Discourse Analysis

Discourse analysis is a qualitative research method that examines how language is used in texts, conversations, and other communicative practices to construct meaning, identities, power relations, and social realities within specific contexts. Rather than treating language as a transparent window onto an objective reality, discourse analysis treats language itself as the object of study, investigating how particular ways of talking or writing produce and reproduce social structures, knowledge claims, and subject positions. It encompasses multiple theoretical traditions, including critical discourse analysis, Foucauldian discourse analysis, conversation analysis, and discursive psychology.

Language is not merely a tool for describing reality; it actively shapes how people understand and navigate the social world. In health research, the way diseases are named and described influences public perception, funding, and patient identity. In policy research, the framing of issues in policy documents determines which solutions appear natural or inevitable. Discourse analysis makes these processes visible, revealing assumptions, power dynamics, and exclusions that may be invisible to those operating within a particular discourse. Researchers in sociology, political science, communication studies, nursing, public health, and education use discourse analysis to examine media representations, policy documents, clinical interactions, institutional texts, and everyday conversation.

Discourse analysis begins with the selection and collection of texts or talk for analysis. The term "text" is used broadly to include written documents (policy papers, clinical guidelines, media articles), transcribed interviews or focus groups, social media posts, visual materials, and naturally occurring conversation. The selection is guided by the research question and the theoretical framework.

Analysis proceeds differently depending on the tradition. Critical discourse analysis (CDA), associated with scholars like Norman Fairclough, examines the relationships between language, power, and ideology. CDA analysts identify linguistic features (word choice, metaphor, passive voice, nominalization) that serve particular ideological functions, such as obscuring agency or naturalizing inequality. For example, a policy document that refers to "health disparities" rather than "health injustice" frames the problem differently, with implications for the types of solutions that seem appropriate.

Foucauldian discourse analysis focuses on how discourses construct objects of knowledge and position subjects within systems of power. Rather than analyzing the linguistic details of individual texts, it examines broader discursive formations, the rules and conventions that govern what can be said, thought, and done within a particular domain. A Foucauldian analysis of mental health discourse might examine how the shift from "madness" to "mental illness" to "mental health condition" reflects changing configurations of medical, legal, and social power.

Conversation analysis (CA) takes a different approach, focusing on the micro structure of naturally occurring talk: turn taking, repair sequences, adjacency pairs, and the organization of action in interaction. CA is strongly data driven: rather than importing external theories or social categories to explain patterns in the data, it grounds its claims in how the participants themselves respond to each other's turns.

Many discourse analytic approaches involve iterative close reading, coding, and interpretation, whereas conversation analysis builds collections of similar sequences from detailed transcripts. Analysts move between the data and their theoretical framework, identifying patterns, contradictions, and silences (what is not said can be as analytically significant as what is said). Findings are typically presented with extensive data excerpts that allow readers to evaluate the analyst's interpretations.

Discourse analysis is appropriate when the research question concerns how language constructs meaning, identity, power, or social practice, rather than when the goal is to discover the "true" attitudes, beliefs, or experiences of participants (for which content analysis or thematic analysis may be more appropriate). It is well suited for examining policy and institutional documents, media representations, professional communication, and any context where the researcher is interested in the productive function of language. It is not appropriate for quantitative research questions, for studies requiring numerical data, or for contexts where the researcher seeks to summarize content rather than analyze how it functions.

A common error is confusing discourse analysis with thematic analysis or content analysis. While all three involve analyzing text, they differ in their theoretical assumptions and analytical goals: thematic analysis identifies patterns of meaning in the data, content analysis systematically categorizes content and, in its quantitative form, counts how often categories occur, and discourse analysis examines how language constitutes social reality. Another mistake is selecting a discourse analytic tradition without understanding its theoretical commitments, leading to analyses that are methodologically incoherent. Researchers sometimes present discourse analytic findings without sufficient data excerpts, asking readers to accept interpretations on trust rather than providing the evidence for evaluation. Overlooking the reflexive dimension, the analyst's own position and assumptions, produces analyses that claim objectivity while being shaped by the analyst's perspective. Finally, treating discourse analysis as a purely descriptive exercise, without engaging with the critical or theoretical implications of the findings, produces shallow analyses that do not fulfill the method's potential.

A researcher conducts a critical discourse analysis of 40 media articles published during a public health emergency, examining how metaphors of war ("fighting the virus," "frontline workers," "ammunition of vaccines") construct the pandemic as a military conflict. The analysis shows that this framing positions citizens as either soldiers (compliant) or deserters (noncompliant), marginalizes nonbiomedical approaches to health, and naturalizes authoritarian public health measures. The researcher argues that alternative framings, such as metaphors of community care or ecological balance, would produce different subject positions and policy possibilities.