Meter in poetry: A new theory (review)

Abstract

Reviewed by: Meter in poetry: A new theory Paul Kiparsky Meter in poetry: A new theory. By Nigel Fabb and Morris Halle. Cambridge: Cambridge University Press, 2008. Pp. 312. ISBN 0521713250. $45. The publication of this joint book by the founder of generative metrics and a distinguished literary linguist is a major event.1 Fabb and Halle take a fresh look at much familiar material, and introduce an eye-opening collection of metrical systems from world literature into the theoretical discourse. The complex analyses are clearly presented, and illustrated with detailed derivations. A guest chapter by Carlos Piera offers an insightful survey of Southern Romance metrics. Like almost all versions of generative metrics, F&H adopt the three-way distinction between what Roman Jakobson called verse design, verse instance, and delivery instance.2 F&H’s theory maps abstract grid patterns onto the linguistically determined properties of texts. The mapping imposes constraints on texts, which define their metrical form. In that sense, it is a kind of template-matching theory. Recitation may or may not reflect meter, according to conventional stylized norms, but the meter of a text itself is invariant, however it is pronounced or sung. Where F&H differ from everyone else is in denying the centrality of rhythm in meter, and characterizing the abstract templates and their relationship to the text by a combination of constraints and processes modeled on Morris Halle/William Idsardi-style metrical phonology. F&H say that lineation and length restrictions are the primary properties of verse, and rhythm is epiphenomenal, ‘a property of the way a sequence of words is read or performed’ (242). This seems inconsistent with their use of bracketed grids to characterize metrical patterns and stress, for bracketed grids represent—indeed are designed to represent—rhythm as periodic alternation of prominence at a hierarchy of levels. F&H’s point is probably that metrical rhythm and textual rhythm are not necessarily articulated in recitation, since meters constrain texts, not performance. [End Page 923] The thesis of the primacy of lineation leads F&H to suggest that rhythm arises as a by-product of counting syllables to fix the length of lines. Traditional doctrine conversely derives lineation and line length from constraints on the hierarchical rhythmic structure that meter imposes on texts (Chen 1980, Kiparsky 2006).3 An argument for the latter view is that it explains the conventional character of lineation. Any place where an obligatory major prosodic break in the verse design divides equivalent units may be a line break, by convenience or tradition. F&H themselves illustrate this point nicely by splitting the half-lines of Arabic verse into separate ‘lines’ forming a ‘couplet’,4 likewise Sanskrit ślokas (222). 4343 ballad quatrains can be printed as fourteener distichs and vice versa, and similarly 3343 quatrains are interchangeable with poulter’s measure. Some editions of the Kalevala print its eight-syllable parallel couplets as single sixteen-syllable lines, and nobody minds.5 Another argument for the primacy of rhythm is that meters constrain only those categories that are prominence-defining in language. For example, meters may require that syllables in certain positions must be stressed or heavy, never that they must have onsets. The idea that meter regulates rhythmic patterns of prominence explains this, for (as we know independently from phonology) onsets are not prominence-bearing. More interestingly, consider empty positions (catalexis). Quatrains typically combine four-and three-foot lines in 4343, 4443, or 3343 patterns, not in 3434 or 3334 patterns. The short lines can be perceived as metrical realizations of four-foot lines with a missing final beat. In songs and other isochronous performance styles, they are typically realized by lengthening the last word, producing a saliency effect characteristic of terminal elements (Kiparsky 2006). F&H’s theory cannot explain the perception of 4343 and 4443 stanzas, and the preference for them, in this way, because it does not assign gridmarks to empty positions. The second leading idea of the book is that the abstract templates are constructed by a bottom-up parsing procedure, from the smallest groupings to the biggest, and that template-to-text mapping is governed by a combination of rules...