Gala2012 Monaco Is Technology Replacing Smaller Ls Ps Christopher Carter

Is Technology Pushing Small and Mid-sized

LSPs out of the Translation Supply

Chain?Chris Carter

@GALA_global

@GALAconf

What is the future of human communication?

Where are we going?

Technology is helping our industry. For now.

But eventually, we will just replace ourselves with technology.

One day, no one will need translation companies at all.

One day, we will all have one of these....

Sir, I am fluent in six million forms of

communication.

Okay. Maybe not.

Confidence in Language

My name is Chris Carter.

General Manager of aLanguageBank

- language service provider- based in New York City- translation, interpretation,

audio and video localization, and various types of consulting services

¡Hola!

お元気ですか？

I started studying linguistics when I was 11, and when I was 12 I started studying Spanish and Japanese.

When I was 13, I decided that I would study linguistics at university, and I did.

And I still study linguistics and languages for fun. Every week.

I have loved languages and linguistics since I was a little kid.

Translation industry is trying to make text available for everyone, regardless of language. But the goal here is not translation, it is communication. Communication is the transfer of information, of thoughts. Communication may not even use any text but still need translation.

It might not even include words at all! Emotions can be communicated too, without words. Examples: dance, theatre, ….. or even mime.

CONTENT TRANSFORMATION

REVOLUTION

We are beginning to realize that we are starting a “content transformation revolution”. Not about translation or interpretation. Not about technology. It’s about content.

Okay, that is a bit of an exaggeration. This revolution is slow.

Actually, it is very slow. But the changes are there. We can already see it starting.

Google is already allowing search browsers to search by reading level. Making content available in a format they want. This has nothing to do with language. It is about letting people access content in the format that they chose.

Similarly, META-NET in the EU has 54 research centers in 33 countries, investigating how technology can unify Europe across languages. It pursues language technology research and engineering to make content available as quickly and easily as possible in all aspects of daily life.

Professional Trainer and Consultant Mark Proffitt has a methodology called Predictive Innovation. One simple aspect of it is that technology improves in cycles.

Until the “perfect solution” is reached. (If ever.)

For example, machine translation. The general market definitely knows about it now. It’s not perfect. The market wants it to be better. We keep trying to improve it.

MT is not perfect, but no one wants to admit that neither is human translation.

The real question is not if MT is perfect. Rather, do the end users care?

So what is going to happen in the near future? More commoditization.  This will continue, and it will continue to commoditize localization services. You can’t fight it, accept it.

In fact, free general purpose MT, like Google and Bing, and trends such as crowdsourcing are making clients think that translation is cheap. “This is your invoice?! Oh just use machine translation on it.” But they are also making the end users, the general public, think that translation SHOULD BE FREE. Commoditization is not going away any time soon.

We are also going to see much more technology integration.

By “integration”, I mean the combination of different types of technology, inside and outside our industry. Technologies that manage content, transfer content, or convert content.

All content storage and even processing off data is moving into the cloud. And we are accessing the cloud more and more from mobile devices.

We all want to interact more, with either each other or with technology, and that all involves content. And it all involves integration of technology.

There are now more ways to access the cloud or the internet. And all this can change what we take for granted.

For example, intelligent searches on mobile devices, such as Apple’s Siri, use geo-location with GPS on our phones or tablets.

With Siri, people are arriving at web sites or mobile content without using a standard search engine.

They ask for something, Siri uses key terms to decide which app to use to get the answer. And when location-based questions or requests are made, Apple admits that Siri relies more on information in sites such as Yelp or Foursquare.

People are accessing mobile content without ever using a search engine like Google.

The more we use intelligent software, or these intelligent computer assistants, the less effective SEO is.

USER

One major change we are seeing already is the shift of who is demanding our services.

The growth of our market is motivated not by clients, but by the general population – the end users.

This has also led to the shift of who determines quality.

Clients or LSPs have always decided what quality means in translation. But now users are telling us. They tell us when they need perfect translation, when they need “good enough”, or when they need something in between.

Another change I see is the shift from when translation happens.

Up until now, most translation happens when it is sent out to the users. Now though, users are receiving content and then deciding they want it translated. Now users look for any content, in any language, and they decide if they want it converted, and when.

If your client can offer technology that lets their customers translate content after receiving it, that client doesn’t need an LSP to translate it.

And this leads to translation, and content conversion, becoming a background step and not always a process by itself. Users just decide how they will receive content, any content.

This could be in their settings on whatever technology they use to receive content. They won’t even have to click a “Translate Me” button.

One day, the master profile on your iPad 13 will just know to automatically translate all content from any language, into the language that you chose. You wouldn’t even know (or care) what language it was originally written in.

We are already seeing this “featurization” of translation. Facebook now uses Bing Translate to allow users to translate comments. Several companies offer some form of instant chat translation.

“These things kind of sneak up on us. By the time they are revolutionary, they’ve been around, actually, for 20 years.”

