Antin's 'Notes for an Ultimate Prosody' Revisited

[NOTE, FOR THE RECORD. Originally published in George Quasha’s magazine Stony Brook (number one, December 1968), Antin’s essay on prosody was accompanied by the following note from the editor: “Mr. Antin wrote these Notes as a paper, originally, which was not amended for publication. I persuaded him to publish it, though he is not happy with the presentation, because I believe it raises crucial questions. It is coherent if not thorough, and it may succeed in bringing about some relevant discussion, hopefully in future issues of STONY BROOK. Part Two of the Notes will begin to investigate the poetics of that area inhabited by himself, Rothenberg, MacLow, Cage, Duncan, etc.”

      There was no followup, however, nor is it certain that one was intended.  The essay, never republished in a print edition, is a clear indication of Antin’s ongoing thoughtfulness in these matters.  It was posted in two segments on the blogger version of Poems and Poetics in 2009 and is here carried forward to Jacket2. (J.R.)]




The Contribution of Meter to the Sound Structure of Poetry is and has been Trivial


Most discussions of prosody begin and end with metrics, but the contribution of meter to the sound structure of all poetry that was neither sung nor intended for musical accompaniment, when it has been at all specific, has been trivial. Yet because most writers on prosody would probably dispute this, and since some recent poets have worked out sound structures on the basis of implicit defects in metrical theory, it's probably worth taking a look at the metrical background.      


Almost all writers on metrics agree that meter is a compositional constraint. In this theory a particular meter is a pattern of distribution of some phonological feature over stretches of language.  A particularly simple example is iso-syllabic verse.  The pattern prescribes that for each language stretch called "the line" there must be an equal number of syllables. The syllable is a phonological feature of most European languages. Even in English, where it is not always possible to determine syllable boundaries, it is usually possible to agree about the number of syllables in an utterance. But "the line" is another matter and has no linguistic existence. It is therefore a matter of metrical convenience where the line is broken. If there were no additional constraints preventing the poet from ending lines in the middle of words or from ending or beginning poems in the middle of lines, the only constraint on the poet would be the requirement that he either count while composing his poem or afterward when

arranging it on paper; and the preceding sentence would class as a kind of didactic isosyllabic verse.


it is therefore

a matter of

metrical con

venience where the

line is broken


This little poem is extra-formal in that the total number of syllables is an integral multiple of the permissible number of syllables per line. While this is a clear-cut example of a compositional constraint (at least insofar as setting down the poem on paper), it is not at all evident how such a constraint enters into the sound structure of the poem. For this to occur there must be some manner of marking off line endings in an unequivocal fashion, say by rhyme, or sounding an instrument, or by some theory of recitation, however arbitrary, by which the line endings could be made audible. Even this would not ensure the perceivability of the number of syllables per line, though it would establish unequivocal line endings. A fairly large number of syllables per line

would make it virtually impossible to listen to the words and count the syllables at the same time. Naturally it is possible, when there is a written text, to inspect the line

endinqs and then read the poem with the conviction that one is "hearing" the syllable count. This is something like a music student at the opera, reading a score of Tristan and using the orchestra as an auditory aid. It may be enjoyable but it is not listening to Tristan. Now I am perfectly aware of the visual and conceptual fascination of printed texts, musical scores and architectural plans. That is the way of concrete poetry. But

printed lines are no more verbal poems than drawn lines are architecture. Up till fairly recently the printed text has primarily been a notation for some language utterance, which must be audible. In his history of German prosody Andreas Heusler mentions a poem of Rueckert's which was composed entirely without the phoneme /r/. Is it reasonable to suppose that this constraint enters into the sound structure of the poem?      


It may be suggested that this is an unreasonable analogy, that English poetry, for example, is typically written in syllable-stress meters and that syllable-stress is

distinctly perceivable in English, but that "our ear is not accustomed to estimate line lengths simply by the number of syllables." This is quite true but not especially relevant, since the ear is not accustomed to estimate line lengths by counting anything at all. It is no more a part of normal linguistic behavior to count syllable stresses than to count syllables. In fact counting is simply not a normal part of linguistic perception.


