Given the increase of political polarization in social networks, particularly during electoral campaigns, this study sought to determine the levels of affective polarization in political communication through the temporal perspective, delegitimizing beliefs, and intergroup emotions. Two studies were carried out: the first in the context of the elections in Andalusia (Spain) and the second during the Colombian presidential campaign. Both studies took the tweets of candidates and Twitter users who commented on them. From linguistic analysis, an index was developed to measure affective polarization in linguistic pieces, such as tweets. It was found that in contexts without political violence, it is common to implement communicative strategies with high indicators of ingroup bias, which implies a lower polarization, but increases with positive emotions and a focus on the future. Whereas, in contexts of political violence, the strategy focused on discrimination of the outgroup and its delegitimization, in which positive emotions decreased polarization, while the focus on the past increased it. It was concluded that affective polarization could be an indicator the disposition to reconciliation processes in contexts of violence, and a map of emotions associated with polarization is provided.