Dogwood Leaves


Ah, Masaoka Shiki…..

The dogwood 
     tree leaves.
There really must be
     red powder
       in the autumn breeze.

Feet, Not Syllables


Michael, thanks for your comments.


In your article you wrote that poets should understand what syllables are and should know how to count them, but the truth is that English language poets have never counted syllables, rather they they have counted the feet which measure the metric value of stressed and unstressed syllables. This is something different than just counting syllables.


English is a stressed timed language, which means that syllables aren’t uniform in the length of time and some are longer than others, this is different from syllable timed languages where every syllable has the same amount of time with no variation between them. This is why we in English count feet rather than syllables.


The first line of haiku that you quote in your article is an example of why this is:


tired old work horse

stands thirsty and sweating in

summer’s sizzling heat


You state that the word “tired” has been mistakenly counted as two syllables here and it is true that “tired” spoken alone is a one syllable word. But what happens when you read it with the same timing as the other one syllable words that follow it?


/           x        /         x

tired old work horse


By the time we get to the end of the line we are all of out breath and just crash and cannot easily connect the sentence nor the thought to the next line. We are hung out at our wit’s end here.


Even when we change the stress pattern:


x            /        x          /

tired old work horse


The same crash at the end happens. Now let’s scan it like the person who wrote it did:


x    /     x        /          x

tired old work horse


By reading tired as having a stress at the end of the word we have placed it in rhythm that leads smoothly into the next line and allows our thought to follow as well.


It’s easy to get fooled into believing that by just counting syllables we can get the music measure of poetry because so much of our good poetry has been written by using similarly timed syllables, but that doesn’t mean that there wasn’t any written that didn’t use syllables that had different time values. I had never thought of the difference of counting feet until I read your article, so I am guilty of it as well.


Of course, there is a way to write this line to make tired read as one foot and still keep it in a natural speech rhythm:


tir’d old work horse


By using the elision the stress is taken out and the word fits into a natural rhythm. Poets in the past used this a lot as a way of to fit the past tense -ed verbs into rhythmic patterns because it knocked off a syllable from the word. It would do the same to the verb which added an extra foot to the second line of your raccoon haiku.


Of course, just using an article like we are supposed to do would do the trick as well:


a tired old work horse


This takes out the stress as well. I’d guess the reason why we stress the word without the article is because we unconsciously fit it into a workable speech pattern. But that is the kind of affection in language which happens when you try to mimic the writing in Japanese.


Stress is, of course, is something that Japanese language really doesn’t have, but it does use pitch and by looking at pitch we can get some ideas about how Japanese words work as feet (and syllables) in relationship to how we see them per our words.


To quote from the translated version of “The Poetics of Japanese Verse” by Koji Kawamoto (pg. 189-190):

“In modern standard Japanese, words of two morale or more can be divided into two broad categories, namely, those in which the first mora only is pronounced with a high pitch (A-me, SU-ga-ta, O-o-ka-mi, KA-ge-bo-shi) and those which start with a low-pitched first mora followed by one or more high-pitched mora. Of these, the latter group can be further subdivided into those words which end on a high pitch (sa-RA, o-N-NA, to-MO-DA-CHI, o-SHO-O-GA-TSU) and those which return to a low pitch before ending (ko-KO-ro, ka-RA-KA-sa, i-NO-CHI-BI-ro-i, ka-MI-NA-RI-O-ya-ji). The arrangement of high and low pitched morae within individual words thus comes down to a simple dichotomy:  “whether a high pitch will stop one mora, or continue for two or more –not forgetting of course the restriction that a high pitch cannot emerge more than once within a word (i.e., with low-pitched morae spaced in between)”  (quote is from Shibata Takeshi’s “Nihongo no akusento”).


The important part to consider in the above is how “a high pitch will stop at one mora, or continue for two or more” because it explains how a high pitch will eat up feet. A word noted above like “tomodachi” shows how the high pitch will take a 4 syllable (mora) word and reduce it two two pitch feet, and how in a Chinese character compound “onyomi” word like “oshoōgatsu”  it moves  across two characters and reduces six mora into two pitches. It can even move across two compound words like “inochibiro” and two separate words like “kaminari oyaji” to reduce 6 mora into 2 pitches and 7 mora into 4 pitches.


In the previous post I mentioned that “kawa” was one syllable and you corrected me by saying that it was two and you are right, ka-WA makes it two pitches. But if you put it together with something like se-MI you will still get the word ka-WA-SE-MI (kingfisher), which is a good example of how the high pitch dominates the low pitch in terms of feet and changes something that you would think naturally would be 4 pitches into being 2.


