Supple Stanzas

by John Carley

The word haikai is used extensively in this text. It refers to a group of literary genres outside the Japanese courtly tradition which includes renku (haikai renga), haiku (independent hokku), senryu, haibun and haiga.

Introduction – message in a Rokku

Renku arose in Japan. Declined in Japan. And, during the course of the last century, was reborn in Japan. Throughout this time the vast majority of sequences were, and still are, written according to accepted norms of prosody described as teikei – strict-form – a principal feature of which is the adherence to shichigochou, a set of cadences based on metrical phrases of five and seven syllables. Strictly speaking these ‘syllables’ are in fact mora, but for the moment the distinction is slight.

In recent years one of the more significant contributions to the continued evolution of renku has been the proposal, by the respected Japanese poet and critic Haku Asanma, of the rokku – a flexible yet demanding sequence in which the penultimate movement abandons the strict-form prosody in favour of a more open approach.

It is tempting to see this as proof that a highly conservative establishment has finally recognised the superiority of Occidental notions of haikai prosody wherein freedom has long been the whole of the law. But the truth is otherwise. The free-verse passages of rokku are simply intended to prevent the lazy and unimaginative from exploiting the sheer constancy of strict-form prosody in order to dignify poor writing. After all, if a given stanza follows the accepted structural formulae is it not at least ‘correct’? If short verse duly follows long verse is the integrity of a renku sequence not guaranteed? No, replies Asanuma-san, renku demands more than this.

The propensity for slack writing is of course a true universal, and the problems the rokku highlights are mirrored in Occidental haikai. Mirrored, and reversed. The fact that there is almost no constraint on English-language haikai prosody means that poets can dash off any old draft and declare it to be definitive. In the absence of any comparator each verse is, in the literal sense at least, peerless.

In the case of haiku such facile bombast may be of little account. A haiku is by definition a verse which exists in isolation. If it succeeds it may do so purely on its own terms. But when applied to renku such randomness of form can spell disaster, a disaster which has nothing to do with the failure to placate tidy minds, or otherwise conform to some inviolable rule.

Renku verses by definition do not stand alone. In a sequence where each added verse is written according to its own unique prosody, most forms of phonic correspondence between verses are either absent or seriously marginalised. Tangible inter-relationships may be established at a conceptual level only. Accordingly, poets are pushed towards meaning-link (kokorozuke) and word or object link (monozuke/kotobazuke) as they seek to ensure that the dependencies between verses ‘make sense’.

At a stroke Basho’s most prized technique is compromised. The deep-level evocative potential of scent link (nioizuke) becomes problematic as, in the absence of phonic ties, the semantic relationships between verses become too tenuous to reward the readers’ wish to engage.

The wider movement of the poem is similarly more difficult to orchestrate; verses crash or stutter as each attempts to assert a largely self-referencing structure. In such circumstances it takes great artistry to establish any form of sequence at all. Too readily we are presented with what is in effect a series: a succession of highly variable and quasi-autonomous verses whose very amorphousness requires us to study their shape.

Necessarily, as we focus on the structural rationale and hard content of individual stanzas, the creative power of the white space between them is lost.

Some phonic tools

One of the more curious aberrations to emerge from the English-language haiku movement during the course of the last century was the absolute conviction, still frequently encountered, that haiku, and by extension all haikai, ‘don’t do poetics’. It is certainly the case that very short stanzas will not readily support lengthy and complex tropes. And the sheer abundance of vowel sounds in Japanese makes nonsense of the idea of end rhyme. But the word ‘poetics’ embraces more subtle temptations than these – temptations to which poets, of every period, have serially succumbed.

Japanese haikai prosody makes extensive use of assonance and alliteration to animate diction. Onomatopoeia is common both in its literal from and in the more abstract sense that certain sounds are understood to evoke particular emotional palettes. This technique in turn is related to that of ‘vowel harmony’ in which heterogeneous sounds are often subordinated to the repetition of a given vowel.

Phonetic and syntactic parallelism are standard. Great play is made of cognates and homonyms – often enhanced by the flexible orthography which favours both fine shading and a creative ambiguity of association. Symbolism, metaphor, oxymoron, hyperbole and extra-textual allusion are meat and drink. In short, the Japanese poet plays with words. Crass writers do it crassly. Good writers do it well.

