return to research page    

Stalking the Wild Onji: The Search for Current Linguistic Terms Used in Japanese Poetry Circles


Publication: Language Issues: Journal of the Foreign Language Education Center 5.1,
Prefectural University of Kumamoto (1999), Kumamoto, Japan;
also Frogpond: Journal of the Haiku Society of America, XXII: Supplement (1999).


Many challenges confront poets and educators in the burgeoning international haiku and tanka poetry movements who, researching Japanese forms of poetic composition in English translation, wish to better understand these genres and skillfully emulate them. Differences between the two languages and inadequate presentations of these differences have created confusions, misusage of terms, and in some quarters a reductionistic sensibility regarding formal aspects of Japanese poetry. Onji, the Japanese term commonly used in the West to count up and define Japanese "syllables," has had a contentious history in North America, having served as one of several loci of controversy regarding how the haiku, particularly, is best emulated in English. This paper investigates the historic usage of onji as a linguistic term in Japan and presents an argument for its removal from usage, as currently construed, by the international haiku community. Linguistic terms that are in widespread use within contemporary Japanese haiku circles are described, defined, and suggested as replacements. It is hoped that such reparations may effectively halt a rapidly escalating situation of cross-cultural miscommunication, occurring in both directions. A brief overview of the evolution of the Japanese writing system and issues relating to the modern Language Reform Movement provides an historical context for terms and concepts discussed.


Synchronic Forays

Finding the Forest

Over the past year, while teaching English Communication to a third-year class of Japanese Literature students at the Prefectural University of Kumamoto, I persuaded the students to try their hand at haiku composition in English. In the course of exploring differences between haiku written in English and Japanese, inevitably the question of which syllables to count has come up, and indeed what to call them; whether in Japanese they are called jion, or as I have always called them, onji. (viz. "In fact, Japanese poets do not count "syllables" at all. Rather, they count onji." Higginson, 1985, p. 100)

My anecdotal research into the usage of Japanese poetic terms began when, in several of my classes, I experimentally wrote lines of hiragana on the board and asked the students to count the onji. In every case, the students had no response at all to my request, there were only blank stares. I was especially surprised that Japanese Literature students had the same reaction as the general education students; I began to wonder if the term itself was problematic. Later, I asked several professors who attended a Kumamoto Gakuen University "Haiku Club" to explain onji. None of the professors I queried recognized the term. I was truly surprised. How could contemporary haiku writers, engaging with the haiku tradition, as well as Japanese Literature students, be unfamiliar with this word, considered in North America to be the one and only term used by Japanese poets to count up "syllables"?[1]

Quizzing Kojyo-sensei of Kumamoto Gakuen University, a scholar of Old English stylistics and haiku poet, I asked, "What is this word onji and how is it used?" Kojyo-sensei didn't recognize the term either. I found the situation rather odd, as Kojyo-sensei had spent many years as a dedicated haiku writer and aficionado. What words do Japanese people commonly use to describe and count haiku kana or "syllables"? What is the meaning of onji? Why isn't this word known in Japanese poetic circles? Is the English use of the term onji related to another word altogether in Japanese? And, how does the English use of the word "syllable" relate to the Japanese language? These are some of the questions I set out to answer in my search for the apparently elusive onji.  

Conduct of the Research

Research has been conducted on several fronts. Primary textual research was conducted with the help of Kojyo-sensei; additional internet-based research and Meiji-era translation was conducted at the Prefectural University of Kumamoto, with Matsuno-sensei, Professor of Information Science; a third research group composed of Japanese Linguistics and Phonology post-graduate research students, directed by Kai Tomoko, also at the Prefectural University of Kumamoto, aided in confirming current academic linguistic usage of terms. In addition to anecdotal evidence, the following resources have served as primary sources:

The results, aided by later confirmations, provide evidence that onji is no longer an appropriately communicative term. Additionally, it is my belief that the two 'counters,' -on and -ji, used for counting Japanese kana or the Japanese "syllables" in haiku and tanka, have been artificially fused or confused with the term onji (or sometimes jion) as used in English.

Recently, the poet, publisher and translator, Jane Reichhold, (see Kawamura & Reichhold, 1998; 1999), wrote to me about the controversies that had first occurred in the 1970s surrounding the English use of the term onji, and methods for counting -on in haiku and tanka. Here is an excerpt from our dialogue:

Did you know that haiku wars were waged in the 70s over this issue of onji and "syllable" counting? Friendships were permanently destroyed. Haiku groups split up. New ones formed. Persons were reviled. There was much sneering, jeering, and rejection. It was terrible. The problem remains and is just now entering the tanka scene. From Japan, one group is pushing that all our tanka be written in 5-7-5-7-7 but 5-7-5-7-7 what? How can we count our syllables and equate them with this unknown factor which the Japanese count and hold in such high esteem? (J. Reichhold, July 11, 1998. Personal communication.) 

It is ironic that there were such bitter arguments over a Japanese word-- and the "syllable" counting battles it typified-- which had exited the Japanese linguistic vocabulary years before the haiku wars

Counting in Japanese, and Some Differences Between English and Japanese "Syllables"

Japanese counting systems use 'counters,' which are special counting terms for things. There are many different counters, or counting terms, for all sorts of things in Japanese, including phonetic characters, alphabetical symbols, spoken sounds, and sound-units in poetry.[2]

Onji is an obsolete linguistic term used to define "phonic characters," that is, characters (ji) which have sound (on), but not meaning. In modern times, this word has been supplanted by the term hyouon moji (similarly): characters (moji) which are representative (hyou) of sound (on), or simply "sound representative characters." Onji and hyouon moji are terms of categorical definition; neither term has ever been used to count up "syllables" in Japanese poetry.

