The Translation Process – 2

Having dealt with the commission side of things, I want now to look at the actual process of translating. As far as I am concerned, this is broken down into three major stages: analysis, translation, checking.

Analysis

The analysis stage covers an analysis of the actual text, to verify that it is complete, to perform an initial word count, and to determine possible causes for concern. Causes for concern would include unknown terminology, unfamiliar concepts and unclear instructions. If the translator feels that the text in isolation does not provide enough context, it is often necessary to approach the customer to ask for background material. This is usually fine, as long as the translator is dealing directly with the end-customer, but if an agency or middleman has set up the job, it might be difficult to get hold of background information.
 
If no background material can be provided, the translator is forced to rely on his or her own wits to solve the issues at hand. Usually this means accessing reference materials. As most deadlines do not lend themselves to permitting research days in local university libraries, the main resource available to the translator in this instance is the Internet. It is safe to say that the Internet has revolutionised the way in which translators, especially freelance translators, work. Not only do very many freelance translators use the Internet to communicate with their clients via e-mail, an increasing number also use it as their primary research resource. Modern search engines, such as AltaVista and Google enable users to search millions of web pages in a multitude of languages within a matter of seconds, and the modern translator has to be able to master this technology to derive the maximum benefit. Often, searches are simply too general, returning 1000s of possible pages, and so the translator has to be au fait with the advanced searching options, and should have an idea of how to apply boolean logic to searches. Susan Larsson, a translator of Swedish who lives in the North West of the USA has a good selection of search engines, as well as tips on how to use them.

Translation

For the actual translation phase, different translators will have different preferences. Some like to work from paper, and start typing their translation into a blank word-processor document. Others prefer to use a copy of the original document (providing it was received electronically in a form that can be edited readily) and overtype a section at a time. If there is complex formatting in the document, which has to be maintained in the translation, it is often a good idea to overtype the original file where possible.

Personally, I nearly always use a translation memory tool to do my translations. This type of tool works with me as I translate, and saves my translations as I progress. When I come to a section that has been translated before, the tool recognises this and suggests my previous translation. The most advanced of these tools also incorporate a “fuzzy matching” engine, which enables them to detect a level of similarity between translations. Thus a tool that has already learned my translation for “Click on OK to continue.” will recognise that “Click on Cancel to exit.” is partially similar, allowing me to decide how to modify the original translation to fit the new source text. This sort of tool enables me to be more consistent in my translation, and that is especially important in the technical texts that I translate. Good user manuals use consistent instruction formulations to aid the reader in remembering complex tasks. A translation of such a manual should be no less consistent in its formulation of those instructions, and a translation memory tool ensures that consistency is maintained throughout. Currently, I use two different tools, both of which offer different advantages.

The first tool I use is called Translators Workbench, developed and marketed by Trados, a German-based software house. Trados software has become the market leader in recent years, and its ability to integrate fully into the most common word processing software, Microsoft Word, means that I can be sure to preserve even the most complex of formatting. In addition to the Workbench software, Trados also produce a fully-integrated terminology database software called Multiterm. I use this to ensure that my terminology is consistent throughout my translation, and to leverage the knowledge gained in one project so that it can be reused in other projects. The second tool is produced by SDL Ltd., a British-based localisation company. The tool, SDLX, performs roughly the same job as Trados, but goes about its work in a slightly different manner. Whereas Trados Translators Workbench and Multiterm integrate fully into MS Word, SDLX offers the translator a separate translation environment. I tend to use this tool when the formatting of my text is less complex or important. The advantage of SDLX is that less computer resources are used in comparison with Trados (which requires Microsoft Word to be running at the same time). The SDLX database access routine also seems to be a little faster than the Trados system.

As well as the improved consistency, mentioned above, an added advantage of using a system such as Trados is that I can perform a provisional check on my translation as I progress through the document. Trados takes me through the document a segment at a time (usually this is the same as a segment), and so I am more conscious of ensuring that all sections of the text are translated than I would be if I were to overtype the translation in a more random fashion. A disadvantage of the Trados is system is that there is sometimes a tendency to lose the cohesion between sentences and within paragraphs. This problem can be overcome with experience of using the tool, however. And there is always time for a thorough check and reworking of the translation after the main translation phase is over.

Checking

This phase is just as important as any other in the translation process, and should not be neglected. Generally, translations should be checked a number of times, with the translator looking for different types of error each read-through. For example, it is advisable to read through first to ensure that nothing is missing from the translation. Then you should check that all numbers, dates, references and personal names are correctly localised. Finally, the last check is the spell-check, relying both on the computer’s internal spell checking program and on the translator’s eye for typographical errors that are not caught by spell-checkers, such as typing “tot he” instead of “to the” or “form” instead of “from”.