Skip to content

Literal Content Parsing

A parsing method — instantiates Premise–Action Decoupling

Splits a message into what it literally asserts or asks and what it merely presupposes, so the presupposition surfaces as a claim in its own right instead of riding in unexamined.

Literal Content Parsing takes a single message — a question, a form field, a clause, a subject line — and separates two layers that ordinary reading fuses: the at-issue content it actually asserts or asks, and the presupposed content it quietly takes for granted. The presupposition is what a reader would have to already accept for the surface message to make sense, and it is precisely the part that never gets stated as a claim. The move that makes this mechanism itself, and not one of its siblings, is that it works purely at the semantic level of the wording — before any judgment about the sender's intent, the force of the act, or whether the buried claim is true. It answers one question only: what does this sentence literally commit itself to, and what is it leaning on without saying?

Example

A product team fields a customer-experience survey whose flagship item reads: "How much has the new dashboard improved your team's reporting?" — answers on a 1-to-5 scale. Parsed literally, the item has two layers. The at-issue content is a scalar the respondent is asked to supply: the degree of improvement. The presupposition, carried by the change-of-state verb improved, is a standalone claim the item never puts up for inspection: the new dashboard has improved reporting. The literal scope is narrower than it reads, too — your team, reporting specifically, since rollout — not "your work overall, ever."

Laid out that way, the defect is obvious: a respondent whose reporting got worse has no honest cell to tick, because every answer on the scale concedes the buried claim. The parse doesn't fix the survey; it hands the team the extracted premise ("dashboard improved reporting") and the true at-issue question ("did it change reporting, and in which direction?") as two separable things, so the item can be rewritten to ask the second without asserting the first.

How it works

The parse is mechanical, driven off the linguistic form rather than a reading of intent:

  • Scan for presupposition triggers — definite descriptions (the backlog), change-of-state verbs (stopped, improved, resumed), factives (realize, regret that P), clefts, and comparatives each smuggle a proposition the sentence treats as settled.
  • Restate each trigger as a bare declarative — turn the presupposition into a flat claim someone could affirm or deny ("the backlog exists and is theirs"), stripping it out of the question or instruction that carried it.
  • Pin the literal scope — resolve the quantifiers, time frame, and conditions actually written, separating them from the broader reach the phrasing implies.

The distinguishing discipline is literalism: it extracts what is on the page, not what the sender probably meant. Deciding whether an extracted premise is illegitimate, or how to answer around it, is deliberately someone else's job.

Tuning parameters

  • Trigger sensitivity — how many presupposition triggers you flag. High sensitivity catches subtle factives and definite descriptions but buries the load-bearing premise under trivial ones; low sensitivity is faster but misses the quiet ones.
  • Restatement literalism — how strictly you paraphrase versus charitably interpret. Strict literalism avoids importing intent but reads as pedantic; charity is smoother but re-fuses the layers you were trying to split.
  • Parse granularity — clause-by-clause versus whole-document. Fine granularity surfaces nested and cross-sentence presuppositions; coarse is quicker and better for triage.
  • Scope strictness — how aggressively you narrow quantifiers and time frames to what is literally written ("customers surveyed" vs. "all customers").

When it helps, and when it misleads

Its strength is that it makes a smuggled claim answerable: once the presupposition is sitting on the table as a flat declarative, it can be accepted, denied, tested, or rewritten — none of which is possible while it hides inside a question. It is the front-end almost every other mechanism in this archetype consumes.

It misleads in two directions. Over-literal parsing treats every presupposition as suspect, but most shared context is legitimate and necessary — flagging all of it is not rigor, it is paralysis, and it shades into weaponized literalism, feigning misunderstanding to dodge a good-faith request. And a clean parse says nothing normative: it does not tell you the buried claim is false (that is a test of evidence) or that engaging would concede it (that is a matter of force). The classic degenerate case it is built to catch is the loaded or complex question, where the presupposition is exactly what the asker wants nodded through.[1] The discipline that keeps it honest is to keep identifying a presupposition strictly separate from judging it, and to escalate only the material ones.

How it implements the components

Literal Content Parsing realizes the semantic-decomposition components — the layer that turns one utterance into separable claims:

  • surface_action — isolates the literal at-issue task the message performs: the question actually asked, the field to be filled, the box to be ticked.
  • embedded_premise — extracts each presupposition and restates it as a standalone claim, lifted out of the question or instruction that concealed it.
  • scope_or_limit — resolves the literal quantifiers, time frame, and conditions of the wording, holding them apart from the wider reach the phrasing insinuates.

It does not classify the act's force or what response it compels — engagement_assent_coupling and response_coupling_rule are Force Classification's; and it does not test whether the extracted premise is warranted (evidence_test) — that is Assumption Audit.

References

[1] In semantics a presupposition is content a sentence takes for granted rather than asserts — it survives negation ("the dashboard hasn't improved reporting" still presupposes there is reporting). The "complex question" (loaded question) is the degenerate case in which the presupposition is the real payload; parsing is the standard way to expose it. ↩