Skip to content
Back to search results

Critical Discourse Analysis — CDA

Introduced by
Norman Fairclough; Teun A. van Dijk; Ruth Wodak
Year
Late 1970s–1990s (systematised ~1979–1995)
Type
Qualitative research method
Family
Discourse Analysis
Font size

What does it do?

Critical Discourse Analysis (CDA) is a qualitative method that examines how language in texts and talk constructs, sustains, and challenges relations of power, ideology, and social inequality.

Drawing on linguistics, social theory, and critical philosophy, CDA treats discourse not merely as communication but as social practice — a site where dominance is reproduced and where resistance can be articulated. Developed in the late twentieth century by Norman Fairclough, Teun van Dijk, and Ruth Wodak, among others, CDA is applied to political speeches, media texts, policy documents, educational materials, and institutional interactions.

The intuition behind the method

Imagine a government press release about immigration policy. On the surface it appears neutral, but CDA asks: whose voices are amplified and whose are marginalised? What metaphors frame migrants as threats rather than people? What is left unsaid? Which social assumptions are treated as common sense? CDA works by reading texts closely at the level of vocabulary and grammar, then connecting those linguistic choices to the broader social and institutional context in which the text was produced and consumed. The analyst exposes the ideological work that language does — often invisibly — and asks whose interests that work serves.

History

Critical Discourse Analysis emerged from the convergence of critical linguistics (Fowler, Hodge, Kress, Trew — Language and Control, 1979) and systemic functional linguistics (Halliday). Norman Fairclough systematised the method in Language and Power (1989) and Discourse and Social Change (1992), proposing his influential three-dimensional framework linking text, discursive practice, and social practice. Teun van Dijk contributed a sociocognitive strand focused on elite discourse, racism, and mental models, articulated across numerous works from the 1980s onward. Ruth Wodak developed the discourse-historical approach, emphasising the role of historical context in analysing argumentation and prejudice. By the mid-1990s CDA had become an established interdisciplinary field, with the founding of the journal Discourse and Society (1990) as a key institutional marker.

When to use it

CDA is appropriate when the research question concerns how language mediates power, ideology, or social inequality. It is well suited to analysing political and media discourse, policy texts, institutional communication, advertising, and educational materials where the researcher suspects that particular representations serve particular interests. CDA requires text or transcript data and a theoretical commitment to the relationship between language and social structure. It is NOT appropriate when the goal is descriptive content frequency, hypothesis testing, or statistical generalisation — for those purposes content analysis or quantitative text analysis are better choices. It is also poorly suited to research that has no critical theoretical anchor; without a clear argument about power relations, CDA risks becoming ordinary close reading.

Limitations and considerations

  • Findings are not statistically generalisable; conclusions apply to the analysed corpus, not to a population.
  • The analyst's own ideological position inevitably shapes what is noticed and how it is interpreted — reflexivity is necessary but does not eliminate subjectivity.
  • Requires substantial expertise in both linguistics (grammar, discourse structure, rhetoric) and relevant social theory; novice analysts often remain at the surface level of word choice.
  • Intensive and time-consuming: thorough CDA of even a small corpus of texts can require weeks of work.
  • Risk of confirmation bias — analysts who begin with a strong critical hypothesis may inadvertently read the texts to confirm it rather than test it.

Source conditions do not replace independent scientific review or validation by an analysis engine.

Strengths

  • Directly addresses the relationship between language, power, and ideology — questions that purely descriptive methods leave unasked.
  • Applicable to a wide range of text types: political speeches, news media, policy documents, social media, institutional talk, and educational curricula.
  • Integrates linguistic precision with social theory, bridging close reading and macro-level analysis of society.
  • Transparent and systematic: explicit analytic frameworks (Fairclough, van Dijk, Wodak) make the process auditable and replicable in broad terms.
  • Particularly powerful for uncovering naturalised or taken-for-granted assumptions that reproduce dominant ideologies invisibly.

Applications

  • Analysing how political speeches and party manifestos construct national identity, frame immigration, or legitimise austerity policies.
  • Examining news media representations of marginalised groups — refugees, ethnic minorities, people with disabilities — to reveal stereotyping and othering.
  • Investigating how educational policy documents and curriculum texts embed assumptions about ability, merit, and citizenship.
  • Studying corporate and institutional communication — annual reports, mission statements, HR documents — to expose how managerial ideology is normalised.
  • Tracing how public health campaigns or medical discourse construct the 'responsible patient' and obscure structural determinants of health.
  • Examining social media and online discourse for the reproduction of populist, nationalist, or extremist ideological frames.