To suggest that the English-language poet should employ only the most utilitarian word choice when composing haikai is an instance not so much of misunderstanding as of self-hatred. It is clearly the case that in the poetry of evocation any linguistic technique which draws undue attention to the mode of utterance itself, to the surface of the text, must count as a failure. But this is question of poor artistry, not evidence of a mistaken philosophy.

For centuries haikai poets have used the phonic tools available to them to help cement both sense and sentiment, be it within a given verse, or between the verses of a sequence. It would be strange indeed to suggest that this approach was appropriate solely in the original linguistic and cultural context.

Appearances – the Japanese model

While discussion of linguistic techniques appropriate to English-language haikai has been scant, and heavily skewed by early and often clumsy attempts to re-versify haiku in translation, issues of stanza form have at least been on the agenda … until recent times that is. What shape should a stanza be? Should it have a shape at all? And how long should it be? Measured by what? Dosed and ordered how? And aren’t the proportions, if any, purely visual?

Clearly the starting point for any such enquiry must be the genre from which our interest derives. So let us look first at how the long and short stanzas are set down in Japanese.

A typical stanza is written as a single unbroken column of characters, though in some circumstances it may be typeset as a horizontal line reading left to right. There are no gaps or punctuation marks to indicate the boundaries of a particular phrase or segment. Nor are there additional spaces to separate one word from another.

The Japanese language permits more forms of inflection than are conveyed by the query and exclamation marks common to European systems of writing. Similarly, in the case of haiku and hokku, there is more than one type of cutting word (kireji) each with an associated emotional tone. However, these modes are signalled not by punctuation, but by the inclusion of a word which is considered to be a normal part of speech and which is voiced when a poem is read aloud.

The reader of a haikai stanza in Japanese is therefore presented with an otherwise undifferentiated string of characters, the individuation of words, phrases, metrical beats and rests taking place entirely within the mind. In the case of strict-form haikai, stanzas are composed of a number of metrical feet of five or seven morae – this latter a technical term drawn originally from the prosody of Latin which indicates a standard phonic time unit, and which in this context it is important to distinguish from ‘syllable’.

Each mora (in Japanese: on or haku) is represented by a single character in the phonetic syllabaries (hiragana, katakana) which, with the exception of a number of free-standing vowels and a nasalised ‘n’ not dissimilar to that found in French, may be equated to an Occidental consonant plus short vowel (onset plus nucleus) and therefore written as such in the Romaji phonetic rendering system.

By way of example, the word (onji), which may also be written phonetically as , and was mistakenly thought to be a core term in Japanese poetics, is composed not of two syllables ‘on’ and ‘ji’ but of three morae: the free-standing ’o’, the nasalised ‘n’ and the onset plus nucleus ‘ji’. Clearly, as is the case here with the element ‘on’, utterances considered as a single syllable in English may comprise more than one mora in Japanese. In other words, when it comes to length, to temporal duration, the English syllable is considerably more variable than is the Japanese mora. The syllable is frequently much longer.

It is not unknown for the sounds, and therefore the characters, which compose a word to straddle a pair of metrical feet, but in almost all cases one word will end, and another begin, at a metrical boundary. However the broader syntax is not necessarily interrupted or modulated by the passage from one metrical foot to another. The transition is therefore qualitatively different to that imparted by the Occidental line ending which always confers some form of semantic colouration.

While instances of over-run (ji-amari) and under-run (ji-tarazu) do occur, the regularly phrased Japanese haikai long verse comprises three metrical feet of 5-7-5 morae, and the short verse two metrical feet of 7-7 morae. These quantities are also found in forms of classical poetry such as choka, waka, and tanka, in scurrilous verse such as senryu and zappai, and in traditional popular song. So engrained are they in the general psyche that they occur frequently in contemporary slogans, epithets and proverbs. In sum, they are profoundly familiar to their audience.

Some modes of emulation

Profoundly familiar. But only within Japan. In fact it is hard to imagine a stanza form more difficult for the early Occidental enthusiast to emulate than a single unbroken column of characters. As noted below, the now almost-universal adoption of the tercet (three-line stanza) as the preferred way of writing the long verse is not of itself proof of suitability. Over the course of the last 100 years or more translators and poets have sought many different solutions, each of which has attracted its supporters and detractors – sentiments which have sometimes been voiced with a vehemence verging on the absurd.

