Average poetic line length is determined in large part by human neurology. Firstly, syllable length and delivery-rate show a rough correspondence to our minimum auditory stimulus response time, which would make some basic evolutionary sense: presumably requests for lion-related assistance from the very fastest talkers went both uncomprehended and unheeded, while the drawlers and the mumblers were eaten before the words were out. Secondly, there’s the matter of the ‘specious present’. William James defined this notional period as a kind of ur-unit of time perception, being ‘the short duration of which we are immediately and incessantly sensible’,1 i.e. the frequency of our conscious being. We now have a measurement for this duration, which it comes in at about 2.5 to 3 seconds. It would appear that this is the length of the ‘auditory present’, our hearing being the sense best equipped to follow and measure a timeevent.2 (Sight is half as quick in its reactions; smell is singularly poor in this regard – try counting fifteen minutes off with your nose. If it’s thinking at all, and not screaming food, fear or filth, your nose is getting all dewy-eyed with the temporal cortex in its role as an access-point for long-term memory.)
We appear to divide time into psychological chunks based on the length of this three-second slot, and prepare our speech (as well as many other short-term actions, from handshakes to tennis serves) in roughly three-second units, with micro-pauses between them. As listeners, we attend to the speech of others in exactly the same way, dividing it up into manageable units of around three seconds. This three-second ‘buffer’ is then parcelled up and sent upstairs for processing, cyclically, over and over – forming a continuous three-second wavelength, a slow ‘pulse’ of attention, both to our own speech and that of others. If the information contained in this buffer sounds important, and needs to be consciously reviewed or memorised, it can be done so via the ‘phonological loop’; this is a short-term memory slot about the length of the auditory present, which can be replayed for a short time. (If you snap your fingers in a silent room and listen to how long you can ‘recall’ the sound, how long the very sound itself seems to echo and persist in the mind’s ear – that’s about the number of times the loop can be replayed before the sound degrades into a memory; the exercise is a profoundly chastening and mortal one.) There is an important data-processing consideration attached. The loop is capable of holding around seven (plus or minus two) bits of information; this is why a seven-digit landline number is very much easier to recall on one hearing than a ten- or eleven-digit mobile number.3
It should not surprise us to learn that most cultures have a default poetic line adapted to the length of the human auditory present, one which takes around three seconds or so to say aloud. (The kind of ‘silent reading’ we sometimes do when scanning a newspaper article or novel doesn’t inwardly articulate the sounds of the words, and greatly increases our reading-speed. When reading poetry, however, this almost never happens; our ‘lips move’. Or at least should.) Lines of a rough length of three seconds are more easily committed to memory, since they can be apprehended as an indivisible auditory unit. The data-limits of the phonological loop constrain us too, as do the number of syllables which can be slowly and deliberately spoken in this time: around ten, with the brain able to process around five or six pieces of content. Overload a line with content, and mind-leakage will occur. (I’m also struck by the coincidence of poetry’s traditionally image-heavy tendencies and the ‘visuospatial sketchpad’, the visual equivalent of the phonological loop, and a major component of the mechanism of working memory.)
The poem, then, is a succession of instances of the human present. This line length of between 2 and 4 seconds roughly corresponds to a universal default, a line of around twelve (plus or minus five) syllables in non-tonal languages, and about half that number in tonal, where the metrical syllable is twice as long. The three-second line is a feature of classical Chinese and Japanese poetry, and of Latin, Greek, Hebrew, Slavic, German, Spanish and Gaelic verse; it corresponds to the French alexandrine, and in English to the 4-strong line. And it is the engine of the iambic pentameter, that bastard offspring of those princely three-seconders, the Italian hendecasyllable and the Anglo-Saxon alliterative line. (It’s also the rough length of time it takes to read the schwa-infested tweet; there was nothing arbitrary in its 140 character limit ‘feeling about right’. The average tweet-length is the same as the average sentence, of course: around 14 to 17 words and approximately a hundred characters, which a conversational reading speed will bring it at around the three-second mark. The recent doubling of the character limit stands a good chance of killing the platform.)
