Taylor and Francis Group is part of the Academic Publishing Division of Informa PLC

This site is operated by a business or businesses owned by Informa PLC and all copyright resides with them. Informa PLC's registered office is 5 Howick Place, London SW1P 1WG. Registered in England and Wales. Number 3099067.


Section B: Development: Approaches to discourse analysis

B1: Approaches to Discourse Analysis

B2: More on cohesion and coherence

B3: Creativity: genre bending and genre mixing

B4: Ideology and Discourses: Big 'C' Conversations

B5: The texture of talk

B6: Face threatening acts

B7: Examples of Ethnographies: Drug Users

B8: Affordances and constraints

B9: Analyzing multimodal interaction

B10: Other techniques in corpus assisted discourse analysis

B1: Approaches to Discourse Analysis

This book covers most of the major approaches to the study of discourse. Below is a short summary of those approaches along with links to more information about them.

Textual analysis
I use the term ‘textual analysis’ to refer to approaches to discourse which focus on understanding how texts are structured beyond the level of the sentence. The focus is primarily on the contributions of M. A. K. Halliday to the analysis of cohesion and the organization of information in discourse.

Useful links

Coherence and Cohesion in Text Linguistics

Coherence and Cohesion, One Stop English

‘A tribute to Michael Halliday’, International House Journal

Genre analysis

Genre analysis examines the structure and communicative purpose of different types of texts (e.g. business letters, newspaper editorials, weblogs) and the way these texts function in the groups of people who use them. Genre analysis examines how texts are structured to fulfill communicative aims and how members of discourse communities use genres to claim and impute community membership. Major figures in genre analysis are John Swales and Vijay Bhatia.

Useful links

Genre and Discourse Analysis, Bev Traynor

John Swales

Critical discourse analysis

Critical discourse analysis aims to uncover ideology and power in discourse by understanding the relationship between textural features and larger social practices. It draws on many other approaches to discourse, including Halliday’s systemic functional linguistics. Major figures in critical discourse analysis are Norman Fairclough, Ruth Wodak and Tuen van Dijk.

Useful links

Critical Discourse Analysis, Tuen van Dijk

Principles of Critical Discourse Analysis, Tuen van Dijk

Critical Discourse Analysis, Norman Fairclough

Norman Fairclough

Ruth Wodak

Tuen van Dijk


Pragmatics grows out of the philosophy of language of Paul Grice, John Austin and John Searle. It examines how people ‘do things with words’ and how they interpret what others are ‘doing’ when they speak. Major contemporary figures in pragmatics include Jacob Mey and Jeff Verschueren.

Useful links

What is pragmatics? Shaozhong Liu

Jacob Mey (Wikipedia)

Jeff Verschueren

Conversation analysis

Conversation analysis is an approach to the analysis of conversation that grew out of the sociological tradition of ethnomethodology. It was developed by Harvey Sacks, Emanuel Schegloff and Gail Jefferson and concerns itself with how people create order in their social interactions through the structure and procedural rules of conversation. Major contemporary figures in conversation analysis include Paul Drew, Paul ten Have and John Heritage.

Ethnomethodology and conversation analysis, Robert O. Keel

Paul Drew

Paul ten Have

John Heritage

Useful Links

Interactional sociolinguistics
Interactional sociolinguistics is an approach to discourse analysis that focuses on how people manage social identities and social activities as they interact using such devices as contextualization cues and conversational style. It is based largely on the work of Erving Goffman and John Gumperz. Contemporary scholars working in this tradition include Deborah Tannen, Deborah Schiffrin and Heidi Hamilton.

Interactional sociolinguistics, Benjamin Bailey

Deborah Tannen

Deborah Schiffrin

Heidi Hamilton

The ethnography of speaking
The ethnography of speaking is an approach to discourse that grew out of linguistic anthropology, especially the work of Dell Hymes. It seeks to discover the knowledge speakers need in order to be competent participants in speech events within communities. Contemporary scholars working in this area include Gerry Philipsen,  Muriel Saville-Troike and Donal Carbaugh.

Useful links

Ethnography of communication, Electronic Encyclopedia of communication

Gerry Philipsen

Donal Carbaugh

Mediated discourse analysis

Mediated discourse analysis grows out of the theories of Lev Vygotsky and Mikael Bakhtin as they were developed by the psychologist James Wertsch. It focuses on how concrete, real time social actions are mediated through discourse. Scholars working in this area include Ron Scollon, Rodney Jones and Sigrid Norris.

Useful links

A Conversation with Ron Scollon

Rodney Jones

Multimodal discourse analysis

Multimodal discourse analysis examines how people deploy different semiotic modes (such as images, gestures, music and architectural layout) in communication. Scholars working in this area include Theo van Leeuwen, Kay O’Halloran, Sigrid Norris and Jay Lemke.
Useful links

Multimodal discourse analysis and digital technology, Kay O’Halloran

Theo van Leeuwen

Kay O’Halloran

Sigrid Norris

Corpus-assisted discourse analysis

Corpus-assisted discourse analysis uses techniques of corpus linguistics (such as word frequency counts, concordances and collocation analysis) in order to aid discourse analysis. Scholars working in this area include John Sinclair, Michael Stubbs and Paul Baker. 

