Talk is the object of analysis, not a channel onto something else

This fifth of ten units argues that discourse and narrative analysis ask different questions of the same transcript while both treat language as action. The analyst's object is what a version of events accomplishes, and how a story orders experience, rather than a count of themes. The change decides what counts as evidence, what a finding looks like, and where two competent analysts may legitimately disagree. The unit's demand follows: name the object of analysis before you claim anything about it. Where the fourth unit read the body, this one reads the words.

Language is the object, not the channel

A transcript can be read two ways. On the first the words report something that happened elsewhere, and the analyst recovers what the speaker meant. On the second the words are the thing that happens, and the analyst describes what they do. Discourse and narrative analysis take the second reading, and the difference is not a matter of taste. Taylor (2013) states the position directly: language is constitutive rather than a neutral channel, meanings are socially derived and negotiated, and language use is a form of social practice.

Discourse analysis is a family, not a method

Hodges, Kuper and Reeves (2008) sort the field for health researchers into three clusters: formal linguistic, empirical and critical. They show how each would study one domain, doctor and patient communication about diabetes.

Their illustration of the first cluster is Lacson and colleagues (2006), who transcribed telephone conversations between nurses and 25 adult dialysis patients over three months and coded all 17,385 words by semantic type and structure, testing whether a computer could compress such conversations into something a physician could read. Counts of word types are what a formal linguistic analysis produces, not what the other two clusters produce. The corpus is telephone talk, and a participant who is deaf or on a poor line leaves gaps that no coding scheme can tell from a topic the nurse chose not to raise.

Gee (2014) supplies a toolkit. Seven building tasks that language performs are significance, practices, identities, relationships, politics, connections, and sign systems and knowledge. He also distinguishes discourse with a small d, language in use, from Discourse with a capital D, the ways of acting and valuing that privilege some groups. Fairclough (2013) adds the critical strand, linking textual analysis to discourse practice and social practice and arguing that texts both reflect and reproduce social relations. A study therefore has to say three things. It must name the cluster, the building task, and the relation between text and social order it is working with.

An account is built to be defensible, not to report an inner state

Discursive psychology gives the clearest demonstration that language is action. Potter and Wetherell (1987) argue that language should be studied as social action and that objects such as attitudes are better understood as situated, occasioned accounts than as inner states that talk reports. Edwards and Potter (1992) add that phenomena such as memory and attribution are things people do in talk, and that descriptions are built to be defensible, so a factual account is produced rather than retrieved.

A participant who says that the old system was fine until the team expanded is not reporting a property of a system. The account attributes the change to circumstances and protects the speaker from blame. Wetherell (1998) widens the frame from the single turn to the cultural resource. She uses interpretative repertoires and ideological dilemmas to analyse a discussion among three young white middle-class men about an episode in one participant's recent sexual history. She insists that both the local ordering of talk and the broader resources it draws on need analysis. The object is by now clearly not a theme count.

Narrative asks how experience is ordered

Narrative analysis puts a different question to the same material: how events are ordered into a story with a teller, a plot and a point. Labov and Waletzky (1967) identify a regular six-part structure in oral narratives of personal experience: abstract, orientation, complicating action, evaluation, resolution and coda. They argue that evaluation is what makes a narrative tellable. Its categories are the vocabulary for asking what a speaker does with the order of events. They were built from a specific corpus. Labov's interviewers put one question first: whether the speaker had ever been in serious danger of being killed. In the best known of them the narrator describes twisting his brother's arm behind his back during a fight, letting go, and finding a knife on the table. The six categories describe the working of those two linked episodes.

The tradition is nevertheless a family of methods with different objects. Riessman (2008) distinguishes four kinds of narrative analysis: thematic, structural, dialogic/performance and visual. Each asks a different question of the same story. Chase (2005) describes five analytic lenses, running from the story as a window onto experience to the story as a form of social action. She argues that the lens chosen determines what counts as narrative evidence; her fifth lens turns the analyst into a narrator as well.

