Machine Translation and Human Translators

March 1, 2020 |

Translation Services Industry Growth

More Corporate Globalization

Trade and the flow of people, information, and capital has increased considerably with time. For companies expanding in international markets, localization is an important part of gaining local community acceptance.

As we’ve mentioned above, an increase in globalization means translating an increasing amount of content into more languages than ever before.

An increase in needs for non-English languages, associated with a global environment in which any company can reach a global foot print, means that demand for language translation services is growing rapidly.

At the same time, however, the shape of the translator market is changing, with machines taking a greater share of the business. While the demand for translation services continues to grow, the role of the human translator will change significantly as his expertise will be enhanced by technology.

This means that, while the demand for translation services continues to grow, the role of the human working as a foreign language translator will change significantly. Human translators will need to work in harmony with machines in the future.

Post-editing Machine Translation

For organizations with consumers who speak different languages, they have no choice but to translate and localize content if they want to adequately serve their customers.

Post-editing of machine translations (PEMT) will be a trend in 2020. Machine translation is an attractive option for many businesses with large translation needs and there is growing evidence to suggest that turning to translation by a machine, then improving the result with a PEMT service is faster (and therefore more cost effective) than translating from scratch, especially for large volumes.

A case study by the University of Groningen (Netherlands) and Dublin City University (Ireland) analyzed three methods of translation of a chapter of the popular novel, Warbreaker:

  • Translation from scratch
  • Post-editing of phrase-based statistical machine translation
  • Post-editing of neural machine translation

The study found that post-editing of phrase-based statistical machine translation produced a 18% greater translation productivity, while post-editing of neural machine translation gave a 36% higher productivity than translating from scratch.

In addition, both machine translation methods saw the translators take fewer breaks during their work (29% for post-editing of phrase-based statistical machine translation and 42% for post-editing of neural machine translation). The breaks they did take, however, lasted longer than they did when translating from scratch, at 14% for phrase-based and 25% neural machine translation.

This study confirms that businesses will try to combine machine translation with post-editing services in 2020. Even though translations produced by machines are far from perfect, technology advances mean that it is often faster to use a machine AND a human.

AI and Translation

Will AI end human translator’s work?

With human translators adjusting their work to satisfy PEMT needs, AI will not replace human translation any time soon. Perhaps it may one day, but we’re nowhere near that point yet.

Here is an example: paste a block of French text into Google Translate and ask it to deliver the copy in English and you’ll see why. Even with the vast resources that Google has at its disposal, the company has been unable to master machine translation in the way that a human translator can. While they are emblazoned in a human brain, humor, cultural references, and semantic subtleties escape the machine translation engine and will do so for a long while.

And that is only using two of the languages that the Google Translate team has been working on for years. What about more rare languages? How many languages are there in the world? There are around 6,500 spoken languages in existence at the present time, however some 2,000 of them have fewer than 1,000 speakers, meaning that the number of languages is declining every year.

A real translator who speaks a lesser used language has almost no reason to fear the impact of machine language translator developments on the translation industry.

The majority of language service providers focus on delivering about 40 of the most commonly used languages. Alpha Omega Translations delivers translations in 210 languages. Whatever your translation needs, the right combination of quality, customer service and price will be essential to your success. While machine translation can become an important technological tool, it is the human translator who can provide the quality of translation and level of customer service that are so important to business translation clients.

Using the Right Process: MT Post Edit or Human Translation

When do I hand off content to a human translator, and when do I let MT handle the task?”

The biggest factor will come down to content priority and quality. Here are some factors to consider:

  • What is the scope of this content?
  • What is the target audience?
  • What need is this content fulfilling?
  • Does this content need to be edited or revised after publication?
  • Are there any legal requirements or strict brand guidelines pertaining to this content?

High priority critical content, for example legal documentation or medical information, will require top quality work thus the expertise of a human translator. Due to the sensitive nature of this content, and the high priority of accuracy, rapid deployment is not always be the best option. One mistake could lead to a cascade of potential technical and legal issues.

Even launching a website, a new landing page for your latest product or service is a perfect example of high priority content requiring quality work: this is where brand voice and style need to shine, and companies should leverage the creative touch of a human translator to reach their goal.

We can go even further and determine whether or not MT is the right choice for each project. This will depend on the complexities involved and your specific needs.

For example:

  • Volume: If you need to translate a large volume of content in as little time as possible, then MT can outpace human translators and bring otherwise massive projects into a shorter time-table.
  • Simplicity: When translating repetitive information, with specific language and terminology, MT can produce the repetitive and simple text.
  • Priority: Low priority content can be processed through MT. For example, content that will be used internally might not require the same level of polish of a website for a new product.

Questions to ask:

Does this content make sense for machine translation?

Do I need the creative input of a professional translator?

Will I be able to correct my course of action, or once I hit send will that email exist, as it is, forever?

At Alpha Omega Translations, we can guide you through the maze of translation and help you find the right solution to your needs. As your organization evolves, let us help you adapt to these changing translation options. Access our free e-books and guides here

As part of the Linguistic Quality Assurance, Alpha Omega automatically saves your translations to a database, known as Translation Memory (TM). We can then leverage that knowledge for future translations, which cuts costs and speeds up terminology harmonization across all your projects.

To discuss how linguistic quality assurance can solve your globalization challenges, call us or click here for a free quote.

Category: Translation Services

Skip to content