Decoding gets you the words. Meaning gets you what they say. CSAF shows you what they are doing.
Educators teach reading comprehension as a single skill. In practice, it has layers. Decoding is one layer. Meaning is another. There is a third that has not been named or taught consistently: the construction layer. What the words are doing to the reader. Whether they move, inform, or frame.
CSAF makes that layer observable. Ten signals. Three categories. One shared vocabulary. Built so every reader, educator, and researcher can name what they see.
CSAF is built to work alongside the tools you already use. It gives them a shared foundation, grounded in language and aligned to UNESCO and Ofcom frameworks.
Reading comprehension at the construction level has historically been taught as a judgment skill: deciding whether a source is trustworthy, whether an article is biased, whether a claim is true. Judgment is essential. But judgment is hard to teach consistently, hard to assess, and hard to defend in a classroom of thirty readers who disagree about what "fair" means.
CSAF treats comprehension at the construction level as an observation skill first. Before readers judge, they identify what is actually present in the article: the language patterns, the sourcing structure, the framing choices, the use of evidence. Observation is teachable, consistent across classrooms, and measurable with pre/post assessment. Judgment then has somewhere solid to stand.
Because the underlying patterns such as emotional activation, selective sourcing, false precision, and missing context appear across all persuasive media, the skill readers build through news transfers directly to media literacy, digital literacy, and AI literacy. News is simply the most structured place to learn it.
Before readers judge, they observe.
That is the methodological shift that makes reading comprehension at the construction level teachable, consistent across classrooms, and measurable with pre/post assessment.
CSAF starts with three observations a reader can make on any article. From there, ten signals provide the analytical depth. Six recurring patterns describe how those signals combine in practice.
How is the article trying to push or pull the reader?
Move language is the part of the article doing emotional or directional work. It is what the words are trying to do to the reader, regardless of what they are saying.
What material is the article giving the reader to think with?
Inform language is the substance. The named, sourced, specific, and verifiable content that lets a reader build their own understanding.
What shape is the article giving the story, and does that shape hold up?
Frame language is structural. It is the choices the article makes about what to include, what to leave out, and what to call things.
Each signal is built the same way — a question a reader can ask of any article, a clear definition of what high and low look like, and a transparent path from the words on the page to a score you can stand behind.
Looking for outside research to supplement our signals? Our Ask Clear-Sight module is built to deliver lateral reading context and sources to any article.
Every signal delivers three layers of validation: a quantitative score, a qualitative explanation of why the article earned that score, and direct evidence pulled from the text. No black boxes. No unexplained numbers. The score tells you where the article lands. The explanation tells you why. The evidence lets you verify it yourself.
How the article tries to push or pull the reader.
Does the article present the subject fairly or does it favor one perspective?
Balance measures the structural choices an article makes — how it frames the argument, whose voice gets space, and whether the reasoning is sound. It is not about word count. It is about whether the construction of the story gives the reader an honest view of the landscape.
Does the article inform or does it activate?
Logic measures the emotional intensity of the language used and the effect it is designed to have on the reader. A high score means the article delivers information in a measured tone. A low score means emotional language is doing more work than the facts are.
Does the article let you reach your own conclusion?
Autonomy measures the pressure an article applies to steer the reader toward a specific outcome. A high score means the article trusts you to decide. A low score means the structure of the story is designed to leave you with only one acceptable conclusion.
The material the article gives the reader to think with.
How much of this article is verifiable fact versus opinion or assertion?
Evidence measures the ratio of checkable, concrete facts to unsubstantiated claims. It does not evaluate whether those facts are correct — it measures how much of the article's weight is carried by verifiable information versus stated opinion.
Who is speaking in this article and how transparent is their role?
Sourcing measures the quality, diversity, and transparency of the voices used to build the story. A high score means sources are named, credible, and represent a range of relevant perspectives. A low score means the article leans on anonymous sources, a single voice, or sources whose interests are not disclosed.
How precise is the language?
Specificity measures whether the article uses concrete, verifiable detail or relies on vague language that sounds authoritative but resists scrutiny. Precision is a marker of rigor. Vagueness is often a marker of something else.
