Right, so you need another study like this one: https://www.tandfonline.com/doi/full/10.1080/00224545.2019.1687415

Although this one was designed to probe whether it is being black or presumptively being part of American Black culture, it directly addresses the name weirdness question because characteristically African names are considerably weirder to the average American than are characteristically American Black names. And yet greater discrimination was shown against American Black names.

This study isn't nearly as strong as the Berkeley one, so it might be wrong, but it's part of a consistent pattern.

Incidentally, it's worth pointing out that the presumption of undesirable cultural traits which are linked to race is still "racism" (and it would be ethic subgroup bigotry even if not, which is hardly any better). It doesn't really matter what goes in the middle.

The core of racism is: Race/ethnicity -> (infer something here) -> differential treatment

And people have done tests that hit the "racial/cultural identity" button harder, and they generally see larger effects. Another moderate-sized study that shows a (surprisingly large) effect is this one: http://www-2.rotman.utoronto.ca/facbios/file/Whitening%20MS%20R2%20Accepted.pdf

See Figure 2, where whitening both name and experience separately has an effect.

So the (tentative) conclusion should be that, yes, it's an identity-related, not pronunciation/memorability-related bias. We should be open to the idea that a combination of other things might combine in just the right way to give the illusion of racism, or that the studies weren't done carefully enough, or something, but at this point the onus is on the "isn't any discrimination" people to show compelling evidence that there isn't. There are plenty of indications that there is a moderate amount. It's a surmountable barrier in general, given the numbers, but it's not absent.