There are dictionaries in Japanese that are only for showing how to accent words properly and using the NHK Japanese Pronunciation Accent Dictionary (NHK日本語発音アクセント辞典) I went and checked where all the high pitch and low pitched was for all the words in the haiku that were in Keiko Imaoka’s article.


yuKU HARU YA/  toRI naKI uO NO ME ni NAmida

Has 11 transitions between high and low pitch.


NEko no meSHI shoUBAN suRU YA / suZUME NO ko

Has 10 transitions.


WAre to KIte aSOBE YA /oYA NO NAi suZUME

Has 11 transitions.


uGUisu no naKU YA / CHIISAki kuCHI aKETE

Has 9 transitions.


So once you start to consider these haiku on the terms of how our language is scanned the Japanese lines actually shrinks in feet, which something that is impossible in English, and thus shows the difficulty of trying to consider what the length of haiku should be in English. You can just look at it uncritically and say that Japanese haiku scans less this 17 syllables,  so English should also be cut down to 11 syllables, but to argue that this is acceptable because 17 syllables in English carries more information than the Japanese misses the fact that the Japanese carries much more information per pitch foot than the English does. Which is how it is able to shrink 17 down to 11 or less in the above.


Keiko Imaoka made mention of how “The mnemonic quality of 5-7-5 Japanese phrases is much closer to that of metered rhymes in English” and the theory of how this works is explained clearly in Kawamoto’s book. In 1922 Doi Kichi developed the idea of the “bimoraic foot,” which is “a unit of time which is always –at least in theory– of the same duration, regardless of the number of (vocalized) morae” (pg. 200). This foot “entails one important condition, namely, that a single mora alone may count for a whole beat under certain circumstances.” This is accomplished when this single mora is made to “either lengthen (the additional mora) or place a stop (pause) after it so that it balances with the preceding two.” (pg. 202)


Doi also argued that bimoraic foot showed a stress pattern and that by “clapping or tapping to the beat” one can understand “where such stresses occur, it soon becomes clear that they fall consistently on the first mora of each two-mora unit.” When it comes to the single mora that counts more than one, the reader “accents” it and “then inserts a pause (or lengthens the vowel) to ensure that an appropriate stress falls on the first mora of the next bimoraic group.” (pg. 204) This idea of that Japanese prosody has stress, and has a stress pattern that is accented and unaccented is something that we can readily understand. It is what feet are in English.


Kawamoto (p.g. 283-284) used two of Basho’s haiku to show how this works in them:

fu-ru | i- ke | ya — | — — |

ka-wa | zu — | to- bi- | ko- mu ||  mi- zu | no — | o- to| — — ||


It is pretty easy to pick up the beats in this if you accent the beginning of each mora. The vowel in “ya” in the first line gets lengthened to complete the bimoraic rhythm and the two pauses that follow it give the reader the breath space to accent on the “ka” that follows it. The vowel of “zu” and the vowel for “no” are lengthened to complete the rhythm.


Kawamoto also gives this Basho haiku as an example of when the first 12 mora are connected:

sa-mi- | da-re |o — || a-tsu- | me-te | ha- ya- | shi — ||

Mo-ga- | mi — | ga-wa ||


And then says that the meter above had been extinct for centuries and it was probably never intended to be read this way and gave this as a better example:

sa-mi- | da-re |o — | —  — ||

a-tsu- | me-te | ha- ya- | shi — || Mo-ga- | mi — | ga-wa | — — ||


Both examples have the single mora with the vowel elongated and the the second version has the five mora lines elongated to fit in with the meter and the two five mora sections have the silent bimoraic feet at the end to fill out the measure.


Kawamoto argues that “the individual seven and five mora verses of Japanese prosody each form a discrete metrical unit compared of four bimoriac feet, or in musical parlance, a four-beat quadruple-time bar” (pg. 223). This is why he adds the  silent bimoras at the end of each 5 section to fill out the measure. I’m not sure what the reason for this is, but we, and maybe it is because our own poetry has a lot of line breaks, don’t have the need to fill out the form so roundly.


Which means, then when you look at this model which is the accepted view on what the metrics of Japanese poetry are, then it is easy to see that what is discussed here is that the lines of a haiku will always scan out as being 6-8-6 feet. So, then 20 feet is what haiku counts out to in the way we scan our own poetry (discounting the two bimoras which are used to fill out the two 5 lines.)


