Systran On the Move - Lets Shoot Language

If you are not interested in the nitty gritty of Machine Translation, skip most of the first page of the interview with SYSTRAN CEO Dimitris Sabatakakis. The scary parts for us, as human translators, and as future readers and speakers of language is this promotion of a stilted usage for financial gains:

Sabatakakis: Today, we have no monopoly but Google and Microsoft created a real market. The primary use of machine translation is to translate from one language to another. There is a second use that is more complex and sophisticated and very interesting. When a corporation localizes content, it needs to provide technical support, [translation of] databases, marketing material, Web material and so on. Up to now, that was a human process. Companies would pay per word and rely on human capacity. Suddenly and progressively, translation software is starting to be used by corporations willing to standardize the way they translate complex documents. Translation software is becoming a key software for [translations by] international corporations. It is not for SMEs or local corporations, but big corporations with employees and people talking different languages who want to communicate more efficiently.

Then further on in the interview:

Weinschenk: Is translating general material – whatever comes through a Web portal – fundamentally different from translating technical information?
Sabatakakis: The more the text is technical, the more restricted the language, the better the translation of the document. Translation works much better with technical material. If you read the New York Times, there are tens of subjects, much general language, many journalists, many different styles. The creators are not offering tools or obliging the writers [to follow specific conventions] in the same way that it is done for technical documentation. Symantec writers [for instance] follow specific rules. They do not have the freedom of the journalist. Their documents contain very little general language that is ambiguous.

Weinschenk: So there is a lot more predictability when people aren’t involved.
Sabatakakis: When you use translation software, the translation always is consistent. In other words, a human being may say “plane,” “aircraft” or “airplane.” When using translation software, you select the terminology depending on the domain or the profile you are creating. The first thing it brings is consistency. It is not a human being, and does not make the mistakes or have the freedoms of a human brain. You also can reuse translated content, which creates more efficient and faster translations. (He is making some seriously mistaken assumptions here).

Weinschenk: How evolved is this technology and the market?
Sabatakakis: Five or 10 years ago, the software was not working properly. You could not use it without extensive training. Machine translation is not a commodity market, but tremendous progress has been made during the last two years. We’ll see in two years what the adoption rate will be. I believe more corporations will automate the process. I cannot say the human translation industry will stay as it is. Up to now, human translators were using little translation software because it was not working properly. Now with the tremendous increase in quality, more and more translators are using our software. Most will adopt it progressively. This will have a very important impact on their jobs. They will translate more and some of the [monetary and other] benefits will accrue to the client and some to them.

Weinschenk: How big is the market, and how does it compare to the human translation market?
Sabatakakis: The human translation industry produces between $10 and $20 billion per year. I do not know the precise figures. The translation software tools industry produces less than $100 million. There is enormous potential for growth.

Let me get out and purchase my R2D2 dictionary.