Absolute silliness.
Feature on BBC, more detail in New Superstitionist.

Headline: New theory about linguistic evolution places origin in southern Africa.

Location isn't the problem. Might be Africa, might not.

The problem (as usual in modern "science") is a really dumb theory.

Theory:

Africa turned out to have the greatest phonemic diversity – it is the only place in the world where languages incorporate clicks of the tongue into their vocabularies, for instance – while South America and Oceania have the smallest. Remarkably, this echoes genetic analyses showing that African populations have higher genetic diversity than European, Asian and American populations.

This is generally attributed to the "serial founder" effect: it's thought that humans first lived in a large and genetically diverse population in Africa, from which smaller groups broke off and migrated to what is now Europe. Because each break-off group carried only a subset of the genetic diversity of its parent group, this migration was, in effect, written in the migrants' genes.

Makes sense in genetics, doesn't make sense in linguistics. New phonemes and new grammatical forms arise quite often.

Vowel phonemes are most obvious. The alleged "mother of all languages" has plenty of consonants including those famous clicks, but only has aeiou for vowels. Many modern languages (including most Indo-European and most Ural-Altaic) have full sets of umlauted vowels which are phonemically distinct from the unrounded. Slavic tongues have a full set of palatalized vowels, also not present in the "mother". And you don't need to compare all the way back to Africa; some of the umlauts developed in historical times.

In grammar, old forms are constantly fading and new forms are developing. The original Indo-European verb, preserved nearly intact in Hindi, has an 'orthogonal' 5-dimensional grid of person, number, tense, voice and mood, done entirely with suffixes. Most of the descendants have dropped most of these suffixes, replacing them with a variety of separate words that allow more complex tenses. (As I was typing this, the BBC announcer said "is going to have to adapt", which is a damn good example! I suppose you'd call this a Future Continuous Necessitative form, and there's no way to form it in pure Latin.)

= = = = =

So the whole theory of steadily decreasing diversity may work for critters but it doesn't work for languages. Humans have a powerful drive to say new things, and a drive to develop more refined ways to say old things. Independent origin of language in several places seems equally likely, until better evidence emerges one way or the other.
  ¶ 4:38 AM