Useful links

Lectures by John Sinclair

Michael Stubbs

Geoffery Leech

Paul Baker

Audio Listen to an audio lecture explaining the different approaches to discourse analysis

Download Download the accompanying PowerPoint slides

Return to the top

B2: More on cohesion and coherence

In Unit B2 I talked about the qualities that give a text texture: cohesion and coherence. Cohesion has to do with the way elements in a text are joined together. I introduced a number of cohesive devices, including:

Conjunction (using ‘connecting words’)
Reference (using a pronoun to refer to another word)
Repetition (repeating the same word)
Substitution (substituting one word or phrase for another word or phrase)
Ellipses (leaving something out)
Lexical cohesion (repeating similar types of words)

Some texts may make use of a lot of these devices, whereas others may use very few of them. Halliday and Hasan refer to the texture of a text as being either ‘tight’ – meaning that there are many cohesive devices – or ‘loose’, – meaning that there are few. What often determines the extent to which these devices are used is how much they are needed for readers to make the kinds of connections they need to make to understand the text.

Communication generally operates according to the principle of ‘least effort’. There is no need, for example, for me to insert the word ‘and’ after every item in my shopping list for me to know that I need to buy tomatoes in addition to buying milk. One of the challenges for people who are producing texts therefore, is figuring out what kinds of connections readers can make for themselves by invoking what they already know about the world and about this particular kind of text (coherence) and what connections need top be spelled out explicitly in the text (cohesion). 

Coherence is the way the overall structure of the text helps it to ‘hold together’.
When you were in secondary school your teacher probably taught you how to produce coherent texts by paying attention to the order of the different parts (like the introduction, the body and the conclusion). Different kinds of texts have different elements and different standard orders. Many stories, for example, have the following basic structure.

Orientation (setting and/or background)
Rising action (in which some kind of problem is introduced)
Complication (in which the problem becomes more complicated)
Resolution (like a ‘happy’ or ‘sad’ ending)

Some kinds of special stories have special structures. For example, the traditional ‘moral story’ in Chinese culture usually goes like this:

Orientation (setting and/or background)
Action (transgression)

Many texts have what we call a ‘Problem-Solution’ structure, which goes like this:

A problem is introduced
A solution or solutions are suggested
The solution or solutions is/are evaluated.

Cohesion is something that we can find in the text. But coherence is not just in the text. It is also in our minds. It depends on the expectations we have about what kinds of elements texts should contain and what order they should come in. These expectations include:

Expectations about people
Expectations about places
Expectations about events
Expectations about logic or reasoning
Expectations about what is good/bad, right/wrong, normal/abnormal

The most important thing to remember is that these expectations will be different in different cultures, groups and communities of practice. So what seems coherent to one person from one culture/group/community may seem incoherent to someone from another culture/group/community. Sometimes our ideas about coherence can cause misunderstandings which might lead to stereotyping (thinking people from a certain culture/group/community are uneducated, illogical, unreasonable or impolite).
It is also important to remember that coherence has a lot to do with ideology. We accept these standard structures or formulae for texts as natural and seldom ask where these rules came from. These expectations about how texts should be put together can cause certain people or certain ways of communicating to be excluded.

Information Structure

Another aspect of texture that we did not mention in this Unit has to do with the way given and new information is ordered in texts. We can call this the ‘information structure’ of the text. This aspect of texture falls somewhere in between the notions of cohesion and coherence. Halliday, for example, notes that even though the following two sentences are joined by both grammatical and lexical cohesion, they cannot really be considered a text:

No-one else had known where the entrance to the cave was situated. What John discovered was the cave.(Halliday 1968: 210)

The reason for this is that the information which begins the second sentence does not follow logically from information given in the first sentence.  In order to explain how information is structured in texts, Halliday begins by looking at clauses. All clauses he says, can be divided into two parts. The initial (first) element in a clause is called the topic or theme of that clause. This may be the subject of the clause, it may be the subject plus some other information, or it may be something else. The rest of the clause is called the comment or rheme. For example:

The subjects in our study (topic/theme) were all healthy, adult males (comment/rheme)

The theme is the ‘jumping off’ idea. It usually contains given information, often linked to previous clauses or sentences. The rheme (or ‘comment’) usually contains ‘new’ or ‘newsworthy’ information. Ideas at the end of a clause are usually more prominent.

Sometimes we organize information in paragraphs by using the same (or similar) themes throughout. Notice how the author of the following paragraph repeats the theme ‘Hong Kong’:

Hong Kong is a city of multiple and shifting identities. It is an international city, using the international language English, plugged into global communication networks, open to the fads and fashions of international consumerism and drawing on a range of international cultural symbols from white wedding gowns to Hello Kitty logos. It is at the same time a Chinese city, sharing with other parts of 'Cultural China' a written (and to some extent spoken) language, and other symbols of Chineseness such as lion dances, dragon boat races, temples and cuisine. And of course since July 1997, Hong Kong has also been Chinese in a political sense, as a Special Administrative Region of the People's Republic of China.