Any line of between roughly eight and twelve syllables is generally thought of as ‘average length’, i.e. neither long nor short. By the same rule, phonological memory decay dictates that there are about three seconds or so on either side of a word in which its sound can be prepared for or echoed, unless it has a salient position, like a rhyme-word: these are noisier – and so can be committed to memory, then recalled even several lines later. (The kind of analyses which point out buried lyric correspondences between non-salient or non-terminal words many lines apart are, alas, acts of either pure Kabbalism or wishful thinking.)
In any art form interested in being memorable, the ease with which it can be memorised is clearly going to be a major consideration; in an art form like poetry whose project is additionally memorialising, it’s going to be a fundamental one. It should be no surprise that poetry’s memorial project should naturally converge on those units of time that best reflect the way our brains encode memories. Those line-units are not discontinuous – one line runs into the next, as one moment bleeds into the next – but they do establish our deep poetic measure: the line is the carrier-wave of poetic sense, with three seconds its rough frequency.
1 William James, Principles of Psychology (New York: Henry Holt, 1890), vol. 1, 631.
2 As mentioned earlier, Pöppel and Turner’s paper ‘The Neural Lyre’ (1983) first proposed and elaborated this idea, but it also drew some very unfortunate reactionary and pro-New Formalist conclusions, using the three-second rule to declare the innate and superior fitness of certain line types over others. This, from a remarkable attack on Charles O. Hartmann’s Free Verse (Princeton, NJ: Princeton University Press, 1980): ‘His argument attempts to save free verse, and therefore defines verse in a hopelessly vague way; ours is content to abandon it as verse unless it consciously or unconsciously employs the human and universal grammar of metre. It may be an admirable kind of word play, and it might even be argued that it is a new art-form of our century. But it is not poetry …’ Fast-forward ten years to New England symposia where young Republicans in bow ties give papers on Yvor Winters. Setting aside the stupid philistinism of Turner’s declaration, while one is at liberty to say what one thinks poetry is, one should never say what it is not: poetry is whatever we call poetry. Besides, as soon as you declare what poetry isn’t, poets will just go and write that instead.
3 This insight originates with George A. Miller’s influential paper ‘The Magical Number Seven, Plus or Minus Two: Some Limits on Our Capacity for Processing Information’, Psychological Review 63 (1956): 81–97. Atkinson and Shiffrin later proposed the dual-store memory model, where a ‘buffer’ can hold a short memory for a very limited time – R. C. Atkinson and R. M. Shiffrin, ‘Human Memory: A Proposed System and Its Control Processes’, in The Psychology of Learning and Motivation, vol. 2., ed. K. W. Spence and J. T. Spence (New York: Academic Press, 1968): 89–195. While this happens, it is bedding down its own semantic associations in the long-term memory – which, in contrast to the short-term memory, has a limitless capacity. Once there, it can be recalled by a contextual prompt. This memorising process is called ‘consolidation’, which is exactly what poetic lines attempt to achieve: if we fetishise their sheer beauty, truthfulness, originality and musicality, through rehearsal and re-rehearsal we will create the kind of meaningful associations by which they can enter the long-term archive. Here they can be strengthened by their periodic recall or ‘recapitulation’. Baddeley and Hitch adapted the Atkinson–Shiffrin memory model to that of ‘working memory’; this uses the short-term slave-systems of phonological loop and the visuospatial sketchpad, which are governed by a notional ‘central executive’, to hold different types of stimulus – A. D. Baddeley and G. Hitch, ‘Working Memory’, in The Psychology of Learning and Motivation: Advances in Research and Theory, vol. 8, ed. G. H. Bower (New York: Academic Press, 1974), 47–89. Later Baddeley added the ‘episodic buffer’, a vague holding-space for information from a number of sensory sources, but within which new kinds of cognitive representations might be created – which might point to the buffer also having a problem-solving role.