Businesses are same all over the world. People work hard to make their products sell, their companies grow and broaden their futures. But while business may be universally understood, the languages we conduct it in are not. One of the exciting recent developments in technology by Google has the potential to change the face of business as we know it. It is known as Neural Machine Translation (NMT), and it promises to break down language barriers to a degree we have never seen before.
“The Internet created a global economy, but there are still numerous friction points, chief among them a substantial language barrier,” says Denis Gachot, CEO of Systran Software Inc., a language technology company. “NMT is a scalable solution to the language barrier problem that can achieve numerous outcomes – allow businesses to rapidly transmit large volumes of documents in different languages, connect small businesses to the global economy that could not operate without professional translation, and even empower consumers to find products and services they couldn’t have before.”
How is it different from Rule-based and Phrase-based translation models?
Neural Machine Translation is the shift from rule-based and phrase-based translation models, which translates word-by-word or in groups of words between languages. Instead, NMT translates entire sentences at a time, looking to discern cultural, colloquial, and technical contexts to create more accurate translations. The technology mirrors human intuition in its ability to pick up on subtleties, but because it is a machine, it can process these faster than we can.
Let’s take an example of an official note that says: パリに出張の時に私はCEOに会いました.
With a phrase-based machine translation (PBMT), you would receive the translated output as: ‘I met Paris in the CEO trip doing business.’ With Neural Machine Translation (NMT), you will get: ‘I met the CEO when I was in Paris on a business trip.’
In Japanese, main verbs are always used at the end, so to make sense of the used phrases within it, you need to reference the end of a sentence.
NMT is also an end-to-end learning system, which means that it gets better the longer it is in use. This deep learning function powers its neural network, which computes translations with such a degree of complexity that often times even its developers are unsure how it arrived at its conclusion. It is, in a sense, very much like the human mind.
The beauty in all this is the ease to conduct international business now. Caring for clients, clarifying concerns with business partners, or trying to reach new markets is possible with the click of a button. All those written communications that would once have required a linguist can be translated with a comparable degree of accuracy using NMT, making business as usual, unusually uncomplicated.
So how should we anticipate seeing the effects of NMT?
Firstly, NMT allows small businesses to bring their product to the global market, and with them, their increasingly high standards for innovative, quality products. If a businesswoman in Poland wants to sell to clients in Japan, she can do so without having to spend weeks laboring over miscommunications and misunderstandings in emails written and read in second languages. This will increase the diversity of products and services available, while also speeding the pace of innovation.
Large corporations are already benefiting from this technology. Google’s launch of NMT technology operates in eight of the world’s major languages, covering 30 percent of the world’s population. Systran’s PNMT facilitates communications in 70 different language pairs.
An influx of both quality and quantity in products and producers will inevitably boost the current of globalization and its roaring marketplace. Language, which was originally a tool for organizing mankind, has in recent times become a barrier. This new wave of advancements in NMT may be just the crossing over point the global market has been looking for but it is more than ready to bridge that language barrier very soon.
This post is part of our contributor series. It is written and published independently of TNW.