directly to quotes from the transcriptions. There are other differences, too, such as very little ‘theoretical integration’ in the discussion and little attempt to say how one could follow up the research.
Thirdly, there is some discussion of the reliability and validity of the method, but this is cannot carried out in the same way as would be done for a report on a lab experiment. Note the fact that a discourse analysis is ‘text intensive’, there being a limited space to cover other aspects. However, one way in which the student has saved words in the main body of the report is to number the quotes and recording them as an appendix. Then in the text, rather than citing from the transcript, the student just refers to ‘Quote 4 in Appendix B’ and so on.
If you have become more comfortable with the more traditional
approaches in psychology, such as the lab experiment, then you may find
the report difficult to read. I have to confess feeling slightly uneasy
about recommending a discourse analysis for a beginner. However, there is certainly something to be said for
this approach and if you do feel happy to use this method, then
Barker, M. (2003). Satanic subcultures? A discourse analysis of the self-perceptions of young goths and pagans. In T. Waddell (Ed.) Cultural Expressions of Evil and Wickedness: Wrath, Sex, and Crime. Amsterdam: Rodopi.
The subculture known as Goth emerged in the 1980s with the music of
post-punk UK bands, such as Siouxsie and the Banshees. Goth is associated
with a particular style of dress, which generally involves wearing black
clothing, silver jewellery, a pale complexion, and dyed hair. However,
there is some variety in this style and Goths may be found wearing silver
and have spiky hair. Paganism is often claimed to be an alternative form
of belief to Christianity that has had a history of being suppressed and
oppressed by Christians. However, many historians claim that Paganism only
emerged quite recently – at the start of the twentieth century. Goths and
Pagans are linked in many ways. Firstly, they are both outside mainstream
culture. Secondly, both often wear jewellery that has Pagan symbols on it.
Thirdly, both groups have been labelled by mainstream culture as
satanic, evil, and dangerous.
Several recent crimes have been linked to Goths. One of these was the
1966 Florida‘vampire murders’ that was supposed to have been done by
members of a ‘vampire cult’ (Whitworth, 1999). Another was the 1999
‘trenchcoat killings’ in which two teenage boys shot and killed thirteen
people at their Denver school. In the UK the London bomber David Copeland
was recently ‘linked’ to the Goth culture (Smith, 1999). In Germany, two
murderers were claimed to have been devil worshipers, being involved in a
Gothic club, bloodsucking, and graveyard parties. Many of the articles
rely on the construct that Goths and violence are linked in a
taken-for-granted common sense way.
The main social psychological theory of how individuals in subcultures
create identities is ‘social identity theory’ (Tajfel and Turner,
1979). This suggests that people organise their perceptions of themselves
and others by a process of categorising others into groups, and then
identifying with one group as opposed to another. Our sense of self-esteem
is said to come from how we compare our group with that of others. People
tend to exaggerate the similarities between members of their group and
exaggerate the differences between members of their group and members of
other groups. This approach has been criticised for its failure to focus
on any groups other than ‘traditional’ groups (Wetherell, 1996). A related
theory is that of Baumeister (1996) who has argued that when we are
attacked by others we tend to fall back into relying on our group
identity, seeing our group as good and innocent and the other group as
sadistic and motiveless. A common theme of these theories is the notion of
‘us and them’ to describe experience. However, Widdicombe (1993) has
argued that such research has ignored how members of subcultures
understand the meaning and significance of their subculture, and that this
can be addressed by using discourse analysis.
Discourse analysis examines how people use language to construct versions of their experiences, and is based on the assumption that people draw on cultural and linguistic resources in order to construct their talk in certain ways to have certain effects. The aim of this investigation is to seek to understand how the accounts of the identities of Goths and Pagans are constructed and what is gained from these constructions.
The conversations were recorded using a mobile digital recorder and were later transcribed by playing the recordings on a computer system. The transcription code used was:
(…) - long pause between words
(12) - a 12 second pause between words
X - means that we could not understand what was said
So she said..
So I said… - bold text was said with an emphasis (i.e., said louder)
The interviewer’s speech begins with the letter I, and other letters
are used to identify individual participants.
The analysis that was conducted revealed a number of themes. The most interesting to the researcher were the identities of ‘persecuted outsider’, ‘otherness’, and ‘being different is good’. Some contradictions were also noted, such as ‘we are just ordinary people under attack’.
All participants agreed that people from outside their subculture labelled them as evil or Satanic, using the words ‘Satanist’, ‘freaks’, or ‘witches’, when asked how they are perceived. Quote 1 in Appendix B shows two examples. It is interesting that both N and D use the same wording in their examples, ‘oo Satanism’. This is known as active voicing (Wooffitt, 1992) when reporting someone else’s speech within an account of what really happened. Utterances reported in this way may be used to imply that they were really said at the time (Hutchby and Wooffitt, 1998). Both D and N also use hypothetical illustration (Wooffitt, 1992). The situation ‘oo Satanism’ is not one single event and is used to imply that it has happened many times. This is supported by D’s remark ‘So it’s really common’.
By putting on a dumb or silly voice when imitating other people, it is obvious that they regard such views as ridiculous. D’s example is especially clear because she voices with bad English ‘you’s all Satanists’ implying her view that people with such prejudices are generally ‘pig ignorant’.
The experienced prejudice was also supported by some stories, such as that of G when someone was violent towards him (see Quote 2, Appendix B). Here G and C both construct themselves as reasonable people in unreasonable situations, using a format Wooffitt (1992) refers to as ‘I was just X … when Y’. They were just doing something normal when they were attacked by ‘drunks’. G also uses the word ‘random’ when describing receiving abuse ‘yelling random things’ and later with his claim of the unpredictability of whether one reacts or not to abuse. These emphasise the random, chaotic nature of the attacks they received. This fits with Baumeister’s (1996) myth of pure evil: the common way in which people construct themselves – we are good, innocent victims provoked by sadistic, chaotic attackers.
