Discourse Analysis
Discourse analysis examines how language constructs meaning, identities, relationships and versions of social reality. Rather than treating words simply as containers of information, it investigates what language is doing in a particular social context. Depending on the discourse-analytic tradition, researchers may examine how language performs social actions, positions individuals and groups, reproduces assumptions, establishes what counts as legitimate knowledge, or reinforces and challenges relations of power.
A major methodological challenge is that discourse analysis is not one standardized technique. Different traditions focus on different aspects of language and social life. Therefore, researchers should not simply state that they “used discourse analysis.” They need to explain what aspect of discourse they analysed, which analytical tradition informed the analysis, and why that approach was appropriate for the research question.
On this page
- Discourse analysis explained simply
- What discourse analysis is
- What counts as discourse
- Main approaches to discourse analysis
- The Dudovskiy Discourse Analysis Lens Framework
- Critical Discourse Analysis
- Foucauldian Discourse Analysis
- Discourse Analysis vs Thematic Analysis
- Discourse Analysis vs Content Analysis
- Discourse Analysis vs Narrative Analysis
- How to conduct discourse analysis
- Application example
- Advantages and limitations
- Common mistakes
- Discourse analysis in business research
- Discourse analysis in the age of AI
- When to use discourse analysis
- Dissertation example
- Exam tip
| Question | Discourse Analysis |
|---|---|
| Main analytical interest | What language does, constructs or makes possible |
| Typical data | Interviews, conversations, speeches, documents, policies, reports, media and digital communication |
| Unit of interest | Language in its social and interpretive context |
| Typical focus | Meaning, positioning, identity, interaction, assumptions, knowledge, ideology and power |
| Research orientation | Primarily qualitative, although specific traditions vary considerably |
| Key methodological decision | What aspect of discourse is being analysed and through which theoretical lens |
| Typical output | Interpretation of how particular versions of reality are constructed and what they accomplish |
| Particularly useful when | Language itself is central to the research question |
Discourse Analysis Explained Simply
Imagine that a company introduces an artificial-intelligence system and managers repeatedly describe the change as “empowering employees to work smarter.”
A researcher could analyse what employees think about the system. That would make employees’ experiences or attitudes the primary object of analysis.
Discourse analysis can ask a different question:
What is the language of “empowerment” doing?
Calling the technology empowering constructs a particular version of organizational change. It positions the technology as beneficial, management as enabling progress, and employees as people who should embrace that progress. Alternative descriptions such as “workplace monitoring,” “automation” or “management control” would construct very different versions of the same organizational development.
A discourse analyst therefore does not merely extract information from what people say. The researcher examines how language participates in constructing the phenomenon being studied.
What Is Discourse Analysis?
Discourse analysis is a broad family of approaches concerned with the study of language in use and the relationship between language and its social context. The term encompasses different intellectual traditions rather than referring to a single universally accepted analytical procedure.
This diversity is important. Some discourse analysts concentrate on how speakers use language to accomplish actions during interaction. Others examine how particular ways of talking construct identities and social realities. Critical approaches may investigate relationships between discourse, ideology and power, while Foucauldian approaches may examine how systems of discourse establish particular forms of knowledge and make certain ways of thinking and acting possible.
Accordingly, the methodological question is not simply:
“How do I code my data using discourse analysis?”
It is first:
“What do I want to understand about discourse?”
Only after answering that question can the researcher make a defensible choice about analytical tradition and procedure.
What Counts as Discourse?
Discourse is broader than spoken conversation. Researchers can analyse interviews, meeting transcripts, political speeches, organizational documents, advertisements, annual reports, policy documents, newspaper articles, websites, social-media posts, professional guidelines and many other forms of written, spoken or digitally mediated communication.
What makes these materials suitable for discourse analysis is not simply that they contain words. Language must matter analytically to the research question.
For example, an interview transcript can be approached in fundamentally different ways.
A researcher conducting thematic analysis may treat participants’ statements primarily as evidence of recurring experiences and identify themes such as job insecurity, workload and resistance to change.
A discourse analyst may instead investigate how participants construct themselves as “adaptable professionals,” describe colleagues as “resistant to change,” or reproduce a broader organizational discourse in which technological adoption is equated with progress.
