Propaganda and mis/disinformation in the media has been identified in the social sciences literature as a major threat to the functioning of democratic society. While mis/disinformation is false by definition, the target of my research is any language that is intended to modify the beliefs or behaviors of the information consumer. This kind of language is manipulative, but does not have to be false and is referred to as propaganda. Propagandistic text is often characterized by the use of rhetorical devices and linguistic style designed to exploit the reader’s cognitive and emotional biases. The aim of my research is to automate the extraction of certain rhetorical devices, syntactic features and word choices as identified in the literature on persuasive language. These features can then be utilized to train machine learning models whose predictions can be explained using the rhetorical features. To take advantage of the advances in deep learning while maintaining the interpretability of symbolic systems, I aim to leverage neuro-symbolic AI (NeSy) towards interpretable propaganda detection in news text.