Attested lineages as a third robustness test for quantitative typology: data and code

Data, code and derived tables for a preregistered robustness analysis that re-tests three published typological findings on attested earlier and later stages of the same lineages, as a third frame beside balanced sampling and statistical bias control. The universe is the 27 supramillennial lineages frozen on 17 August 2026 (OSF KY6JW). Version 1 contains: shcherbakova_metric.py, a Python implementation of the nominal and verbal coding metrics of Shcherbakova et al. (2023, Linguistics Vanguard 9(s1): 155-167) following their published R code; Shcherbakova_Tier1_reproduction.xlsx, which reproduces their published 244-language scores from their own data to their rounding and applies the metric to the Grambank v1.0.3 doculects of the lineage universe; and the preregistration document deposited on OSF Registries. Results tables for the Universal 43 and noun-phrase order tests will be added in a later version. Grambank data are used under CC BY 4.0 (Skirgård et al. 2023, Science Advances 9(16): eadg6175).