The same interview can therefore produce very different research depending on the analytical question.
Main Approaches to Discourse Analysis
There is no single authoritative taxonomy of discourse analysis, and labels sometimes overlap. Nevertheless, several influential traditions illustrate the methodological diversity of the field.
Critical Discourse Analysis (CDA) investigates relationships between discourse and broader structures of power, inequality and ideology. It is particularly appropriate when researchers are interested in how language legitimizes or challenges social arrangements.
Foucauldian Discourse Analysis draws on Michel Foucault’s work on discourse, knowledge, subjectivity and power. Rather than treating discourse simply as language used by individuals, it examines broader systems of meaning that establish what can be regarded as true, normal, legitimate or possible.
Discursive Psychology examines how psychological concepts and versions of reality are constructed and used in interaction. Instead of assuming that talk straightforwardly reveals internal psychological states, researchers can examine what particular descriptions accomplish.
Conversation Analysis closely investigates naturally occurring interaction, including features such as turn-taking, sequencing, pauses, repair and how participants organize social actions through talk. Although closely related to discourse-oriented research, it has distinctive theoretical assumptions and rigorous analytical conventions.
Other discourse-analytic traditions exist, and researchers should avoid treating these approaches as interchangeable labels.
Dudovskiy Discourse Analysis Lens Framework
A common methodological error is to select “discourse analysis” as the analytical method before deciding what aspect of discourse actually matters to the research question.
The Dudovskiy Discourse Analysis Lens Framework reverses that process.
The framework does not propose a new form of discourse analysis. Instead, it synthesizes established methodological concerns into a practical decision aid for determining the researcher’s principal analytical focus.
| Analytical lens | Central question | Possible analytical focus |
|---|---|---|
| Language-in-use | What is language accomplishing here? | Actions, interaction, rhetorical functions, conversational practices |
| Meaning construction | What version of reality is being constructed? | Categories, descriptions, assumptions and representations |
| Identity and positioning | How are people or groups being constructed and positioned? | Identities, roles, subject positions, inclusion and exclusion |
| Power and ideology | Whose interests or assumptions are normalized or challenged? | Dominance, inequality, ideology and legitimacy |
| Knowledge and possibility | What is presented as true, normal or possible? | Knowledge systems, norms, subjectivity and institutional discourse |
| Social consequences | What becomes easier or harder to think, say or do because of this discourse? | Organizational practices, policy, behaviour and social consequences |
The framework should be used as a lens-selection tool rather than a six-stage procedure. A study does not need to investigate all six dimensions.
Instead, the researcher begins with the research question and identifies which analytical lens is most important. That decision helps determine the theoretical tradition, analytical procedure and type of evidence required.
For example, a researcher studying how corporate leaders justify redundancies may primarily investigate meaning construction and power. A researcher studying how customer-service employees handle complaints during telephone conversations may instead be interested primarily in language-in-use and interaction.
The central principle is:
Do not choose a discourse-analysis technique first and justify it afterwards. Identify what language needs to reveal for your research question, and then select a discourse-analytic approach capable of revealing it.
Critical Discourse Analysis
Critical Discourse Analysis is particularly concerned with relationships between discourse and social power. Researchers may investigate how particular representations become normalized, whose interests they serve, which groups are given authority, and which perspectives are marginalized.
For example, consider corporate reports that consistently describe employees as “human capital.”
A CDA study would not stop at observing the frequency of this phrase. It might examine how this language represents employees primarily as economic resources, how that representation relates to managerial assumptions, and what alternative understandings of employees are obscured.
CDA therefore requires more than finding politically or socially significant words. Researchers need to connect detailed analysis of discourse with the wider social structures and power relationships relevant to the study.
Foucauldian Discourse Analysis
Foucauldian Discourse Analysis is influenced by Foucault’s understanding of discourse as intertwined with knowledge and power.
Its analytical interest often extends beyond what an individual speaker intends. Researchers may examine broader systems of discourse that establish particular categories, forms of knowledge and subject positions.
Consider workplace discussions about the “productive employee.” A Foucauldian analysis might investigate how productivity is constructed as a measurable and desirable characteristic, which forms of employee behaviour become categorized as productive or unproductive, and how individuals are encouraged to monitor and regulate themselves according to those expectations.
