Calculated Definiteness Across Tokens and Kinds in Jordanian Arabic and Its Implications for Article Typology

Jordanian Arabic uses a single definite article, al-, across functions that other languages split across multiple morphological forms. This paper introduces the Jordanian Arabic Definite Article Corpus (JADAC), the first spontaneous-speech corpus documenting the functional distribution of al- (N = 2658 NPs; approximately five hours of YouTube speech across three formats). When quantified against a crosslinguistic referential typology, the distribution follows a clear three-way pattern: 41.5% of NPs serve canonical definiteness functions, 18.1% build genus-level kinds in generic statements, and the remainder follow the expected indefinite pattern, except for a small subset, roughly 5%, showing structurally governed form-function mismatches. We argue that al- operates through two operations. The first isolates unique or maximal referents in episodic contexts; the second builds abstract kinds from predicates in generic ones. Most apparent mismatches follow from these same mechanisms rather than from semantic attrition. Within this mismatch subset, we identify a previously undescribed reverse pattern. Elative and ordinal constructions appear without the article despite denoting semantically definite referents, the inverse of the well-documented case where definite marking fails to deliver it. These findings supply new empirical data for formal accounts of kind-reference and article typology from a language and construction type not previously documented.