Japanese generally uses two different counting terms for counting "syllables." One counter is -on. The other is -ji. They are two separate words. On means "sound," and Ji means "character." One can count up the 17 "syllables" in the typical haiku using either term:

1. go-ji shichi-ji go-ji  
literally, "five characters, seven characters, five characters." Similarly we can count:  
2. go-on shichi-on go-on  
which has a meaning similar to #1 above: "five sounds, seven sounds, five sounds."

If we refer to the number of "syllables" in a poem, we can ask, "How many -on?" or "How many -ji?" These questions ought to be quite clearly understood by modern Japanese people, when referring to poetry. For counting the total number of "syllables" in a poem, two terms are generally used. The first, and more strictly correct is -on, as in: "There are 17-on in that haiku." Also, people informally use moji. Moji is not generally used as a counter, but its meaning is virtually identical to -ji: a letter or character, that is, any written character. So people also say, "That haiku has 17-moji." Ji is not commonly used to count totals of "syllables" in poems. So, the most communicative terms used in contemporary Japan to count "syllables" in poetry are -on or -ji or sometimes moji, for totals.

'Syllables' has been enclosed within quotation marks, because with only few exceptions due to dialect, there is virtually no perception of English-style syllabification of words on the part of adult Japanese speakers. Natsuko Tsujimura mentions that "Specifically, English speakers divide words into syllables while Japanese speakers divide words into morae. Due to this difference, a native speaker of English divides 'London' into two syllables, while a native speaker of Japanese considers the word as consisting of four morae. [lo/n/do/n] . . . Mora is considered as a timing unit, especially within the larger context of words" (Tsujima 1996, pp. 64-6).

Basically, this means that, perceptually-speaking, Japanese speech is composed of small, timed units of sound, rather than syllables. Mora (plural, morae) is the term that both Japanese and English linguists often use to identify the 'time-unit sounds' of speech, which when put together, compose words in spoken Japanese.[3] With regard to Japanese poetry, the terms -on and -ji identify these same time-unit sounds. It is this time-sense division of sounds, rather than syllabification, which accounts for how words are parsed by Japanese speakers.

One term that can be used for the English-style syllabification of Japanese is on-setsu. It is both a name and a counter. On-setsu has several conflicting definitions, and there is some controversy right now in Japan about its appropriate use. Here, following one of the two main definitions advanced by Hattori (1961; cf. Campbell et al, 1993, p. 670), on-setsu will be used to indicate English-style syllabification of Japanese, and additionally, the perception of on-setsu in non-moraic languages.

If one gives Japanese students the task of separating English words into syllables, and then asks them to describe what they are doing, they will say: "These are English on-setsu." On-setsu, then, is indicative of the closest available Japanese concept to the English-speaker's perception of "syllable." However, there are some compelling differences between syllables and on-setsu, as applied to Japanese. For a start, there are no on-setsu in Japanese which are longer than two combined -on. Japanese on-setsu are always either one or two mora in length. Therefore, we can say that spoken Japanese is a language composed of either long or short on-setsu, the long on-setsu being more or less exactly twice as long as the short on-setsu. When English-speakers hear the word nihon, they will perceive the word syllabically as "ni/hon." Are these two English syllables reasonably similar to what is meant by the two on-setsu: "ni/hon" in Japanese? Clearly not, because on-setsu remain rooted in Japanese language perception. They carry a precise time-sense that is fundamental to the language. English syllables are not only vastly more variable in length, but further, are paradigmatically disjunctive to moraic timing in Japanese.[4] Thus, the term "syllable" is conceptually counter-intuitive to the way in which native-speakers of Japanese perceive and cognize their language.

Though the use of the term "syllable" may seem expedient and practical in its application to Japanese poetry from an English-language standpoint, our main concern here is to promote clear bilingual, cross-cultural communication. When using the word "syllable" in referring to the individual sound-units which we may naively perceive and count up in haiku or tanka, we are not being correct, certainly, in terms of Japanese usage or sensibility. Further, we run the risk of distancing ourselves from cogent factors that are innately a part of those original poetic works which are serving as models, or as a basis for emulation, translation, or study in English. It seems more elegant, as well as accurate, to use the terms -on or -ji, and avoid the use of the term "syllable," if possible, when counting up the separate sounds (hyouon) in Japanese poetry.

From this point, terms typically used in Japan will be used to distinguish between English-style syllabification and individual sounds, as follows: on-setsu to indicate English-style syllabification, and -on to indicate the separable sounds (hyouon), or time-units of speech (mora), which are what we want to count up in Japanese poetry.

There are two Japanese phonetic alphabets in contemporary use, i.e. the kana alphabets. Kana can refer to an individual alphabetical character, or group of characters. The kana alphabets have often been described as syllabic alphabets, but we can consider each kana character as one -on. Each kana character is, then, representative of a separable, timed, sound-element of language.[5]

Here is an example of the difference between on-setsu and -on counting: nihon (ni/hon) has two on-setsu. Notice that a single sound (-on) can sometimes function as an on-setsu, as in the case of ni in ni/hon. With "nihon," we count the three -on as: ni/ho/n. Some further examples which illustrate problems that can arise when counting -on in romaji, the Roman letter alphabet, are detailed below:

Enpitsu (pencil) is made up of four kana: e/n/pi/tsu, so there is a total of 4-on. If we are reading romaji, we might mistakenly count enpitsu as en/pi/tsu. Parsing enpitsu this way separates the word into three on-setsu. To get an accurate count of hyouon moji in Japanese poetry when we are using romaji, we need to count the sounds just as they would be represented by the kana alphabets. Here are a few other examples: eigo (English) has three -on: e/i/go. But two on-setsu: ei/go. Nihongo (Japanese language) has four -on: ni/ho/n/go. And three on-setsu: ni/hon/go. Koukousei (High School student) has 6-on: ko/u/ko/u/se/i. And three on-setsu: kou/kou/sei. A word like this may be written in romaji as kokosei in some variants, with the -u kana left out, making it more difficult to accurately count the total -on. Usually, an omicron over the 'o' will indicate the presence of the additional -u kana. Kekkon (marriage) has a doubled consonant, which indicates the presence of the small "tsu (t)" kana, a "stop" or pause in speech of one mora. Kekkon has 4-on: ke/(t)/ko/n. And two on-setsu: ke(t)/kon." Remember, each -on or -ji can always be represented by one (or one digraphic) kana character in Japanese.  

