Critical Discourse Analysis With a Worked Framing Table

A policy notice can name who made a choice or make it seem to happen by itself. Critical discourse analysis asks how words show people and their place in society. It links that reading with questions about power.

Start with a question you can check in selected texts. Keep the words, who they name, what you know about the setting and your reading together. A reader can then trace the claim and see where it might need to change.

What critical discourse analysis examines

Critical discourse analysis, or CDA, is a family of ways to study language in relation to social life, power and inequality. Its critical focus asks how social relations are made and sustained. State which approach you use.

The LSE authors' guide names several traditions, including dialectical-relational, sociocognitive and discourse-historical approaches. They differ in their ideas about language and in how they study it.

Choose an approach that helps answer your question. In a study of how policy shows people who seek aid, you might ask who gets blamed, whose account is heard and who gets to make the rules.

Broader discourse analysis can study language in use without the same critical question. Genre analysis asks about a text's purpose and common forms. Knowing how a policy notice works can help CDA, but you still need to build the power claim.

State your critical question and lens

A first question such as "Is this policy unfair?" is too broad for a close text check. Try "Who do these notices blame for missed visits, and who gets to explain when the rule cannot be followed?"

That question points to wording, a social relation and texts to compare. It leaves room for a surprise. A notice may name the office's role more clearly than you thought it would. Keep such findings in the account.

Write a short lens statement before you pick passages, and state what evidence you will seek. In this fictional example, the lens concerns responsibility and voice: who acts, who must follow the rule, and whose account counts when they cannot do so?

List what could weaken your reading. A later paragraph might explain how to appeal or name the office that caused a delay. Keep that paragraph in view. Leaving it out could give a false picture of the notice.

If you compare lenses, keep their differences clear. Theory triangulation helps you place more than one reading beside the same text. Check the basis of each, even when the readings seem to agree.

Select texts and contextual evidence

Define your corpus, or set of texts, before claiming a pattern is widespread. Record versions, dates, who issued each text, who it addresses and why you chose it. Keep a claim about two notices within that scope.

Choose parts that can be compared. You might read the paragraph that states a rule in each notice, while keeping the text around it. A short headline and a full policy manual may serve different purposes and need a different kind of check.

Keep sources about context separate from the texts you study. A guide to the rules may show who can grant an exception. An interview may show how one person read a notice. They support different kinds of claims.

The Institute of Historical Research guide links words and grammar with how texts are made, read and set in social life. Seek those sources when a sentence cannot support the wider claim on its own.

Use texts you are allowed to study and upload. Keep enough context to read them fairly, while following the study's consent and privacy rules. Check these rules before adding any sensitive account to a shared source set.

Build a worked framing table

The following notices and context are fictional teaching material. Notice A says "Appointments were withdrawn after repeated non-attendance." Notice B says "The scheduling team cancelled appointments after two missed visits. Recipients may request a review."

Both notices describe a limit. They differ in who and what they name. Keep those facts apart from your reading of them. Another reader needs to see the reasoning behind a label such as "oppressive language."

Fictional comparison of appointment notices through a responsibility-and-voice lens

Table 1: The first row shows that A's sentence does not name who made the choice. You do not know from that sentence whether the writer meant to hide blame. To make that stronger claim, seek sources about how the text was made.

The second row makes a quick judgment harder. A count makes the threshold clear, but leaves the reasons for missed visits unknown. Wording can give more detail in one respect and still leave a gap in another.

The third row raises a question about voice. B names a way to seek review, but you do not know whether people can use it easily. Look at the rules and how people experience them before you make that claim.

Revise the table when new evidence appears. If A's next paragraph provides a clear review process, change "A excludes recipient voice" to "A's opening sentence does not foreground the review process described later."

That change keeps the finding about the opening words while narrowing its scope. Your critical reading must fit the texts you chose, including parts that challenge it. A useful account makes room for those parts.

