The kingdom serialized its memory

Claim (verbatim)

The largest fixation of oral literature anywhere in Oceania was not an ethnographic project. Between 1834 and 1948 Hawaiians produced over a hundred Hawaiian-language newspapers — roughly 125,000 pages — and used them, among everything else, as the publication venue of the old corpus: serialized moʻolelo (Haleʻole's Lāʻieikawai ran in 1862–63 before becoming the first Hawaiian novel), Samuel Kamakau's histories running for years in the weeklies, mele of every genre, and kanikau — composed laments — by the thousand, printed by kin as an ordinary funerary act. Because this fixation ran through commerce and community rather than scholarship, it was never treated as the archive it is: Puakea Nogelmeier's study of the corpus (Mai Paʻa i ka Leo, 2010) puts the portion ever translated or seriously indexed at under two per cent. The structural consequence is a claim about witnesses: material fixed by ethnographers got re-collected, cross-filed and cited, while material fixed by the nūpepa mostly went nowhere else — so the print archive of Hawaiian orature should be overwhelmingly a sole-witness corpus, a library of unica in plain sight, and the loss of any volume of it (the runs that survive in single files, the papers with no surviving file at all) is the loss of the only witness. Prediction: drawing a systematic sample of traditional-genre items (mele, kanikau, moʻolelo installments) from the digitized nūpepa corpus and searching each against the manuscript and ethnographic record, over 70% will prove to have no witness outside the newspapers, and traditional-genre items across the corpus will number over ten thousand (primary clause: the 70% sole-witness share; the verdict follows it). Kill (not yet built, from real sources): a sole-witness census over the digitized newspaper corpus in Ulukau and the Papakilo Database, checked against the Bishop Museum Library's mele and Hawaiian Ethnological Notes collections, with Nogelmeier, Mai Paʻa i ka Leo (2010) as the framing control.

Prediction clause (verbatim)

Prediction: drawing a systematic sample of traditional-genre items (mele, kanikau, moʻolelo installments) from the digitized nūpepa corpus and searching each against the manuscript and ethnographic record, over 70% will prove to have no witness outside the newspapers, and traditional-genre items across the corpus will number over ten thousand (primary clause: the 70% sole-witness share; the verdict follows it).

Kill-dataset (verbatim)

Kill (not yet built, from real sources): a sole-witness census over the digitized newspaper corpus in Ulukau and the Papakilo Database, checked against the Bishop Museum Library's mele and Hawaiian Ethnological Notes collections, with Nogelmeier, Mai Paʻa i ka Leo (2010) as the framing control.

Novelty / leakage triage

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

Nogelmeier's Mai Paʻa i ka Leo (2010) establishes the frame — over a hundred Hawaiian newspapers, roughly 125,000 pages, under two per cent ever translated or seriously indexed — and the nupepa as the great uncollated archive of Hawaiian orature is a recognized point (Silva, Aloha Betrayed, 2004, works the same corpus). But the primary clause is an un-built census: a systematic sample of traditional-genre items searched for outside witnesses to establish a sole-witness share above 70%, plus a count over ten thousand items. The kill is honestly flagged not-yet-built; the framing is in print, the census un-run. Adjacent.

• M. Puakea Nogelmeier, Mai Paʻa i ka Leo: Missionary Influence on Hawaiian Orality and Literacy (Bishop Museum Press, 2010)
• Noenoe K. Silva, Aloha Betrayed: Native Hawaiian Resistance to American Colonialism (Duke University Press, 2004)

Predictions

No prediction registered yet.