Translation experimentation

The next real big step for the Internet in revolutionising this planet is, I am convinced, mass accurate translation so that the language barriers that have divided and enclosed the world ever since we as a species first started talking, will be if not exactly cast away, certain diminished.

And so I have installed a translation plug-in for this blog as an experiment. There are about five out there that plugin to WordPress. I would like to say I will test them all, find the best, and write about it but I already know I won’t have time. Anyway, let’s see how it handles this post.


This is terrific – looks at the little flags under the search on the right-hand-side. Just click on your flag and the site is translated. I have a choice of Google translation engine or BabelFish. Gone with Google for the meantime. Now the big question is: is the translation good enough to be worthwhile for people to read or is it just frustrating? It’s a very difficult thing to judge. Any readers whose first language is not English that may try this, please, please do comment below on what you think.

  1. German is missing, so I can’t say if it’s any good now. But from past experience with these translations I wouldn’t expect too much. It helps to get an idea of the page but is not a pleasure to read.

  2. Hey Armin,

    I’ve stuck German up and pulled down Portugese – it’s more to do with only having one line of flags.

    See German version here:

    Have a look, give the translation a rating out of 5 would you?

    More importantly though – is the translation good enough for you to read a post out of interest in that language, or is it too bad to be worth the trouble?


  3. How do I put this mildly? Absolutely useless. I hadn’t expected it that bad, but what this translation is doing is beyond hopeless. In some cases it might help me to figure out what an entry is about, but it’s hard work. That’s a 1. If someone bribes me massively (as in I’ll never have to work or worry about money again) I might give it a 2.

    Funnily enough I used earlier today for a Spanish to English translation, that actually was fairly readable. I don’t know if it was the contents or if Spanish is more suitable for machine translations.


  4. Ah really? That’s a shame. I was hoping it was at least good enough to read.

    It’s so hard to quantify these things. For example, it occurs to me that the translations in other languages might be really good – as you appear to suggest. But how would you ever know.

    How do you get people to rate quality of translations? And then, more importantly, how do you get the collective efforts of people online to build a learning system that can get close to real translation?

    Anyway, thankyou very much for your feedback Armin. Sorry the result is so poor.


  5. I think the original English is too complicated. Shorter sentences might yield a better translation. I rate the Japanese a 2 out of 5. Disclaimer: my first language is English.


  6. Yes, I agree Jim, I came to the same conclusion. I write blog posts in quite complex English: chatty, full of reference, phrase and unusual sentence construction. There’s just no way a computer can understand that. But I wonder how it would cope with straight, plain business language.

    I think I might write something in a news story format and then see how the translation deals with it.

    Thanks for looking at the Japanese, Jim.


  7. Tried the Italian, it’s a step up from the usual babelfish mess but not by a lot. The first half of the entry works much better — would be curious to look at another post in simpler language to see how it pans out…

  8. […] Three weeks ago, I added a translation module to this blog as an experiment with automated translation software. […]

  9. As for me, I use WordPress and I’m completely satisfied.

Comments are closed.