At least one researcher (Joseph Jordania) has a theory that articulated speech arose in East Asia and back migrated. He hypothesizes that speech disorders like stuttering are more common in European and African populations, the last to acquire articulated speech in his model and thus having less time to winnow out bad alleles.
A rather unlikely theory[1], if only because backwards gene flow was until recently anathema within the mainstream of human evolutionary genetics, but one that is readily testable. Unfortunately the jury is still out on the prevalence of stuttering in, e.g., China. Current studies suggest that stuttering is very rare in China. But the counterargument is that social stigma keeps stutterers hidden and nobody has looked hard enough to discover them. And to his credit Jordania agrees that the existing data, while favorable to his hypothesis, is underwhelming and that true prevalence remains an open question.
[1] I like his approach more generally, though. Jordania's broader theory for the emergence of human cooperative behavior is the only one I've ever encountered that works in a classical Darwinian genetic model. All the others are implicitly or explicitly dependent on group selection, which is facially contradictory to Darwinian genetics and otherwise has yet to be shown in nature. Group selection might still be a thing, but regardless of whether Jordania's model is literally true in the specifics, Jordania successfully shows that you can find a series of eminently plausible steps that remain perfectly consonant with the selfish gene model; you don't need recourse to an unproven and seemingly contradictory extension.