It’s unfortunate that Imaoka never went into how 5-7-5 in Japanese related to English metrics, rather instead she inferred that if English haiku was written in something that approximated the length of the “mnemonic quality of 5-7-5 Japanese phrases” it meant that we were “concerning ourselves too much with the outward form of haiku” and thus “can lose sight of its essence.” Its a little hard to understand how, given how our language and our poetry relies on sound as a means to create added meaning, giving up on the mnemonic qualities would make English language haiku the poetic equal of Japanese haiku.


One gets the sense the Imaoka was the one who was more interested in form over language. If we look at this example she gave us we find this to be true:


across the arroyo

deep scars

of a joy ride


could be rewritten to approximate the 3-5-3 form as


across the

arroyo, deep scars

of a joy ride


without affecting the meaning. As it is, doing so sacrifices too much in the natural flow of words and interferes with the image.


She argues that the second version has an unnatural flow to it  and that this blurs the image, when in fact the opposite is true. The first line in the first version is unnatural, and because it isn’t in a natural speech pattern we run out of breath at the end of the line and our mind just flattens out. This unnatural breath pauses damages the images in the haiku because we spend time and energy just to find the proper linguistic syntax to fit it into, and by doing so are made unable to focus on the imagery. However, by rearranging the lines and inserting a comma, the second version is now into a natural speech rhythm and we are able to see the images a lot brighter and clearer because of it.


Imaoka’s statement that the “the type of unnatural line breaks seen in the latter is a problem associated with the 3-5-3 (or other short) form, whereas the 5-7-5 form is long enough to accommodate natural line breaks dictated by the English grammar,  due to a greater degree of freedom provided by the extra syllables,” is only an argument about form because it missed the point about how the English language works and how unnatural the whole first line of “across the arroyo” is in the first place. Besides adding the proper punctuation in, if we use proper grammar and and rearranging the imagery to “there are deep scars across…..” etc. the writer is able to use language to reinforce and add unstated emotional content into this haiku.


Which returns us to my point about poetry being more than just imagery. Poetry is a mix of sound and images to produce the emotional energy that the poet feels towards what they are writing about. The reason why English language haiku is such a dead dry language is because people have conned themselves into believing that all you have to do is to create poetry is to write out imagery and this was OK because that is what the Japanese did in their haiku in for centuries. Of course, these ideas about Japanese haiku were completely wrong and are now cached in the dust bin of history. This imagery only theory about haiku lingers on, but people are only fooling themselves because in the end all haiku written in English will be judged on the what the standards of English language poetry are and those standards will always insist that both sight and sound are essential for good writing.


I am not advocating a strict mandate that states that all English language haiku must be 20 feet in length, but I do advocate that haiku has enough space so one can write in a natural voice within a haiku. I find that Imaoka’s ideas about the length of haiku to be based on uncritical thinking and are ideas which deaden of the use of language in haiku. Instead of continuing to come up with schemes that try to emulate Japanese haiku, it is high well past the time that writers in English start to write within the poetic traditions of their own language.


I know that this sticks in the craw if a lot of people in haiku because they have been deceiving themselves for such a  long about how they are “so different” from “mainstream poetry.” The truth is that haiku writers aren’t different and that they have only made themselves worse poets by thinking that they are. It’s time to realize this.

The Problems With Keiko’s Rules


One of the problems with trying to understand Keiko Imaoka’s article from the perspective of someone who lives in Japan and can speak and read Japanese is that she never clearly differentiates between what a Japanese person counts when they write a haiku and how English speaking people count syllables.


Anyone with even a rudimentary understanding of the Japanese language understands that syllables are not what is being counted when a haiku is written by a Japanese person. All they are counting is the number of letters from the Japanese alphabet to fit into a 5-7-5 pattern.


To explain it more, the word for river in Japanese is made from two letters of the alphabet “ka” and “wa” to form the one syllable word “kawa.” When it is used in a haiku the word counts as two because that is the number of alphabet letters it is,  even though it is a one syllable word. Trying to say that it is two syllables is like saying that a one syllable word like “sky” is two syllables (sk/y) because it has two phonetic sounds.


Yet, it is not as if she didn’t know the differences between the two for if we scan both the original Japanese haiku and her literal translations of them, counting out the syllables in both the way we do in English, we find that her translations exactly match English syllables with the originals. Why she didn’t bother to first explain what the differences are is something that one can only guess at.


