A hypothetical
language family, or
macrofamily, bringing together many of the established families of
Europe,
Asia, and
Africa. The Nostratic
hypothesis is held by a minority of
linguists and viewed by many more as interesting and
unproven, and possibly incapable of proof.
There are competing variant theories of Nostratic, differing in which language families they include and what sound laws they propose. They all include Indo-European, and the name (coined by Holger Pedersen in 1903) comes from Latin nostrates 'our people'. The first reliable and substantial work on Nostratic was by Vladislav Illich-Svitych in the 1960s, whose list of families to be united was Indo-European, Afro-Asiatic, Uralic, Altaic, Dravidian, and Kartvelian; and by Aaron Dogopolsky, who also included several small language groups of Siberia, such as Inuktitut.
Joseph Greenberg has proposed a group called Eurasiatic covering the north Eurasian Nostratic branches and Japanese, Korean, and a few more north Asian isolates. He regards the other branches of Nostratic as more distantly related. Greenberg has also proposed a genetic connexion with Amerind, a group he (though few other people) believes he has reconstructed to cover almost all languages of the Americas.
There is certainly a lot of vocabulary in common between some candidate branches, enough to give reasonable grounds for them to be related, but the similarities could usually be equally well explained by borrowing. It is easy enough to find ancient words in common between Indo-European and Semitic, or Indo-European and Finnish, but to find material that occurs in Semitic and Finnish and Georgian, material that is more likely to be ancestral than borrowed, is much more difficult. Further, the appropriate genetic comparand is not the Semitic branch but its deeper family Afro-Asiatic, which is mainly African and shares less material with Indo-European.
Ancestral sounds have to be postulated and ancient laws of sound change proposed that could consistently give rise to the observed forms. Several rival theories exist in this area but none is convincing enough.
Similar sporadic resemblances in grammar can be found. The pronouns and personal inflections for Indo-European and Uralic are alike: first person -M, second person -T, third person -S. Some of these are found in other branches too, but the endings being so small, this could well be due to chance.
The normal methods of comparative reconstruction become progressively harder to apply as you go back in time, because entropy has obliterated so many distinctions that once existed that any remaining resemblances could have occurred as random noise. This is generally held to happen after between 5 000 and 10 000 years of divergence, and the period when Nostratic existed, if it existed, was probably earlier than that. This is why many linguists, even if not hostile, do not believe the hypothesis can be supported.
Nostratic is a reasonable hypothesis, extending the existing accepted methods and results of linguistics. It should be distinguished from highly contentious hypotheses, espoused by very few serious linguists, that take this process further back: to attempt to construct a correlation between language groups and either 'race' or the migrations of non-African humans revealed by mitochondrial DNA genealogy. The mtDNA studies do have some relevance and significance for language study, but the situation is complicated.
For example, some linguists believe that many non-Nostratic languages of Eurasia and America can be grouped into a macrofamily called Dene-Caucasian, and this covers all sorts from Basque to Chinese to Navaho. Older attempts to find an ancestral language for the 'white race' are universally rejected.