Thread poster: Owen owenfw
This is partially a technical question and partially a process question. I do quite a bit of proofing, but am always looking for ways to do it better/faster.
So, take a current large EU job: I have the source, the target, bilingual files, and the TM from the translator. Some questions:
1) If you intend to compare documents (so real, full-blown verification of the translation), how do you set up your workstation? Do you a) just open the documents side-by-side, b) step through the bilingual files, or c) produce a side-by-side table of the translation units (my medical clients do this).
2) If you do the side-by-side table thing, how do you create the file? My clients that do this use Wordfast, but I would be interested to hear how to create this type of file in either Wordfast or Trados, but mostly Trados.
Note: I edited this post to remove a part that was leading people to focus on something that wasn't the question I cared about...so the first couple of replies seem odd...it's just because I removed something so as not to get bogged down in irrelevancies.
[Edited at 2008-08-19 18:38]
| || || |
| | Karen Stokes
Local time: 19:46
French to English
| Reviewing or proofreading || Aug 19, 2008 |
Owen Witesman wrote:
1) How would you approach this in general? Do you a) compare every target sentence to every source? or b) focus on the target unless something feels wrong? I know this is often a question about what you've been hired to do, but often there is ambiguity with some clients.
I don't think you can even begin until you've resolved any ambiguity about what the requirements are, because the amount of time it takes to review/edit (compare target against source) is completely different from how long it takes to proofread (target only).
I'd also be very wary of focusing on the target (if you really are reviewing and not just proofreading). I've reviewed plenty of target texts that read perfectly well in themselves but don't necessarily bear much relation to the source. Not to mention checking that figures etc. have been transferred properly.
| | Owen owenfw
Local time: 13:46
Finnish to English
| They always want more... || Aug 19, 2008 |
Some clients are certainly very on top of things and understand different "levels" of proofing, editing, reviewing. Some are just going through the motions for their quality process and just want someone to blame if it isn't right.
So, when you are doing a full verification, when you are comparing source and target, how do you do it?
| | ViktoriaG
Local time: 15:46
English to French
| Tricked into reviewing at proofreading rates? || Aug 19, 2008 |
Karen is right - there seems to be confusion about the distinction between reviewing and proofreading. Better make that clear with your client if you don't want to be paid EUR 8 per hour...
My approach to this is to let the client know, as soon as s/he asks me to proofread for them, what my conception of that task is, how I do it and what it includes (what the value is for the client). If they say "no, we want more than that", then I offer to review, explaining again what that means and what that includes.
Now, for the technical part. It is best to have a few different approaches, adapted for certain document types. I don't review a parts list the same way I review an MSDS, and I don't review an MSDS the same way I review a manual. Here is the approach I most frequently use (for flowing text like manuals):
If the translation is of good quality, then I start by reading the target text first, without looking at the source. At that stage, I don't worry about the meaning being accurately translated - I worry more about the terminology, the readability of the text, and I pretend to be part of the target audience, e.g., I pretend I am about to change a filter in a compressed air system. In other words, I take off my translator hat and put my handyman hat on. I read the target text as though I was using it. I highligh all phrases that I have any kind of trouble with. Once I am done with this, I correct all parts of the text that I highlighted, occasionally looking at the source in case I am unsure whether I am faced with a writing mistake or a translation mistake. Once I am done with this, I start working on the bilingual document. However, my focus at that stage is really only to compare source and target - making sure the correct concept was rendered in French, making sure the target did keep secondary words that we tend to forget about, like very and adjectives. Once I have made sure that the target and the source mirror each other, I do one run of straightforward proofing. I simply read the target text without looking for anything in particular, and if something catches my eye, I correct.
My approach is about splitting the process up into several individual tasks. This way, I don't need to look out for too many things at once, and I can concentrate better. This usually enhances my attention to detail.
I have an interesting solution to propose for your third question. If you do have the source document and the target document, you can align them with your CAT tool, which will first provide you with a side-by-side view, segment by segment. I don't know about Wordfast, but with Trados (WinAlign), you can then read each segment side by side and make corrections as you deem fit directly in the WinAlign window. Those corrections are carried over into the TM you then produce by saving your alignment project. Then, all you need to do is take the source document and apply the TM to it - piece o'cake! Of course, you would still need to do a simple proofread of the target text to make sure the operation worked fine, but that doesn't really add to your workload.
[Edited at 2008-08-19 18:22]
| || || |
| | Owen owenfw
Local time: 13:46
Finnish to English
| Nice approach || Aug 19, 2008 |
I like your approach. I think I definitely fall into the "doing too many things at once" camp sometimes. Sometimes I do a one-pass blend of what you do. I'll read a paragraph of the target, so things are in context, and then read the same from the source. I find it extremely distracting to go sentence by sentence, except when the document is highly demanding and every word matters, as it were. For example, a medical device document obviously needs a different level of care than something more prose-like. In general, with somewhat less-demanding texts, I find that errors "jump out" as it were in the target reading and that the reading of the longish source section is sufficient to call attention to omissions.