—— Ray Kurzweil

Here is the basic supply chain for document translation or localization.

I separated Small and Mid-sized LSPs (SMEs) from Large LSPs. Why?

There are technologies that could possibly threaten the way LSPs do business. But Large LSPs are big enough to provide the services that SMEs do, and also to provide technology.

Some of the technology that threatens SMEs actually belongs to Large LSPs. They are large enough to invest in or develop them. Examples:

- Lionbridge’s GeoFluent offers clients MT for chat (using IBM’s “real time translation”)

- SDL offers their EasyTranslator and BeGlobal software options that do not use any humans in the translation process.

One way new technology companies will say that they can offer lower rates is by removing all of that expensive, wasteful project management.

These companies are not offering “good enough” translation. They are trying to compete against you by saying professional translation can be done without any human project management.

The general market sees new web sites such as MyGengo, WebTranslateIt, Bewords, and Cloudwords and they think that translation can be done extremely fast and extremely cheap.

Venture Capital is investing heavily in these technology companies.

- Cloudwords: several million (including over $3 million in April 2011)

- myGengo: over $5 million late last year- Smartling: $14MM in less than two years

But most of these companies that say they remove project management still use human translators. Whether TEP or MTPE, human translators are used.

It appears that these new technologies that don’t use PMs are connecting our clients directly to the linguists.

These companies are saying clients don’t need PMs. You need to tell your clients why they do need PMs. That is how you compete.

What does a PM offer? Why is it better than these sites? The answer depends on your own company’s process, and the clients you serve. Look at your clients.

This new model might be better for a few. If so, concentrate on your other clients, and tell them why they need project management. And then keep reminding them.

But this model doesn’t actually remove Project Management. It just shifts it to the client.

Some sites ask the client to upload glossaries, upload TMs, clean TMs, search through large databases to choose the appropriate translators, review and quality assure each step, and manage accounting paperwork.

See, no PM! You just saved money.

Maybe your client’s time is more important than a little money.

These new technologies still use human translation with TEP or MTPE.

But most of the growth in market demand in coming years will not be human translation. Users (not clients) will require automation of translation. It is what the market, users, now realize that they want and are not getting. Now they want it and will want it more and more. Technology companies dominate this area and always have.

Integration of technology will especially be seen in the growing industry of fully-automated translation.

WordLens, for example, lets you point your mobile phone camera at something with text on it and it will translate that text. It uses optical character recognition to scan the image, then uses MT to translate it, then it covers up the original text in the image on your screen, and finally places the translation in the image where the original text was. Amazing!

Of course, it only works with English to and from Spanish. And the MT isn’t great. But users see the idea of what this kind of technology could be. And we’re back to the technology cycle. They want more. Now we have to deliver.

已经看到了，但是 我兴奋看又。 这是

sa pou yo kout zegwi sou a, Se konsa fin pa

Speaking AND hearingLANGUAGE #2

Speaking AND hearingLANGUAGE #1

Voice Recognition+

Machine Translation+

Text-to-speech

Some technologies even let you speak to someone on the phone in one language, while the other person hears a translation in another language. I said translation, because no interpreters are used.

The technology uses voice recognition software to transcribe what you say. Then it uses MT to translate. And then a text-to-speech software “speaks” that translation into the other phone.

NTT Docomo and Lexifone are examples of this. Microsoft is developing their own version of this as well.

These are a great example of multiple types of technology coming together to create a user experience from daily life where content conversion happens in the background.

The idea sounds great! There are still a lot of problems with the software. But users will want more.

Households all over the world have those boxes that can see and read your body movements. One example is Kinect for XBox. It was created for gaming systems. But why can’t we use that technology to transcribe sign language?

Capture the movements, transcribe the gestures using one of the many existing sign language notation systems, translate that notation into another language, and use text-to-speech to “speak” that translation to a hearing person.

Machine Interpretation for Sign Language.

Yesterday Tomorrow

Glo

baliz

ati

on

Com

modit

izati

on

Experi

menta

tion

/ Te

ch

inte

gra

tion

Separa

tion /

E

duca

ting c

lients

For many decades, our clients thought all translation was the same. Quality was just quality.

At the beginning of globalization, clients wanted to start reaching other cultures, so they sought out specialists. Us. LSPs raised their prices.

Then commoditization started, and prices started to fall.

(And they still are.)

Demand for our services is growing, especially by users. More new technologies will try to support that. We are entering a time of experimentation. Everyone trying to determine what everyone wants.

Tech companies and LSPs have to educate the clients and the users about the differences between the different kinds of localization. Some is free, and some is not.

As everyone finds their match, the definition of translation will split, and thus pricing models will split.

Yesterday Tomorrow

Glo

baliz

ati

on

Com

modit

izati

on

Experi

menta

tion

/ Te

ch

inte

gra

tion

Separa

tion /

E

duca

ting c

lients

Most daily content will be translated by machines, creating rough but usable translations. People will know that it isn’t perfect.

