Simply Haiku: A Quarterly Journal of Japanese Short Form Poetry
Contents Archives About Simply Haiku Submissions Search
Autumn 2006, vol 4 no 3

A column by George Swede

In the psychology of art, the term "top-down" describes a quest for knowledge in which ideas are formulated first and then applied to works of art and their creators. A good example involves Sigmund Freud who, after developing the theory of psychoanalysis, used it to explain the art and creativity of Leonardo da Vinci (1947). The term "bottom-up" denotes the opposite strategy—data is gathered first and then used to build a theory. For instance, psychologist Colin Martindale (1975) tabulated the number of times poems contained unusual imagery over hundreds of years of published poetry in England and France. He then employed the results to originate his theory of poetic change.

In 1980, Eric Amann and I published some bottom-up research involving a survey of all the haiku published in Cor van den Heuvel's The Haiku Anthology (1974) and in a work I had edited, The Canadian Haiku Anthology (1979). We wanted to see how closely the haiku composed by the best writers of the time adhered to the then current and top-down definitions of the haiku, and, in particular the one issued by the Haiku Society of America (HSA) in 1973 (with slight modifications in 1976):

(1) An unrhymed Japanese poem recording the essence of a moment keenly perceived, in which Nature is linked to human nature. It usually consists of seventeen jion [sic] (Japanese symbol sounds).
(2) A foreign adaptation of 1. It is usually written in three lines of five, seven, and five syllables. (The Haiku Society of America Twentieth Anniversary Book Committee, p. 82)

Among our findings was that reference to nature occurred in 90% of the haiku, while the balance dealt exclusively with human affairs, thus strongly supporting definition one. We also found, however, that about 80% of the haiku were free-style with the remainder in 5-7-5 mode—a result greatly at odds with definition two.

Missing in our research, however, was an empirical examination of the phrase "in which Nature is linked to human nature"; in other words, how does the linkage actually occur? To answer this question, I decided to do another survey that looked in more depth at haiku with nature content, namely, how many involve only nature references and how many also include mention of humans (or human artifacts). Put another way, how many haiku involve only nature and thus serve as metaphors for human existence and how many make overt connections with humans (or their products)?

This time (without Eric Amann) I did a bottom-up analysis of Cor van den Heuvel's second edition of The Haiku Anthology (1986), focusing on 350 haiku by the twelve most prominently-featured poets (Swede, 1992). The analysis revealed that their haiku could be divided into three content groups: those involving only nature (N) at 22.8%; those that contained humans and/or their artifacts along with nature (N + H), 61.4%; and, finally, those that were concerned only with humans (H), 15.7%. The fact that N plus N + H haiku accounted for 84.2% of the haiku (down from the 90% in the 1980 study) was another general confirmation of the 1973/76 HSA definition. But my bottom-up data made clear the way in which the twelve poets dealt with nature—most of their work connected it to humans and/or their artifacts.

My results also suggested that the almost 16% H-haiku (up from the 10% in the 1980 study) should be considered as senryu rather than haiku because they had no nature content, other than that related to humanness. Such an interpretation was supported by the 1973/76 HSA top-down definition of the senryu:

A Japanese poem structurally similar to the Japanese haiku . . . but primarily concerned with human nature. It is usually humorous or satiric (The Haiku Society of America Twentieth Anniversary Book Committee, p. 82).

What needed clarification was how the senryu is "primarily concerned" with human nature. The 1992 results suggest that the sample poets considered a senryu to be a haiku-like poem, but with no nature content.

Eight years later, I published a chapter, "Towards a Definition of the English Haiku," in an anthology edited with Randy Brooks (2000). Using bottom-up data from the two studies above, I concluded that, for elite haiku poets, the haiku typically involves five factors:

1. brevity—about a breath-length long;
2. some aspect of nature other than human nature;
3. sense images, not generalizations;
4. the present tense:
5. a feeling of awe or transcendence (with modifications, from p. 33).

While these five guidelines seemed to be a significant finding at the time, four subsequent attempts at haiku definition have ignored these data-driven criteria. Instead, their focus has been on top-down or personally-generated meaning (Spiess, 2000; Missias, 2001; HSA, 2004; Verhart, 2006). For instance, if the writer has a Zen background, his or her view involves Zen principles in the definition. The same can be said for those with other backgrounds, such as existentialist, experimental, post-modernist, realistic, surrealist, symbolist, traditionalist, etc. In all cases, little attempt is made to reconcile personal views with what has actually appeared in the best haiku publications.

