Old Korean

Authored by: Nam Pung-hyun (南豊鉉)

The Languages of Japan and Korea

Print publication date:  May  2012
Online publication date:  June  2012

Print ISBN: 9780415462877
eBook ISBN: 9780203124741
Adobe ISBN: 9781136446597




We divide Old Korean (OK) into Early, Mid and Late Old Korean (EOK, MOK, LOK). EOK was the Korean of the Three Kingdoms period, roughly from the start of the fifth century until Silla unified the Three Kingdoms in the 660s. MOK was the Korean of the Unified Silla [Sinla] period, from the 660s until the 930s when Koryŏ [Kolye] re-unified the country. LOK was the language of the earlier part of the Koryŏ dynasty from the 930s till the mid-thirteenth century.

 Add to shortlist  Cite

Old Korean

3.1  Periodization

We divide Old Korean (OK) into Early, Mid and Late Old Korean (EOK, MOK, LOK). EOK was the Korean of the Three Kingdoms period, roughly from the start of the fifth century until Silla unified the Three Kingdoms in the 660s. MOK was the Korean of the Unified Silla [Sinla] period, from the 660s until the 930s when Koryŏ [Kolye] re-unified the country. LOK was the language of the earlier part of the Koryŏ dynasty from the 930s till the mid-thirteenth century.

The reason why I treat the language of the earlier part of the Koryŏ period as late Old Korean is because the grammar of ‘interpretive’ kugyŏl, presumed to be data from the tenth to the mid-thirteenth century, and that of ‘consecutive’ kugyol, which constitutes the Early Middle Korean data of the later thirteenth century, show quite significant differences. The social factor that separates the OK and MK worlds is taken to be the Mongol invasion and occupation. this resulted in large population movement, and the nomadic culture that subsequently entered Korea greatly changed the traditional culture that had continued since ancient times, heavily affecting language and culture.

The EOK period followed the establishment of the Three Kingdoms: Silla [Sinla] 新羅, Paekche [Paykcey] 百濟 and Koguryŏ [Kokwulye] 高句麗. AS the Three Kingdoms were independent kingdoms, some believe that each must have had its own independent language, but the territory unified by Silla was roughly the territory of the early city-states (the ‘Samhan’ period, from first century bc), and the texts left by the Three Kingdoms are not so diverse that we should regard the Three Kingdoms’ varieties as separate languages. Moreover, it is preferable to treat what has been left from this period as texts in the Korean language, and the characteristics shown in them suggest dialectal differences of a single language.

3.2  Data

It is a universal fact of any language that the further back we trace it into the past the scarcer the written materials are, and OK texts are in addition very crude considering that they extend over 900 years or more. I shall divide them into internal and external materials.

Internal materials are texts which are documented by borrowing Chinese characters. We call this writing ch’aja [chaca] 借字 or ‘loan character’ writing, and texts written with ch’aja writing have been divided previously into idu [itwu] 吏讀, hyangch’al [hyangchal] 鄕札 and kugyŏl [kwukyel] 口訣. We can include vocabulary lists as a fourth category.

I shall start briefly with vocabulary list materials. These consist of the recording by means of Chinese characters of native Korean words. They record proper nouns such as toponyms and names of people, and the names of official ranks or roles and names of things. There are many written inside Chinese texts (epigraphs) and idu or hyangch’al texts, and there are also collations of vocabulary such as the geographical chapters (地理志) of Samguk sagi [Samkwuk saki] 三國史記 (mid-twelfth century). These are difficult to decipher and different researchers suggest different views, but in the case of EOK for which texts are scarce we have to rely on them to obtain crucial information. The folk vocabulary recorded in the Kyerimyusa [Kyeylim yusa] 鷄林類事 (c. 1103) and the mid-thirteenth century Hyangyak kugŭppang [Hyangyak kwukuppang] 鄕藥救急方 are also lexical list materials. There are records of EOK/MOK vocabulary in the Samguk yusa [Samkwuk yusa] 三國遺事 (late thirteenth century) in addition to its hyangga poems, and we can also use the Koryŏsa [Kolyesa] 高麗史 (mid-fifteenth century) for snapshots of LOK vocabulary.

3.2.1  Idu data

‘Idu’ in the broad sense sometimes refers to all the ch’aja texts, but in practical terms, once we separate off hyangch’al and kugyŏl, it refers to prose in ch’aja. Most kugyŏl texts, in terms of what is extant, consist of administrative texts.

Extant EOK idu texts are in the form of texts on metal or stone, known as ‘pseudo-Chinese’ (pyŏnch’e hanmun [pyenchey hanmun] 變體漢文) or ‘early idu’ texts. Koguryŏ idu texts include the King Kwanggaet’o [Kwangkaytho] stela (廣開土大王碑銘, 414), the Sŏbong-ch’ong [Sepong-chong] silver bowl inscription (瑞鳳冢銀盒杅銘, 451?) the Chungwŏn Koguryŏ [Cwungwen Kokwurye] stela (中原高句麗碑銘 590s?) and the four P’yŏngyang Koguryŏ castle wall engravings (平壤高句麗城壁刻字). The Kwanggaet’o stela is typical Koguryŏ Chinese, but its word order and use of declarative 之 display features of early idu. The Chungwŏn inscription is in word order and its use of idu a typical Koguryŏ idu text.

Paekche idu texts consist of the silver bracelet inscription (銀釧銘, c. 520) and the so-called sukse-ga [swuksey-ka] 宿世歌 (‘songs to past lives’) written on wooden slats excavated at the temple Puyŏrŭng-sa [Puyelung-sa]. Their word order and expressions are Korean.

The extant idu texts of Silla are the most numerous of the Three Kingdoms. Abbreviated names are given below, with a full name in Chinese characters in parentheses.

  • EI-i. Naengsuri [Nayngswuli] inscription (迎日冷水里碑), 503?
  • EI-ii. Pongp’yŏng [Pongphyeng] inscription (蔚珍鳳坪新羅碑), 524?
  • EI-iii. Ch’ŏnjŏlli [Chencenli] first inscription (蔚州川前里書石原銘), 525?, and additional inscription, 539?
  • EI-iv. Chŏksŏng [Cekseng] inscription (丹陽新羅赤城碑), 540?
  • EI-v. Imsin oath inscription (壬申誓記石銘), 552?
  • EI-vi. Ojak [Ocak] inscription (戊戌塢作碑), 578?
  • EI-vii. Sinsŏng [Sinseng] inscriptions (南山新羅碑) 1 to 9, 591?
  • EI-viii. Myŏnghwal-san [Myenghwal-san] inscription (明活山城作成碑), 611?

(EI-i–iv) mix Korean and Chinese word orders, whereas (EI-v) and (EI-viii) are arranged completely in Korean word order but use characters as ŭmdokcha (see 3.3). This writing continued in later periods, when it constituted one of the styles of writing. (EI-vi–vii) are also arranged in Korean word order, but use hundokcha. There are many collocations used in these that continue in later administrative texts.

Early idu style continued into the MOK period; the following are representative texts:

  • MI-i. In’yang-sa [Inyang-sa] inscription (昌寧仁陽寺碑銘), 810
  • MI-ii. Chungch’o-sa [Cwungcho-sa] pillar inscription (中初寺幢竿石柱記), 827
  • MI-iii. samhwa-sa Buddha inscription (東海市三和寺鐵佛造像銘), 860

A sentence of (MI-ii) is parsed in (1):

























9th month




‘At Sŭngak to the east of Chungch’o-sa they split a stone into two. On the 28th of the same month two groups began shaping them, and on the 1st of the 9th month they arrived here.’

MOK texts emerged with ‘t’o [tho] 吐 marks’ added to early idu style, displaying features of full-blown idu texts. Twenty-five or so such idu texts are known at present, of which the following are representative:

  • MT-i. Kamsan Amida [Kamsan Amita] inscription (甘山寺阿彌陀佛像造成銘), 720
  • MT-ii. Mujin-sa [Mucin-sa] bell inscription (无盡寺鐘銘), 745
  • MT-iii. Hwaŏm-sa [Hwaem-sa] sutra notes (華嚴經寫經造成記), 755
  • MT-iv. The Silla [Sinla] Register (新羅帳籍), pre-758
  • MT-v. Karhang-sa [Kalhang-sa] pagoda inscription (葛項寺石塔), 758
  • MT-vi. Ch’ŏpp’ogi [Chepphoki] (正倉院所藏毛氈貼布記) notes, mid-eighth century
  • MT-vii. Vairocana-Buddha inscription (永泰2年毘盧遮那佛造成銘), 766
  • MT-viii. Ch’ŏngje [Chengcey] inscription (永川菁堤碑貞元銘), 798
  • MT-ix. Sŏllim-wŏn [Senlim-wen] bell inscription (禪林院鐘銘), 804
  • MT-x. Kyuhŭng-sa [Kyuhung-sa] bell inscription (竅興寺鐘銘), 856

These are inscriptions on metal or stone, and old manuscripts. There are, in addition, texts on wooden slats. (MT-i) is a relatively simple record, but shows the earliest t’o. (MT-iii) is a typical idu text of the period, displaying diverse idu symbols. (MT-iv) is a Silla administrative document that displays diverse administrative terms and idu symbols. A sentence of (MT-iii) is parsed in (2):































(Underlined characters are t’o.)

