Abstract
An original unstressed vowel *ō generally develops to a in Old English. In some categories, however, both a and u are found as reflexes. The traditional explanation for this phenomenon posits that *ō developed to u when the following syllable also had a *u. A statistical analysis of the distribution of a and u in such forms in an Old English text finds no support for this theory. Since shortened vowels tend to raise and unstressed vowels in medial syllables are shorter than in final syllables, I hypothesize in this paper that *ō shortened and raised to u in medial syllables. A statistical analysis of the same text strongly supports this new hypothesis. This vowel raising process can be explained by neuromuscular and perceptual properties. Producing short low vowels increases the risk of undershooting the vowel target, which translates into vowel raising. At the same time, a shortened low vowel can be misperceived as a higher vowel, since high vowels are shorter than low vowels. With these explanations, a formal grammatical analysis is not needed to explain why vowel raising takes place.
Vowel reduction in verbs in King Alfred's Pastoral Care