Moreover English "syllable stress" meter is complicated still further by the circumstance that no one can be quite certain what phonological feature is distributed, whether it is a single phonological feature, or whether it is a phonological feature at all. For a more or less thorough discussion of this problem it is worth looking at Seymour Chatman's A Theory of Meter. The metrical theory he advances is not convincing, but his review of the phonological problems is fairly up to date.  Briefly, the main difficulty in identifying the metrical ictus of English poetry with English stress is that it is now by no means certain how we are to identify English stress. Traditionally linguists and grammarians agreed that there was a feature of emphasis marking either prominent syllables or

their syllabic vowels. Formerly this emphasis in the Germanic languages was believed to consist of increased intensity of articulation resulting in increased loudness or acoustical

intensity. Experimental phonetics has indicated that increased loudness is by no means the most significant factor in the perception of this emphasis. Most work in speech

synthesis has suggested that the main factor in perceiving stress is increased duration of the syllabic vowel, and that vowel quality and pitch deflection are also of considerable

importance. These results do not lend support to the acoustical intensity theory of stress, but it is quite possible, nevertheless, that syllable duration, vowel quality and pitch deflection are simply cues to recognition of the increased articulatory force required to produce the emphasized syllables. Moreover it now begins to appear that stress within word boundaries -- lexical stress, may have to be distinguished from phrasal accent, the relative prominence of a syllable in phrasal grouping. The most convincing descriptions, which are still no more than tentative, suggest that phrasal accent.results from the interaction of lexical stress rules with rules for pitch contours.


 The most recent work pointing in this direction is Chomsky's The Sound Pattern of English. But the earlier work of Kenneth Pike and Dwight Bolinger also tends in this direction. In any event we are no longer confronted with the single syllabic emphasis of

lexical stress or even with the marvelously complex but symmetrical theory of four independent stress phonemes and four equally independent pitch phonemes hypothesized by Trager and Smith back in 1951. Since the problem of assigning ictus in English syllable stress meters is the problem of comparing immediately adjacent syllables for relative prominence regardless of word boundaries, phrase boundaries, or even discourse boundaries, it is more reasonable to assume that any available sign of relative prominence will be used. What is in fact distributed is then "syllable prominence," which may result from a variety of phonological, syntactic and discourse factors. This is apparent if one merely looks at the standard account of the conventions for distributing metrical accent in typical English meters. One degree of accent is recognized, so that there are only accented and unaccented syllables; and the accent is

determined by comparing immediately adjacent syllables in a left to right direction in sets of two or three (according to whether the meter is duple or triple) for relative

prominence. The process of comparison is abruptly terminated when the "line ending" is reached. Generally it is quite possible to reach agreement in comparing syllables, but there are certain cases in which the comparisons are, to say the least, difficult and others in which they are probably nonsensical. Within a word it is always possible to determine

which of two adiacent syllables is more prominent. In a phrase group, given an understanding of the domain of the speaker's emphasis, it is also often possible to agree more or less unequivocally. But when the comparisons of relative prominence have to cross phrase juncture or sentence juncture boundaries, it is frequently impossible to reach any meaningful decision. In a phrase like " . . . in chase of him . . . " when the central accent falls on "chase" and the pitch contour begins to fall from "chase" and falls smoothly to "him," where the fall in pitch is abruptly cut for what Trager and Smith used to call "single bar juncture," who can say with certainty whether "of" or "him" is more

prominent? Or in the sequence "He left me: I called after him." who can with any assurance assign relative prominence by comparing "me" and "I". That there is a convention that legalizes such successions of two unaccented syllables ~ the pyrrhic foot) or in some cases two accented syllables (the spondee) is not the point. The introduction of successions of this sort makes it impossible to rely on the number of prominent syllables to determine the location of the line ending or to count the feet by adding up the number of accents. It is usual to refer to these substitutions as occasional variations on a well established pattern, but they are not occasional and what metric pattern is established in a passage like the following, from which they come?


…quickening then the pace  (line 131)

of the unwieldy creature he bestrode,

He left me: I called after him aloud;

He heeded not; but with his twofold charge

Still in his grasp, before me, full in view, (line 135)

Went hurrying o'er the illimitable waste,

With the fleet waters of a drowning world

In chase of him; whereat I waked in terror,

And saw the sea before me, and the book

In which I had been reading, at my side. (line 140)


