As a concept, machine translation still feels quite futuristic—the idea of machines having the capacity to understand and manipulate language. In reality, it’s been around for a long time – half a century, in fact – which has given this technology time to get really rather clever.
In its simplest form, machine translation is the process of automating the translation process from one language to another with the help of artificial intelligence (for more details, read our blog What is machine translation?). Alone, this is still an imperfect science. But when teamed with human intelligence? That’s where you truly see the scope and benefits of this practice.
Trending: technical content
One such area of opportunity is technical content—which is naturally on the rise as we become a more technical species. The employment growth for technical writers alone in the US is expected to increase by 12% by 2030.
From electric and smart cars that require detailed manuals, to a surge of interest in health and medicine following the pandemic, the public is savvier than ever when it comes to technical language and jargon. What’s more, the sheer volume of content that we are able (and often demand) to access as consumers is massive. With a single car manual able to clock up tens of thousands of words, it’s no wonder brands and businesses are looking to machine translation to support this global content boom.
The benefits of machine translation in the technical field
The benefits of machine translation mostly boil down to speed, which, in turn, improves cost and ROI. Given the high-volume, high-wordcount nature of a lot of technical content, it’s becoming an increasingly popular option for those in the technical fields.
But it’s not just speed that comes in handy. Machine translation often works hand in hand with Computer-Assisted Translation (CAT) tools, Translation Memories (databases of past translations that a company leverages to reduce the workload of new content) and glossaries of technical terms to improve the accuracy and consistency of translations. One can also use customised models that are trained with both monolingual and bilingual data sets. These are managed separately from the baseline engines offered by the tech companies and, though they tend to give a higher level of accuracy, data protection and privacy, they also go hand-in-hand with a higher price tag.
It’s tech advancements like these that have significantly reduced the likelihood of errors when using machine translation, meaning less time spent in the editing and proofing stages. Naturally, this leads to an enhanced global user experience as materials are available at scale and in the end user’s language—no matter where they are in the world.
A human touch
As mentioned, however, machine translation is not perfect. Only when combined with human post-editors can it achieve total accuracy, as well as impeccable style, tone of voice and level of cultural nuance. The need for accuracy is especially important for a lot of technical content, as it frequently deals with issues of health and safety, as well as legal and liability issues. If these sorts of materials are incorrectly translated, you could be in for more than a slap on the wrist—it could cause significant cost and irreversible damage to your brand.
What’s more, you need human intelligence involved at the very beginning of your project – way before translation is involved – in order to check that your project is even right for machine translation. At Comtec, for example, we don’t recommend machine translation for professional video localisation or high-quality, user-facing content such as advertising, brochures and brand literature. This kind of content requires a more creative translation process, or a process called transcreation—an advanced form of localisation that combines translation skills and creative licence to preserve the core meaning of the source text for its new region.
It’s also important that the post-editors you are using have a profound knowledge of the subject matter in order to translate texts in the technical field which may not extend to MT’s capabilities. This helps to avoid the risk of mistranslations and overly literal translations.
The only way to know for sure which method and which linguist is right for your project is to work with an expert.
Where tech content happily meets MT
It’s extremely important to work out where translation fits in your process and which kind of service you need based on your requirements, budget and content, whether that’s machine translation, human translation or transcreation.
If you’re producing:
- User or instruction manuals
- Standardised health and safety content
- Technical training or elearning content
- Engineering or product-related content
- Parts manuals
- Standardised software strings
- Patent documents
- Electric vehicles
- VR (virtual reality) or AR (augmented reality) instruction manuals
you may want to consider machine translation to support you.
Research has shown that technical fields have an especially high post-editing score (aka fewer mistakes prior to the human linguist’s work) at around 80%. Comparing it to travel and hospitality or business and education content, for example, both scored at around 60% (all based on an English to Arabic language pair), the benefits of adopting MT as an effective and convenient automation step is very clear.
It’s also interesting to note that some language pairs work better than others when using machine translation. English and Simplified Chinese, for example, are a tricky match, with an accuracy score of only 54%, whereas Spanish and Portuguese have an impressive accuracy score of 81 and 79% respectively (both examples are based on translations into English).
Preparing content for machine translation – the golden rules
To save you considerable time (and headaches!) down the line, you should follow these golden language rules before creating content for machine translation:
- Sentence length and structure should be short and concise. The longer and more complex your sentences, the more likely you are to find errors when using MT.
- Avoid using the passive voice. An active voice is easier to translate.
- Tempting though it might be (especially if you are a writer!), avoid using unusual words, dialectal variations, abbreviations or humour. Your machine won’t get the punchline…unfortunately!
Reviewing your source content to make sure that it is machine translation friendly is a key part of the process.
An informed localisation strategy in tandem with an expert partner
At Comtec, we specialise in all of the above, pairing translation technology with our professional human translators to get accurate results that speak to your customers—in your language, in your tone of voice. You can join the ranks of happy customers like Quadient Industries, who rely on us for technical content translation.
We’re a dab-hand at localisation, understanding which markets will help your growth strategies and how to go about scaling into them. We work closely with many of our clients from the initial stages to inform strategic planning, content strategy, implementation and process guidance, and generally freeing up your resources!
For further info on how we can help you with your translation queries or requirements, please give our friendly team a call on +44 (0) 1926 335 681 or email info@comtectranslations.com. We’re always happy to talk translation!