Another way we sometimes organize information is to make the theme of a clause or sentence the same as the rheme of the previous clause or sentence, as in the following example:

Lady Gaga was pleased to announce the release of her second album, The Fame Monster. This album featured her new hit single Bad Romance. The song talks about the difficulties involved in falling in love with one’s best friend. Love and friendship are common themes in Gaga’s music.

It is not necessary to use either of these two patterns, and few writers follow these patterns strictly. Sometimes we use other ways to organize information, especially when we want to create special effects in our writing (such as emphasizing certain pieces of information or contrasting one piece of information with another). 

The default (most common) thematic structure in English is to use the subject of the clause as the theme. There are many other choices as well, choices which can help you to emphasize certain parts of the sentence, make your text more organized or create other effects. All choices other than the default choice (subject as theme) are called marked themes. Marked means unusual. The most important point is that these choices are used to create particular effects, and when you use them at times when you do not want to create this effect you can confuse your reader. This is essentially what is wrong with the example given by Halliday that I referred to above.

No-one else had known where the entrance to the cave was situated. What John discovered was the cave.

Here, the construction ‘What John discovered’ is a marked (unusual) theme (called a ‘cleft’). It is usually only used when we want to especially highlight the information that occurs after the linking verb (the cave) and to background information occurring before it (John). This, of course, does not make sense in the context of the previous sentence. A more logical arrangement of information might have been:

No-one else had known where the entrance to the cave was situated. John discovered it before they did.

An example of the proper use of the marked theme might be:

No-one else had known where the entrance to the cave was situated. It was John who discovered it before they did.

In other words, what this structure does is create a contrast between the other people and John.


M. A. K. Halliday (1968) ‘Notes on transitivity and theme in English Part 3’. Journal of Linguistics, 4, pp 179-215 

Useful Links


Texture in Text: A Discourse Analysis of a News Article Using Halliday and Hasan’s Model of Cohesion, Paul A. Crane

The Textual Metafunction, Hilde Hasselgård

Return to the top

B3: Creativity: genre bending and genre mixing

Although genres normally seem to come with ‘rules’ you have to follow and being considered a competent member in a discourse community requires following those ‘rules’, ‘expert’ genre users are often able to break the rules very effectively.   In fact, using genres effectively sometimes means manipulating and sometimes confounding people’s expectations. But without the rules, without the expectations, creativity would be impossible.

At http://www.datelance.com/billboard.html you can find a good example of a text which builds upon and then appropriately breaks the rules of a genre of the dating ad which I discussed in Unit B3. This ad contains many of the same moves, especially the move of self-introduction and stating the activity (dating) which is desired, that we saw in the other dating ads. But this person, rather than posting his ad in the newspaper or on the internet, posted it on a billboard. In other words, he combined two different genres: that of a personal ad and that of a roadside billboard. We call this kind of creativity ‘genre mixing’.

If you go to the URL written on this billboard however, you will find that this is not the end of the story. This ad is in fact, not a dating ad at all, but rather a part of a recruitment campaign by the software company Logoworks. And so another layer of genre blending becomes apparent. This is not just a dating ad impersonating a billboard. It is an employment ad impersonating a dating ad impersonating a billboard.

A closer look at the website reveals yet another instance of creative genre blending in the form of Lance’s ‘personal profile’ (see http://www.datelance.com/index.html), which includes all of the moves you would expect to see on the profile of a person on a dating or social networking website such as information about Lance’s hometown, his age, interests and favourite restaurant, along with a gallery of photos of Lance. This profile page is really a clever way the company has devised to communicate to potential employees the kind of person they believe would fit in best at their company.

Useful Links

Interdiscursivity in Critical Genre Analysis, Vijay K. Bhatia

Genre Conventions, Speaker Identities and Creativity: An analysis of Japanese wedding speeches, Cynthia Dickel Dunn

Return to the top

B4: Ideology and Discourses: Big 'C' Conversations

In Unit A3 we introduced the idea of ‘capital D Discourses’ – combinations of ways of speaking and writing, ways of acting and ways of relating that promote particular ideological agendas. There are lots of different Discourses in the world and they are not all compatible with one another. The Discourse of Marxist, Leninist, Mao Zedong thought (which used to be the main Discourse in mainland China) is in many ways at odds with the Discourse of global capitalism (which is now the main Discourse in mainland China) and both of these are very mainland China during the dynastic period).

In complex societies like ours, most texts and situations are affected by more than one Discourse. Just as when two or more people come together they have ‘small c conversations’, when two or more Discourses come together in a particular text or situation, they have ‘capital C Conversations’. Understanding how these Conversations occur can help us to understand how texts exert ideological control over us.  