Before moving on to the contemporary mainstream it may be useful to briefly survey some common alternatives. For the moment our comments deal principally with haiku, as renku has been so little written until recent times.

Short Prose

One of the earliest translators of haiku, Lafcadio Hearn, pioneered this approach. Unkindly characterised by later critics as a mere ‘dribble of prose’ these snippets of text adopt capitalisation and standard punctuation, sometimes with the addition of a line break or extra spaces in combination with parataxis. With so few constraints the technique is highly flexible and tends to the discursive. It may be objected, though, that by its very openness it bears little resemblance to the tightly structured original.

Concrete and/or Variable Use of Space

Striking effects engineered by the concrete arrangement of text and the variable use of space have been of particular interest to English-language haiku poets since the 1970s. The active juxtaposition of text against blank space is supposed by some to echo the fundamentals of Buddhist aesthetics, drawing on notions of the dynamic tension between presence and absence. There is also a belief that the importance of visual presentation may in some way echo the significance accorded to the execution of brushwork in earlier times. The fact remains however, that the source genre relies little, if at all, on any form of overtly visual effect. The artistry of the Japanese verse resides overwhelmingly in effective word choice and the conceptual, not concrete, use of space.

The One-liner

Perhaps the most demanding of all the stanza forms, in skilled hands the one-liner, (a wholly unpunctuated monostich) lends itself to complex disjunctive and/or agglutinative effects ranging from the subtle to the profoundly disconcerting. Though there are character spaces between words it also looks, superficially at least, very much like its Japanese counterpart.

Yet this apparent similarity is in truth the source of its difference. The form of the Japanese stanza is very familiar to its reader, as is the activity of divining discrete semantic and metrical units in an otherwise unified bloc of text. The English-language reader by contrast is unused to the monostich as anything other than a vehicle for epithets, aphorisms and slogans. Most tellingly, the presence of some or all of capitalisation, punctuation, spaces, and line breaks is normally required to identify both semantic and metrical divisions. The unbroken line is, therefore, rather alien to the reader.

The Two-liner

The two-liner as applied to English-language haiku is more distich than couplet in so far as this latter term suggests end rhyme, which is rarely employed, in contemporary work at least.

Visually the stanza is familiar to the reader, the relatively long lines allowing a relaxed phrasing. However, the temptation to employ strong syntactic and metrical parallelism between the pair runs counter to the core semantic structures inherent in the source literature which, if divided at all, tend to a one-third versus two-third split.

The Quatrain

For the general reader the quatrain, the four-line stanza, is perhaps the most familiar form of all. However, like the distich, it does not easily lend itself to handling a marked one-third/two-third division as it falls most readily into two pairs of lines. The very brevity of these lines is a further problem; there is the tendency for each to attempt to carry a complete phrase, albeit compressed. The cumulative effect, it is argued, leads to a more extensive treatment of a subject than is typical of the source genre. It tends to be over long.

The Zip

First proposed at the turn of the 20th century the zip attracted a great deal of initial interest for its unlikely combination of high flexibility with a recognisably fixed form – this latter quality still considered a minimum condition of acceptance by some of the more conservative Japanese commentators when considering the validity of English-language ‘haiku’.

In skilled hands the zip can handle any type of semantic structure, adapt itself to any style of phrasing, and enable everything from soft pivots to harsh syntactic disjuncture. However, both in print and as (computer language) html, the stanza is difficult to type-set correctly; it requires formatting that is suppressed by some digital media, and it can look rather daunting to the beginner.

The Three-liner – Obviously

If the alternatives have failed to catch the collective imagination perhaps that is simply because the three-line model, freely interpreted, is nothing less than correct. As anyone who likes a good snail will tell you: 50 million Frenchmen can’t be wrong. Yummy!

Research carried out in 1997 by the American poet and editor George Swede into the preferred stanza form for English-language haiku as represented by those published in the journals Modern Haiku and Frogpond, confirmed the anecdotal impression that in North America at least, during the final decades of the last century, a tercet of free verse had indeed become the accepted norm. Over 90 percent of poems counted were written on three lines, and more than two-thirds to no fixed syllable count. Further, during the period surveyed, the number of those which did adopt a 5-7-5 structure had decreased.

