Does Machine Translation Take Longer than Human Translation?

I am always talking about a term: ‘translation editors’. In my opinion, today’s ‘translators’ will change into ‘translation editors’ considering the advances in machine translation. However, I have read an article and the author is more pessimistic than me! The focus of her article is the current situation of translation agencies. She thinks some translation agencies are unreliable because they use machine translation. This is not the case in many translation agencies of course. That’s why you should be careful while choosing an agency. Agencies, like AIM Consulting,  offer the full services -not just translation but also a second round of editing and then a third round of proofreading. That’s the way it should be. Other than translation agencies, she is also talking about post-editing and how machine translation fails when the source text does not have a neat and simple grammatical structure. S/he has found this comment on web, which is quite right: 

“I was with a company back in 1995 that sold consumer translation software for PCs, and they marketed it as something magical: input an English business letter or marketing brochure’s text, and out comes a French or Spanish version. So simple, so inexpensive. No more expensive human translators.  But linguists laughed at the French and Spanish output which was often not only inaccurate, but offensive. Then they thought up the idea to combine machine translation (MT), as it is commonly-called, of business text, often marketing materials, with a low-paid, non-trained cadre of foreign language speakers, not translators, for a service offering to produce faster, accurate translation, but it turned out that this was not a faster process since even those linguists could not quickly “post-edit” poor quality machine translation of marketing content. It takes longer and is much more difficult to do that than just translating manually. Here we are years later facing the same issues. Most marketing material is not written with translation in mind, and contains abbreviated, “jargony” English language that is nearly impossible to translate accurately by machine. “Robo translators” can only work if the source language is carefully controlled, written in a simple grammatical style, and key term dictionaries are developed in advance that  can be used to handle a company’s specific terminology. The “crowdsourcing” model for translation for business purposes is a disaster waiting to happen in my opinion. For a global business, a careful, well thought out, culturally appropriate, quality localization project cannot happen magically with “robo translators” and volunteers.”

And conclusion:

So, why the ‘most’ translation agencies suck? Because it is not about human translation any more. Just learn from the example of Fortune 500 companies and try to understand why they don’t trust your “human” translation offerings. It’s a boiling soup, ladies and gentlemen. It’s time for a wake up call, or you’d be part of statistics.”

For the rest of the article, click here.

Visit our Facebook page.

Advertisements

8 responses to this post.

  1. Of course it does. Fixing the problems afterwards is more time-consuming and expensive. Whether businessmen will eventually figure this out is debatable, as with most gimmicks they like to play with.

    Reply

  2. Posted by cole on November 21, 2011 at 17:26

    First of all, there is a lot to be said for community translations, especially in areas where companies do not deem it necessary to localize for other markets (such as fansubs of Japanese anime etc.) or there simply isn’t a company behind a product (open source software anyone?).

    But using this as a “model” in order to get your stuff localised for free, you deserve to fall on your face. Quality has it’s price, and quality will always reward itself. So if you are running a business and trying to make the most out of the opportunities presented by globalization, using professional translation company is the only way to go.

    And yes, we have seen enough of crappy MT that takes a few seconds to produce and then hours and hours to sort out and turn into something remotely useful. But yes, all the people who are obsessed with the numbers just don’t seem to learn.
    Luckily, I can’t see machines taking over the jobs of human translators in the near future, as they have done with so many other professions.

    Reply

  3. Posted by Richard Boulter on November 21, 2011 at 22:12

    Richard Boulter • Hi Rad, It’s nice to hear from your extensive background in translation again, through Muge. I suppose you don’t remember me among the many, but I contacted you in 1998 as an already-experienced translation pro, to ask for some insight on how to make a deal with agencies and to perform well on assignments. I was transitioning from a trained criminal-legal interpreter mode into civil/commercial legal translation and interpretation, at the time; your generosity with information and your observations about our favorite art & trade have been invaluable, over the years! Considering Muge’s several contributions, I suspect that her own approach to fellow lingistic service providers is similar to yours. My own attitude toward colleagues is similar.

    Regarding the ‘proofing’ of purely machine-generated translations, I agree that they lack … well, everything. A proofer/editor must simply research and render every expression in the document, in order to assure a reliable rendition of the source text. Having accepted a number of these assignments, I now decline them or return them with an offer for a translation at full translation prices. Perhaps you neglected to include the ‘ symbol around ‘editing’ as well as around ‘translation/proofing/editing’.
    Best regard, Richard Boulter

    Reply

  4. Posted by Richard Boulter on November 21, 2011 at 22:28

    Richard Boulter • Hi Rad, It’s nice to hear from your extensive background in translation again, through Muge. I suppose you don’t remember me among the many, but I contacted you in 1998 as an already-experienced translation pro, to ask for some insight on how to make a deal with agencies and to perform well on assignments. I was transitioning from a trained criminal-legal interpreter mode into civil/commercial legal translation and interpretation, at the time; your generosity with information and your observations about our favorite art & trade have been invaluable, over the years! Considering Muge’s several contributions, I suspect that her own approach to fellow lingistic service providers is similar to yours. My own attitude toward colleagues is similar.

    Regarding the ‘proofing’ of purely machine-generated translations, I agree that they lack … well, everything. A proofer/editor must simply research and render every expression in the document, in order to assure a reliable rendition of the source text. Having accepted a number of these assignments, I now decline them or return them with an offer for a translation at full translation prices. Perhaps you neglected to include the ‘ symbol around ‘editing’ as well as around ‘translation/proofing/editing’.
    Best regard, Richard

    Reply

  5. As with so many things, it probably depends… But in general terms, for languages and content types that are well suited, there is now a lot of evidence that post-editing of MT takes less time than translating from scratch, at identical final quality. At Autodesk, a typical translation service buyer, we’ve carried out several productivity tests using fairly standard open-source MT technology. Some of the results can be consulted online at http://translate.autodesk.com/productivity.html.

    The use of this technology raises evidently a lot questions, but the most relevant questions are maybe not anymore about whether the technology as such is working but more related to its ownership in the supply chain and the fact that the productivity gains it enables first of all benefit the customer and others further up in the supply chain — unlike other professions where the choice of technology is more transparent to the end customer and the distribution of its benefits controlled by the ones producing the work (or their employers). I wonder if rejecting the technology out of principle doesn’t actually carry the risk of worsening this situation from a translator’s perspective.

    Reply

  6. Posted by ÜMİT YAKUP DURAL on December 8, 2011 at 12:03

    I have two conclusions on this issue that I have been stating since many years:
    1- Every sentence and every translation text has a unique SOUL AND SPIRIT because they are written by different persons with PERSONAL APPROACH AND SPIRIT. Therefore, only a human translator can translate texts such deep texts. As a result, a machine translator can not be more than a dictionary or an idiot translator.
    2- Beside these, a machine translator can be our servants that may help our work.

    Reply

  7. Hi!
    As per my knowledge goes I think human translation is the best method to continue with the term. Machine translation to some extent can remain accurate but it lacks the human emotion which usually should not be missed in the translated version. Translation does not mean only translating the original document. Instead translation has got a meaning of translating a document with its every other thing keeping intact.

    Reply

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: