Skip to content

Wording Test

A phrasing review — instantiates Premise–Action Decoupling

A sender-side check that vets a question, notice, or clause before it goes out — asking whether every permitted response would concede the same buried premise, and returning a neutral rewrite when it does.

Wording Test is a check applied to a question, survey item, notice, or clause before it is sent: does its phrasing force the responder to concede something just by answering? It runs the answer-space test — asking whether every permitted response ratifies the same buried proposition — and, where it does, returns the smuggled premise as an explicit claim plus a neutral rewrite that asks the real question and lets the premise stand or fall on its own. Its defining move is that it is evaluative and preventive: it does not merely describe what a message presupposes, it judges a candidate phrasing against a neutrality standard and certifies or rejects it, so the loaded question is caught at the drafting board rather than after someone has answered. Where Literal Content Parsing decomposes a message you have received, the Wording Test vets a message you are about to send.

Example

A trial attorney is preparing questions for her own witness on direct examination, where the rules bar leading questions — ones that put the answer in the witness's mouth. Her draft includes: "And after he threatened you, what did you do?" A Wording Test flags it: the phrase presupposes he threatened you, a contested fact the witness has not yet established, so every answer — "I left," "I called security" — silently ratifies the threat. The test returns the buried premise as its own claim (there was a threat) and a neutral rewrite that decouples it: first "What, if anything, did he say to you?" to establish the words, then "What did you do next?" The surface task — getting the witness's account — is preserved; the conclusion that it was a threat is no longer smuggled in by the question and must instead be built from testimony. Run before trial, the test also pre-empts a sustainable "leading" objection that would have interrupted the examination.

How it works

  • Draft, then interrogate the answer space. For each candidate item, lay out the realistic responses and check whether they all concede one proposition; if there is no answer that engages without conceding, the wording is loaded.
  • Extract and restate the smuggled premise. Name the presupposed claim explicitly, lifted out of the question, so it can be judged on its own.
  • Reword to decouple, then re-test. Produce a phrasing that asks the genuine question and leaves the premise separately answerable, and run the answer-space test again on the rewrite.

The distinguishing discipline is that it certifies phrasing against a neutrality bar — a pass/fail with a fix — rather than parsing a received message for analysis.

Tuning parameters

  • Neutrality bar — how strict the standard for "loaded." A high bar catches subtle leading but trips on ordinary shared ground no one disputes; a low bar lets soft framing through.
  • Answer-space completeness — whether you enumerate every realistic response or a sample. Fuller enumeration catches the no-honest-answer case but is slower.
  • Materiality filter — flag every presupposition, or only ones that would matter if conceded. Flagging trivially-shared assumptions is noise that buries the load-bearing one.
  • Rewrite ambition — from a minimal de-load to a full redesign of the item. More ambition yields cleaner instruments but reopens the design.

When it helps, and when it misleads

Its strength is catching the loaded question while it is still just a draft — the cheapest possible point to fix — and giving form-designers, pollsters, interviewers, and drafters a repeatable guard against leading their own respondents.

It judges form, not truth, and can be run backwards — used to find the minimal rewording that still lands the premise while passing as neutral, laundering a leading question into a subtler one. It also over-triggers on benign shared assumptions, and a bad-faith author can wave "that's a loaded question" at a perfectly fair item to kill it.[1] The discipline that keeps it honest is to treat a flag as "isolate and state this premise," not "this premise is false," and to gate on materiality so it fires on premises that actually matter.

How it implements the components

Wording Test realizes the phrasing-vetting components — the ones a neutrality check produces on a candidate message:

  • embedded_premise — its output: the proposition a candidate phrasing would smuggle, extracted and stated as an explicit claim.
  • ratification_inference_rule — it identifies the "answer ⟹ concede" inference the wording relies on (every permitted response ratifies the premise), which is exactly what makes an item loaded.
  • materiality_threshold — it flags only presuppositions material enough to matter, so ordinary shared ground is not treated as smuggling.

It vets outgoing phrasing but does not decompose a received message into its literal and presupposed layers (that's Literal Content Parsing), test whether a real-world convention's preconditions for ratification actually hold (that's Context-Condition Testing), or reserve a position on the record when a loaded item can't be avoided (that's Presupposition Challenge Marker).

References

[1] On direct examination most evidence rules — for example U.S. Federal Rule of Evidence 611© — bar leading questions, those that suggest the answer or assume an unproven fact, precisely because the form lets the questioner, not the witness, supply the premise. The wording test applies the same scrutiny to any instrument that must not lead its respondents. ↩