20th International Conference on English Historical Linguistics, University of Edinburgh, Edinburgh, UK.
(with Warren Maguire)
The insertion of an unstressed vowel in lC and rC clusters of non-rising sonority (fil[ə]m), is widely known as characteristic of Irish English. It is frequently cited as a substrate feature carried over from Irish, where this process is productive (Ní Chiosáin 1999), as it is in the related Scottish Gaelic. However, this feature is also known from Older Scots, modern Scots dialects (Maguire 2017), and traditional varieties of English. Therefore, it might be an areal effect in the British and Irish Isles; notably, a similar phenomenon is also attested in Welsh (e. g. Schumacher 2011, Iosad 2017)
We reconsider whether areality best explains the prevalence of epenthesis in English. Epenthesis in various rC and lC clusters recurs throughout the history of English: Luick (1914) identifies two separate sound changes involving this epenthesis in Old English, with a third one in Middle English (Jordan 1934), all with different trigger contexts and patterns of quality of the epenthetic vowel. In fact, epenthesis in such clusters is extremely common across West Germanic (cf. Howell & Somers Wicka 2007): it is attested in Old High German (Reutercrona 1920, Braune 2004) and retained in Middle High German (Michels 1979) and many modern dialects (Schirmunski 1962); in Old, Middle, and Modern Dutch (van Loey 1976, Warner et al. 2001), and in Frisian (Steller 1928, Visser 2017). Epenthesis appears rarely in ancient North Germanic, but is widely attested at least in Danish dialects (Hansen 1962). Common to Germanic epenthesis patterns are a diversity of affected clusters and qualities of inserted vowel, a lack of regularity of epenthesis as a sound change, and erosion over time due to syncope processes targeting both epenthetic and original vowels. This makes it clear they cannot all be inherited from a single sound change in the protolanguage.
We argue that parallel developments of post-sonorant epenthesis across (at least West) Germanic represent an instance of drift arising from variation at an earlier stage (e. g. Joseph 2013). The recurring epenthesis ‘events’ are not separate, but instead are best seen as attestations of a variable phenomenon always present to some extent in the West Germanic languages. This makes them distinct from exceptionless excrescence, which arises from retiming of vocalic gestures (Hall 2006), as attested in Scottish Gaelic and in the Sámi languages. We argue that these differences undermine the case for epenthesis as an areal feature in the British and Irish Isles.