16 Jul 2013

Social Media and Machine Translation: A Recipe for Disaster?

This month Twitter began translating Arabic language tweets from high-profile Egyptian politicians. Following the recent political upheaval, the former president Mohammed Morsi, opposition leader Mohammed ElBaradei and Arab Spring activist Wael Ghonim all had tweets made available to non-Arabic speaking readers. You can view the list of popular Egyptian Twitter accounts being translated here.

The tweets were translated using Microsoft’s Bing Translator, a popular machine translation tool used every day by millions of people around the world to translate text into different languages.

“As part of our experiment with Tweet text translation, we’ve enabled translation for some of the most-followed accounts in Egypt, so people around the world can better understand and keep up with what’s happening there,” a Twitter spokesman said in a statement.

However, Twitter admitted that this new functionality was very much at the experimental stage and that the service had not been officially launched or expanded into other languages or territories.

Facebook has offered a translation function (also powered by Bing Translator) for some time. If you’re an English-speaker reading a Facebook public page and encounter a comment in Spanish, for instance, you’ll see a ‘translate’ button next to it, which once clicked allows you to view the comment in English in a pop-out window.

There’s no question that machine translation is incredibly efficient at translating foreign language content. However, given the unreliability of machine translation services, is the automation of social media translation such a great idea?

There have been many recent cases of Twitter and Facebook users getting into trouble for comments they’ve posted on the social media networks.

Sally Bercow, wife of the Commons Speaker John Bercow, had to pay damages to Lord McAlpine, after the High Court found that a tweet posted by her which falsely linked him to an allegation of child sexual abuse was highly defamatory. The tweet in this case was “Why is Lord McAlpine trending? *innocent face*” – in reference to allegations by Newsnight that the high-profile Tory politician was a paedophile.

Mrs Bercow said she had “promptly apologised publicly and privately to Lord McAlpine for the distress I caused him. I also made two offers of compensation.”

“The High Court found that my tweet constituted a serious libel, both in its natural meaning and as an innuendo,” she said. “To say I’m surprised and disappointed by this is an understatement.

“However, I will accept the ruling as the end of the matter. I remain sorry for the distress I have caused Lord McAlpine and I repeat my apologies.”

In another case, a Twitter user called Paul Chambers from Doncaster was convicted in May 2011 of sending a “menacing electronic communication”.

The message Chambers tweeted stated:

“Crap! Robin Hood airport is closed. You’ve got a week and a bit to get your shit together, otherwise I’m blowing the airport sky high!!”

He was found guilty by Doncaster magistrates, fined £385 and ordered to pay £600 costs. He claimed it was a joke and eventually had his conviction and sentence quashed. However, this served to highlight how causing offence on social media can land you in quite a lot of trouble.

So can you trust a machine to translate your social media messages without inadvertently causing offence by not understanding humour or sarcasm?

Also, if social media interactions are translated from the source language into different target languages and this is repeated many times, is there not a real risk of machine automated Chinese whispers?

We’d like to hear your thoughts.

 



 
 

Sign up to our newsletter

Get our blog articles straight to your inbox.