Read actor representation in a recruitment poster

James Montgomery Flagg's 1917 recruitment poster gives you a real text with a short message to inspect. The Library of Congress record names the artist and date and places it in U.S. Army recruitment.

James Montgomery Flagg, 1917, Library of Congress, digital ID ppmsc.03521. The archive reports no known restrictions on publication. This exact public-domain reproduction was encoded as WebP.

You can record three features of the words and image before asking how viewers may have felt:

• "I" gives the appeal a speaker, with Uncle Sam standing for the institution.
• "You" directly addresses the viewer, reinforced by the pointing finger and outward gaze.
• "Nearest recruiting station" connects the appeal with a place where the viewer can act.

A lens concerned with power might read this as a personal appeal made for an institution. That reading is separate from a claim that viewers felt forced to enlist. The poster does not tell you how a given viewer felt or acted.

The figure who stands for the army matters alongside the words. A semiotic reading asks how signs like this work. CDA still needs to link its reading of words and images with supported claims about the social setting.

The poster shows a limit of the corpus too. One text cannot show all wartime discourse or how different groups responded. To answer those questions, seek other artifacts and historical accounts of how people read them.

Test the interpretation against alternatives

Return to the whole text before explaining one sentence. A writer may leave out an actor because of a common form, lack of space or a name used earlier. Compare these readings with the proposed power claim before you choose one.

In their CDA guide, Dahlborg and colleagues separate description, interpretation and explanation. They also discuss reflexivity: thinking about how your own position affects the work.

Use these ideas with care for the approach you chose. The LSE guide to CDA traditions shows why methods can differ. State your basis so a reader knows which account of the method to use.

In the policy example, description notes the passive wording. Interpretation asks what it does to the actor's visibility. Explanation links the reading with social relations, using sources about context and the lens you stated.

Keep the source needs clear at each step. A grammar claim needs the passage. A claim about who can make a choice needs rules or office records. For reception, follow the IHR guide's distinction between text and setting and seek accounts of how people read it.

State your own position and starting ideas in the memo. Explain why the lens helps, which texts you left out and which readings you checked. This lets a reader inspect the choices and judge the support for each claim.

Code labels can help you find passages, and axial coding can relate categories in a grounded theory study. CDA needs its own stated method for language and power. Explain how any coding step serves that method.

Compare texts and context in Atlas

Create an Atlas project with the notices you may upload, the text around each extract, your question and the context sources. Give each a clear name so you can tell Notice A from the guide to its rules.

Choose Ask a question. Use @ to select the texts and context sources, then Send a bounded request. Name the lens and ask for the words to stay separate from claims about what they mean.

For this example, ask: "Compare who acts and whose voice is heard in Notices A and B. Show the exact words, actors named or left out, sources about context, other readings and claims the sources cannot support."

Open the numbered citations beside the answer. Notice A can show its own sentence, but may not show who approved the policy. Check whether a claim about that person's power points to a source that supports it.

Correct an answer that says A "denies appeals" when the selected sentence just leaves them out. Read the whole notice. If a later part gives a way to appeal, ask for a narrower account that includes it.

Use New and Note to save the checked table and reading. Wait for Saved. Keep open questions with the note so the next check starts with what you know and the gaps you still need to address.

Atlas helps compare the sources you supply. You still own the lens, the corpus, the account of context and the final critical claim. Review those choices alongside the cited answer before using it in your report.

Write a qualified analytical account

A defensible account states the scope before the claim: "In these two notices, A's opening sentence makes the decision-maker less clear than B's. We need to check the wider texts before saying this leaves recipients without a voice."

Then link what you saw with the sources you have. Explain what the rules show, which other readings remain plausible and why the lens helps answer the study's question. Keep the source of each part clear.

End the memo with the next source you need. Here, that might be the full appeal rules and interviews about how people read them. Seek that data to address an open claim, including findings that could change your first reading.