Old Japanese

From Infogalactic: the planetary knowledge core
Jump to: navigation, search
Old Japanese
Region Japan
Era Evolved into Early Middle Japanese during the Heian period
  • Old Japanese
Chinese characters and Man'yōgana
Language codes
ISO 639-3 ojp
Linguist list
ojp [lower-roman 1]
Glottolog oldj1239[1]
This article contains IPA phonetic symbols. Without proper rendering support, you may see question marks, boxes, or other symbols instead of Unicode characters.

Old Japanese (上代日本語 jōdai nihongo?) is the oldest attested stage of the Japanese language. It was spoken by the Japanese ethnic group from an unknown beginning point until it evolved into Early Middle Japanese in the eighth century, during the Heian period, although the precise separation of these two languages is controversial. Old Japanese was an early member of the Japonic family; no conclusive links to other language families have been drawn, although the Altaic family is frequently suggested.

Old Japanese was written using Chinese characters, using an increasingly standardized and phonetic form that eventually evolved into man'yōgana. Typically for a Japonic language and for a step in the evolutionary line of modern Japanese, Old Japanese was a primarily agglutinative language with subject–object–verb word ordering. However, the language was marked by a few phonemic differences from later forms of Japanese, such as an eschewing of diphthongs. It distinguished between a few pairs of syllables with identical pronunciations—a phenomenon known as Jōdai Tokushu Kanazukai—but the function of this differentiation is not known.


Linguistic changes are gradual, and the periodization of Japanese is "both delicate and controversial", with multiple competing methods and criteria for division.[2] For both practical and conventional reasons, these divisions often correlate to political events.

The lower boundary, i.e. the latest possible date for the end of the Old Japanese period, is AD 794,[2] when the capital Heijōkyō moved to Heiankyō. However, the upper boundary is more difficult. A limited number of Japanese words, mostly personal names and place names, are recorded phonetically in ancient Chinese texts such as the "Wei Zhi" portion of the Records of the Three Kingdoms (3rd century AD). Wooden tablets and relics with fragments of text written on them have also been excavated. However, the first extant written text of substantial length is the Kojiki of 712. Without necessarily excluding such fragmentary early evidence, the upper boundary is generally discussed as from circa 712 for practical reasons. This coincides well with the Nara period (710-794). A more formal dating would not recognize an upper boundary and merely date it as through 794.

Besides Kojiki, the other oldest literary sources include Fudoki (720), Nihon Shoki (720), and Man'yōshū (c. 759).

Writing system

Calligraphy during the Nara period, near the end of Old Japanese's lifespan

The earliest texts found in Japan are written in Classical Chinese, although they may have been meant to be read as Japanese by the kanbun method. Some of these Chinese texts show the influences of Japanese grammar, such as the word order (for example, placing the verb after the object). In these "hybrid" texts, Chinese characters are occasionally used phonetically to represent Japanese particles. Over time, the phonetic usage of Chinese characters became more and more prevalent, until Man'yōgana, a system of using the Chinese characters phonetically to record Japanese, was born. This system was already in use in the non-prose part of Kojiki, and was used in a highly sophisticated manner in Man'yōshū.


The study of Old Japanese phonology is based on the comparative study of synchronous pronunciation of Chinese, reverse analysis of diachronic change in Japanese pronunciation, and comparative study of the Ryukyuan languages. Although the majority of Old Japanese writing represents the language of the Nara court in central Japan, some poems in the Man'yōshū are from southern and eastern Japan, and represent different dialects of Old Japanese. Some of these dialectal differences are still found today.[citation needed]

Old Japanese differed phonetically from later periods of the language. An analysis of Man'yōgana reveals a peculiar system known as Jōdai Tokushu Kanazukai. See the Syllables section for details.

The transcriptions of Old Japanese words given in Kojiki (712) differ from those found in Nihon Shoki (720) and Man'yōshū (c. 759) in that it distinguishes the syllables /mo1/ and /mo2/ whereas the latter two do not. This has been correlated with the historical record of Kojiki being compiled earlier than Nihon Shoki, and thus preserved an older distinction that soon vanished.

Other characteristic differences of Old Japanese as compared with its modern counterpart include:

  • words do not begin with voiced plosives or /r/
  • open syllables: no syllable-final consonants of any kind.

Some scholars have suggested that there might be a link between Old Japanese and some of the extinct languages of the Korean peninsula, including the Gaya language, but the relation of Japanese to any language other than Ryūkyūan remains undemonstrated. See Classification of Japanese for details.


