What is particularly interesting about the distribution is that the inheritance of languages often is quite different from the genetic ancestry. Of course, we can see modern examples: Jamaicans speak English, but have only a little Anglo-Saxon in their ancestry. In Ethiopia, where there are South Semitic languages and the "Cush" group (the forum won't let me write "Cush" with the "-itic" suffix!), there is a minority genetic type with as dark a skin-tone as other Africans but a different facial structure, lacking the broadened nose and lips of the "Negroid" face; this used to be called the "Hamitic" type but now the anthropologists scramble for a different word for it because "Hamite" is thought of as derogatory ("Negroid" isn't considered acceptable by everyone either).
So in that case one can see that the languages were brought in to the African population by immigrants from further north mixing in with the pre-existing population: but in the Chaddic case further west, the Hausa and other tribes speaking these languages are thoroughly Negroid, indistinguishable in visible traits as well as subtler genetic markers from the tribes around them speaking African languages of the Niger-Congo superfamily. There too we have to suppose that an immigrant group brought the language and imposed it on everybody, but however successful they were as a conquering aristocracy they were so outnumbered that they have left little genetic trace.
So what's puzzling is that there is not much correlation between the size of an incoming (often conquering) group, as reflected in the percentage of genetic contribution they make to the descendant population, and the linguistic outcome. The old native language may be completely displaced: as among the Hausa; or in cases like Egypt and Morocco, where the actual number of Arabians moving in during the Muslim Conquest was trivial compared to the populations (Egypt has always, of course, supported a lot more people than Arabia; the Arabs did not slaughter everybody and breed like rabbits to refill the country!) but Arabic has taken over. Or the old and new languages may mix to some extent: in Norman England, the resulting English language was mostly a continuation of Anglo-Saxon, but with an import of a much bigger percentage of French vocabulary than the percentage of actual French immigrants, and considerable reshaping of the grammar; while in Roman Gaul, the resulting Old French was almost entirely a vulgar Latin, little of the old Gaulish vocabulary surviving but with the Latin words deformed by Celtic pronunciation habits like dropping consonants.
Or, the newcomers can adopt the native language and leave no trace of their old language: in post-Roman Gaul, the German-speaking Franks dominated totally, but there is no Frankish at all in French. And this is what seems to have happened with the "Abrahamic" immigrants to the Middle East. We do find shared between Jews and Arabs (Arabs of the peninsula, Palestine, Jordan, and Syria that is: those with lots of Arabian ancestry, not Arabic-speakers from further afield) genetic markers of a population not shared with other Middle Easterners, related to the Caucasus mountains. That is, there is scientific substance to the tradition that descendants of the survivors of a big flood around Mt. Ararat moved down south. But, the languages in the Caucasus are very strange, and what we now of the ancient languages from around there indicate that the Caucasian groups have long been quite alien; yet there is no linguistic trace of this whatsoever in the Central Semitic speeches (the Canaanite/Arabic/Aramaic grouping).