The “18 Manchu Speakers” Myth

I keep seeing Manchu described on forums and in YouTube videos as an “endangered language,” sometimes even as outright extinct. Most of the speaker counts floating around, though, come from seriously outdated data. The familiar claim that “Manchu only has a dozen or so speakers left” traces back largely to Sanjiazi in 2007, when all 18 recorded native speakers were already over 80.

So I did some basic math. Nineteen years have passed between 2007 and 2026. If that figure were somehow still current, even the youngest person in that group would now be over 99. Meng Shujing, one of the speakers documented at the time, was 82. She would be 101 in 2026. Repeating “a dozen or so native speakers” nearly two decades later without ever updating the source is just nonsense. By 2024, research was already stating outright that Manchu had no native speakers left, as noted in the ACL paper ManWav: The First Manchu ASR Model.

Bring Xibe into the picture and things get broader. Linguistically, Xibe is extremely close to Manchu. Veronika Zikmundová goes as far as describing Sibe as probably best treated as a dialect of Spoken Manchu, while also noting mutual intelligibility between Sibe and Sanjiazi Manchu.

Separating Manchu and Xibe follows directly from the way the Chinese state sorts ethnic identities into official categories. Once Manchu and Xibe were recognized as separate ethnic groups, their languages got shoved into separate boxes as well. Linguistically, the distance between them is tiny. Meanwhile, Sinitic languages with far greater differences, often enough to prevent mutual intelligibility, still get lumped together as “Chinese dialects.” There is no consistent linguistic standard behind that language-versus-dialect boundary. It follows the political needs of the state.

Administrative classification warps the linguistic picture from the ground up. Once state-made categories start overriding the actual relationships between languages, political boundaries begin to look natural simply because they have been repeated for long enough. The result is a linguistic map built on inconsistent criteria and a broken conceptual framework.

Manchu and Xibe make the mechanism especially obvious because the ethnic boundaries are drawn first with linguistic status built around them afterward. Over time, those categories begin to pass as objective facts and once they have been repeated for long enough, people stop noticing that the boundaries were constructed in the first place.