Morae and Prosody

We shall return briefly to the term mora, and its plural, morae. These are English linguistic terms which we have also found used as Japanese loan-words. In typical Japanese spoken style, each -on takes approximately the same amount of time to speak. In fact, each kana, including digraphs like kyo, jyo, gyo, etc., takes about the same amount of time to speak. This time-sense or time-count is defined by the term mora, or we could say by kana or hyouon time-units.

Mora is a technical term, not generally used outside the field of linguistics. The more commonly known term is haku. Currently, mora(e) are undergoing intensive linguistic studies, which show connections between spoken and written Japanese that reveal underlying relationships not altogether unlike English prosody. Recent research shows that the perception of moraic length and timing on the part of native speakers is highly complex, being influenced by the accent, pitch shift, duration, and volume-level of words.[6] Consequently, it has become somewhat reductive, linguistically, to consider morae purely as measures of abstract time-units, a view which was widely held some 20 years ago. This is an exciting area of research and may yield new methods for emulating haiku and tanka in English. Mora research is mentioned, then, to call attention to layers of prosodic complexity in Japanese language and poetry that go far beyond -on counting alone, which if taken as a singular, defining formal feature of haiku, leads to painfully reductive structural interpretations. In terms of written Japanese, the number of mora will always agree with the number of kana (with digraphs considered as single kana), and therefore the number of -on.

A Glossary of all the terms covered in this paper will be found in the Appendix. 

Diachronic Expeditions

A Brief History of Kanji and Kana

Here is a brief history and overview of the Japanese language from an orthographic viewpoint. "It is generally believed that kanji came to Japan from China through Korea, [between 300--400 CE]. No record of a written language exists in Japan before this time" (Mitamura & Mitamura, 1997, p. xi). Kanji themselves are much more ancient, "attributable to the scribes of the Yin Dynasty [17001050 BCE]" (p. xi). The oldest kanji descend from hieroglyphs or pictographs ("shoukei moji," op cit, p. xiv). Japanese kanji, as presently used (with some exceptions) are ideogrammatic, not phonetic. Each character represents a singular concept or idea rather than a singular sound. The group of characters from which kanji spring are called: hyoui moji, meaning, "characters (moji) which are representative (hyou) of meaning (i), not sound." An older term for hyoui moji is iji: meaning (i) + characters (ji).

In China, the meaning and sound of the kanji were originally directly related, but when kanji were imported into Japan, some interesting changes occurred. First, the original Chinese sounds which came along with the kanji were changed to accord with Japanese, which does not use pitch to ascribe meaning, as does Chinese. (This is called the ON reading of the kanji.) Then, pre-existing Japanese words which had the same meaning as the Chinese kanji were added to each kanji. (This is called the KUN reading of the kanji.) Over the centuries, the same kanji was reintroduced to Japan, sometimes repeatedly, from various regions of China, and from succeeding Dynasties. With each reintroduction to Japan, the same kanji took on yet another form of pronunciation.[7] Kanji were also created and further adapted in Japan, and rarely, a kanji from Japan went back to China.[8]

The typical kanji now has two or three ON readings and two or three KUN readings, while some of the commoner kanji, such as "life" and "below" can have as many as ten fundamentally different readings. .  . As a result of fundamental differences between the monosyllabic Chinese language and the polysyllabic, highly inflected Japanese language the Chinese writing system proved decidedly unsuitable in the case of inflected items such as verbs. .. . the potential for confusion was obviously considerable . . . (Henshall, 1988, p xiv).

"The number of kanji in actual use probably did not exceed 5,000 or 6,000 [before 1946]" (Campbell et al, p. 669). Eventually, novel 'problem solving' phonic characters (absent of intrinsic meaning) were evolved in stages to distinguish between alternate readings of the same kanji and determine verb inflections, among other uses.

Today, there are two phonic alphabets, hiragana and katakana, in general use. Japanese is written with a mixture of kanji and kana, mostly hiragana. There may be a sprinkling of katakana and the merest dash of roman letters, usually for foreign names or places, and technical words. Current linguists now consider contemporary Japanese to have three separate alphabets, plus the 1,945 joyo kanji approved for general use-- unless you are a haiku writer, in which case you probably know many more kanji than the average person on the street. (Joyo kanji instruction is not completed until the end of the 9th grade.) When we read a Japanese poem in romaji, we are looking at a special form of hyouon moji, called tan-on moji: a "single-phoneme character" alphabet. However, outside of a few historic social and literary experiments, romaji is not ordinarily used by the Japanese themselves in written discourse. Romaji is mostly used for the benefit of those who cannot read the kana alphabets.

