What is Discourse Analysis?

Front matter

RESEARCH METHODS SERIES

What is Discourse Analysis?

Stephanie Taylor

Contents

Series foreword

Acknowledgements

1 Introduction 2 Theories and common concerns 3 Four examples of discourse analysis

Introduction

This book has been written as an introduction, to explain and demonstrate some of the possibilities which discourse analysis offers for contemporary social research. It is for novices as well as experienced researchers in the social sciences and related disciplines who want to read about other people’s discourse analytic research and perhaps conduct their own similar projects. The two major themes of the book are as follows: discourse analysis does not refer to a single approach or method, and its varied forms cannot be distinguished without an understanding of their premises. The aim is therefore to help readers understand the connections between the theoretical assumptions which a discourse researcher makes about the social world, the research topic or problem being investigated, the data chosen for analysis and the aspects of the data which are studied. Any research project represents a unique combination of these features, and indeed part of the attraction of discourse analysis is the wide range of possibilities which it encompasses.

Chapter 1 develops an initial definition of discourse analysis and then outlines the content of the subsequent chapters, with some suggestions for using the book effectively as a practical guide, and building on its content.

What is discourse analysis?

This question is difficult to answer succinctly because the term ‘discourse analysis’ refers to a range of approaches in several disciplines and theoretical traditions. Discourse analysts in sociolinguistics, sociology and social psychology, to mention just a few possibilities, are likely to differ in the sources they refer to, and also, to some extent, in the problems and research questions which they set out to investigate. This variety is potentially confusing, but also positive in that it expands the possibilities for new researchers and projects. This book provides an overview of discourse analytic research as a rich and interdisciplinary field which continues to change and develop in new directions. Later chapters will discuss some of the theoretical background and examples of research studies. This section begins a discussion on the meaning of the term discourse analysis.

One starting point is that discourse analysis usually refers to a research approach in which language material, such as talk or written texts, and sometimes other material altogether, is examined as evidence of phenomena beyond the individual person. To understand this, imagine looking at some old letters, written several decades or even centuries ago. Each letter will of course be interesting for what it conveys about the writer’s situation, opinions and feelings. However, it can also provide more general evidence of society at that earlier time. For instance, passing references may suggest what is taken for granted, including the priorities and values shared by members of society. Some of the words used may even be offensive to a contemporary reader, for instance, because they are linked to assumptions about class or gender or race which have since been questioned. In addition, some of the writing may seem ‘old-fashioned’ in its style and level of formality because there have been changes in the directness with which people express opinions or state disagreements. If letters between both parties to the correspondence have survived, they will provide further insights into how people communicated at that time, including the conventions which operate in particular relationships, business or personal. In short, each letter, however private its original purpose, is potentially of interest as evidence of social phenomena, in a way that the writer could not have anticipated. This is the level of interpretation employed by a discourse analyst.

To explain the variety of discourse analytic research, it is useful to distinguish two lines of academic work. They do not encompass all the variations in discourse analysis and in practice, they tend to converge particularly around the study of difference and inequality and the workings of power in society, as some of the studies discussed later in the book will indicate. Nonetheless, the distinction is helpful as a starting point.

The focus of the first line of academic work is the nature of language. Traditionally, linguistics, or ‘linguistics proper’, can be defined narrowly as ‘the study of “grammar” in a broad sense: the sound systems of language (“phonology”), the grammatical structure of words (“morphology”) and of sentences (“syntax”) and more formal aspects of meaning (“semantics”)’ (Fairclough, 20012, p. 5). However, sociolinguists and many other academics consider language as inseparable from its social contexts. Among many

Introduction 3 other aspects of language use, they study differences in how people speak (and write) which are linked to class and other social categories, or to a particular activity, situation, role and purpose. Summarized somewhat crudely, this line of writing and research can therefore be said to have extended from the concerns of ‘linguistics proper’ to explore features of language linked, again, to social phenomena. Discourse analytic research in this line includes investigations of the details of how language varies across contexts and can mark social difference, and how children acquire competence in language use and, again, how that competence is linked to identity and social difference.

