The roles of phonology and analogy In Old English high vowel deletion

Abstract

The problem under consideration is the exact nature of the application of High Vowel Deletion (HVD) in Old English. As discussed in §1, according to what has come to be the most prevalent view, a form such as nom.acc.pl. hēafdu‘heads’ in Late West Saxon, from earlier *hēafudu, is the phonologically regular result of the application of HVD. In §2 a recent alternative explanation is discussed, whereby *hēafudu should have produced *hēafd, which was subsequently reformed to hēafdu in West Saxon on the basis of analogy. Some initial difficulties that confront this latter explanation are discussed. The earlier analysis of Eduard Sievers maintains that neither of these analyses is correct, and that hēafudu, one of the forms actually attested in early texts, including the Vespasian Psalter, represents the purely phonological result of HVD, which applies vacuously to this form. Evidence is adduced in §3 demonstrating that the treatment of the phonological results of HVD in interrelated declensional categories in the dialect of the Vespasian Psalter are preserved with impressive conservatism, evincing little or no analogical disruption. This conclusion lends strong support to Sievers’ analysis of forms like hēafudu and renders it extremely improbable that West Saxon hēafdu can be anything but an analogical creation or that the proposed *hēafd could ever have existed. In §4 a rationale is offered for the changes, analogical or otherwise, that must be assumed for West Saxon in forms like hēafdu, as well as in ja‐stem neuter nouns with plurals like rīċu, wītu, along with feminine nouns in Germanic *‐iþō, such as and .

Early Old English Foot Structure

“…We have seen that a number of theorists have attempted to describe syncopated hēafdu under the assumption that this is the phonologically regular form of this word. Much recent work, however, has shown that the oldest form of this word is most likely hēafudu , with preservation of both final vowels (Bermúdez‐Otero : 32‐4; Fulk ). The attestations of * hēafudu show considerable variation, with hēafudu , hēafdu , and hēafod all occurring (with further variation between u and o , and sometimes occurring with a distinct ending ‐a ).…”

“…The importance of foot structure in conditioning the loss of high vowels in Old English can be simply illustrated by the nom.‐acc.pl endings of neuter a –stem nouns (Hogg [] 2011 : 220). The pre‐OE form of this ending was * ‐u , and is preserved as such after initial light syllables: fat–u ‘cups’ < * fat–u .…”

“…Such words show a great deal of variation in what vowels are actually found in our texts, and it is not always immediately obvious which developments are straightforwardly phonological, and which have been altered morphologically. The single lexeme hēafudu alone is found in three of the four logically possible variants: hēafudu , hēafud , and hēafdu ; only × hēafd does not occur (Fulk : 137). A robust theory of high vowel deletion needs to not only describe a coherent original phonological process, but also provide a plausible morphophonological account of the later attested forms.…”

“…Ringe () has argued for a far more sweeping loss of high vowels, with a consequently greater role for morphology in restoring lost vowels. The opposite approach is taken by Bermúdez‐Otero (), and Fulk (), who have both made compelling arguments that there was no vowel loss at all in hēafudu , with both unstressed vowels retained phonologically. This last view has the most to recommend it as representing the oldest pattern, both textually and in terms of the morphological assumptions required to account for variants.…”

Review of Ringe & Taylor (2014): The development of Old English

“…Ringe argues that the consensus view that syncope and apocope of unchecked *ĭ and *ŭ after stressed heavy syllables apply at the same time (Campbell 1962;Hogg 1992;Fulk 2010) is incorrect, and that syncope in fact preceded shortening of word-internal unstressed long vowels, which preceded apocope ( §6.8). This predicts that the outcomes of regular sound change applied to prehistoric *hēaβudu "heads" should be hēafd, *mae riþu "fame" should be mae rþ, etc.…”