The term hyouon moji is very useful in Japanese linguistics, because historically, various phonic characters were employed before the later development of the kana alphabets currently in use. The Man'yoshyu, 759 CE, the earliest collection of poetry, is written in Man'yogana, a script in which certain kanji were designated as 'sound-only' kanji (see Campbell et al, pp. 730-31). Hyouon moji can be used to describe all of these phonic characters, and hyouon can be used more generally to describe phonic representations, without the need to discuss characters, specifically.

Hiragana in its modern form is composed of 48 kana characters. "Hira means 'commonly used,' 'easy,' rounded' (Campbell et al, p. 731). Hiragana was developed from simplified kanji, and takes its name

because the [kana] are considered rounded and easy to write [when compared with the original kanji]. In its early [9th century Heian era] forms, hiragana was used by women [who were not permitted to learn the Chinese script], while the unsimplified kanji were used by men; for this reason, the earliest hiragana was also called onnade, "women's hand." By the end of the 9th century, onnade ceased to be a system limited to women and . . . [only] gained full acceptance when the imperial poetic anthology [Kokinshu, published in 905] was written in onnade (op cit). 

"The kata- in katakana means 'partial, 'not whole, 'fragmentary' (op cit). This name stems from the fact that many katakana were taken from only a part of the original kanji. In its earliest use, "katakana was a mnemonic device for pronouncing Buddhist texts written in Chinese" (op cit). By the mid-10th century, poetic anthologies had been composed in katakana

The Rise and Fall of Onji

The year of the Meiji Restoration, 1868, marked the beginning of the modern Japanese language reform movement. "Although much of the development of modern Japanese proceeded spontaneously, the role of planned development was considerable. . . . It was necessary to select a single variety of Japanese . . . to increase literacy, [and] create an extensive modern vocabulary" (Campbell et al, p. 669). As well, grammatical and stylistic usages began to be codified, "[and Japanese began] to be liberated from its dependence on classical Chinese. [It was felt by a number of eminent scholars that] the only way to modernize the language-- and the minds of the people that spoke it-- was in affiliation with the languages by means of which the knowledge of the developed West [could be] introduced" (op cit).

It seems likely that the entrance of onji into the Western lexicon was a result of the publication of Nishi Amane's (1829-1897) landmark text, the "Hyakugaku Renkan," in 1870 (see Campbell et al, p. 1098), soon after his four-year sojourn in Europe. Amane equated the pre-existing Japanese term onji orthographically with "letters." In this first Western-style encyclopedia, "patterned after the works of Auguste Comte . . . Amane introduced the full spectrum of Western arts and sciences to Japan" (op cit). The encyclopedia contains hundreds of Western terms, which are correlated with Japanese terms or concepts (and vice versa). Amane's translation is preserved in the "Nihon Kokugo Daijiten," which annotates the "Hyakugaku Renkan" as the root-translation of onji to the English "letters" (see Nihon Kokugo Daijiten, 1988, Vol. 4, p. 159). The "Hyakugaku Renkan" was an important source of English-Japanese and later, Japanese-English translations (based upon Amane's correlations) throughout the Meiji era (1868-1912), during which time the Japanese vocabulary developed with phenomenal rapidity.

Meiji-era grammarians typically used onji to describe phonic characters, while its sister-term iji was used to describe ideogrammatic kanji characters. Prior to 1900, language reform groups were urging the government to take steps to modernize the Japanese language, but no major language reforms occurred. The changes that did result were serendipitous. Two varieties of Japanese emerged: the classical standard, based on pre-Meiji styles, used only in writing; and the colloquial standard, rooted in the spoken language and more or less identical to modern Japanese. Literacy was also rapidly improving through the implementation of mandatory education.

By the early 1900s Ueda Kazutoshi, also known as Ueda Mannen (1867-1937), a professor at Tokyo University (influenced by his studies with Basil Hall Chamberlain and "the first Japanese trained in Western linguistics..." Campbell et al, p. 670), had become a member of the National Language Research Committee. Ueda introduced Western linguistic research methods into Japan, trained researchers, and contributed greatly to national language reform policies. In a relatively early monograph "Kari-ji Meishyou-kou" [The Origination of Kana], (1904 or prior), he used the term onji exclusively. One of Ueda Mannen's crowning achievements was the "Dai Nihon Kokugo Jiten" [Japanese Language Dictionary], (1972, reprint), originally published in four volumes, 1914-1919. "Containing over 200,000 entries, it became the standard work for editors of later dictionaries . . .  it is distinguished by its cautious treatment of etymologies and its policy of including only information of unquestioned accuracy" (Campbell et al, p. 266). Within, onji is cited, and a brief definition is included (p. 254). There is no reference to hyouon or hyouon moji. In fact, examining citations for hyou, we find no entry (see p. 1666). Evidently, neither hyouon nor hyouon moji had entered linguistic parlance prior to publication.

Hashimoto Shinkichi (1882-1945), a noted Japanese linguist and grammarian, graduated from Tokyo University in 1906 and served as assistant to Ueda Mannen from 1909 to 1927, when he succeeded Ueda as professor of Japanese at Tokyo University. We find in his "Kokugakugairon" [An Outline of Japanese Linguistics], (1967, reprint), originally published in 1932-33, a chapter titled "moji no shurui" [Types of Characters]. On page 104 is the statement: "Concerning onji, it is synonymous with hyouon moji and onhyou moji" (my translation). In the remainder of the chapter, onji is the term Hashimoto prefers, and uses it on two occasions. This situation is reversed some ten years later in a paper written in 1943, titled "nihon no moji ni tsuite" [About Japanese Characters], (see "moji oyoubi kana ken no kenkyu" [Research into Kana and the Usage of Characters], 1976, reprint, pp. 226-36). The paper's subtitle "moji no hyoui-se to hyouon-se," means roughly, "The Ideogrammatic and Phonic Nature of Characters." On page 226 we find: "Concerning hyouon moji, it is synonymous with onji." Hashimoto uses hyouon moji throughout the paper (some eight times), and does not use onji. So, though it is unclear when exactly the term hyouon moji entered the linguistic lexicon, it seems apparent that it was becoming more popularly used by grammarians by the 1930s, and likely was becoming or had become the preferred term just prior to the post-war period.