‘The Way of making the sutra is to pour perfumed water on the root of the paper mulberry tree, mature it, … and to give commandments to the errand boys of the calligraphers of the Buddhist statues and have them take ritual meals, and …’

Sixty-plus Koryŏ period idu texts are known, and 20 or so of these are LOK. The most important of these are:

  • LT-i. Myŏngbong-sa [Myengpong-sa] inscription (醴泉鳴鳳寺慈寂禪師塔碑陰銘), 941
  • Lt-ii. Yaksa (= Bhaisaiyagura) inscription (校里磨厓藥師坐像銘), 977
  • LT-iii. Kaesim-sa [Kaysim-sa] pagoda record (醴泉開心寺石塔記), 1010
  • LT-iv. Pulguk-sa [Pulkwuk-sa] pagoda record (佛國寺無垢淨光塔重修記), 1024
  • LT-v. Chŏngdo-sa [Cengto-sa] pagoda record (若木淨兜寺五層石塔造成形止記), 1031
  • LT-vi. Pulguk-sa [Pulkwuk-sa] west pagoda record (佛國寺西石塔重修形止記), 1038

(LT-i) is an administrative document in which the Top’yŏng-sŏng [Tophyeng-seng] 都評省, a Koryŏ central government office, authorizes the founding of the temple Myŏngbong-sa. (LT-iv) and (LT-vi) are records of repairs to the pagoda at Pulguk-sa and show various idu symbols. In particular, they use characters abbreviated as – or according to the principle of – kugyŏl abbreviated forms, and this is the same logical principle as the use of old katakana in Japan. (LT-v) has been known for a long time. Part of it is parsed in (3):
























































‘District commoner Kwanghyŏn expressed the desire to build a five-storey stone pagoda, and on the seventh of the fifth month of the sixth year of the Ch’ŏnhŭi era when he had not been able to build it he departed this world because of illness, but deputy hojang [= Koryŏ period local headman] P’umyu who was his elder brother encouraged the noble monk Kagyu to inherit this wish and to complete the construction, and …’

3.2.2  Hyangch’al data

Hyangch’al was mainly used to write hyangga [hyangka] 鄕歌 songs in sentences recorded entirely in Korean. Scriptural commentaries that explained difficult passages of sutras are also included here. The following 14 hyangga have been preserved in Samguk yusa:

  • H-i. Mojukchirang-ga [Mocwukcilang-ka] 慕竹旨郞歌
  • H-ii. Hŏnhwa-ga [Henhwa-ka] 獻花歌
  • H-iii. Anmin-ga [Anmin-ka] 安民歌
  • H-iv. Ch’angip’arang-ga [Chankiphalang-ka] 讚耆婆郞歌
  • H-v. Ch’ŏyong-ga [Cheyong-ka] 處容歌
  • H-vi. Sŏdong-yo [Setong-yo] 薯童謠
  • H-vii. Toch’ŏnsugwanŭm-ga [Tochenswukwanum-ka] 禱千手觀音歌
  • H-viii. P’ung-yo [Phung-yo] 風謠
  • H-ix. Wŏnwangsaeng-ga [Wenwangsayng-ka]願往生歌
  • H-x. Tosol-ga [Tosol-ka] 兜率歌
  • H-xi. Chemangmae-ga [Ceymangmay-ka] 祭亡妹歌
  • H-xii. Hyesŏng-ga [Hyeyseng-ka] 彗星歌
  • H-xiii. Wŏn-ga [Wen-ka] 怨歌
  • H-xiv. Ujŏk-ka [Wucek-ka] 遇賊歌

Samguk yusa was written in the later thirteenth century, so (H-v), (H-vi) and (H-viii) are thought, because of how they are written and their grammar, to have been recorded not long before the thirteenth century. I treat these as LOK texts. (H-ii) is deciphered in (4):


































fact-TOP (= if)










‘Beside the purple rock [of azaleas] you made me let loose the cows [because of your beauty], and if you do not feel ashamed of me I shall pick a flower and give it to you.’

This is deciphered initially in MOK grammar and in MK lexis and morphology. Underlined elements are t’o.

Another 11 songs, the Pohyŏn sibwŏn-ga [Pohyen sipwen-ka] 普賢十願歌, or ‘songs of Samantabhadra’s ten great vows’, are included in the Kyunyŏjŏn [Kyunyecen] 均如傳:

  • P-i. Yegyŏng chebul-ga [Yeykyeng ceypul-ka] 禮敬諸佛歌
  • P-ii. Ch’ingch’an Yŏrae-ga [Chingchan Yeray-ka] 稱讚如來歌
  • P-iii. Kwangsu kongyang-ga [Kwangswu kongyang-ka] 廣修供養歌
  • P-iv. Ch’amhoe ŏpchang-ga [Chamhoy epcang-ka] 懺悔業障歌
  • P-v. Suhŭi kongdŏk-ka [Swuhuy kongtek-ka] 隨喜功德歌
  • P-vi. Ch’ŏngjŏn pŏmnyun-ga [Chengcen peplyun-ka] 請轉法輪歌
  • P-vii. Ch’ŏngbuljuse-ga [Chengpulcwusey-ka] 請佛住世歌
  • P-viii. Sangsu purhak-ka [Sangswu pulhak-ka]常隨佛學歌
  • P-ix. Hangsun chungsaeng-ga [Hangswun cwungsayng-ka] 恒順衆生歌
  • P-x. Pogae hoehyang-ga [Pokay hoyhyang-ka] 普皆廻向歌
  • P-xi. Ch’onggyŏl mujin-ga [Chongkyel mucin-ka] 總結无盡歌

3.2.3  Kugyŏl Data

Kugyŏl is a way of reading Chinese out as Korean by adding t’o marks to a Chinese text, but the term is also applied to the language. Most t’o are supplementary symbols for Korean morphology, such as particles and endings, and word-final sounds, but also include numerals and punctuation added to help understand the Chinese text. Kugyŏl is divided into ‘interpretive kugyŏl’ (sŏktok kugyŏl [sektok kwukyel] 釋讀口訣) for decoding Chinese and reading it as Korean and ‘consecutive kugyŏl’ (sundok kugyŏl [swuntok kwukyel] 順讀口訣), which intones the Chinese text following Chinese word order faithfully, while inserting Korean t’o in places corresponding to natural breaks in the text. Extant texts before the mid-thirteenth century are only interpretive kugyŏl; the extant consecutive kugyŏl texts date only to after the late thirteenth century but these are MK materials. Kugyŏl is divided into character-type kugyŏl and dot-type kugyŏl. Dot-type t’o are dots and lines, and also marks joining them. These dot-type t’o are marked in fixed positions inside or outside the imagined square occupied by a Chinese character with a stylus or in ink, and express pronunciation or morphology.

Character-type interpretive kugyol texts are:

  • K-i. Sŏk-Hwaŏm kyobun-gi [Sek-Hwaem kyopun-ki] 釋華嚴敎分記 (mid-tenth century)
  • K-ii. Hwaŏm-gyŏng-so [Hwaem-kyeng-so] 華嚴經疏 (estimated end of eleventh or start of twelfth century), volume 35
  • K-iii. Hwaŏm-gyŏng [Hwaem-kyeng] 華嚴經 (a text of garland Sutra, estimated first half of twelfth century), volume 14
  • K-iv. Happu Kŭmgwangmyŏng-gyŏng [Happu Kumkwangmyeng-kyeng] 合部金光明經 (a text of golden Light Sutra, estimated early thirteenth century), volume 3
  • K-v. Kuyŏk Inwang-gyŏng [Kwuyek Inwang-kyeng] 舊譯仁王經 (a text of Benevolent King Sutra, estimated first half of thirteenth century), volume 1, five pages
  • K-vi. Yugasaji-ron [Yukasaci-lon] 瑜伽師地論 (a text of Yogācāūrabhūmiśāstra, 1246), volume 20

(5a) reproduces horizontally (90° anticlockwise) lines from the original manuscript of (K-v), parsed in (5b–d):

























again-do-NMR = also







〢 七















‘also there are uncountably numerous people in other worlds, and …’

(5b) reorders the text into Korean word order, while (5c) converts the kugyŏl characters in (5b) into their source Chinese characters. Underlined characters in (5c) are hundokcha (see 3.3). These (5c) characters are identical to hyangch’al texts, so we know interpretive kugyŏl formed the basis for hyangch’al. (5d) is a transcription; Chinese loans are transcribed in capitals based on their MK forms.

