A language branch, or subfamily, is a smaller group inside a language family whose members share a more recent common ancestor. Germanic is a branch of Indo-European, for example. Branches make a family tree readable.
Key takeaways
- A branch groups languages that share changes the rest of the family did not undergo.
- Indo-European has 8 branches on the map, Sino-Tibetan 17 and Uralic 9.
- The number of branches does not track the number of languages: Austronesian has 5 branches for 1,226 languages.
- Where to draw branches is a classification choice, and experts sometimes disagree.
How branches work
A language family is a group that descends from one ancestor, its proto-language [1][2]. Large families do not split into all their languages at once. They divide in stages, and each stage produces smaller groups. These smaller groups are branches, also called subfamilies or subgroups.
The test for a branch is shared innovation. Members of a subfamily share features that developed after the larger family's ancestor existed [1]. Germanic is the standard example: it is a subfamily of Indo-European, and its languages share changes that other Indo-European languages do not [1]. Shared old features do not count, because they were inherited by everyone.
Branches can contain smaller branches, and so on down to single languages. Glottolog, which our classification follows, treats every one of these units as a "languoid", defined as "a dialect, language or language family" [3]. On this map, a family page lists its languages and the number of top-level branches Glottolog gives it.
Branches are part of a tree, and trees simplify. Experts disagree about the shape of some trees, and languages that stayed in contact can look more closely related than their descent suggests [1]. Where the evidence is thin, a published classification is one reasoned proposal and not the last word.
An example from our data
Our data give each family a count of branches.
- Indo-European: 485 languages in 8 branches.
- Sino-Tibetan: 496 languages in 17 branches.
- Afro-Asiatic: 332 languages in 4 branches.
- Austronesian: 1,226 languages in 5 branches.
Branch counts and language counts do not move together. Austronesian has only 5 branches for 1,226 languages, while Sino-Tibetan has 17 branches for 496. The counts reflect how the classification is drawn at the top of each tree, not how varied the languages are. The map has no data on the internal structure below that level, so you cannot see individual branches on it. The family filter shows a whole family at a time.
The thing people get wrong
People often assume a branch is a geographic region. It is a group by descent. Languages on one branch can be spoken far apart, and neighbouring languages can sit on different branches or in different families. The coloured areas on the map show where a language has traditionally been spoken, while the branch is a separate fact about ancestry.
A second mistake is to read a count of branches as a measure of age or diversity. It is a count of units in one published classification, and a different classification could draw the top of the same tree differently.
Frequently asked questions
What is the difference between a branch and a family?
A branch sits inside a family. The family is the whole group descended from one ancestor, and a branch is a smaller group within it that shares a more recent ancestor.
How are branches decided?
By shared innovations: changes in sound, grammar or vocabulary that appeared in one group after it split from the rest of the family.
Sources
- Language family, Wikipedia.
- Proto-language, Wikipedia.
- Glottolog glossary, Glottolog.
- 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).