At this time, the field of linguistics was developing quite rapidly, and a great number of language reforms were being implemented. Significantly, Hashimoto, in concert with a group of linguistic scholars, was responsible for the "Supplement to the New Grammar" (shin bunten bekki), the official school reference grammars established by the Ministry of Education, as late as 1939. Hashimoto's grammatical ideas were widely distributed and remain influential to the present.

Further references to onji in Hashimoto's papers on Japanese phonology written in the 1930s and 1940s have not been located. It is likely that onji was removed from educational grammars during or following the 1930s reforms, being supplanted with hyouon moji. As to the exact date of the disappearance of onji, and a significant referential notation regarding its disappearance, none has been found to this date. Okajima Teruhiro, Professor of Japanese Language and Linguistics at Fukui University, who has done extensive research on the Language Reform Movement and the works of Hashimoto Shinkichi, searched for particular references and reported that there were indeed very few references to onji whatsoever. Okajima suggests: "As the field of Japanese linguistics developed, there arose confusions in the usage of onji with another term, on-se kigou, a term used to denote the special characters used for phonetic representation. It is likely that the more accurate term, hyouon moji, was chosen to replace onji in order to avoid confusions caused by variant usages of onji" (Personal communication, July 30, 1998, my translation).

Onji is no longer included as a reference in many current Japanese dictionaries. We do not find a reference in the "Nihongo Hyakuka Daijiten" (Kindati, Hayashi and Sibata, 1988), an encyclopedic linguistics dictionary, wherein hyouon moji receives a lengthy treatment (p. 307). A top-line electronic translation dictionary, the 1998 Canon IDX-9500, containing nearly 400,000 Japanese word-entries and over 250,000 English word-entries, finds hyouon moji, only, when a search is performed using "hyoun" as a keyword. Keying in "phonogram" in English, the dictionary translates the term as hyouon moji, only. Searching with the keyword "onji" results in zero hits. Onji is not contained in the database. The above facts must give one pause. We do find references in some academic dictionaries, for instance, the 20 volume "Nihon Kokugo Daijiten," (1988), as previously mentioned. In the "Koujien," (Shimura, Ed., 1998), perhaps the best and largest of the single-volume unabridged dictionaries in Japan, we find a citation for onji (p. 419), which refers us to hyouon moji. It is only under the citation for hyouon moji (p. 2275), that we find a definition given. We also find a single reference in the largest of Kenkyusha's Japanese-English translation dictionaries, (1978), where onji is translated by the word "phonogram," with a reference to "see: hyouon." Further investigations into the disappearance of onji will no doubt yield more definitive results. Questions concerning regional variations of usage and specialized technical usages remain.

Clearly, onji is not recognized in contemporary Japanese poetic circles. In contrast to onji, the term hyouon moji is fairly well-recognized as a defining term, in both linguistic, and, as we have found, literary environments. Onji was first brought to the United States haiku audience via a letter published in the Haiku Society of America's "frogpond" (1978), written by Tadashi Kondo, in response to the Haiku Society of America, under the direction of Harold G. Henderson, having previously adopted the term jion. Kondo wrote that "Jion is a specialized term from linguistics relating to the pronunciation of a Sino-Japanese character. Onji means 'phonetic symbol' (or 'sound symbol,' and seems to be the term desired. . . . [however] while the concept of onji has often been translated into English as 'syllable,' it would be more accurate to say that the onji is a 'mora.' . . " (op cit). Following the "frogpond" publication, onji became an important part of the English language haiku-study vocabulary. Hopefully, current research will suffice to change this usage, which still remains active throughout the West. Kondo recently commented that

I knew Professor Henderson through letters exchanged three or four times before his death in 1974. I did not meet him in person, since his death came the day after I arrived in this country. I know he was a fine linguist, and would never make such a trifling mistake. My personal speculation is that the mistake, of having jion instead of onji, could have simply been a typo. It could have been done either by Professor Henderson himself or by the printer. This type of mistake could have happened easily; as most people do not pay much attention to these words, the mistake might have passed by many peoples' eyes. [Nevertheless] in the early 70s . . . Haiku poets who would come to the HSA monthly meetings at the Japan Society were using jion, as in: "haiku is written in 5-7-5 jion," which is absolutely wrong. . . . When I found this mistake, my simple reaction to it was to flip the word order, from jion to onji, because that seemed the most reasonable correction . . . ([When you] look up onji in older dictionaries, you may find explanations given, instead of a simple direction to see hyouon moji or onpyou moji.) . . . I found the problem and decided to write a letter to correct the misunderstanding. So the issue is basically technical and I was not concerned about its popular usage. It is another issue. (T. Kondo, April 3, 1999. Personal communication.)