Both groups said that they felt that ‘people’ or ‘they’ perceive them in a negative way. This suggests that they construct themselves as being ‘outside’ mainstream culture (see Quote 3, Appendix B). The construction here is of an identity of a group that defines themselves more favourably than the norm. In discussing the perception that they are linked to violence most made it clear that they are strongly opposed to doing harm (see Quote 4, Appendix B). Many cited this ‘harm none thing’ as a device to show that they could not be involved in evil in any way (see Quote 5, Appendix B). Here C uses stake inoculation (Potter, 1996); she initially expresses doubt about the truth of her claim (I don’t have much experience of this). Potter argues that such expressions of uncertainty allow speakers to establish that they have no stake in what they are saying. She combines this with expert knowledge to display that she actually does know about Satanism and can show that it differs from her own beliefs. She uses the externalising device ‘what I’ve read is …’ to construct her understanding of Satanism as factual (Edwards and Potter, 1992). When A talks about not being evil she uses the same devices (see Quote 6, Appendix B). Here A uses two three-part lists (Jefferson, 1991) to illustrate quite how different she is from the assumptions that are made about her by the ‘god squad’. Jefferson says that a three-part list is a culturally available resource for list construction which we often use in everyday conversation. Listing these behaviours together indicates a broader class of things that A does not do, backing up her contention that she is gentle and not evil.
The interviewees spent some time talking about the claim that their style of dress is merely to get attention. G and C both counter this notion by stating that they get more of a reaction from people they know when they wear non-Goth clothing (see Quote 7, Appendix B). So they do admit that they partly aim to get a reaction or at least enjoy getting a reaction from people by dressing as they do. Occasionally they suggested that they wear Goth clothes to divert attention away from other physical aspects of themselves, such as their height or weight.
Often the interviewees told the history of Paganism as the story of a group of persecuted outsiders. This is echoed by the stories they tell about their own persecution.
Reliability and Validity
Edwards, D. and Potter, J. (1992) Discursive Psychology. London: Sage.
Hutchby, I. and Wooffitt, R. (1998). Conversation Analysis. Cambrdige: Polity Press.
Jefferson, G. (1991). List construction as a task and a resource. In G. Psathas and R. Frankel (Eds.) Interactional Competence. Hilldale, New Jersey: Lawrence Erlbaum Associates.
Potter, J. (1996) Representing Reality: Discourse, Rhetoric and Social Construction. London: Sage.
Smith, A. (1999). Have you got what it takes? The Sunday Times, 9th May, Culture, p.12.
Tajfel and Turner (1979). An integrative theory of intergroup conflict. In G.W.Autin and S. Worchel (Eds.) The Social Psychology of Intergroup Relations. Montery, California: Brooks/Cole.
Wetherell, M. (1996). Identities, Groups and Social Issues. Milton Keynes: OU Press.
Whitworth, D. (1999) Gloomy tribal craze that was born in Britain. The Times 22nd, April, p.5.
Widdicombe, S. (1993). Autibiography and change: Rhetoric and authenticity of Gothic style. In E. Burman and I. Parker (Eds.) Discourse Analytic Research: Repertiores and Readings of Texts in Action. London: Routledge.
Wooffitt, R. (1992). Telling Tales of the Unexpected. Hemel Hempstead: Harvester Wheatsheaf.
N: the symbol of the pentagram (.) um (.) which people (.) when they see it they automatically think (.) ‘oo Satanism’ (…) coz a friend of mine went to get a tattoo (.) and he wanted a pentagram (.) and the guy said ‘oh you’re a Satanist are you’ (laughs).
D: always worn a pentagram ring (5) people have always looked at it and gone ‘oo oo Satanism’ [dumb voice] and (..) ‘you’s all Satanists’ (.) yeah I’ve often had to defend myself against allegations. So it’s really common.
G: I mean when we were walking up [road name] there was about twenty (.) odd people who started yelling random things at us and one guy decided he was gonna come down and do something about it and he hit me around the head with a beer can
C: It was um (.) I think they were all drunk and we probably should’ve ignored them more than we did (.) it’s hard when people make sort of (.) stupid ignorant comments not to really
G: yeah I mean I’ve tried it both ways tried ignoring them and then you get something whack on your back
G: I just don’t get on with townie normals because they have the view that you must dress like this you must wear named brands and you must like clubby music
S: I view people who dress in a conformist way as people who (.) don’t think for themselves people who (.) take the easy option and wear what they are told to wear
C: we don’t try to (.) harm others (.) if you (.) if someone actually does like to start (.) doing curses or whatever that person (.) couln’t be classaed as a Pagan because they have the harm none thing
C: I don’t really have much experience of this (.) but what I’ve read is um the idea is that Satan rebelled against god so (.) Satan was saying that I’m (.) I am you know (.) on my own I can you know do things for myself (.) I don’t need you and that’s what a lot of Satanists think
A: I just say that I can’t be a Statnist I don’t believe in god (.) you
can’t have one without the other (.) but they don’t seem to get it … I
mean I am (.) like a (.) I’m basically quite a gentle person (.) I’m not
actually that violent (.) I don’t steal I don’t do drugs (.) certainly
don’t go robbing things and bashing
S: at school I always dressed differently and that always caused problems on non-uniform day everybody would sort of (.) stare at me (.) but I started to thrive on it (.) and play up to it and now I’ve just come to accept that they will stare and I should perhaps take it as a compliment although in the daytime I dress down because I still don’t want to get (.) stared at too much.
© Eamon Fulcher, 2012