This article is really a mishmash: "Molly Babel, a linguist at the University of British Columbia in Vancouver, Canada, replicated Rubin's study this year. More than 20 years later, Babel uncovered exactly the same bias that Rubin had. Students actually understood Chinese Canadian and Caucasian Canadian speakers equally as well when they didn't see either of their faces, Babel says. "It's only when listeners saw a picture of the Chinese Canadian speaker that their ability to accurately understand them went down."
In other words, Rubin was right in his hunch. It was the students' own biases, and not the teachers' bad communication, that led to poorer ratings of foreign professors. The students were simply not listening to their teachers."
This really undermines the discussion of the Lev-Ari study that it precedes. It's no surprise to anyone familiar with the McGurk Effect that we're predicting what people are going to say, and this can influence comprehension of different speakers. But the Babel study emphasizes that it's the stereotype that creates that perception, not merely a preference for acoustic similarity to one's own lect.
The discussion of the 2013 study on British/working-class New York accents is also a mess. The author somehow reads it as a preference against foreign accents, when in fact the subject used inference more with the New York accents. Which sounds more like a preference for high-status accents (British) over low-status (working class New York) accents.