Polish is one of the languages which can employ V-stranding ellipsis in answers to polar (*yes/no*) questions (V(erb-echo) answers; see esp. Holmberg 2015, and see Ruda 2014; Mendes 2020; Arregi & Pietraszko 2021 for some (preliminary) discussions of V-stranding in Polish). Having revised evidence that Polish V-stranding in this case results from X[+V]P ellipsis (as opposed to *pro*-drop/argument ellipsis; see esp. Landau 2018, 2020a,b for relevant discussion), including facts related to indefinite, &P, QP, and NPI subjects (see also Mendes 2020 and references therein), I show that the elided projection is much lower in Polish than what is standardly assumed for this type of V-stranding (TP ellipsis; see e.g. Holmberg 2015; Gribanova 2017; McCloskey 2017; Mendes 2020; Sato & Maeda 2020; see also Holmberg 2015; Gribanova 2020; Portelance 2020 for low VP ellipsis analyses). In particular, while determining the category of X[+V]P in V answers in Polish is a challenging problem due to the inadequacy of the standard diagnostics for determining the height of V movement in the language (see Ruda 2014 and references therein), based on arguments from the answering options, I show that in Polish X[+V]P cannot be higher than AspP and based on feature mismatch options, I show that it is lower than AspP and at least in some contexts can be lower than VoiceP. This provides empirical evidence against Landau’s suggestion that the movement of X across a spell-out boundary (i.e. outside the complement of a phase head, e.g., V-to-Voice movement, where Voice is a phase head) bleeds XP ellipsis (e.g. V-stranding VP ellipsis). To this I add a theoretical argument: if ellipsis is a cyclic operation (see Murphy & Müller forthcoming; Sailor 2021), as it needs to be in a phase-based derivation (otherwise the complement of the first phase (e.g. VP) would have undergone morphophonological processing before ellipsis could apply, ellipsis currently conceived of as non-pronunciation licensed by syntax (see Merchant 2001; Aelbrecht 2010; Saab forthcoming a.m.o.)), any V-stranding X[+V]P ellipsis necessarily employs the step which is forbidden in Landau’s system: V needs to move to Voice and this needs to be followed by VP ellipsis (V is stranded, so it escapes the ellipsis site and VP is the complement of a phase head, so its non-pronunciation needs to be licensed before/at its spell-out).