Binary tense systems (Verkuyl 2008) assume a basic distinction between two temporal domains that relates to the time of speech. In his discussion of the present domain, Verkuyl (to appear) discusses an extended present that is split into an actualized part and a non-actualized part. This notion of present allows to explain the uses of present tense that applies both to phenomena already actualized as well as to phenomena that are still to be realized, as in Dutch, German, and also English English (cf. Copley 2008 for “futurates"). The zero-dimensional boundary (Varzi 2008), the horizon that splits what has happened from what will happen is contained within the domain of the present. In my talk I will point out that there are languages that opt for a different choice, where this horizon seems to be located outside of that domain. These are languages that are traditionally described as having a nonfuture — future distinction, or more appropriately, a realis — irrealis distinction. I will give a very short overview of this distinction and then describe its workings with data relating to my own field work on the language Daakie spoken in Ambrym, Vanuatu. This will be semantically modelled in a frame with branching times that captures not only what is traditionally understood as irrealis (better, potentialis), but also other domains such as unrealized past and unrealizable future developments (so-called distals).