For matters related to payments, access to training sessions and technical issues, please submit a support request for faster assistance.
Machine Translation for agencies
| Access to the course || Oct 28, 2010 |
Access is granted immediately after payment is processed. I see you are in the session. Hope you enjoy it!
| | Jeff Allen
Local time: 18:40
| comment about experience of translation agencies with MT || Oct 29, 2010 |
Thanks for the feedback in the trainer feedback form about why more emphasis was not placed on experience of agencies with MT.
I'd like to answer that here. Having been in the translation industry for 20 years and in MT for last 15 of that period, and wrote many reports for the European Commission on the progress of the language technology industry based on surveys we conducted at ELDA. I have seen how many different translation agencies/LSPs have tried to adopt MT, and have seen the successes and failures. In general, as I mentioned during this webinar and in all of the webinars and panel sessions, the use of any type of customized MT (irrespective of the MT system approach) is always more successful than anyone trying to use non-customized MT (such as Google Translate). Now, there are a lot of translation agencies out there who think that MT = Google Translate, and so that is what they try to work with. I never recommend doing this, except for exotic languages like Haitian Creole where the only MT systems that exist for it are the pure non-customizable SMT systems.
The other factor is personnel training on their tasks, and any project which has not put into place training sessions to show people how to adapt to new roles from the mindset of what they are used to doing, can put the project at risk. And the first page mentioned looking at one's set up as a company with in-house vs external freelancers. Who will pay the freelancers for being trained on postediting principles, or on rule-based MT dictionary building, or on bug reporting to various MT systems, etc? This is often overlooked because it is assumed that it will all take place.
However, translation agencies factor this into their own human translation projects as project admin which is often a percentage of the project, or is a dedicated part of the project localization experts who know how to use specific tools, or even special costs for specific tool training, or higher cost per word by
specific translators who have special skills.
But how do these project related costs get factored in to produce a successful project.
For many translation agencies, this was unknown territory. Some tried, and failed. Some waited and watched the failures and asked a lot of questions. And most of the 5,000 translation agencies worldwide have been watching and waiting for many years.
All those who seek good objective advice and study it carefully for their own context have a very high chance of being successful in their decision.
There have been many failures of translation agencies trying to use MT over the past 15 years, and usually it is due to such issues as those mentioned above.
A few how been successful with it because they chose to specifically become well-informed before moving in that direction, and now they are talking about it at conferences.
* Pangeanic - they finally converted their own translation service offer into an MT solution for others
* PROMT - PROMT is an MT provider, but also has a language services dept that uses MT to process the content of the translation services that are provided.
* CLS : wrote a few papers about their implementation several years ago (after 2000) but I haven't seen any updates on it.
SDL has been providing MT and postediting services for many years, since they acquired the Transparent MT system back in 2000-2001.
Lionbridge acquired the MT system that was part of the Mendez translation company portfolio that had been part of L&H, then got acquired by Bowne (BGS) (in about 2001-2002) and then acquired by Lionbridge (2005ish)
So these companies all have been using MT internally for a long time, but don't talk much about it, or how. Lionbridge has been mentioning over the past 2-3 years at conferences and also some on LinkedIn that they are using MT internally and plan to continue to do so.
Yet, any failure of using MT for a project had great impact. It's like hotels and restaurants.
How often do you recommend an average hotel or restaurant to friends? : never
How often do you recommend a really good hotel or restaurant to friends? : well you might if you remember to take the business card and put it somewhere, so likely you do it occasionally
How often do you talk about a bad experience with a hotel or restaurant? : you tell everyone in the world and might even take a picture of it with your phone and put in on your facebook wall.
The same goes for MT. The bad experiences are constantly brought to the surface. But then I always ask, why did they go bad? And it is not surprising that some of the project management and project factors mentioned higher above were at the root of the problem.
It is good that a few agencies are now having a good experience with MT, based on really focusing on customized MT implementations and doing it appropriately. And hopefully this will help compensate for all of the previously bad experiences.
| || || |
To report site rules violations or get help, contact a site moderator:
You can also contact site staff by submitting a support request »
Machine Translation for agencies
|SDL Trados Studio 2017 only €415 / $495|
|Get the cheapest prices for SDL Trados Studio 2017 on ProZ.com|
Join this translator’s group buy brought to you by ProZ.com and buy SDL Trados Studio 2017 Freelance for only €415 / $495 / £325 / ¥60,000
You will also receive FREE access to our getting started eLearning program!
More info »
|Déjà Vu X3 |
|Try it, Love it|
Find out why Déjà Vu is today the most flexible, customizable and user-friendly tool on the market.
See the brand new features in action:
*Completely redesigned user interface
*Inline spell checking
More info »