Alliteration and Sound Change in Early English (review)

This remarkable book is a history of English syllabic onsets with special reference to the poetic treatment of onsets known as alliteration. The book makes diligent use of recent philology and goes beyond it to find new tests for important claims about syllable structure.

Minkova begins with a summary of requirements for alliterating syllables (Chs. 1, 2, and 3). In Old English poetry, these must always be stressed. Matching of the first consonant in the onset is often present and often suffices. There are two interesting wrinkles, however: (i) any stressed syllable spelled with an initial vowel alliterates with any other such syllable, and (ii) in the clusters sp-, st-, and sk-, both consonants must match. Chs. 4 and 5 provide full discussion of these wrinkles and their linguistic implications. In Middle English poetry, syllables spelled with initial vowels are less frequently employed for alliteration (166). Although sp-, st-, and sk-still show a strong tendency to alliterate as clusters, we find a significant number of alliterations that match only the initial s- (239). At the same time, the frequency of bisegmental matching is enhanced in a variety of other clusters, most commonly in those with initial s-, but to a significant extent also in clusters like br- and gr- (306). The highly favored yet less strict cluster alliteration of sp-, st-, and sk- seems to have been reanalyzed as a violable preference for cluster alliteration generally. A particularly interesting tendency is ‘splitting’, whereby the second consonant of an onset cluster in one alliterating syllable appears in the coda of its alliterating partner (300). In Ch. 6, M uses several kinds of evidence to argue that such discontinuous matching of cluster elements was more highly valued than ‘skipping’, in which the second element is simply not matched (301).

The sort of juncture displacement found in expressions like the ton and the tother (< that one and that other) plays a significant role during the later period. In the Middle English poems, we encounter a kind of ‘elision alliteration’ (167–69), whereby that on, for example, alliterates on the word-final [t] that moves at phrase level from the coda of that to the onset of on. In Old English poems, by contrast (145–50), the meter shows scrupulous respect for morphological word boundaries, which provide barriers not only to alliteration of underlying coda consonants but also to resolution, whereby a short stressed syllable and a following unstressed syllable become equivalent to one long stressed syllable (157). In Old English, verbal prefixes are unstressed, hence ineligible for alliteration, because they lie outside the phonological word that contains the major-category verbal root (179). In the Middle English poems, alliteration on these prefixes has become a significant option (57). It seems reasonable to conclude with M that the prefix and the verbal root were incorporated into the same prosodic word during the later period (180). On this hypothesis, variant pronunciations with stressed prefixes would be likely to arise, providing a phonetic basis for the change in alliterative practice.

M’s most valuable contributions to our understanding of syllable structure are situated in the later period. In Ch. 6, she provides many kinds of evidence for the claim that a pair of Middle English consonants with a high frequency of cluster alliteration—what she calls a ‘cohesive’ pair—is unlikely to be split by metathesis, whereas less cohesive pairs with a lower frequency of cluster alliteration are more likely to undergo this type of historical change. Cohesive pairs are also less likely to be produced by syncopation of an intervening vowel. In Ch. 7, M provides a similar variety of evidence for the claim that noncohesive pairs are more likely to be simplified into monosegmentals. Gradations that emerge from M’s survey tell against the hypothesis that clusters are cohesive because they violate the principle of sonority sequencing (Vennemann 1988, Lutz 1991). M argues instead for a concept of cohesion...