How to Combine (Free) AI and (Paid) Human Translation Forces
Translation technology in the form of Computer-Assisted Translation (CAT) tools is improving every day. Most people have taken advantage of free or cost-effective online translation tools like Google Translate or Babel Fish to stock up on useful holiday phrases or to speed up their work as a professional translator or transcreator using Memsource and DeepL.
CAT tools, however, deliver a raw-machine translation which is often stilted. It’s not always correct in its vocabulary, colloquialisms or syntax. It doesn’t quite have the human touch so isn’t pleasant to read – even for search engines! And it certainly won’t help your Google rankings. At greatcontent we show you how to combine AI with human knowledge and experience, saving time and money to further develop your marketing strategy.
What can AI do?
AI (Artificial Intelligence) is the use of machines to simulate human intelligence. At the moment, AI is commonly used in:
- Expert systems where it simulates judgement
- Machine vision where it gives computers the ability to see
- Natural language processing (NLP) where computers can understand spoken language
To do this, AI programming involves the replication of three cognitive skills – learning, reasoning and self-correction. Data is gathered (learning) and turned into algorithms (rules). The computer selects the right algorithm (reasoning) to complete a task and achieve the desired outcome (self-correction). Algorithms are constantly fine-tuned for greater accuracy. All these processes, of course, are carried out at a much faster speed than human cognition could ever achieve. Prominent users of AI include:
- Google AI:which is developing artificial intelligence tools to improve the quality of life
- Amazonan early adopter of AI. Its customer recommendation engine now drives 35% of its sales!
- DeepL developed from a machine learning system that replicates human language, its online translation tools are among the most successful in the world
CAT tools use advanced AI to translate written or spoken words along (generally) with their meaning and (sometimes) the sentiment behind the meaning. Now everything from basic product descriptions to highly specialised and technical texts can be translated using CAT tools.
If AI is so good at translating, does the world of content creation still need translators, transcreators and post-editors? Most definitely. Machine translating is comparatively new and doesn’t always get it right – especially when it comes to colloquialisms, idioms, syntax and specialised professional vocabulary and phrases. Also, CATs don’t really do empathy, irony or subtle cultural differences. Sometimes, they simply get the context wrong.
How to combine AI and human linguists
Start with the best CAT tools …
greatcontent successfully combines artificial intelligence with an experienced and knowledgeable post-editing workforce. When deciding on a CAT tool, look for easy-to-use interfaces and special features such as translation memories, terminology management and context consistency. Good examples include:
- Memsource: with its simple interface this cloud-based CAT is ideal for travel use as well as more complex projects
- SDL Trados Studio: one of the earliest CAT tools, Trados software is popular with professional translators
- CafeTran Espresso: supported across various platforms including Windows and Mac, this is used by both freelance and commercial translators
… then add in the skills of experienced post-editors
Once a machine translation has been made the skills of a human workforce are needed. While traditional translators are experts at translating the spoken word or whole written texts, bilingual post-editors have the skills needed to turn the segments of machine-translated text into a coherent whole. They deliver exactly the context, style and tone the client wants.
This is only testing quotations. SInce Prismic doesn't support it as an HTML tag we are using preformatted text to write quotations.
Skilled linguists working with CATs will:
- Fine-tune: making translations smooth to read and easy to understand
- Add context: internalising meaning and empathising with the text – something machines can’t do (yet!)
- Add colloquial fluency: combining creativity and language knowledge to make sure the target text is fluent and natural
- Correct and standardise: writing dates, numbers, units of measurement etc. according to the brief’s instructions
Bringing it all together
For AI to be a successful part of a marketing strategy, there must be an efficient and cost-effective match between machine translating and linguists. At greatcontent, we have a large pool of multilingual post-editors skilled in working with CAT tools. We develop their skills further by offering regular tutorials covering transcreation and post-editing skills. If you think greatcontent is a good match for your marketing strategy contact us now.
Text: Leonie Yeates
Image: James Pond on unsplash.com
Les 5 bonnes raisons de faire relire et corriger son site internet
Avez-vous déjà parcouru sur un site web un texte criblé de fautes d'orthographe, à la grammaire plus ...
5 février, 2020