The well-known haiku poet James Kirkup, a frequent resident of Japan, has commented that "few [Japanese] have ever heard the word onji" (frogpond, 1995); we assume he is referring in particular to the Japanese poetic community. Due to the obscurity of onji, some rather ironic cross-cultural miscommunications have been occurring with greater frequency at one of the most active areas of international exchange-- internet haiku and tanka websites. In a typical scenario, an aspiring non-Japanese haiku poet wishes to discuss how to count -on in haiku (a popular research topic), and uses the term onji in an e-mail to a newsgroup. Japanese respondents, who are rarely fluent in English, do not recognize the term, and assume it is yet another mysterious English word. A potentially edifying cross-cultural dialogue is thus aborted; the topic under discussion, -on counting, is never effectively broached. Terms such as -on, -ji, or -moji, while they might not always be applied with technical precision, would remain communicative. One can also imagine that the use of communicative terms, combined with a more refined understanding of Japanese language issues on the part of haiku poets who are otherwise unfamiliar with Japanese, would create an atmosphere more conducive to a multi-cultural exchange of poetic ideas. Communication problems, as in the example above, only serve to maintain historical patterns of isolation and insularity between the Japanese, North American, and increasingly, international haiku and tanka cultures which use English as a medium of exchange.[9] 

Eight Reasons Why Onji May Have Persisted in the English Poetic Lexicon

It is now possible to summarize some probable reasons for the persistance of onji:

1. A number of influential translators and Japanese authors used the term, following the Meiji restoration (1868), through the 1930s. Later translators, seeking translation sources, may have continued to follow English translations by previous translators.

2. Onji continues to be indexed in the best available Japanese-English translation dictionary as well as in some of the academic and collegiate Japanese dictionaries. This is a primary source of information for those living outside Japan. Although in all the cases we found, hyouon is referred to as an operative term, onji is implied by description as archaic only in the "Nihon Kokugo Daijiten." Kenkyusha's translation dictionaries, good as they may be, are no substitute for an unabridged Japanese-language dictionary, which does provide enough information to discern an archaic attribution, if one researches the indicated references.

3. Under the influence of Western linguistic methods and the pressing need for language reform, the Japanese language has undergone rapid change in the 20th century, especially in the use of grammatical terms. The Japanese pay a great deal of attention to their language: Japanese language history, grammar, and phonology are taught in public schools, and knowledge of such terminology is often required for college entrance exams. Any changes made by Ministry of Education linguistic research groups tend to be rapidly implemented in future textbook changes. Given this atmosphere of change, it is possible for a term to quickly exit the Japanese lexicon, say in a period of 20 or so years from the date of textbook removal. Obviously, World War II also had an enormous impact in creating a 'break' from some aspects of pre-war linguistic usage: outmoded aspects of pre-war language dropped from usage with extreme rapidity. This scenario implies that a Japanese word could enter the English language while shortly thereafter disappearing from the Japanese lexicon.

4. A 'culture-gap' can occur through purely written communication with Japanese correspondents. In correspondence with Japanese writers in which I was soliciting information about onji, informants never mentioned their personal experience. If they were questioned about an unfamiliar word, a dictionary was consulted. Not finding the word in abridged dictionaries, informants tended to seek out a dictionary such as the "Nihon Kokugo Daijiten," whose onji information was then quoted. However, the archaic nature of the term, which takes additional research to verify, was not mentioned. Generally, informants tended not to offer personal opinions or their lack of acquaintance with onji in personal communications, unless they were teachers of Japanese linguistics and phonology, had access to numerous sources, and perhaps therefore considered themselves informed enough to comment. Due to differing cultural styles of communication, English informants, on receiving communications from correspondents, may have been lead to believe that onji was a viable term of discourse.

5. There is little readily available research into contemporary Japanese terms utilized in Japanese linguistic phonology in English translation. Japanese Linguistics texts often apply English linguistic terms and categories to the Japanese language. Much of the linguistic commentary on Japanese poetry (in English) has taken a Western-oriented approach to the language, as is the case with the term "syllable." This has been true even within Japan, though the situation is changing. In addition, the disappearance of onji is not of particular significance within Japanese phonological studies.

6. The Japanese use of 'counting' terms has caused misunderstandings in English. The specific counting terms and concepts utilized in Japanese poetic circles have not been broadly introduced.

7. Few Western poets or translators with an interest in Japanese poetry have lived and worked in Japan, gained Japanese language ability, joined haiku circles, and inquired about linguistic terms. Short visits and overseas inquiries may not have elicited the necessary cultural information.

8. Research into onji twenty to thirty years ago might have yielded different results. If Japanese grammarians had been consulted, researchers may have found that some scholars had knowledge of the term, having been educated during a time when onji was still in usage. 

Beautiful Language and the Spoken Word

One day, after Kojyo-sensei and I had satisfactorily laid onji to rest, sensei dug up a gorgeous book of collected haiku, and read--mostly Basho and Buson, but some modern haiku also. And we talked about each haiku a bit; he made instant translations after reading the commentaries, and we counted kana and clapped out the mora. This experience was most satisfying. Having had so many haiku spoken out loud by a sensitive reader, I now have a better sense of the spoken haiku rhythm, and why classical writers may have sometimes added or subtracted an -on or two, here or there. As poets know, the feeling of the language and its living texture often takes precedence. Language is not, after all, some pure mathematical equation with exact answers; words play, stretch, bend into each other, eddy and separate. The supposedly "tight" form of haiku and tanka has served to mask the fact that whether a line has 4, 5, or 8-on, in Japanese, these lines can sound shorter or longer, staccato or legato, be phonemically complex or have extreme brevity, regardless of the "count." No doubt this information does not translate easily into English, especially because this has to do with sound and speech, not just reading--the standard medium for approaching Japanese poetry in English translation. It was wonderful to hear a sensitive native-speaker reciting haiku, to "get into it" in this way. I think this was why Kojyo-sensei, at a certain point, just put down the texts and began to read out loud. 


I would like to thank Jane Reichhold, publisher and co-editor of "LYNX: A Journal for Linking Poets," for her encouragement, informed discussion, and editorial support; Prof. Ryoji Matsuno, who devoted many hours to internet research and English translation; and referees Prof. Judy Yoneoka of Kumamoto Gakuen University and Prof. Morio Nishikawa of Kumamoto University for their careful reading of this article and valuable comments. Any errors that remain are my own. 