The story does something for someone

The strongest version of the claim is that stories act. Bamberg (1997) moves beyond structure to positioning, showing how narrators situate themselves and others in relation to the story, the audience and themselves, so that a story performs identity work for a particular occasion. Frank (2010) goes further with socio-narratology, treating stories as actors that shape the lives of those who tell and hear them. He argues that the analyst should study what stories do rather than only what they represent. The move has a price: if a story acts on its audience, the occasion of the telling belongs in the account, and an elicited interview narrative is one occasion among many.

The warrants can be combined once each states its scope

Alejandro and Zhao (2024) offer the constructive answer. They distinguish four families of method: Discourse Analysis, Foucauldian Discourse Analysis, thematic analysis and qualitative content analysis. They review how the families have already been combined across disciplines, and set out a framework for combining them without collapsing one into another. The framework produces questions that neither approach asks alone, which is what the contested question needs. Rather than arguing that the local order or the cultural resource is the real object, a study can state where each does its work.

Contested question: the local order of talk, or the cultural resource behind it?

The literature agrees that language is action and disagrees about how far a claim may travel from the transcript. One position keeps the analysis close to the participants' own displayed orientation, so that a claim about power or culture is warranted only where participants themselves orient to it. This is the conversation-analytic reservation that Wetherell (1998) answers rather than ignores. The wider position is held by Fairclough (2013) and Gee (2014), for whom a text is tied to discourse and social practice and to the distribution of social goods. On that view an analysis confined to the turn misses the ideology the text enacts. Wetherell (1998) occupies the synthesis position, arguing that interpretative repertoires and positioning let the local ordering and the broader cultural resource be analysed together.

A second disagreement runs through narrative. Labov and Waletzky (1967) take the elicited personal narrative as the model case. De Fina and Georgakopoulou (2011) argue for the small story told in ordinary interaction. They treat narrative as situated communicative practice and criticise the over-reliance on the single elicited story.

What would settle either dispute is one corpus analysed twice, under each warrant, and reported so that a reader can see which inferences survive the change of frame. For the discourse question the test is whether an analyst can point to the cultural resource in the participants' own orientation; a critical reading fails when it needs material the participants never treat as relevant. Chase (2005) explains why the disputes are hard to close: the lens fixes the evidence, so the two sides can pass each other without contradiction. A demonstration would first have to hold the question constant.

The two warrants are argued rather than tested against each other. The wider reading draws on frameworks and programmatic books (Fairclough, 2013; Gee, 2014; Potter and Wetherell, 1987), which show what a critical analysis can do on one corpus. The local reading is disciplined by demonstrations a reader can check turn by turn. Wetherell (1998) shows the synthesis on one discussion between three speakers, so she illustrates the method rather than settling its scope. While the dispute is open, a designer studying talk should name the object of analysis before making any claim, and mark every inference that needs material the participants never treat as relevant. The strongest objection is that a study confined to displayed orientation can describe the order of talk but cannot say what that order is doing politically.

Where the argument is heading

The direction of travel is from a single named method towards a stated relation between transcript and social order. De Fina and Georgakopoulou (2011) move narrative's model case from the elicited story to the small story in interaction, and Alejandro and Zhao (2024) turn discourse analysis into a set of combinable analytic logics with declared points of contact. The vocabulary also travels: Hodges, Kuper and Reeves (2008) carry it into health research.

The sources this unit rests on

Each note states what the cited document did and found. The bibliographic records were checked against registered metadata on 30 October 2026.