The second academic line can be said to have originated in the study of society and people as social beings, especially in sociology and social psychology, and then developed to incorporate a focus on language. In other words, the move has been in the opposite direction, from social phenomena to language. For example, discourse researchers have analysed public and private language use as a way of accessing the collective, though not necessarily coherent, ‘world view’ of a society. Some aspects would be the ways in which people and their activities are categorized, valued and located in relationships of dominance and subservience. Language use may also be analysed as one activity or practice (some would say, the most important practice) which people engage in as part of their ongoing social lives and relationships. Through the analysis of language and language use, the researcher therefore builds up a picture of society and how it functions.

The variety in discourse analytic research is also partly given by the kinds of data used by researchers. Discourse analysts can study other forms of representation from language, such as pictures and film, or consider language use alongside other practices, but most work with some kind of language data (the focus of this book). Some researchers investigate historical material, like the old letters already mentioned, but most take as their data contemporary material related to language and communication. As already noted, this is analysed following the principle that language provides evidence of social phenomena. For example talk, perhaps from focus groups or people participating in interviews, will be of interest not (just) as straightforward reporting, like witness statements. (Many analysts will also avoid interpreting it as a direct report of the speaker’s inner thought processes and feelings.) Instead, the analysis may focus on how the interac tion of speakers is shaped and constrained by its social contexts, from the immediate conversational interaction to the larger situation of the place, time, state of the nation and so on. The analysis may explore the styles of speaking, and the ‘to and fro’ of the talk. It may consider the details of how the talk is constructed, including grammatically, and the implications of the choice of one possible word or structure over an alternative. It may consider the functions of the talk, or examine the assumptions which apparently underlie what is said. Words, expressions and arguments may be analysed as social and cultural resources associated with certain social groups and contexts, recycled and modified by individual language users in particular instances of communication. The analysis may explore more dramatic aspects of the talk, for instance, when someone speaks ‘as’ a certain character or role (a concerned parent, an authoritative expert) or appears to repeat a well-rehearsed anecdote, or address an audience beyond the immediate situation. Some of these foci are associated with particular named approaches to discourse analysis, but in many or even the majority of studies, the features to be studied will be decided according to the priorities of the research, drawing on the wide range of possibilities given by previous published work.

The remainder of this book looks more closely at some of these possibilities. To assist the exploration, it will be useful to give an initial answer to the question ‘What is discourse analysis?’ as follows: Discourse analysis is the close study of language and language use as evidence of aspects of society and social life. Later chapters of the book return to this definition and extend it.

The chapters

Chapter 2, ‘Theories and common concerns’, presents an overview of the main theoretical traditions relevant to discourse analysis in the social sciences. Although the focus is on key ideas, not individual theorists, the chapter does introduce some of the ‘names’ readers may encounter in their reading of other sources. It is intended as a starting point for further reading which researchers can build on as they follow their own interests and develop new projects.

Chapter 3, ‘Four examples of discourse analysis, discusses four recent articles published in academic journals, each of which presents findings from a discourse analytic study. The examples have been chosen to demonstrate the variety of discourse analytic approaches and also of

Introduction 5 the kinds of research problems they have been used to address. Together, the four articles refer to a range of subject fields, project designs and types of data. The first article, by Kirsten Bell, is based on a study conducted with people attending cancer support groups; the second, by Jovan Byford, on an investigation of political rhetoric; the third, by Elizabeth Stokoe, on a close analysis of talk from neighbours engaged in disputes; and the fourth, by Ruth Wodak, Winston Kwon and lan Clarke, on a study of meetings in a business organization. The four articles variously discuss racism and prejudice, health, risk and personal responsibility, morality and gender, and leadership and consensus. The chapter provides an overview of each research project, its theoretical grounding, the empirical work and the data which were analysed, and the discourse analytic or discursive approach which the researcher has adopted.