Dot-type interpretive kugyŏl texts are:

  • Hwaŏm munŭi yogyŏl [Hwaem munuy yokyel] 華嚴文義要訣 (estimated mid-eighth century)
  • Chinbon Hwaŏm-gyŏng [Cinpon Hwaem-kyeng] 晋本華嚴經 (the Chinbon text of the garland Sutra, estimated ninth or tenth century), volume 20
  • Yugasaji-ron [Yukasaci-lon] 瑜伽師地論 (estimated first half of eleventh century), volumes 3, 5, 8
  • Chubon Hwaŏm-gyŏng [Cwupon Hwaem-kyeng] 周本華嚴經 (the Chubon text of the garland Sutra, estimated first half of the eleventh century), volumes 6, 22, 33, 34, 36, 57
  • Pŏphwa-gyŏng [Pephwa-kyeng] 法華經 (lotus sutra, estimated tenth or eleventh century), volume 1
  • Happu Kŭmgwangmyŏng-gyŏng [Happu Kumkwangmyeng-kyeng] 合部金光明經 (golden Light Sutra), volume 3

The symbols used in this dot-type kugyŏl are as in Figure 1. The clause 令一.衆生悉能除滅諸障閼業 is from line 7, page 2 of volume 20 of (ii), and the three characters 生, 能 and 業 in the sutra have dot-type t’o attached to them. I shall use these to explain the way of reading dot-type kugyŏl. (6) is the sentence reordered into Korean word order and with character-type kugyŏl added.

Dot-Type Kugyŏl

Figure 1   Dot-Type Kugyŏl



















‘causes all living beings to be able to surmount every obstacle, and …’

A single dot at mid-low position on the left side outside of 生 expresses the accusative particle -(ɨ)r. This shows that 一切衆生 is used contextually as the object. A single dot at top right within 能 expresses ti and shows that this 能 is interpreted as the adverb 能-ti. Three dots are clustered on 業. The dot at mid-low position on the left side outside it is the t’o that expresses accusative -(ɨ)r, as seen above. This shows that 諸障閼業 is read as an object. What looks like an exclamation mark made by a dot under a slant line at mid position on the right side outside 業 is the t’o expressing causative -hɐi. This corresponds to the function of 令 in the original Chinese, but one reads this t’o suffixed to the verbal 除滅 rather than read the 令. The dot at the bottom left outside 業 expresses -miɐ, a suffix conjoining two clauses.

There are many dot-type interpretive kugyŏl marks that are not yet deciphered. However, the single dot was used most. We know that in most cases double dots, lines or complex marks represent expanded versions of morphemes: in the Chubon Hwaŏm-gyŏng, for example, · at mid-low position on the left outside the Chinese character represents -r, while in the same position ·· represents -ter,: represents -holir, and ·| represents -holter, all of which include accusative -r.

Table 3.1 gives charts that show the various positions of dots relative to the imagined square occupied by a Chinese character.

Tabel 3.1   Dot-Type Kugyŏl Systems of Chumbon Hwaŏm-Gyŏng and Yugasaji-Ron

In the library at the temple Tōdaiji in Nara is an abridgement of the Hwaŏm-gyŏng, volumes 12–20, believed to have been brought to Japan before the 740s. This is currently being studied because it has Silla-language t’o produced with a stylus as well as annotations and all sorts of marks. There are characters and forms in this that do not appear in Silla idu or hyangga.

External materials include Chinese histories: Shiji (史記), Book of Han (漢書), Book of Later Han (後漢書), Records of Three Kingdoms (三國志), Book of Jin (晉書), History of the Southern Dynasties (南史), History of the Northern Dynasties (北史), Book of Sui (隋書), Book of Tang (舊唐書), New Book of Tang (新唐書). These fragmentarily show us OK lexis. Japanese histories are Kojiki (712) and Nihon shoki (720). These also present us with data on Three Kingdoms lexis, but mostly they preserve information on Paekche OK. Also, Japanese metal and stone inscriptions and old manuscripts present data on EOK. The Kyerim yusa was recorded by Song dynasty scholar Sun Mu, and within it 350 or so Koryŏ words are recorded under the heading of ‘dialect’.

3.3  Adaptation of Chinese Characters and Their Development

This notation developed together with the way that Chinese characters and Chinese were learnt and read. This development had already begun in the EOK period and developed into a rather precise notation by the LOK period.

In studying Chinese, Koreans read Chinese characters out as Chinese (ŭm [um] 音) or as semantically equivalent Korean words (hun [hwun] 訓). Hence, by borrowing the characters’ pronunciation and meaning, a method of representing Korean came to develop. Their usage came to differ depending on whether they used a character’s meaning as a semantogram or ignored the meaning and used only its pronunciation as a phonogram. Table 3.2 shows the division of ch’aja use into four types. The semantographic use is called ‘read’ (tok/dok 讀) in the sense that they are read just as they are, and the phonographic use is called ‘borrowed’ (ka/ga 假).

Table 3.2   Adaptation of Chinese Characters to Old Korean

Semantographic value (讀)

Phonographic value (假)

Sino-based (音):

ŭmdokcha [umtokca] (音讀字) Sino-based semantogram = Chinese loan morpheme pronunciation

ŭmgaja [umkaca] (音假字) Sino-based phonogram = phonogram based on Chinese

Koreo-based (訓):

hundokcha [hwuntokca] (訓讀字) Koreo-based semantogram = native Korean morpheme of similar meaning

hun’gaja [hwunkaca] (訓假字) Koreo-based phonogram = phonogram based on the native Korean morpheme’s pronunciation

All ch’aja notation, i.e. the ch’aja used to notate sentences and words, corresponds to one of these categories. Sometimes only ŭmdokcha and ŭmgaja are written, and sometimes only hundokcha and hun’gaja are written. However, the most widespread device was the structure semantogram + phonogram.

Ŭmgaja were used from the earliest period in which characters were used to represent Korean. These correspond to the jiajie category of character which the Chinese used to write loanwords, and proper nouns that appear in the Kwanggaet’o stela are almost only in ŭmgaja. The hun device is a native device that Koreans developed. It is already recognizable in the metal and stone inscriptions of the Three Kingdoms period, and the Silla official title 大舍 used in the Ch’ŏnjŏlli inscription (?525) is written in eighth-century texts as 韓舍. This shows that 大 was a hundokcha read as 韓 han. Also, in Ch’ŏnjŏlli supplementary inscription (?539) the Silla title 波珍干支 appears, where 波珍 means patul ‘sea’ and so 珍 is a hun’gaja read as tul. On the other hand, the Koguryŏ minister 淵蓋蘇文 is recorded in Nihon shoki as 伊梨柯須彌 irikasumi. This shows that in Koguryŏ the character 淵 was a hundokcha for iri. No Paekche texts are extant, but considering the point that Old Japanese literary activity was carried out by migrants from Paekche, one must see that using the hun device also developed in Paekche. These facts indicate that all the Three Kingdoms used notation that used the hun device.

There are proper ‘Chinese’ texts extant from the Three Kingdoms. Koguryŏ’s Tongsu epitaph (冬壽墓誌, 357) and the Kwanggaet’o stela, Paekche’s King Munyŏng tomb inscription (武寧王陵誌石銘, 525) and Sat’aek Chijŏk [Sathayk Cicek] inscription (砂宅智積碑銘, ?654) and Silla’s inscription of King Chinhŭng’s visit (眞興王巡狩碑銘, 540s or later), etc. are proper Chinese texts. In contrast, the Chungwŏn Koguryŏ inscription of Koguryŏ mixes Chinese and Korean word order and uses ch’aja such as 之 or 節 associated with later idu. The P’yŏngyang castle wall engravings include Korean linguistic elements. The Ǔnch’ŏn inscription excavated from Paekche King Munyŏng’s tomb is Korean in word order, while the sukse-ga mix Korean and Chinese word orders. The Naengsuri, Pongp’yŏng, Ch’ŏnjŏlli and Chŏksŏng inscriptions, etc. from Silla also mix Chinese and Korean word orders. However, the Imsin oath and Myŏnghwal-san inscriptions arrange Chinese, i.e. ŭmdokcha, entirely in Korean word order. These are early idu texts written only in semantograms. On the other hand, notation that arranges hundokcha entirely in Korean word order also developed in this period, as in the Ojak and Sinsŏng inscriptions.