• Taylor, S. (2013). What is Discourse Analysis? Bloomsbury. Introduces discourse analysis as the close study of language use as evidence about society, arguing that language is constitutive rather than a neutral channel and that language use is social practice. It stresses that there is no single discourse analysis. https://www.bloomsbury.com/uk/what-is-discourse-analysis-9781849669030/
• Hodges, B. D., Kuper, A. and Reeves, S. (2008). Discourse analysis. doi:10.1136/bmj.a879 Sorts discourse analysis into three clusters, formal linguistic, empirical and critical, and illustrates each on doctor and patient communication about diabetes. It is the unit's evidence that the label names a contested field rather than a method.
• Lacson, R. C., Barzilay, R. and Long, W. J. (2006). Automatic analysis of medical dialogue in the home hemodialysis domain. Journal of Biomedical Informatics, 39(5), 541-555. doi:10.1016/j.jbi.2005.12.009 Codes each turn of recorded nurse and patient telephone consultations by semantic type and structure, reaching 73 per cent classification accuracy, and tests automatic summarisation. It is the primary study behind the dialysis coding example.
• Gee, J. P. (2014). An Introduction to Discourse Analysis. doi:10.4324/9781315819679 Sets out seven building tasks that language performs, significance, practices, identities, relationships, politics, connections and sign systems and knowledge, and distinguishes discourse in use from Discourse as a way of acting and valuing. It supplies the unit's practical toolkit for asking what a transcript builds.
• Fairclough, N. (2013). Critical Discourse Analysis (2nd ed.). doi:10.4324/9781315834368 Develops critical discourse analysis as the study of how language is tied to power and ideology, linking textual analysis to discourse practice and social practice. It argues texts both reflect and reproduce social relations, the position the unit weighs against the conversation-analytic reservation.
• Potter, J. and Wetherell, M. (1987). Discourse and Social Psychology. Sets out discursive psychology, arguing that language should be studied as social action and that attitudes are better understood as situated, occasioned accounts. It shows how participants construct versions of events to manage blame. https://archive.org/details/discoursesocialp0000unse
• Edwards, D. and Potter, J. (1992). Discursive Psychology. Develops the claim that memory and attribution are things people do in talk rather than inner states that talk reports, and analyses how descriptions are built to be defensible. It grounds the unit's distinction between an account and a theme that is tallied. https://archive.org/details/discursivepsycho0000edwa
• Wetherell, M. (1998). Positioning and interpretative repertoires. doi:10.1177/0957926598009003005 Brings conversation analysis and post-structuralist discourse analysis into dialogue through interpretative repertoires and positioning, and analyses a group discussion with three young men about a personal episode. It argues both the local ordering of talk and the broader cultural resources it draws on need analysis, the unit's synthesis position.
• Labov, W. and Waletzky, J. (1967). Narrative analysis: oral versions of personal experience. Identifies a regular six-part structure in oral narratives of personal experience, abstract, orientation, complicating action, evaluation, resolution and coda, and argues that evaluation makes a narrative tellable. It is the standard structural model that later work uses and contests. https://www.degruyter.com/document/doi/10.1075/jnlh.7.02nar/html
• Riessman, C. K. (2008). Narrative Methods for the Human Sciences. Distinguishes four kinds of narrative analysis, thematic, structural, dialogic/performance and visual, and shows how each asks a different question of the same story. It carries the unit's claim that narrative analysis is a family of methods with different objects. https://archive.org/details/narrativemethods0000ries
• Chase, S. E. (2005). Narrative inquiry: multiple lenses, approaches, voices. Reviews five analytic lenses in narrative inquiry, from the story as a window onto experience to the story as social action. It argues the lens chosen determines what counts as narrative evidence. https://methods.sagepub.com/book/the-sage-handbook-of-qualitative-research
• Bamberg, M. (1997). Positioning between structure and performance. doi:10.1075/jnlh.7.42pos Develops positioning analysis to show how narrators situate themselves and others in relation to the story, the audience and themselves, going beyond structure to the identity work a story performs. It argues that narrative analysis must account for the interactive occasion of the telling.
• Frank, A. W. (2010). Letting Stories Breathe. doi:10.7208/chicago/9780226260143.001.0001 Develops socio-narratology, arguing that stories are actors that shape the lives of those who tell and hear them, and that the analyst should study what stories do rather than only what they represent. It supplies the unit's strongest claim that narratives act on the world.
• De Fina, A. and Georgakopoulou, A. (2011). doi:10.1017/cbo9781139051255 Reviews discourse and sociolinguistic approaches to narrative, including small stories research and positioning analysis, and argues for studying narrative as situated communicative practice. It criticises the over-reliance on the single elicited personal narrative as the model case.
• Alejandro, A. and Zhao, L. (2024). Multi-method qualitative text and discourse analysis. doi:10.1177/10778004231184421 Develops a framework for combining qualitative text and discourse analysis, identifying four families of method and reviewing how they have been combined across disciplines. It shows that the combination produces questions neither approach asks alone.