top of page

The rise of machine translation

The translation industry is booming. As the need for translation grows, technology steps in to fill the need. Increasingly, machine translation presents itself as the future of translation. But is it really?

The global translation boom

First, let’s look at the translation industry‘s strong growth. In 2016 the global translation industry reached an estimate of $40 billion. By 2020, the industry is likely to cap $45 billion. The industry’s annual growth rate through 2018 sits robustly between 6.5-7.5%.

Clearly, global market practices have accelerated the need for large amounts of content to be translated into a multitude of languages. As demand rises, the market has been searching for an instrument to mass produce translation-related services.

Translator machines to the rescue?

In response, the language technology industry has grown and is estimated at $33 billion. The necessity for fast, affordable translations has sparked the development of automation technology, with the objective of creating an automatic language translation machine.

There are currently different types of translation machines: rule-based machine translation (RBMT) and statistical machine translation (SMT). RBMT is the more expensive of the two since it requires laborious human intervention, rule-setting and an analysis to be performed by a competent linguist.

On the other hand, SMT is a more simple translation process and uses statistical data to generate translations.

One of the most famous SMT products is , of course, Google Translate. Launched in 2005, Google Translate is a cloud-based SMT technology. There are currently 104 languages available, and users are generally non-professionals who have no knowledge of the source language.

In November 2016, Google launched Neural Machine Translation (GNMT). The technology uses an artificial neural network in an effort to raise accuracy and fluency levels. Generally speaking, GNMT does mark an improvement for machine-translated texts in terms of accuracy and fluency.

Is machine translation the future of the translation industry?

But even with the improvements that GNMT, the technology still is not adequate for all types of content. For example, GNMT is not suited to highly technical or legal translations. The GNMT system also requires massive input that totals more than three times more data than SMT.

In many ways, some form of automated translation is likely here to stay. However, concerns over choosing an approach to translation based on price alone would invariably impact accuracy and translation quality.

Another concern that is perhaps less discussed is the ethical issues raised by machines that translate. As we look to the future of the rapidly growing translation industry, the ethics of machines are critical to address.

Stay tuned for an upcoming article on the ethical issues surrounding machine translation!

This article is based on an academic study by Kaori Myatt, titled “ The ethics of machine translation and crowdsourcing.”


bottom of page