Machine Translation: How to Implement it Successfully

Machine Translation: How to Implement it Successfully

Machine Translation: How to Implement it Successfully 3968 2976 Sergio Guillén

How to Get the Best Results from Machine Translation

Translation is becoming increasingly automated. Machine translation (MT) allows for startups of practically any size to easily translate documents. But MT isn’t as simple as copying and pasting a document into a translation window—the human touch is needed to deliver the best results.

What Content is MT Friendly?

Not every document can be translated successfully by machine. Generally, it works best on information-based content, such as guides and instructions. Simple, direct language is the most easily translated.

Translation software often struggles with content which is creative or literary. Metaphors, culture-specific references, and other layered messaging will often be difficult for machines to handle.

Does the Provider Offer Suitable Confidentiality?

What happens to the data you input into an automated translator? You’ll want to check the service’s privacy policy before sending over any documents.

If you’re translating publicly available information, such as the text of an existing website, confidentiality probably isn’t a huge deal. But if you’re translating internal business documentation or other sensitive information, make sure the provider has appropriate confidentiality constraints.

Use Trainable Machine Translation

MT can be trained to produce a desired tone and voice. Ideally, you want to train the system with your own data, which will be previously translated documents.

Don’t worry if you don’t have any existing data. Documents similar in style and content can be inputted to help refine future translations.

Does the Service Speak Your Language?

Translating from one widely-spoken language to another is usually no problem for most machine translators. For example, most of them can handle English, Chinese, French, and Spanish. But you can run into problems with other languages.

Global by Design explains how a lack of language support is a huge limitation of many neural machine translators. For instance, Amazon Translate supports just six languages.

According to a discussion of experts on Linguistics.com, machines have the easiest time translating between languages with a similar syntax, such as English to German.

MT has the most difficulty translating to and from agglutinative languages. Basically, these are languages where complex words are created using smaller words. Examples of these types of languages are Turkish and Japanese.

The Importance of the Human Touch

Recently, Microsoft announced their machine translation achieved human-machine parity. At first read, this seems like a big deal which would put human translators out of business.

But the definition of parity used is a bit shaky. Only individual sentences are scored. So, a human judge would look at a translated sentence in isolation and score its accuracy. Using that specific testing method, humans and machines had results which were statistically the same.

Quite frankly, machines get trounced when larger documents are compared. Paragraphs translated by people are often far superior. Human translators have a more nuanced understanding of the material, which helps retain the original meaning of the translated text.

Lifehacker reviewed the five top translators but found they all fell short compared to a human translator. So, does this mean automated translation is useless beyond simple sentences? Not at all.

Use Post-Editors to Add Professional Polish

No matter what software you use, you’ll probably also need to use a team of post-editors from a translation company.

Post-editing is the process of reviewing and adjusting the output. It’s a different skill than straight-up translation.

You want post-editors who can make quick decisions about whether or not to keep the output. If the post-edit team can’t work relatively quickly, then you’ll lose one of the main advantages of automation, which is speed.

But post-editors also need to be detail oriented. One of their main jobs will be to correct the translation for meaning. The output might be grammatically correct but not make sense culturally.

A translation company can keep your local content up-to-date. Plus, they’re able to manage the machine translation processes. But beyond organization services, they’re also able to provide a cultural perspective to help with localization best practices.

Finding the Best Translation Methods for Your Startup

Translation is more than a simple word swap. Which means machines can’t do it all. But they can speed up the bulk of the work, especially if the text translated is heavily information-based.

Speed is a major benefit of machine translation. Any large-scale translation project will almost certainly use some automation, at least as a starting point.

When expanding a business into a foreign market, being first is a huge advantage. You want potential customers to see your content and brand before they discover your competitors. Automation lets you get the translation process started right away.

But human involvement plays an important role, too. Expert translators and experts in dealing with machine-generated output can both be used to help shape your messaging, so it appeals to your new customer base.

If you’d like to learn more, check out this white paper about how translation technology can help your startup expand internationally.

Sergio Guillén

As the CEO of Accelingo, I believe in a globalized world where businesses flawlessly share their message across different cultures. I strive to empower startups to scale up globally and take over the world by providing compelling multilingual content that preserves their brand’s message and identity while increasing their reputation and visibility among global clients and partners. If you’re an ambitious startup hungry for international success, get in touch with me today and let’s find out how your startup can benefit from my expertise!

All stories by: Sergio Guillén