Arranging only ŭmdokcha in Korean word order was used even in the MOK period, but proper idu and hyangch’al added t’o to early idu text. Although with idu and hyangch’al there is a difference between practical and literary texts, the idu often omits the representation of particles and endings, while hyangch’al rarely does so. T’o developed from kugyŏl but was applied to idu and to hyangch’al. The oldest extant text that uses t’o is the idu Kamsan Ami inscription (720), so it is assumed that kugyŏl had developed at an earlier period. In Samguk sagi and Samguk yusa, the late seventh/early eighth-century Confucianist Sŏl Ch’ong [Sel Chong] 薛聰 is said to have taught the Confucian scriptures and literature to the young by reading them in Korean. This shows that Sŏl Ch’ong created the interpretive kugyŏl of Confucian scriptures around the later seventh century. There are documents from which it is assumed that among Buddhist priests kugyŏl t’o had already been used in the time of Ǔisang [Uysang] 義湘 a generation earlier. From such facts we can assume that they interpreted Chinese texts in the Three Kingdoms period but t’o for this purpose had not developed in detail, and we can make a division before and after the unification of the Three Kingdoms in terms of the development of t’o and of ch’aja notation.

As most t’o writes with ŭmgaja or hun’gaja, the writing structure of idu or hyangch’al texts that used t’o appeared as semantographic + phonographic. This agrees with the structure of interpretive kugyŏl that used t’o. In (7) I analyze the types of characters used in the interpretive kugyŏl example (5) seen above.




hundokcha + hundokcha + ŭmgaja



ŭmdokcha + ŭmdokcha + ŭmgaja



ŭmdokcha + ŭmgaja + ŭmgaja



hundokcha + ŭmgaja + hundokcha + ŭmgaja



hundokcha + ŭmgaja + hun’gaja + ŭmgaja + ŭmgaja




hundokcha + ŭmgaja + hundokcha + ŭmgaja


This shows that the writing of a single word was structured semantogram + phonogram. The semantogram part was written the same as in the original Chinese and the phonogram part was written in t’o for interpretation. Among the t’o hundokcha like 爲 or 在 kiǝ were used but these represent the roots of auxiliary verbs and phonograms express their suffixes, so this is also a semantogram + phonogram structure. This semantogram + phonogram structure was also applied to the representation of hyangch’al virtually as it was. In (8) I give one interpretation of one line of the Hŏnhwa-ga given above (4) and analyze it.




hundokcha + ŭmgaja



hundokcha + ŭmgaja



hundokcha + ŭmgaja



hundokcha + ŭmgaja + ŭmgaja





hundokcha + hundokcha



hundokcha + hundokcha + ŭmgaja


This also writes the beginning of words as semantograms and their end as phonograms, agreeing in structure with that of interpretive kugyŏl. Such a structure was also applied to idu texts. Because idu had a lot of omission of its t’o we have to read it by supplying the missing t’o. When we analyze them together, we can call the development of reading devices for Chinese, or kugyŏl which was the device for reading Chinese, the womb for the development of ch’aja writing.

To write t’o in kugyŏl, sometimes ch’aja were used unchanged in their proper written form but in most cases unique kugyŏl forms of characters were used. If it was a simply written ch’aja the whole character was used, but mostly an abbreviated character was used by missing out strokes of the original character. Typically this abbreviated form involved taking the first part or the last part of a square- or grass-script character. Its value was determined according to whether it used the ŭm or the hun of the character. Kugyŏl characters used in some interpretive kugyŏl were generally about 55 characters, but depending on text we also find increased numbers of characters and different forms of the same character.

In Table 3.3 are presented the kugyŏl characters used among the interpretive kugyŏl of kugyŏl texts (ii) and (vi). Following the kugyŏl character is its reading, the character it is derived from, then a classification into whether it is based on square or grass script, whether it derives from the early or latter part of the full character, and whether it is ŭm or hun. We can tell by this the process by which kugyŏl characters came about.

Table 3.3   Interpretive Kugyŏl Characters


square, bottom, ŭm

grass, whole, ŭm


square, top, hun


square, bottom, ŭm


grass, bottom, hun


square, bottom, ŭm


square, whole, ŭm


grass, whole, hun

square, whole, ŭm


square, whole, hun


square, top, ŭm


square, bottom, ŭm


square, bottom, hun


square, bottom, ŭm

square, top, hun


grass, whole, hun


square, whole, ŭm


square, whole, ŭm




square, whole, ŭm


square, whole, ŭm


square, top, ŭm


grass, whole, ŭm


s/g, whole, hun


square, whole, hun?


square, top, hun?


square, whole, ŭm


square, whole, hun?


square, top, ŭm


grass, whole, hun


square, top, ŭm


grass, whole, ŭm


square, top, ŭm




grass, top, ŭm


square, whole, ŭm

square, whole, hun


square, bottom, ŭm


square, top, ŭm


square, top, ŭm


square, whole, hun


square, top, ŭm


square, top, ŭm

grass, bottom, ŭm


square, whole, hun


grass, top, hun

亠, 〻

s/g, top, ŭm

乊, ろ


s/g, top, hun?


square, whole, ŭm


square, top, ŭm


grass, top, hun


square, whole, hun


grass, whole, hun


grass, whole, ŭm


square, whole, ŭm

ノ, 乊


square, top, ŭm


square, whole, ŭm

grass, top, hun


grass, top, hun

Most of these kugyŏl characters are phonograms, but auxiliary verbs such as hɐ, kiɐ, sɐrp, i, hɐi, etc. and bound noun čahi are mostly used as hundokcha.

We can easily trace these abbreviated kugyŏl characters to their original Chinese characters, and most of the latter are used to write idu and kugyŏl t’o.

3.4  Early Old Korean

I regard Early Old Korean as a single language rather than as Koguryŏ, Paekche and Silla being separate languages. The three-language approach is based on the Koguryŏ toponyms in the geographical chapters of Samguk sagi. However, Koguryŏ ruled the territory corresponding to these toponyms for not even 100 years. This territory was not greatly separate from the territorial range of the Samhan period, which preceded the division into the Three Kingdoms, and I regard the latter’s language as a single ‘Han’-branch variety.

If we regard the Han-branch varieties as the ‘Southern branch’ and, from the fact that Koguryŏan succeeded the ‘Gojosŏn’ and ‘Puyŏ’ languages in the north, we regard it as the Northern branch, we can show how the two branches were related as follows. It is extremely rare to be able to decipher with Koguryŏ texts but the spelling 伊梨柯須彌 irikasumi recorded in Nihon shoki for 蓋蘇文 cited above shows that Koguryŏan was close to the Han branch. In other words, 伊梨 iri, which is the hun of 淵 (a Chinese character meaning ‘abyss’), corresponds to the 乙 ɔl of 奈乙 naɔl (‘well’) in Sillan and to the 於乙 äɔl of the Koguryŏan toponym 於乙買串, which in the Chinese version of its name (泉口縣) means ‘Spring Water Wellmouth county’. Also, the fact that 蓋kaisumun of 淵蓋蘇文 is also found as 蓋 shows that ‘metal’ (金) was 蘇 su in Koguryŏan. 蘇 su is the same as Sillan 素/省 su ‘metal’ and is connected with MK soj ‘iron’. Also the original name of 鉛城 (lit. ‘lead castle’), a castle north of the Yalu river, was 乃勿忽, allowing us to obtain the reading 乃勿 namül for ‘lead’, and this matches with Old Japanese namari ‘lead’ and with 那勿 namɐr given in the statement ‘lead is colloquially known as namɐf’ (鉛俗云那勿) in the Hyangyak kugŭppang which reflects the early Koryŏ language. I shall discuss more regarding the commonality of the Three Kingdoms’ varieties in the following section.

It is still too early to discuss the phonology of EOK, and I shall adopt the romanization used for MOK.

3.4.1  Lexis

I shall describe EOK lexis predominantly in connection with toponyms in Samguk sagi, with reference also to Samguk yusa and Chinese and Japanese texts. I list Koguryŏ (9), Silla (10) and Paekche (11) EOK words, followed by romanization and their meanings.

