Introduction to Renku

by John Carley

Types of Renku

There are several types of renku, but among those being written today are:

  • Kasen – 36 verses
  • Triparshva – 22 verses, divided into 3 “sides” of 6, 10 and 6 verses respectively (invented by Norman Darlington of Ireland)
  • Nijuin – 20 verses
  • Junicho – 12 verses.

Ways of Composing

There are two ways of composing a renku – degachi in Japanese, which more or less means ‘competitive’ and describes a system whereby participating poets go head to head for each verse position, or hizaokuri, which translates as ‘passed along the knees’ conjuring up the image of the kaishi (manuscript paper) being passed from one poet to the next. Hizaokuri means ‘by turns’.

As in all things there are pros and cons – hizaokuri is generally a bit quicker and also puts participants on the spot to really tackle the demands of a given verse position as there is not the entirely natural wriggle room of expecting suitable offers to come from someone else. The typical practice of working by turns is to ask the person on the spot to submit three candidate verses simultaneously which are then discussed by the company.

However, a ‘pass’ is an option open to all people at all times when working by turns – sometimes pressure is not all it’s cracked up to be. In the late medieval and early modern period contenders in Japanese haikai competitions had to complete their offers before a little spiral of incense burnt down. No pressure there then! Still, it kept the insects at bay.

In my opinion though, the single-most important feature of working by turns is that participants have to be prepared to consider amendments and redrafts of their work by any and all other participants, with the Sabaki (renku leader) being judge, juror and executioner. In my experience this can be quite a significant psychological barrier to overcome – my own first reaction to witnessing this was ‘how ******* dare she!’.

However, common practice in the Edo period was for the renga master to rework the entire sequence before publication. This is very rare in Japan now, and any such liberties could only be taken by a Sosho – a Master.

Hokku & Wakiku (first & second verses)

The first verse in a renku, no matter which form you are using, is the hokku, while the second is the wakiku. The hokku is structurally and functionally indistinguishable from a haiku. In the Edo period very many hokku were also written as a figurative greeting or augury for the assembled company and/or the prospects for the poem.

The hokku may take ‘moon’ as its topic. In a Junicho alternatively, and unusually, it may take ‘blossom’ or ‘flower’ as it topic. Please note the ‘may’ – neither moon nor flower are obligatory. In terms of season words – I do not advocate using kigo which belong wholly or principally to Japanese literature.

Typically the hokku is a ‘cut’ verse, employing the technique of juxtaposition or combination called, in the Japanese, toriawase. The two ‘parts’ of the stanza are articulated or intensified by the presence of a ‘cutting word’ (kireji) – the resulting verse being in all aspects ‘self-sufficient’ or ‘stand alone’ (tateku).

The hokku is therefore the precursor to the later ‘haiku’ and is the only stanza in a renku sequence which may usefully be considered as ‘like a haiku’.

In most forms of renku the relationship between hokku and wakiku is specially close – often compared to that between the upper and lower halves of a tanka (or, more accurately, of a bi-partite waka). For this reason, until recent years, the wakiku always took the same season as the hokku. The Junicho is unusual in that it admits to a wakiku that may be non-season (there are very few things in any form of renku that are obligatory). Even where the relationship is less strict the wakiku is however, constructed to provide a sense of ‘closure’ or ‘completion’.

Daisan (third verse)

The daisan is the ‘break-away’ verse. Whereas hokku and wakiku might read as a unit, with the arrival of daisan, the sequence begins to unfold.

Links between daisan and wakiku will tend to be more free than that between wakiku and hokku, while tone, setting and narrative perspective can all be expected to differ markedly from the initial pair.

Daisan are expected to open outwards – to be both germinal and unfinished, suggestive of multiple possibilities.

Ageku (last verse)

The name implies not just an ending but also the fulfilment of anticipation: ‘at last’. In classical renku the ageku takes the same season as the preceding verse (spring blossom), though in some recent variants it may be any season, or none.

Whatever the seasonal aspect, the ageku has a function mirroring that of the hokku – this time combining elements of summary, salutation and augury. To have the freedom to meet these demands the ageku may be largely exempted from the more rigorous demands of link, shift, and variety that condition all other verses (hokku excepted).

The composition of the ageku is like that of the hokku, a special honour. The same poet would not be expected to figure in both, an exclusion which generally includes the wakiku, and may also extend to daisan.

