Research indicates both similarities and differences in the rhetoric employed by far-right and violent jihadi extremists. This article employs natural language processing and linguistic analysis to investigate identity construction in selected far-right (FR) and jihadi extremist (VJE) English material. We assess inter-group differences at the level of the subject, e.g. “us,” “you,” and “them,” along linguistic dimensions such as moral foundations, psychological processes, and social processes. The findings reveal that the construction of the “us” identity in both FR and VJEs demonstrates a strong presence of social processes while showing a relative absence of cognitive and analytic features compared to neutral data. Conversely, features associated with cognitive processes and morality are more prevalent in the construction of audience identities and the “other.” Additionally, both the FR and VJE corpora exhibit a stronger focus on the past compared to the neutral baseline. We also discuss specific linguistic features critical for identifying extremist subjects, namely “us,” “you” and “them.”