(9) The Koguryŏ dialect:

  • Nouns:ka ‘great man, noble’; 皆 kɔi ‘king’; .璃/類利/孺留/累利 # rüri ‘world, society’; 旦/呑/頓 tɔn ‘valley’; 內/奴/惱 nui ‘earth’; 達 tar ‘mountain’; 巴衣/波衣/波兮 # paɔi ‘rock’; 於乙 är ‘spring’; 勿 # mr ‘water’; 買 mɔi ‘water, river’; 古衣 # kui ‘swan’; 功木 # kumuk ‘bear’; 首 #siu ‘ox’; 烏斯含 usɔham ‘rabbit’; 夫斯/扶蘇 püsɔ ‘pine tree’; 蘇/素 # su ‘iron’; 乃勿 # namr ‘lead’; 也次 # iach. ‘mother’; 伊伐支 #ipärti ‘beside’; 鄒牟/朱蒙 čümu ‘skilled archer’; 忽次/古次 kučhɔ ‘mouth’; 斤乙 # kr ‘written character’; 吐 tu ‘dike’; 溝漊 kürü, 骨 kur, 忽 hur ‘district’. e.g. 奈兮忽 nahiäi-hur =白城郡 ‘White Fort district’, 沙伏忽 sapuk-hur = 赤城縣 ‘Red Fort county’, 買忽 mɔi-hur = 水城郡 ‘Water fort district’, 述爾忽縣 süni-hur = 峰城縣 ‘Peak Fort county’, etc.
  • Numbers:mir ‘three’; 于次 üčhɔ ‘five’; 難隱 nann ‘seven’; 德 täk ‘ten’.
  • Verbals: 多勿 tamr ‘reclaim territory’; 於斯 #äsɔ ‘lean’; 伯 paik ‘welcome’; 伊 i ‘enter’; 沙非斤/沙伏 sapuk ‘be red’; 位 # üi ‘resemble’; 屈火 # kürpür ‘bend’; 今勿 # kmr ‘be black’.

(10) The Silla dialect:

  • Nouns:kan, 邯 kam, 今 kɔm ‘leader’; 次次雄 # čɔčüng, 慈充 čɔčüng ‘king, shaman’; 居西干 käsäkan ‘king’; 尼師今 # nisɔkm ‘king, elder’; 吉士 kirsɔ ‘14th-rank title’; 儒禮˙弩禮・世理 # nüriäi ‘world’; 內 nɔi ‘world’; 只 ki ‘fortress’; 乙 r ‘well’; 勿 #mr ‘water’; 那 na ‘river’; 比自/比斯 # pis. ‘light’; 一利 # iri ‘star’; 素/省 # su ‘metal/iron’;首 # siu ‘ox’; 巨老 # käru ‘goose’; 毛 # mu ‘mosquito’; 三 # sam ‘hemp’; 朴 # pak ‘gourd’; 伊史/異斯 # isɔ ‘moss’; 柯半 # kapan ‘breeches’; 麻立 # marip ‘stake’; 嘉俳 # kapɔi ‘15th of 8th lunar month, mid-autumn festival’; 吐 tu ‘dike’; 洗 # siän ‘shoes’; 尼叱/尼師 # nis ‘tooth’; 閼知 # arti ‘infant’; 都 # tu ‘barbarians north of tumen river’; 牟羅/慕羅 mura ‘county’; 火 # pr, 伐 pär ‘district’.
  • Numbers:mir ‘three’.
  • Verbals:# mir ‘push’; 阿火 # apr ‘join’; 南 # nam ‘exceed’; 內 an ‘be good’; 異次/異處 # ičhɔ ‘hate’; 今勿 #kmr ‘become shady’; 吉 # kir ‘be long’; 居柒 # käčhr ‘be rough’;弗矩 # prkä ‘be red; bright’; 赤牙 # prkäm ‘be red’; 翰 # han ‘big’, 韓 # han ‘big’.

(11) The Paekche dialect:

  • Nouns: 鞬吉之 känkirči ‘king’; 於羅瑕 äraha ‘king’; 吉士/吉師 kirsɔ. ‘noble’; 二リム # nirim ‘lord’; ハシカシ # hasikasi ‘wife’; 所非 # supi ‘forest’; 斯摩/斯麻 # sɔma ‘island’; 仇池/仇知 # küti ‘metal/iron’; 珍惡 # turak ‘stone’; 只/己 ki ‘fort’; 牟羅 mura ‘village’; 忽 hur ‘fort’; 夫里 # püri ‘city, village’.
  • Verbals:# na ‘come out’; 近 # kn ‘big’; 沙 # sa ‘new’; 烏 # u ‘be lonely’; 所比 supi ‘be red’; 勿居 # mrkä ‘be clear’; 今勿 kmr ‘be black’.

Three Kingdoms lexical items we can identify are extremely few in number. We can see that the number of words in common between them is significant. In particular the number of words connected to MK is significant and this fact shows that they were all Han languages. I mark such words with # above.

The 忽 hur in Koguryŏ county and district place names was used not only in the Samhan territory but also mostly north of the Yalu river. In the ‘eastern barbarians’ chapter of the Wei volume of the History of the Three Kingdoms, it says that 溝漊 denotes a Koguryŏ castle, and if we refer to the fact that a Koguryŏ city name was 紇升 then we can assume the development 溝漊 kürü (> 骨 kur) > 忽 hur. This name was used as an administrative term meaning ‘district’ or ‘county’. This 忽 hur did not come from the corresponding area’s spoken language but was established uniformally in the centre. We can confirm that in the southern dialects 忽 hur was used in 慰禮國 in Silla and 伏 in Paekche, and so it reflected the indigenous language. 夫里 püri is an administrative term used in Paekche toponyms and its distribution is relatively wide, and 只/己 ki is also used several times in the sense of ‘fort’. The corresponding administrative term in Silla was 火 p /伐 pär. We trace these, 夫里 püri and 火 p.r, to the 卑離 piri in the Mahan language. On the other hand, 牟羅/慕羅 mura was widely used across Paekche, Silla and Kaya. This was used as a toponym for relatively large settlements in the south before the break-up of the Samhan. The corresponding northern word is 牟婁 murü, and it is assumed that Japanese mura ‘village’ is cognate.

Words corresponding to ‘king’ differ between the Three Kingdoms, and this is thought to derive from the different processes by which the dynasties were established. In contrast, we can see that the titles of officials or for ‘noble’ are in common.

The word for ‘rabbit’ is in common between Koguryŏ (usɔham) and Japanese (usagi). Though 吐 tu ‘dike’ does not occur in MK, it is found in both Koguryŏ and Silla. The number 密 mir ‘three’ is found in Koguryŏ and Silla, and 沙非斤/沙伏 sapuk ‘be red’ in Koguryŏ and 所比 supi ‘be red’ in Paekche show the same heritage, so we know that both are Han languages. 今勿 kmr ‘be black’, despite semantic differences between the Three Kingdoms, is shared between them. The dialects of the Three Kingdoms show differences, but these are small compared with what they have in common.

EOK has many words in common with Japanese. The following are often discussed and have a high credibility:




OJ tani



OJ kuma



OJ mira ‘leek’

ku i


EMJ kofu



OJ pe



OJ sima



OJ itu



OJ mura



OJ mure ‘mountain’

OJ mori ‘forest’



OJ tera



J nata ‘hatchet’



OJ mi



OJ nami ‘wave’



EMJ usagi



OJ kuti



OJ ki



OJ mi



OJ towo

kɔpɔr > kur


OJ kopori



OJ wata



OJ pata



OJ mu-

3.4.2  Grammar

In EOK the verb root had a very independent nature. In MK, the verb root could be made into an adverb, e.g. kót- ‘be like’: kót ‘like’, two verb roots could be combined into a compound verb, e.g. tat-ni- ‘run’+‘go’, verb roots could modify nouns like pswus ‘rub’ in pswustol ‘whetstone’, and verb roots could occur as nouns, e.g. sin- ‘put on [shoes]’: sin ‘shoes’ or póy- ‘be pregnant’: póy ‘belly’. We know the verb root’s independent nature was far greater in EOK. Koguryŏ 伯 paik ‘welcome’, 伊 i ‘enter’, 多勿 tamur ‘reclaim territory’, 於斯 äsɔ ‘yellow’, 今勿 kmr ‘black’; Silla 居柒 käčir ‘rough’, 吉 kir ‘long’, 密 mir ‘push’, 阿火 apr ‘join’, 異次/異處 ‘hate’; and Paekche 烏 u ‘lonely’, 所比 supi ‘red’, 奈 na ‘go out’ indicate that this sort of syntax was found in the Three Kingdoms. 南內 naman, renamed in Chinese 餘善縣 (lit. ‘exceedingly good county’), shows a compound of the roots nam ‘leave’ and a ‘good’. Even though we do not have many Three Kingdoms lexemes, there is a lot of this sort of commonality.

