One of the earliest goals for computers was the automatic translation of text from one language to another.

Automatic or machine translation is one of the most challenging artificial intelligence tasks given the complexity of human language. At first, rule based or statistic based methods were used for best translation. More recently, neural network models are being used to achieve high end results.

Basically in machine translation, a computer takes already given symbols and texts in one language and is tasked to turn them into symbols and texts in another language. However, there is no single best translation from language to language. The human language is much more complex than a computer can handle. Language has natural ambiguity and flexibility making it difficult for a machine to accurately translate.

Machine translation is evolving. Statistical Machine Translation uses statistical models to be best translate sentence A into sentence B. It’s statistical, given sentence A, the error in translation is minimized by choosing sentence B, the most probable. The source is data driven. It requires a large amount of examples in both languages to accurately find the best translation. The issue with Statistical Machine Translation is its narrow focus, the translation process goes sentence to sentence, thus losing the broader nature of the text. In Statistical translation the computer loses the syntax that would be easily recognizable by a linguist.

Neural Machine Translation is another type of evolving Machine Translation. Neural Machine Translation creates a neural network model to learn and source from. What’s cool about this is a single system can be trained on a source/target text. Similar to our own neural systems, the machine allows itself to learn and remember movements, context, etc. from a text to apply to future translations. However, the machine does run into issues when the translation need becomes too long, creating errors in translated words or skipping words overall. Although effective, the neural machine translation systems still suffer some issues. The machine has a hard time recognizing rare words and scaling to a larger vocabulary. Neural machine translation machines also face an issue of speed; their processing time is much slower than desired.

Machine translation is a modern technology that everyone can admit is pretty cool. The development and success of these machines can and will alter the language industry greatly. However, for now, the best translations out there are those done by humans.

GrandViewResearch.com announced that machine translation (MT) market revenue is anticipated to reach $983.3 million by 2022.

We know that the MT market hold will continue to strengthen and grow, and translation providers that blindly look the other way will surely be disheartened by this powerful market shift. But how far will machine translation go? Let’s start with the basics.

What is machine translation and why is it so important?

A Brief Introduction to MT

Machine translation, according to Wikipedia, is a “sub-field of computational linguistics that investigates the use of software to translate text or speech from one language to another.” On the most basic level, MT switches words in one language to another. This word-for-word substitution cannot produce a quality translation because translation is based upon the meaning of a complete message or a section of multiple phrases. What machine translation does accomplish is a cost-effective, very fast and easy way to manage large volumes of content. MT costs are significantly lower than human translation costs, machines by nature are faster than humans and managing high volumes is easier with technology.

The Importance of Machine Translation

The sheer volume of website content paired with the increased demand for cost efficient translation services exceeds the capacity of human translation. This intense need is expected to catapult the machine translation market and drive it into massive growth. Why?

• Globalization increases the need to address linguistic diversity of internet content for local audiences
• Online content is expanding at record-breaking speed
• Emerging markets are demanding access to content, goods and services
• Global collaboration amongst employees and work teams is pushing current translation capacities worldwide

There are many challenges for the machine translation market including a serious concern about the overall translation quality mixed with a technology that is difficult to use in an industry that is resistant to change. Any MT project will require post-editing skills that are not currently abundant within the translation providers and these skill will have to be learned quickly to reach the $983.3 million dollar mark.

Machine Translation Market by Sector

The graph from GrandViewResearch.com (click to view) highlights the different market sectors that use machine translation on varying levels. The graph illustrates the growth for machine translation users. The following are the sectors with the highest MT use in terms of volume:

• Automotive: Crucial communication pieces must be quickly and accurately translated for consumers
• IT: Specific vocabulary and terminology must be consistently applied across all channels
• Military & Defense: Time-sensitive document and intelligence translation is critical to military and defense efforts
• Healthcare: Physicians need communication supplements to effectively reach patients and research
• Electronics: Efficient and cost-effective translation required to speed up time-to-market

The machine translation market is here to stay and with the projected revenue of $983.3 million dollars by 2022 is not a number that translation providers can hide from. At INGCO International, we have worked with a variety of MT programs and technology solutions for our clients over the past 10 years. Our post-editing capabilities are consistently expanding and growing stronger to meet and exceed our customers expectations when speed-to-market is at the forefront of everyone’s agenda. Are you looking at machine translation as an option for your high-volume translation needs? Reach out to us today to learn more about the MT solutions that will help mitigate your multilingual communication challenges.

One of my favorite activities is to test out Google Translate to get a good laugh.

I should have been working this morning and answering the hundreds of emails in my inbox, but instead I was sidetracked by Mr. Google Translate. Oops. But I did get a good laugh!

I love the great example of “my nose is running.” Native English speakers know this doesn’t literally mean that my nose is out for a jog. Or that my nose is working like a machine would be working. (That would be just weird and make you think that I was a robot of some sort.)

I love to use this example when talking about how humans will never be replaced by machines when it comes to translation. Language is a living, breathing entity that ebbs and flows just like the tide. It’s constantly changing and evolving and our understanding of words and their meaning evolves as well. Machines as they stand today just cannot do that. To prove my point, I asked Google Translate how to say “my nose is running” in Spanish.

Oops.

This time my nose is like a robot and working. Just like a machine. Google Translate is a great tool for certain purposes, but not for a professional translation. To achieve better accuracy and a greater ability to capture the tone and style of the original documents, humans are where it’s at.