On July 21, 2020, a pop idol named Yamy (郭颖), captain of the group Rocket Girls 101, posted a recording to Weibo that would put a name to something millions of Chinese office workers had felt but couldn't quite describe. In the clip, her boss, talent agency founder Xu Mingchao, can be heard telling a room full of employees that Yamy was "超级丑" (super ugly) and that her singing was "难听" (unpleasant to listen to), not in private, but in a company-wide meeting, with her present.
The recording went viral within a day. The next day, Xu responded publicly, not by denying it, but by owning it: "我就是那个职场PUA的中年男老板" ("I am that middle-aged boss guilty of workplace PUA"). He still refused to apologize to Yamy directly. The hashtag #职场PUA went on to collect hundreds of millions of views on Weibo, and a survey from recruitment platform Zhaopin later found that more than 63 percent of Chinese white-collar workers said they had experienced something similar themselves.
What makes this case genuinely useful, beyond the drama, is that it gave an entire culture a shared vocabulary for something that had previously been nearly impossible to name. 职场PUA, workplace PUA, borrows its acronym from the American "pickup artist" subculture, systematic techniques originally designed to manipulate someone into romantic interest, retooled by Chinese internet culture to describe how a boss manipulates an employee into compliance, self-doubt, and silence. The term stuck because it did something a policy violation report never could: it gave the pattern a shape, and once you can see the shape, you start recognizing it everywhere.
A Beijing employment lawyer, Zhong Lan'an, commented publicly at the time that workplace PUA could constitute a genuine infringement of an employee's personal dignity rights (人格权) under Chinese civil law, a real legal theory, not just an internet buzzword. But that's precisely what makes this genuinely difficult territory: proving it usually requires the accumulated pattern, not a single violation. One harsh comment is just feedback. The same comment repeated in a specific rhythm, alternating with just enough warmth to keep someone hoping it'll get better, is something else entirely. That rhythm is the actual evidence, and it lives almost entirely in language.
In the aftermath of the Yamy case, Chinese media and workplace commentary converged on a fairly consistent set of recurring tactics. Here's what each one actually sounds like in a real meeting or a real chat message.
"Criticized for your own good." Nothing you do well gets acknowledged as simply competent; it was expected. Anything short of perfect gets a disproportionate response, dressed up as tough love: "how do you not know how to do this?", "does this really need explaining to you?" The framing makes pushing back feel like proof you can't take feedback, which is exactly the point.
"A slap, then a piece of candy." The harshness is followed by an apology, but the apology is only ever for tone, never for substance. "I just have high standards," "I can't let something like this reach the boss," never "I was wrong about you." The employee ends up managing the boss's emotional volatility while still absorbing the original criticism at full strength, and often starts feeling grateful for the "kindness" of the apology itself.
"Dressed-up exploitation." Extra, unpaid, unreasonable workload gets reframed entirely as opportunity: "you're young, take on more," "this will look great on your resume." No actual promotion or raise is ever promised outright, only vaguely gestured at, which means it can never technically be broken as a promise.
"Quitting won't help you." A pointed reminder, often delivered casually, that the boss knows people at every company in the industry, that leaving would follow the employee wherever they go next. Nothing here is a direct threat you could point to. It doesn't need to be.
None of these four patterns require an insult. That's what makes them durable and hard to report. A forensic read of workplace language isn't looking for a single cruel sentence, plenty of ordinary tough bosses say sharp things under pressure without this pattern being present. What it's actually tracking is structure: does criticism ever arrive without a path to actually improve? Does the apology ever address the substance, or only the delivery? Is the promised reward ever concrete, with a date and a number, or does it stay permanently vague? Is a threat ever explicit, or does it stay just deniable enough that repeating it back sounds like an overreaction?
Any one of these in isolation is unremarkable. All four, repeating in a cycle over months, is the actual signature.
A few concrete linguistic devices show up constantly across all four tactics, and naming them makes the pattern easier to spot in real time, not just in hindsight.
Presupposition traps. "Why can't you get this right?" isn't a genuine question. Grammatically, it presupposes the failure as an established fact and only asks for a reason, there's no version of answering it that lets you contest whether the premise itself is true. This is a well-documented rhetorical device, and it's most damaging precisely because it doesn't feel like an accusation, it feels like a question you're simply failing to answer well.
Rhetorical questions functioning as verdicts. "Does this really need explaining to you?" isn't asking for information either. It's a declarative judgment wearing a question mark, and because it's technically phrased as a question, responding to the judgment underneath it, rather than just answering literally, reads as defensive.
Commissives without commitment. A genuine promise ("you'll get a raise in Q2") is a commissive speech act, it obligates the speaker to a future action. "This will pay off down the line" sounds similar but commits to nothing checkable. No date, no number, no way to ever call it broken.
This behavior often doesn't break a specific rule, and job security is exactly why most people say nothing. That's not a personal failing. It's the entire design of the pattern working as intended, deniability is the feature, not a side effect.
The one thing that does help, even without HR involvement or a lawyer: writing it down as it happens, dated, in the person's own words where possible. Not because a single message will ever be damning on its own, but because the pattern, once it's visible in writing over weeks and months, is exactly the kind of evidence a case like Yamy's turned out to need. She had a recording. Most people won't have that. But most people can have a timeline, and a timeline is its own kind of forensic record.