In April 2001 I published my article entitled "Test TranslationsTo Do or Not to Do?". This publication generated quite a lot of feedback from all over the world, with all correspondents and freelance translators supporting my idea that test translations are practically useless and should be ignored. The conclusion substantiated in my article was based on my first nine months of marketing my translation services worldwide.
However, after three years of active translation for the worldwide translation market, I have had to revise my initial unconditionally negative attitude to tests. I would now like to share my new vision of the issue with my colleagues. My new approach may be of interest to them, particularly since it has enabled me to win such customers as Volvo Cars Russia, Philips France, Ford Motors Russia, Babylon.com, and Ericsson (Mobitex) as long-term clients.
There two main points I would like to emphasize:
1. There are two kinds of test
translations. A translator should learn to differentiate between them and respond
Let me explain what I mean.
Some test translations are sent to translators as a response to their application letters (sometimes with a CV attached). Many translation agencies respond in a knee-jerk mannerthey send a test translation. This is done even if the agency in question does not work with the language pair of the applicant, or if the agency has a huge database of translators working in this area. In both cases, your chances of getting a real job are non-existent, even if you do not know it. Consequently, I throw such testswhich are not related to a real projectin the trash basket as soon as I detect them in my mail box.
Some examples: Softitler (Italy), Xerox translation department (GB), Wordbank (GB). Even though my test translations were evaluated positively, I never received a real job from these companies.
Tests of another kind are sent to freelance translators when an agency has already won (or trying to win) a real project from a customer and is looking for the subcontractor most suitable for the job. In this case, a test text is a part of the translation project. In such cases, I usually try to do the test using my own quality assurance system.
Here are the principles of this system.
1. Do the test translation as soon as possible.
2. Understand who this translation is meant for, i.e., the end user, and base your terminology research on this information. E.g., if you (let's say, a Moscow-based translator) receive a test from Belgian agency containing some technical documentation of Yamaha, find your local (Moscow) Yamaha representative, distributor or dealer, visit the office or showroom, and get as many reference materials as possible. My motto is: the best translation (from the customer's point of view) is the one containing terminology mistakes the customer is already used too ;-))). This is joke, of course, but only partially, since it reflects a sad reality. Usually, the customer uses its local office for evaluating test translations. Therefore, let the evaluator see what he/she wants to see in your translation.
3. Have your translation proofread by one of your colleagues specializing in the subject of test translation (make sure he is a real friend!). You may establish a long-term cooperation of this kind, which will be a mutually beneficial two-way help. This way you will avoid typical mistakes such as omissions, etc.
4. Next, have your translation proofread by a local expert in the relevant field of knowledgeengineer, marketing manager, etc. You may find such a person easily through the Internet. Don't forget to pay himyou may need this person's terminological advice in the future, after you win the translation project.
5. After that I usually do my own final proofreading using the DejaVu Database maintenance interface (a DejaVu tool designed for alignment of source and target files)this is actually a table letting you to compare each target sentence against corresponding source sentence.
Only then is the test translation ready to be submitted to the agency or direct client.
My business results for 2002, my third year of distant translation practice, prove the efficiency of this approach to test translationsin total I received a workload of about 750,000 words of source language (English) from the customers mentioned above and several others. In many cases, I won these clients through test translations. And of course I was able to do this huge workload only thanks to daily use of Wordfast, my favorite translation memory tool, which, unlike Trados, never hangs my computer.
Please see some ads as well as other content from TranslationDirectory.com: