Are English-language tests fair? With two grants, NAU linguist studies the need for varieties of English in tests, learning apps

Okim Kang

In the English language, there is not one correct way to pronounce the word “pecan.”

In English-language-learning apps and tests, there is. That means students who pronounce it differently are assessed that it’s wrong.

Okim Kang, a professor of applied linguistics with a long history of research into how accents are perceived, says this idea of “standard” English, which almost always means American or British English, needs to give way to a more inclusive curriculum that includes different accents and ways of speaking. She has received grants from DuoLingo and the International English Language Testing System (IELTS) to study how tests can be altered to better meet the ends of the diverse test-takers.

For Kang, this is social justice work. In the field of second language assessment, scholars are discussing how tests need to be fair to test-takers and institutions ought to be fair to test-takers. Not only does this provide a more level playing field for all students taking the test, it also provides more useful information to society. Prioritizing certain English models, known in the field as prestige models, in the English test while downgrading other varieties, as has been the norm in the field, can have negative impacts on individuals and society.

“The whole issue of linguistic discrimination happens right here in this practice,” she said. “If we are not treating all English forms in an equal manner, we are not treating people equally.”

In a perfect English-learning situation, all learners would experience different accents and varieties of English. This prepares learners better for real-world situations, as standard American and British English are not the entirety of the English language, and it adds fairness into the learning environment for people who may be familiar with a different variety of English that is not represented in language-learnings apps and curricula.

With the $95,000 grant from Duolingo, she aims to promote the idea of incorporating different varieties of accents (instead of using just an American accent) in language learning and testing apps. The $30,000 grant project from IELTS will investigate how language learners or test takers cope with their learning in the pandemic.

This is not exactly a simple fix, though. There are many variables, including creating effective tests and the current reality that students, when asked, want to speak prestige English because of the way it’s perceived. This is even true of English language teachers, who want to teach American or British accents because that’s what their students want to learn.

“This starts with social, economic and political power, I think, and sometimes even cultural power,” she said. “It’s not just a linguistic issue; it’s much more complicated than that. Accent is an insignia of group identification. Prestige accents represent social, economic and political power. Learners of English want to have that.”

Her efforts, alongside those of her colleagues in the field, are working to intentionally educate teachers, learners and test practitioners about the importance of producing intelligible speech instead of certain prestige accents. This also serves to empower learners, whose learning goal should not be to sound “American” but to learn English in a way that does not take away from a learner’s social identity, ethnic group membership or belonging.

The language learning apps and other curricula right now are stuck between the competing needs of fairness and user demand; developers don’t want to build a product that people don’t want, so the twin goals of normalizing different varieties of English and developing programs with those different varieties are happening simultaneously, even as the first is somewhat reliant on the second. Duolingo is trying to move forward with changes, Kang said, which is why the company is funding research like hers.

This research builds on the work Kang has done in her career. She has conducted a number of studies on speech production and how speech is perceived by listeners and the testing process for English language learners. Her overarching research goal is investigating the nature of accented speech of non-native speakers.

Learn more about Kang’s research.

Scholars and researchers are asking these questions about fairness and justice in the intellectual sense, but test-takers aren’t seeing a difference and the same patterns remain. This research will offer evidence that developers and educators can use in their approach to language learning.

“As long as the speaker’s English is clear or intelligible, it doesn’t matter if they have an accent,” Kang said. “I want to provide data that shows accent is not a stigma that you have to get rid of and it’s OK to have an accent. Non-native speakers of English or speakers with accents cannot be marginalized just because they have an accent.”

To participate in the research, email

Northern Arizona University Logo


Heidi Toth | NAU Communications
(928) 523-8737 |

NAU Communications