The announcement in December of the sale of Systran – the grandmother of all MT – to CSLI Co. Ltd. in Korea sent a small shock wave through the translation technology sector. Is Systran giving up after some forty years of chasing the holy grail of fully automatic high quality translation? Or is this the best deal that one could expect?
The question also creeps up: is there a market for MT, really? On the one hand we see MT everywhere, mostly as a free service or utility on web sites and in apps, soon also in the Google Glass and NTT Docomo smart eyewear. On the other hand we don’t see commercial MT development companies being very successful. A strange paradox, isn’t it?
It certainly wasn’t for the money when Jean Gachot bought the rights to Systran from the original legendary developer Peter Toma in 1986 and 1990. He was an industrial valve magnate and owner of the Gachot group of companies. The couple of million dollars he needed to acquire Systran Institut in France and Toma’s affiliations in La Jolla (USA) wouldn’t have bothered him much.
In 1990, Jean convinced France Telecom to offer Systran MT on the Minitel network in France, a national Internet-type network 'avant la lettre', and an early predecessor of the now-common MT buttons we find everywhere on the Internet these days. In fact, for a short period of time, anyone in any country could access this service via telecom link from a PC (Minitel acted as a modem), paying $1.20 for a page of translation!
It was the vision that took possession of Jean. The same vision that captured many ideologists and over-optimistic investors in the translation technology sector since the eighties. What if I can make this computer translate fluently?
Jean Gachot wasn’t alone in his dreaming. The European Commission spent 35 million Euros from 1982 till 1987 on the Eurotra project. Eurotra was aimed to deliver reliable automatic translation between the then nine official languages of the European Union. In fact Eurotra was to beat Systran, among others. But it failed and we don’t talk about Eurotra anymore.
Just like we don’t talk anymore about Logos, Globalink, Tovna, Weidner, Metal (Siemens), Rosetta (Philips), DLT and other dream MT systems that absorbed tens of millions of dollars in the eighties and nineties to crack the language barrier.
We are living through the Gartner Hype Cycle, going from the peak of inflated expectations into a trough of disillusionment, and there is nothing we can do about it. Before we start climbing up the slope of enlightenment (still in Gartner terms) it seems that we really have to shake off the past completely.
Old-style companies cannot survive in this modern age of free software and free services. Apart from a few dozen government organizations and closed corporations there aren’t that many customers anymore for expensive license contracts. Is this why, after Language Weaver was acquired by SDL, Systran is now becoming part of a Korean language service provider?
And yet, we see MT everywhere. We (we at TAUS) even believe that MT becomes the new lingua franca. Then where is the money? And who is getting big, rich and famous? Market analysts believe that the global machine translation market grows 18% per year to a size of several hundreds of millions or even a billion plus in the course of this decade. We doubt it. We don’t see the growth of pure MT revenues.
What we do see is that MT becomes part of other services that drive revenue. The MT technology itself becomes very cheap or is even free. Google offers free MT engines to the European Patent Office in exchange of corpora of text in multiple languages. These corpora help Google to improve their overall automated translation and that supports the core business of Google.
Similarly we see Microsoft offering MT customization to many corporations and translation companies for free or at a very low volume rate. Obviously large corporations like Google and Microsoft have different ways to justify and monetize their investments in MT, but can we count them in for what we frame as the market for MT?
More and more translation companies build MT engines into their service offering. They connect with Google Translate or the Microsoft Translator Hub or they build their own engines using the open-source Moses MT technology. The MT service gives them a competitive edge and helps them to sustain and grow their translation and post-editing services, basically larger volumes at lower rates.
MT is becoming a must-have feature for all providers of translation services and translation management systems. And yes, this does lead to the emergence of a new generation of MT companies that define their niche as being the alternative to Google and Microsoft. If there is a market for MT, we need to zoom in on the 21st century translation technology companies. How do they differentiate from Google and Microsoft and how do they create a space for themselves?
We invite the entrepreneurs of all of these new companies to help us understand and predict the market for MT as we see translation becoming a utility embedded in every app, device, sign board and web site and converging with other technologies such as speech technology. Please click here to fill in a short survey.