To avoid social media police, Indian trolls go vernacular

When Shruti Agrawal retweeted Delhi student activist Gurmehar Kaur's posts on social media, she found herself the target of vicious, obscene messages and rape threats.Reporting the trolls did not work, because many of her abusers were attacking her in Hindi, which evaded Twitter's filtering process. In another incident, a well-known journalist from Chennai was targeted when she tweeted that she did not find Tamil actor Vijay's latest movie interesting.She was soon at the receiving end of more than 45,000 derogatory tweets, many of them in Tamil. Many of the software professionals who were part of the troll wave were making use of the language loophole.
The regional-language trolling problem is amplified by the fact that India has more than 122 languages and 1,599 dialects. An overwhelming 234 million Indians access the internet every day in regional languages, compared to 175 million English users in 2016, according to a Google-KPMG report.
To combat this, firms like Facebook have started employing native languages speakers to monitor content. We have real people (not bots) looking at reported content, 24 hours a day, 7 days a week. We have hired a lot of native language speakers as they are better able to decipher the true meaning of words and the context,“ says Ankhi Das, public policy director, Facebook India, South & Central Asia.
Twitter has not responded to calls and emails seeking comment.
Another new issue cropping up is preying on minors via social media using nonEnglish content. Youngsters, who are depressed, at times send SOSs to potential “masters“ or “curators“ in Romanian or Russian using Google Translate. This way they cir cumvent filters set to monitor English language content. These curators are the ones that usually set tasks for participants in the suicidal Blue Whale Challenge.
The European social network VKontakte (where the notorious Blue Whale challenge is said to have originated) and Facebook are combating this scourge with all linguistic and cultural know-how. A hashtag search for depression, body cuts or razor art will immediately have Instagram (owned by Facebook) pop up with: “Can we help?
Posts with words or tags you're searching for often encourage behaviour that can cause harm and even lead to death. If you're going through something difficult, we'd like to help.“
The Data Security Council of India (DSCI) is also working with the ministry of electronics & information technology for ways to battle this. “We have recommended that the ministry engage with social media service providers to identify any groups that violate their usage policy . There need to be broader appeals to internet users to refrain from forwardingdiscussing about any such viral messages that originate from a non credible source,“ says Ra hul Sharma, senior consult ant, DSCI.
Knowledge of regional language is necessary to gauge the nature and intent of certain posts. A 21-yearold woman in Salem, Tamil Nadu killed herself last July after being at the receiving vicious posts with obscene written material and morphed nude photographs on Facebook.
To combat tragic incidents like this, Facebook is focusing on improving monitoring in regional languages. “We are actively working to remove content that appears to purposefully target private individuals with the intention of degrading or shaming them,“ says Das.