Preparing content to translate it efficiently, cost-effectively and more consistently.
Every day, companies publish a lot of duplicate or similar information that generate large volumes of text and affect translation costs and project schedules. Professional translators use CAT tools (Computer Assisted Translation) to process these large volumes efficiently. CAT systems will detect and propagate exact repetitions in their documents and use translation memories to recycle similar translations coming from previous projects. However, existing CAT tools cannot detect many types of near-repetitions that frequently occur in the original text.
Content optimization aims to improve the effectiveness of tools, to better control costs and improve the quality of translations.
Where do these duplications come from?
Creating new content often starts with taking an existing document that will be re-used, edited and completed to to be used in a new context.
This practice allows writers to quickly generate new documents that remain perfectly consistent with those already published.
A company can then publish a full range of materials for each of its products or services. Some contents are very specific, while others are common or very similar.
The multiplication of media (leaflets, presentations, web pages, mobile applications, etc.) also contributes to the profusion of replicated and near-identical contents. Duplicated content is often slightly modified to suit each medium (change of format, code, illustrations, etc.).
Why is that a problem?
Whether calculated by word, by page, by character, by line or even by hour, the cost of a translation is always linked to the volume of text that the translator will have to process.
When translating a new document, the full content is accounted for, even when it contains only a small portion of unique information.
We can identify duplicates from content already translated in other documents, so that they are not fully re-translated. This process limits the costs and reduces errors and inconsistencies with existing versions.
Existing tools: CAT
For over 20 years, professional translation services have relied on Computer-Assisted Translation (CAT) technology to become more productive and more reliable.
CAT systems are generally developed around translation memories (TMs), which are databases that store all the existing translations for a given customer.
During a translation, the system analyzes each sentence to be translated and compares it to selected translation memories. If it finds that the same sentence or a similar version has already been translated, the translator will instantly receive a suggestion based on previous translations.
Translation memory tools should not be confused with machine translation services, where software generates translated sentences according to a predefined algorithm. We currently limit the use of machine translation to a few very specific situations. A professional translation should clearly reflect the original content without any ambiguity. Machine translation is not ready for this task, but it offers an efficient solution for understanding a text when the quality is less important than the cost or speed of the results.
We consider this technology as an understanding enabler for an informed reader who will be able to identify translation errors and understand the reason for any incongruities.
Repetitions and perfect matches
CAT tools are very effective for repeated sentences that generate perfect matches from the translation memory. They can process repetitions quickly and consistently. The same “source” text found in different documents will use the same translation to ensure perfect consistency between translated documents.
Of course, the translator will still need to verify that the translation suggested by the TM is correct in its new context. Indeed, the same source sentence may legitimately require different translations, for example, to agree with a masculine/feminine or singular/plural subject.
Similarities and “fuzzy” matches
To detect partially identical content and suggest relevant proposals, translation memories rely on a rating system based on “partial matches” also called “fuzzy matches”.
In that case, the translation memory will offer existing translations whose content is similar to the new text to translate. It will also display the “matching rate” to indicate the degree of similarity with the new sentence to translate. The translator can use the proposed translation, adapting it to the new context, while maintaining a uniform style and terminology between documents.
Processing partial matches requires a good working knowledge of the language in order to distinguish the nuances between several propositions.
Limitations of CAT and translation memories
CAT tools offer translators great efficiency, but they cannot yet deal with small variations of the same source texts.
Although already highly perceptive, their algorithms cannot detect certain similarities that a human translator will immediately identify and adapt effortlessly.
What is content optimization?
Content optimization is the process of preparing documents before they are translated, in order to anticipate and work around the current limitations of CAT tools.
By carefully analyzing the texts to be translated, we can quickly identify similarities not detected by CAT tools and optimize the documents to benefit from our innovative processing.
In addition to improving the translators’ efficiency and significantly lowering the costs, content optimization ensures more consistent translations and better compliance with quality control requirements.
Optimization can also detect errors and inconsistencies in the source text to correct them as necessary.
Translating less, but better.
Kevrenn content optimization: An all-inclusive service… also available à la carte
Backed by more than 20 years of experience with CAT systems, Kevrenn has designed and developed some unique solutions to optimize your translation projects.
These solutions are aimed at simplifying the management of your translations and lowering their costs. They apply to all types of documents and can automatically identify and process similarities that are not detected by existing CAT tools.
Optimization is particularly effective for projects containing similar sections or frequently updated materials (catalogs, brochures, websites, marketing documents for different product ranges, etc.).
We systematically optimize all the translation projects entrusted to us.
But you can also take advantage of these solutions to optimize your projects before you translate them in-house or outsource them to your regular external translators.
Please contact us for more information about our optimization solutions or to request a free demo of DFR to evaluate the benefits using your own data.