Filed under the wrong tongue

Claim (verbatim)

Colonial and early post-colonial catalogues of West African manuscripts were built by Arabists, for Arabic, with a 'language' field that in practice meant Arabic; African-language texts in Arabic script — Hausa and Fulfulde Ajami above all — were routinely swept into the Arabic count or left untagged, because the cataloguer read the script, not the language. The mechanism is category error at the point of description: an Ajami poem looks Arabic to a fast hand, so misclassification is systematic and one-directional (vernacular-read-as-Arabic, never the reverse), which means every genre and language statistic built on those catalogues overstates Arabic and understates Ajami by a measurable rate. The test is a recount: where a collection catalogued in the colonial style has since been re-examined by scholars who actually read the vernaculars, the corrected language tags should reveal a substantial buried Ajami fraction. Prediction: in at least one large West African collection whose items were catalogued before circa 1970 and later re-identified, re-examination will reclassify at least ten percent of items originally recorded as 'Arabic' or left untagged as containing substantial Hausa or Fulfulde Ajami, and the reclassification will run almost entirely one way, toward Ajami (primary clause: a ten-percent-or-more Arabic-to-Ajami reclassification rate on recount; the verdict follows it). Kill: not yet built as a single rate — construct it by diffing the original finding-aid language fields against current re-identifications for the Northwestern University Herskovits collections (Falke, Hunwick, Paden) and the West African Arabic Manuscript Database (AMMS), with the Boston University African Ajami Library as the vernacular-reading control.

Prediction clause (verbatim)

Prediction: in at least one large West African collection whose items were catalogued before circa 1970 and later re-identified, re-examination will reclassify at least ten percent of items originally recorded as 'Arabic' or left untagged as containing substantial Hausa or Fulfulde Ajami, and the reclassification will run almost entirely one way, toward Ajami (primary clause: a ten-percent-or-more Arabic-to-Ajami reclassification rate on recount; the verdict follows it).

Kill-dataset (verbatim)

Kill: not yet built as a single rate — construct it by diffing the original finding-aid language fields against current re-identifications for the Northwestern University Herskovits collections (Falke, Hunwick, Paden) and the West African Arabic Manuscript Database (AMMS), with the Boston University African Ajami Library as the vernacular-reading control.

Novelty / leakage triage

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

The systematic bias — Arabist cataloguers reading script not language, so Ajami is swept into the Arabic count — is a well-worked observation of the Ajami-studies literature (Mumin & Versteegh; Ngom). But the primary clause is a measured, one-directional reclassification rate: at least ten percent of a pre-1970-catalogued collection's 'Arabic' or untagged items re-identified on recount as substantial Hausa or Fulfulde Ajami. The conjecture states outright this 'is not yet built as a single rate,' and no published recount of a named collection reports such a finding-aid diff. The neighborhood is thoroughly worked; the specific rate is un-run.

• Meikal Mumin & Kees Versteegh (eds.), The Arabic Script in Africa (Brill, 2014)
• Fallou Ngom, Muslims beyond the Arab World: The Odyssey of Ajami and the Muridiyya (Oxford University Press, 2016)
• John O. Hunwick & R. S. O'Fahey (eds.), Arabic Literature of Africa (reference series, Brill, 1994-2003)

Predictions

No prediction registered yet.