Iberian Lead Plaques: Empirical and Computational Decipherment via the Fractional Operator -erter (1/2) and The Roman Institutional, Cadastral, and Contractual Architecture — Executive Synthesis (EN / ES / DE)

EXECUTIVE THESIS: THE BINOMIAL KEY (-ERTER [1/2] + ROMAN INSTITUTIONAL ARCHITECTURE) ACROSS N = 2,090 HESPERIA RECORDS • 1. The Archival Premise (>90% Economic & Administrative Corpora): Over 90% of ancient non-monumental archival texts on clay, lead, or wooden tablets (Proto-Cuneiform Uruk [Nissen, Damerow & Englund 1993], Mycenaean Linear B [Ventris & Chadwick 1956], Greek commercial lead plaques from Pech Maho and Emporion [Lejeune, Pouilloux & Solier 1988; Sanmartí & Santiago 1987], and Roman tablets from Pompeii, Puteoli, and Vindolanda [Bowman & Thomas 1994; Camodeca 1999]) are accounting ledgers, cadastres, and commercial contracts. Traditional philological classification (Untermann, MLH III) prudently cataloged most recurring Iberian stems as personal names (anthroponyms), while computational decipherment models using phonetic priors on undersegmented scripts (Luo, Hartmann, Santus & Barzilay 2021, TACL 9, building on Snyder, Barzilay & Knight 2010 and Luo, Cao & Barzilay 2019) faced the absence of an explicit institutional and syntactic prior for lead plaques. • 2. The Decipherment Engine (The -erter [1/2] Operator + Roman Cadastral & Contractual Parallels): Combining the consensus Iberian fractional operator -erter / eŕte- ('one-half', 1/2; Orduña 2005; Ferrer i Jané 2007, 2009) with the institutional architecture of Roman cadastral and partnership law (Forma Censualis, Ulpian Dig. 50.15.4; Societas, Dig. 17.2 and Gaius Inst. 3.152) provides a testable functional model. The direct suffixation/prefixation of -erter / eŕte- to baśir (baśirerter in V.13.03 Pos[92]; eŕtebaśká in GI.10.11 Pos[12]), ortin (ortinbeŕeteŕeikiaŕ in T.07.03 Pos[01]), and laku (lakueŕterḿi on the Alloza ceramic vessel TE.06.05; cf. śorlakuḿiuneri : berter ... anaterter in V.13.02 Pos[14–17]) proves empirically at a first level that these stems denote quantifiable magnitudes divisible by 50%, which we model at a second institutional level as divisible patrimonial estates (dimidia pars fundi) and partnership shares (societas dimidiae partis; alongside metrological alternatives on ceramic vessels, Ferrer i Jané 2009). • 3. Corpus-Wide Traceability (N = 2,090 Hesperia Records; 4,771 Tokens): Audited algorithmically in Corpus_Iberico_Auditado_y_Clasificado_2090_v4_RevC_ES.csv, 159 of the 165 long Iberian inscriptions (>= 5 words; 96.4% of long texts, totaling 2,333 tokens) are classified within the syntactic functional catalog, achieving 55.64% strict automated token recognition (1,298 / 2,333 tokens) and 60.61% unweighted mean recognition across the 398 glosable inscriptions of Table A.2 (2,915 tokens; 57.50% weighted: 1,676 / 2,915 tokens).