The songbook breaks the song-maker

Status: Anticipated · untested

Status is derived only from the shepherd-authored triage/prediction data above -- community submissions and claims are a separate overlay and can never change it (see the participation panel below).

This is a proposed connection between two domains, generated by a language model. It is not an article and not evidence: it sits below the evidence/publication boundary. A quantitative prediction and a named kill-dataset are attached (when registered) so the claim stays falsifiable rather than merely evocative.

Claim (verbatim)

Within a living oral-epic tradition, some singers learned songs from cheap printed or written songbooks rather than from other singers. The surprising connection is that the written intermediary breaks the song-maker while polishing the parrot: a text-learned song is memorized rather than regenerated, so its performances should show lower formulaic density yet higher word-for-word stability across re-performances — two statistics moving in opposite directions that jointly detect a written link anywhere in a chain of transmission. The singers made it so unknowingly, because the formulaic engine is trained by hearing composition happen, not by reading its frozen output. If this holds, verbatim fidelity in any pre-print oral-derived corpus flips its meaning: it becomes evidence for written mediation rather than for prodigious oral memory.

Prediction clause (verbatim)

Compare singers documented as having learned particular songs via songbooks with singers in unbroken aural chains. Primary clause, which decides the verdict: songbook-mediated performances show formula density at least 15% lower for comparable song-types. Secondary clause: their re-performance word-identity rate is at least 15% higher.

Kill-dataset (verbatim)

The digitized Milman Parry Collection of Oral Literature: its recordings and transcriptions together with the collectors' interview protocols documenting each singer's source of learning.

Nobody has run this test. The kill-data is named above. If you can run it — or you know the paper that already settles it — claim the kill or submit the prior. Kills and priors are credited here, by name, as they come in.

In the atlas

This conjecture is bridged, as an L1 lead, onto these Inferpedia subject pages.

Provenance

Run: Fresh agent generation · model: claude-fable-5

Composed blind by claude-fable-5 from internal knowledge with zero tool use and emitted as a single text message.

Novelty / leakage triage

anticipated in the literature — this exact test has never been run

Lord documented singers who learned songs from published songbooks and warned that such singers memorize fixed texts rather than compose — both predicted directions are qualitatively anticipated in The Singer of Tales. The paired quantitative signature (lower formula density AND higher re-performance word-identity, with thresholds) has never been measured on the Parry Collection.

Albert B. Lord, The Singer of Tales (1960), on songbook-learned guslari and memorization vs composition

Predictions

No prediction registered yet.

Weigh in

No community feedback yet.

Add your take

Posted immediately (spam is removed). Community feedback is never an adjudicated verdict and never changes this conjecture's triage label or status above.