22/07/2025

Professional translation is in a state of flux. Machine translation and AI are changing the requirements. For companies that rely on professional translation, this poses the question: what role do translators play in this new world? The answer may come as a surprise: their importance is growing – only the tasks they do are changing fundamentally.

Today, translation services begin long before the first word is translated. Feasibility analyses for the use of machine translation have become a core competence of language service providers and translators. This involves checking whether the text type, domain and language combination are suitable for machine pre-translation and what savings can realistically be achieved. This assessment requires a deep understanding of both the languages and the technological possibilities and limitations.

The relevance of this expertise is clearly evident at oneword: over 75 per cent of our top clients already use machine translation solutions, which are always preceded by a thorough review by our specialist experts. Because not every project is suitable for MTPE (Machine Translation + Post-Editing). At oneword, the decision is made by people with years of experience who know exactly when technology offers added value and when traditional human translation is the better choice.

A misconception that is still widespread in many companies is: “We let the machine translate it and then someone just reads over it.” In reality, however, post-editing – the checking and editing of machine translations by humans – covers a broad spectrum of demanding tasks.
The machine output is evaluated according to the proven “rule of thirds”: only if at least one third of the output can remain unchanged, another third requires only minor corrections and at most one third needs to be heavily revised, is the use of MTPE efficient and sensible. This assessment requires not only linguistic but also analytical skills.

The subtle errors of modern AI systems are particularly challenging. While earlier MT systems also produced obvious grammatical errors, today’s systems – whether MT or AI – produce fluent-sounding and linguistically correct texts that can still contain many errors. Hallucinations – i.e. information invented by the AI but sounding plausible – require just as much attention and expertise as omitted words or inappropriate technical terms.

With the advent of NMTs and LLMs, the skills profile of translators has expanded significantly. What used to be primarily specialist language work is now developing into a combination of linguistic, technical and strategic expertise.

Terminology creation, for example, is a highly specialised task that becomes even more important with the use of MT. Translators not only research the appropriate technical terms, but also manage their consistent use across different systems. For example, it could be that the term “Einspritzdüse” must be reliably translated as “injection nozzle” and never as “injector” or “spray nozzle”, as this can lead to costly misunderstandings. Thanks to glossary integration, terminology can be integrated into NMT and LLM systems, but requires a specific structure and assignment of terms. Preparing the data requires an understanding of how different MT systems and LLMs process terminology.

Another area of expertise for translators and MTPE service providers is systematic error analysis. Post-editors and MTPE experts develop into data detectives to recognise patterns in translation errors and derive rules for system improvement.

Error analysis is particularly important with regard to the use of LLMs, as previous sources of error can be reduced or avoided entirely by using appropriate instructions (linguistic prompts). Language experts develop customised linguistic prompts for a wide range of requirements, from the formal implementation of certain specifications and stylistic adaptation for different target groups to the integration of company-specific terminology and cultural localisation.

However, language engineering, i.e. the strategic optimisation of language data and processes, is also one of the core skills in the translation sector. Translators analyse and clean up huge translation memories, for example, which can serve as a data basis for the use of AI in companies. The creation and expansion of terminology data and the design of workflows for MTPE and feedback analyses are also tasks that are becoming increasingly important.

Quality assurance has also developed from simple error correction into a multidimensional process. Internationally recognised standards form the basis for this: DIN EN ISO 17100 defines requirements for translation services, while DIN ISO 18587 specifically regulates the post-editing of machine translations. Supplemented by DIN EN ISO 9001 for quality management systems, the result is a comprehensive quality framework that we have fully implemented as one of the leading language service providers.

These standards are more than just certificates on the wall. They define specific processes in which human expertise remains indispensable, from the selection of qualified translators to multi-stage quality checks and systematic error analysis. ISO 18587 in particular makes it clear that post-editing goes far beyond a “quick read-through”. It requires specifically qualified post-editors with proven expertise in machine translation as well as established processes for analysing feasibility and managing feedback.

In practice, this means translators develop into process architects. You create detailed guidelines for different text types and domains. Technical documentation, for example, is subject to completely different criteria from marketing texts: while the former emphasise absolute terminological precision and clarity, the latter require cultural adaptation and creative freedom.

The detailed feedback process that we have implemented at oneword demonstrates the importance of this systematic approach. The thousands of post-editing projects we have worked on provide valuable insights into typical sources of error in various MT systems and LLMs. This data flows directly into the continuous optimisation of our processes. It’s a task that is only possible with the expertise of experienced language professionals.

The translation industry is undergoing a transformation towards ever more demanding and complex tasks. In addition to excellent language skills and in-depth industry-specific expertise, highly qualified translators today increasingly also have technical, organisational and consulting skills.

What this means for your company is that the expertise described in the article makes professional translation service providers and experts a decisive success factor in turning AI potential into real business benefits.

Translators today are language engineers, quality managers and strategic consultants – in other words, experts who combine a deep understanding of language with technological expertise. They ensure that your international communication is not only linguistically correct, but also culturally appropriate and technically optimised.

Would you like to find out how modern translation processes can bring concrete benefits to your company? Our experts will be happy to show you how you can use our oneSuite platform and customised MTPE solutions to reduce your translation costs and increase quality at the same time. Please feel free to arrange a no-obligation consultation.