Chapter 4, ‘The usefulness of discourse analysis for social science researchers’, begins with a general discussion of the reasons for using this form of research and for analysing language data. The chapter also discusses two ‘half-reasons’ which are based on a partial misunderstanding of the premises of a discourse analytic approach. The chapter then describes some of the different kinds of data which can be analysed discursively and considers practical aspects of obtaining discourse data. The chapter includes sections on the collection of new data, the selection of already existing or ‘found’ material as data and the production of transcripts.

Chapter 5 considers some of the problems or challenges faced by the discourse analyst. These include deciding on the right discourse analytic approach, and beginning to analyse data and develop an interpretation and argument. The discussions of interpretation and analysis continue the practical guidance provided in Chapter 4.

Every research approach has its critics and Chapter 6 addresses some common criticisms of discourse analysis. For example, the chapter answers the challenges sometimes made that discourse analysis is out of date as an approach or is ‘just’ about words, or that it does not take enough account of people or has limited practical applications. The chapter also discusses some more specific criticisms, such as that discourse analysts should not collect data through interviews.

Chapter 7 summarizes the key points covered in previous chapters and presents a list of suggestions and references for readers who would like to build on the practical introduction to discourse analysis which is presented in this book. The book also includes a glossary of key terms.

Using this book

Like other text books, ‘What is discourse analysis?’ has been written as a guide and reference tool. Readers are likely to turn to chapters and sec tions which seem pertinent to their immediate concerns and practical problems. However, the book has also been written to be read from start to finish. The concepts and terms which are introduced in Chapter 2 are referred to in the very detailed discussions of empirical studies presented in Chapter 3. The discussions in Chapters 4 and 5 refer back to those studies, and to previous chapters, and the arguments presented in Chapter 6 presume some awareness of theories and concepts introduced earlier in the book. Readers are therefore recommended to work through the chapters in order, at least initially. It will also be useful to refer to the glossary of key terms at the end of the book.

Chapter 1 has introduced the topic of the book, offered an initial answer to the question ‘What is discourse analysis?’ and outlined the contents of the remaining chapters.

Introduction

Chapter 1 proposed a definition of discourse analysis as ‘the close study of language and language use as evidence of aspects of society and social life’. This chapter explores the theoretical underpinnings of discourse analysis and some of the main concerns of researchers, including the different terms they use. The first three sections will discuss theories which establish a connection between language and social phenomena and therefore inform discourse researchers’ arguments for the status of language data as evidence. The fourth section discusses some relevant terminology, including different definitions of the key term ‘discourse’. The fifth and sixth sections introduce some of the most important concerns which underlie discourse analytic research in different traditions, and the seventh section outlines the links between discourse analysis and social psychology.

To introduce the first two sections, it may be useful to return to the example of letters, introduced in Chapter 1. The discussion there of course assumed that readers understand what is meant by ‘a letter’, even though the use of letters is increasingly superseded by electronic communications. A formal dictionary definition, from the Shorter Oxford Dictionary (1965), is that a letter is ‘a missive in writing, an epistle’, a definition which might prompt further consultation of dictionaries. A simpler definition is that a letter is a communication on paper which is conventionally sent through a postal system. This probably explains the ‘meaning’ of a letter sufficiently for someone to understand what the word refers to, for instance, in order to translate it from one language to another. However, it does not encompass the kinds of meanings which might be of interest to a social researcher. Consider, for example, the difference between using a letter or an email. A letter would probably seem more appropriate to an invitation for a special occasion. Similarly, in a personal relationship, especially one in its early stages, a love letter would probably carry greater significance than a declaration of love in an email. More negatively, in a correspondence about a dispute, perhaps in a workplace or between neighbours, sending a letter might raise the issue to a new level of seriousness. In short, a letter can carry meanings of formality and occasion (the invitation), commitment and lastingness (the love letter), seriousness and legality (the dispute). These meanings are cultural, made familiar through the many experiences of being part of a society, including seeing the post delivered, receiving letters and sending them, and reading or viewing stories in which letters mark key dramatic points. Additional meanings will attach to details like the quality of the paper, whether the letter is printed or handwritten and whether it is addressed to a named person or merely ‘The Occupier’. And there are many other associations, of emotion and of images, such as a bundle of letters tied with red ribbon, a card decorated with balloons and so on.

