On Chinese-language social media, something quietly remarkable has been happening for years. Taiwanese internet users have gotten very good at spotting when someone claiming to be Taiwanese is actually from mainland China. They do not need to see an ID card or check a IP address. They read the language.
It sounds like a party trick. It is actually a real-world application of forensic linguistics, practiced instinctively by millions of people who have never heard that term.
The answer requires a little background. China has long maintained a sophisticated online influence operation. Among its various tactics, documented by researchers and journalists over many years, is the use of paid commenters who post on social media to shape public opinion. In mainland China, these operators are commonly called the "50 Cent Army" (五毛党, wumao dang), a reference to the small payment they reportedly received per post in the early days of the operation.
One of the goals of these operations, as documented in academic research and investigative reporting, has been to influence opinion in Taiwan, particularly around questions of cross-strait relations and Taiwanese identity. To do this effectively, it helps to appear Taiwanese. A post expressing support for unification is more persuasive if it appears to come from a Taiwanese person than if it obviously comes from the mainland.
The problem, from the impersonator's point of view, is that written Chinese is not one uniform thing. Taiwanese Mandarin and mainland Mandarin have diverged in vocabulary, idiom, character usage, and cultural reference in ways that are immediately visible to a careful reader. And Taiwanese internet users, having grown up immersed in their own variety of written Chinese, are often very careful readers indeed.
The most obvious tell is character set. Taiwan uses Traditional Chinese characters, which are the older, more complex forms that have been in use for centuries. Mainland China uses Simplified Chinese characters, which were introduced in the 1950s and 1960s as part of a literacy reform program. The two systems are not interchangeable. Some characters look completely different. Others have been simplified in ways that a Traditional Chinese reader finds immediately jarring.
A person who grew up reading and writing Traditional Chinese develops a feel for how text looks on the page. When Simplified characters appear in a post claiming to be from a Taiwanese person, or when a mixture of both systems appears in ways that no natural writer would produce, it is an immediate signal. A genuine Taiwanese writer who occasionally uses a Simplified character might do so intentionally as a stylistic choice. A mainland impersonator mixing the two systems tends to produce inconsistencies that feel unnatural to a native eye.
Vocabulary differences are equally revealing. Taiwan and mainland China have developed different words for many of the same concepts over the past seven decades of separation. Some examples that come up frequently:
Beyond specific words, there are broader stylistic patterns. Taiwanese written Chinese has absorbed different influences over the decades, including Japanese loanwords and expressions that reflect Taiwan's unique cultural history. Mainland written Chinese, particularly on political topics, carries traces of a very different rhetorical tradition. A post about cross-strait relations written in the cadences of mainland political discourse will feel subtly wrong to a Taiwanese reader even if no single word is technically incorrect.
Language does not exist in isolation from culture. A genuinely Taiwanese writer will make casual references to Taiwanese television, politics, geography, food, and daily life that a mainland impersonator, however well-prepared, often gets slightly wrong. They might reference a political figure by a name or nickname that Taiwanese people do not actually use. They might describe a geographic location using mainland Chinese place name conventions. They might reference a cultural phenomenon that is big in mainland China but not particularly familiar in Taiwan.
None of these individual signals is definitive on its own. A single unusual word choice could have any number of explanations. But forensic linguistic analysis, whether practiced by a professional or by a sharp-eyed internet user, looks at patterns across a body of text. When multiple signals point in the same direction, the cumulative picture becomes compelling.
What is striking about this phenomenon is that ordinary Taiwanese internet users have essentially developed an informal forensic linguistics practice without any formal training. They discuss specific tells on social media. They share examples of posts they have flagged as suspicious. They have developed a shared vocabulary for talking about these linguistic signals.
In doing so, they are applying, intuitively, many of the same principles that forensic linguists use professionally: looking for patterns in word choice, character usage, syntax, cultural reference, and stylistic register that place a writer in a specific linguistic community. The methods are informal. The underlying logic is sound.
It is a reminder that language is identity. And that trying to fake an identity through language, without genuinely inhabiting the linguistic community you are impersonating, is harder than it looks. The words give you away in the end.