‘Samhan’ or ‘the three Hans’ is the coverterm for the 78 states of the Mahan 馬韓 (west), Chinhan [Cinhan] 辰韓 (east) and Pyŏnhan [Pyenhan] 弁韓 (south) confederacies. 韓 han is formed from the verb ha ‘great’ and the nominalizer -n and has come to have the meaning ‘chieftain’. This han probably designated the chiefs of each of the 78 tribes, and developed as a nation name when the tribes formed confederacies and became the Samhan. The 韓 han in Silla and the 翰 han of Paekche toponyms are also cognate. Another form is 干 kan and it is thought that it diverged from the same word and the two forms co-existed very early on. With the growth and development of tribal societies we find the development of titles for ‘king’ by adding modifiers to this morpheme or such variants as 邯 kam or 今 km, giving 居西, 居瑟, 尼師 or 麻立. Also it developed the meaning ‘noble’ and was used in Silla for such official ranks as 伊伐 ipärkan , 波珍 patorkan , 阿 akan or 一吉 irkirkan , as well as in the title of nine Karak officials, 刀 tokan . 近 kn in the names of the Paekche kings Kŭnch’ogo [Kunchoko] 肖古, Kŭngusu [Kunkwuswu] 仇首 and Kŭngaeru [Kunkaylwu] 蓋婁 is also a variant of kan, with nominalizer -n added to the root 幹 k ‘great’. The 鞬 kän ‘great’ of 吉支 is also a variant. Some think that the 加 ka ‘noble’ used in Koguryŏ titles derives from the root underlying kan, and the 瑕 ha of the Paekche royal title 於羅 is also of the same root.

From the above we can confirm the existence of nominalizer -n, and the nominalizer -ɔn can be confirmed by analyzing the name of Silla’s founder 弗矩內/赫居世 as prk ‘red’ + -ɔn nominalizer + ɔi.

In Three Kingdoms inscriptions the respectful suffix 智/知/支 ti is widely used. It is used in the Naengsuri inscription, as 智 ti in people’s names (斯夫, 乃, 斯德, 子宿, 爾夫) or as 支 ti in titles (阿干, 居伐干, 壹干) or the personal names of lower-ranking officials (斯申, 蘇那, 叟, 須). On the other hand, 智 is used in titles such as 壹今, showing 支 and 智 were homophones. In the Pongp’yŏng inscription 智 is used in personal names and 支 is used consistently in the title 干, but 智 is used in names of low-ranking officials such as 邪足, 小舍帝, 吉支, 小烏帝. In the Sinsŏng inscription 知 ti is used, but in the name of a local man 之 is also found. 智/支/知/之 ti was used in this way as a Silla EOK honorific suffix, but it is the same root as 智 ti in the Mahan chieftain 臣, 支 ti in Paekche 揵吉 above, and 支 ti in Koguryŏ 莫離.

It was noted above that EOK idu texts arrange ŭmdokcha in Korean word order. Koguryŏ’s Chungwŏn inscription mixes Korean and Chinese word orders. (13) is from the Korean word-order part:
















‘When he commanded that all ranks come and receive their robes, they knelt before the camp.’

The sequence 上下衣服來受 here is Korean SOV word order. The declarative suffix 之 and the verb 敎 are used in this inscription and are connected with Silla idu texts. These are ŭmdokcha but must be viewed as underlyingly spoken Koguryŏ.

The inscription of the queen’s silver bracelets excavated in King Munyŏng’s tumulus is typical of Paekche idu (14).














1/24 oz.


‘Tari made this in the second month of 520. It is for the king’s wife. It is 9.59 Chinese ounces.

This is also entirely in Korean word order. The sentence-final 耳 is a lexical particle and umdokcha used to correspond to the Paekche declarative suffix.

The Imsin oath inscription is a typical Silla idu text, ordering all the text entirely in Korean word order. In contrast is the Sinsŏng inscription (15), an idu text which uses hundokcha.


























*il *-ta








‘When they built Namsan Sinsong, if it is built in accordance with the regulation and if it crumbles within three years, report it to the king to be punished, and the builders accordingly took oaths.’

The text concerns those who undertook the building work on the castle swearing an oath to the king.

The grammatical morphemes that appear in idu texts of the period are identified as follows. The particles 以 lo (instrumental) and 者 n (topic) occur, and 中 is used to represent spoken a or k i. 之 as a suffix is used to represent declarative -ta. 在 kiä is an auxiliary verb expressing continuing action, and 節 tiüi is a conjunctive adverb. These were inherited into later idu.

3.5  Mid Old Korean

3.5.1  Phonology

T’o developed in the MOK period and there are more texts than EOK. Yet it is only after the 1446 promulgation of han’gŭl [hankul] that a full description of Korean becomes possible. OK research is based predominantly on tracing back in time from fifteenth-century MK texts, applied more vigorously with phonology. The MOK consonants had two sets of obstruents: plain (p t č k) and aspirate (?ph ?th čh kh). However, aspirates’ functional load was extremely low, and so it is dubious whether ph th existed. Fricatives (s h) and nasals (m n ng) were the same as in MK. Liquids are thought to have distinguished r (乙) vs l (尸). MK z did not exist in MOK, and the phonemic status of β is uncertain even in MK, and we cannot trace it back into OK. The fifteenth century had a seven-vowel system with vowel harmony, so we presume these applied more strictly in OK. Vowel harmony is not reflected in the texts, so this is hard to verify. Some believe that a systematic obeying of vowel harmony meant an eight-vowel system, but the basis for this is weak. Current research regards it as least risky to trace back and adapt the MK seven-vowel system, which I also accept here (Lee 1972: 72; Lee and Ramsey 2011: 67; for a recent alternative, see Whitman, this volume):


ü (= Mk wu)

u (= Mk o)

(= Mk u)

ɔ (= Mk ó)

ä (= Mk e)


We cannot say that the MK semi-vowels had developed. I represent MK semi-vowel + vowel sequences as two vowels, i.e. ia, iä, iu, iü, ua, üä, etc. A phonological process that we can identify in this period is monophthongization of diphthongs, e.g. the use of 令 (liäng) – thought to be used for * liä – in its abbreviated form 仒 for li in the twelfth century, or the development of nominative 亦 iäk > 弋 ik > 〢 i (see 3.6.2.(ii.)). The writing of one king’s name as 儒禮, 弩禮, 儒理 or 世里 shows monophthongization of 儒 niü > 弩 nu and 禮 liäi > liä > 理/里 li. ɔ was weak even in MOK, so such contractions as ɔ+i > i or ɔ+iä > occur.

3.5.2  Grammar

I shall divide MOK grammar between that in the hyangga of Samguk yusa and that in idu texts. The writing of grammatical morphemes in hyangga is largely full, but as they were composed at a different time to when they were committed to writing, they are difficult to date accurately. There have been many decipherment attempts, but not only are there so few texts but we also do not understand OK grammar enough, so decipherments have conflicted with each other. Following the recent discovery and decipherment of interpretive kugyŏl and ancient idu texts, we are now able to close in on hyangga grammar, but parts still remain undeciphered, so we must bear this in mind when we describe the grammar. Idu texts’ dates are certain and they are accurately decipherable, but the writing of grammatical morphemes is cursory and we certainly do not have enough of these texts. We can describe the grammatical system to an extent by combining these two types of text, but gaps are inevitable. In Tables 3.4 to 3.9 I present morphemes from hyangga on the left and from idu on the right. Superscripta = fusion with a stem-final sound, b = fusion with the bound noun ‘fact’, and c = special cases.

(i) Noun-following particles (Table 3.4). I shall start with case particles in hyangga. Nominative -i was used as in 雪是 snow-NOM (H-iv), 爲賜尸 知 do-HON-NMR fact.NOM = ‘the fact that X does’ (H-iii). Genitive i was used of sentient beings (心未 際叱 heart.m-GEN edge.s = ‘edge of the heart’ (H-iv)), -s of non-sentients, and -l of direction-nouns (東尸 汀叱 east-GEN shore.s (H.xii)). 肹 -hir was used as accusative. It is unclear if the h characteristic of MOK hyangch’al in the accusative and locative was actually pronounced, and does not appear in LOK. The locative has - i and -a types (心未 masɔm-i heart.m-LOC, 人米 saram-i person.m-LOC, 前乃 front-LOC), as well as combinations of -a + -ki/hi. The connective was used as in 邊也 藪耶 shore-CONN woods-CONN ‘the shore and the woods’ (H-xii). The writing of case particles was cursory in idu: neither nominative nor accusative are written, and the genitive only occurs in 經之 成. 法者 sutra-GEN make-CONSID[-NMR] way-TOP ‘the way in which one makes the sutra’ (example (2)).

