Emergency contact number: 078 905 76 02 (Sat. and Sun. from 08.00 to 17.00 hrs)
DE   FR   EN
ITSA | Inter-Translations SA
ITSA | Inter-Translations SA
Getting your message across.
      Whatever the language.
Connecting worlds for you – since 1974

Machine translation for professional translations

The accuracy of machine translation (MT) has improved tremendously in recent years, especially since the advent and continued use of neural machine translation (NMT). In NMT, the translation tool, analogous to the human brain, links various artificial neural networks with the help of algorithms to learn the translation process and infer word order based on probabilities. The tools that are available today can all be trained to continuously improve the quality of translated files, provided that they are used efficiently. But it is the human translator who stands at the end of the machine translation process. It is he or she who finesses the machine-generated text with their specialist knowledge during a process termed post-editing.

lightbulb-o

Blessing or curse?

MT tools are (as they currently stand) neither the holy grail nor a tool of the devil for customers or for translators. They are tools that, when used proficiently, can improve the quality of a text, reduce turnaround times and reduce costs. In spite of all this, human expertise remains irreplaceable as the final arbiter of desired meaning. Anyone wishing to use MT as a stand-alone tool to provide high-quality translations with immediate efficiency gains will soon discover that, in such instances, the cost of post-editing by a human translator will usually outweigh that of a new translation.

check

Appropriate application

Machine translation will not be suitable for all types of texts – only where it is used in combination with a CAT tool (i.e. SDL Trados). MT is initially more time-consuming and costly since, besides ensuring that the correct terminology is used, the quality of the source text will also be critical in determining the final outcome. Unless the source text has been worded entirely unambiguously, the risk of mistranslations cannot be ruled out. When it comes to context recognition, machine translation still remains imperfect. In order to achieve first-rate results, the source text should always be appropriately prepared via pre-editing.

question-circle

Is MT also right for you?

Machine translation will only be worthwhile if a long-term partnership is being sought in which the customer and translation service provider both remain in constant dialogue about terminology and possible applications, and MT can be combined appropriately with a CAT program. A use of MT that will not conflict with your existing processes therefore needs to be planned carefully. Take advantage of our experience and get in touch with us: we would be pleased to meet with you and discuss whether the use of MT might prove advantageous for you.