CDA-Activist Lite

A streamlined version of the interventionist Critical Discourse Analysis framework, optimized for shorter AI-related texts like press releases, announcements, op-eds, and corporate communications. Same theoretical depth, faster analysis.
Why a "Lite" Version?β
The full CDA-Spicy (Activist) framework asks for 10+ instances per taskβideal for long-form interviews, policy documents, or speeches. But when analyzing a 500-word press release, that density becomes counterproductive. The model either pads with weak examples or, doing what LLM's do, crafts "plausible" instances.
CDA-Spicy Lite solves this by:
| Aspect | Full Version | Lite Version |
|---|---|---|
| Instances per task | 8-12 | 4-6 |
| Ideal text length | 2,000+ words | 300-1,500 words |
| Analysis time | 60-90 seconds | 30-45 seconds |
| Token cost | Higher | Lower |
| Theoretical depth | Full | Full (unchanged) |
The theoretical apparatus remains identical: Halliday's SFL, Fairclough's CDA, and Critical Theory (Adorno, Marcuse, Jacoby). Only the instance counts are reduced.
What This Framework Analyzesβ
The Lite framework executes the same five-task structure as the full version:
Task 1: Agency & Accountability Audit
Who acts, who is acted uponβand why it matters
Identifies how texts manipulate agency through:
- Erasure β Passive voice, agent deletion
- Delegation β Attributing human decisions to systems or processes
- Diffusion β Spreading responsibility across abstractions
- Inversion β Positioning powerful actors as servants or benefactors
- Collectivization β "We all" constructions that manufacture consent
- Personification β Granting agency to technologies or markets
Task 2: Ideology & Common Sense Audit
The politics of word choice
Exposes how "neutral" vocabulary smuggles contested values:
- Euphemisms that sanitize harm
- Metaphors that naturalize power
- Cultural models that privilege certain worldviews
- Lexical choices that foreclose alternatives
Task 3: Positioning & Solidarity Audit
Creating "Us" and "Them"
Reveals how texts construct social relationships:
- Pronoun strategies (inclusive "we" vs. distancing "they")
- Register and formality as power markers
- Presuppositions that assume agreement
- Voice representationβwho speaks, who is spoken about
Task 4: Discourse Strategies
Pattern synthesis across tasks
Identifies 2-3 overarching strategies that cohere across the text, citing specific instances from Tasks 1-3.
Task 5: Structural Mystification
Reification, Amnesia, and False Separations
Applies three Critical Theory concepts:
- Reification β Where social relations appear as natural objects or autonomous forces
- Social Amnesia β What historical struggles or alternatives are systematically forgotten
- False Separation β How structural problems are privatized as individual issues
Plus a synthesis connecting all three mechanisms.
Conclusion: Stakes & Counter-Discourse
The framework demands:
- Naming the ideology β What worldview does this text construct? Whose interests does it serve?
- Counter-discourse principles β What would resistance sound like?
- Concrete reframings β Original passages rewritten to restore agency, recover history, and reconnect to structure
Theoretical Foundationβ
This framework operationalizes insights from:
Systemic Functional Linguistics (Halliday)
- Language as social semiotic
- Transitivity and agency distribution
- Register and social positioning
Critical Discourse Analysis (Fairclough, van Dijk)
- Language as practice of power
- Ideology naturalized through discourse
- Intertextuality and social cognition
Critical Theory (Adorno, Marcuse, Jacoby)
- Reification β social relations appearing as natural objects
- Social amnesia β systematic forgetting of alternatives
- False separation β privatizing structural problems
- The dialectic of individual and society
The prompt treats the boundary between individual and society as ideologically constructed rather than natural. It operates from a tradition that recognizes all discourse analysis already embeds political commitmentsβthe question is whether those commitments remain tacit or are made explicit and accountable.
Limitations & Caveatsβ
This is a probabilistic text generator executing instructions. It cannot "think" or "know" anything about the world. The outputs are pattern-matched responses shaped by the prompt's constraints.
What works well:
- Identifying linguistic mechanisms (passive voice, nominalization, etc.)
- Generating alternative framings
- Structuring analysis according to theoretical categories
- Producing counter-discourse examples
What requires human judgment:
- Validating whether identified instances are genuinely present in the text
- Assessing the political/material stakes (the model has no stakes)
- Determining whether the analysis serves your research goals
- Connecting findings to specific historical or institutional contexts
The prompt makes its commitments explicit: it centers those harmed by mystification, recovers suppressed histories, and demands imagination of alternatives. It treats neutrality as a stance that, by default, reproduces existing power.
Licenseβ
Discourse Depot Β© 2026 by TD is licensed under CC BY-NC-SA 4.0 Contact Me
See Alsoβ
- CDA-Spicy (Full Version) β For longer texts requiring deeper analysis
- CDA-Soft β Forensic approach without interventionist stance
- Metaphor Audit β Cognitive linguistics framework for AI discourse