Honesty is one of the major pillars of professionalism but has yet to become commonplace in the field of legal translation. I have analyzed several reasons for the phenomenon, and overall can find three major contributing factors. First, universities are not fulfilling their mission to adequately prepare students, rather operate as diploma mills instead. Second, the low supply chain visibility strategy used by Language Service Providers (LSPs) deliberately provides plausible deniability and cover for translators to do shoddy, dishonest work for clients. Third, clients tend to overestimate their ability to identify good quality translations and often make excessive demands for cost-cutting and fast turnarounds.
Student Dishonesty
In my own experience teaching college-level translators, the vast majority of students cheat. This particular problem is so severe that I’ve come across pitches from translation software vendors offering educational software for students designed to screen out their cheating issues. The basic modus operandi since 2017 seems to have been to use neural machine translation or generative AI to complete homework assignments and even tests. Tools like DeepL, and even Google NMT, will generally outperform student translators because they are trained on data from more experienced translators. Even if those translation tools and the translators that train them are usually wrong, they are at least slightly better than the student translators. Thus, students who don’t use machine translation will likely be at a disadvantage compared to those who do use AI. Professors’ workload with student dishonesty in academic papers wouldn’t skyrocket when students turn to machine translation, but it would for translation since the number of mistakes increases dramatically with machine translation, basically overwhelming the entire system.
The current financing models used by educational institutions generally give students a lot of power to enforce dishonesty. While tuition payments and credentialism are not unique to translator training, they certainly play an outsized role. Credentialism is the tendency and behavior of employers and clients to value credentials over actual worker ability as the determining factor in hiring decisions. However, the true value of a credential is largely determined by the institution’s reputation for excellence in the educational program, rather than what any individual invisibly does within the institution. Combined with large amounts of tuition money, this creates a tragedy of the commons effect where students are empowered to minimize the effort required to get their degree credentials, thus watering down entire programs. Translation is particularly affected because it’s very much hands-on and requires a lot of diligence and hard work to produce something that’s high quality, whereas most students prefer to pressure the system to give them lots of praise for shoddy, dishonest work, rather than shaping them into competent professionals.
Finally, in a smaller number of countries, tuition fees are relatively low and professor compensation is also low. These professors are not sacrificing anything for the greatness of the country and society. Rather, they are using alternative funding schemes. A particularly popular alternative funding scheme is to set quotas for translation internships that require a certain number of words translated per student, such as 100,000. Professors are then hired by outside parties to complete translation work, often government or state-owned, which is subsequently handed off to students who complete the work for free, while the professor is paid for their work. This translation work might be worth 5 cents per word and would imply asking a professor to work 16 hours a day every day to meet a deadline, earning them $15/hour. The reality, therefore, is that each student’s free labor is transformed into a new revenue source. In turn, there is very little incentive for the students to do more than extremely shoddy work, as there is likely to be little or no oversight or correction.
Supply Chain Visibility
Most translators released into the translation ecosystem are essentially trained cheaters. How does the industry react? By covering it up. A typical method used to facilitate cheating in today’s translation industry is supply chain visibility. If you look at what translation vendors are saying about translation, and the discourse of professional project managers, you will see a lot of discussion about the translation supply chain. With that territory comes the concept of visibility, which is the idea that since the supplier is an independent organization, you are unable to see what they are doing during their work process. This is not an unreasonable way for an LSP to arrange its relationships with translators, as indeed a single translator may work with dozens of different LSPs when performing work and is effectively an independent contractor while on the job. This provides the LSPs with a great deal of plausible deniability when its translators act dishonestly but nonetheless makes it easy to mislead clients.
Looking at the marketing materials and what LSPs tell clients, it would be easy for a typical client to come to the conclusion that the translators ‘work for’ their choice of LSP, are somehow under their control, and that assurance is very high. The reality is largely the opposite, and limitations on supply chain visibility give extensive cover for translators to be dishonest. The project management staff generally lack any expertise in translation in itself, much less a particular language, and do not employ the services of experts. Moreover, an LSP will almost never employ a genuine expert or even someone with minimum competency certification, such as ATA or CIOL, to audit or validate their translators, as this could dispel the plausible deniability about dishonesty. Instead, what translation companies do is send ultimatums to their translators to promise and deliver perfect work—or else! The ultimatum is never seriously enforced, however, and it serves mainly to give the impression that an ISO-compliant quality control policy is in place.