An example of such a Conversation occurs in this McDonald’s ad.

Video Watch the You tube clip at:

In this ad there are at least two Discourses conversing with each other. There is what we might call the Discourse of Romance, which has its roots in the Discourse of Chivalry from medieval Europe as well as the Romantic movement of the 18th century. This Discourse brings with it a whole system of beliefs, presuppositions and rules about how men and women are supposed to treat each other. The ideology of this discourse states, among other things, that human relationships should be governed by emotions, that a given person can only be ‘in love’ with one other person, and that ‘true love’ lasts forever. Practices in this Discourse include things like giving gifts or tokens of one’s affection to prove one’s love. This Discourse is so powerful that many people consider these beliefs and practices to be ‘natural’ rather than a matter of culture and history. But, you don’t have to look very far to find cultures in which these beliefs are not considered natural, cultures where people marry for financial rather than emotional reasons, for example, or cultures in which a man can have several wives. In fact, this Discourse is in many ways at odds with both the Utilitarian Discourse (which believes that human relationships should be governed by rational self-interest) and the Confucian Discourse (which believes that human relationships should be governed by duty).

What is interesting about this ad is that it does not just contain the Discourse of Romance, but also the Discourse of Economics, which is about the ways resources (such as French fries) are distributed in society. The reason this ad is effective is that the advertisers create a Conversation between these two Discourses. In this way, they are able to make a connection between an economic practice in one Discourse – buying a meal at McDonalds and sharing it with another person – and a romantic practice in the other Discourse – giving gifts or tokens of affection to prove one’s love. In other words, they are able to convince you that your boyfriend’s practices of distributing French fries is a measure of his love for you and by extension, your boyfriend’s willingness to take you to McDonalds and purchase their products is also a measure of his love for you. Don’t believe it. If you have a boyfriend who only wants to take you to McDonald’s, dump him.

Another Sample Analysis

Lips Inc.
Follow this link to see a sample analysis of a text from a woman’s magazine promoting lipstick:


Useful Links

Critical Discourse Analysis of Ads, Rick Beach

Return to the top

B5: The texture of talk

In Units A5 and B5 I talked about principles of pragmatics and conversation analysis and how they explain how we make sense of conversations and how conversations hold together as coherent texts.

Pragmatics explain conversation coherence based on the kinds of expectations people bring to interaction, chief among them being that they expect people to be cooperative.

For example, although there is nothing grammatical or lexical that connects the two utterances below, most hearers would assume that they are somehow related and proceed to make inferences about their relationship (the most logical being that B is unwilling to answer the door).

A: Someone’s at the door.
B: I’m in the bath.

That is to say, when conversations do not seem coherent, people naturally try to make them coherent by making inferences about the relevance of one utterance to another.

There are times however when conversations may seem coherent, but the relationships between utterances that speakers assume turn out not to be true. In the book I gave the example of the conversation in The Pink Panther Strikes Again between Inspector Clouseau and a hotel clerk. You can watch a clip of this conversation below:

Video ‘Does your dog bite?
Watch the You tube clip at:

Indirect Speech Acts
As I mentioned in Unit B5, people often express speech acts indirectly. That is, the locutionary force of the utterance is very different from the illoucutionary force. In order to understand such indirect speech acts, we need to refer to the conditions under which they are produced. In the video below Ben Loka gives some possible interpretations of common indirect speech acts. Here are some of the examples he gives:

Sentence meaning

Possible speaker’s meaning

I’ve got a really sore throat and my head hurts.

I require sympathy.

Excuse me, do you know the time?

I really just want to talk to you, that’s why I’m asking.

Ha, ha. This video is really funny.

I didn’t really watch your video, but here’s a comment that showed that I tried.

It’s a really nice day out. Have you seen the sun?

I have nothing to talk about, but I can’t stand silence.

Video Speech Acts (Ben Loka)
Watch the You tube clip at:

Examples of the analysis of conversations

Examples of Conversation Analysis, Charles Antaki



One of the most important aspects of the texture of talk from the point of view of conversation analysts is the mechanism of turn-taking in talk – how we decide who gets to talk and when. Turn-taking entails:

  • who may speak (take the speaker role)
  • when they can begin to speak
  • how they can achieve this right
  • how they can hold on to this right
  • how they can relinquish this right (freely or otherwise).

We refer to the right to speak in interaction as 'the floor'. Rules of turn-taking tell us how to 'get the floor', to 'hold the floor' and to 'give up the floor'. Generally, the person who is speaking has the most rights over the floor. They can usually hold the floor for as long as they want, can select who will speak next, and can constrain the next turn by controlling the topic. 'Getting the floor', 'holding the floor' and 'giving up the floor' involves a whole series of signals, some of which can be rather subtle. The most common signal that someone is ready to give up the floor is pausing. Other signals include falling intonation (in English) and looking at one’s interlocutor. People also have various signals for when they want to take the floor, including breathing in to show they are about to talk and tensing their bodies.

