I've never studied verse formally. I've read a few books and a lot of essays, but never in any sort of systematic way. For a long time I've struggled with the various ideas about rhythm in English language poetry as much of what a lot of poets say about rhythm doesn't really make a lot of sense. Ultimately, I've been left with the definite suspicion that much of what poets believe about rhythm is largely unconnected to what they practice when they're writing. Most important, there is a flaw in the concept of poetic rhythm being regular in the same way that music is regular. Existing systems of scansion that attempt to regularize poetry rhythm are therefore flawed at root and make for a dull and difficult tool for the analysis of poetry. Far more systematic and interesting is, I think, the study of prosody from a linguistic point of view and there is a great deal of very good literature available on phonology and phonetics which is illuminating when applied to poetry.
I'd like to open what promises to be a rambling and jumbled essay on the subject by stating up front that I place a high value on the printed nature of poetry. Poets tend to talk a lot about oral traditions and the necessity of sound to poetry and the value of poetry being read aloud. I have a number of problems with this sort of idea, the most important being that it ignores a fundamental aspect of the printed word. The technical term is "subvocalization," which is the silent inner voice we "hear" ourselves speak when we read to ourselves. It's not something that we should take for granted as it has not been a necessary or even common aspect of reading at all times in history. Subvocalization is a salient factor in the way we read in English and I think that its implications for English language poetry are worth considering, although I'll only touch on the issue for now. Obviously, when a reader reads a poem silently to herself, she subvocalizes in her own internal voice. This inner speaking of the text is, I think, an underappreciated avenue of poetic intimacy between the author of a poem and each individual reader. A cursory googling of various search strings returns little more than speed reading programs and information about NASA's attempt to do subvocal speech recognition for communications systems.
Poetry is a communicative artform that works more or less constantly with various levels of semantic density. This is, I think, non-controversial whether or not one accepts Pound's maxim correlating greatness in literature with the level to which it is "charged with meaning." The various devices and methodologies employed by all poetry are all in some way involved in semantic content and it's transfer from author to reader by means of symbols. One of the distinguishing features of poetry is the relative semantic weight each symbol is expected to bear and the various techniques of poesy are all methods of increasing that weight. In some cases this is fairly obvious as with simile and particularly metaphor as Borges described in his Charles Eliot Norton lectures. Prosodic devices like rhyme and meter have less obvious semantic consequences, and these consequences I think are often ignored or believed to be mere vanity and decoration by many poets. This ignorance and vanity, and the fact that it is the use of these techniques that have the most profound effect in the intimacy of subvocalization, results in vast swaths of unexplored territory in this ecclesiastically postmodern age where there's supposedly nothing new under the sun. Prosodic devices work as controls on the reader's subvocalization process, forcing him to repeat sounds or emphasize and de-emphasize certain syllables and give guides toward intonation. Since so many prosodic devices work by repetition, whether phonetic or rhythmic, they function as subtle temporal connective tissue in the threading space-time the poem occupies as it is being read. Reading poetry silently, when understood correctly, is an act of following a pattern for the weaving of a tapestry of meaning in which every element of language can be employed by the loom.
So rhythm then. What's to be made of rhythm in the silent poem? The most obvious factor is the fact that every English speaker is a speaker of some dialect not necessarily well aligned with the dialect of the writer. In fact, a poet can and should expect that there will be rhythmic variations from her intentions depending on the dialect of the reader. Various dialects of English drop consonants, have vowel shifts that effect the length of syllables, replace single vowels with diphthongs or monophthongize common diphthongs, and particularly the widespread and irregular tendency among all dialects to break up consonant clusters by inserting unstressed vowels, among many other phonological differences between regions, classes, and cultures, all of which have various effects on the rhythm of the subvocalization of a given text. Certainly there are general rules that are going to apply in almost all cases. Syllables containing long vowels and dipthongs in their nucleus and with complex codas are going to have some sort of stress on them. Syllables with short vowels and particularly schwa's are commonly going to be unstressed. But there are exceptions. There are many English speakers for whom English is a second language, who do not possess the schwa vowel, for example, and may substitute an /a/ vowel in its place in which case what most native English speakers will naturally read as an unstressed syllable may be stressed for them. The exact implications of this are subtle but profound, and ultimately amount to the fact that a poet can never completely design and construct the meter of a poem. Rather than a problem, I find this to a be a good thing. Whereas with many other aspects of composition, it is necessary to do a great deal of thinking in order to insert telling and skillful ambiguities and uncertainties—two qualities I place a great deal of value on—in the case of rhythm it's impossible for them not to be there, and a poet who is well aware of various dialects and the differences between them can manipulate these ambiguities as easily and effectively as one might deploy parataxis or some other syntactic or imagistic device to a similar end.
