Some believe that there are two untils in English: one durative and apparently polarity-insensitive (1); and one punctual and requires a (local) licensing negation (2).
Advocates of this view (e.g. Karttunen (1974) and many others), have claimed that an alternative single-entry view would have to explain not only the distributional heterogeneity observed in (1,2), but also a number of differences in the inferences that are licensed by durative uses of until and by its punctual uses. In this talk I review some of the observations that have guided this debate, and examine Iatridou and Zeijlstra's (2021) recent unifying proposal.