Machine vs Human Translation. Which Is Right for You?

6 Minute Read

Human Translation vs. Machine Translation

Every year there is a new report on the improvements in machine translation and its narrowing gap to human parity.

As with all machine learning applications today, there is much excitement about the possibilities to make our lives easier and in the case of MT, our content faster and cheaper to translate.

But like many automated systems, machine translation is not a silver bullet. Varying degrees of human translation expertise is still needed, and this will not change anytime soon.

Let’s set aside the hype and focus on what’s important to you – finding the right balance of price, speed, and quality for your content.

About Machine Translation

Machine Translation, also known as MT, has come a long way since its origins more than 60 years ago. Statistical and Rule-Based Machine Translation were the norm for many years, using corpora and grammatical rules to determine the best fit for a translation. However, the output was frequently clunky, and the systems could not learn on the go.

Neural MT, a relative newcomer, has set out to resolve these issues. It can “learn on the fly,” reducing the size of the data needed and providing more natural-sounding translations. Google Translate is by far the most popular neural MT, used by millions every day.

Machine translation is often used when “getting the gist” is the primary objective. Typical uses are in eCommerce sites, automated chat, and the free, instantaneous results provided by google translate. MT is favored in these examples since it can handle enormous amounts of text at instant speed. 

In fact, in 2018, Google stated that 143 billion words are translated every single day on google across 100 languages —more than all of the world’s human translators could handle in a year.

Human Translation vs. Machine Translation

Speed and volume are the main advantages of MT. However, due to its limitations in precision, style, and cultural fluency, its applications are still limited.

In comparison, human translators can use judgment and experience to understand the style, context, and tone of a text, even if the source language has errors or typos. 

A translator that works in their native tongue will have the cultural fluency necessary to pick up on nuances and ensure that the foreign audience clearly understands the content as intended.

The Translation Services Continuum

While Machine Translation and Human Translation are quite different translation processes, both, or a hybrid of the two can be used in a single project or campaign.

Many Translation companies today offer both MT and human translation services. With Computer-Assisted Translation (CAT) tools, translators can get the best of both worlds, leveraging automation and human expertise to produce high-quality translations consistently.

Listed below are five main types of translation. Their uses differ according to the desired speed, quality, and resonance of the translated content.

  1. Standard Machine Translation: Machine translation with no human review.
  2. Machine plus Human Translation: A model in continuous improvement that attempts to combine MT speed with human expertise. While MT post-editing is often not practical for quality sensitive material, adaptive MT is showing promise.
  3. The single translator approach: Done by one translator, often using translation memories and CAT tools to leverage past translations as well as guarantee correct and consistent terminology, voice, and tone.
  4. Translation plus editing: Done by two or more translators to provide the highest quality possible. It also leverages CAT tools and translation memories.
  5. Transcreation: Also known as creative or marketing translation and often used in global marketing campaigns. Transcreation is similar to copywriting and requires the highest amount of cultural fluency. Since a word for word translation would often be too clunky or irrelevant to foreign cultures, new phrases need to be created to ensure conceptual consistency and resonate with international audiences.

Which Is Best for You?

The best application for you will largely depend on what type of content you need to translate and how important quality, precision, conceptual clarity, and brand resonance is for your target audience. 

If the objective is for your foreign readers to get the gist of the text, like whether or not an eCommerce review is good or bad, machine translation could be a good fit.

On the other hand, if you are translating quality-sensitive material such as medical documents or IFUs for highly technical products, you’ll need a level of quality that can only be guaranteed by human translators and editors.

And for marketing material, only a human translator will possess the cultural fluency needed to guarantee your collateral will resonate with foreign target audiences.

When consulting with your language service provider, be clear about what your objectives are for your translations to find the right balance between speed, quality, and resonance.

More Articles