Old Japanese distinguished 88 syllables, assuming OJ has the same gaps of *yi and *wu that later stages of the language had:

a i u e o
ka ki1 ki2 ku ke1 ke2 ko1 ko2
ga gi1 gi2 gu ge1 ge2 go1 go2
sa si su se so1 so2
za zi zu ze zo1 zo2
ta ti tu te to1 to2
da di du de do1 do2
na ni nu ne no1 no2
pa pi1 pi2 pu pe1 pe2 po
ba bi1 bi2 bu be1 be2 bo
ma mi1 mi2 mu me1 me2 mo1 mo2
ya   yu ye yo1 yo2
ra ri ru re ro1 ro2
wa wi   we wo

Shortly after the Kojiki, the distinction between mo1 and mo2 was lost, reducing the syllable count to 87.

Several hypotheses have been proposed to explain the syllable doublets, involving:

This issue is hotly debated, and there is no consensus.


The transcription does not imply a particular theory, and the subscript 1 or 2 could apply to either the consonant or the vowel.

There are several competing transcription systems. One popular system places a dieresis above the vowel: ï, ë, ö. These represent i2, e2, and o2, and assume that unmarked i, e, and o are i1, e1, and o1. This is problematic in that it neglects to distinguish words where the 1 / 2 distinction is not clear, such as the o in /toru/. To address this, ǐ, ě, ǒ may be used for i1, e1, o1. Another common transcription is yi, ye, o for i1, e1, o1, and iy, ey, wo for i2, e2, o2. This reflects their suspected derivation in proto-Japanese.


The number of vowel distinctions may have been anywhere from five to eight, depending on the hypothesis. These are the four principal (more common) vowels a, u, i1 (yi), o1 (o), and the four secondary (less common) vowels i2 (ï/iy), e1 (e/ye), e2 (ë/ey), o2 (ö/wo). Miyake reconstructs the four principal vowels as /a, u, i, ə/ and the four secondary vowels as /ɨj, ʲe, əj, o/. The four principal vowels descend from pre–Old Japanese/proto-Japanese *a, *u, *i, *ə. /u/ was prototypically rounded, unlike in Modern Japanese. /ʲe/ descends from pOJ *ia and *iə, and may have been either [je] or [e] preceded by a palatalized consonant: /pje/, etc. /ɨ/ descends from pOJ *ui and *əi.[3]


The number of consonants likewise depends on the hypothesis. Miyake reconstructs the following inventory, in addition to a zero (vowel-initial) onset /∅/:[4]

Labial Coronal Palatal Velar
Voiceless obstruent *p *t *s   *k
Prenasalized voiced obstruent *ᵐb *ⁿd *ⁿz   *ᵑɡ
Nasal *m *n    
Approximant/Flap *w *r *j  


The voiceless obstruents /p, t, s, k/ had corresponding voiced consonants which were prenasalized. This prenasalization is seen through Late Middle Japanese, and in the case of /ɡ/, as an intervocalic allophone [ŋ] to the present day. /t, d/ were not, and /s, z/ may not have been, palatalized before /i/, as they would be later.

Voiceless labial obstruent

Modern Japanese /h/ was phonetically realized as [p] in Old Japanese. This assumption is predicated upon the following textual and phonological analyses:

  • The modern /h/ causes a discrepancy in the pairing of voiceless vs. voiced consonants. Thus, /k, ɡ/, /s, z/, /t, d/, and finally /h, b/. The pair /h, b/ does not fit. The voiceless version of /b/ is /p/.
  • Comparison with Ryukyuan language shows [p] where mainland Japanese has [h]. As these two languages split at some point in history, this can be taken as evidence that the Japanese [h] was once pronounced identically to the Ryukyuan [p] (although the comparison alone does not directly address the issue of what its pronunciation was in Old Japanese).
  • A look at /h/ in modern Japanese shows that it becomes [ɸ] when followed by /u/. Looking further back, Portuguese missionaries visiting Japan in the early 17th century wrote the entire /h/-row of kana as "fa, fi, fu, fe, fo". Korean visitors in the same century suggested a voiceless bilabial fricative sound, i.e. [ɸ].
  • In 842, the monk Ennin writes in Zaitōki a description in which he states that Sanskrit "p" is more labial than Japanese. This is taken as evidence that Japanese /h/ was pronounced [ɸ] rather than [p] by that time. This was after the Old Japanese period, however.

There is general agreement that between the 9th and 17th century /h/ was [ɸ]. The dialectal and distributional evidence suggest that at some point it must have been [p]. Recent scholarship reconstructs this as *[p] for Old Japanese.[5]

Phonological rules

In 1934, Arisaka Hideyo proposed a set of phonological restrictions permitted in a single morpheme. These are known as the "Arisaka Laws".

