Impact of NMT on Productivity: Paying Translators an Hourly Wage
May 29th 2018 07:40 - 08:00
When a translator uses his own linguistic resources, be it his memory, his dictionaries, his glossaries, his TMs, it makes sense to measure his work by the number of words translated, and accordingly pay him a word rate. When most of the resources are supplied with the job, such as client glossaries, LSP TMs, in-domain NMT and platform-generated quality check, there is intense pressure to reduce the word rate. And by how much? Difficult to answer when the productivity boost provided by the neural translation engine cannot be evaluated ex ante. Confronted with this problem, Lingua et Machina has implemented a pay-per-hour scheme on its online platform Libellex. Time spent on translation and revision is tracked automatically.
We will go over the pros and cons of this approach, with a solid conclusion that the pros win.