Freedom of expression on social media often gives rise to conflict, debate and even division in society. One of the social media features is the comments column. Apart from being a positive tool, this can also be used as a negative tool which can lead to language wars. A form of language war can be found in the comments column, namely in the form of hate speech. Hate speech texts usually contain insults, defamation, blasphemy, provocation, incitement and unpleasant actions, as well as other aspects regarding SARA and people with disabilities. Hate speech can take the form of words, phrases and sentences depending on the context. To be able to understand hate speech in comments, it needs to be studied using semantic and pragmatic studies. This research aims to describe the forms of illocutionary speech acts in comments on social media. The method used in this research is descriptive qualitative. The data in this research are words, phrases and sentences containing illocutionary speech acts, while the data source is hate speech comments. Data collection techniques use documentation techniques. The data analysis technique was carried out using descriptive methods. The results of this research show four forms of illocutionary speech acts, namely assertive acts, directive acts, expressive acts, and commissive acts. Assertive actions are used to bind speakers to the truth of their speech, such as stating, insinuating, reporting, showing, and accusing. Directive acts function to encourage the other person to do something, both verbally and nonverbally, such as begging, demanding, ordering, suggesting, requesting, and opposing. Expressive acts function to express the speaker's feelings towards the person he is talking to, such as congratulating, apologizing, criticizing, praising, criticizing and complaining. Commissive actions refer to the interests of the interlocutor, such as promising, offering, swearing and threatening. Keywords: illocutionary speech acts, comments, semantics, pragmatics