Mitigating Accent Discrimination with AI

Non-native English speakers can be at a disadvantage when communicating their thoughts, concerns and ideas in business settings. Learn how AI can help.

By Ofer Ronen in AI innovations 06/18/24

Clear communication is key to conducting business effectively. However, this has become something of a challenge in an increasingly global marketplace. In a world in which English has emerged as the de facto language of commerce, science, medicine, and many other key areas, non-native English speakers can often be at something of a disadvantage when it comes to communicating their thoughts, concerns and ideas. 

Filipino call center female agent with a headset sitting at her desk at home, viewed from the back. She is in front of a computer monitor.

Accent coaching and training have been the go-to methods of addressing accent nuances in the past. However, the emergence of accent softening AI tools has introduced new efficiencies to the enhancing of speech clarity and, by extension, reduced the potential for accent discrimination. 

What is Accent Discrimination?

Everyone speaks with an accent. However, people whose enunciation does not meet the definition of the Standard American English (SAE) accent often experience systemic linguistic oppression. This can manifest itself in a number of different ways. 

People whose accents do not conform to the norms of SAE are often looked upon as being less credible than those whose accents do conform. They can sometimes be seen as unqualified, or even dishonest as well. Managers have also been known to give substandard performance reviews to employees whose verbal inflections differ from what is considered to be the norm. These concerns are particularly amplified when it comes to positions requiring extensive verbal communication with the public. 

Accent Discrimination and Customer Service

One of the principal tenets of business is the provision of quality service in as cost efficient a manner as possible. Toward this end, customer service centers are often located in countries in which lower infrastructure and overhead expenses can be leveraged to positive effect. However, this also means the talent pools these operations draw from are comprised of people for whom English is a secondary language. 

This can become problematic when the customer base is located in an English-dominant country such as the United States. Remember, among the manifestations of accent discrimination is the perception that non-native speakers are less credible, unqualified and sometimes even dishonest. These preconceptions can give rise to open hostility from callers in some cases. Given the fact that many callers are experiencing a degree of frustration when they phone, it’s important to take this factor into consideration when making the decision to locate a call center offshore.

Mitigating Accent Discrimination With AI

Traditional accent training programs can be costly, time consuming and ultimately ineffective. Moreover, there’s a lot more to being a good customer service representative than the way candidates speak. They also need a strong degree of product knowledge, customer service skills and problem-solving abilities.  

Time spent on accent training is time that could be used to enhance a potential rep’s abilities in those other—arguably more important—areas. What’s more, caller expectations are likely to vary, depending upon the region from which they are calling. AI softening technology, such as Tomato.ai’s solutions, clarifies the voices of offshore agents as they speak, which improves their intelligibility and reduces customer frustration. It’s also capable of adjusting to compensate for regional differences.

Accent Softening and Cultural Respect

Some people have argued mitigating accent discrimination with AI is discriminatory in and of itself. The concern is that it subjugates the cultural norms of one group of people to those of another. To that end, Tomato.ai’s algorithm was specifically built to preserve the original speaker’s voice and elements of their natural way of speaking. In this way, the algorithm respects the identity and cultural background of the speaker, while simultaneously making them easier for Standard American English speakers to follow and comprehend. This makes support calls smoother and less stressful for both the speaker and the listener.

In Summary

Accent discrimination is a real thing. However misguided it might be, there is a persistent belief that credible individuals speak in a manner that is consistent with the mores of SAE. This can be a hindrance to individuals, business and society in general. Mitigating accent discrimination with AI can help. However, there needs to be a shift in this thinking. The world is a much smaller place than it was in the past.

By Ofer Ronen in AI innovations 06/18/24