What Old Frisian Can Tell Us about the History of i-Umlaut across West Germanic

Joseph C. Salmons

INTRODUCTION In Germanic historical linguistics, and perhaps in historical linguistics gene- rally, there is a clear and longstanding bias toward the oldest data, and per- haps even toward data from ancestors of the languages with the most widely spoken modern descendants.1 As a result, relevant data from languages like Old Frisian have at times been ignored by scholars making far-reaching claims about Germanic. While Old Frisian is robustly attested only much later than , and , it can nonetheless fill in critical pieces of broader West Germanic linguistic history, as Smith shows in her comparison of Old Frisian ‘vowel balance’ to patterns of syncope and apocope in Old Saxon, Old English and Old High German.2 When it comes to i-umlaut, Germanic historical linguists might expect to find virtually nothing, given the salient and early general of /a/ in Frisian, as well as the early unrounding of the reflexes of mid and high umlaut outputs.3 In this paper, I argue that Old Frisian data can provide decisive support for one of two competing views about the unfolding of West Germanic i- umlaut. One view sees umlaut as having arisen extremely early, well before

1 This paper grows out of years of work and discussion with Greg Iverson, as well as Rob Howell, Dave Holsinger and Laura Catharine Smith, among others. Additional comments and discussions directly on this paper came from Greg Iverson, Monica Macaulay, and I happily acknowledge that help. The comments of an anonymous reviewer were greatly helpful as well. 2 L. C. Smith, Cross-Level Interactions in West Germanic Phonology and Morphology. Ph.D. dissertation, University of Wisconsin-Madison, 2004. 3 C. V. J. Russ, ‘Umlaut in Frisian and the : Some Thoughts on its Genesis and Development’, NOWELE 28–29 [= Festschrift Nils Århammar] (1996), 501–12, provides an excellent general overview of how Frisian umlaut fits into the broader picture ofGermanic, but that paper was written before the post-Twaddellian view (as defined below) was widely known, and thus he does not focus on the core issues treated here. 368 Salmons the dawn of directly recorded Germanic linguistic history, even Proto-Ger- manic, while the other view dates umlaut back to the dialectal period, ca. 7–8th century C.E., just before the emergence of our earlier extensive written documents in Germanic. I aim to show that Old Frisian evidence underpins the latter view, namely that umlaut was an emphatically dialectal pheno- menon, in line with its traditional dating in Old Frisian and across West Germanic. Umlaut cannot date back to earliest Germanic, as argued bymany in the classic American Structuralist tradition (Antonsen, Penzl, others). This paper thus brings the chronology of Old Frisian sound changes to bear on Germanic umlaut, following conservative and virtually theory indepen- dent assumptions about Frisian linguistic history. More importantly, it adds a distinct new type of argument to the debates over Germanic i-umlaut, namely from relative chronology, and does so in a theory-neutral way. After a brief sketch of the two just-noted and fundamentally opposed views of umlaut (§1), I recapitulate some generally accepted points of Frisian phonological history(§2), showing that Frisian palatalization of velar stops had to pre-date the rise of even allophonic umlaut. In §3, I briefly out- line arguments that changes in low vowels must have preceded umlaut, and that monophthongization of West Germanic *ai and *au almost surely did as well. §4 concludes. While what follows may be largely a case of stating the obvious to - nists, the evidence reviewed here and its implications for a widespread view of umlaut across Germanic seem remarkably unknown among the broader community of Germanic linguists, at least on the topic of umlaut, and this, I hope, shows how Old Frisian data can be directly important for Germanic comparative linguistics.

1. TWO COMPETING VIEWS OF UMLAUT For present purposes, let us distinguish three approaches to how umlaut unfolded, with Twaddell’s structuralist view chronologically in the center, flanked by earlier traditional or pre-Twaddellian views on one side and post- Twaddellian views on the other. I will focus on the Twaddellian or American Structuralist view of umlaut, and how it compares to a newer post- Twaddellian or ‘Wisconsin School’ view. The turning point at b, Twaddell’s note of 1938,4 anchors an analysis of Old High German umlaut firmly in classic phonemic theory, rigorously distinguishing the behaviour of condi- tioned or allophonic variants from the behaviour of independent phonemes. Whereas many earlier views had assumed that umlaut had unfolded gradually, one very direct implication of Twaddell’s groundbreaking article

4 W. Freeman Twaddell, ‘A Note on Old High German Umlaut’, Monatshefte für deutschsprachige Literatur und Kultur 30 (1938), 177–81.