These meanings and associations are not fixed. They have accumulated over time and may now be fading because of the increasing use of electronic communication. In addition, the meanings of particular letters will depend on their context, including the context of a time (today, 20 years ago or a few years into the future), the purpose of the letter, and the relationship between the sender and recipient(s). The relevant point for this discussion is that a letter is not just a letter but potential data for a social researcher. It may provide evidence relevant to the exploration of society and social phenomena such as celebrations, or changes in intimate relationships, or the settlement of disputes, or institutionalized communication or the legal profession: the exact focus will be given by the topic which the researcher has chosen to investigate.

This example need not be elaborated any further but it introduces the three approaches to meaning which will be discussed in the following sec tions, in relation to language. The first is the exploration of social meanings as systems or aggregates, accrued over time. The second is the study of meaningful practices, such as inviting, declaring, threatening and also writing, signing, witnessing, consulting a lawyer, entering a contract and many more. The third concerns the use of language for the communication of meanings. All three sections adopt a broad approach and at some points juxtapose theorists who are usually associated with separate traditions of academic work. As indicated in Chapter 1, the aim is to introduce the field and enable readers to begin pursuing lines of interest to be followed up in further reading.

Theorizing language as systems or : aggregates of meanings i

The example of the letters introduced the notion that there are aggregates or groups of meanings which ‘hang together’ by association rather than logical links. The theorization of such systems of meaning is strongly informed by the work of the French philosopher and social theorist Michel Foucault (1926-84). His work is perhaps less academically fashionable than it was two or three decades ago, but its influence persists, including in work which may not acknowledge it directly. Foucault’s writings are too wide-ranging and complex to be summarized easily. His own research was primarily historical in the sense that he attempted to unravel the foundations of contemporary knowledge by tracking connections between ideas back through a variety of documentary sources. For example, in one famous study, he suggested that madness acquired some of the stigma of leprosy partly because of the chance circumstance that the same buildings which had accommodated leprosy suffers were later used to incarcerate people who were mentally ill (Foucault, 2001). He argued that although the connection was subsequently forgotten, the negative association was transferred from one set of sufferers (lepers) to the other (people with mental illness). It influenced the treatment of the mentally ill and persisted into contemporary thinking, given fortuitously by history yet having important social consequences for social attitudes and practices around mental illness.

This example also illustrates a second premise: such an aggregate of ideas and meanings or, in Foucault’s term, ‘discursive formation’ is a form of knowledge which is inextricably linked to the workings of society. To understand this point in relation to the example mentioned above, consider, first, the difference between the terms ‘mad’ and ‘mentally ill’, and second, how the categorization of people as mentally ill inevitably involves a chain of connections. The authority of the experts who can ascribe such categories is linked to a justification for treating supposedly mentally ill people in certain ways, the traditions and institutions which provide the ‘knowledge’ underpinning the justification and the power structures and institutions through which the treatment is administered. The words and terms used in various contexts, officially and colloquially, are part of the chain; the language is inseparable from the social phenomenon.

The notion of the social world as shaped by historically accrued, partially contingent knowledge and language has further implications. It inverts the commonsense assumption that knowledge, recorded or unrecorded, follows faithfully from the details of the world which pre-exists it, like a description of what ‘is’. A classification like ‘mad’ may even be seen to produce what it purportedly describes (a person different to most others?). This is a challenge to the possibility of objective knowledge and it points to a further premise: people do not necessarily learn by observing. Rather, what they see is largely determined by already existing, socially circulated knowledge. This point undermines the concept of ‘truth’, changing its status from what ‘is’ (or was) to what is generally believed and accepted. There can be more than one truth! It also implicates knowledge with power since the workings of the social world will be closely connected to the accepted truths of that world, such as which people ‘are’ right, or best, or most important, or deserving of reward, and which have the power to dictate or influence the accepted truths. This is discussed further in the section ‘Knowledge, Truth and Power’.