Here are the translations (with the English syllable count of both original and translation at the end of the translations):

yuku haru-ya tori naki uo-no me-ni namida    – Basho

spring passing – birds cry, tears in the eyes of fish   (11 syllables)


neko-no meshi shoubansuru-ya suzume-no-ko    – Issa

sampling the cat’s food –  a baby sparrow       (9 syllables)


ware-to kite asobe-ya oya-no nai suzume  – Issa

come play with me – you motherless sparrow.  (11 syllables)


uguisu-no naku-ya chiisaki kuchi akete    – Buson

uguisu singing – (uguisu : a nightingale-like bird) with the small mouth open     (9 syllables)


Now that we understand the syllable count of the originals, we must consider Imaoka’s statement how “as to the form, some American poets advocate writing in 3-5-3 syllables or 2-3-2 accented beats.”


Since all the examples above in the original language all are over 7 syllables in length, then we immediately have to dismiss this idea of 2-3-2 form in English as being too brief for haiku, for how can something that is shorter in form than the original convey about the same amount of information as well as the brevity and the fragmented quality found in Japanese haiku”? Even Imaoka herself didn’t attempt this in her translations.


It’s a bit absurd too think that by using less syllables than the Japanese that you could be as brief as they are, rather you would end up being briefer than they are, so we can summarily dismiss the 7 syllable form as simply missing the point of it all.


Being a native English speaker, it is easy to see where Imaoka cropped the language to make the translations fit to the count of the originals. To get a better idea of what a truer syllabic ratio between the two we first must rewrite the translations into normal English speech patterns:

spring is passing – the birds cry and tears in the eyes of fish  (14 syllables)


The second haiku has a translation problem. The “shoubansuru” is a verb that means to “partake at the expense of someone else.” I will use the colloquial phrase for it.

mooching off of the cat’s food –  a baby sparrow    (12 syllables)


Another translation problem with this one as well. “Oya” means “parents” not “mother” and the phrasing has to change because of it.

come and play with me –  you sparrow without any parents.   (13 syllables)


The only reason why she used the Japanese word “uguisu” instead of the English “bush warbler” must be because it would throw off the syllable count.

A bush warbler is singing – its small mouth opens   (12 syllables)


Now that we have the translations into natural English speech patterns we find that all of them have more syllables than then even the 3-5-3 form that Imaoka noted above.


First, this reconfirms about the 2-3-2 form being too short in English, for if you need more syllables just to match the information in the original Japanese, as putting these translations into more natural English show, then surely cutting down the amount of syllables to even less than what the originals have severely hinders the ability to express as much as the Japanese could.


Secondly, although only 4 haiku is hardly a true sample of what the parallels between Japanese and English haiku might be, in the context of the article that Imaoka wrote this is what was presented as being the basis of premise of how “the fact that English carries significantly more information per syllable than Japanese indicate that using the 5-7-5 form does not necessarily provide an analogous condition for writing haiku in English.” From what we can gather from putting her translations into a more native form is that the English language needs more syllables to get  the same amount of information that the Japanese has and that rather than carrying more information per syllable the English language actually carries less.


And if English needs more syllables, then how can cutting down to  3-5-3 syllable count provide an analogy to Japanese haiku?


Another issue with a set syllable count in English is that every haiku must be exactly the same count, which isn’t true in the four Japanese examples above, rather they alternate between two different syllable counts.


With all these problems we find in trying to align the length of English language haiku to what it is in Japanese, we come back to what Keiko Imaoka never mentioned, that the way the two different languages counts syllables out is diametrically opposite. Given this, and the fact that the Japanese use of seasonal words (“kigo”) add unwritten tropes which the writer in English doesn’t have access to, let alone not discussing the way using Chinese characters reads differently than the way the phonetic alphabet of English does, it is better to leave off trying to do the impossible task of trying to align the length of one with the other.


Sound has long been a problem in haiku written in English, and it is the bane of the shortened 2-3-2 and 3-5-3 form. With so little space, the writer doesn’t have the space to build up sound to write poetry. Some might argue that all you need is imagery to create poetry, but they are just kidding themselves because poetry is both sound and imagery. You can’t sacrifice one in favor of the other, which unfortunately the haiku writer has to do when confronted with these two forms. Imagery without sound only leads into intellectualism and pedantry, not emotion and real expression. It is an affected style that ignores the innate power of its own language by uncritically trying to copy the Japanese language. And affected language never creates effective poetry.