  • -o1 and -o2 do not co-exist
  • -u and -o2 generally do not co-exist
  • -a and -o2 generally do not co-exist

These rules suggest two groups of vowels: /-a, -u, -o1/ and /o2/. Vowels from either group do not mix with each other; -i1 and -i2 can co-exist with either group. Some[who?] take this phenomenon as evidence that Old Japanese had vowel harmony as found in Altaic languages.

Syllable structure

The Old Japanese syllable was CV (consonant-vowel).

  • A bare vowel does not occur except for word-initially.
  • /r/ is not found word-initially (with the exception of two foreign loans: /rikizimahi1/ and /rokuro/).
  • Voiced plosives do not occur word-initially.

Vowel elision occurred to prevent vowel clusters:

  • Second vowel is dropped: /hanare/ + /iso1/ → /hanareso1/
  • First vowel is dropped: /ara/ + /umi1/ → /arumi1/
  • Two continuous vowels merge into a separate vowel: i1 + a → e1, a + i1→e2, o2 + i1→i2
  • /s/ is inserted between the two vowels: /haru/ + /ame2/→/harusame2/ (It is possible that /ame2/ was once */same2/)



  • First person: wa, a, ware, are
  • Second person: na, nare, masi, mimasi, imasi, ore
  • Third person:
    • Proximal: ko, kore, koko, koti
    • Mesial: so, si, soko
    • Distal: ka, kare
  • Interrogative: ta, tare, idu, idure, iduti, iduku, idura


Old Japanese distinguished between eight verbal conjugations: quadrigrade, upper monograde, upper bigrade, lower bigrade, k-irregular, s-irregular, n-irregular, and r-irregular. Lower monograde does not exist yet.[6][7][8]


Verb Class Irrealis Adverbial Conclusive Attributive Realis Imperative
Quadrigrade -a -i1 -u -u -e2 -e1
Upper monograde - - -ru -ru -re -(yo2)
Upper bigrade -i2 -i2 -u -uru -ure -i2(yo2)
Lower bigrade -e2 -e2 -u -uru -ure -e2(yo2)
K-irregular -o2 -i1 -u -uru -ure -o2
S-irregular -e -i -u -uru -ure -e(yo2)
N-irregular -a -i -u -uru -ure -e
R-irregular -a -i -i -u -e -e

Consonant / vowel stem

Verbs having a base that ends in a consonant are known as consonant-stem. These are exhibited by the Quadrigrade, the R-irregular and N-irregular verb classes.

There are only three N-irregular verbs, which were originally verbs in the Adverbial conjugation combined with the negative auxiliary -nu, but which were later reinterpreted as verbs in their own right.

Verbs having a base that ends in a vowel are known as vowel-stem. Upper Monograde represents monosyllabic vowel stems ending in -i.

Upper Bigrade is made up of polysyllabic roots ending in the vowel -i, while Lower Bigrade represents polysyllabic roots ending in -e. This ending vowel was elided in the conclusive, attributive, and realis conjugations.

K-irregular and S-irregular represent verbs whose stems appear to be composed of only one consonant each. These verbs behave as verbs with consonant stems, but they are originally vowel stems whose irregular conjugations are the result of vowel mutation and elision.

Irregular verbs

There are several verbs with irregular conjugations.

  • K-irregular: k- "come"
  • S-irregular: s- "do"
  • N-irregular: sin- "die", in- "go, die"
  • R-irregular: ar- "be, exist", wor- "be, exist"

The conjugation class for each is named after the final stem consonant.


There were two types of adjectives: regular adjectives and adjectival nouns.

The regular adjective is sub-classified into two types: those where the adverbial form (連用形) ends in -ku and those that end in -siku. This creates two different types of conjugations:

Adjective Class Irrealis Adverbial Conclusive Attributive Realis Imperative
-ku -ke1 -ku -si -ki1 -ke1
-kara -kari -si -karu -kare -kare
-siku -sike1 -siku -si -siki1 -sike1
-sikara -sikari -si -sikaru -sikare -sikare

The -kar- and -sikar- forms are derived from the verb ar- "be, exists". The adverbial conjugation (-ku or -siku) is suffixed with ar-. The conjugation yields to the R-irregular conjugation of ar-. As Old Japanese avoids vowel clusters, the resulting -ua- elides into -a-.

The adjectival noun has a single conjugation:

  Irrealis Adverbial Conclusive Attributive Realis Imperative
Adjectival noun -nara -nari -nari -naru -nare -nare


The Man'yōshū includes poems written in an eastern dialect.[citation needed]


Four-vowel system

The following fusional reductions (in most instances these are monophthongizations) took place:

  • *i1a > /e1/
  • *ai1 > /e2/
  • *ui1 > /i2/
  • *o1i1 > /i2/
  • *au > /o2/
  • *ua > /o2/

Thus, the proto vowel system can be reconstructed as /*a, *i, *u, *o1/, where o1 represents schwa..

