Thousands of translations and terminology entries are centralised in one database for long-term storage and interactive use. The concept of a translation memory is to store and help re-use previously translated, edited and approved content. This drives speed and consistency and reduces costs and time spent in the long term. Translation memories help the team to work simultaneously on a project, makes it possible to perform fast edits, updates and pretranslations, thus helping you focus on the most important tasks in a project and shortening the time spent.
No doubt it's the personal power and intellect of our linguists that is of utmost importance in our working process and in which we place greatest trust. But it's also true that we are early adopters of anything new in the field and try to constantly learn and integrate in our work new methods and tools to assist and enhance productivity. There are no set applications we use, which means we are open to any solutions and it's of high chance that we'll be able to adapt your needs for a specific and customised tools upon requests. Apart from that, we use on a daily basis solutions such as SDL Trados Studio, Across, MemoQ, former SDL (Workbench and SDLX), Alchemy Catalyst, Deja Vu, Wordfast, Logoport, LEAF, Helium and more.
It's humans that drive the change, but machines are there to do the hard work. Adopting Machine Translation (MT) in our work is comparative new to us, but if the cllient wants us to, we can make it possible to integrate MT along with TM.
MT applications work based on the so called statistical or rule-based algorithms. Besides the corresponding rules and statistics for the particular language pair, they also require specific information and guidelines for the given text. The MT engine is trained in multiple cycles to “learn” to produce the desired results. That means: With each new training cycle, one of our MT specialists gives the MT engine the exact translation information that will bring about an improvement in the next round of translation.
This training is carried out before the machine translation begins and continues during the course of the project. During this process, text-specific information and references, such as glossaries, dictionaries and various translation memories could be imported into the engine. The system subsequently delivers results that can be optimised further by specially trained translators, if the client so desires, in order to produce translations that meet a high linguistic standard.
Although the source texts often need to be prepared and the machine translations often need to be post-edited, it’s worth the effort: Clients can realise substantial savings on certain projects, depending on the quality of the source material.