The important question becomes not merely what a particular person says, but what forms of knowledge and subjectivity make that way of speaking possible and authoritative.
Discourse Analysis vs Thematic Analysis
The distinction between discourse analysis and thematic analysis is especially important because both can use interviews and other qualitative textual data.
| Discourse Analysis | Thematic Analysis | |
|---|---|---|
| Primary interest | How language constructs or accomplishes something | Patterns of meaning across a dataset |
| Treatment of language | Language itself is analytically important | Language often provides access to experiences, perceptions or meanings |
| Context | Often central to interpretation | Importance depends on the form of thematic analysis |
| Typical question | How is X constructed through discourse? | What recurring patterns characterize experiences of X? |
| Output | Interpretation of discursive construction/practice | Themes representing patterned meaning |
Suppose employees are interviewed about remote working.
A thematic analysis might identify recurring themes such as flexibility, isolation and work-life boundaries.
A discourse analysis could investigate how employees construct the idea of the “responsible remote worker,” how productivity is talked about, and how particular descriptions legitimize either autonomy or managerial surveillance.
Neither method is inherently superior. They answer different questions.
Discourse Analysis vs Content Analysis
Content analysis generally focuses on systematically categorizing the content of communication. Depending on the approach, researchers may quantify categories or interpret their meanings.
Discourse analysis goes further when the research question concerns how language constructs reality or performs social functions.
For example, a content analysis of CEOs’ letters could classify references to employees, customers, innovation and sustainability.
A discourse analysis could examine how CEOs linguistically construct corporate responsibility, position their organizations as responsible actors and manage tensions between commercial interests and environmental obligations.
The boundary is not always absolute, particularly across qualitative traditions, but the researcher’s analytical purpose should determine the choice.
Discourse Analysis vs Narrative Analysis
Narrative analysis is particularly interested in stories: how experiences and events are organized into narratives, including sequence, turning points, identity and meaning across the story.
Discourse analysis is broader in its focus on language and social construction and does not require the material to form a narrative.
For example, an entrepreneur’s account of building and losing a business could be examined narratively by studying the sequence through which failure is transformed into a story of learning and eventual success.
A discourse analyst might instead investigate how the entrepreneur draws upon broader cultural discourses of resilience, individual responsibility and entrepreneurial heroism to construct that account.
The two approaches can overlap, but the primary analytical object differs.
How to Conduct Discourse Analysis
There is no universal sequence suitable for every discourse-analytic tradition. The analytical process must be consistent with the theoretical approach chosen. Nevertheless, a defensible study normally requires several connected methodological decisions.
1. Define the research question. Determine why discourse matters to the phenomenon being investigated. Questions beginning with “How is X constructed…?” or “How does discourse position…?” often signal discourse-oriented inquiry.
2. Establish the theoretical orientation. Decide which discourse tradition best addresses the question and explain its assumptions about language, knowledge and social reality.
3. Define the corpus. Identify the material to be analysed and justify why it provides appropriate evidence. The corpus may consist of interviews, naturally occurring conversations, corporate documents, policies, media texts or combinations of sources.
4. Determine the analytical focus. Decide whether the study primarily investigates language-in-use, meaning construction, identity, power, knowledge or another discourse-related phenomenon.
5. Become familiar with the material and context. Read or examine the corpus repeatedly while considering who produced the discourse, for whom, under what circumstances and within which institutional or social environment.
6. Identify relevant discursive patterns. Depending on the approach, researchers may examine recurring descriptions, categories, contrasts, metaphors, rhetorical strategies, subject positions, assumptions, interactional sequences or other linguistic practices.
7. Examine function and consequences. Move beyond identifying language features to asking what those features accomplish and what version of reality they support.
8. Develop an interpretation. Connect detailed textual evidence with the theoretical framework and research question.
9. Test the interpretation against the data. Examine variation, contradictions and alternative readings rather than selecting only extracts that support the emerging argument.
10. Report the analytical logic transparently. Explain how interpretations were developed and support them with carefully selected extracts from the corpus.
The process is usually iterative. Researchers often move repeatedly between data, theory, context and emerging interpretations.
