Introducing your new colleague: MT from TextShuttle

A quick glance at industry headlines confirms what we already know: machine translation (MT) is here to stay. And while post-editing will remain an important component of our work as language professionals, increasingly high-quality MT is opening up more creative and exciting ways of working. Supertext is now testing a new approach with the help of our partners at TextShuttle.

Who is TextShuttle?

TextShuttle is a Swiss company that offers tailored neural machine translation solutions. Tired of editing MT output to match client style guides or glossaries, over and over again? TextShuttle’s custom engines integrate client translation memories, termbases and style guides, and can be adapted to specific language regions. Say goodbye to tweaking British English output to suit a US target market!

At Supertext, we’ve put TextShuttle’s system through an in-house evaluation process, tailoring it even more closely to our clients’ needs. TextShuttle’s own case studies suggest that this could result in time savings of more than 40% in the future. What’s more, the TextShuttle system transmits data in encrypted form and does not store the texts it processes, avoiding the privacy concerns raised by generic MT systems.

What does this mean for you?

While we’ll still be carrying out standard post-editing jobs for our clients, investing in TextShuttle’s system gives us the opportunity to explore innovative new ways of working with MT. So, in the future, we’ll also be providing pre-filled MT output in Trados for some normal general and specialist translation jobs. You – the language experts – can then decide for yourselves what use to make of this output. Incorporate the good bits directly into your target text, use the terminology as a reference or simply delete it all and start over – it’s up to you.

This new approach keeps you in control of your own work while still improving efficiency. You can use MT content to inform and inspire your own translation, just as you would use a translation memory. The difference is that the MT output is already pre-filled in each segment in Trados, so you don’t have to search for it. And, importantly, these jobs will be paid at our standard rates for general and specialist translation for the time being, not at post-editing rates. They’ll also undergo the usual bilingual review by a human proofreader.

We’ll begin testing this approach with selected clients in the next few weeks. We hope you’re as excited as we are to discover all the benefits of maximum technological efficiency combined with maximum human control!

Cover image via Pexels

Related posts

Leave a Reply

Your email address will not be published. Required fields are marked *