Abstract
The paper deals with offensive language gleaned from Twitter, encoded by (abusive) insults and verbal reactions to them. The study aims to analyse how they are constructed in terms of (i) the cognitive strategies employed by the insultees, (ii) rhetorical figures and (iii) linguistic devices used by both the insulters and insultees. The tweets are illustrative of the following shifts occurring in responses relative to insults: register clash, changing explicit insult into implicit, figure/ground reversal, syntactic echoing, changing abusive language into jocular or into an ironic insult, etc. The cognitive strategies employed by the insultees comprise: ignoring the insulting content, agreeing with the insult, and attacking the insulter with an explicit or implicit insult.