Sound Change and Analogy, Again: Brugmann's Law and the Hunt For O-Grades in Indo-Iranian*

This article revisits the interaction between regular ‘Neogrammarian’ sound change (defined as a purely phonological process) and subsequent morphological change (especially changes subsumed under the term ‘analogy’) in the development of the outcomes of Brugmann's Law (BL) in Indo-Iranian. The traditional formulation of BL states that Proto-Indo-European *o became Indo-Iranian /ā/ in open syllables and /a/ elsewhere, positing a purely phonological context of application. Alternatively, Kiparsky (2010) has argued for a revised version of BL in which the accent and ablaut properties of the affected forms play a role, hence essentially for a synchronic morphophonological rule. I argue that this revised version fails both from the perspective of comparative reconstruction and as a synchronic rule based on a detailed study of the forms cited as evidence for the revised rule. Rather, in order to identify inherited *o in Indo-Iranian the effects of the ‘blind’ Neogrammarian rule must be separated from the synchronic morphological rules of the attested languages. This paper thus makes a methodological contribution in defence of the Neogrammarian approach to sound change, but also an empirical contribution by showing that this approach, in combination with a strictly lexical definition of analogy, can account for some conspicuous exceptions to traditional BL.