Conversational coherence depends a great deal on the mechanics of turn taking functioning properly. When it breaks down, confusion can occur. One example of this results from the fact that people from different cultures are used to pauses of different lengths. People in New York for instance, tend to use very short pauses between turns, whereas people from most other parts of the United States use longer pauses. So when a New Yorker is talking to, say, a Californian, the New Yorker might think that the other person is giving up the floor after a very short pause, when actually they may just be pausing to take a breath. This is why many people in the United States consider New Yorkers rude and aggressive.

Interruption is another important and often misunderstood part of interaction. Many people think that people interrupt in order to get the floor and that interrupting is rude. Actually, most interruptions are what we call 'cooperative interruptions’ – they are not meant to take the floor away from the speaker but rather encourage the speaker to continue talking.


In my discussion of cohesion in Unit B2 I noted that repetition of words or phrases is a common cohesive device in written texts. It is also common in spoken conversations. In fact, it is even more common.

In her book Talking Voices (2007), Deborah Tannen argues that in casual conversation people regularly and frequently repeat themselves and other people. An example of such repetition can be seen below:

Lori: Would you consider yourself a dog person or a cat person?
Michael: Oh, I’m a dog person, I can tell you straight away.
L: Really?
M: Absolutely, 100% confirmed, dog person. You bet.
L: [laughter] Dog person… Have you ever had a dog?
M: Yes, I have. Yes, I had — let me see, this would be about 10 years ago. I had my own doggy; I had a Norwegian Elk Hound.
L: Oh, was he cute?
M: He was very cute; he was painfully cute. He was great. But I don’t have a dog nowadays, but I have…I walk my neighbors’ dog.
L: Oh right, yeah, Bertie.
M: That’s right, I was telling you the other day. They’re a lovely old couple, who I’ve known for years, but they’re getting on a bit and they’re not so active, so I take their dog out for a walk quite often.
(from: http://hw.libsyn.com/p/9/9/e/99e956f80efbfdef/0201_dog_doo.pdf?sid=111b416cbdaa7e040ecf2387165035f4&l_sid=23550&l_eid=&l_mid=2134506)

According to Tannen, such repetition serves a to make conversations coherent not just because the meanings of words are repeated, but also because repetition sets up a kind of rhythm in conversation in which the phonological patterns produced by certain words or phrases are repeated (see Companion Website material B9 – coherence in multimodal texts). Other functions of repetition in conversations are:

  • It aids production by giving people a chance to think about what to say next.
  • It aids comprehension by making conversation less lexically dense.  
  • It helps to divide conversations into ‘chunks’.
  • It gives you a sense that the other person is ‘with you’ (involvement).
  • Expanding on what was previously said.
  • It can be used by speakers to hold on to the floor.

Tannen, D. (2007) Talking voices: Repetition, dialogue, and imagery in conversational discourse. 2nd Edition. Cambridge: Cambridge University Press.

Useful Links
Schematic Analysis of Adjacency Pairs

A Practical Example of the Application of Conversation Analytic Concepts, Pascal Belouin

Return to the top

B6: Face threatening acts

Any act that has the potential to threaten somebody's face (whether it be their negative face or their positive face) is called a 'face threatening act' (FTA). Actually, nearly every social action is potentially face threatening. If you insult somebody (or ignore them) you can threaten their positive face (their desire to be liked), and if you smile at someone, show them concern or invite them to the cinema, you can threaten their negative face (their desire to be left alone). So, every time we encounter someone we are faced with a number of decisions about how to manage our respective 'face threatening acts.' 

One choice we have is to not do the face threatening act at all (to avoid asking the girl to the party, to pretend that you didn't hear when your friend asks you how you like her new hairdo). If we choose to perform a face threatening act, we are faced with another choice. Either we can do it 'directly' (baldly) (for example, 'your new hairdo is horrible!' or, in the case of a threat to negative face, 'kiss me!'), or we can do it 'off-record' (pretend that we are not doing it), or we can use the kinds of politeness strategies like those in Unit B6.

It should be clear from the above that politeness is not just about 'being nice'. It is about managing the power relationships you have with others, the degree of closeness you enjoy with them, and the way you exchange goods and services with each other. It should also be clear that within these systems and strategies which we have to figure out and choose from constantly in the ongoing flow of interaction  there is a lot potential for error and misunderstanding. 

Audio Listen to an audio lecture on politeness

Download Download the accompanying PowerPoint slides

Framing and face threatening acts

Face strategies also contribute to the management of conversational activities (especially those involving ‘face threatening acts’) and framing strategies are often central to the discursive construction of identity. Just as face strategies of involvement and independence, while primarily providing information about relationships, also give clues as to what we think we are doing and our attitude towards it, framing strategies, while primarily signaling what we think we are doing, also play an important role in managing relationships.

In an article called ‘Talking the Dog’ (2004) (see Useful Links below) about how people use pets to frame and reframe their utterances in interaction, Deborah Tannen gives the following example of a conversation between a woman, Clara, and her husband, Neil, in the presence of their dog, Rickie.

