Why do accents trump skin color?

Scientific American has published a piece on work by Katherine Kinzler and colleagues showing evidence that kids, when presented with pictures/voices of other kids, identify the ones they would like to be friends with more by accent / language (namely, speaking English 'with a native accent') than by 'race'. That's not very surprising, I'd say.

A cliche has developed around here about articles in places like Nature, Science and PNAS — a view forged by people who haven't published in any of those places, I think — that the deal is that you have to do serious (or serious-looking) science up to the very end of the article and then you can go wild with speculation about broader ramifications that are highly improbable. This is being called the "Last Five Lines" principle.

So too, it seems, here. Here's the last paragraph:

Why was accent more important than race? “Race, as a psychological category, may be relatively modern in terms of human evolution,” explains Kinzler, now at the University of Chicago. In prehistoric times, “a neighboring group might have sounded different even if they did not look different,” she says. Preference for our own race might have developed later, after the more ancient preference for our own accent. The next step is to see whether living in bilingual or multilingual countries might change this early inclination.

I would have figured that kids just generally prefer the familiar to the unfamiliar, and that socialization of even relatively young kids has clued them in about who to hang out with. If so, going for deep evolutionary explanations is a massive stretch. I guess this is less surprising from Scientific American than the big general science outlets.

The most interesting followup might be not with 'foreign' accents, but rather with regional ones. Would you get the same results with Wisconsin, Alabama, Massachusetts, and so on?