American Heritage Dictionary. (1992). New York: Houghton Mifflin.

Campbell et al. (Eds.). The Japan Encyclopedia. (1993). Tokyo: Kodansha.

Donegan, P. and Ishibashi, Y. (1998). Chiyo-ni: Woman Haiku Master. Tokyo: Tuttle.

Hashimoto, S. (1976). Nihon no Moji ni Tsuite [About Japanese Characters] In Moji Oyoubi Kana Ken no Kenkyu. Tokyo: Iwanami Shoten. (Republication of original work published 1943). [Research into Kana and the Usage of Characters]. Available on the World Wide Web at

Hashimoto, S. (1967). Moji no Shurui [Types of Characters]. In Kokugakugairon. Tokyo: Iwanami Shoten. (Republication of original work, 2 vols., published 1932-1933). [An Outline of Japanese Linguistics]. Available on the World Wide Web at

Hattori, S. (1961). Prosodeme, Syllable Structure and Laryngeal Phonemes. In Linguistics in Japan, 1980, vol. 2. 186-218. Shibata, T., Kitamura, H. and Kindaichi, H. (Eds.). Tokyo: Taishukan.

Henshall, K. (1988). A Guide to Remembering the Kanji Characters. Tokyo: Tuttle.

Higginson, W. (1985). The Haiku Handbook. Tokyo: Kodansha.

Inamoto, N. (1993). Colloquial Japanese. Tokyo: Tuttle.

Kakehi, K., & Hirose, Y. (1997, September). Mora Identification and the Use of Prosodic Cues. Transcripts of the Technical Committee on Psychological and Psysiological Acoustics. H-97-68. [Japanese].

Kawamura, H. and Reichhold, J. (1998). White Letter Poems: Fumi Saito. Gualala, CA: AHA Books.

Kenkyusha's Japanese-English Dictionary. (1973). (4th ed.). Tokyo: Kenkyusha.

Kindati, H., Hayasi, O. & Sibata, T. (Eds.). (1988). Nihongo Hyakuka Daijiten. Tokyo: Taishukan Shoten. [An Encyclopedia of the Japanese Language].

Kirkup, J. (1995) [quoted by Sato, H.]. Frogpond: Journal of the Haiku Society of America, 18:1, 39.

Kondo, T. (1978). In support of onji rather than jion [Letter to the editor, dated January 29, 1976]. Frogpond: Journal of the Haiku Society of America, 1:4, 30-31.

Mannen, U. (1904 or prior). Kari-ji Meishyou-kou [The Origination of Kana]. Retrieved July 23, 1998 from the World Wide Web:

Mannen, U. & Matsui., K. (Eds.). (1972) Dai Nihon Kokugo Jiten. Tokyo: Fuzanbou. (Republication of original work, 4 vols., published 1915-1919; first single vol. ed., 1952). [Japanese Language Dictionary].

Mitamura, J. and Mitamura, Y. (1997). Let's Learn Kanji. Tokyo: Kodansha.

Nihon Kokugo Daijiten. (1988). (20 vols.). Tokyo: Shogakukan. [The Encyclopedic Dictionary of the Japanese Language].

Omuro, K., Baba, R., Miyazono, H., Usagawa, T. & Egawa, Y. (1996, December). The Perception of Morae in Long Vowels. Proceedings of the Third Joint Meeting, Acoustical Society of Japan. Honolulu, Hawaii.

Reichhold, J. (1999). AHA! POETRY Website. On the World Wide Web:

Shinmura, I. (Ed.). (1998). Koujien (5th ed.). Tokyo: Iwanami Shoten. [Encyclopedic Dictionary of the Japanese Language].

Tsujimura, N. (1996). An Introduction to Japanese Linguistics. Cambridge, MA: Blackwell.  


[1] I first became interested in this subject nearly twenty years ago, while attending the Naropa Institute, in Boulder, Colorado. I encountered the traditional haiku masters through the auspices of Patricia Donegan (Donegan & Ishibashi, 1998), and became involved with various American haiku circles over succeeding years. This paper focuses on linguistic issues related with the most popular traditional Japanese poetic forms, both within and outside of Japan: haiku and tanka. Nonetheless, one can likewise apply this discussion to other forms of Japanese poetry, where counting hyouon moji comes into play.

[2]Some examples of 'counters,' also known as 'numerators' or 'numeral classifiers': round slender objects take -hon; flat objects take mai;, (postal) letters take -su; footgear take -soku; vehicles take -dai; animals take -hiki, etc. (cf. Inamoto, 1993, pp. 69-73.)

[3] Apparently, mora, defined by the American Heritage Dictionary, 1992, as "(Latin) The minimal unit of metrical time in quantitative verse, equal to the short syllable," has entered Japanese linguistic circles as a loan-word. I have found this term used frequently within Japanese as well as English linguistics contexts and have also noted its use as a 'counter in Japanese.

[4] Hence the controversy, from the Japanese side, regarding the use of the term on-setsu, when applied to non-moraic languages like English.

[5] Incidentally, this includes the digraphs, or 'double-characters,' such as kyo, jyo, gyo, etc. These combined characters too are each counted as one -on (or one -ji), and they take about the same amount of time to speak as ko, ji, go, etc. For simplicity's sake, and the benefit of a readership possibly unfamiliar with the kana alphabets, the digraphic kana are being included here under the appellation "single kana characters," though they are actually composed of two kana characters, which produce a single "combined" character.

