Hate-inducing language, which has become a recurrent decimal in Nigerian socio-political discourse, is not unconnected to the deep-seated boundaries existing amongst different ethnic groups in Nigeria. Linguistic studies on hate language in Nigeria have largely utilised pragmatic and critical discourse analytical tools in identifying the illocutions and ideologies involved but hardly paid attention to the metalinguistic forms deployed in hate speeches. Therefore, the present study, aside adding to the research line of the Natural Semantic Metalanguage (NSM)—which has unduly focused on language typology, explores the metalinguistic evaluators that index hate speech in Nigeria, and relate them to specific pragmatic strategies through which hate speech producers’ intentions are communicated. To achieve this, three full manuscripts of hate speech made by three groups (i.e. Arewa Youth Consultative Forum, Youths of Oduduwa Republic, and Biafra Nation Youth League) from three (northern, western, and eastern, respectively) regions of Nigeria are purposively sampled from Google directories and Radio Biafra archives, subjected to descriptive and quantitative analysis, with insights from the NSM theory and aspects of pragmatic acts. Two categories of metalinguistic evaluators were identified, positive (GOOD) and negative (BAD) evaluators; and these are associated with three pragmatic strategies; namely, blunt condemnation, unshielded exposition, and appeal to emotion. While the condemning and exposing strategies largely utilise negative evaluators in initiating hate on target groups, the emotion-drawing strategy largely employs positive evaluators in boosting the image of the hate-speech producing group in the eyes of the audience. With these findings, the study takes existing scholarship on violence-inducing language a step forward, especially in providing a pragmatic explanation to the proliferation of hate crimes in Nigeria. It also offers a holistic linguistic database and critical meta-language for the teaching of hate-related language and crime, especially in second-language situations.