Demos study shows 14,000 Tweets a day contain racial slurs

A British research organization trawled nine days of Twitter's live feed looking to see just how people use racial slurs online.

Their results are counter-intuitive. While more than 14,000 messages a day typically have an identifiable racial slur, that's less than one in 15,000 English-language messages, according to Demos' Anti-Social Media study published this week.

The most common slur used: White boy.

Also common were "paki," "whitey," "pikey," "nigga," "spic," "crow," "squinty" and "wigga." Roughly half of the instances involve the use of derogatory language "in a non-offensive, non-abusive manner, to express in-group solidarity or non-derogatory description," Demos wrote. These top ten contained about 87 percent of racial slurs in the sample. "Relatively few tweets – from 500 to 2,000 per day – were directed at an individual and clearly abusive."

Demos noted that racial abuse online is doubtlessly broader than the sample, since many people hurl racial invective without using obvious racial slurs. But if someone uses Twitter to call someone a racially-offensive name, in the public feed, in a way that is clearly abusive, it's incredibly rare and probably notable since fewer than one in 75,000 messages contain language used that way.

Demos researchers led by Jamie Bartlett used a combination of natural language processing algorithms and human analysts to categorize subset samples of the data. The team used Wikipedia's crowdsourced terms considered derogatory as the sample set.

The classification of terms appears to have been fraught with mixed-messages and unclear cultural context, analysts noted. Many messages contain "casual use" slurs that reflect social norms within the sender’s personal community, but not overt hostility.

"The racially-prejudiced tweets appear not to be uniform in nature; we suggest that a proportion of such tweets might be described as showing a ‘casual use’ of slurs," Demos wrote. "They contain phraseology that might be deemed insulting, abusive, or threatening, and they use terms in a prejudicial fashion, but the terms are used in an off-hand or casual manner. ... It suggests underlying racism, but conveyance of that racist viewpoint is often not the point of the message."