The Colombian peace agreement, signed in 2016 between the government and the Colombian Revolutionary Armed Forces (FARC), claimed to put the victims in the center by building a comprehensive system of transitional justice, a Comprehensive System of Truth, Justice, Reparation and Non-repetition. To evaluate the declared centrality of victims, we aim to understand the language around them used during the negotiations that took place from 2012 to 2016, by analyzing all documents published in Havana using combined natural language processing techniques and a close reading of some key documents. Our argument is that, in order for language to become an ameliorating factor of the transitional justice process, language around victims included in the peace agreement needs to pass the language test that guarantees its effectiveness even beyond the end of the conflict. However, in the case of Colombia, results show a small statistical presence of victims in the conversations that contrast with the expressed statement about making them the focus of the agreement and the attempt to expand its description including the diversity of identities underneath this condition. We recommend that peace processes use data analysis techniques to ensure that the discourse really reflects the intention of the parties. This would avoid any gap between goals and semantics, facilitating the deployment of the post-agreement legislation in a fashion that closely reflects the parties’ intentions and the victims’ rights.