One of the features of Uli is the redaction of slurs. Why did we build this? Not all abuse is explicit but the use of slurs, coded language and dog-whistling to target gender minorities is common on social media. And we have become normalised to it.
First, there are explicit slurs. Most slurs, even if undirected, are sexualized. Many are used to target members of the dominant group by describing their relation to women (mother or sister).
Dog-whistling is the use of phrases/terms, possibly understood only by an ‘inner group’ to target a community. It is a form of coded language. But coded language can also include misspellings or the use of words that sound similar to a slur.
This is especially common in mixed-language communities — one can use English words that sound similar to a Hindi or Tamil slur.
Such language is so pervasive on social media that one doesn’t even register when one encounters them. One is also likely to say things online that one would not say to someone in person. The slur redaction feature interrupts the unconscious scrolling to call attention to the presence of such language.
The slur list crowdsourced for Uli can be found here. A search on Twitter with some of these terms will reveal just how abusive Twitter can be to individuals of marginalised genders.
Coded language evolves rapidly. Someone comes up with a novel way to evade automated moderation everyday. And now we’re looking at a world where the new owner of Twitter wants to pull even further away from detecting and acting on such content.