It’s been a while since I posted here, I’m sorry, but I should really finish what was meant to be a trilogy of posts before I start on new material.
“In for a penny” suggested that per-word rates were stagnant because of increases in efficiency among translators, not because of the ‘usual suspects’ who are often blamed. “That we might stroll where unicorns graze” pointed out the trap of those who sell the ‘change management’ snake oil to those who patently don’t need it, just to line their own pockets.
Both my earlier posts were deliberately optimistic, and dismissive of the doomsayers. I am still dismissive of the doomsayers, but I’m aware that this post might not be received so well. I’ll even give you the bad news first, to let you digest it: per-word translation rates at the bulk end of the market are going to continue falling, and this will drag down per-word rates in the semi-specialised market as well.
How can I be so sure of that? My argument is that rates have been dropping in real terms because of a rise in productivity. Computer-assisted translation (CAT) tools have enabled translators, especially experienced translators, to translate more words per hour. When translators are paid by the word, that means more pounds, dollars or euros per hour of work. In the competitive market that freelance translators work in, the real per-word rate has tended to drop to keep the real per-hour rate stable.
But CAT tools are now in almost universal use among commercial translators, surely rates should now be rising again, in line with normal inflation… but they’re not. Because there is another technological advance that is making its way into Translatorland. And it’s not machine translation as the doomsayers claim, and have been claiming for eighty years now (see “I need technology; technology threatens to replace me”).
CAT tools needed an increase in affordable computing power to become established as part of the commercial translator’s toolkit. Speech recognition needs even more computing power, but that power is now within the reach of most professionals. Of course successful translators have long used dictation to increase their turnover, but now they no longer have to find (and pay) an audio-typist: they can choose from one of the several or many (depending on your target language) speech recognition solutions that are available.
Estimates of the gain in productivity from using speech recognition (dictation) range from two-fold to four-fold. That means that translators using speech recognition are currently earning two- to four-times more than their colleagues who insist on typing. To me, such a difference does not seem sustainable in the long term. The logical conclusion is that rates in the bulk market, where it doesn’t particularly matter which individual does the translation, will be cut by at least a half.
There will be a certain irony if my prediction comes true. If the rates in the bulk market fall to such an extent because of the productivity gains of speech recognition, human translation will become cheaper than the prices currently quoted for commercial machine translation with post-editing (MT-PE). Post-editing is not a task that is particularly suited to speech recognition, at least not at the moment. The computers will have beaten the computers (at least until the next battle), and human translators will still be in place. We live in interesting times.