A source atlas is a published book or map collection from which a language's area was taken. Our map draws its shapes from digitised atlases rather than from new surveys, so each area reflects what its atlas showed, often many decades ago.
Key takeaways
- Every area on the map comes from a published atlas, and each language page names its atlas.
- The atlas behind most areas is Asher and Moseley 2007, which supplies 3,790 of them.
- In all, 13 atlases are used, and 2,714 languages have no atlas area and appear as dots.
- An atlas shows where a language was traditionally spoken, so its age and purpose shape what you see.
How source atlases work
A language atlas is a published collection of maps showing where languages are spoken, usually compiled by researchers from surveys, censuses, fieldwork or earlier literature. Each atlas answers its own question. One might cover a continent, another a single country, and each was drawn at a particular time with the knowledge then available.
Our map does not survey languages itself. Its shapes are taken from atlases that have been digitised by the Glottography project, which publishes them as open data [1]. The word source atlas is our shorthand for the atlas that a particular area was taken from, and each language page names it.
This has consequences for how to read the map. An atlas draws a boundary as a generalisation: a smooth outline around a region where a language was traditionally spoken. It cannot show every village, every city or every household. Because atlases were made at different times and for different purposes, two neighbouring areas can be drawn with different levels of detail, and where two atlases or two languages cover the same ground the shapes overlap.
Names and identifiers are matched to Glottolog [2], so an area from an atlas is attached to the right language entry even when the atlas used a different spelling. Where no digitised atlas area exists, the map has nothing to draw and shows the language as a dot instead.
An example from our data
One atlas supplies most of the map. Asher and Moseley 2007 is the source of 3,790 areas, and 13 atlases are used in total. You can see which one applies on any language page: Welsh, Basque and Navajo each name Asher and Moseley 2007 as their atlas.
In all, 4,008 languages (59.6%) have an atlas area. The other 2,714 have none and appear as dots. By region, the dots are 648 in Africa, 850 in Eurasia, 628 in Papunesia, 356 in North America, 128 in Australia and 104 in South America. The data sources page lists every atlas and its licence.
The thing people get wrong
Many people read a coloured shape as a current census of a language. It is not. It is a copy of a historic atlas drawing, and a language may have spread, shrunk or moved since. Atlas dates differ, so a shape from one atlas may be much older than the shape beside it. The accurate reading is: this is where a published atlas placed the language, as one source among several, and the area marks a traditional range rather than the home of every speaker.
Source atlas vs language area
| Source atlas | Language area | |
|---|---|---|
| What it is | The published work an area comes from | The shape drawn on the map |
| Number on the map | 13 | 4,008 |
| Where to see it | On each language page and the data sources page | On the map itself |
Frequently asked questions
Which atlas does the map mostly use?
Asher and Moseley 2007, which supplies 3,790 of the areas. In all, 13 atlases are used.
Why do some languages have no area?
Because no open, digitised atlas area was available for them. They appear as dots at their Glottolog position.
Are the areas up to date?
Not necessarily. Many atlases are decades old, so an area shows where a language has traditionally been spoken, not where its speakers live now.
Sources
- Glottography, Zenodo (Glottography community).
- Glottolog, Max Planck Institute for Evolutionary Anthropology.
- Every figure from our own map is worked out from the datasets on the data sources page (checked against Glottolog and Glottography, retrieved 13 October 2026).