no warranty · an illustration · not a teaching resource
This is not an educational resource and not something to cite. It is an illustration — a way to look at the shape of the thing and find connections in it. It claims nothing more.
No warranty. The data is simplified and assembled from widely available surveys. Errors are likely and they are mine.
Dates are estimates. For proto-languages the spread is centuries to millennia and different schools differ by thousands of years. For attested languages the start is usually the first surviving text, not the real beginning — a language is always older than its oldest record.
The classification is one of several possible ones. Where the field does not agree — the Nilo-Saharan and Khoisan languages, the Papuan and Australian areas, the Altaic, Nostratic and Dene-Caucasian hypotheses — it is marked here as an areal grouping or a hypothesis. But disputes run inside families drawn here as certain too.
The map is not a map. Areas show schematic reach, and for older slices they are a rough guess. Languages also overlap and mix in ways no flat shape can capture.
Speaker numbers are order-of-magnitude and sources differ on them by tens of per cent. Read them as a size class, not a number.
Territories are built from modern borders where a language has a national or regional footprint; elsewhere they are approximate shapes. A country in the colour of a language does not mean everyone there speaks it.
Wikipedia links point to hand-picked articles. I do not vouch for their content and they may disagree with the dates given here.
For anything that matters — schoolwork, teaching, publication — use scholarly literature. This page is a signpost and a picture, not a source.
Luko Karavan · lukokaravan.cz · tree of language families, 2026 edition