This of course gives the lie of various methods of scansion such as, to name just a few, traditional system syllabic feet borrowed from classical Greek and Latin poetry, Gerard Manley Hopkins's sprung verse, Dana Gioia's ham-handed attempt to graft Anglo-Saxon meter on to modern English, and the attempts of other New Formalists to revive iambic pentameter. For an illustration of the limitations of all such approaches, one only has to look as far as Dub poets like Mutabaruka or Oku Onuora, who, while being primarily performance poets working more closely with the tradition of art in the African diaspora than they do with the tradition of English language poetry in the Americas, never-the-less in their patois compositions illustrate the futility of all such reductive systems of scansion. Only a meticulous phonological analysis acknowledging variations in vowel length as important for syllable strength along with at least primary and secondary stresses and probably tertiary stresses in addition to unstressed syllables, which syllables stand as the basic pulse of English speech, can adequately represent the rhythmic complexity of such work.
That having been said, to say that there is no regularity to meter is not to pledge allegiance to Dan Schneider's reductive metric fallacy either. Schneider, while correctly rejecting the reduction of meter to the diad of stressed and unstressed syllables, goes on from that point to argue that because there are multiple levels of stress that stress is therefore infinitely divisible and entirely relative. It is not at all obvious that such a conclusion follows, and Schneider's examples are easily dealt with by a standard four level stress analysis. I should note that the four level analysis I'm advocating is merely a convenience and nothing absolute. I like it because I find that the rhythm of natural speech makes sense when scanned in such a way and that certain features of poets like Frank O'Hara and Charles Bukowski become more clear under such analysis. Among Phonologists, the four levels of stress analysis is still controversial, with Ladefoged being one of the prominent linguists who believes that there are other, more illuminating ways of accounting for such differences. From a poetics standpoint, I think such controversies in Linguistics can only be good, as they provide poets and readers with further tools which can be used to construct and decipher a line, and current trends in Linguistics are sufficiently away from the reductivism of previous generations to bring no cause for alarm on that note. It's unlikely that Laboratory Phonetics is going to announce someday soon that all of the rhythm of natural speech and therefore of language is in fact capable of being generated by a set of simple algorithms. Fixed meter can be safely left to gather dust in the historic literature of previous centuries.
But where does this idea come from that we ought to have these regular meters to follow, that scansion should be reductive and simple, even though it's patently obvious that there is no regularity to meter even in much strictly formal poetry (take a look at the distribution of secondary stresses in any sixteen lines of Milton if you don't believe me)? My suspicion is that it comes from the unilluminating and foolish analogy between poetry and music.
While I haven't studied poetry formally, I have studied music formally, and it is relatively clear in my mind what constitutes rhythm in music. Primary among it's components is the presence of a pulse at a fixed interval in a given piece of music. These beats may then be organized in various ways by accentuating certain pulses in a regular pattern and through the subdivision of the pulse in various ways. English as a more or less stress-timed language does not have a fixed interval between pulses and while at times the interval taken by a string of syllables will be more or less regular. Here the analogy to music falls apart obviously and completely. Misleading at best, the analogy to musical rhythm has caused generations of poets to attempt to bend language into a structure that it simply will not fit. As a result, various language features are de-emphasized or ignored completely—such as unstressed syllables in Gioia's take on accentual verse. This imposes a limitation on the compositional possibilities of a poet. Given that style requires a certain anarchy and chaos in order to thrive and diversify—a fact that I hold to be true of all artforms but the argument for which is beyond the scope of this essay (interested readers should read Paul Feyerabend's Against Method and notice that what he says about the practice of science is even more true of the practice of art)—no system of prosody which glosses or de-emphasizes language features, and thereby impose a compositional limitation, is an acceptable tool either for analysis or design.
To that end, I suggest the abandoning of ideas of feet, of "Strong and Weak" syllables, of terminating attempts to graft ideas based on the poetics of other languages whether that language be Latin, Anglo-Saxon, or Old English, and of approaching scansion on a case by case basis neglecting no factors save those which are usefully neglected at any given time. The caveat of course being that committed as I am to methodological anarchy, a writer should always completely ignore any theoretical statements that interfere with working, including this one as the case may be.