Following the trend of “overemployment,” machine translation tools are providing companies’ in-house employees working on Chinese-English translations with an excellent means to claim work they never performed. This trend became especially prominent during the pandemic and has only accelerated since; I now see more companies routinely sending out machine translations and asking for an expensive revision of work that is often fundamentally incoherent. In this article, I’ll give an overview of the problem and explain why it’s such a thorny issue for companies.
Around 2020, right when news stories about “overemployment” began hitting the international media, a strange phenomenon began occurring in China. Companies began sending machine translations apparently “painstakingly” translated by their employees, asking for revision and perfection. A lot of these articles were incoherent, mind you, and had not even been checked. For example, “COVID-19” would often be called “new crown,” because that’s what Google Translate put out. Moreover, pretty much no matter what, these companies would keep sending the same stream of broken English translations even if advised that the quality coming in was too low to realistically work with. Managers are typically totally unaware of what’s being done and actually think their translators spent 6 hours out of a day writing these 2,000-word articles. The bosses are totally unfamiliar with translation and, therefore, cannot tell that what they’re looking at is robot output from Google Translate. In the absence of expert supervision, this low-quality work then gets pumped out unabated.
I’ve talked to a number of people around the industry to try to figure out what is going on here and have concluded that the recent trend of “overemployment” is driving the behavior exhibited by these in-house translators.
Overemployment is a Global Trend
Overemployment refers to the act of holding multiple jobs simultaneously and is increasingly common in the modern economy, with the rise of the gig economy and remote work acting as major contributors to the trend.
The Wall Street Journal investigated a remarkable remote work phenomenon. According to their findings, white-collar workers in the tech, banking, and insurance industries seemed to have discovered a way to double their salary by working two full-time jobs from home. Those involved were instructed to keep the situation discreet and to avoid putting in too much effort. The paper even obtained employee contracts, pay stubs, emails, and other documentation to verify the accuracy of these claims. Remarkably, those who held two jobs reported they earned between $200,000 and $600,000 annually, including bonuses and stock.
Away from the scrutiny of their supervisors and co-workers, these multiple job holders “switch between two computers” and timetables. They describe handling it all as almost like having a third or fourth job. Sometimes, you need to sign into two conferences simultaneously or use vacation time from one job to work on the other’s assignments. And you’d better keep up with the interminable Zoom meetings! Monitor your LinkedIn profile and remember who you are messaging to avoid confusing people and sending an important document to the wrong executive. These strategists unashamedly declare that they don’t usually work more than 40 hours per week for both positions. After the financial crisis and large-scale layoffs during the Covid-19 pandemic, coupled with the absence of loyalty from big corporations and watching billionaires gain even more wealth, they feel okay about exploiting the system.
In the past few years, a news topic about “touching fish” (see SCMP report) has also become extremely popular in China as a metaphor for slacking at the job. According to Chinese encyclopedias, the phrase originates from a traditional metaphor meaning “to muddy the waters make it easy to catch fish.” The idea behind this comes from rice paddies in ancient China, where carp would grow and swim; muddying the waters made it easier to catch them in a net. Thus, “touching fish” really means “catching fish” and, in the modern workplace, the fish caught is the company’s money with the means of catching it being the theft of company time. Machine translations thus offer an excellent opportunity for in-house employees to “catch fish” when given translation assignments.
Whether an in-house translator is overworking or underworking is extremely context-specific and hard to pin down. A lot of people these days are overworking on-site by listening to things like Coursera lessons on earphones while pretending to work.
Translation as an Overworking (or Underworking) Opportunity
Above, I’ve established that when workers try to avoid doing work, the reason is either to attempt to achieve overworking, i.e., secretly doing two jobs at once, or underworking, termed in China as “touching fish” — where the employee is just trying to avoid working altogether. The key shared factor is that the employee has identified a good way to make it appear as if they are working while they’re actually doing something else. On my end, I can very quickly tell whenever this is happening because the boss said their person spent six hours doing this “human translation,” when it totally matches Google Translate and, for that matter, does not make any sense. The company managers are not translation experts, so they really have no effective defense against this kind of behavior. When anyone is getting service in a field they don’t understand, that person has to rely on the professionalism of their expert. When you don’t understand the work and ask a non-professional to do it, it creates a lot of room for things like overworking.
Why are these general managers at businesses resorting to having their internal employees do translations? In part, the boss thinks the person just has bad English, so has them send out the translation for editing and thinks that this revision will help the employee learn when in reality it’s just reworking a machine translation. Translation outsourcers—misleadingly calling themselves language service providers—also have the same exact flaw that causes them to send out huge volumes of low-quality, usually robot-produced incoherent translations. They are staffed by general managers who don’t understand translation and, in turn, hire the cheapest freelance translators they can find, not professionals, to do the work.
This can be easily seen by looking at translation company guidelines and standards that imply their translators are earning about the minimum wage in China, despite being uniformly well-educated. The industry rule of thumb used by the general managers for newbie, just-starting translators, is to hit them right away with standards saying “1,000 words in 3 hours.” This, however, isn’t all that realistic for newer translators who still don’t know how to translate any of these things. The speed standard was actually derived in a 1990s academic study as a general standard for experienced translators who know what they’re doing.
In the Chinese-English translation industry, what actually happens with translators is that they start out as interns doing 1000 words in 3 hours. They then speed up a lot from there to earn a fair salary, to the point where translators behind the scenes say they are doing 10,000 — 20,000 words a day. The quality of their work never improves from when they were just fresh-faced interns because few translation companies reward improved quality. They all, however, reward improved speed. Thus, there is an industry phenomenon where highly experienced translators are producing intern-level quality and still claiming they translate 1,000 words in 3 hours for 1 cent a word, which nominally means they earn $3.35 an hour. Translation managers actually believe them, and most of the industry’s work planning revolves around these fictions where 20-year veteran translators are actually overworking to make $15/hour for five different companies at once. Translation clients notice that even 20-year translators are sending intern-quality work, so they simply have their internal staff do the translation instead. Those internal staff quickly discover that they too can easily overwork using machine translation.
The widespread broken Chinese-English translation quality cannot be solved by attempting to simply take the work in-house and have untrained employees have a go at it. The result of that approach is that those employees start doing exactly what the freelance translators are guilty of: turning over the entire translation process to artificial intelligence while claiming to be working very hard. For an unprofessional translator, the temptation is too great—the only viable solution is to seek a genuine professional to do the work.
