Syllable structure and stress in Spanish: A nonlinear analysis By James W. Harris (review)

REVIEWS933 Syllable structure and stress in Spanish: A nonlinear analysis. By James W. Harris. (Linguistic Inquiry, monograph 8.) Cambridge; MA: MIT Press, 1983. Pp. xi, 158. Cloth $30.00, paper $15.95. Reviewed by Rafael Núñez-Cedeño, Boston University In 1969, Harris published Spanish phonology, which has long been recognized by Hispanicists as the first comprehensive treatise on the subject as seen from the generative-transformational viewpoint. In the book under review, H confronts some of the same issues and presents them afresh, grounding his analysis on the most recent advances made in generative phonological theory— in particular, on the insights offered by studies treating syllable structure and metrical phonology. Any reader familiar with H's writings will be pleased to observe that this monograph is in keeping with his reputation for care, succinctness , and clarity. The monograph is divided into two parts plus an appendix. In the three chapters of Part I, H presents an explicit discussion of the structure of the syllable and the principles that underlie its organization, followed by an illustration of how the theory handles issues and problems in Hispanic phonology. In the two chapters of Part II, H incorporates the insights of metrical theory to provide a more adequate description of word stress. He also discusses the well-known topics of cyclic and non-cyclic stress assignment, and ends with an extrametrical account for the assignment ofproparoxytone stress in Spanish. The appendix deals with the stress shift of some highly exceptional forms. Chap. 1 lays the bases for the generative study of syllable structure, which H sees as a 'legitimate object of linguistic investigation' (p. 4); this statement may seem extraordinary to traditional linguists, since the notion of the syllable has never escaped their attention.1 H reviews the early literature ofthe syllable in generative phonology, with special attention to the description of Saporta & Contreras. That work proposed a phrase structure grammar, generating a syllable with a ternary-branching organization. With convincing argumentation and counter-examples, H demonstrates that their grammar generates potentially unacceptable syllables, thereby failing observationally. For H, the syllable has two immediate constituents: a rule-constrained optional onset, and an obligatory rhyme (which minimally contains a vowel in Spanish). In essence, he dispenses with the notion ofa coda, since its recognition entails a significant weakening of generalization in accounting for phonological processes. It is presumed that onset and rhyme operate independently of each other, although their internal organization is constrained by language-particular rules and universal conditions. H supports his claims of a binary-branching syllable by arguing, first, that the maximum permissible length of a syllable is five segments; and second, 1 Various non-generative studies have considered the structure of the Spanish syllable, starting with Malmberg 1948. Granda 1966 provides us with a more detailed diachronic/synchronic analysis, which in part resembles that of Saporta & Contreras 1962, but is less lucidly explicated. Both Malmberg and Granda are overlooked by H. The natural generative syllabic model proposed by Hooper 1976 receives passing attention. Harris dismisses her arguments for not providing 'an adequate characterization of syllable organization' (137, fn. 4). 934LANGUAGE, VOLUME 60, NUMBER 4 (1984) that antepenultimate stress is impossible if the penult is checked by a consonant . The first point is a novel one in Hispanic linguistics; its significance will be shown below. The second argument has its roots in Latin, and was alluded to in previous literature; but H highlights its importance, to underscore its relationship with prosodie restrictions. Once the syllable is divided into two constituents, it is easy to understand why any single vowel in words may qualify as a legitimate syllable: oía 'heard' has three sonority peaks, and therefore contains three syllables, o-i-a. But Spanish syllables may also begin with one or more consonants. H captures this fact by proposing a rule, in Chap. 2, in which onsets are said to be maximally binary-branching trees whose branches dominate consonantal segments. According to this Spanish-specific principle, a rhyme may be preceded by any one of the consonants of the phonological inventory. The well-formedness of two-segment onsets is constrained by a universal sonority scale, and by several language...