The translation process can be time consuming, so it’s no wonder that companies around the world are turning to machine translation. It’s a cheap way to translate content and can speed up the localization process.

But is machine translation a good option for your business? 

This article looks at the developments of machine translation and some of the main benefits and challenges involved.

The progress of machine translation

Companies like Amazon, Google, Facebook and Microsoft have built powerful translation engines in the last decade. 

The quality of translations has improved in that time period, due to advances in AI technology and the amount of data available for machine learning. 

There’s also been progress to overcome some of the most common challenges.

A great example is Google’s work to remove gender bias from its translations. The machine was automatically replicating existing gender biases, like using masculine terms for words like “doctor” and feminine for words like “nurse”.

It now reviews its translations from gender-neutral source languages to check if it’s produced a gender-specific translation. If it has, the translation engine provides both a feminine and masculine translation for users. 

For example, if you type “o bir doktor” in Turkish, Google Translate will produce both gender-specific translations: “she is a doctor” and “he is a doctor”.

How do human and machine translations compare?

Despite the improvements to machine translations, they are still far from being fluent. 

There are certain aspects of language that machine translations can’t replicate. And, while advances in machine learning might lead to accurate, high-quality translations in the future, it’s not the case right now.

Human translation is still a vital part of localization

You need to be able to connect with your target users and communicate with them in a way that feels easy and familiar. To do that, your translator needs to understand the culture and history of your target market. 

Machine translation remains a long way off understanding things like etiquette, idioms and humor. If you keep these aspects of language in your text, a machine translation will likely produce results that confuse or alienate users. Remove them from your language altogether and your messaging can sound robotic and sterile.

Simply put, there’s no substitute for using a human translator if you want your content to be accurately localized. 

Using machine translations at the development stage

While we don’t recommend relying on machines to translate your content, they can still be a useful tool.

Machine translations can help developers estimate the amount of space the translated content needs. 

This means you can avoid common issues related to text expansion or contraction and your features won’t just be left blank while you wait for the professional translation.

Text can expand or contract for different reasons: grammar, word length and sentence structure are all contributing factors. 

Some languages will expand or contract depending on the subject matter. 

Having a rough estimate of how your translated text will look early on can prevent the finished content from being too cramped or sparse.

How different languages expand and contract

If you translate from English to Korean, the text should take up around 10% less space because the script is more compact. An English to German translation should take up more space, because German typically has long, compound words. For example, “food intolerance” becomes “nahrungsmittelunverträglichkeit”.

You need to think about how you can make the text as easy to read as possible. While Chinese, Japanese and Korean typically contract, the characters are more complex, so your developers will need to make sure there’s enough space between each character.

Using machine translations for your business

Machine translation is a useful tool to imitate the human translation process. It will give your developers a better idea of how your localized product will look, so they don’t need to work with a blank canvas.

However, it’s crucial to use human translation to enter a new market with the best possible version of your product. An experienced human translator will be familiar with specific regions or dialects, and can address any cultural nuances that are relevant to your text. 

A bad translation can be a major red flag for new users, so don’t be tempted to take shortcuts. We always recommend using human translators to create accurate content and a professional, engaging experience for your users.