This section has indicated one aspect of discourse analysis, that is, the study of language as evidence of a system or formation of meanings and the connections of those meanings to society, including the power relations within society. However, such an analysis can be criticized as static because it maps meanings onto language as if they do not change. This potentially conflicts with two other premises which are particularly relevant to the analysis of talk. The first is that meaning is ‘indexical’, that is, it depends on and changes with context. A simple example is provided by the reference of a pronoun such as ‘her’, ‘him’ or ‘it’. The meaning of the pronoun will be given by the surrounding physical environment of objects and people, or by the preceding talk. For example, ‘it’ may be something which a speaker points to or touches, or something which is not present but has already been made a topic of conversation.

The second premise is that meanings are constituted through what is ‘done’, that is, through practices and processes. One example would be how information can be made ‘official’ through processes such as publication or the release of statements. Another would be in talk, the way that a speaker’s identity may be indicated not by what is said but by how it is said, for example, with authority or obedience. Each of these points opens up further directions for investigation, some of which are illustrated in the studies discussed in Chapter 3. The current chapter continues with a discussion of theories which have contributed to the study of social activities or practices, and in particular, language use and talk as an everyday practice.

Theorizing language as social practice

A key writer in this tradition is Harold Garfinkel (1917-2011) who developed the sociological field of ethnomethodology. The following short quotation sets out its main points:

Ethnomethodologists seek to understand how we ‘do’ social life and how ‘the properties of social life which seem objective, factual and transsituational, are actually managed accomplishments or achievements of local processes’. (Zimmerman, 1978, p. 11, quoted in Stokoe, 2003, p. 320)

Ethnomethodology is therefore concerned with the study of practices and processes. In addition, it begins with the study of what is ‘local’. It challenges macro-level explanations of social order by explaining the functioning of society as based in micro-level, rule-governed interactions between people. Garfinkel famously encouraged his students to disrupt the order of local situations, for example, by behaving at home as if they were guests rather than members of the family. The purpose was to demonstrate the tacitly understood rules and forms of behaviour which regulate ordinary life practices and become obvious only when they are ‘breached’. In this view, as people competently function in each small situation of daily life, drawing on the knowledge and skills they have acquired as members of society, their myriad activities combine to constitute an orderly whole. Society is therefore seen to operate from the ‘bottom up’, although ethnomethodologists would reject any neat macro-micro hierarchy.

Following from ethnomethodology, talk and spoken interactions can be analysed as orderly social activities. This is the approach of conversation analysis (CA), based on the work of Harvey Sacks (Jefferson, 1992). It is sometimes referred to as a form of discourse analysis and sometimes as a separate field, also known as the study of ‘talk in interaction’. Conversation analysts are interested in talk as a social practice and in the meanings which are established within the immediate conversational context. The data they analyse are usually ordinary conversational interactions (often referred to as ‘mundane’), such as phone conversations or talk at the dinner table. These are electronically recorded and then transcribed in close detail, for example, to include emphases, pauses, overlapping talk and sometimes other details like intonation. There is a preference for analysing ‘naturally occurring talk’ which would have taken place whether or not the researcher (or recording equipment) was present, rather than an interview set up for the purposes of the research. This kind of investigation of talk as a social practice potentially (some would say definitely!) conflicts with the analysis of meaning systems which is associated with the tradition discussed in the previous section. Meanings are assumed to be established within the context of the talk, rather than being given by society and then imported into the situation being studied.

