Early Old English Foot Structure

Abstract

The variable operation of high vowel deletion in Old English has long been a point of difficulty, both descriptivelya prehistoric form like *h eafudu is attested variably as h eafudu, h eafdu, and h eafodand theoretically. Recent work, especially by Berm udez-Otero (2005b) and Fulk (2010), has indicated that plural forms like h eafudu are most likely original, but accounting for why the medial *u is preserved in this case form, and not in h eafde, the dative singular of the same word, has remained theoretically problematic. These difficulties arise from attempting to describe the prehistoric Old English process of high vowel deletion on the basis of later Old English phonology. At an earlier stage, the nominative-accusative plural *h eafudu could be exhaustively parsed into two precisely bimoraic feet: *[h ea][.fu.du]. The dative singular historically ended with a long vowel, *h eafudaē, in which the medial *u could not be accommodated within a bimoraic foot: *[h ea].fu[.dǣ ]. High vowel deletion is therefore best characterized as the deletion of unfooted high vowels in early Old English, initially operating while length in unstressed vowels remained contrastive. Both this quantitative system and the preference for precisely bimoraic units receive support from Kaluza's law, an archaic metrical phenomenon in Beowulf which prohibits resolution in secondary metrical ictus if the resulting unit would have more than two moras, and which is sensitive to prehistoric length distinctions. This original system was obscured, linguistically and metrically, in later Old English by the shortening of unstressed long vowels, triggering various morphological reanalyses of the effects of high vowel deletion. A review of these changes suggests that the system of metrical phonology described here provides a more plausible starting point for the reworkings that produced the forms found in later Old English than do alternative accounts such as those of Campbell (1983) or Ringe (2002). 12 Erfurt 275; spelled hurnitu in Corpus (CorpGl) 603. 13 Epinal 718; so also Corpus 1439. Erfurt has aebitu for aelbitu.

Reduction of unstressed vowels in Proto-Frisian and the Germanic ‘Auslautgesetze’

“…10 for Riustring Old Frisian, it can be mentioned that Early Old English shows a different i-mutation effect of *-i vs. *-ī/j (Versloot forthc.). Effects of vowel length contrasts in pre-Old English until a fairly late date are also mentioned by Goering (2016). *-ōn merged.…”

Kaluza's Law and Secondary Stress

“…The details are too complicated to treat here, but two initial points are worth making. First, *‐ æ always comes from long vowel sources (or from diphthongs) in earlier Germanic, and there is no linguistic evidence that shortening took place here earlier than for eOE *‐ ā ; that at least some instances of *‐ æ̅ retained their length is undisputed and linguistically supported (Goering 2016a: 178–80), so the question is whether we should reconstruct an extra wave of shortenings that affected some, but not all, instances of older *‐ æ̅ . It seems to me that this extra shortening requires more assumptions, and cannot be taken for granted without specific evidence.…”

“…From a phonological perspective, the strangest thing about resolution in Beowulf is not that LH sequences are sometimes suspended, but rather that words such as * su‐n ā , * wræ‐cæ̅ , and cy‐ning are ever resolved (Minkova 2021: 122‐8). Prosodically, Old English, like other Germanic languages, clearly shows a preference for moraic trochees, that is, prosodic units with exactly two moras, the first of which is stronger than the second (Kiparsky 1998; Goering 2016a; 2016b: 280–9). As a cross‐linguistic unit, the bimoraic trochee is very well established (Kager 2007: 203–204), and in cognate systems which feature resolution, such as Latin or Greek, the phenomenon is strictly limited to LL sequences.…”

“…This contrasts with other case forms, such as the dative plural * hēafudum or singular * hēafudæ̅ , both of which historically had the shape HLH: the middle L syllable could not be accommodated with in a precisely bimoraic second foot (the final LH sequence could not ‘resolve’), and so the medial * u was deleted, giving attested hēafdum , hēafde , etc. (Goering 2016a: 177–180).…”

“…Without this licence, resolution of an LH sequence is impossible. On the other hand, in (Gēa)t ā (cy‐ning) , the inflectional syllable is defooted (Goering 2016a: 179–80), meaning that cyning is initial to a new sequence of feet, and so resolves with the overheavy licence.…”