Skip to content

Compare methods

Compare source explanations, strengths, limitations and references for up to four methods.

9 results
Criterion Discourse Analysis Qualitative Content Analysis Critical Discourse Analysis
Overview
  • Discourse analysis is a qualitative research methodology that examines how language, communication, and power shape meaning, identity, and social reality. Developed across linguistics, sociology, and psychology (particularly by Norman Fairclough and Jonathan Potter), discourse analysis goes beyond content to analyze language use as a social practice that constitutes and reflects power relations, ideologies, and social structures.
  • Qualitative Content Analysis (QCA) is a systematic, inductive method for analyzing textual or visual data by identifying and categorizing meaning units into content categories. Developed and formalized by Klaus Krippendorff (1980), QCA can be purely qualitative (inductive, exploratory) or combined with quantitative counting; it analyzes manifest content (explicit, surface meanings) and latent content (underlying, interpretive meanings).
  • 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.
Use and interpretation
  • Use discourse analysis when your research question concerns how language, representation, or communication shape meaning, power, and social reality. Ideal for studying media representations (of disease, groups, events), policy language, professional discourse, or organizational communication. DA is valuable for understanding how social problems are framed, how groups are represented, or how power operates through communication. DA is less suited for research seeking to understand individual subjective experience or to test hypotheses about objective conditions; instead, it illuminates how people talk and think about those conditions. DA is powerful in critical research aimed at revealing ideologies or exposing how language reproduces inequality.
  • Use QCA when you have large volumes of text or visual data and need systematic, transparent categorization. Ideal for analyzing documents (policies, clinical guidelines, media), interview transcripts, social media content, or any communicative artifact. QCA works well for research questions asking 'What themes or topics are prominent in these documents?' or 'How is a particular concept represented in media or discourse?' QCA is less suited for research seeking to generate new theory (use Grounded Theory) or explore lived experience (use phenomenology). QCA excels in applied research, policy analysis, media analysis, and organizational research where systematic documentation of content patterns informs decisions.
  • 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.
Level of measurement
  • Written texts, media discourse, political speeches, institutional documents, interview transcripts
Sample size
  • Purposively selected texts or episodes; sample size varies from a single document to dozens of texts
Strengths
  • Reveals how language constructs meaning and power; exposes ideologies and taken-for-granted assumptions embedded in communication.
  • Applicable to diverse texts and contexts: media, policy, professional communication, social media—anywhere language matters.
  • Systematic and transparent: detailed linguistic analysis can be verifiable; readers can see evidence for claims about language use.
  • Practical insights: findings often inform communication improvements, challenge stereotypes, or reveal how language reproduces inequality.
  • Systematic and transparent: explicit coding scheme and rules enable reproducibility and inter-rater reliability assessment; ideal for rigorous analysis of large datasets.
  • Combines rigor with flexibility: can incorporate both manifest (objective) and latent (interpretive) content, bridging quantitative and qualitative approaches.
  • Suitable for diverse data types: text, images, video, social media, documents—applicable across many research contexts.
  • Code frequency quantification provides summary of emphasis and distribution, complementing qualitative interpretation with counts and proportions.
  • 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.
Limitations
  • Requires linguistic and critical sophistication; analyzing discourse well demands training in language analysis and critical thinking.
  • Can be time-consuming: close linguistic analysis of rich data is labor-intensive; even a few documents can yield months of analysis.
  • Potential for over-interpretation: language is ambiguous; different analysts may find different meanings in the same text. Validity depends on justified interpretations grounded in evidence.
  • Limited scope: analysis of media or policy discourse may describe language patterns without directly measuring social impact or change.
  • Risk of over-interpretation in latent content analysis; assigning meaning to underlying messages is inherently subjective and may reflect coder bias.
  • Creating and refining coding schemes is time-consuming; changing categories mid-analysis undermines consistency.
  • Does not generate theory or deep contextual understanding; content analysis describes what is present, not why or how it came to be that way.
  • Inter-rater reliability, while valuable, can be artificially inflated if coders simply follow rigid rules without understanding meaning; agreement ≠ validity.
  • 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.
Common pitfalls
  • Confusing discourse analysis with discourse (the thing being analyzed); DA is a method for analyzing discourse, not synonymous with discourse itself.
  • Imposing predetermined political or ideological conclusions; analysis should be grounded in text evidence, not driven by a priori ideological commitments.
  • Analyzing language in isolation from context; discourse is social action; understanding context (who is speaking, to whom, when, why) is essential.
  • Claiming discourse causes social outcomes (e.g., 'media discourse causes discrimination'); discourse is implicated in social processes, but causation is complex.
  • Imposing a rigid, inflexible coding scheme too early; categories should emerge partially from data and be refined iteratively.
  • Miscounting or inconsistently applying codes; systematic checking and regular coder meetings are essential.
  • Confusing content analysis with discourse analysis; discourse analysis emphasizes how language shapes meaning and power (e.g., linguistic features, rhetoric), while content analysis focuses on categorical content.
  • Relying solely on code frequency without qualitative interpretation; raw counts are meaningless without context and analysis of what patterns signify.
  • 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.
Introduced by
  • Norman Fairclough; Jonathan Potter and Margaret Wetherell
  • Klaus Krippendorff; refined by Margrit Schreier
  • Norman Fairclough; Teun A. van Dijk; Ruth Wodak
Year
  • 1989 (Fairclough); 1987 (Potter & Wetherell)
  • 1980
  • Late 1970s–1990s (systematised ~1979–1995)