An analysis of talk as a social practice also challenges many established ideas about communication and about people themselves. The analyst studies how talk is shaped to the interactional context, to perform func tions within that context. To understand this, consider the kind of talk which might be labelled ‘apologizing’. This would often be assumed, by people other than ethnomethodologists, to follow on from some prior mental activity: a person feels apologetic or thinks something like ‘I was wrong’, and then expresses that feeling or thought in talk. As another example, telling a story about something that happened in the past might be assumed to follow from the mental activity of ‘remembering’. However, in the practice-focussed logic of ethnomethodology and conversation analysis, apologizing or storytelling talk occurs in response to other talk, as part of a sequence of steps or moves like a social dance. The telling of the story or the apology occurs at the point where it is made appropriate by the ongoing social interaction. It is therefore understandable as part of that interaction. In summary, what a person says can be accounted for only with reference to interactional processes; it is not described or explained in terms of some kind of individual and internal mental or emotional prompt. This view has implications not only for interpretations of language but also for models of people as language users. It has contributed to important debates in psychology, discussed in more detail in the section ‘Discourse analysis in psychology’.

A rather different emphasis on talk as an activity, usually associated with the concept of performance, derives from the work of the philosopher J. L. Austin (1911-60). He emphasized the distinction between the label given to an activity and the words used to carry out that activity. Interrupting, for example, may be accomplished using a variety of expressions (‘Excuse me’, ‘Just a minute’, ‘But’), not to mention non-verbal sounds and actions (clearing the throat, coughing, raising a hand and so on) but it seldom involves the words ‘I interrupt’. This may seem a rather obvious point but Austin’s work is important for discourse analysts because it draws attention to the functions of language, especially but not exclusively of talk. These functions can be understood only by considering its use in context.

This section and the preceding one have discussed two different theorizations of language and meaning, in terms of systems and practices. Together and separately they challenge many everyday assumptions about language, including how it works for communication. The next section discusses theories of language as communication.

Theorizing language as communication

Communication between two people can be represented very simply using a three-part model of sender, communication channel and receiver (Wertsch, 1990). A piece of information which originates in the person who is the sender, perhaps as a thought or observation, is conveyed through the channel to the receiver. If the model is elaborated to include language as the channel of communication, the information is first encoded or translated into language by the sender, then decoded or translated again by the receiver. The model implies that the information is communicated without distortion: neither the encoding and decoding processes nor the transmission in language alters its meaning in any way. The communication process functions so efficiently that it can be ignored. This is the assumption which operates when what someone says is assumed to be entirely intelligible as the utterance and intended meaning of that individual speaker.

However, this model can be challenged. The problems with it become obvious if it is applied to the example discussed earlier, that is, if the communication channel consists of letters. The model would imply that the meaning of a letter originates entirely in the writer and has no additional meanings which are given by its social context. It would therefore discount any meanings which arise as part of the social practices around language, such as the style of the letter, and it would discount too any indexical meanings belonging to the correspondence or communication itself. These might include the kinds of emotion- or affect-laden meanings which develop over an extended communication (such as pet names or special occasions referred to by lovers, or a particular grievance which becomes the charged focus of a dispute between neighbours.) The model would imply, too, that the language of the letter has only one meaning which will not change in different contexts, including different times.

Most, if not all, academic discourse analysts would reject the simple three-part model, as James Wertsch (1990) discusses. The point of presenting it here is to draw attention to this rejection and, in doing so, to indicate some of the interests of academics who study language as communication, particularly sociolinguists. Their ‘social view of language’ (Kress, 2001, pp. 29-38) can be presented as a rejection of the study of languages as stable, coherent, intact systems of words and grammar which can be understood without reference to the contexts in which they are used, or to the language users themselves. A social study of language would assume that a standard named language, such as English or French, has many different versions and variations. There are the ‘dialects’ associated with communities defined by geographical location, class, age, multilingualism, among other possibilities. There are the differences in how people use languages in specific contexts and relationships, and for different purposes: these various ‘registers’ incorporate distinctive vocabulary and grammar (think of the different ways in which children speak to their friends and to their parents or teachers). Most of these variations carry social meanings and values (think of the preferred speaking style for a TV news presenter, or a government spokesperson). Some of the variations are more within the language user's control than others; in a specific situation (the pupil in trouble who is talking to the head teacher) the user will make choices about the language to be used but some variations, such as class and regional dialects, may be relatively inflexible.