People will eventually understand better that human translations are at a higher standard. They will know that they need to pay big money for this high quality translation that automated technology does not provide.

MT and tech integration could actually one day De-commoditize human translation.

For the past few years, the assumption has been that the industry would split. But between TEP and MTPE. This is what linguists have been afraid of and fighting against.

But as different technologies keep providing new and different solutions, the market will separate in new ways. This will create a new kind of segmentation.

The market will split between fully-automated translation, and what I call “human touch” translation. That is translation that still requires some human linguist to work on it. TEP, MTPE, even linguistic or subject matter QA. Anything that is not completely automated using technology.

And if this data deluge is coming, or is just starting, there will be more data than we can handle. With more and more data, and more people will want more access to more data more often.

New jobs will be created. Jobs that don’t exist today will be created to manage the large amounts of data, to sift through the data and focus in on only the relevant or needed data. We will see more cross-over between the localization industry and text analytics and data mining.

Most SMEs are currently on the human touch side of the industry and will want to stay there. Large LSPs might be big enough to be on both sides.

continue educating your clients on the differences between human translation and automated translation.

LSPs providing human translation will need to show their value over automated translation. And the best places to do this are areas where the standard of quality must remain high.

Be experts, not general full-service LSPs. So many companies do everything.

The future is specialization.

LSPs should specialize in certain industries or fields to create an advantage.

That doesn’t mean you can’t or shouldn’t provide other services. But become well-known for 1-2 specialties. It is harder for clients to see a small company as a “one-stop-shop” for everything.

This will be especially important if you add MT into your workflow. Each MT takes a large investment of time and money. The more work you do in one industry, the more return you can get on one investment with an MT engine.

Stay with fields that need accuracy or craftsmanship, like marketing, medical, financial, or legal localization. Regulated industries, high-stakes content, or confidential and sensitive content will also be good for specialization.

And where is all of this going? Technology will continue to work together. We will find more ways to put the pieces together to do more. But the tech cycle says we will keep improving these new ideas.

Imagine combining TM with MT with Voice Recognition software with Intelligent Search or even artificial computing programs. A computer could listen to us and understand more than just verbal commands. Communication includes both verbal and non-verbal information.

A computer could understand context. If we also add Facial Recognition software or movement scanning technology like the Kinect for Xbox, it could read our body language and know that we mean more than we are saying. Or it knows when we’re lying. A computer could even have an intuition.

The historian and linguist Nicholas Ostler has said that MT is becoming the world’s lingua franca. There may be some truth to that.

Humans are shifting their interactions with each other more and more to technology. Phones. Email. Text messaging. Instant messaging. Chat. Social media. I have friends who will call someone just to get them right away, and they say “Text me” and then they hang up. And the person texts them, and they have an entire conversation in text message. ......... But you were just on the phone!

More and more of the freelance translators that my company works with want us to instant message them in chat instead of call them on the phone.

I see young people everywhere having conversations in text messages, when they are in the same room next to each other. (My nephews love to do this.)

Is technology becoming the middleman for all human interactions?

“Oh no, Chris. That’s ridiculous. Technology is cold, distant. Human interaction is special, intimate.”

Twenty years ago, computers were very common. But who knew back then that today so many people would be having sex through a computer.

IBM scientists are already working in the field of bioinformatics. They are trying to develop ways for machines to read brain waves. This research had a medical beginning, scientists trying to create a prosthetic arm that the person could control by thinking.

But why would we stop with prosthetic arms and legs. If we can control and communicate with computers with our thoughts, and those computers talk to other computers, which talk to people’s brains.

Just remember:

Content is merging, technology is merging. The future is a crazy mix of possibilities, but that won’t happen tomorrow.

The amount of fully-automated translation being used in the world is going to grow exponentially in the next few years.

Don’t be afraid of technology that goes around LSPs. Look for markets that cannot or should not skip the expertise of an LSP, or industries that honestly do need to be managed by humans – some form of project management or guidance.

And if a client or an industry needs the expertise and human management that an LSP offers, do us all a favor, and educate them. Keep educating them. We have to education of the market about the differences before we can de-commoditize localization and translation.

Just remember:

Content is merging, technology is merging. The future is a crazy mix of possibilities, but that won’t happen tomorrow.

The amount of fully-automated translation being used in the world is going to grow exponentially in the next few years.

Don’t be afraid of technology that goes around LSPs. Look for markets that cannot or should not skip the expertise of an LSP, or industries that honestly do need to be managed by humans – some form of project management or guidance.

And if a client or an industry needs the expertise and human management that an LSP offers, do us all a favor, and educate them. Keep educating them. We have to education of the market about the differences before we can de-commoditize localization and translation.

Answer

The only SMEs that will be pushed out of the translation supply chain are the ones that will keep trying to do the work that automated technology will soon be able to do.

Answer

The only SMEs that will be pushed out of the translation supply chain are the ones that will keep trying to do the work that automated technology will soon be able to do.

Thank you

Chris CarteraLanguageBank

[email protected]