Data Sources & Licenses
I built this on top of several open linguistic datasets. Every analysis you run surfaces data from one or more of them. Their authors did the work; this page names them and their licenses so attribution travels with the data.
| Source | License | What I use it for |
|---|---|---|
| Wiktionary | CC BY-SA 4.0 | Etymology text, definitions, pronunciations, related words |
| Glottolog | CC BY 4.0 | Language families, tree structure and representative coordinates. Coverage varies by artifact; the research sources name each normalization and checksum. Glottolog 5.3: DOI 10.5281/zenodo.18840935. |
| Lexibank | CC BY 4.0 (per dataset) | Expert-annotated cognate sets (4,981 sets, 25,741 members). Cards on the etymology page mark Lexibank-sourced entries with a ★ badge. |
| IE-CoR | CC BY 4.0 | Indo-European Cognate Relationships: 25,731 lexemes across 160 languages with LIV²/NIL references. Distributed via Lexibank. |
| WOLD | CC BY 4.0 | World Loanword Database: documented target forms with separate immediate and earlier source relations, borrowing score, source certainty and comments. The pinned repair rebuild retains language and sense identity; ambiguous matches do not receive a donor annotation. Dates belong to the target-form record, not every earlier source. Attribution to WOLD is not a definitive historical reconciliation. Haspelmath, Martin & Tadmor, Uri (eds.) 2009. WOLD. Leipzig: Max Planck Institute for Evolutionary Anthropology. |
| ASJP | CC BY 4.0 | Automated Similarity Judgment Program: 40 Swadesh-list concepts surveyed across 11,540 languages. Powers the "attested in N languages worldwide" coverage stat. Wichmann, Søren, Eric W. Holman & Cecil H. Brown (eds.) 2022. The ASJP Database (version 20). |
| CLICS³ | CC BY 4.0 | Database of Cross-Linguistic Colexifications: which concepts share a single word across language families. Powers the "in other languages this word also means..." section. Rzymski, Tresoldi et al. 2020. The Database of Cross-Linguistic Colexifications, reproducible analysis of cross-linguistic polysemies. |
| PHOIBLE | MIT — pinned CSV; CC BY 4.0 — CLDF derivative; CC BY-SA 3.0 — website | Phoneme inventories in the versioned research tools. Diachronica uses the checksum-matched CSV at revision 7030ae0; World Languages Explorer uses the separate CLDF derivative at 5c477f1. Retain the original inventory sources and artifact-specific notices. |
| WALS | CC BY 4.0 | Typological features in /api/research/v1; preserve observation source keys and individual chapter attribution. |
| ISO 639-3 | Open (SIL attribution) | Canonical three-letter language codes |
| COCA | Licensed (paid) | Modern American English word frequency, 1990–2019. Licensed from Mark Davies / english-corpora.org; redistribution restricted, summary statistics surfaced here. |
| Wikimedia Commons | Varies (mostly CC BY-SA) | Illustrative images and pronunciation recordings; each image caption and audio credit links back to the file page with its specific license |
| Google Books Ngrams | CC BY 3.0 | Word frequency per decade, 1500s–2010s (English corpus, version 3), accessed via ngrams.dev. Powers the usage sparkline. |
| kaikki.org (wiktextract) | CC BY-SA (from Wiktionary) | Machine-readable pronunciation data (IPA and audio recording links). Tatu Ylonen: Wiktextract: Wiktionary as Machine-Readable Structured Data, LREC 2022, pp. 1317–1325. |
| Natural Earth | Public domain | Simplified 1:110m coastlines behind the journey map in the mobile app, via world-atlas. |
A note on share-alike
Wiktionary-derived content carries its applicable share-alike terms. PHOIBLE terms depend on the exact CSV, CLDF derivative or website material. A graph or table can combine separately licensed sources: keep their notices, identify modifications and check the terms for the material being reused. This page does not grant one blanket licence over all underlying content.
Corrections · 8 October 2026
- Glottolog 5.3 now cites its own DOI; existing filtered catalogues retain their artifact-specific coverage.
- WOLD immediate and earlier borrowing sources are separate, and target-language/sense ambiguity is withheld.
- PHOIBLE CSV, derivative and website terms are distinguished.
Something missing?
If I surfaced data from a source not named here, or if an attribution needs fixing, let me know. Reach out to luke@lukesteuber.com or open an issue on the code at github.com/lukeslp/diachronica.