machine translation

Machine Translation (MT) is a fascinating branch of computational linguistics that captivates both linguists and computer scientists alike. For the translation industry, advancements in MT and the accompanying increase in automation present new opportunities to enhance productivity and remain future-proof.

With the globalization-driven increase in demand for information and documentation, machine translation has become an indispensable prerequisite for intercultural communication. It is now a vital component in the world of language service providers and localization experts, and we have been successfully employing it when explicitly requested to do so by our clients.

ISO 18587 certification

In May 2024, we received ISO 18587 certification for post-editing of machine translation services.

This certification confirms that, alongside human translation, we employ future technologies such as Artificial Intelligence to the highest quality standards in our translation services.

The ISO 18587 certification standardises essential processes:

  • Project preparation
  • Production process
  • Post-editing

These processes are firmly integrated into our workflow, ensuring consistent quality.

For further information, please feel free to contact us.

Since the inception of MT research, various approaches and methods have emerged in the development of MT systems, differing significantly in strategy and complexity. The two primary approaches are the traditional linguistic approach, which relies on linguistic rules and multilingual dictionaries, and the statistics-based approach, developed through digital progress, which bases translation on the mathematical calculation of statistical probabilities.

A new milestone is the neural machine translation, still under research and development. Like the statistics-based MT, it utilizes extensive data from bilingual text corpora and uses artificial neural networks and training data to learn the relationships between the source and target languages to make translation decisions.

The use of machine translation, particularly statistics-based and neural MT, is gaining increasing importance. Combined with the well-established translation memory technology in the translation sector, it offers enormous potential for a high-quality approach. MT systems are specifically trained with the language resources already available in most companies (translation memories, glossaries, etc.), which then serve as a validation basis for machine-generated translation decisions.

The interface between powerful MT systems and available translation memory resources opens up potential for new translation solutions. Especially in today’s world, where product launch times (time-to-market) are becoming shorter and shorter, the use of high-performance and reliable MT systems for the translation of documents such as technical documentation, operating instructions and product information has become essential.

Many of the MT systems currently used in the translation industry can also be individually adapted to the company-specific requirements and demands of customers (translation style, language and domain-specific specialist content, terminology) and integrated into their own standardised and automated translation environments.

When introducing a customized MT solution, it is therefore particularly important to analyse and evaluate the selection of the right MT system and its implementation in day-to-day business, as well as the economic objectives and the requirements of the MT service spectrum.

Post-editing – an integral part of the translation process

Post-Editing

Depending on the source document and calculated project costs, the various applications of machine translation (technical documents, manuals, digital content) must consider the guidelines of DIN ISO 18587 and different levels of post-editing.

Sometimes raw translations without linguistic correctness are sufficient to understand the core message. For other translations (e.g. published documents), light to full post-editing is recommended to achieve a high-quality end product.

While the degree of post-editing always depends on the client’s objectives and quality requirements, post-editing must always be understood as an integral part of the entire machine translation process, involving all participants in the MT project from the start.

Our ISO 18587-compliant services

Reviewing the source documentation for suitability for machine translation in terms of content, target audience and purpose, with pre-editing if necessary

Linguistic/formal preparation of the source text to improve MT results and facilitate post-editing, including spell check and formatting

Creation and/or optimization of language resources (translation memories, glossaries) as a basis for building a customer- and subject-specific MT engine

Light post-editing for readability and content comprehension or full post-editing for grammar, punctuation and style based on client-specific terminology

Use of qualified, native-speaking post-editors in accordance with the processes and guidelines of the post-editing standard DIN ISO 18587

Machine Translations – with or without post-editing – are conducted exclusively in consultation with our clients.

Any questions?

Questions about machine translation, post-editing, language combinations, subject areas, file formats, an upcoming translation project, or your internal workflows in general? We can look back on 30 years of experience in the translation and localization sector and work with you to develop customized solutions and optimized workflows for your company and your translation projects.