Historical Linguistics/Subgrouping and Linguistic Classification

As a language changes, given time and distance or isolation, its speakers will no longer be able to understand each other, creating separate languages. For instance, you are probably familiar with the fact that Spanish, French, Italian, and Romanian are all descended from Latin, which is descended from Proto-Indo-European. As such, we call the Romance languages, however, these languages are not special in that regard; we can classify most languages into groups, subgroups, and supergroups based on the various changes they share.

We can often reasonably construct language groups among related languages purely by comparing data from the languages, without knowing any of the historical background behind subgroupings. This is done by comparing shared innovations from whatever we reconstruct or have attested in the protolanguage. The more shared innovations and the less likely they would be to occur independently, the stronger the evidence for a subgrouping. It should be noted that shared conservation of older forms are not evidence for a group: it is expected that languages will have features from their parent languages.

Remember that sound changes are often crucially ordered. This restriction often makes some proposed shared innovations incompatible with one another.

Language Families

All of known natural languages can be grouped into approximately 420 different families, including isolates and dead languages.

The Indo-European language family, including all those languages descended from Proto-Indo-European, is the largest language group in the world by number of speakers (by number of languages, the Niger-Congo family, including languages found predominantly in sub-Saharan Africa, is the largest). The common descent of Indo-European languages has been known longer than any other major language family, and consequently the historical research in this family is extremely robust.

Language Isolates

Some languages, known as isolates. have no known related languages. Well-known examples might include include Basque, Ainu, and Burushaski. It is unlikely that there have never been divergences in their language families, but instead it is probable that all other members of the families simply no longer exist. For example, the Siberia language Ket is known to belong to the Yeniseian family, which includes several known dead-languages, but no other living languages. There are also known cases of languages once thought to be isolates for which related languages have been discovered.

Some languages are not well-known enough to be classified, including extinct languages with few attestations, and modern languages which have been scarcely studied. These languages are considered distinct from isolates, as if more information were available, they might potentially be classifiable as members of an existing language group. An example would include Pictish, the language spoken in today's Scotland in the Early Middle Ages.