In Japanese-to-English translation, the need to cut costs and the long-held folk beliefs by some native Japanese speakers have often taken precedence over the value of front-loading quality into the translation process.
Back in the late 1970s, most Japanese-to-English translation was done by native Japanese speakers (NJSs), with varying degrees of ability to translate and to write in usable English. There is strong evidence that the demographic makeup of at least the humans still translating remains largely the same, particularly when you consider in-house translators in Japanese entities.
Those translations were often then subjected to what was then (and even now, apparently) called a “native check,” which very often, particularly back then, was characterized as a “brush up” performed by a native English speaker (NES) who would sometimes not be Japanese-capable, making referring to the source text impossible and meaningless. The salient characteristic of this workflow was that frontloading of quality was not part of the translation process. Quick and cheap, followed by repair was the norm.
Numerous hapless foreigners were employed in such work in Japan, some were English teachers, others were recruited from among in-house foreigners working in Japanese companies, but who were not necessarily hired as translators or editors. They just had the misfortune to be standing around when a translation needed a “native check.”
As they became more numerous, NES translators started receiving more JA-EN translation work. There was still, however, a lingering uneasiness among many Japanese clients that a non-Japanese person couldn’t actually understand all the nuances of something written in Japanese. And could they really read the characters? I have been asked as much at translation conferences, even after it was clear that I am a JA-EN translator. One such encounter happened when I attended a translation gathering of the Japan Translation Federation as the head of Japan Association of Translators.
Perhaps more importantly as a business reality, however, NES translators were considerably more expensive than their NJS counterparts, so the process of making a rough translation and having it repaired continued for years. Again, front-loading of quality took a back seat to other concerns.
Decades went by and the number of NES translators who could convince clients of their capabilities gradually grew.
Enter machine translation, years before the AI hype. This gradually turned many translation opportunities for freelancers into opportunities to repair not poor translations done by an NJS human, but rather the output of an MT system, resulting in their earnings tanking even further, as translation consumers demanded cost cutting.
What is happening is that professional humans are being replaced in the translation process by collections of software commands with insufficiencies that are supposed to be fixed by professionals humans, but at a fraction of the earnings that were possible when those humans were still being trusted to translate. Post-editing repair crews come to the rescue. Many translators are not eager to take on such work, but numerous translators have few other options, particularly ones in mid-career or later. And that is where freelance translation sits as I write this.