It has been hypothesized that separable complex verbs (SCVs, for example, ópzoeken ‘look up’) and inseparable complex verbs (ICVs, for example, doorzóeken ‘search’) form part of the same historical development, SCVs representing a stage intermediate to constructions with syntactic resultatives and ICVs. This paper shows that such a hypothesis is untenable, since many SCV preverbs and ICV preverbs are nonpredicative and thus semantically different from resultatives. Instead, it is claimed that predicative preverbs and nonpredicative preverbs result from two independent historical developments. In addition, the particular semantic and structural properties of SCVs are assumed to suggest a specific SCV structure, to be positioned in between syntactic phrases and morphologically complex words.I would like to thank Geert Booij, Hana Filip, Beth Levin, Andrew McIntyre, Ariane van Santen, and two anonymous reviewers for their valuable comments on earlier versions of this paper. Of course, any remaining errors are entirely my own.

This paper pursues an “ingenerate” or phonetically based account of i-umlaut as it unfolded in North Germanic. We focus on a famous problem relating to umlaut distributions in i-stem nouns: In the long stems of that class (gestr ‘guest’, from earlier *gastiz), where umlaut is arguably less motivated phonetically, it is generally reflected throughout the paradigms, but in short stems (staðr ‘place’, from earlier *staðir), where it is more expected, umlaut is generally absent. A central feature of our understanding of these and other Norse facts is the interleaving of processes of sound change and analogy, the latter of which, by an assumption validated elsewhere, comes into play only under extraordinary circumstances. In contrast to previous work on the conundrum of umlaut in Old Norse, we situate this account in the context of umlaut as a general phenomenon, with parallels in development to that of its West Germanic sisters.Besides many members of the audience at the XVIth International Conference on Historical Linguistics (Copenhagen, August 2003), where a preliminary version of this paper was presented, we thank two anonymous readers for this journal and the following colleagues for their comments and discussion: Anthony Buccini, Rob Howell, Monica Macaulay, Richard Page, Michael Schulte, J. C. Smith, as well as Laura Catharine Smith and other members of the UW Phonology Group. As may become clear in the course of the paper, these individuals do not necessarily agree with what follows; of course, any shortcomings remain our own. The translations of Danish and Norwegian quotes are likewise our own.