Client behavior in general provides LSPs with a lot of encouragement to facilitate or cover up translator dishonesty. As some industry think tanks have pointed out, the two things clients frequently pressure LSPs about are rapid turnaround times and low costs. Consider the negotiation tactics of a Shenzhen-based electronics maker who is no doubt aware of the Chinglish problem: when a vendor suggested translators spend more time to optimize quantity, the maker’s executives replied that a competitor translation company provided a very rapid turnaround time with no qualms. The question of assurance as to whether the process is viable never crossed their minds. Under these circumstances, it would be very easy for someone to use AI to fully automate translations from Chinese to English and allow poor-quality translations to pass unnoticed by the company’s own translation team, who themselves are likely unable to actually recognize what a high-quality translation looks like.
Another plausible way for a translation vendor to meet very tight deadlines in such a context is to split a large document into many tiny files to assign to individual translators. For example, a 20,000-word document might be split into 20 tiny individual files. Clients and project managers doing this mistakenly view translation as a specifically linear task, like driving a car down a road. In actuality, translation is not a linear task. The 20,000-word document is likely to have 100 terminology items that make up the bulk of the material, which the translators will need to research and verify to avoid reliance on highly inaccurate online dictionaries. By splitting the task up into tiny fragments to meet a specific deadline goal, the LSP basically forces translators to achieve results akin to machine translation, even if no actual machine translation is used to complete the work. Consequently, individual translators effectively wind up in a race to the bottom, working under conditions so impossible that they cannot perform above even the most minimal level. Moreover, they are also financially incentivized to do just this—client translation rates will often be set at the speed minimally necessary to cover machine translation errors clients themselves are likely to notice.
What Can Be Done?
While translators have relatively little power over what each individual client does, and clients often make poor decisions, it’s fully possible to collaborate with cooperative clients who respect expertise. Moreover, those clients are more likely to succeed than their competitors. For example, I previously worked with a Shanghai-based law firm advising its Fortune 500 clients about China’s latest developments in information security law, which included things like the anti-espionage law. I pointed out that the typical approach that an associate attorney with a high school English background would want to take to translate the China law provisions into English would actually encourage a scheme of illegal information theft inside China. Furthermore, the data would be stolen in such a manner that it could easily be dropped into the hands of foreign governments. None of the clients I have worked with on translating compliance requirements were ever raided or otherwise sanctioned, whether they were Chinese companies under USA law or American companies under China law. Nonetheless, famous American companies like Bain and Capvision were raided by the Chinese authorities. Later, China stated that its espionage prevention law had been misunderstood, which is also a point consistent with my university lectures on the subject.
Despite the numerous headlines criticizing the move, I honestly believe that bad legal advice and white-collar dishonesty by these companies, perhaps due to pinching pennies on compliance, led to the issue. In my own experience, the people who translate official PRC laws insist that the free versions are not to be relied on, noting that their focus is on things like negotiating with the FATF or IMF about coordinating domestic legislation with international trade. Yet, despite this advice, most law firms’ attorneys in China will simply copy and paste the free version and send it to their clients. The reason is not particularly nefarious, rather most of them have not heard of Skopos theory and instead rely on a theory called Xin Da Ya (see related article here). However, having been cheated by LSPs so many times, they decide the translation can only be done in-house. Often, they write legal advice in English only for the benefit of foreign executives, which is then implemented by these same executives in an English-only environment in a top-down manner, resulting in much of the law being simply lost in translation.
From the perspective of the law firm an honest translator is working with, their own clients are well-protected, and the law firm can point to how other law firms’ clients are being raided. As a result, business opportunities in the country will tend to flow toward those firms willing to embrace expertise-driven translation practices. In other words, instead of simply just giving up and scamming clients “like everyone else,” translators can help the “right” clients gain a competitive advantage over their dysfunctional rivals.