Love Verses

Love verses should really only deal with relationships which might find sexual expression. Therefore they feature adults. Get tactile if you like, but love verses stop short of pornographic levels of detail, or really coarse suggestiveness.

Both Edo period and contemporary renku (Shomon haikai-no-renga) do have love verses which centre on homosexual attraction, or are capable of such a reading. By contrast Japanese friends have told me how scandalised they were to read Occidental sequences that had love verses talking about children or animals, before realising that there was a very unfortunate misunderstanding at play!

A confession: I really don’t like love verses. In this I’m similar to Basho – who kept them down to the absolute minimum or made sure his mates got them. I suspect his reasons were the same as mine – we’re both rubbish at writing them.

Link, Shift & Separation

The core generative mechanism of renku is that which operates within any given trio of verses: link to the preceding, shift from the one before that. Beyond that the turnover of fresh material is governed by sarikirai – the idea of ‘minimum separation’… but minimum separation of what?

The answer is, things that belong to the same category. And the precise number of verses we have to wait depends on the ‘primacy’ of the category. Really prime things have to have greater separation.

To have a full appreciation of ‘shift’ (the argument goes) you have to have a shared cultural appreciation of what belongs in which category, and of the relative primacy of any given category. If ever you’ve had the misfortune to witness a ‘renga master’ tell a participant that they can’t have ‘moth’ as their candidate for verse #32 because we’ve had ‘mansion’ at position #15, and both have ‘wings’ – you’ll know that our concerned theorists have a point. A really good one.

There’s a brilliant section on classical theories of both link and shift in Herbert Jonsson’s Haikai Poetics: Buson, Kitō and the Interpretation of Renku Poetry.  Herbert reproduces one of the ready-reference tables of categories that many people used (and still use) when composing renku.

These arguments about cultural specificity are absolutely central to the prospects for genuine art in English-language haikai (or indeed for any haikai in any language other than Japanese).

There is a lot of misunderstanding about repetition in renku. In English the term most frequently seen is ‘backlink’. Unfortunately this gives the impression that, rather than having forward momentum, renku spends all its time looking over its  shoulder; and proposes that the generative force of renku is governed by a sole aesthetic principle – that to create good poetry it is sufficient to avoid all and any repetition.

In contemporary renku there are three basic principles which counteract repetition: uchikoshi (more properly kannonbiraki); sarikirai; and torinne.

The late and truly great Master Meiga Higashi identified uchikoshi as the sine-qua non of renku composition. He proposed that even if every other convention and consideration were disregarded any piece of poetry which respected ideas of uchikoshi would have to be treated as renku. Personally I think he was being too liberal, but he was a Master.

The core dynamic of renku resides in any set of three verses (looking backwards):

  • Added verse (tsukeku)
  • Head verse (maeku), so named because it is the lead-in verse for the added verse
  • Last-but-one verse (uchikoshi).

So, in the sequence K, L, M, N: ‘k’ is uchikoshi to ‘m’, and ‘l’ is the head verse for ‘m’. Similarly ‘l’ is uchikoshi to ‘n’, and ‘m’ is the head verse for ‘n’.

The crucial link and shift dynamic means each new verse must link to its head verse, but be entirely different from the last-but-one. When, for instance, ‘k’ and ‘m’ do not show sufficient difference this failing is called kannonbiraki. The word means ‘double doors’ and refers to the tabernacle of the Buddhist altar which open outwards to either side, symmetrically framing the centrepiece. Rather confusingly the word uchikoshi is sometimes used as an alternative to kannonbiraki – so it may refer either to the ‘leap-over’ position or to the undesirable similarity between added verse and last-but-one. The principal of uchikoshi (kannonbiraki) means that there should be no similarity between added verse and last-but-one, other than possibly belonging to the same seasonal segment, or to the ‘love’ section.

Rinne is another term drawn from Buddhism. It might be given as ‘reincarnation’. In medieval renga theory rinne designated any situation in which the norms of minimum separation (sarikirai) were breached. It must be borne in mind that medieval renga manuals were very heavily codified and proscriptive. In contemporary terms the meaning of rinne has changed to embrace the wider sense of poetic sensibility needed to avoid gross repetition in shorter sequences. In this it closely resembles the medieval renga principal of torinne or ‘distant reincarnation’, and many contemporary theorists have adopted this terminology.

