Defining Haiku by Iterations

I wrote the following essay in March, June, and October of 2006 (with a few revisions in August of 2007 and February of 2012), in response to an essay by Kai Falkman in the Haiku Society of America’s journal Frogpond, which in turn was written in response to an essay by Cyril Childs in the previous issue of Frogpond (Cyril’s essay also appeared in Presence #22, perhaps in a different form). I was not quick enough in submitting my essay for publication in the subsequent issue or two of Frogpond or elsewhere, so my essay lost a measure of its timeliness as a response. Nevertheless, I offer it here, and hope it might be of interest, and not just to those who wish to look up the Falkman essay that prompted it (unfortunately not online). You can, however, find at least a version of Cyril’s essay, “On Not Defining Haiku,” online on the Presence website, together with a response by Brian Tasker [the preceding two links no longer work]. The following essay is not previously published.

“Nothing against 5-7-5, but I prefer haiku.” —Ralf Bröker

“In English, most of the best haiku are not 5-7-5 syllables, and most 5-7-5 poems are not haiku.” —David Oates 


Kai Falkman’s “Swedish View of Defining Haiku” in the Winter 2006 issue of the Haiku Society of America’s journal Frogpond (XXIX:1), and by extension Cyril Child’s essay on definitions in the preceding issue, compels me to respond. It seems that the only true definition of haiku has to be the entire corpus of poems written in the genre in all languages, plus all related criticism, commentary, and translation. If one could possibly get one’s head around it all, that would be a definition—and surely the only true one. But a shorter synopsis is helpful, especially for the newcomer, even though anyone’s choice of what to say will inevitably leave out or de-emphasize something that’s important to someone else. The recent new Haiku Society of America definition has not erred much more than have most other definitions, and it seems unfair, on the surface, to criticize it for what it leaves out when the objective in the first place seems to have been to provide something brief and general enough to satisfy a basic curiosity. Generalist definitions will never satisfy the specialist. Consequently, it seems understandable that Kai Falkman may be dissatisfied with the HSA definition because he wants a specialist definition—as do I and many others.

        Let’s be more specific—both agreeing and disagreeing on various points. First, as David Cobb and others have written, perhaps it’s not nature that haiku is really after, but a seasonal reference. Just as the 5-7-5 canard dies hard, so does the popular belief that haiku is meant to be a nature poem. Maybe not. If haiku were just after nature, whole swaths of every saijiki—the parts about human festivals, for example—would have to be removed. Consequently, one may question Falkman when he says “I agree with Cyril Childs that seasons are a part of nature and not necessary to include in a definition.” No, it’s the other way around: nature is a part of the seasons (in haiku terms), and the seasons of life on earth include more than just nature—or terms such as “Tanabata” or “Christmas” would not be listed as season words. And remember that a kigo is a season word, not a nature word, so of course seasons—not necessarily nature—need to be included in a definition of haiku.

        It also seems ill-advised to “not differentiate between haiku and senryu,” at least not for the reasons suggested. Falkman seems to be saying this based on the practice, in Sweden, that haiku address both nature and human situations, presumably based on the assumption that haiku is defined as nature, and senryu is defined as human nature. But it’s easy to disagree with this perception, and thus to disagree with the conclusion to not differentiate between haiku and senryu. Just as the hokku (starting verse) of a renga or renku is differentiated from the interior verses because it has a kireji and a seasonal reference (and the internal verses should generally not have a kireji or be like haiku), so too senryu is largely differentiated from haiku by not intentionally having kigo or kireji. It’s not just the “nature versus human” or “serious versus humourous/satirical” distinctions that differentiate haiku and senryu as many people seem to believe. So it seems to me that Swedish haiku poets, to the extent that Falkman can speak for them and offer a “Swedish” definition of haiku, are using the wrong reasons to not differentiate between haiku and senryu. Both haiku and senryu can have nature content; rather, it’s the presence or lack of seasonality that matters more. Both can be serious or humourous; it’s the use or lack of the kireji or two-part juxtapositional structure that matters more. There seems to be little problem considering senryu as a subset of haiku (because there are many similarities between the two), but it does seem important, conceptually, to distinguish haiku from senryu (which is not the same as segregating them, which may well be unnecessary in haiku journals, including Frogpond). In any event, in actual poems, it’s worth remembering that the distinction is more of a continuum rather than a black-and-white dichotomy.

        Falkman offers an excellent reminder, though it’s far from new, when he asserts that haiku should employ “concrete” images. Perhaps it’s merely a matter of degree or emphasis to add “concrete” to the definition, but the fundamental value of employing “images” is to recognize the effectiveness of depicting what we can experience through our five senses (“image” is therefore not just visual). Thus, it seems worthwhile to clarify Falkman’s suggestion to say that haiku employ “concrete sensory images.”