There is no evidence however, that this impressive degree of uniformity, which continues to obtain at the time of writing, had been achieved as a result of universal scholarship, or a definitive advance in the field of relative linguistics. Rather, it seems to describe a default position arrived at through the collision of a number of factors, not least the contrast between a desire to emulate the ‘strictness’ of the original Japanese model and a fundamental urge to freedom echoing the old Imagist proposition that ‘the individuality of a poet may often be better expressed in free-verse than in conventional forms’. Somewhere in there is a goodly portion of self-satisfaction too, plus inertia and unease in equal measure.

As another leading haijin, Lee Gurga, observed in that same year: ‘I think that the world’s haiku poets recognise our common heritage in Japanese haiku, and at the same time acknowledge that Japanese and American haiku will likely grow apart’. Two years later the deal was done, the tortuous convolutions of the 1999 Matsuyama Declaration reading more like some surreal armistice than a discourse on poetic technique. Henceforth it seemed that anything would do as, ‘forcing the fixed form of Japanese haiku and accompanying techniques on other languages is nonsense’.

From the outset, even before Hearn had dabbled with his aforementioned dribbles of prose, another early translator, B. H. Chamberlain, had concluded that the tripartite metrical structure of the Japanese haiku might be best emulated by a tercet of English verse. His 1888 Handbook of Colloquial Japanese contained a number of examples of hokku in translation along with the observation that ‘this particular kind of stanza […] consists of three lines of respectively five, seven, and five syllables’. As we have seen this characterisation of the Japanese metrical foot as a ‘line’ and of a mora as a ‘syllable’ is highly contentious. As a shorthand it is fine. As a linguistic axiom it is false. But a precedent had been set which, for the non-specialist at least, remains the ubiquitous definition of a ‘haiku’ to this day. Other than in Wikipedia.

In fact Chamberlain did not attempt to meet the 5-7-5 syllable standard in his English, choosing instead to reinforce the perception of haiku as legitimate verse by larding his translations with the significantly poetic devices familiar to the bourgeois parlour: end rhyme, strong rhythm, high register language, and archaisms. The suitability or otherwise of such poeticisms did not go unremarked at the time and a slow-burning debate was ignited, pitting artifice against the unadorned, which would eventually feed into the Calvinistic strand of hyper-minimalism that enjoyed a vogue in the United States many decades later.

Interest in Japanese verse during the early part of the 20th century remained slight, with the haiku known principally for its curiosity value. However, exploratory works were written in the majority of European languages throughout, the principal, though far from universal approach, being to write on two or three lines, using techniques drawn from the poetic traditions of the language of composition.

With the end of  World War 2 radical changes in the geopolitical situation saw the rapid development of internationalism in all its forms along with, in the West at least, the beginnings of the counter culture. It is to this period that the true genesis of haiku orthodoxy may be traced, and to a particularly significant trio of author/translators.

R. H. Blyth, Harold G. Henderson and Kenneth Yasuda dominated the nascent haiku movement of the 1960s. Though very different one from another they shared a number of important characteristics. Firstly, unlike the contemporary assortment of proto-Beats, Imagists, Minimalists and devotees of syllabic verse who wished to push the boundaries of Occidental literature, they did not set out to deal in verse influenced by Japanese poetics, these were not haiku-like poems or haiku analogues, they were instead proposed as haiku, pure and simple. The assertion was a bold one: true haiku could exist in a language other than Japanese. And these gentlemen would tell us what they should look like.

As befits such an undertaking, the three sought to base their theorising on something resembling serious study – seeking to deduce the shape of the English genre from an examination of the history, aesthetics, and prosody of the Japanese art. When it came to stanza form, be it for their translations or the writing of original works, all opted for the tercet, though without any thoroughgoing exposition of the reasons.

In the matter of syllable count the approach was more nuanced. All acknowledged some difference in the nature of English and Japanese ‘syllables’, and there were vague intimations that perhaps this might account for the way in which a strict 5-7-5 count in English often felt more restrictive than its Japanese counterpart. Their prescriptions differed though: at least initially Yasuda stuck firmly with 5-7-5, Henderson wavered back and forth, and Blyth, conscious perhaps of the exigencies of a truly faithful approach to translation, allowed for more variation as a matter of course.