Application of Discourse Analysis: an Example
Consider a study investigating:
“How do major UK corporations construct employee wellbeing in their annual and sustainability reports?”
The researcher collects reports published by 30 large companies over a five-year period. Rather than simply counting references to wellbeing, the study examines how organizations construct the meaning of a “healthy employee” and how responsibility for wellbeing is allocated.
During analysis, the researcher identifies recurring language around resilience, self-care, flexibility and personal responsibility. Employees are frequently positioned as active managers of their own wellbeing, while structural factors such as workload, job insecurity and managerial practices receive comparatively little attention.
The researcher then examines how this discourse operates within the broader organizational context. The analysis suggests that corporate wellbeing discourse may simultaneously acknowledge employee welfare and individualize responsibility for maintaining it.
The methodological contribution therefore comes not from discovering that companies discuss wellbeing, but from demonstrating how a particular construction of wellbeing organizes responsibility and makes alternative interpretations more or less visible.
Advantages and Limitations of Discourse Analysis
One major advantage of discourse analysis is its ability to reveal dimensions of social phenomena that approaches focused only on reported experiences or content may overlook. It allows researchers to examine how apparently ordinary language constructs categories, identities and assumptions and how these constructions operate within particular social contexts. This makes discourse analysis especially valuable for studying organizational culture, professional identities, policy, institutional communication, inequality and power.
Its interpretive depth is also a limitation. Discourse analysis can be methodologically demanding because the researcher must make theoretically informed interpretations rather than simply apply a standardized coding procedure. Different discourse traditions operate from different assumptions, so a vague statement that “the data were analysed using discourse analysis” provides little methodological transparency.
Another limitation concerns generalization. Discourse analysis generally seeks contextual and theoretical understanding rather than statistical representativeness. Researchers should therefore avoid presenting an interpretation of a particular corpus as evidence that all members of a population think or communicate in the same way.
Finally, researcher reflexivity is important. Analytical interpretations are produced through the researcher’s theoretical perspective and engagement with the data. This should be acknowledged rather than disguised as completely neutral observation.
Common Mistakes When Using Discourse Analysis
A particularly common mistake is treating discourse analysis as thematic analysis with more sophisticated terminology. Identifying recurring topics such as “leadership,” “stress” and “motivation” does not by itself constitute discourse analysis. The researcher needs to examine what particular ways of talking about those topics accomplish.
A second mistake is failing to specify the discourse-analytic tradition. Because different approaches make different assumptions, simply stating “discourse analysis was used” leaves the methodology underdeveloped.
Researchers can also overinterpret individual words without considering their context. A meaningful discourse analysis normally examines patterns, functions, relationships and contexts rather than assuming that one expression has a fixed hidden meaning.
Another problem is beginning with a predetermined ideological conclusion and selecting extracts that confirm it. Even explicitly critical approaches require systematic engagement with the corpus, including contradictions and alternative constructions.
Finally, researchers sometimes detach discourse from the circumstances in which it was produced. Who is speaking, to whom, within what institution, under which constraints and for what purpose can be essential to interpretation.
Discourse Analysis in Business Research
Discourse analysis is particularly useful in business and management research because organizations operate partly through language. Strategies, identities, leadership roles, organizational changes and stakeholder relationships are not merely implemented; they are also described, justified and negotiated through communication.
Researchers can investigate how executives construct organizational crises, how companies communicate sustainability, how entrepreneurs narrate responsibility, how managers position employees during organizational change, or how industries construct emerging technologies as either opportunities or threats.
Consider the phrase “digital transformation.” Organizations may use it to describe technology investment, restructuring, automation or changes in employment. A discourse-oriented study can investigate how the term makes certain organizational actions appear necessary or progressive and how alternative interpretations are marginalized.
Discourse analysis therefore enables business researchers to investigate not only what organizations do, but how language helps make particular organizational actions understandable, legitimate and possible.
Discourse Analysis in the Age of AI and Digital Research
Generative AI creates particularly interesting methodological issues for discourse analysis because AI systems increasingly participate in producing the very discourse researchers may wish to study.
