Translating with pen and paper? Dragging around dictionaries and printing out terminology lists? If you hear those questions and picture yourself back at school before the new millennium, perhaps you will feel it even more keenly when you learn how the translation process has changed since that time thanks to the introduction of modern technology.

The first major revolution was in computer hardware, which was becoming increasingly powerful, along with the associated research into automated translation workflows. The industry-wide use of so-called CAT tools (computer-aided translation), i.e. software that could intelligently reuse text that had been previously translated as part of a previous project, was not far behind.

Machine translation has ushered in the second major revolution. So, should we send all the staff home and celebrate the miracle of machine translation?

We don’t think so…

Anyone who wants to add machine translation to their existing processes in a way that is efficient and sustainable for the long term will require a carefully considered quality management concept that incorporates the work of our talented language experts.

Hitting the mark, not missing it completely – we show you what matters!

Machine translation (MT) in day-to-day work

The influence of MT technology on modern life cannot be denied. Sometimes it is discreetly in the background as you scroll through supporting documents, sometimes it is obvious such as the use of translation software to overcome language barriers. The increase in globalisation coupled with the human desire to consume content in our own language have led to daily growth in the need for translations.

A variety of use cases have arisen out of this with sometimes very different requirements – from a simple transfer of information between colleagues to texts featuring complex language and content destined for target markets with demanding clientèle.

The demands on MT systems are high: More content in less time at a better price point. In order to keep up, good training and regular retraining is necessary!

Good training is half the battle

The story of MT goes back to the 20th century, but it has only been in recent years, thanks to advances in the areas of language processing and deep learning, that it has been distilled into a technology with enormous flexibility that has shown clear advances in quality in comparison with the earlier versions.

Using what are known as neural machine translation engines (NMT), bilingual or multilingual text corpora containing verified translations are collated, cleaned and then language structures are defined with the help of deep learning algorithms. Over several training rounds, the results are checked and further perfected. With NMT, the information is even contextualized in the form of word clusters, which the system can use to decide the probability of certain word combinations appearing.

Impressive, isn’t it? Yes, but it’s not without errors.

The quality of the MT output is only as good as the material used to train it.

Do your texts contain incorrect terminology, inconsistencies or reference errors? Then the MT engine will almost certainly produce these errors as well.

When it comes to introducing an MT solution, we take you through the most important questions, step by step.

Click here to download our checklist

Would you like to efficiently and sustainably deploy machine translation for your projects?

Discover more in our white paper. We help you to implement the best solution for you.

First name *

Last name *

E-mail *

Privacy *

• I hereby agree to the Privacy Policy.

Submit

More information about machine translation

STAR proudly presents: Our colleagues, Birgit M. Hoppe and Birgitta Geischberg have added to the existing work by Jean-Marc Dalla-Zuanna and Dr. Christopher Kurz with their article “Quality of terminology processes in corporate contexts/Agreeing and harmonizing terminology”.

In this age of digital transformation, we are faced with the challenge of clearly converting the hazy notion of translation quality into a world dominated by concepts such as key figures, artificial intelligence and automation. We are confronted on an almost daily basis with the redefining of our profession and work environment.

The new BDÜ collective volume is a bold and successful attempt to bring some order to the system and outlines a strong link between theory and practice. On the one hand, it reflects findings, tips, strategies and suggestions from day-to-day work, but on the other hand it also addresses basic theories and offers a reflective and multifaceted introduction to the issues surrounding quality and translation.

It includes considerations about translation quality, factors that have a significant influence on translation quality, importance of the elements of the translational ecosystem for achieving translation quality, as well as application in practice. Read more

“Terminology management” is a key phrase in the standalone section of the 550-page volume that explores factors that have a major impact on translation quality. While extremely important, the impact of consistent, well-managed corporate terminology is often overlooked.

This is reason enough for STAR to put the topic in the spotlight as part of STAR terminology weeks, running from now until the end of the year.

Standardised, manufacturer-independent reference terminology for the automotive industry is now available. It came about as part of an EU project with the goal of facilitating access to vehicle manufacturers’ repair and maintenance information (RMI) by “authorised independent operators” (independent workshops, testing institutes, etc.) (EN ISO 18542 standard).

As a central partner, the STAR Group was responsible for preparing the initial data set, providing a terminology manager and undertaking a large part of the project management. Working closely with experts from automotive manufacturers and authorised independent operators, search-relevant English terms were identified, validated and then translated into German and French. To facilitate this collaborative creation and maintenance process, the STAR Group implemented a central, web-based workflow solution based on WebTerm 7 (terminology management) and STAR CLM (corporate language management).

The aim was not to collect as many terms as possible, but to collect as many terms as relevant to the vehicle category as a whole. The digital reference terminology currently contains 521 detailed data records and is now available in three languages from our collaborative partner DIN Software GmbH. Read more.

Following its initial publication, experts are once again working on the terminology: It will be expanded yearly with up-to-date terms for all vehicle classes. In 2020, the main focus will be on the “heavy goods vehicle” sector. The system and data architecture are set up in such a way that additional sectors in the EN ISO 18542 standard (such as motorcycles and agricultural vehicles), as well as additional languages, can be easily included. This means the new digital terminology also offers users an interesting long-term perspective.

Click here to download our PDF