YouTube’s Algorithm Is Pushing Kyrgyz Children Toward Russian, Threatening an Indigenous Language

23

Kyrgyzstan spent a century under Russian imperial and Soviet rule, yet the Kyrgyz language survived. Today, it remains widely spoken by adults and is a cornerstone of national identity. However, a new study reveals a different threat to its future: not political oppression, but the recommendation algorithms of YouTube.

Researchers from the University of Michigan, UC Berkeley, and other institutions have found that the platform’s algorithm systematically prioritizes Russian-language content over Kyrgyz-language videos, particularly for children’s interests. This digital shift is inadvertently reversing decades of linguistic resilience, steering a new generation away from their native tongue and toward the language of their former colonizers.

The Algorithmic Bias

The concern first emerged during fieldwork by Ashley McDermott, a doctoral student at the University of Michigan. While researching in Kyrgyzstan, she encountered a recurring complaint from parents and educators: children in rural villages, where Kyrgyz was the dominant household language, were spontaneously picking up Russian phrases and accents.

Adults in these communities identified a single culprit: YouTube.

To test this hypothesis, McDermott and a team of five researchers conducted a simulation of user behavior on the platform. They collected nearly 11,000 unique search results and video recommendations, focusing on topics popular with children, such as cartoons, fairy tales, and mermaids.

The results were stark. When searching for these topics in Kyrgyz, the algorithm rarely returned content in that language. Even more concerning, the team simulated users who had already watched 10 children’s videos featuring Kyrgyz speech—a clear signal of preference for that language. Despite this strong behavioral data, the algorithm recommended fewer Kyrgyz-language videos to these users than it did to simulated “bots” that showed no language preference at all.

“Kyrgyz children are algorithmically constructed as audiences for Russian content,” said Nel Escher, a co-author and postdoctoral scholar at UC Berkeley. “There is no good way to be a Kyrgyz-speaking kid on YouTube.”

A Digital Colonial Legacy

The findings highlight how digital platforms can reinforce historical power dynamics. Under Soviet rule, proficiency in Russian was essential for economic and social success in Kyrgyzstan. Today, while many adults are bilingual and Kyrgyz is mandatory in schools, Russian remains the lingua franca of commerce and, increasingly, digital entertainment.

The researchers found that when searching for terms spelled identically in both languages—such as Harry Potter or Minecraft —the results were predominantly in Russian. Overall, only 2.7 percent of the videos analyzed featured ethnically Kyrgyz people.

This imbalance creates a powerful socializing effect. The researchers argue that YouTube teaches young users to view Russian as the default language of technology and entertainment, while framing Kyrgyz as peripheral or uninteresting. This is reflected in offline behavior: Kyrgyz children are increasingly quoting Russian internet slang, mimicking Russian accents, and adopting Russian syntax in their daily speech.

The Human Cost

For parents trying to preserve their children’s connection to their heritage, the algorithmic bias presents a frustrating obstacle. McDermott recalled speaking with a mother in 2023 who explained that she deliberately paid her internet bill a day late each month. This ensured one day without internet access at home, giving her family a brief respite from YouTube’s influence.

The issue is not a lack of content. The researchers noted that there is ample high-quality Kyrgyz-language material available. In 2024, D Billions, a Kyrgyzstan-based children’s content studio, ranked as the 35th-most viewed YouTube channel globally. Its dedicated Kyrgyz-language channel boasts nearly 1 million subscribers, proving there is a significant audience for native-language content.

Despite this, the algorithm fails to surface this material to users seeking it. YouTube, which has publicly committed to amplifying indigenous voices, did not respond to requests for comment regarding these findings. However, the research team is attempting to engage with YouTube’s parental controls team to discuss the implementation of language filters that could help mitigate this bias.

Conclusion

This study serves as a critical warning about the unintended consequences of algorithmic curation. While technology connects the world, it can also erase local cultures if left unchecked. The survival of the Kyrgyz language now depends not just on schools and families, but on the ability of digital platforms to recognize and promote linguistic diversity rather than reinforcing historical hierarchies.