For many people, their first exposure to Machine Translation (MT) came through Google’s automated engine. For years, Google Translate has been at the forefront of free, widely available machine translation technology. It also has a reputation. Some will (perhaps justifiably) say the results are never good enough for publication. Others, after comparing throughputs from different engines, will conclude that Google’s is among the most efficient machine translation engine. In the world of machine translation, however, efficiency does not equate to publishable quality. If you have content requiring translation and are wondering whether machine translation is suitable, this blog will help you discover which approach to machine translation would best fit your needs.
FAUT (Fully Automatic Useable Translation)
Machine translation developers and their customers dream of a world where FAUT is the norm – a technology so powerful it can produce good enough translations without any human involvement. Unfortunately for them, the number of cases where this strategy can be implemented is extremely limited. FAUT might be suitable if the text you need translated is short and basic (e.g., simple sentences, place names, etc.), or if the quality of the target language does not really matter as long as the translation gets “the gist” across (e.g., hotel booking websites with numerous customer-generated reviews, many of which are poorly written). If accuracy is important, however, you should not leave it to machines alone.
Post-edited Machine Translation
SimulTrans can take your content, run it through Google’s Neural MT engine, and have it post-edited by a professional translator. Once this is done, the text is also reviewed by a second translator. This way, you make the most out of MT’s capabilities (productivity gains, better consistency, etc.), and can rest assured that your content has been properly translated. The end result is perfectly publishable translated text – accurate, grammatical, and fluent – all for a considerably lower price. Suitable types of content for this approach include user interfaces, user manuals, technical guides, websites – generally straightforward content.
For more information on machine translation quality and post-editing criteria, check out our blog about Quality in the Machine Translation Workflow!
Translation and Transcreation
On the other hand, if your content is not as straightforward (marketing materials, blog articles, or white papers), you may want to pursue a human-only translation option, whereby translators will handle the job without any influence from machine translation, which might otherwise incite them to use phrases that are too literal for these types of content. When fluency and quality are of utmost importance, human translation is your safest option.
Finally, if your content requires transcreation (adaptation of underlying concepts described by the text rather than just adaptation of the text itself), simply take machine translation out of the equation. Machine translation does not really allow the level of freedom and creativity needed for transcreation. Transcreation ensures that your brand, slogan, catchphrase, and other advertising content are perfectly suited for the target audience and will have maximum impact in the target country. This can only be achieved by a professional human translator and a transcreation approach.
Still unsure which translation approach is best suited to your needs? It’s possible that more than one approach may be required, depending on your content for translation. For example, machine translation and post-editing may be best for user manuals, while white papers and product brochures may be best translated using (human) marketing translators. Likewise, transcreation may be best for advertisements you plan on using in multiple countries, in order to best leverage the idiosyncrasies of each culture. If you have content requiring translation, click below to receive a machine translation suitability report, courtesy of SimulTrans.