Cross-Lingual Visibility in a Bilingual Urban Crime Corpus: Reproducibility Package for Khulna, Bangladesh, 2019–2026

This reproducibility package accompanies the manuscript “Cross-Lingual Visibility in a Bilingual Urban Crime Corpus: Event Selection and Reporting Ecology in Khulna, Bangladesh.” The deposit provides the public statistical and documentation materials supporting a longitudinal bilingual reconstruction of publicly reported, web-retrievable crime events in Khulna Metropolitan Police jurisdiction, Bangladesh, from January 2019 to June 2026. The analytic corpus contains 427 reconstructed events and evaluates whether each event has at least one linked English-language document within the bilingual reconstructed event universe. Contents include: • events_public_statistical.csv: event-level analytic data used for reported descriptive statistics, regression models, tables, and figures.• search_log.csv: 1,260 pass-level retrieval-log records across 90 monthly search packages.• source_families.csv: 699 period-specific source-family records supporting the reporting-ecology analysis.• corpus_codebook.csv and retrieval_and_adjudication_protocol.md: field definitions, inclusion and exclusion rules, event-linkage procedures, and the 14-pass monthly bilingual retrieval protocol.• Editable Excel source files for Figures 1–4 and Supplementary Tables S1–S7. The deposit does not redistribute copyrighted article text. To reduce re-identification risk for sensitive crime cases, it also excludes article titles, publisher URLs, precise locations, free-text incident descriptions, ambiguity notes, claim-level narratives, and restricted linkage/provenance files. The released event-level data retain only the analytical variables necessary to reproduce the reported results. The primary outcome is English visibility within a bilingual reconstructed corpus; it is not an independently reconstructed English-only crime dataset. Users should preserve the stated uncertainty, allegation-status, and responsible-data-use safeguards when reusing these materials. For access to restricted corpus-construction files for legitimate scholarly verification, contact the corresponding author and use an appropriate data-use agreement.