The paper is full of grammatical mistakes and typos, and full of obvious errors in how they classify statements in their own examples. I don't think it 'confir

The paper is full of grammatical mistakes and typos, and full of obvious errors in how they classify statements in their own examples. I don't think it 'confirms' anything. And using a reddit dataset as a way to classify "abusive" language makes zero sense if you're using it to evaluate movies from the 1950s. What counts as abusive language is extremely context sensitive -- to give one example: the word "retarded" used to be seen as a euphemism and not a term of abuse. It'd be highly abusive in the 1980s or 1990s, but in the 1950s, it might be a kind way to refer to someone with a mental handicap.