Man versus machine

Zurück zur Übersicht
06.10.2017

No doubt about it: IT-supported processes and digital translation tools that help to automate workflows wherever possible are now standard for translation service providers. This is the only way they can fulfill the increasingly complex requirements of their clients from various industries and provide these companies with high-quality, speedy translations at reasonable prices.

Working conditions have changed dramatically: in the past, for example, one hundred pages of text would need to be translated into English or another language, with a timeframe of 30 days. Today, companies generally work with many small text modules, which are then used to update large documents or marketing texts

Digital communication channels have also had an enormous impact: if a piece of text is changed or added to a company's German website, all the other languages on the site should also be changed accordingly by the next working day. “This is only possible with a well-established global network of native-language translators, IT-support and a high degree of process expertise,” says Transline's CEO, Katja Schabert. 

Software assists translators

One of the most important IT tools for translators is the “translation memory”. This customer-specific analysis and research system trawls through a connected database of completed sentences or text modules, extracts them and offers them to the translator. The translator can then decide whether to accept or change the suggestion. This greatly accelerates the translation process and cuts costs. 

However: “Without the ability of human experts to understand background information and context, and to use these to make a correct, professional decision, the whole thing doesn't work,” notes Katja Schabert. “So far, no computer has been able to replicate the feeling for language and power of judgement that are necessary for a high-quality translation.” 

Undue complexity 

It's impossible to foresee that computers will completely take over the work of humans. Because: “Language is a highly complex, rapidly changing phenomenon that always takes place in a context of culture and meaning. For example, certain terms are understood completely differently in different regions and at different times.” In addition, many words have multiple meanings – and the correct one can only be taken from the context. Meanings and how they continually resonate in languages cannot be detected by any computer program to date. Complex grammar phenomena have barely been researched, let alone solved in software technology.

Automatic machine translation is still being developed. So far it does not fulfill all the requirements that companies rightly place on a translation service provider, according to the Transline CEO. Linguists are of the opinion that for certain languages, such as Finnish, machine translation will never be able to produce the same quality that a human translator provides.

“On the other hand, other languages, such as Norwegian, are comparatively well suited to machine translation. This is already being used commercially, if not yet in the areas of technical translations or marketing.” So far, computer translations have not been able to produce the necessary accuracy for technical topics or the emotionality that marketing texts require.

Quality begins with the source text

A further problem: the computer does not report on any errors in the source text, but rather translates these word for word. “A machine translation is only as good as the source,” explains Katja Schabert. If the source text is not clearly formulated, has multiple meanings or is ambiguous, this requires the computer to do something which it cannot do: make a judgement on the plausibility and evaluate the context. “This is where humans are needed.” 

Lastly, Katja Schabert gives a rather pragmatic reason for continuing to rely on the power of human judgement: “It's essential to provide both reliable quality and confidentiality for a company's critical content. An incorrect translation can lead to a liability lawsuit.“ 

In summary: companies that want an eloquent linguistic presence in their target countries need to employ the expertise of professional service providers. They require this expertise to decide when it makes sense to supplement the workflow with machine translation and when human intervention is required. Katja Schabert: “Transline is completely open-minded when it comes to new technologies, and of course we use IT-systems and computer programs to increase our efficiency. But a real person always has the final say.”

Katja Schabert is the CEO of Transline Deutschland GmbH. With more than 100 employees in Europe and approx. 5000 specialised translators worldwide, Transline is one of the largest translation service providers in Germany. Its experts translate and localise technical documentation and marketing materials for more than 160 local markets, effectively reaching 93% of the world's population.

E-Mail schreiben
Anrufen
Beratungstermin buchen
Anfrage senden