We’re happy to say that the short answer is yes, with the right provider. If you’re in the mood for a longer explanation, however, please read on.
As you might know, machine translation (MT) can be considered an enfant terrible of translation industry. This is mainly because when MT was in its nascence, it was used to produce horrible, stilted target texts – often results of horrible, stilted source texts that were fed into the engine in the first place. The MT itself was not advanced enough to properly handle most texts and required extensive pre- and post-editing processes, which were often skipped to cut costs even more, thus adding the insult to injury and contributing to the misunderstanding of the process.
Luckily, things have changed and it’s time to put post-edited MT where it belongs – in your arsenal as a great method to cut costs without compromising on quality. To successfully do so, two questions must be answered:
Or what texts lend themselves well to Machine Translation?
We advise using MT to enhance translation process in informative or technical text types, like instructions, user manuals, descriptions of services, online help pages and messages, product catalogues and similar offerings. With these texts we can easily use MT to reduce costs and keep the quality.
We can even use MT for regulated legal and medical texts, however this demands extensive post-editing and quality control, garnering us smaller cost-reductions.
We don’t advise using MT for highly idiomatic, less-than-straightforward texts or such texts where the individual style is of great importance as they don’t lend themselves well to the strict rigors of MT.
We use the process some call augmented translation, where your text is post-edited by a translator and then independently verified by another expert, thus fulfilling the requirements of ISO 17100. For this, we need a domain MT engine for a given language pair. Alternatively, we can “train” universal MT engines for a given language pair using more than a hundred thousand words’ worth in client- or field-specific texts from existing translation memories.
This process allows us to reduce the cost by approximately 25-30% compared to human-only translation. The result faithfully reflects the accuracy, fluency, and other quality markers of the source text, while the time to perform the service stays the same. In fact, machine-translated and post-edited texts can rank higher in terms of quality and consistency than texts translated by humans only as the latter’s quality of work can fluctuate wildly depending on the agency and ever-changing service providers.
We hope we’ve shed some light on the complex and often misunderstood nature of Machine Translation and its post-editing process. If you still have questions, however, please get in touch and we’ll be more than happy to answer.