Table 3.4   Mid Old Korean Case and Discourse Particles



a. Case particles:


是 -i, 知 b t-i



矣 - i, 未 a m-i


叱 -s

之 -s


肹 -hir



衣/矣 - i, 希 -hi, 中 ki, 未/米 a m-i

中 - i


良 -a, 乃 a n-a


惡希 -a-hi, 良中 -a-ki


以 -lu

以 -lu


也…耶/也 -ia…-ia



良 -a, 也 -ia


下 -ha



之叱 - i-s, 阿叱 -a-s

b. Discourse particles:


隱/焉 - n

者 - n

呑/等/等焉 b t-n


置 -tu


那 (乃) -(i)na


沙 -sa


爾/厼 - m

厼 - m

Topic 隱 and rarely 焉 were used in hyangga, some believing the difference being vowel harmony. There was no special conditional converb in OK and the topic marker fulfilled this function. Emphatic -sa either directly attached to a noun or followed another particle. Distributive 爾 -km is used as in 八切爾 eight-cut-DISTR ‘eight cuts each’ (H-xii), or 經心內中 一收 舍利厼入內如 sutra.heart inside-LOC one śarīra-DISTR insert-CONSID-DEC (MT-iii) ‘inserts one śarīra inside each sutra’.

(ii) Verb enders (Table 3.5). Declarative -ta was used as in 人是 有叱多 person-NOM exist.s-DEC ‘there’s someone’ (H-xii), 植內之 plant-CONSID-DEC ‘planted them’ (MT-iv), 妹者 敬信大王嬭在也 younger. sister-TOP Kyŏngsin-king maternal.aunt[-COP]-CONT-DEC ‘his sister is King Kyŏngsin’s aunt’ (MT-v). 如 emerged in the mid-eighth century and was used till the end of the nineteenth, and 也, though rare, had been used since EOK. 矣 was mostly used for a converb -tɔi, but there are MOK examples for -ta: 鐘 成內矣 bell[-ACC] make-CONF-DEC ‘they (had) made the bell’ (MT-x). -čiäi was used for an objective statement, but in MOK idu -čɔi was also used for a wish: 後代 追愛 人者 此 善 助在哉 posterity cherish people-TOP this virtue assist-CONT-DEC ‘may people who cherish this in later times assist in this virtue’ (MT-i). Interrogative -ku and -ka were both used only as polar interrogatives, unlike the LOK polar -ka vs wh -ku distinction. 民焉 狂尸恨 阿孩古 爲賜尸 知… people-TOP foolish child-QUOT do/say-HON-NMR fact-NOM ‘that the people are called foolish children’ (H-iii) illustrates quotative -ko.

Table 3.5   Mid Old Korean Verb Enders



a. Sentence-enders:


如/多 -ta

之/如/也/矣 -ta

Declarative 2:

齊 -čiäi

哉 -čɔi


古/遣/故 -ku,去 -ka


也/邪 -ia, 丁b-t-iä

下 -ha, 下是 -hai, 多羅 -tara


叱多 -sta


古 -ku


羅 -ra, 羅良 -ma


賜立 -sa/-siä

b. Converbs:


良 -a, 可 a k-a, 也 a i-a


惡只 -ak


古/遣 -ku


古只/遣只 -kuk, 古音 -kom


音 -m


米 -m-ɔi


旀 -miä

旀 -miä


乃 -na


維 -ra


哉 -čɔi


欲 -kua


厼 -km


矣 -tɔi

c. Nominalizers:

Established fact:

隱 -n

u-n, 者 -(/ɔ)n


尸 -l, 乎尸 u-l

A few examples of converbs are: 折叱可 käsk-a pick-INF ‘pick and’ (H-ii), 執音 乎 手 čap-ɔm hɔ-n sɔn hold-DUR-AUX-NMR hand ‘the hand holding it’ (H-ii), 此矣 有阿米 here-LOC exist-CONF-CAUSE ‘because X is here’ (H-xi), 望阿乃 look.at-CONF-ADV ‘although I stare at it’ (H-xiii), 二于萬隱 吾羅 two lack-NMR I[-COP]-CAUSAL ‘because it is I, who hasn’t even two of them’ (H-vii), 若 大小便爲哉…若 食喫哉 爲者 if defecation.urination-AUX-PARA… if eat.drink-PARA do-TOP ‘if X defecates and urinates or eats and drinks’ (MT-iii).

Nominalizer of an established fact, -n, most commonly modifies a noun, e.g 去隱 春 pass-NMR spring ‘the spring which has passed’ (H-i), but also functions like MK -ni: 白乎隱 花良 汝隱 say(HUM)-INTENT-NMR flower-VOC you-TOP ‘I say to you, O you flowers’ (H-x). It also occurs on the scope of the negative anti in nominal sentences. In idu there is no ch’aja that expresses the nominalizer by itself, and so we have to reinstate it according to context. The prospective nominalizer -l expressing factual or future meaning is mostly used to modify nouns, but in for example 愛尸 知古如 love-NMR know-ASSERT-DEC ‘they should know that he loves them’ (H-iii) it is the object of a verb.

(iii) Auxiliary roots (Table 3.6). I use ‘auxiliary root’ for affixes that when attached to verb roots form new roots. They have been called ‘pre-final suffixes’, but this is not as appropriate in OK because they are also used as enders, and ‘pre-final suffixes’ are by nature uncharacteristic of agglutinative languages. In previous research the causative and passive have been regarded as auxiliary roots, but here I only deal with non-derivational forms.

Table 3.6   Mid Old Korean Auxiliary Roots




賜/史 -

賜 -


烏/乎/屋 -u

乎 -u


阿/惡 -a, 於 -ä, 邪 -ia

去 -ä


古/遣 -ko


音叱 -ms


里/理 -ri

Past tense:

如 -, 頓 -tä-n

Present tense:

奴 -nu

飛 -


立 -siä


省 -su


將來 -riä

Some view that the intentional mood -u expresses ‘the speaker’s psychological proximity’ (Kŏno 1996). -a/-ä/-ia are used as the confirmatory mood; in idu there are examples of 去 - used word-finally: 法界 有情 皆 佛道中 到內去 誓內 universe sentient.beings all Buddhism-LOC reach-CONSID-CONF (nirɔ-a-kä) swear-CONSID ‘he swears that all sentient beings of the universe will certainly arrive at Buddhism’ (MT-ix). This illustrates the independence of roots. -ku combines the speaker’s desire in the confirmatory mood: 白遣賜立 say(HUM)-ASSERT-HON-IMP ‘make sure you tell [Amitābha]’ (H-ix). Essential -ms expresses the sense ‘that is how it should be’. -ri expresses factual or conjectural meaning, and intention when the speaker is the subject: 去賜里遣 go-HON-PROSP-Q ‘[Moon,] will you go?’ (H-ix), 獻乎理音如 give-INTENT-PROSP-ESSEN-DEC (above). -nu/-nɔ and - express present and past tense respectively: 去奴隱 處 go-PRES-NMR place ‘the place X goes to’ (H-xi), 仰頓隱 面 look.at-PAST-NMR face ‘the face I used to look at’ (H-xiii). -siä is a fusion of honorific - and confirmatory -ä: 陪立羅良 accompany-HON-CONF-IMP ‘please accompany him’ (H-x).

(iv) Derivative suffixes (Table 3.7). Verb noun -m continues into later Korean: 誓音 swear-VN ‘oath’ (H-ix) > MK tati-m > taci-m, 岳音 urɔ-m climb-VN ‘mountain’ (H-xii). The adverbalizer -ü/-u in hyangga and idu differs from MK in that it can have an argument: 白雲音 逐于 white cloud .m follow-ADV ‘following white clouds’ <H-iv>.

Table 3.7   Mid Old Korean Derivative Suffixes



Verb noun:

音 -m


于 -ü, 乎 -u

(于) -ü

(v) Auxiliary verbs (Table 3.8). OK auxiliary verbs often link directly to verb roots without a linking suffix. - is an auxiliary verb ‘do’ that covers all verbals: 狂尸恨 foolish.l-AUX-NMR ‘foolish’ (H-iii). - is widely added to Chinese verbs to Koreanize them, e.g. Chinese 供養 ‘perform a memorial service’ > 供養爲 perform.memorial. service-AUX[-NMR] (MT-iii). Considerate 內 -a relates to performing an action towards someone senior, and accepts the doer’s action as appropriate, and came also to be used for humble mood. The source verb 內 a means ‘be good, appropriate’. -kisɔ is an honorific auxiliary, which developed from the meaning of a command issued by someone senior or ‘instruction’ so was much more honorific than -. -ti is assumed to derive from ‘fact’ and -i, meaning ‘to be certainly the case’. Humble -sɔrp appears in (MI-iii) from the 860s and began to take over the humble function of -a. 在 -kiä < kiä ‘exist; put’ expresses progressive and resultative. In the MK period is- ‘exist’ assumes this function, and -kiä now remains fossilized in honorific kyesi- ‘exist’ only.

