Quentin Atkinson of the University of Auckland, New Zealand, has done a computer analysis of certain sounds (phonemes) in about 500 languages, and concludes that all of them likely descended from African tongues. African languages, like African human genetic endowments, are the richest in diversity. A greater diversity of phonemes in African languages and less diversity in diaspora languages (with the least being the Polynesian languages) would make sense if humans brought language out with them when they exited Africa 50,000 to 70,000 years ago.
The New Scientist notes that many historical linguists reject Atkinson’s conclusions. Those who use ‘glotto-chronology’ or the study of how fast languages change, believe that there is a 10,000-year barrier to studying language relatedness. Let us say you had a band of people 10,000 years ago in Central Asia. Let us say that they split, and some went to India, and others went to Europe. Their languages would then evolve independently. After 10,000 years they would have changed so much that the similarities between them would be reduced to the same level as random chance. So you can’t prove them related any more, because while there might be some related words or cognates, they cannot be scientifically demonstrated to be such.
Atkinson took a detour around this roadblock by a quantitative approach, in which he used diversity of phonemes as a proxy for ancientness. Me, I think it makes perfect sense that most languages originally evolved in what is now Africa, along with homo sapiens sapiens itself.