English is the world’s most widely used language not only for international communication but also as a common technical language. It is widely used for all kinds of instructions, manuals, diagrams and other reference material for equipment, designs, software and so on. Statistically more than 50% of all instruction manuals and technical publications are originally written in English. When adapting these instructions in other countries, in our case in the Russian Federation, it is English that most often has to be translated into Russian. Moreover, almost 90% of all information found on the Internet is written in English.
The experts at ATT technical translation agency have analyzed the situation for the period of 2019-2023. Over the entire period, there has been a natural shift towards English as the highest priority. According to our estimates, the volume of orders for technical translation from English was about 92%, three out of the remaining eight percent were in German, and only 5% in other languages (in particular Italian). The most problematic languages are, as always, the English versions of the original texts translated from, for example, Japanese, Chinese (most frequently) or Korean.
There are currently a huge number of companies on the Russian market of translation agencies, the majority of which are mainly concentrated in Moscow, St. Petersburg and the Moscow region. In their advertising campaigns, many agencies claim to be able to translate documentation from nearly all known languages of the world and at the speed of technical translation ranging from 30 to 50 pages a day, and technical translation from English even faster.
Let’s talk about technical translation timelines, both theoretical and real. If we present a two-dimensional graph (X-Y) with “technical translation time” and its “quality” along the axis (at a certain average fixed volume), we can easily identify an optimum (extremum) on this curve, which allows us to estimate a reasonable turnaround time required for high-quality translation of technical documentation. Without going into details, we can demonstrate that the average speed of translation per translator is 6 to 8 pages per day. This is an effective amount that a single translator’s brain can handle. Plus a couple of days for scientific editing (for a typical order, about 50-80 pages). Thus, the average number is about 6-7 pages per day of finished, edited translation for a non-urgent order.
Sometimes, when faced with a large volume of material, they need to split the document into blocks and translate it by several translators, and then stitch it into a single file (by one scientific editor). In this case, the time spent on stitching and editing the material will directly depend (upwards) on the number of translators involved in the project.
The best solution for small and mid-sized projects is “one translator + one scientific editor”. For larger projects (for example, translating a single monolithic document that can be divided up into chapters), it is possible to use two to three translators and one editor, but not preferably more. As a rule, using more than four people for a single document is not optimal in terms of technical translation quality. Experience has shown that a patchwork quilt can be difficult (time-consuming) to stitch together after several translators, even with the much-publicised translation memory software. This is because the suggested bundles and sentence fragments will in any case need further checking, not to mention any time consuming substitutions, insertions etc. The only exception is with different documents from the same project, which can easily be “scattered” among translators.
As I mentioned above, if there is a large volume of work, a two-three translator/editor approach will produce 12-16 pages of high quality translation and editing per day. We have every reason to believe that such rates are optimal for today’s technical translation projects and it makes sense to expect a reasonable turnaround time of 30 days (and not 5-8 days as some agencies manage to claim) for a 400-page workload.