Beyond the mysterious properties of the alleged ‘syllable’ a further doubt over exact, or near-exact, 5-7-5 counts soon arose – that of the relative compression of one language in respect of another. Simply put: did the 17 syllables of the English strict-form tercet cause more to be said than those of their Japanese counterpart? Did they give rise to padding? Were 17 English syllables more than were necessary to write a Japanese-identical haiku? And if so, what was the correct number?

Ultimately this was a massively complex question which the linguistic theory of the time could barely frame, let alone answer. The issue was further complicated by an ill-defined feeling that English-language haiku should perhaps omit certain parts of speech as a matter of course – articles, prepositions and conjunctions coming in for particular scrutiny. However, as the 60s turned into the 70s the outlines of a consensus began to emerge: English did indeed require fewer phonemes to convey a message than did Japanese. In all likelihood 17 syllables were too many.

So what to do? One solution lay in the widespread belief that rigour and discipline were important virtues in both the study and writing of haiku. The beginner was after all a mere novice , and there were perfectly good examples of haiku which just happened to work out well at 5-7-5. Therefore the best advice might be for the neophyte to adhere to the strictest of formulae, while the master would be allowed to break the mould. This suggestion sat rather well with the faux-Oriental notions of hierarchy which had already begun to bedevil the world of English-language haiku and one might expect Blyth, with his less than attractive enthusiasms, to be its main proponent. But Blyth was a better man than his prejudices suggest and he had already begun to look elsewhere for a solution to the 5-7-5 conundrum. The metre of haiku in English, he suggested, might best be viewed as accentual rather than syllabic. Written on three lines certainly, but with a short/long/short structure provided by two, three and two stresses respectively.

Remapping – the Future

Those of a sensitive nature can hardly fail to be depressed by the fact that, with the tentative proposal by Blyth of accentual metre, and given the presence of the ever-combustible kire, the stage was now set for a generation or more of strife, by turns apocalyptic and apoplectic, in which the same arguments would be endlessly rehashed with absolutely nothing of merit being achieved beyond the revelation of the general unpleasantness of so many of the protagonists.

To count or not to count? Syllables or stresses? Fixed or free? Cut or continuous? Enemy or friend? This was the epoch that gave us the Haiku Wars, as in Star, and in which that most overweening of compound nouns, the American Haiku, acquired an Archive, though little else. No wonder those polite people at Matsuyama were keen to catch the next train home, and the wider literary community driven to conclude that haiku probably wasn’t poetry at all.

Not that the wider literary community was exactly covering itself in glory. At least not when it came to the metrics of English-language verse. Or should that be ‘scansion’? ‘Rhythm’? ‘Da-dum-di-dum’?

While the illuminati of haiku-land were busy savaging each other over whether or not an em dash should be counted as one of the 17 thingies, those interested in proper poetry were getting very exercised indeed about some hitherto inviolable precepts of accentual-syllabic theory. Thus we find the eminent poeticist T. V. F. Brogan opining in 1993 that the metrical foot is ‘more than a mere analytic tool, a device of scansion: it is a principle of structure’ yet it is ‘almost certainly not an element of performance’. From which he concludes, with majestic circularity, that ‘it can be used to describe and analyse verse whose regularities support it, and not verse which does not’.

The truth is that while the men in high towers seemed reasonably happy with their understanding of ‘syllable’ they were having no such luck with ‘accent’. Or stress. Even as some of us were filling the pages of our copy books with italic slashes and breves, Chomsky & Halle were arguing that there were at least four levels of stress, and probably more. By the time we reach 1998 Peter L. Groves had mercifully dropped this back down to three, as long as account was also taken of whether a syllable was Independent, Dominated, Subordinated, Inhibited, Demoted or… Accented! Simple. Or not, as the case may be. William of Occam would not have been pleased. Nor C. S. Lewis who had wryly observed in his essay Metre, ‘If the scansion of a line meant all the phonetic facts, no two lines would scan the same way’. As the venerable Yoda would doubtless have remarked: ‘Counting the wrong things they are’.

Elsewhere further kernels of academic wisdom were undergoing stress tests of their own, courtesy of the men in lab coats. Researchers into automatic speech recognition, seeking to develop digital algorithms based on the long-held distinction between stress-timed and syllable-timed languages, soon discovered that in practice the theory was of limited utility. Or, to use a technical term, it was ‘pants’. This proved unsurprising to the neurologists and psycholinguists who had long since begun to voice suspicions that patterns of speech production might be largely universal thanks to hard-wired structures and periodicities, not just in the brain, but in the lower-order processing nodes.

