ИСТИНА

Since at least (Chomsky & Halle, 1968), stress has been taken as the domain that is most evidently expected to require cyclic rule application in rule-based phonology (which are preferable to constraint ranking on independent grounds (Vaux, 2008)) — including, of course, in Russian (Halle & Vergnaud, 1987, pp. 81– 83, 97–98) (Melvold, 1989). However, phonological cycles do not match with syntactic cycles, aka phases (D’Alessandro & Scheer, 2015); and if that were a feature of variably applied Phase Impenetrability Condition, as the paper suggests, we would expect cycle-bound postsyntactic operations to match with phonology rather than syntax, which is plainly not true in Distributed Morphology (Halle & Marantz, 1993), where a number of operations are phase-bound (for instance, Amalgamation (Harizanov & Gribanova, 2018)) but none are bound by phonological cycles. In Nanosyntax (Starke, 2009), of course, there are no postsyntactic operations to speak of, yet spell-out itself is so intertwined with syntax that it is difficult to understand how it can generate phonological cycles unless one accepts the idea (espoused by (Melvold, 1989)) that all affixes are cyclic. Against this background, it is high time to re-introduce boundary segments. Most of their rebuttals (so in (Rotenberg, 1978) and (Steriade, 1982)) crucially rely on accessibility of syntactic structure to phonology, which is doubtful on independent grounds (Scheer, 2008) and doesn’t explain why postsyntactic agreement is structure-insensitive, as shown by Wurmbrand (2016). One then has to show that stress can be accounted for in generative phonology without cycles, using boundary segments. There are pre-generative (and thus obviously non-cyclic) descriptions for Russian stress, most famously (Zaliznyak, 2010). I shall set out to show that their main insights can be adapted into cycle-less generative phonology without losing generality.