Machine translation is becoming increasingly important. As a matter of principle, we do not use any machine translation system (MT system) without consulting our customers. Today’s state of the art are neural machine translation systems (NMT systems), which make different errors than human translators or rule-based or statistical MT systems.
.
Post-editing ultimately determines how successfully machine translation can be used. Too much effort negates the economic and time benefits of MT. Too little effort increases the risk of errors and can lead to serious mistakes. Post-editing must be learned. It is a specialization in itself, different from editing human translations.
Do you need a translation? We will send you a quote within the shortest possible time. Send us your request with this quote form.
The ISO 18587:2018 standard, which defines the requirements for revising machine translation, distinguishes between two levels of post-editing: light post-editing and full post-editing.
In light post-editing, the post-editor must correct errors in content, such as comprehension errors, numbers, or incorrect statements. However, it should not eliminate all syntax errors as long as they do not distort the meaning of the sentence. The goal is an understandable text that does not necessarily have to reach the quality of a human translation.
In full-post editing, the goal is to achieve a translation of similar quality to that of a human translator. The ISO 18587:2018 standard stipulates that post-editors should have the same qualification as for specialized translators as defined by ISO 17100.
It is not easy for a post-editor to quickly decide what to accept from machine translation and what to correct. It takes several months for a post-editor to achieve the desired output. On average, and depending on the quality of the MT output, a post-editor can correct about 7000 words per working day.
We support our post-editors and give them feedback on the result of their work. Post-editors gradually learn the specific errors MT systems make with certain topics, clients, and text types, and over time become much more targeted with machine-translated texts.
We provide our post editors with our quality assurance software, ErrorSpy, which has its own editor for corrections and automatically detects and displays a wide range of errors, such as consistency errors, numerical errors, or terminology errors.
Despite all quality assurance measures and technologies, a residual risk remains with machine translation + post-editing. Some machine translations sound so good that the post-editor may read them too quickly. If the meaning is wrong, if a crucial word is missing, or if a homonym is misinterpreted, errors can be overlooked.
Lorem ipsum dolor sit amet, consetetursadipscingelitr, sed diam nonumy eirmod tempor.
Loremipsumdolor sit amet, consetetur sadipscing elitr, sed diam nonumy eirmod tempor.
Lorem ipsum dolor sit amet, consetetursadipscingelitr, sed diam nonumy eirmod tempor.
Loremipsumdolor sit amet, consetetur sadipscing elitr, sed diam nonumy eirmod tempor.
We can offer you machine translation + post-editing or postedit machine translations that you have produced yourself. If such projects are regular, then we will discuss with you in advance aspects such as your quality criteria, the possibilities for optimizing the source texts, the specialized terminology to be used or the processes for improving the machine output (feedback from post-editing)
Our services for you are as follows:
We have organized numerous MT projects. Benefit from our experience. We will discuss how and to what extent MT is suitable for you. Just get in touch with us!