Clients of language services are always looking for the best translation for the best price. This is only natural. In the absence of any way of judging the linguistic qualities of the service provider, one of the ways of differentiating between offers is the translation process. Agencies often highlight their methodology, backed up by ISO standards. However, they all end up looking the same, precisely because of these standards. Things are a little more complex, however, when you look not at an agency, but at the translator himself. While there are as many translations as there are translators, each one also has its own method, and very few describe it. Here’s the method I use when I’m faced with a new text.
Process and translation agencies
Translation agencies very often highlight their translation methods and other workflows, along with the ISO standards that go with them. But, contrary to what the agency may lead you to believe, this emphasis is not a real differentiating factor from the competition. It’s a safe bet that your neighbour has exactly the same qualifications. What’s more, if they don’t have the famous ISO 17100 and ISO 9001 standards, it’s more than likely that they apply the same procedures.
In fact, translation processes are not as numerous and disparate as we are led to believe, with a few variations here and there, it is generally limited to 3 offers:
- Translation only
- Translation -> Proofreading
- Translation -> Editing -> Proofreading
So we have 3 levels of service that agencies can offer, whether they have the standards or not. Just because an agency has ISO certification does not mean that it cannot offer you a translation-only process, and there is no guarantee that the proofreading (also known as QA, QC or quality control) will not be carried out by the project manager, which is often the case. In any case, these stages must be carried out by different people, qualified if possible. I say ‘different people’, but I should also mention the machine, because in a post-editing process, it is the machine that purely and simply replaces the translator. This in no way detracts from the process employed.
Finally, while methodology is not a real differentiating factor between agencies, it is between agencies and freelance translators. By definition, freelance translators work alone, so they cannot carry out all the stages. But there’s nothing to stop the client using several freelancers to carry out all the stages (which is what agencies do) or, better still, asking the translator directly if he has any colleagues he could work with.
What about freelance translators?
That’s why freelance translators don’t talk about methods. They work alone and cannot offer anything other than a translation. However, the method used by the translator on his or her text is also very interesting and a determining factor in quality. While everyone has their own methods, I am far from suggesting that one method is better than another. I firmly believe that everyone should find the method that suits them best.
It is not my intention to set out the ideal method. I simply want to present the one I use most often, whether consciously or not.
General reading
I always start with a general reading of the text. This may seem obvious, but in a hurry it’s easy to forget to read the text. With an estimate to turn around very quickly, it’s easy to import the text into our CAT tool and stop there. We check diagonally that everything is translatable, we look at the statistics and we forget to read the text. But reading the text is essential. It allows you to get to grips with the subject and anticipate certain problems. It gives you an idea of the research needed and a thousand other little details that will save you time. This is also where the first questions may arise that the customer may be able to clarify.
CAT and analysis
It is only after this first reading that I import my text into my CAT tools. I’ve already mentioned the translator’s tools in a previous article, so I won’t go into detail here. But it’s also an opportunity to check that the entire text has been imported correctly, that the segmentation is correct and, if necessary, to use machine translation and translation memories.
As far as translation memories are concerned, there are different ways of managing this aspect. I’ll talk more about this in a separate article. Personally, I always add 2 or 3 memories:
- The memory provided by the customer, if it exists;
- My historical memory with the client. I generally apply a 1% penalty to it (to avoid perfect/context matches and other 101% matches);
- My memory in the text domain. Here I apply a 2% penalty (so that 100% matches don’t occur).
The penalties applied to these memories are designed to avoid skipping over segments too quickly in the new context. They also serve to draw my attention to certain points. But I’ll expand on that in the article on memories.
Quick translation
This is where things get serious. I do an initial quick translation in my CAT tool. By quick translation, I mean: a correct translation, but without detailed research. The CAT tool has the option of assigning statuses to segments. I take advantage of this to validate the segments that I’m completely sure of and leave in draft form those that I have doubts about or that I’ll have to come back to later. This is also the time when I add comments to the segments. Whether it’s for information to pass on to the client (typos in the source text, for example) or for myself during my research.
Advanced translation
Once my first quick translation is complete, I start a second translation. After a thorough proofreading, I’ll go back over each section left pending and my comments more carefully. This is the time when I’m going to waste 30 minutes on the word that’s bothering me by doing extensive research. I could perfectly well include this stage in my first translation, but I prefer to separate them. This allows me to manage my time better and to have peace of mind when I ‘waste’ an hour looking for information.
Rereading sentences that have already been validated also gives me a better idea of the logical sequence. So I sometimes make stylistic or editorial changes at this stage.
Proofreading with tools
Like everyone else, I make mistakes when I write fast, reword a lot and change elements here and there. Hence the importance of proofreading.
I carry out an initial careful proofreading within the CAT tool. I take advantage of the advanced features in terms of QA and other automatic processes.
I then generate my target text in the correct format and carry out a second proofreading. This is done by myself and using an external spellchecker, such as Antidote, which provides a lot of information.
Reading aloud
Normally, my text is already ready, but I like to make sure it flows. One of the best ways I’ve found so far is to read it out loud. I try to take advantage of the times when my wife is away to do this. That way she has fewer doubts about her husband’s mental health and, above all, I don’t breach any confidentiality rules.
Layout
The last stage in my translation workflow is the page layout. Whether I’m working with a simple Word document or an InDesign page layout file, it’s imperative that the client finds what they have provided. Bolded, highlighted and underlined text, as well as spaces provided for text, must be found in the translation. There is nothing more frustrating for a customer than having to redo the layout of their document. How can they know which words should be bolded in a sentence when they don’t know the language? The rate of expansion when switching from English to French can also cause texts to slip out of the boxes provided. This is regularly the case in PowerPoint and InDesign files. These are all small points that I adjust before delivering my document.
One method among others
This is my working method. It’s not necessarily better than any other. It is also relatively time-consuming, given the number of passages in the text. But it’s the one that suits me best at the moment.
Don’t hesitate to let me know in the comments how you do it!