The lexicon outruns the landfill

Claim (verbatim)

The great Old Babylonian thematic lexical lists enumerate the material world by category - Ura = hubullu marches through trees and wooden objects, reeds, containers, textiles, leather, metals, animals, and foodstuffs across its 24 tablets, and the list of professions (Lu) names hundreds of titles. Many named categories are things that, had they generated documents or borne writing, should surface in the administrative record - yet some never do. A lexical list preserves a culture's taxonomy of the possible; the documentary and artefactual record preserves only the sampled actual. The gap between the two is a measure of what existed unrecorded or unrecovered - the lexicon reaching past the landfill. The mechanism is that scribal lists aimed at completeness of the category system, whereas excavation samples only the documents that a given office happened to produce and bury.

Prediction clause (verbatim)

Prediction: cross the object, commodity, and profession terms enumerated in Ura and Lu (via ORACC DCCLT lemmatization) against their attestation in dated administrative and legal documents (BDTNS for Ur III, Archibab for Old Babylonian, ORACC administrative corpora); at least 20 percent of the lexically-enumerated concrete nouns and profession titles have zero attestation outside lexical and literary contexts (primary clause: the lexical-only fraction >= 0.20 among concrete nouns and titles). Disambiguation: restrict to concrete common nouns and profession titles, excluding divine names, toponyms, and rare logograms flagged as scribal artifice; 'attested' means the lemma appears in a non-lexical, non-literary document. Coverage guard: separate terms that are hapax within the list itself (possible scribal invention) from well-integrated entries, and report the fraction both ways.

Kill-dataset (verbatim)

Kill (not yet built): the join between DCCLT lemma inventories and the administrative lemma indices of BDTNS, Archibab, and ORACC is not a single existing dataset - it requires assembling a cross-corpus lemma concordance from the four instruments' public exports. The sources are real and the operation is defined; the concordance itself must be built (per-instrument lemma lists are downloadable now).

Novelty / leakage triage

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

That the lexical lists enumerate past the documentary record - entries carried for the completeness of the category system with little or no attestation in use - is a stated theme of the lexical-tradition literature (Veldhuis on the lists' systematizing reach; Civil on lexicography's relation to usage), but the >= 0.20 lexical-only fraction over a DCCLT-by-BDTNS/Archibab lemma concordance has never been computed, and the kill itself concedes the concordance must first be built. Minor factual note: the 24-tablet division of Ura is the first-millennium canonical form, not the Old Babylonian recension the claim leads with.

• N. Veldhuis, History of the Cuneiform Lexical Tradition (GMTR 6, Munster, 2014)
• M. Civil, 'Ancient Mesopotamian Lexicography', in J.M. Sasson (ed.), Civilizations of the Ancient Near East (New York, 1995)

Predictions

No prediction registered yet.