Clara: You leave the door open for any reason?
((short pause, sound of door shutting))
—> Rickie,
—> he’s helpin burglars come in,
—> and you have to defend us Rick.> 
Tannen 2004:413)

In this example, Clara shifts frames from talking to her husband to talking to the dog by altering her voice quality (adopting the high pitched and playful tone of ‘baby talk’). In a sense, though, she is still talking to her husband, communicating to him ‘through’ the dog the potential seriousness of leaving the door open. By addressing her remarks to the dog, however, and by adopting a different tone of voice, she shifts the frame from scolding to playing, allowing her to get the message across without threatening her husband’s face.

Audio Listen to an audio lecture on interactional sociolinguistics

Download Download the accompanying PowerPoint slides

‘Talking the Dog: Framing Pets as Interactional Resources in Family Discourse’, Deborah Tannen

The pragmatics of cross cultural communication, Deborah Tannen


Outline of Gumperz’s Discourse Strategies

Conversational Terrorism

Return to the top

B7: Examples of Ethnographies: Drug Users

Ethnographic approaches to discourse have been used to study the language and communication practices of a wide range of different cultures, sub-cultures and speech communities. Some of this work has been done in communities outside the mainstream of society. Work with these groups can be interesting in showing how norms of communication function to regulate group membership and participation when it comes to speech events that are sometimes secretive or illegal. They also present ethnographers with a difficult methodological dilemmas since access to such speech events is often difficult to negotiate and sometimes dangerous. Sometimes researchers are forced to rely on methods other than direct observation to obtain data.

One example of such research is Michael Agar’s ethnography of urban drug addicts reported in his book Ripping and Running (1973). During his research, Agar interviewed addicts in a rehabilitation centre and asked them to perform simulated situations of the various speech events associated with taking drugs. One thing he found was that speech events around the taking of heroin tended to be ‘ritual events’ in which participants focused a great deal on things like the sequence of acts, the key and the setting. He also found that different speech events were inseparably bound to each other so that competence in one required competence in others. The primary speech event in the lives of heroin addicts (‘junkies’) is ‘getting off’ or injecting heroin in a specific place in a specific ritualistic fashion. Successful participation in ‘getting off’ however, requires participation in another ritualized event known as ‘copping’ or buying drugs from a dealer. Obtaining the money (or ‘bread’) to purchase drugs depends on competence in yet another kind of speech event known as ‘hustling’.

Another example of such a study is my own research with adolescent drug users in Hong Kong (Jones 2005) in which I asked participants to make videos about their drug use and subsequent recovery. One of the things I observed was that just as the life of an active drug user involves the learning of a special language and the mastery of various rituals, so does participation in a recovery programme. One important genre in sharing sessions in recovery programmes is the ‘recovery story’, which involves its own specialized lexicon and sequence of events. Below is an example of such a story from one of my participants:

I’m (name). I’m now twenty-one years old. Well, I began to take drugs a long time ago and I’ve been doing it for about five years. However, I never thought that I could be here ... to learn some video techniques. Well, I never thought about it. As a drug addict, I just looked like a beggar who was always squatting on the roadside. Although I am young, only twenty-one ... I took drugs and at last my family, even my friends and relatives began to give me up. They just thought that I could not be saved. Meanwhile, I thank God for letting me have the chance to know him. He found me again and again although I had left him. God did not give up on me... he loves me very much. As the Bible says, David left God many times. But David wrote in a Psalm: ‘You take me back anytime I leave you, please hide my tears into your leather bag.’ Now I ask God to hide my faults, my tears into his leather bag, so that I can be honest and brave enough to stand to face his judgment. Well . . . I . . . wish I can be . . . although I don’t know tomorrow’s path, I deeply believe . . . with the lead of God, I will not be afraid no matter what problem or things I face. Even if my friends and relatives lack the confidence in me, I will not be afraid since I know that God is my power and shelter. Well, I wish all my glory and praise belong to him. Is it one minute already? (OD1:175/263)1

Like the other recovery stories in my data, this one follows a particular format, beginning with an introduction followed by a description of the number of years the teller had used drugs and often what kinds of drugs he had used, followed by the portrayal of a ‘decline’, in which the increasingly negative consequences of drug use are depicted, a description of ‘hitting bottom’, a moment in which the narrator’s drug use reached a crisis point, a description of ‘surrender’, involving a decision to seek help or ‘give their lives to God’ and finally, an assessment of the future, marked by some degree of uncertainty and a strong dependence on outside forces (usually God or family members).

Membership in groups of ‘recovering drug users’ was in many ways contingent on participants’ competence in the speech event of the sharing session, which depended on their mastery of the genre of the ‘recovery story’. In a sense, then, drug use and recovery from drug use are not just a matter of physical behavior and biological addiction, but also a matter of social identity and competence in the key speech events of various speech communities.


Agar, M. (1973) Ripping and running: A formal ethnography of urban heroin addicts, Seminar Press.

