Screening Claims for Epistemic Structure

This paper sets out a screening protocol for sorting claims by their epistemic structure: by how a claim could be settled rather than by whether it is correct. The screen asks one prior question that most disputes skip: is this the kind of claim an observation could bear on, and if so, could any observation ever count against it? It returns one of three outcomes. A claim is testable when some observation could show it false. It is unfalsifiable when it is dressed as a claim about the world yet nothing could ever disconfirm it. It is non-empirical when it is not a claim about the world at all, such as a value, a definition, an aesthetic judgment, or a matter of proof. Where a claim's structure is not yet determinate, because context is missing or because readings genuinely diverge, the screen flags the claim as contested rather than forcing a confident label. The procedure is four steps: strip the framing to find the actual claim, ask whether it is empirical at all, apply a falsifiability test, and show the reasoning that produced the verdict. The value of the method lives in its hard cases: false-but-testable claims, retrodictions disguised as predictions, self-sealing structures, and claims that read as vague only because their source was withheld. It also lives in the method's honesty about a residual band where no confident label is warranted. The method scales from a single sentence to a whole paper, where the unit of judgment is not a simple average but a census of claim types together with the load-bearing claim or claims on which the rest depends. Throughout, the screen judges structure rather than truth. It runs before fact-checking and does not replace it.