Quoted Language Hypothesis
Tags: abbey-root • research • voice-analysis • hypothesis
Quoted Language Hypothesis
The quoted-language evidence supports a narrower claim than “uses quotation marks for humor.”
HYP-004 proposes that quotation marks sometimes function as stance markers. They tell the reader that wording is disputed, nonliteral, inadequate, or being replaced with a comic alternative.
The hypothesis deliberately excludes ordinary titles, neutral direct quotation, reported speech, copied material, and malformed punctuation.
Validation will not reuse the canonical evidence set. It starts from the 165 reviewed candidates and removes the supporting examples, comparisons, and rejections cited by EVID-004.
The remaining holdout must meet four requirements: a 30% core rate, at least ten cases of each core subtype, recurrence across at least three chronological bands, and at least twenty comparison cases.
The comparison threshold matters. A method that labels nearly every quotation as support has failed to distinguish the proposed characteristic.
HYP-004 also contains a narrowing rule. If distancing replicates but comic renaming does not, or vice versa, the aggregate result cannot hide the failed subtype.
The chain now has an observation, complete evidence review, and a testable hypothesis. The next artifact can calculate VAL-004 deterministically from the already reviewed holdout.