Russian Yers and Prosodic Structure

In this paper I investigate a previously neglected aspect of the phonology of Russian vowel-∅ alternations involving verbal prefixes and prepositions. There is evidence both for and against the notion that prepositions and verbal prefixes — which I will call P, for short — have identical phonological behavior. There are at least two processes that suggest that P forms a unified phonological category. Final devoicing and vowel reduction both treat the unit consisting of P and the following material as a single phonological word (for a detailed argumentation, see Matushansky 2001, Steriopolo 2007, Gribanova 2009, 2010). Crucially, P is unified for the purposes of these two processes, in that prefixes and prepositions are phonologically indistinguishable. On the other hand, the vowel-zero alternation known as yer realization appears to distinguish the two categories (Gribanova 2009). This leads Gribanova to posit a stratal analysis where the two members of P attach at different levels in the phonology. In sum, the diagnostics on the nature of P are paradoxical. The key to solving the paradox is that the empirical complexity of the yer facts has been underappreciated. Here I take a closer look at the data on yer realization with prepositions, using both the Russian National Corpus (RNC) and Google. Once the facts are sorted out, and once the intricate phonotactic, syntactic, and lexical factors that affect yer realization are brought to light, it is possible to resolve the two paradoxes by inferring the correct prosodic representation of the structures. In the following section I outline the yer facts relevant to the phonology of prepositions, and propose a prosodic analysis. Section 3 summarizes the facts of the so-called rule of stress retraction, whose behavior parallels that of yers in relevant ways, and supports the analysis. In Section 4, I argue that the behavior of yers and of stress retraction in verbal prefixes, while superficially dissimilar to that in prepositions, in fact shows the same characteristics, and must be analyzed in the same way, vindicating the phonological unity of P.