Human Translation vs. Machine Translation

neural-network.jpg

Machine translation (MT) tools like Google TranslateBing Translator and even our own Lexigo.EMT (Enterprise Machine Translation) are on the rise ,  and they’re getting smarter — but when is it suitable to use machine translation as opposed to a professional human translation?

To answer, we need to look at the disadvantages and advantages of both solutions by understanding how everything really works.

It’s also important to note that Machine Translation that exists in professional organisations like Lexigo is built and operated purely for clients, this should arguably provide more accurate and specific, tailored results for clients in comparison to free tools. This adds yet another dimension to the long-standing debate  —  for simplicity, let’s compare Google Translate with Human Translation.


How machine translation like Google’s, work

Google Translate works on something called Neural Machine Translation which basically learns from loads of data over time, attempts to find patterns and continually improves to offer you the best combination of translation in available language pairs.

Advantages

It’s free: Google Translate offers its services at no-cost. That’s a business advantage everyone can appreciate but there is a minimal cost if you’ll be using Google Translate through an API.

It’s instant & online: Google Translate is conveniently available online for cross-platform use. Translations are instantaneous and for immediate use.

Get the gist: If you’ve ever used Google Translate, you probably have quickly determined its limitations. It’s good at getting the ‘gist’ of something so you can understand the basics. You could easily determine what a site’s subject is about or the primary points of an email.

Multiple languages: Google Translate is available in a large number of languages already, with more on the way.

Disadvantages

It’s limited: You can’t get everything translated with the service. And you won’t get a comprehensive translation of the text, either.

Not every language is offered: While Google’s translation gives plenty of languages, there are almost 7000 languages in the world and at least 300 of those are important business languages.

It won’t give natural, fluent translation: Google Translate gives a mechanical and direct translation of words. You won’t get text that reads naturally across those languages, although the neural foundation to its technology aims to solve this problem.

Context is not part of the equation: Google Translate won’t know the context of your content. For example, who is your target audience, what are their ages, education level and cultural sensitivities? In addition, what is the content being used for? eg. for a speech, marketing copy, website, white paper, instructional content or something else? What is your brand’s voice and what terms does your brand prefer to use (eg. automobile vs. car).

It’s only a machine after all: Translating keywords for SEO and SEM or translating a brand name requires lots of research, study and further localisation  —  all things machine translation can’t take in to account.


So, what’s Google Translate good for?

Google Translate is suitable for all of the following purposes:

Determining a document’s native language: Google can automatically detect a language, so you don’t have to spend time figuring out its native tongue.

Understanding the primary points of a website or section of text: You can use the translation service to get a vague understanding of what a webpage or other correspondence entails. Similarly, you can translate your message into a different language if you simply need to transmit something quickly.

Informal or casual communication: The service makes it easy to quickly get a message to or from different languages. This type of quick and simple translation is great for informal or casual correspondence. Anything official within a company would risk miscommunication or other related issues.

Nothing official: Related to the above points, Google Translate should never be the chosen method to communicate anything official within company business. It’s great for casual and informal messaging — especially if the receiving party already knows you’re using it. That way any unintentional errors won’t be a problem.


How professional translation works

Limitations in automated translation services showcase the need for active professionals to conduct your translation work. I am personally very excited about where machine translation is headed and don’t believe it will have to be an epic battle between human and machine in the future.

That’s why at Lexigo our approach is to combine both and I take the view of human and machine rather than human vs. machine  —  this is why technology plays an extremely important factor in our human translation process. Machines are excellent at remembering things humans don’t need to or don’t have the capacity to. For example, how can a translator remember every brand term, glossary of terms and styles that they need to apply to specific clients. They also can’t quickly scan thousands of words and find repeated content or previously translated content.

This ultimate combination of human and machine makes for high-quality translation results suitable for:

Any official communication: Letters, Messages, Emails, Etc. When your company name is on correspondence in an official capacity, you need it translated by a professional. Machine translation alone could inadvertently make your company look silly, convey an important point inaccurately or miss critical aspects of your message. Why would you want to risk this with anything official? It’s your name and reputation at stake, after all.

Localising marketing materials: Language doesn’t always reduce to the nuts-and-bolts provided by Google Translate. In most cases, context is extremely important. Localising text to specific cultures or languages simply cannot be done with current automated technology. This is precisely an area where a professional human translator will provide the best results.

Corporate and other market-facing content: Market-facing content already looks and reads great in your native language. Shouldn’t it look and read in the same effective way no matter what language it’s in?

Multi-language services: In many corporate marketing or communications, multiple languages are an element of your campaigns. Each language will carry specific requirements in terms of conveying your brand  —  and message  —  without losing anything in translation.


Conclusion

There is a time and a place for Google Translate which features world-class artificial neural networks. In the same breath, there’s also a time and a place for more thorough human translation services. Once you understand the benefits and limitations machine translation provides, you can better determine your own specific translation and localisation (or localization!) needs.

I’d love to hear your experiences, good or bad, with machine translation and professional translation  —  leave a comment below.


Written by Mark, founder and CEO of Lexigo: With his passion for technology, business and globalization – Mark is responsible for realising Lexigo's vision and strategic development of Lexigo technologies, with a particular interest in translation process automation, machine learning and multicultural marketing. His experience in a diverse range of fields gives Lexigo a dynamic and fresh approach to the delivery of its products and services.