I doubt that anyone will deny the brilliance of the checked music of this passage from the Fifth Book of The Prelude, but it is hardly a metrical pattern that determines the musical effects. In order to scan these lines as "blank verse" we must assume at least two pyrrhic feet in line 132, two more in line 133, an inverted (trochaic) foot in 134, another one in 135, two more pyrrhics and one anapest in line 136, two more pyrrhics in 137 plus a possible spondee, a probable pyrrhic in 138 and an additional unaccented syllable at the end of the line, a pyrrhic in 139, and two or (conceivably) three pyrrhics in line 140. Add to this that while 7 of the lines have some clear junctural breaks at the line endings,

there are at least 7 junctural breaks within the lines, so that the caesura, which is the basis of the sound of the passage, cannot be used as a clue to the line ending. For the

sake of clarity, this is my conjectural scansion of the passage:


… quíck  ên/îng thén /thê páce (line 131)

ôf thê/ûnwiél/dy créa/tûre hê/bêstróde,/

Hê léft/ mê:  Í/ câlled áf/ têr hîm/â lóud/

Hê héed/ êd nót,/ bút wîth/ hîs twó/fôld chárge/

Stíll în/ hîs grásp,/ bêfóre/ mê, fúll/ în víew,/  (line 135)

Wênt húrr/ yîng  óer/ thê îl  lî / mî  tâ/ blê wáste,/

Wîth thê/ fléet wá/ têrs ôf/ â drówn/ îng wórld/

În cháse/ ôf  hím;/ whêreát/Í wáked/ în térrôr/

Ând sáw/ thê séa/ bêfóre/ mê, ând/ thê bóok

În whích/ Î hád/ bêen réa / dîng, át / mˆy síde./ (line 140)    


But this seems like a foolish exercise, since it could not be resolved by any ear. It hardly seems likely that the lines were composed with the metrical restraints in mind. The poet

like the centipede would have too many options for the "metric" to provide him with an unequivocal way of going on. Nor is the passage atypical.




"Metrical irregularity" is more normal than exceptional throughout the history of blank verse. Shakespeare provides as good examples as Wordsworth because there is no neat single convention of syllable - stress meter in English. The reason for the very complicated set of options available to poets writing in these meters is that from the start there were two conflicting conventions, one of which was an iso-syllabic convention adapted from Romance practice and the other a quantitative convention adapted to English with the substitution of accent for quantity. This conflict is apparent in the first essay on prosody in the English language ~George Gascoigne's Certain Notes of Instruction Concerning the Making of Verse or Rime in English. It is clear that Gascoigne regards the number of syllables as the measure of the verse: "I say, then, remember to hold the same measure wherewith you begin, whether it be in a verse of six syllables, eight, ten, twelve, etc." At the same time he defends Chaucer's "failure" to observe the syllable count on the grounds that Chaucer's lines are isochronic: "Also our father Chaucer hath used the same liberty in feet and measures that the Latinists do use, whosoever do peruse and well consider his works he shall find that, although his lines are not always of the selfsame number of syllables, yet, being read by one that hath understanding, the longest verse and that which hath most syllables in it will fall to the ear correspondent unto that which hath fewest syllables in it; and like wise that which hath fewest syllables shall be found yet to consist of words that have such a natural sound as may seem equal in length to a verse which hath many more syllables of lighter accents."


The earliest English blank verse was, of course, strictly decasyllabic, as was Gascoigne's own The Steel Glass. But with the adaptation of blank verse to the theater there was no possible significance in the purely page oriented, strict syllabics and the poets on grounds of expedience abandoned it and presumably justified the expedience on grounds of a "Classical" practice in which the "long" and "short" syllables of Greek and Latin were identified with the "accented" and "unaccented" syllables of English, Even Milton, whose blank verse in Paradise Lost is clearly decasyllabic, has lines which can only be resolved by some accent-counting convention.[1] The historical situation resulted in a situation of an illusory blank verse metric; whenever one convention acted as a compositional constraint the other convention provided the loophole. The situation with rhymed verse was essentially the same, except that rhyme is audible and the use of terminal rhyme made the line endings distinctly perceivable. This did not establish "the meter," but it did mark off the stretches of language separated by rhyme as equivalence units regardless of how they may have been varied in duration, accentual weight and so on. Consequently rhyme, as far as the sound structure was concerned, was of far greater importance than meter in the history of English syllable-stress poetry.


