The project aims to reconcile conventional assumptions regarding the licensing of Negative Polarity Items (NPIs) with observed exceptions, with a focus on English, German, and Romanian. The initial hypothesis posits that weak NPIs necessitate at least a weak licenser in the at-issue meaning, while strong NPIs require a strong licenser, which may appear in either the at-issue or non-at-issue content. The project’s empirical approach will combine quantitative and qualitative data analyses, including acceptability judgments. The ultimate objective is to develop an empirically grounded theory of NPI licensing within a constraint-based grammar framework.