[6] "Previous researchers insisted that only the duration of vowels affects the perception of the number of morae. . . . however it is clear that not only the duration of the vowel sequence but also the accentual change has an important influence on the perception of Japanese subjects" (Omuro, Baba, Miyazono, Usagawa and Egawa, 1996, p. 6). ". . . Various kinds of information (pitch, rhythm, duration, lexical information) contribute in segmentation of three or more consecutive vowels" (Kakehi and Hirose, 1997, p 1, abstract).

[7] For example, the kanji i, as in iku (to go), takes the sound gyou in gyouretsu (procession-- introduced to Japan, 5th-6th century CE), takes kou in koushin (march, parade-- introduced, 7th-9th century CE), and an in anka (foot warmer-- introduced, 10th-13th century CE). (cf. Mitamura & Mitamura, p xiii)  

[8] "In a handful of cases new characters were created in Japan using Chinese elements, such as 'dry field' and 'frame,' and some of these have since been borrowed for use in Chinese (such as 'work.' These 'made in Japan' characters usually-- but not necessarily-- have KUN readings only" (Henshall, p xiv). 

[9] It is only in the last few years that contemporary tanka and haiku poets have begun to be published with some frequency in English translation. Very little is known about contemporary Japanese haiku and tanka culture in North America, and vice versa. Translations from English haiku and tanka into Japanese are still relatively rare. 


Lexical Glossary

Moji-- Any written character. Moji describes all of the characters that are used in Japanese, which can include both kanji and kana. Moji is also sometimes used informally to count totals of -on or -ji in poems. We can also say that, "this haiku has 17 moji."

Hyoui-- "Representative of ideas." Ideogrammatic representation.

Hyoui moji-- "Characters (moji) which are representative (hyou) of meaning (i)." Kanji are hyoui moji. Hyoui moji are ideogrammatic. A number of hyoui moji are directly traceable to hieroglyphs or pictographs (shoukei moji).

Iji-- An archaic term for hyoui moji. "Meaning (i) characters (ji)." No longer in use.

Hyouon-- "Representative of sound." Phonic representation.

Hyouon moji-- "Characters (moji) which are representative (hyou) of sound (on)." "Sound-representation characters." Hyouon moji include the hiragana and katakana alphabets. (Romaji is also a form of houon moji.) In modern Japanese, hyouon moji are individual phonic units, each of which is a separate alphabetical character (including digraphs) in the hiragana and katakana alphabets.

Onji-- An archaic term for hyouon moji. "Sound (on) characters (ji)." No longer in use.

On-setsu-- A term which has a variety of sometimes contradictory definitions in contemporary Japan; one of the definitions, given by Hattori, defines on-setsu as a term indicative of English-style syllabification as applied to Japanese, as well as English, among other languages (e.g. "English on-setsu"). The word koukousei has three on-setsu: kou/kou/sei, but six -on: ko/u/ko/u/se/i. On-setsu is the closest available term to the English "syllable." But this would be an improper method for parsing words in Japanese poetry (see "ji" and "on," below). Word-parsing by on-setsu is counter-intuitive to the perception of Japanese native-speakers.

On-- (lit. "sound")-- A term used to count kana, or individual phonic units (hyouon) in poetry. It is this counter (or see "-ji," below), which we want to use when we are counting the hyouon in Japanese poetry, as in: "The first line of this haiku has 5-on." On is also used to express the total number of kana (or phonic units) in a poem, as in: "This tanka has 31-on." koukousei has 6-on: ko/u/ko/u/se/i. Properly, we can say that most haiku contain 17-on.

Ji-- (lit. "character")-- Along with -on, another often-used counting term to count kana, or individual phonic units (hyouon) in poetry. We can use -ji when we count up the kana in Japanese poetry, as in: "The first line of this haiku has 5-ji." Ji is not generally used to express the total number of kana in a poem. The word: koukousei has 6-ji: ko/u/ko/u/se/i. As with -on, above, this is another appropriate method for counting kana in Japanese poetry.

Kana-- This term has several referents. It can refer to the hiragana or katakana alphabets, as in: "Write it only in kana; no kanji, please." It can refer to an individual alphabetical character, as in: "Which kana is that?" "It's ko." Most kana are single characters; however the group that uses -y glides, like kyo, jyo, etc., are made up of two combined characters. These are digraphic kana. In this article, for simplicity, digraphs are here included as "individual alphabetical characters." English-speakers will use the term informally, when speaking in English, as in: "I tried to write a poem in Japanese. I used 24 kana." Kana are hyouon moji.

Kanji-- Kanji are generally classed as ideograms, originally imported from China, with many later additions and alterations in Japan. Kanji are hyoui moji, with some exceptions.

Romaji-- Romaji is a transcription alphabet which uses roman letters. Romaji is not a kana alphabet, nor is it a properly Japanese alphabet. It was primarily designed to aid those unable to read Japanese kana. The Roman alphabet is one variety of hyouon moji, known as tan-on moji, or single-phoneme characters, in Japanese linguistics. Many contemporary Japanese transliterations now use the modified Hepburn romanization (see Campbell et al, pp. 665-68).

Mora-- A linguistic term used to identify the sense of "phonic (hyouon) time-units" or "time-lengths" in Japanese speech. Mora, and morae, its plural, are English linguistics terms and also Japanese loan-words. In written Japanese, the number of mora will always agree with the number of -on in a poem. The actual number of perceived mora may differ in spoken Japanese.

Haku-- A synonym of mora. Mora is the technical linguistic term, while haku is the more commonly known term, familiar to most Japanese people. Both are used to count phonic (hyouon) time-lengths in Japanese. Haku is also a 'counter.'

- - - - - - - - - - - - - - - - - - -

return to research page