By 1996, while Groves was still struggling heroically to get beyond the iambic pentameter with his 11-stage scansion notation, having strapped countless people to chairs and subjected them to trial by metronome, the cognitive psychologists Port, Tajima and Cummins had concluded: ‘It appears that speech timing at this gross, phrase and foot level, appears to have many previously unnoticed similarities to other kinds of motor behaviour. One of the most striking of these similarities is self-entrainment’. Oscillator attraction and all that … the mind generating regular loops from partially randomised information. Which is all very interesting. But a long way from poetry.

Well, that depends. Specifically, it depends on Richard Gilbert, Judy Yoneoka, and their paper ‘From 5-7-5 to 8-8-8 (An Investigation of Japanese Haiku Metrics and Implications for English Haiku)’ published in 2000 by Kumamoto University. As that snappy strap line suggests, by the simple ruse of combining a little bit of scientific method with a willingness to suspend prejudice, Gilbert and Yoneoka show convincingly that when uttered aloud the 5-7-5 morae of a typical haiku are in fact remapped by the speaker on to a more constant substructure composed of a rolling eight pulse.

As with all blindingly obvious revelations this is only blindingly obvious once it has been revealed. What Gilbert and Yoneoka have demonstrated in respect of haiku prosody is what every musician or dancer has always known: the timing and extent of the expressive phrase and gesture is related to, but not limited by, the underlying beat. In the words of Bob Marley: ‘We’re jammin’ right straight from Yah’.

The implications of this work are profound. There is every indication that the nightmarishly complex entanglement of metre and rhythm with which English-language prosodists have been wrestling has simply served to obscure the true dichotomy, that of an elastic superstructure overlaying a more constant substructure. Unless of course we wish to posit that the findings from Kumamoto hold true only for Japanese. The Japanese that is. Something genetic perhaps?

Though these particular findings deal with haiku, nowhere are they more significant than for English-language renku. Long syllables, short syllables, diphthongs, pauses – high stress, low stress, no stress – all can be harnessed to an underpinning beat. By a poet. Using artistry. In so doing choices can be made in respect of that underlying music: it can just as easily have a tangible form as not. And naturally, what holds true within a verse holds equally true without.

So, given that the tercet and the couplet/distich have become the expected standards for writing all kinds of haikai, for haiku and for renku equally, can they be made to work?

Organisation – the Japanese model

The ability of English to code more densely than Japanese has been overstated. Reading different translations of the same Basho text soon demonstrates that the terse minimalism characteristic of much English-language haikai in the latter part of the last century is a matter of aesthetic sensibility rather than a perfect paradigm drawn from the source genre. However the accumulated weight of evidence based on sheer trial and error suggests that those early suspicions pointed at the truth: when writing a long verse in English 17 syllables tend to over-extend the semantic information, the sense, the directly stated meaning. Clearly if this is the case for the 17 syllables of the long verse then 14 syllables will be equally too stretched for the renga and renku short verse also.

Not unreasonably, the first instinct is to set a quantity of syllables that will more closely approximate the ‘substance’ of a typical Japanese verse. The commendably moderate William J. Higginson in his Haiku Handbook of 1985 suggested 12 syllables, which appears to be something of a contemporary consensus, others favouring 13. As we will see there are good phonological reasons why too much precision is unwelcome here, but to do so it is first necessary to examine in a little more detail exactly how the Japanese stanzas arrange their hard information. We have already learnt that on the page the text is presented as a continuous and undifferentiated stream, but how does the reader experience it?

Looking at a hokku or haiku, certainly those employing an internal cut, and many employing a terminal cutting word, though the metrical arrangement is 5-7-5 the semantic structure is most accurately described as ‘bipartite’. The poem separates into two image-sets delivered via a long segment and a short segment mostly, though not always, of 12 morae followed by five morae, or five morae followed by 12 morae. In the majority of strongly bipartite stanzas therefore one image-set occupies a single metrical foot, the other a pair of metrical feet. By implication, poems which do break in the middle of the central metrical foot, if written well, are very striking. If written poorly, simply strident

In the case of the internal long verses of a renku sequence, for which the generic name is chouku, while cutting words are effectively disbarred, it is common for stanzas to employ a kind of soft pivot delivered by a semantic and syntactic disjuncture at the end of the first or second metrical foot. The one-third to two-third semantic proportions are therefore less marked, but still in evidence.

