Getting stuff off my chest...

[alexpgp]

Increasingly, agencies run new documents past their databases of document "segments" (typically, sentences) that have been translated before. Heck, I do this all the time myself, with databases of segments I've translated previously, as that is the entire point of the "translation memory" concept.

The problem arises when they embed found segments in documents sent out for translation and insist on paying a reduced rate for "editing" said translations.

Traditionally, editing a translation pays less than writing the translation to begin with. And that's because of the unspoken assumption that the translation has been competently written.

What is a competent translation?

Well, back when I worked in-house in Houston, and had a voice in defining "competence" in this context, we decided that a translation containing more than one major error per 250 words was poor, and one with more than two major errors per 250 words was unacceptably poor.

If one assumes that an "average" sentence consists of 25 words (an assumption that might be called "conservative," in the engineering sense), that works out to one major error per 10 sentences for a poor translation, and one major error per 5 sentences for an unacceptably poor translation.

So if you consider that a "pretranslated" segment that is less than an 100% match to the corresponding bit of source text is guaranteed to have a major error in it (after all, it's not a 100% match!), and that typically, less than 1 in 5 pretranslated segments are a 100% match, the end result is like editing an utterly and completely incompetent translation, for which one is offered a rate typically offered for editing a competent translation.

So in my current assignment, I'm not only finding the less-than-100% matches requiring attention, but the 100% matches as well. To wit:

{0>
Такая укрупненная частица (флок) осаждается более быстро.
<}100{>
Sedimentation of a large particle, or floc, is faster.
<0}
My version:
Such an enlarged particle (floc) settles more quickly.

{0>
Это происходит за счет совокупного действия нескольких процессов.
<}100{>
The effect can be explained by the following processes.
<0}
My version:
This is due to the combined action of several processes.

{0>
Параметризация турбулентности основана на сформулированном в (Озмидов, 1986) подходе.
<}100{>
The turbulent parameterization is based on the Ozmidov’s approach (1986).
<0}
My version:
Parameterization of turbulence is based on the approach formulated by Ozmidov (1986).

{0>
Турбулентность может быть представлена на основе вложенных вихревых структур разного масштаба, определяемого океаническими процессами.
<}100{>
The turbulence may be presented as the enclosed eddies of a different scale depending on oceanic processes.
<0}
My version:
Turbulence may be depicted on the basis of embedded vortex structures of various scale, determined by oceanic processes.

In addition to being annoying as hell, this kind of stuff slows me down significantly (I had, in fact, expected higher quality and factored it into my turnaround estimate).

Ah, well, there's nothing to be done about it now. I shall have to be more careful in the future, though, about accepting work from this client.

Cheers...

P.S. I repeat myself, I know, but it's therapeutic, okay?

Comments

[bandicoot.livejournal.com]

Accepting your definition of poor translations, I'm afraid most of today's writings in English I see in various publications would already qualify ;p

[alexpgp.livejournal.com]

Well, part of being a "major error" is for the source and the target to be saying different things, so it's not just errors in grammar and usage.

But I understand what you mean. :^)

Cheers...

[brightflashes.livejournal.com]

Gees, Sounds very tedious. Good luck with that. : )

[alexpgp.livejournal.com]

Thanks.

Tedium can most times be relieved with cash, y'know?
:^)

Having a technical background helps, too.

Cheers...