Select a language and its wider kin appear as dimmer rings, one ring per step
further up its lineage. A chord states classification and nothing else — shared
descent as catalogued, not migration or contact — and each coordinate is a single
reference point standing for a whole language, not the ground it covers.
Data provenance
Classification, names, identifiers, coordinates and macroareas are built directly from Glottolog 5.3 — the languoid export (doi:10.5281/zenodo.15525265, accessed 2026-03-02, CC-BY-4.0). All 8,618 language-level languoids are included; family, branch and lineage are the real parent chain, the 8,304 coordinates are Glottolog's published representative points, and 8,526 macroareas come from the matching CLDF release. Cite as Hammarström, Forkel, Haspelmath & Bank 2026, Glottolog 5.3, Leipzig: MPI-EVA.
Speaker figures come from Wikidata. Glottolog publishes no speaker counts, so each figure is the best-ranked number of speakers (P1098) on the Wikidata item for that glottocode, carrying its point-in-time qualifier where one exists — so every number is individually attributable and datable, and the query that produced them ships with the page. Wikidata covers a minority of languages: 1,795 of 8,618 carry a figure. The rest have none published, and are shown without one rather than as zero — they keep their location and every relationship regardless.
Glottolog's own eight non-genealogical buckets — bookkeeping, unattested, unclassifiable, pidgins, mixed and artificial languages, speech registers and sign languages — are pooled as Unclassified & placeholders and stay off the globe unless selected, since their members are not relatives.
Coastlines are Natural Earth 1:110m land.