I stumbled upon Funny Translator and got a good laugh by making successive machine translations (back and forth) through this site.

They provide an example on their site about how translation gets mixed up:

Original, from Lady Gaga’s Bad Romance: “You and me put on a bad romance”
54 Translations later: “I love you?”

I was not convinced, so I ran the same original sentence a couple of times through the mill. Note that each translation takes the previous one as its input. I came up with the following results:

1st run:

translation #4: You and I have a bad novel

translation #22: The novel of the past UI’ma

translation #56: Oh, ma news.

As the example translation did not pop up, I tried again.

2nd run:

translation #1: You and I sit on a bad romance

translation #4: I sat in a bad novel

translation #26: There was a bad novel

translation #28: It is a terrible novel

translation #29: This is a terrible novel

translation #30: This is a terrible story

translation #32: This is a terrific story

translation #33: It’s a great story

translation #35: Good story

translation #44: Anecdote

translation #55: Phillip

When you look through it, you can actually see how the whole thing evolves, which does not make it nonetheless less hilarious.

So why these differences? The individual translation pairs are picked randomly, and will be different for each iteration. I checked the translations pairs that I actually understood, and detected no errors. Unfortunately, language is not unambiguous, and a same word can often be translated by several different words. But that does not mean that the translation in the different direction is the same…

There are a lot of hilarious translations on that site. One that caught my eye:

Original: “Machine translations are useful for getting a general idea about what text written in a foreign language means. However, “general idea” isn’t exactly accurate; the translator literally translates (word for word) the text which often results in grammar”

10 translations later: “Machine translation is the text of the general concept is a foreign language to write a useful resource. The idea of “general”, a literal translation of the letter was not very clear (and accurate) text, which often leads to G”

So I am not worried that machine translation will replace me anytime soon… 😉

Be Sociable, Share!