Replication package: a computational census of the Chinese-language research apparatus on digital intangible-heritage safeguarding (2006-2026)

Replication package (version 1.4) for a computational census of the Chinese-language research apparatus on digital intangible-heritage safeguarding, 2006-2026. Version 1.4 adds an adversarial coverage audit of the critical-vocabulary dictionary (data/derived/adversarial_audit.csv): twelve candidate terms from the canonical Anglophone critical-heritage lexicon were scanned across all 2,553 title, keyword and abstract fields; only four of the nine additional terms produced any hit (11 unique records, 8 outside the operationalised dictionary, 5 of them in 2025-2026), moving the pooled 2025-2026 critical rate from 6.2% to at most 6.8%, while the canon's political core (colonial, Western-centr*, Eurocentr*, Orientalism, anti-heritage) stays at zero hits across all 21 years. The package contains: (1) versioned research instruments (five UNESCO-domain lexicons, a preregistered 18-term critical lexicon, a six-group critical dictionary, and funding-channel regular-expression rules with a manual-audit inference rule); (2) 24 aggregate tables covering domain prevalence and funding rates, funding-channel counts, a price-of-presence index with bootstrap confidence intervals, logistic funding models with a domain x document-type interaction test, critical-turn z tests, a six-group change-point scan with bootstrap, Cochran-Armitage trend tests, close-reading inter-coder agreement, a journal-tier (CSSCI) composition check, citation-rank construct validation, and the adversarial coverage audit; and (3) a single Python script that either recomputes every statistic from the licensed raw census (placed in data/raw by the user) or audits the aggregate tables against locked values (64 checks, all passing). The licensed bibliographic exports (CNKI custom export, Wanfang Data query results, Web of Science Core Collection export) are not redistributed; exact query strings and inclusion rules are documented in the preregistration at https://osf.io/wunza/. Prepared under anonymous review; no author or journal identifying information is included. License: CC-BY-4.0.