        Where Falkman’s perspective seems most problematic is his insistence on mentioning syllables. They are not irrelevant to haiku, of course, but, in a definition of English-language haiku, the concept of haiku as a syllabic form has got to be put to bed (shot to death, rather) and be dispensed with once and for all. Every single poet writing haiku—or new to haiku and wondering about the “syllabic” nature of haiku—owes it to himself to get a copy of Kōji Kawamoto’s The Poetics of Japanese Verse (Tokyo: University of Tokyo Press, 2000) and read Appendix I: “The Japanese Mora.” This brief exposition of Japanese linguistics far supersedes Keiko Imaoka’s still valuable but essentially amateur essay “Forms in English Haiku,” which I first published in Woodnotes #29 (Summer, 1996), now available online. There is likely no more definitive exposition and implication than Kawamoto’s for why haiku should never be considered “syllabic”—in both Japanese and English, as well as other languages. The Japanese haiku emphatically does not count syllables but morae (the plural of mora). As Kawamoto emphasizes, “the counting units which it [Japanese verse] employs are not syllables but morae” (p. 297; translation by Stephen Collington), and he explains how a simple English word such as “sign” contains just one syllable but three morae. Abigail Friedman, in The Haiku Apprentice (Berkeley, California: Stone Bridge Press, 2006), points out that the one-syllable word “scarf” becomes four sounds when said in Japanese: su-ka-a-fu (p. 87). Even the word “haiku” itself shows the difference in languages—it’s two syllables in English, but three sounds in Japanese. So of course, speaking strictly, seventeen syllables in English is too long compared to haiku in Japanese. Haruo Shirane further emphasizes the point in introducing Kawamoto’s book, should anyone not be paying attention. Let the proper point be forever understood: Quite simply, as Shirane declares, “the term syllable is an inaccurate way of describing the actual metrical units of Japanese poetry” (p. viii). Indeed, no definition of haiku in English need mention syllables ever again. Please!

        With Kawamoto’s evidence in mind (not summarized here), it seems either misguided or overly simplistic to say, as Falkman does, that “haiku traditionally consists of seventeen syllables, though often less in contemporary practice.” To the extent that the first part of this claim has ever been true in English, it should be considered an error, and it would seem harmful to perpetuate this mistaken understanding in any definition for English-language haiku. To say that “haiku traditionally consists of seventeen syllables” isn’t even true in Japanese, as Kawamoto makes clear, let alone what haiku should be in other languages. In English, the syllabic count of 5-7-5 is not only the wrong target, but aiming at such a target often occurs at the expense of (or out of ignorance of) other more important targets. No, if anything, a definition should say that traditional haiku, in Japanese, consists of seventeen morae. The term “morae” might not be widely understood, but it’s a Western linguistic term applicable to Japanese, not a Japanese term (although it is now used in Japanese as a loan word), so one would hope that a Western audience would know the term or take the initiative to study it. Many dictionaries give it a simplistic definition, however, so perhaps a deeper linguistic understanding of the term would be advisable, but Kawamoto’s essay provides ample explanation for those who are determined enough to read it. Because of further differences in language, however, it may well be best not to offer a limiting English equivalent to seventeen morae, but to dwell, instead, on the formal characteristics of English-language haiku typically having three short lines, a seasonal reference, and a two-part juxtapositional structure. I would also encourage the reading of William J. Higginson’s essay, “Haiku by the Numbers, Seriously,” recently published online.

        Speaking of lines, the three-line structure common to haiku in English is actually a Western construct, for haiku in Japanese are typically written in a single vertical line, as many of us know. The three lines of a Western haiku emulate the three parts of the Japanese rhythm, of course, but haiku need not be limited to three lines so long as it still meets the more important requirements of the juxtapositional structure and seasonal reference. Whether in two or three or more lines, the characteristic of having line breaks also helps signal to the reader that he or she is reading poetry rather than prose (or something else), which seems helpful in English rather than requiring a single line to emulate the Japanese presentation of haiku in a single vertical line. One-line haiku can be effective in English, but they often leave most Western readers unaware that they are reading poetry (despite Hiroaki Sato’s insistence on monostich haiku). Thus, it is no wonder that a lineated form for haiku has taken strongest root in English, with the three-line variation being the most common and most logical equivalent to the three-part structure of haiku in Japanese. Dhugal Lindsay and others have proposed that some metrical form may turn out to be appropriate and be canonized in English, but that form, if any should ever be appropriate or dominate, should definitely not be 5-7-5 syllables.

        A vital point, too, is that haiku seems to be best thought of as a genre of poetry, not a form, and that form is only one aspect of the genre. A widespread adoption of this perspective would help to counteract the undue emphasis on form by most Western poets new to haiku—and even mainstream poets who think themselves more informed on haiku than they really are.

        So what is accomplished here? I should like to hope that what I can offer is a more refined definition than the seemingly generalist one from the venerable HSA definition committee, and the specialist ones offered by Falkman and Childs. But one man’s treasure is another man’s trash, and any of hundreds among us could offer varying—and worthwhile—definitions. This discourse won’t be in vain, however, if the following proposal might at least contribute an iterative step towards an agreeable specialist definition for haiku in English, the first sentence of which might serve as a generalist definition:


Haiku in English is typically a three-line poem that uses concrete sensory images to convey or imply natural and human seasonal phenomena, using a two-part juxtapositional structure as well as simple and primarily objective language. Originally a Japanese genre of poetry, now written and adapted in many languages worldwide, traditional haiku in Japanese consists of seventeen sounds (not to be confused with syllables) in a pattern of 5-7-5. Because of differences in language, this rhythm is generally not followed for literary haiku in most languages other than Japanese. As intuitive and emotional poems, haiku often capture a sense of wonder and wholeness in presenting existence such as it is. Rather than presenting one’s emotions, haiku present the cause of one’s emotions, thus empowering the reader to have the same intuitive reaction to an experience that the poet had.


        Rather than end with the preceding definition, I’d like to quote Maureen Ritter, a student in the Global Haiku Traditions program run by Randy Brooks at Millikin University in Decatur, Illinois. In her introduction to Caroline Gourlay’s Lull Before Dark (Brooks Books, 2005), Ritter candidly—and laudably—writes the following (p. 8): “Haiku may be defined as a simple, brief poem containing a seasonal reference and employing a crafty method of ‘cutting,’ but anyone who truly knows and loves haiku will tell you that defining the techniques of writing will never do justice to the essence of haiku.” Indeed.