Indogermanische Forschungen. — 2015. — Vol. 120. — p. 269-278.
Abstract: In this paper the author argues that Szemerényi’s Law operated in the proto-language and that its operation should be restricted to *-Rs# and *-RH# sequences only. Moreover, it is shown that the development of *-Ts# > *-ss# > *-s#, often assumed as part of Szemerényi’s Law (cf. Neri 2003: 2035; Lipp 2009: 2, 93) as in PIE *pód-s > *póss > *pṓs > Doric Greek πώς ‘foot’ (Hesych.), is hardly credible (cf. also Kim 2001: 12724 but without further discussion). The length of the vowel in the Doric form is not certain (cf. LSJ⁹: 1456) and surely secondary in Attic poús as the short vowel is found in compounds (cf. Homeric trípos). Furthermore, if Szemerényi’s Law operated as *-Rs# > *-R# we would expect such sequences to appear as *-T# < *-Ts#. I will show that such cases are secondary rather than direct outcomes of Szemerényi’s Law.
Keywords: Szemerényi’s Law, lengthened grade, historical phonology