If we extend the meaning of rhyme to cover alliteration, it has been of far greater importance than meter throughout the entire history of English poetry.[2]  The effect of meter seems to be either visual or moral. Either it is a page image of regularity and pattern, something like capital letters at the beginning of verse lines, or else an imaginary sense of constraint that has allowed certain poets to sleep at night. Given its largely fictitious existence one might wonder why poets felt any need to liberate themselves from it. Why free verse? The reason is more or less obvious. The image of meter invariably refers to other poetry. It is a visual framing effect and places whatever language is set within the frame in a context of "literature." It is not a musical device, it's a sentiment. "Metrical poetry" normally comes in a bundle together with syntactical and lexical habits that are much more effective in establishing the presence of the past, but this is not necessary. Whitman is able to embed the sound of a full-blown "folk song" in the free verse of Lilacs without the appearance of metric:


O how shall I warble myself for the dead one there I loved?

And how shall I deck my song for the large sweet soul that has gone?

And what shall my perfume be for the grave of him I love?


Sea winds blown from east and west,

Blown from the Eastern sea and blown from the Western sea, till there on the prairies meeting,

These and with these and the breath of my chant,

I'll perfume the grave of him I love.


O what shall I hang on the chamber walls?

And what shall the pictures be that I hang on the walls

To adorn the burial house of him I love?


And it's possible, like Auden, to come on as Noel Coward till somebody counts the long and short of the syllables and decides "My God, it's in minor Alcaics!" Which is terribly

chic. And poetry that is not at all scannable may still appear metrical if it is sufficiently conventional in its attitudes. Thus Yvor Winters decided that "Gerontion" was written in "Websterian blank verse." This isn't incorrect, it's nonsense. Neither Eliot nor Webster are scannable in any reasonable way, and to say that in Websterian blank verse "the blank verse norm is feeble" is such a grotesque understatement it sounds like a joke. Anyone who scans "Gerontion's" seventy-some lines and finds a handful scannable -- by applying conflicting analyses of the hypothetical pattern -- is not entitled to write "in defense of reason." All that Winters meant to say was that "Gerontion" sounds like Webster, which is neither accurate nor a prosodic statement. Anyone with a perverse sense of humor or a morbid interest in literary criticism can compare Winters' attack on Eliot in In Defense Of Reason with Harvey Gross's defense in Sound and Form in Modern Poetry. What emerges is the conclusion that the briefest suggestion of scannability is a gravitational center around which prosodists cluster like moths around a light. Eliot is, however, partly responsible for this sort of discussion. It is the kind of inanity he made possible by his 1917 essay on free verse: " . . . the ghost of some simple metre should lurk behind the arras in even the freest verse." When Polonius is summoned he always appears.




1. The idea of a blind man composing poems in a purely page oriented syllabic measure is so unlikely that alternatives ought to be suggested. Either Milton’s blindness was mythical, which doesn’t seem likely, or he was not responsible for the final page arrangement of the poem he dictated, in which case the metric of Paradise Lost was largely due to the editors.  This may explain why the irregular lines were never corrected.


2. I am aware that there are a number of very elaborate and ingenious theories of old English versification. but the only one that can count as reasonably "metrical" requires the hypothesis of either musical accompaniment or some special recitation technique which would allow for isochrony, and the evidence for this is slight. The more commonly accepted Five Type Foot Theory worked out by Sievers and subsequently rejected by him may describe fairly accurately the phrasal rhythms of Beowulf, but it is hard to see what that has to do with metrics. As a compositional constraint the theory supposes an immense variety of options. There are really six types of "foot" and numerous loopholes that constitute subspecies. It is not really clear what is excluded by this theory and, if significant Old English phrasal rhythms are excluded by it, and whether this is due to the rhetorical habits of the Beowulf poet or Old English poetry in general. (It is worth pointing out that an observer of chess games might wait a very long time before ever seeing White open by moving his king pawn to king three; it is nevertheless quite legal. Moreover the Sievers theory depends upon the existence of "lines," and there are no lines in Beowulf that were not established by editors. And the lines established by the editors are not satisfactory and require the assumption of numerous "hypermetric lines," which then require still more explanation. It would seem much more economical to assume that there is no "metric" and there are no "lines," that there is a continuum of language punctuated by alliteration, a habit of bipartite phrasing, and perhaps a consistency in placement of the caesura. If we accept the oral formulaic theory of composition for Beowulf there is even less reason to suppose a metric, especially a metric so laughably complicated as Sievers'. It seems extremely unlikely that the theory of Old English meter is based on anything more solid than nineteenth-century expectations.