Introducing Human Translation
Let's get talking about what I really came to Substack to write about
I’ve been on Substack now for a while, and have only really posted about the human side of me (personal interests, life in Vienna, the odd cathartic piece, the reflections of my youthful folly…) But when I set up my account here as “Human Translator” my intention was to use it as an outlet for some of the pieces I have blogged about as an in-house translator. Both the translation profession and the translation industry are in a tricky spot at the moment, in this human-machine translation era.
The traditional “triple constraint” of speed, cost, and quality has been up-ended
For the professionals, the issue has been one of rate collapse due to commoditisation of translation and then the attack from GenAI/LLM-based “translation”, which offers frequently deceptively fluent output for a fraction of the cost. Some people argue that it is free, but that is a fallacy. If the translation is free, the chances are that you are paying by training the model in some shape or form. In addition, the traditional “triple constraint” (sometimes referred to as Burns’ Iron Triangle) of speed, cost, and quality has been up-ended, especially due to the hit taken by quality, as “good enough” seems to be “all” many people need, rather than premium quality.
Meanwhile in the industry, GenAI/LLMs and Neural Machine Translation have all had a massive effect on agencies - who can no longer charge those premium human translation rates - and in many cases are going “all in on AI”. In turn they have been squeezing the cents out of the rates they pay their freelancers in order to maintain margins. And there has been considerable consolidation, with larger agencies squeezing out many of the smaller boutique agencies. Freelancers are having to be proficient in ever more tools, which are frequently counterintuitive compared to the tools they feel most proficient in.
And then there are the manufacturers of CAT tools. In the age of TechBros in the GenAI era, who consider translation to have been solved, even the largest and established manufacturers are struggling to get a place at the top table to push the further development of their tools, replete with AI features.
Back to the profession, which is to a large extent self-employed or freelance, and the lack of job/career security is causing professionals to reconsider their professional futures. Some are trying to make it over the line to retirement, others are considering broadening their services offered, whether by offering additional language combinations, value-added services, or trying valiantly to futureproof their businesses. For some, the “out and out” translation has dried up to a slow trickle, and work is largely as the “human in the loop” reduced to post-editing of machine translation.
In house, the profile of new hires has changed: the classic translator profile seems to be a dying one. Jobs are no longer widely advertised for a full-time translator. New hires are frequently for fractional positions (i.e. either part-time, or as a percentage of a full-time position). Not only are retiring translators no longer being replaced on a like-for-like basis, but there are also fewer junior translators coming through as fresh graduates from translation degrees. And with seemingly diminishing prospects for well-remunerated jobs in translation, there are declining numbers of students in degree courses. Of course, it is not solely down to the career prospects - in some countries their own primary and secondary education policies have done far more to decimate the numbers of students of Modern Languages in Higher Education, as has been demonstrated in recent years in the United Kingdom.


Interesting times indeed.