Jones, R. (2005) ‘Mediated addiction: The drug discourses of Hong Kong youth’, Health, Risk and Society 7 (1): 25-45.

Audio Listen to an audio lecture on the ethnography of speaking

Download Download the accompanying PowerPoint slides

Useful links

The Ethnography of Speaking, Dell Hymes

Summary of SPEAKING model

Verbal Communication and Research, James R. Baldwin

Ethknoworks, Michael Agar

Return to the top

B8: Affordances and constraints

Lately I’ve become fond of listening to audio books on my iPhone. One of the affordances of audio books is that they allow me to catch up on my ‘reading’ while I’m walking down the street or working out in the gym. But there are also some constraints involved. For example, it is much more difficult to search through an audio book to find important information in the same way I can do with a printed book (this affordance is even greater if I’m reading the printed book in PDF form on my computer and I can search through it by typing keywords). It is also impossible to highlight things with my yellow pen when I’m listening to my audio book.

Here’s a similar example. Lots of people like to read the newspaper on their iPad nowadays. There are a number of actions that iPads make easier. But there are some actions that paper-based newspapers make easier and which if you try to do with your iPad, you’ll regret it. The video below illustrates my point:

Video Watch the You tube clip at:

How to do mediated discourse analysis
Different kinds of discourse analysis have different purposes. The aim of conversation analysis for example, is to understand how people make their interactions orderly, and the aim of genre analysis is to understand how different text types help to define and sustain different discourse communities. The goal of mediated discourse analysis is to help us to understand the role of discourse in social action and the extent to which different kinds of discourse make some actions and social identities more possible and others less possible. In other words, it helps us to understand how discourse can affect what we can do and who we can be.
Ron and Suzanne Scollon call the methodological procedures of mediated discourse analysis nexus analysis. They divided nexus analysis into three stages:

  • Engaging the nexus of practice
  • Navigating the nexus of practice
  • Changing the nexus of practice

Engaging the nexus of practice involves choosing a site of engagement you are interested in and doing some preliminary research including informal observations, interviews, and reviews of documents and other textual artifacts. The purpose of this stage is to answer the most basic question that is at the heart of mediated discourse analysis: ‘what’s going on here?’ or, to put it another way:

  • What are the important actions for this group of people in this particular site of engagement?

Navigating the nexus of practice involves exploring the role that discourse plays in these actions and how this affects things like power and social identity. The key questions for this stage of the research are:
What is the role of discourse (and other cultural tools) in these actions?

  • How do the affordances and constraints of these cultural tools amplify or constrain people’s participation in these actions?
  • How do people use these actions to claim and impute membership in different social groups or communities of practice?
  • How are these actions linked together to form social practices?
  • How do these social practices help define group membership?

Like critical discourse analysis, mediated discourse analysis has an activist agenda. In other words, it sees its role as not just to describe the world, but also to make it better. One way to do this is to understand something about how people creatively engage with discourse, adapting and remixing it in ways that expand their scope of social action. Another way is to help people become more aware of their actions and the consequences of those actions moment by moment. The key questions for this stage of the research are:

  • To what extent are the different cultural tools in the site unequally distributed and what are the consequences of this?
  • How do people adapt and mix tools creatively to overcome constraints and change existing power relationships?
  • How can people in this site be made more conscious of their actions and the way these actions are affected by cultural tools and by the social practices enforced by the communities participate in?

Audio Listen to an audio lecture on mediated discourse analysishttp://www9.english.cityu.edu.hk/ebs/rodneyjones/discourse/Mediated_Discourse_Analysis.mp3

Download Download the accompanying PowerPoint slides

Useful links

Mediated Discourse Analysis

Mediated discourse analysis: researching young children’s non-verbal interactions as social practice, Karen E. Wohlwend

Georgetown Working Papers on Language, Discourse and Society, Special Issue in Honor of Ron Scollon

Return to the top

B9: Analyzing multimodal interaction

One problem with analyzing multimodality in face-to-face interaction is that the spatial boundaries of interactions are not always as clear as the spatial boundaries of texts. While the frame of an image clearly marks what should be considered as belonging to the image and what should be considered external to it, a conversation in a coffee shop is not so clearly bounded. In analyzing such an interaction, how much of the surrounding modes should be taken into account? Should the analyst consider for example, the signs and posters on the walls, the conversations occurring at other tables, the ambient music playing over the P.A. system and the sounds of milk being steamed? What about the smell and taste of the coffee?

Sigrid Norris and her colleagues deal with this problem by taking an action oriented approach to multimodal analysis. Modes themselves are seen as cultural tools that social actors deploy to accomplish specific actions. Deciding what modes to analyze is a matter of deciding what modes are important to social actors themselves.

Modal Density

Central to this determination is the concept of modal density. In different interactions and at different moments in a single interaction, some modes are more salient than others. In other words, some modes are more important for performing the particular action that the people involved are engaged in.

According to Norris, modal density is achieved either through modal intensity (the degree to which a mode takes on primacy in an interaction to the extent that the action in question depends on the deployment of this mode) and modal complexity (the degree to which a mode is integrated and interdependent with other modes).