Co2 as Cwo

Distributionally, there may have once been *po1, *po2 and *bo1, bo2. Note that the distinction between /mo1/ and /mo2/ is only seen in Kojiki and vanished afterwards. If this is true, then a distinction was made between Co1 and Co2 for all consonants C except for /w/. Some take this as support that Co2 may have represented /Cwo/.

See also


  1. Described as "The ancestor of modern Japanese. 7th–10th centuries AD." The more usual date for the change from Old Japanese to Middle Japanese is ca. 800 (end of the Nara era).


  1. Nordhoff, Sebastian; Hammarström, Harald; Forkel, Robert; Haspelmath, Martin, eds. (2013). "Old Japanese". Glottolog. Leipzig: Max Planck Institute for Evolutionary Anthropology.<templatestyles src="Module:Citation/CS1/styles.css"></templatestyles>
  2. 2.0 2.1 Shibatani 1990, p. 119
  3. Miyake 2003, p. 198
  4. Miyake 2003, p. 196
  5. Miyake 2003, pp. 164–166
  6. Yamaguchi 1997, p. 18
  7. Kondō 2005, p. 41
  8. Omodaka 1967, pp. 37–38


  • Kondō, Yasuhiro; Masayuki Tsukimoto; Katsumi Sugiura (2005). Nihongo no Rekishi (in Japanese). Tōkyō: Hōsō Daigaku Kyōiku Shinkōkai. ISBN 4-595-30547-8. <templatestyles src="Module:Citation/CS1/styles.css"></templatestyles>
  • Miyake, Marc Hideo (2003). Old Japanese: A Phonetic Reconstruction. London; New York: RoutledgeCurzon. ISBN 0-415-30575-6.<templatestyles src="Module:Citation/CS1/styles.css"></templatestyles>
  • Omodaka, Hisataka, ed. (1967). Jidaibetsu Kokugo Daijiten: Jōdaihen (in Japanese). Tōkyō: Sanseidō. ISBN 4-385-13237-2. <templatestyles src="Module:Citation/CS1/styles.css"></templatestyles>
  • Shibatani, Masayoshi (1990). The Languages of Japan. Cambridge: Cambridge University Press. ISBN 0-521-36918-5.<templatestyles src="Module:Citation/CS1/styles.css"></templatestyles>
  • Yamaguchi, Akiho; Hideo Suzuki; Ryūzō Sakanashi; Masayuki Tsukimoto (1997). Nihongo no Rekishi (in Japanese). Tōkyō: Tōkyō Daigaku Shuppankai. ISBN 4-13-082004-4. <templatestyles src="Module:Citation/CS1/styles.css"></templatestyles>

Further reading

  • Frellesvig, Bjarke (2010). A History of the Japanese Language. Cambridge: Cambridge University Press. ISBN 978-0-521-65320-6.<templatestyles src="Module:Citation/CS1/styles.css"></templatestyles>
  • Martin, Samuel E. (1987). The Japanese Language Through Time. New Haven: Yale University Press. ISBN 0-300-03729-5.<templatestyles src="Module:Citation/CS1/styles.css"></templatestyles>
  • Ōno, Susumu (2000). Nihongo no Keisei (in Japanese). Tōkyō: Iwanami Shoten. ISBN 4-00-001758-6. <templatestyles src="Module:Citation/CS1/styles.css"></templatestyles>
  • Tōdō, Akiyasu; Yoshimitsu Kanō (2005). Gakken Shin Kan-Wa Daijiten (in Japanese). Tōkyō: Gakushū Kenkyūsha. ISBN 4-05-300082-3. <templatestyles src="Module:Citation/CS1/styles.css"></templatestyles>
  • Vovin, Alexander (2005). A Descriptive and Comparative Grammar of Western Old Japanese. Part One: Sources, Script and Phonology, Lexicon, Nominals. Folkestone, Kent: Global Oriental. ISBN 978-1-901903-14-0.<templatestyles src="Module:Citation/CS1/styles.css"></templatestyles>
  • Vovin, Alexander (2009). A Descriptive and Comparative Grammar of Western Old Japanese. Part Two: Adjectives, Verbs, Adverbs, Conjunctions, Particles, Postpositions. Folkestone, Kent: Global Oriental. ISBN 978-1-905246-82-3.<templatestyles src="Module:Citation/CS1/styles.css"></templatestyles>