Torinne, in the modern sense, is more subjective than any strict category type of rule such as sarikirai. Its action is not limited by a designated proximity. If an added verse strongly recalls another verse from anywhere in the poem the accusation of ‘distant reincarnation’ can be levelled. But torinne does not operate at single word/idea level. It is applied to the complex of the verse’s meaning and/or phrasing.

Beyond these three broad principles there is one convention worth mentioning which does resemble the simplistic notion of ‘backlink’. Many renga masters will disbar all and any repetition of core semantic elements which have appeared in the hokku. In English…. well, if we have ‘lighthouse’ in the hokku we might want to question ‘streetlight’ anywhere else as both contain the element ‘light’.

Renku is not the search for novelty at any cost. Our overriding concern in respect of repetition has to be directed not at word level, but at stanza level, and ultimately in the context of stanza-to-stanza relationships.

Kigo (season words)

We all have to be able to refer to our own experience, our deep-level cultural referents, how can this be art otherwise? Anybody who has a problem with ‘Christmas barbecue’ can safely be ignored. And an editor who can’t run to a footnote to clarify the occasional seeming anomaly, or rejects a work for fear of one such, is editing something fit only for the nail on the dunny door.

The reason why Basho and his buddies sold so many books was because they successfully meshed entirely separate worlds of experience – those of the highest and lowest classes of early modern Japanese society. The differences between our various national frames of reference might be otherwise in scope, but they are hardly greater in degree.

I fully agree that where participants are from different hemispheres it is unhelpful to use month names as kigo markers, this also goes for festivals which have their origin on one hemisphere but are now universal. There is in existence at least one Triparshva written in India to a schema which formally adopts the six season that are locally identified.

This is the future. Pretending we were all born in Kyoto is so damned odd as to be deeply unfortunate.

Not a String of Haiku

Because the misconception is so commonplace that the interior verses of a renku sequence can be considered as a string of haiku I, and many other persons wishing to extend a more accurate understanding of the genre, push the message that only the hokku is in fact indistinguishable from a haiku.

This is a true statement, and in terms of mainstream renku theory – which of course means ‘Japanese renku theory’ – it is unequivocal. Unfortunately, because English-language haikai prosody is in a state of flux it is not so easy to make such absolute distinctions.

And if we look at translations: the shortest rendering of Basho’s blighted frog that I’ve had the misfortune to encounter was courtesy of Lucien Stryck at, from memory, seven syllables, while my all-time hero and sometime mentor Nobuyuki Yuasa gave us a graceful 22 (again, from memory).

Does this matter? The simple answer is ‘no’, and ‘yes’.

No, it probably isn’t crucial when we’re writing haiku because the verse exists in its own right, and on its own merits. Yes, it matters terribly when we’re writing renku because the verse does not exist in its own right, or generate its resonance on its own merits.

So, an important point here – renku is not a succession of individual verses, it is a sequence of dependencies. The consonances and dissonances of utterance, and the organisational structures that govern the phrasing of our individual verses, also govern the relationships between verses.

In the genesis of renku this was unexceptional as Japanese haikai prosody was, and overwhelmingly remains, a matter of fixed form (teikei). Trying to equate this in English, or indeed in any language other than Japanese, is a fascinating challenge.

A renku makes use of all the features of language as it strings the pearls together. We disregard phonics at our peril. The fact that there is a direct trend to do so in English-language haiku is a different argument.

Tomegaki (debrief)

In contemporary formal Japanese renku circles it is customary, on the completion of a poem, for the sabaki to post a tomegaki. This word has the same root as that for ‘clasp’ and is a kind of debrief that draws some important strands of the compositional discourse together. The participants will also post a kanso. This word means something like ‘appreciation’, though it is not absolutely de rigeur to be unfailingly flattering.


It is essential to remember, when learning these conventions, that renku is art. It is not a forensic investigation, or a high school debating society. Renku is about periodicity and modulation. It deals not so much with absolute novelty as with recontextualisation. Renku cannot be written by adherence to ‘rules’. We are artists. We must understand our materials. And create.


Editor’s note: 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.

This article has been formed from comments John made as Sabaki for the Yellow Moon junicho at Issa’s Snail, as well as from articles posted elsewhere. It appears here with the author’s permission and with information updated.