OWhat comes to the mind of the average person upon the mention of
machine translation (MT) is probably Google translate, but translation
enthusiasts know that MT covers way more. 

Everything from the different types
of MT, its end-use applications, and the role of Artificial
Intelligence, makes it apparent just how extensive Machine Translation is.

The emergence of the pandemic brought along interruptions in
every sphere of life, with businesses requiring innovative ways of conducting
their operations.

This blog post is an overview of the present state of MT and how
it is of relevance to translation companies and clients alike.

Overview

Although the pandemic may have caused disruptions in the
translation industry, one segment thriving is the machine translation
category. 

The sector, which had a market
size of $622.5 million in 2020, has projections to grow to $1.5 billion by
2027 with a CAGR of 13.9 percent during this period, with the U.S
market acting as a principal center on this front, accounting for a whopping
30% ($183.4
million) of the entire market share.

Principal drivers for the growth amidst the pandemic are; the
need to translate
bulk health-related materials into more languages, the need for
faster translations coupled with the necessity for saving costs at this
critical moment. 

If you want to know more about machine translation, check out
one of our previous posts.

Trends in Machine Translation

1. NMT going
   mainstream

As the
demand for machine translation continues to grow, so does the technology.
Neural Machine Technology (NMT), which made its debut just a few years ago, is already
causing a paradigm shift in the industry.

Many
industry players are turning to NMT due to its superiority in the accuracy of
most translations.

Unlike
statistical translation that adopts a rule-based (phrase, word, syntax)
approach to predict and translate text, which has a fair share of errors, NMT
sequentially utilizes deep learning to create translations that are much closer
to human levels of accuracy.

It is
worth pointing out that neural machine translation still struggles with proper
names, rare terms, and contextualized translation formats.

• Machine Translation in
  Video

Video
content consumption is on the rise. Besides the wildfire called social media,
the increasing adoption of virtual events and
e-learning due to the pandemic plays a pivotal role in the surge of video
content consumption being witnessed. 

These
effects are also having a snowball impact on machine translation. Many video
content providers are now combining automatic speech recognition with machine
translation in order to translate the transcribed source text to the desired
target language, thereby providing captions and subtitles to deliver a better
viewing experience to international audiences.

• Combination of NMT with
  MTPE in healthcare on the rise

With
the need to complete clinical trials at the fastest possible duration the
primary priority of pharma companies during the pandemic, translation companies
are exploring the feasibility of combining neural machine translation alongside
post-editing when translating clinical
trial documents to keep up with
time requirements.

It
would still involve the extensive training of machine models so that errors
during translation remain minimal.

• Newer MT models bursting
  forth

As
neural machine translation continues to take center stage, researchers are
coming up with innovative models in an attempt to disrupt the space. A model of
worthy mention is work done
by a group of researchers to create a multilingual NMT model to translate
biomedical data from five languages to English. This model, which was intended
to help with large-scale multilingual analysis as it concerns the Covid-19
pandemic, was trained on over 350 million sentences to produce state-of-the-art
results.

Another
notable model used an attention
mechanism to improve the performance of the NMT model.

Importance
of MT in the prevailing translation industry

There
might have never been a better time for both translation companies and clients
to look for creative ways to cut down operating costs, and the use of MT
provides a viable solution. Besides cost savings, MT grants translation
companies performance and efficiency upgrades by delivering closer to
human-levels of accuracy when compared to the previous (statistical machine
translation). 

Conclusion

Machine
translation adoption in the translation sector is clearly on the incline, and
the data is there to back it up. Correspondingly, the segment is also
witnessing a tilt from rule-based machine translation to neural machine
translation, which has proved over time to be more accurate in its end
result. 

Although
much work still needs to be done to improve its performance on context-related
translation.

Translation
companies and customers alike, to save cost, must see the option of machine
translation as a good thing and as an opportunity to adapt their workflow to a
more efficient way of carrying out their operations.

Translationsinlondon is
your one-stop-shop for all your translation and localization projects. Our
process leverages cutting-edge technologies and tools to create accurate,
quick, and efficient translations for your business.