More troubling is that the proponents of the top-down approach do not acknowledge by any means, not even footnotes, that bottom-up data even exist. For instance, nary a mention of such work is made by A.C. Missias in her 2001 Frogpond article, "Struggling for Definition." Her strategy was to find common elements in twelve (twenty-five words or less) definitions (including one of mine and the 1973 HSA version) published earlier in an article, "Definitions of Haiku," by Robert Spiess in Modern Haiku (2000). At first glance, her analysis seems bottom-up with its use of two graphs and terms such as "objective" and "subjective." But, it employs a survey of definitions rather than of haiku to arrive at its conclusions. This is akin to determining a world champion in boxing on the basis of how convincingly twelve leading boxers have bragged about their abilities to win the prize. Her analysis, while thoughtful, could have benefited from some presentation and discussion of bottom-up data. The twelve sample poems Missias includes in her article do not count because they were carefully chosen to support her final arguments; that is, they are the result of top-down processing.

Perhaps in response to the Spiess article and the later discussion of it by Missias, the HSA re-tooled its 1973/76 definition in 2004:

A haiku is a short poem that uses imagistic language to convey the essence of an experience of nature or the season intuitively linked to the human condition. (Haiku Society of America)

A few explanatory notes follow and some comments are made about general practice, but no actual data are provided about how the best writers intuitively link the human condition to nature or the season. Thus, HSA's 2004 version remains essentially a top-down definition, like its 1973 and 1976 predecessors.

The most recent example of a top-down strategy similar to that of Missias is the work of Max Verhart who asked 29 poets from 19 countries to give their definitions of haiku and then, like Missias, tried to find common threads (2006). While enlightening, Verhart's results can be criticized for the same reasons as those of Missias and the HSA: absolutely no mention is made of bottom-up data.

I could provide more examples of the neglect of bottom-up findings, but I think my point has been made. In my opinion, the ideal way to deal with haiku definition is to come equally from each direction, top-down and bottom-up. While such a balanced perspective is hard to achieve, fair compromises can be made. A scholar with a top-down bias can at least acknowledge that there is a bottom-up way of getting to the same place and a scholar with a bottom-up bias can do the same in regard to top-down methodology. Then readers will at least be given an opportunity to follow up on whichever approach appeals to them more.



Freud, S. (1947). Leonardo da Vinci. New York: Vintage Books (first published in 1916).

Haiku Society of America (2004). Report of the Definitions Committee Adopted at the Annual Meeting of the Society, New York City, 18 September 2004.

Martindale, C. (1975). Romantic Progression: The Psychology of Literary History. Washington, D.C.: Hemisphere Publishing.

Missias, A.C. (2001)."Struggling for definition." Frogpond: The Journal of the Haiku Society of America, 24:3, 53-63.

Spiess, R. (2000)" Definitions of haiku." Modern Haiku, 31:3, 74-75.

Swede, G. & Amann, E. "Towards a definition of the modern English haiku." Cicada, 4:4, 3-12.

Swede, G. (1992). "Elite haiku: Hybrids of nature and human content." Modern Haiku, 23:1, 65-72.

Swede, G. & Brooks, R. (2000). Global Haiku: Twenty-five Poets World-Wide. Cullercoats, Northumberland, UK/Oakville, ON, Canada: Iron Press/ Mosaic Press, 2000.

Twentieth Anniversary Book Committee (1994). A Haiku Path: The Haiku Society of America, 1968-1998. New York: Haiku Society of America.

Verhart, M. (2006). The essence of haiku as perceived by western haijin. Max Verhart sent an email attachment of this paper to me in May, 2006. The paper contained the following footnote:

A German version of this paper was read by the author on 29 April 2006 for the Haiku Kreis in Frankfurt am Main, Germany. The text will be published in two parts in Sommergras, the German Haiku Society's quarterly. A Dutch version will appear in Vuursteen (Autumn 2006), the Haiku Circle Netherlands' quarterly.