This paper proposes that an account of PP-stranding in ellipsis in non-PP-stranding languages needs to take into account how prominence via Focus/Contrast is computed. The proposal here is that prominence is computed on a phase-by-phase basis (Selkirk and Kratzer 2007). Cases of PP-stranding in Polish will be argued to be instances of non-constituent ellipsis (Bruening 2015), but with one remnant. PP-stranding is possible when there is a smaller phase than the PP where prominence can be established. This is the case when the remnant of sluicing is a D-linked wh-phrase (but not a simple wh-phrase). When prominence needs to computed in more than one phase within the domain of ellipsis, as is the case when there are multiple remnants, then one of the remnants needs to move or be an adjunct. In either case PP-stranding becomes impossible as is shown in non-constituent ellipsis, multiple wh-movement ellipsis, and in sprouting.

Abstract from FASL NYU program