Some haiku, and a great many renku internal verses, deliver what is effectively a single extended image without any form of internal contrast or juxtaposition. Where the language employed is abbreviated the phrasing may pause slightly, though the sense does not turn, at the end of one or other metrical foot. Alternatively the stanza simply employs run-on syntax.

Notable by their rarity are long stanzas which break syntax with each metrical foot. Indeed the majority of schools of both haiku and renku teach that such three-part structures are conceptually flawed. They are nothing less than wrong. For all that the determined will find a piece of calligraphy that seems to indicate otherwise, a Japanese haiku is not a poem written in three chunks. It is most certainly not conceived in three parts, the same holding true for the long verses of a renga or renku sequence.

As to the renga and renku short verse, the tanku: its 14 morae divide into a pair of metrical feet of seven morae each. To all intents and purposes such short verses avoid the most stark forms of haiku-like juxtaposition, either between the metrical feet, or at any place within a metrical foot. Cutting words are unknown. As with the long verse, soft pivots between a complementary pair of image-sets may occur, as may pauses in the phrasing of a single image-set, for the most part at the juncture of the two metrical feet. Similarly the verse may simply employ run-on syntax.

What will be apparent here is that the prosody of the Japanese haikai verse does not automatically link the number or integrity of image-sets to the number or integrity of metrical feet. The semantic and the phonic structures are partially decoupled. The primary level phonics of the long verse, for instance, do indeed run short/long/short, but the image-sets do not.

In Japanese the transition from one metrical foot to another may work in conjunction with an element of syntax, and this in turn may accompany a degree of semantic turn, but it may in fact be no more marked than the mid-line passage from one foot to another in a phrase of English anapestic tetrameter. By contrast, even with the most subtle of enjambments, the transition across an English line ending always involves an inflection of meaning.

In case of the hallowed 5-7-5 strict-form English haiku stanza then, at least as problematic as the 17 syllables are the number and placement of the line endings themselves.

There is another issue which needs to be considered briefly before we return to the question of syllables and accents. Although the significance can be over-stated, grammatical differences between the two languages mean that Japanese enjoys a slight advantage in flexibility of word order. It is possible, so wishing, to segment the semantic components of an expression so that their phonics do correspond to predetermined metrical boundaries while the sense reads seamlessly through. Importantly this syntactic modularity can be achieved even when the boundaries are short. The brutal chopping known in English-language haiku circles as ‘Tontoism’ is therefore more easily avoided.

A personal approach

So to metre and extent: what is the coarse measure of our stanzas; how long should they be? From personal experience – of translation, original work, and reading the work of others – it would seem that Blyth was on the right track. The metre of the English language haikai tercet is most naturally described in terms of accent, the norm being seven (where the distinction is simply between a stressed or an unstressed syllable). The total syllable count will vary depending on word choice but the innate phonology of English leads to a typical range of between 12 and 15 in total, i.e., seven stressed syllables with somewhere between five and eight additional unstressed syllables.

Blyth’s English accent we may heartily approve of. His lineage, or at least his lineation, is a different matter. It is perfectly possible to write good verse following the suggested standard of a 2-3-2 stresses, but the differences between the English line ending and the Japanese metrical foot discussed at length above, taken together with the more modular nature of Japanese syntax, mean that a pattern of 2-3-2 accents offers a far narrower set of verbal constructions than do the5-7-5 morae of strict-form Japanese.

An insuperable problem? No. Since the English line ending cannot be decoupled from the sense, the solution is to allow it to float. Again from personal experience: 2-2-3 and 3-2-2 are distributions which feel perfectly well balanced. When handled with skill, 3-1-3 can be very effective too. By contrast 1-3-3 and 3-3-1 almost invariably fail.

A close mathematical correspondence of the ratio between the 17 morae of the Japanese long verse and the 14 morae of the Japanese short verse would suggest that, if seven accented syllables is a correct English analogue for the long verse, the English short verse should comprise six stressed syllables (the calculator’s suggestion of 5.8 being impractical). And indeed most authors who touch on the subject of English-language linked-verse prosody suggest that these six stresses should divide into equal lines of three stresses each, as this reflects, allegedly, the way in which the Japanese short stanza divides into a matched pair.