One way to visualize modal density is through what Norris calls modal density circles.

In the picture below, taken from the interaction I analyzed in the book, the relative intensity of the different modes the actors are making use of are represented by circles of different sizes. As you can see, different modes are being deployed by the tutor (on the right) and the student (on the left). The most salient modes for the tutor are spoken language and object handling (her handling of a pen to make corrections on the student’s essay). In fact, these two modes are not only the most intense, but they are also interdependent upon one another as the tutor explains the corrections that she makes as she makes them. These two modes working together, and in conjunction with other modes like gaze and head position, constitute for this tutor the higher-level action of ‘giving the student feedback on his writing’. That is not to say that this combination of modes always constitutes this higher-level action or that there are not other ways to perform this action (see below). The point is that this particular configuration of modes represents how the tutor sees teaching: teaching is a matter of explaining to a student what he has done wrong and correcting it.

The student deploys much fewer modes than the tutor in this situation, which tells you something about the amount of power and control he has. It also tells you something about how he understands the higher level action of learning: learning is a matter of sitting quietly and listening to the teacher, deploying the modes of head position and gaze to display listenership.

Figure 1

Figure 1

In Figure 2, depicting a different tutor and a different student in the same writing centre, we can see a very different configuration of modal density. In this picture, the person on the left is the tutor, and the person on the right is the teacher. In this picture, the modes of written text and object handling which were so prominent in the example above are not so important. They are not absent—the student’s essay is on the table between the tutor and the student and the tutor is holding a pen—but they are not being deployed at this particular moment in the interaction. Instead, the dominant mode is spoken language, and this mode is being deployed by the student. The tutor on the other hand, uses gaze and gesture (holding his hands together in a prayer like fashion) to display listenership.

The configuration of modes in this particular moment of interaction constructs the higher-level actions of teaching and learning in a very different way. In this case, teaching is constructed as listening and learning is constructed as an active process of explaining one’s ideas.

Figure 2

Figure 2


Useful links

Sample multimodal transcription, Sigrid Norris

Multimodal Research Centre, Auckland University of Technology

Multimodal Discourse Analysis: Media, modes and technologies,
Odysseas Constantinou

Return to the top

B10: Other techniques in corpus assisted discourse analysis

It should be clear from the material in Unit B10 that corpus assisted discourse analysis always involves strategic mixing and different analytical procedures and techniques. Most importantly, it always involves the combination of qualitative and quantitative approaches. Researchers often start with the quantitative analysis (looking at frequency and distribution), then move to a more qualitative approach examining concordances for patterns and isolating sample texts or excerpts to analyze, using more conventional tools of discourse analysis.

Making corpus comparisons

Some procedures in corpus assisted discourse analysis such as keyword analysis requires that the main corpus under consideration be compared to another corpus, know as a reference corpus. Some analysts, however, make the comparison between two corpora more central to their analysis in order to answer specific questions about discourse. There are many kinds of corpus comparisons one can make. For example:

  • comparing corpora of two different genres
  • comparing corpora of the production of two different speakers/writers or groups of speakers/writers in the same genre.
  • comparing corpora of texts from two different groups with opposing views on a particular issue.
  • comparing corpora from the same group of speakers/writers or on the same topic from different time periods

The Web as Corpus

In a way, the World Wide Web is just like an immense corpus, and some scholars have considered ways it might be analyzed in the same way other corpora are. There are of course problems associated with this. One is that search engines like Google operate using different kinds of algorithms that corpus analysis tools do. Another is that the web is in some ways like a giant echo chamber: people constantly cut text from one site and reproduce it on other sites. This phenomenon introduces problems for the accurate measurement of things like word frequency and ‘keyness’. Some scholars have tried to solve this problem by, rather than treating the web as a corpus, creating corpora from representative texts taken from the web (see Useful links).

Dispersion plots

In Unit B10 I mentioned dispersion plots but did not go into too much detail about how they are used. Generating dispersion plots for different words or phrases in texts can be an extremely good way to understanding something about the discursive structure of texts. With this technique, one or more features are tracked through a text to determine how the feature/s contribute/s to some aspect of discursive development such as the progression of topics, the arrangement of ‘moves’, the order of information, or the rhetorical organization. Most software packages produce a graphic representation of the occurrence of the feature through the text. With this method, multiple texts can be compared in order to search for consistent or divergent patterns.

Statistical analysis

Some people who use corpus assisted discourse analysis submit their results to various kinds of statistical analysis. One of the key statistical methods in corpus linguistics is called multidimensional (MD) analysis. This method uses factor analysis to analyze co-occurrence patterns of different features in a corpus. This form of analysis has been used to examine stylistic differences between texts of different genres or different registers.

Useful Links

Basic statistics for corpus analysis, J.B. Jenset

Using the Web as a corpus

Pragmatics/ text linguistics: dispersion plots, Stefen Gries

Return to the top