How it works

How it works

1

The analyst begins by locating the text within its social and historical context: Who produced it, for what audience, in what institutional setting, and at what historical moment? This contextualisation draws on background reading, document analysis, and sometimes ethnographic observation to establish the conditions of production and reception that shape the text's meaning.

2

A purposive corpus of texts is assembled to address the research question. Selection is theoretically driven rather than random: texts are chosen because they are prominent, contested, representative of a genre, or pivotal in a social controversy. The corpus may be a single landmark speech, a set of newspaper editorials, or a collection of policy documents — richness and relevance take priority over representativeness.

3

The selected texts are subjected to close linguistic description. The analyst examines vocabulary choices (lexicalisation, naming, framing), grammatical structures (transitivity, passive voice, modality), rhetorical devices (metaphor, presupposition, intertextuality), and argumentation strategies. This level corresponds to Fairclough's textual analysis or van Dijk's microstructural analysis.

4

Linguistic features identified in description are interpreted as discursive practices: how is the text produced, distributed, and consumed? What genres, styles, and discourses does it draw on? How does intertextuality — the way a text echoes or contradicts other texts — position particular social actors? The analyst moves between the text and the broader discursive field to interpret meaning as contextually situated.

5

The interpretive findings are explained in relation to social structure and power. The analyst asks: which ideologies, power relations, or social inequalities are reproduced, legitimised, or challenged by this discourse? Fairclough's three-dimensional framework links textual features to discursive practice to social practice; van Dijk's sociocognitive approach examines how elite discourse shapes mental models; Wodak's discourse-historical approach traces how historical contexts shape argumentation strategies.

Common pitfalls

  • Doing discourse analysis without the 'critical' — describing language patterns without connecting them to power relations or ideological work produces content analysis, not CDA.
  • Treating findings as objective — CDA is explicitly value-laden and politically positioned; claiming neutrality undermines the method's epistemological coherence and misleads readers.
  • Ignoring context of production and reception — a text's meaning cannot be fully grasped from the text alone; the institutional, historical, and audience contexts are constitutive, not supplementary.
  • Overinterpreting isolated linguistic features — a passive construction or a particular metaphor is not automatically ideological; patterns across the corpus, not single instances, carry analytical weight.
  • Conflating different CDA frameworks — Fairclough's three-dimensional model, van Dijk's sociocognitive approach, and Wodak's discourse-historical approach differ in their analytic foci and theoretical assumptions; mixing them eclectically without justification weakens coherence.

Frequently asked questions

How is CDA different from ordinary discourse analysis?

Discourse analysis describes how language functions in context — how meaning is constructed, how turns are organised, how genres work. CDA adds an explicit critical and normative dimension: it asks whose interests are served by particular discursive choices, how power relations are reproduced or contested through language, and what ideological work texts perform. CDA is politically positioned; it does not claim value-neutrality and takes a stance against discursive forms of domination and injustice.

Do I need a linguistics background to do CDA?

A working knowledge of linguistics — at minimum an understanding of transitivity, modality, lexical choices, and rhetorical devices — is important for rigorous CDA. Without it, analysis tends to stay at the level of paraphrase. However, most CDA frameworks are designed to be learnable by social scientists, and textbooks by Fairclough, Jorgensen and Phillips, and Wodak provide accessible introductions. Collaborating with a discourse linguist strengthens the analysis considerably.

How large does the corpus need to be?

CDA does not prescribe a corpus size. The goal is depth of analysis rather than breadth of coverage, so a focused corpus of a few key texts is often more analytically productive than a large corpus treated superficially. The guiding principle is theoretical saturation with respect to the research question: enough texts to address the power or ideology question being investigated, selected on theoretical rather than statistical grounds.

Can CDA be combined with other methods?

Yes. CDA is frequently combined with ethnography (to understand the social context of text production), corpus linguistics (to identify patterns across large datasets that then receive close CDA treatment), interview methods (to understand how discourse is received or resisted by audiences), and historical document analysis. The combination of corpus methods and CDA — sometimes called Corpus-Assisted Discourse Studies (CADS) — is particularly productive for large text collections.

Is CDA appropriate for my thesis if I am not an expert in linguistics?

CDA is feasible for thesis work if the student commits to learning the linguistic analytic categories their chosen framework requires. A dissertation that labels a few word choices as 'ideological' without systematic analysis will not pass CDA scrutiny. The practical advice is to choose one well-established framework (Fairclough's three-dimensional model is the most widely taught), apply it consistently, and be explicit about both the analytic steps taken and the researcher's critical standpoint.

Related methods

Sources

  1. Fairclough, N. (1992). Discourse and Social Change. Polity Press.

    Open publication reference
    Copied.
  2. van Dijk, T. A. (1993). Principles of critical discourse analysis. Discourse and Society, 4(2), 249–283.

    Open publication reference
    Copied.