Is your business up to date on global trends? If so, you probably know all about machine translation and machine translation post-editing.
In the past few years, machine translation and MTPE have managed to carve out a niche for themselves in the global market. Businesses are becoming more and more aware of the possibilities for growth that these solutions can open up for them.
But not all businesses know how to make best use of these solutions. They’re similar, but each has different strengths and different tradeoffs. In this article, we’ll explore what these differences are, and what are the best use cases for both raw machine translation and MTPE for businesses.
When talking about machine translation, we often refer to raw machine translation, which is the pure output of a machine translation engine. The advantage of raw machine translation is that it’s available instantly; just input the text or document to be translated, and you get the output with minimal waiting time.
This leads to another advantage of raw MT, which is that it’s endlessly scalable. For one, the length of the text is not an issue with machine translation. This is unlike human translation, where the longer the text, the longer the turnaround time due to the limits of human ability to read, translate, and write.
Machine translation is also scalable across different languages. MT systems can handle many different languages, some even reaching over a hundred. It’s just a matter of choosing the right systems for the job, and your text can be made available in many different languages.
Because of this, machine translation shines when used in automatic workflows, particularly those involving multiple languages. Think, for example, text on ecommerce websites. Having an option to translate your webpages will make them more accessible to consumers, which means they are more likely to make a purchase. This is what happened in the case of eBay, where machine translation helped increase sales by over 10%.
Another good example is the case of AirBnB. AirBnB is an app that caters to travelers from all around the world who are looking for lodging. And there is often a linguistic gap that exists between the “hosts” and “guests” who use the app.
Before AirBnB integrated machine translation into their app, hosts would have to create the same listing in different languages, often with the help of Google Translate or some other external MT engine.
What AirBnB did was build a proprietary MT engine that seamlessly detects a user’s preferred language and automatically translates text into that language. This way, hosts no longer need to create multiple versions of the same listing in different languages.
There’s more to learn about AirBnB’s case, which you can read about here:
AirBnB’s own machine translation engine draws increasing returns on performance and consumer experience
Raw machine translation is also a great solution for texts that need to be circulated but aren’t consumer-facing, such as internal memos and the like. Often, these are texts with short shelf-lives, and it’s enough to get the gist of the text to understand the message and act on them. For companies with a global workforce that speaks different languages, machine translation can quickly bridge communication gaps.
Machine translation systems have come a long way in terms of quality, almost uncannily so. But they aren’t perfect quite yet, which is why there are still many use cases where the human touch is required to ensure quality.
Machine translation post-editing sits in the middle of the spectrum between raw machine translation and human translation. As the name implies, raw MT output is given to a human editor, who works on the text with varying levels of intervention.
There’s light post-editing, where the editor simply proofreads the text and fixes unclear or mistranslated sections. Then there’s full post-editing, where the text is given full editorial treatment to sound as close to human translation as possible.
Want a more in-depth look at machine translation post-editing? Check out our articles here:
What is MTPE?: An introductory guide for businesses
Machine translation post-editing: Boon or bane for translators and the industry?
The advantages of MTPE are best seen when compared to human translation. For one, it’s faster. Today’s machine translation systems use sophisticated AI machine learning techniques built on neural networks, which have vastly improved the quality of MT systems.
Before, MT output was the stuff of humor and amusement; to use it for post-editing was simply not feasible. But those days are gone. Now, machine translation output has proven to be a viable solution for faster-paced workflows than are possible with pure human translation.
But since machine translation isn’t perfect yet, there are times when editing is required, particularly for consumer-facing text. This can include articles, whitepapers, more copy-driven webpages, and the like.
These kinds of text demand a higher standard for quality that raw machine translation cannot yet guarantee, but they are often of a middle importance that doesn’t necessarily justify the cost of human translation. And as is often the case, they come in large volumes and need faster turnaround times, which compound the cost.
For these kinds of text, machine translation post-editing provides a quicker, cost-effective solution that mitigates the disadvantages that come with pure human translation. However, this doesn’t mean that MTPE is set to replace human translation. There will always be critical or high-value texts that require the careful eye or creative flair of humans, which no machine can replicate.
Machine translation and MTPE are solutions that aren’t necessarily opposed to each other. It’s not an either-or situation where you can only pick one and, in fact, they can complement each other very well.
Take the case of Zendesk, for example. Zendesk uses MT to translate their help pages en masse, helping people find vital information in their own language and lessening the need for human agents.
But it doesn’t stop there. Zendesk understood that quality remains an issue with machine translation, so they took note of traffic to help pages in different languages to see which ones people needed the most, and spent resources on MTPE for those pages, helping them provide a better consumer experience.
Learn more about Zendesk’s case here:
How Zendesk achieved a targeted, cost-effective localization strategy through machine translation
From this, we can extrapolate that machine translation works well alongside web analytics to determine consumer behaviors, providing vital information about where MTPE could be best used. It’s an innovative way to use both, that many businesses could learn from.
Machine translation and MTPE are unique solutions for the linguistic challenges posed by the global aspirations of any business. We hope you’ve learned enough to make an informed decision about which one you need, and how you can utilize them to reach your goals for growth.
© Copyright 2023 Tomedes All Rights Reserved.