Machine-translation-01We have traditionally thought of translation as an intricate process performed by human hands. We think of a linguist carefully working line by line, with another coming behind them to review and edit. Depending on the scope of the project, this can be a long process – though fruitful in the end.

Then along came Machine Translation (MT). MT has existed in some form or other since the 1950s, but recent technological developments have helped it become a viable option in many cases.

The promise of machine translation is that it saves money and time, which makes it appealing to many organizations. Reality is a bit more nuanced.

Yes, most organizations can benefit from MT, but they must understand when it is appropriate and when it is not.

We will address this issue in this blog. First, however, we’ll provide you with a quick primer on Machine Translation and the ways in which it is often used.

What is Machine Translation?

It translation, you have the original source language and the target language, which is the language into which the text is to be translated. Simply put, Machine Translation is an automated process where computer software translates text from the source language to the target language without human involvement. The source and target languages are compared and matched against each other by a translation engine.

CASE STUDY: How Rocket Mortgage Uses LanguageLine to Expand Its Audience

As mentioned above, MT has a history going back nearly 70 years, but recent advances in artificial intelligence and natural language processing have brought it into the mainstream.

There are three basic types of Machine Translation:

  • Rules-based: Uses grammar and language rules developed by language experts and dictionaries that can be customized to a specific topic or industry.
  • Statistical: Learns how to translate by analyzing large amounts of human translations. Does not rely on linguistic rules or words.
  • Neural: Teaches itself how to translate by using a large neural network. This method has transformed MT as it provides better results for most language pairs.

Machine Translation can be used alone or in combination with human post-editing, also known as MTPE.

Benefits of Machine Translation

Machine Translation can be an important tool in the translation process. It has three primary benefits:

  • Speed: Machine Translation is capable of translating large bodies of text in seconds. Its ability to interface with content management systems (CMS) further enables fast translations.
  • Memory: Machine Translation uses AI to get smarter the more content is translated. It has the ability to memorize key terms and reuse them when appropriate.
  • Cost: Because it requires less human involvement, Machine Translation has the potential to significantly lower costs.

Considerations for Machine Translation

Machine Translation is a major breakthrough, but it is not appropriate for every translation project. Understanding the nature of their content gives our clients the ability to choose whether or not MT is appropriate for them.

Machine Translation really comes with two main drawbacks: liability and lack of consistency.

From a liability perspective, MT may inadvertently create a meaning change in text. This can create legal exposure for the client.

As far as creativity, MT follows a literal set of rules when translating text. It does not see or inject nuance; therefore, it is not bound to fully comprehend humor or other creative copy.

LEARN MORE: LanguageLine’s Complete Suite of Machine Translation Tools

Machine Translation is a great solution for many needs, but it shouldn’t be considered a replacement for human translation. Traditional human translation is still the best option in many instances.

A general rule of thumb is that Machine Translation becomes a more viable solution when the need for translation quality is low and the need for a quick turnaround is high. For example:

Machine Translation alone is acceptable when:

  • Translation quality: low
  • Examples: help tickets, quick-turn emails, chat sessions, customer ratings and comments, non-clinical patient intake notes.

Machine Translation plus Human Post-Editing is acceptable when:

  • Translation quality: medium to high
  • Examples: health plan correspondence, patient letters, static corporate websites

Human Translation is preferred when:

  • Translation quality: high (due to high levels of creativity or potential liability)
  • Examples: marketing materials, human resources and training manuals, product manuals, training documentation, patient education materials, manuals and brochures

Human Translation is a must when:

  • People could die or be injured if the translation is wrong
  • There could be legal or monetary implications if the translation is wrong
  • It would negatively impact a business’ image or reputation if the translation is wrong

LanguageLine can help

LanguageLine offers traditional Human Translation for projects that are client-facing, highly creative, or contain some level of liability for the client.

We also pride ourselves on providing the industry’s most robust suite of Machine Translation tools. Our MT Hub offers 40 different MT engines and uses best-fit technology. This allows us to select the best engine depending on the type of content and language pairing.

We also offer Machine Translation with Human Post Editing, in which our MT Hub translates the content, and then a human linguist reviews and polishes it.

We are most happy to consult on which translation solution is best for you. Please contact your LanguageLine representative, reach out to us through our website, or call 800-878-8523.

2018-Guide-to-Translation-and-Localization