Matthew Bennett Matthew Bennett
Learn Spanish online. Think about life and British & Spanish business, law, politics & culture. Translator, teacher, blogger. Running, reading, wine. Read more about me and my blog...
Matthew Bennett Twitter
Matthew Bennett RSS

How could a professional translator help his clients more with machine translation?

Audio clip: Adobe Flash Player (version 9 or above) is required to play this audio clip. Download the latest version here. You also need to have JavaScript enabled in your browser.

As a professional translator, I am interested in helping my clients in any way I can and I try and follow trends in my industry (translation) and theirs (law, international commerce) as far as I can in order to see where things might be going.

Recently in the translation industry, there has been much talk of crowdsourcing and newer versions of machine translation tools like the Firefox Translation plugin, which on the surface make things much easier for clients and end-users to ‘translate’ web pages and documents almost instantaneously and for free.

Firefox translation plugin

Firefox translation plugin

I think we might all end up with less global understanding if we carry on like this, but that’s an idea for another post.

Professional translators and their savvy clients, however, know that the quotation marks around ‘translate’ in the previous sentence are there for a reason – machine translation has advanced a lot over the past few years but still has a long way to go before it could be considered a good translation.

The texts produced by most machine translations would currently still fail degree-level translation exams, never mind real-world business, legal and medical document translation tests.

Nevertheless, in the minds of many people, these new translation tools and advances allow them to ‘translate’ their texts quickly and for free. It’s a strong enough offer to demand our professional attention.

It’s all very well rallying against these tools and shouting: ‘that’s a crap translation!‘ – and we all know our clients are wise enough not to use these tools to translate important documents – but I’m sure most of us also know that most of our clients use them in one way or another for less important texts.

As translators, we know they’re doing something which is useful for them in some way but we can also see glaring errors in the results.

If we wish to look after our clients, there might be ways in which we can plug the gaps in a productive manner that takes into account both the behaviour and desires of our clients and our own professional criteria, at an acceptable cost for them and an acceptable profit for us.

What could professional translators do to help their clients about the gaps in meaning which are evident once you know what you’re looking at? Where could we draw a new line between translation quality and the everyday, harsh reality of our clients’ companies?

If you’re a professional translator, what would you offer to do? If you work with a professional translator, how would you like them to help you? I will post my suggestion on Monday.

Your Comments
  1. Hi Matthew,

    Again another fascinating blog. I was particularly interested in this comment of yours: “Nevertheless, in the minds of many people, these new translation tools and advances allow them to ‘translate’ their texts quickly and for free. It’s a strong enough offer to demand our professional attention.”;

    True! In fact the whole idebate about free translator engines has led to some research at the Universitat Autonoma of Barcelona (where I work) to start a ranking. Universities are ranked so why not FTEs? Below are the top ten according to ourselves at the Faculty of Translation and Interpreting at UAB. The ranking is based on a score for a text which is machine translated by each of the FTEs four categories fidelity (4 points), accuracy (4 points), style (2 points) and ease of use (+,-), making a total of 10 points. However, these categories have been hotly debated – I wonder whether other readers, bloggers would agree with the way of marking a text. It is the one that three or four of us have agreed on at the FTI but it is not necessarily the best way of doing it?!
    Our article in the making starts like this:

    Despite the fact that there are many free translation engines on the market, there are no league tables or ´ranking `systems as far as we know. There is increasing user acceptance by the general public of machine translation (MT) for ´gist`. For example Smith ([1] 2001) cites evidence of Transparent Language, which runs a on-line free translation service of receiving 2,000 translation requests in typical one-hour period. Eight years later, the website (www.freetranslation.com) is still going strong with two million visits a week. There is also a huge range of free translator engines such as: Babelfish, Translator Online, Foreign Word, Web Trance, Prompt, Verbalis, Google, Systran and Intertran.
    As the interest in and demand for free translations, from translator engines such as Google, increases, we believe a league table or ranking system of this multitude of engines will be useful to the user to decide which one suits him or her. We provides a clear description of the criteria used to judge the quality Although rankings provoke fierce criticism ([2] Dill and Soo, 2005), they are completely credible at mainstream public opinion and higher education policy levels ([3] Marginson, 2007).

    RANKING OF WORLD FREE TRANSLATOR ENGINES
    (SPANISH TO ENGLISH)

    WORLD RANK ENGINE NTIs Logo Ease of Use
    1 Google 8.0 *****
    2 Dictionary.Com 8.0 ****
    3 Babylon5.0
    4 BabelFish4.0 *****
    5 Systrans4.0 ****
    6 Bing4.0 ***
    7 Worldlingo4.0 **
    8 SDL3.0 *****
    9 PROMT3.0 ****
    10 Reverso2.5
    11 Foreign Word0.0

    Cheers Matthew!

    (Dr) Steve Hampshire

    steve hampshire
  2. Hi Steve, thanks for that good information, I wasn’t aware anyone was coming up with a list like that. Please let me know when your article is ready.

    From the user’s perspective, that kind of advice could be very useful. Do you know if anyone has done any research on mixing machine translation with human translation?