Corporate announcements, customer-service interactions, marketing communication and social-media content can now be partially or entirely AI-generated. Researchers therefore need to consider whether a text represents an individual’s discourse, an organization’s institutional discourse, an AI-mediated production process, or some combination of these.
AI can also assist researchers by searching large corpora, identifying linguistic patterns, locating potentially relevant passages and comparing terminology across documents. These capabilities can make exploratory work with large textual datasets substantially faster.
However, the interpretive task cannot automatically be delegated without changing the methodological situation. If an AI system identifies a discourse, assigns its meaning and explains its social consequences, the researcher must ask:
Whose interpretation is being reported—the researcher’s or the model’s?
This question is especially important because language models themselves reproduce patterns learned from enormous bodies of existing discourse. Their interpretations may import assumptions that are difficult for the researcher to observe or reconstruct.
A defensible AI-assisted discourse analysis should therefore distinguish between AI-supported exploration and researcher-led interpretation. Researchers should document how AI was used, retain responsibility for analytical claims and test AI-generated observations against the original corpus, theoretical framework and social context.
AI consequently does not make discourse analysis obsolete. It makes methodological transparency about who or what performs the interpretation considerably more important.
When to Use Discourse Analysis
Discourse analysis is particularly appropriate when:
- the research question concerns how a phenomenon is constructed through language;
- language itself is part of the phenomenon rather than merely a source of information;
- the researcher wants to investigate identity or positioning;
- institutional or organizational communication is central to the study;
- relationships between discourse, ideology and power are important;
- the study concerns how particular versions of reality become accepted or contested;
- naturally occurring talk or interaction is analytically important;
- the social and historical context of communication matters to the research question.
It is generally less appropriate when the researcher’s primary goal is simply to summarize participants’ experiences, measure the frequency of predetermined categories, or identify recurring themes without analysing the function of language.
Dissertation Example
A Master’s dissertation investigates how commercial banks construct responsibility for cybersecurity in customer communications. The researcher collects cybersecurity webpages, customer guidance and fraud-warning communications from eight major banks and adopts a discourse-analytic approach because the research question concerns how responsibility is linguistically constructed rather than how frequently cybersecurity topics occur.
In the methodology chapter, the researcher explains that the analysis focuses primarily on meaning construction and positioning. Particular attention is paid to how banks position themselves as security providers while simultaneously positioning customers as responsible actors who must recognize threats, protect credentials and follow prescribed security behaviours. Recurring descriptions, contrasts, imperatives and responsibility statements are examined within the institutional context of banking and fraud prevention.
The dissertation justifies discourse analysis over thematic analysis because the purpose is not primarily to identify recurring topics across the documents. Instead, it seeks to understand how particular linguistic constructions distribute responsibility between banks and customers and what assumptions about the “responsible customer” those constructions produce.
Exam Tip
When explaining discourse analysis in an examination or viva, avoid defining it simply as “the analysis of written or spoken language.” That definition is too broad.
A stronger answer explains that discourse analysis investigates language in context and how language constructs or performs social reality. Then specify the analytical focus—for example, meaning construction, identity, interaction, knowledge or power—and explain that different discourse-analytic traditions emphasize different aspects.
If asked why you selected discourse analysis, connect the answer directly to the research question:
“I selected discourse analysis because my research question concerns how the phenomenon is constructed through language, rather than simply what participants say about the phenomenon.”
That distinction demonstrates methodological understanding rather than memorization.
Still not sure if discourse analysis is the right method for your research?
Dudovskiy Research Assistant can analyse your research topic and help you determine whether discourse analysis—or another qualitative analytical method—fits your research question, with a clear academic justification for the choice.
John Dudovskiy
References
Fairclough, N. (2013). Critical Discourse Analysis: The Critical Study of Language. 2nd ed. Routledge.
Foucault, M. (1972). The Archaeology of Knowledge and the Discourse on Language. Pantheon Books.
Gee, J.P. (2014). An Introduction to Discourse Analysis: Theory and Method. 4th ed. Routledge.
Potter, J. and Wetherell, M. (1987). Discourse and Social Psychology: Beyond Attitudes and Behaviour. Sage.
Wodak, R. and Meyer, M. (eds.) (2016). Methods of Critical Discourse Studies. 3rd ed. Sage.