Are the numbers and statistics in this article used honestly?
Claims measures how the article handles quantitative information — whether data is presented with appropriate framing, whether statistics are contextualized, and whether numbers are used to illuminate or to persuade. Clear-Sight does not fact-check claims against outside sources. It evaluates how claims are constructed and used within the article itself.
The shape the article gives the story, and whether that shape holds up.
Does the article give you what you need to understand the story?
Context measures whether sufficient background, history, and surrounding circumstances are provided for the reader to fully evaluate what they are reading — or whether the story assumes knowledge the reader may not have.
Does the article treat its subject with the complexity it deserves?
Nuance measures whether the story acknowledges competing interests, legitimate uncertainty, and the inherent difficulty of most important issues — or whether it reduces them to something simpler than they actually are.
Does the article hold together?
Consistency measures whether the article's claims, framing, and tone remain coherent from beginning to end. A high score means the story does not contradict itself. A low score means the article shifts in ways that undermine its own argument or presents information that does not align across sections.
Take it further
Clear-Sight has identified twelve patterns so far, and the library continues to grow as we analyze new articles. The six examples below are the ones that anchor the current curriculum because each teaches something distinct about how media construction works. They are designed to be used in classrooms, newsrooms, conference talks, lesson plans, and academic papers.
What it teaches: Low sourcing is not always a red flag. Context across the full profile tells the real story.
Both good and bad breaking news have low Evidence and low Sourcing. But everything else diverges. The bad article compensates for thin sourcing with emotional activation, directional framing, false precision, and selective context. The good article compensates with measured tone, honest uncertainty, and internal coherence.
In breaking news, judge the article not by what it knows but by how it handles what it does not know.
What it teaches: Precision is not the same as proof. Specific-sounding language is one of the most effective credibility signals an article can manufacture.
The signature is high Specificity paired with low Evidence. Numbers, names, and precise language create an impression of rigor. But underneath, the claims lack verifiable grounding. Consistency is high because it is constructed, not reported.
The more specific an article sounds, the more carefully you should ask whether that specificity is grounded. Precision is easy to manufacture. Verification is not.
What it teaches: Emotional activation and directional pressure together are the signature of content designed to move people, not inform them.
Logic and Autonomy are both low. Emotional language dominates and only one acceptable conclusion is presented. Facts may be present but in service of the emotional narrative. Nuance is absent because complexity would interrupt the emotional momentum.
When Logic and Autonomy are both low, ask what the article wants you to do. Because it wants you to do something.
What it teaches: A perfectly coherent article can still be a narrow one. Consistency is not a substitute for completeness.
This is the pattern that feeds algorithmic reinforcement. It feels credible because it never contradicts itself, but only because it never introduced anything that would. One perspective dominates throughout. Tone is often measured. Direction is embedded in the framing, not the tone.
Ask not whether the article holds together but whether it ever introduced anything that could complicate it. An article that never challenges its own premise is not rigorous. It is selective.
What it teaches: Well-attributed quotes from a single perspective is still one-sided journalism. Attribution is not balance.
This is the most sophisticated pattern because it passes every basic credibility check a reader would run without a framework. Sourcing is high — named, credentialed sources. The article looks rigorous. But Balance is low. Only one side's complexity is explored. Only one side's sources are sought.
Before you trust an article because it is well-sourced, ask whose sources they are. Attribution tells you where the information came from. Balance tells you whose information was sought.
What it teaches: Clear-Sight does not detect AI authorship. But AI-generated content tends to exhibit a specific construction pattern — high surface credibility with low depth underneath.
The signature is a cluster of high surface signals — Specificity, Consistency, Logic — combined with low depth signals — Nuance, Context, Evidence. The article feels complete. It reads smoothly. Nothing jars. But when you look for genuine complexity, genuine tension, genuine depth of context, it is not there.
The question is not whether an article was written by a machine. The question is whether the article has the depth that genuine reporting produces. High surface credibility combined with low depth is the pattern worth learning to recognize.
CSAF is the same framework wherever it goes. The conversation it opens depends on who you are. Choose the path that fits.