We may reject the notion of equal halves outright. For the reasons already adduced in respect of both comparative grammar and the image structure of the Japanese verse, the direct comparison of line and metrical foot is too crude to provide a useful paradigm. Less easy to demonstrate, in truth less easy to understand, is the fact that, after hundreds of Shomon verses in translation, and thousands written in active collaborations, the present author finds that it is five, and not six, stressed syllables that results as being the natural norm for a short verse responding to a seven-stress long verse.

Rather than advance some partially formulated or entirely bogus rationale it is best to simply observe that five stressed syllables, and associated unstressed syllables to a typical range of between nine and 11 in total, seems to be the most effective and dynamic model for the English-language haikai short verse when this is written over two lines. As may be suspected, the stress pattern will equally be 3-2 as 2-3, with 4-1 and 1-4 making an occasional appearance.

Supple Stanzas

The suggestion is that the tercet and couplet/distich may be employed to good effect for all types of original haikai poetry in English, and considered a fair emulation of their original counterparts when they are approached not as fixed form in the narrowest sense, nor as purely free verse, but as supple stanzas employing an accentual metre of seven stressed syllables for the long verse (typical lineation being: 2-3-2; 3-2-2; 2-2-3; 3-1-3) and five stressed syllables for the short verse (typical lineation being: 3-2; 2-3; 4-1; 1-4).

Further, while there will always be the occasional expression which is disproportionately long or short when carried between languages, the implication is that these stanza structures will serve as an effective vehicle for the vast majority of haikai verses in translation, and, by extension, be equally suited to the emulation of other verse forms and genres based on the shichigochou, the fives and sevens cadences of Japanese versification, such as tanka.

Listening and Responding

The reader is respectfully reminded that this proposal of supple three-line and two-line stanzas is necessarily formulated at a coarse level, and that simple conformity to the suggested parameters is no guarantee of artistic success. All the skill lies in the execution, in the judicious employment of those features which the classical prosodist refers to as comprising the ‘rhythm’ of language. And, as Richard Gilbert and Judy Yoneoka demonstrate, these in turn are bent towards the source of a yet more fundamental set of cadences. It is significant, for example, that, when using this supple tercet model, the pause generated within a strongly cut haiku does not need to be accounted for by the superficial arithmetic. As the language of the 1990s would have it: the em dash is not one of the 17 onji.

One thing at least is certain: the contention that elasticity of syllable length, variability of stress, or indeed any other ‘quirk’ or ‘difficulty’ of English somehow precludes the realisation of tangible form is illiterate nonsense pedalled by the crass to the credulous. There are any number of good reasons why a poet may choose to write their English-language haiku as free verse. A lack of alternative is not one of them.

But when it comes to linked verse the argument is different. There are compelling reasons why a sensitivity to form is absolutely central to the development of English-language renku. The Basho school of haikai renga most certainly is concerned with the cadences of verses, the phonic transitions between them, and the structure of extended passages. The alternating beat of long and short stanzas, the balancing and meshing of metrical proportions, these are crucial tools employed by the poet to boost both grace and cogency. As the earnest men in white coats have been trying to explain: part of the pleasure the audience derives from its experience of poetry is to find expectations based on pattern recognition sometimes met, and sometimes subverted.

This article focuses on ways in which the default models of English-language haikai prosody can be made to work to good effect. Ultimately though, the single-most important lesson to be drawn from a reading of the classics is that, irrespective of questions of stanza form, in order to write renku one must first listen. And having listened one may then respond. Not just to the meaning of the preceding verse, but to its totality. And that includes its sound.


Editor’s note: Most of this article appears in 2 chapters of the Renku Reckoner (Darlington Richards, 2015), which are substantially longer. It was offered to Haiku NewZ for publication in this form by the author.

John Carley (1955-2014) lived in Lancashire, England, was a former renku editor for Simply Haiku and the author of the Renku Reckoner (Darlington Richards, 2015). He invented the four-verse yotsumono, celebrating the form with a collection written with several authors, The Little Book of Yotsumonos (Darlington Richards, 2012). A nijuin led by John won the 2013 Haiku Society of America Einbond contest. Read Morning Heat. Working in association with the poet and renku theorist Eiko Yachimoto John has (in recent years) published numerous retranslations of classic Basho kasen.