Table 3.8   Mid Old Korean Auxiliary Verbs




爲 -, 恨 c -hɔ-n, 好 c -h-u

爲 -

Considerate (Humble):

內 -a

內 -a


敎 -kisɔ (只 -ki


支/支 -ti, 多支(支) -tati


令只 -siki *


白 -sɔrn


在 -kiä


*  There is also the view that this was read as the causative of hɔ, hɔ-ki > hɔi

(vi) Copula. The copula is mostly 是 -i, but in Silla hyangga there are also examples of 以 -i. In idu 是 -i was used with largely the same function as in Modern Korean, but differed in that it could combine with the continuative: 內物是在之 inside thing-COP-CONT-DEC ‘it is something remaining inside’ (MT-vii). -i was often omitted in writing.

(vii) Quasi-grammatical forms (Table 3.9). There are forms that perform an intermediate function between content words and grammatical forms. They are grammaticalized content words.

Table 3.9   Mid Old Korean Quasi-Grammatical Forms




等/冬 , 知 c t-i

‘across (to)’:

念丁 nämtiä

‘when, after’:




‘make into’:


‘starting from’:

初/兀 pirs



The bound noun 等 of the factual mood was widely used with the meaning of ‘certainly the case’, which is the meaning of declarative -ta. nämtiä (cf. MK nem- > nem- ‘go over’) occurred in 西方念丁 去賜里遣 west-across go-HON-PROSP-Q (H-ix), Chinese loan 第 tiäi ‘in sequence’ grammaticalized to ‘when, after’. The verb ki, which since vanished, is assumed to have meant ‘instruct’ from the fact that it was used in -kisɔ, but at present it is difficult to be accurate. sam originally meant ‘make’, but was grammaticalized as ‘for the purpose of’. pir s was a verb root and expressed the meaning of ‘begin’ in constuctions such as ‘beginning from X till Y’: 二月 十二日 元 四月 十三日 此 間中 了治內之 2nd.month 12th.day begin 4th.month 13th.day this period-LOC complete-CONSID-DEC ‘they completed [the repairs] in the period from the 12th of the 2nd month to the 13th of the 4th’ (MT-viii).

(viii) Negation. As in LOK nominal negation and verbal negation differed: 不喩/安攴 anti was the nominal negator. In 不喩 慙肹伊賜等 NEG be.ashamed-HON-NMR fact-TOP ‘if you do not feel ashamed’ (H-ii), anti negates the nominalizer n. Preverbal 不冬 antɔr was the verbal negator and 毛冬 mutɔr the potential negator.

(ix) Conjunctions. Converbs function to link clauses in Korean, and it is thought that the use of conjunctions in hyangga as a part of speech was due to Chinese influence: 但 tamɔn ‘however’ and 唯只 očik ‘only’ (H-xiv).

3.5.3  Lexis

Because ch’aja writing used semantograms to express the content part of a word, it usually does not show us its pronunciation. Just occasionally a word stem is written in phonograms, e.g. 阿孩 ahɔi ‘child’, 於冬是 ätɔri ‘where’, 爾處米 k mčh mɔi ‘because X stops’, 是史 is ‘exist’, 毛冬 mutɔr ‘cannot’, deciphered based on context and corroboration from other materials. However, it is extremely difficult to decipher words written only in phonograms. Words such as 阿冬音, 乃叱好支, 窟理叱 and 乃乎尸 remain hard to decipher.

It is easy to understand when the ch’aja writing is of the structure semantogram + phonogram, common in hyangga, and we can work out a word’s form based on the final phonogram and on its MK form. For example, because of the addition of phonogram 音 m to 心 ‘heart’, we can work out that 心音 is the OK cognate mɔsɔm of MK mózóm ‘heart’. The following are OK words identified in this way.


  • Nouns: 雲音 kürɔm ‘cloud’ (H-iv), 夜音 pam ‘night’ (H-i), 誓音 tatim ‘oath’ (H-ix), 憂音 sir m ‘worry’ (H-i), 道尸 kil ‘road’ (H-i), 二尸 tüp l ‘two’ (H-iv), 秋察尸 kasɔl ‘autumn’ (H-xiii), 皃史 č s ‘looks’ (H-iv), 栢史 čas ‘pine’ (H-xiii), 母史 äs ‘mother’ (H-iii), 際叱 kɔs ‘edge’ (H-iv), 兵物叱 piängkas ‘weapon’ (H-xiv), 物叱 kas ‘thing’ (H-xiii), 枝次 kač ‘branch’ (H-iv), 蓬次 tapɔč ‘mugwort’ (H-i), 一等 tɔn ‘one’ (H-xi), 千隱 čm n ‘1,000’ (H-iv), 紫布 tɔrpɔi ‘purple, azalea’ (H-ii), 岩乎 pahɔ ‘rock’ (H-ii), 國惡 nara ‘country’ (H-iii), 沙矣 mɔrɔi ‘sand’ (H-xiii), 風未 pɔrɔm-ai ‘wind-LOC’ (H-xi), 前乃 čiän-a ‘front-LOC’ (H-ix)
  • Verbs: 有叱 (H-i) is ‘exist’, 折叱可 sk-a ‘pluck-INF’ (H-ii), 修叱如良 task-a ‘wash/polish-INF’ (H-viii), 待是 kitɔri ‘wait’ (H-xi), 持以 tini ‘hold’ (H-iv), 使以 pri ‘use’ (H-x), 遣知 kiti ‘leave’ (H-iv), 慕理尸 k ri-l ‘long.for-NMR’ (H-i), 慙肹伊 püskri ‘be shy’ (H-ii), 改衣 kɔtɔi ‘mend/cure’ (H-xiii), 望良 ra ‘desire’ (H-xii), 哀反 srp-n ‘sad-NMR’ (H-viii), 白反 sɔrpɔ-n ‘say(HUM)-NMR’ (H-xii), 爾處 kümčhiä ‘stop’ (H-iv), 集刀 ‘gather’ (H-ix), 直等隱 tɔ-n ‘straight-NMR’ (H-x)

The following words had a word-medial r, which fell out by MK.


  • 川理 nari ‘stream’ (H-iv) > Mk nay > nay
  • 倭理 iäri ‘comet’ (H-xii) > Mk yey
  • 世理 ri ‘world’ (H-xii) > Mk nwi, but nuri
  • 舊理 niäri ‘old times’ (H-xii) > Mk nyey > yey
  • 邀里白 muri- sɔrp ‘to worship’ (P-i) > Mk moy-zóp-> mosi-

Separate from these are examples where syllables or vowels later dropped out. MK tól ‘moon’ appears in hyangga (H-iv) as 月羅理 rar-i moon-NOM, the middle ra dropping out in MK. MK mom ‘body’ appears in LOK (P-i) as 身萬隱 muma-n body-TOP, the word-final a falling out in MK. On the other hand, 星利 piäli ‘star’ (H-xii) is the same as MK pjel ‘star’, the word-final i dropping out. Added final phonograms appear in MOK idu only in 令只 siki. Added final phonograms are useful in working out OK word forms, but it is still hard to work out the forms of 掌音 ‘palm of hand’ and 花判 ‘flower’ (H-iv) despite the final phonograms, as there are no cognates in MK.

Loanwords from Chinese entered by various routes, but central was importation via Confucian and Buddhist scriptures. The increase in Chinese words was so great that King Kyŏngdŏk [Kyengtek] tried changing the names of the provinces, districts and counties of Silla to Chinese names in the mid-eighth century. Amongst Chinese loanwords are made-in-Korea Sino-Korean words. Characteristic of them is the arrangement of Chinese characters in Korean word order. 楮皮肹 <paper.mulberry + bark + peel> ‘peeler of paper mulberry bark’, 肹皮練 <peel + bark + treat> ‘person who treats the peeled bark’, 紙作人 <paper + make + person> ‘person who makes the paper’, and 苦離樂得 <suffering + avoid + joy + obtain> ‘to avoid suffering and obtain joy’ are examples, but they are coined according to the earliest idu grammar which arranges ŭmdokcha in Korean word order. Administrative terms such as 古有人 ‘old resident’, 烟受有畓 ‘paddy field allotted to a category of household’ are used in (MT-iv), but these too are Sino-Korean coinages with the characters arranged in Korean word order.

There are even Chinese characters invented in Korea. The 畓 ‘paddy field’ above was formed by combining 水 <water> and 田 <field>, and is a made-in-Silla character. 太 ‘bean’ is also thought to have been created at this period, writing 大 <big> and 豆 <bean> in sequence vertically, changing 豆 to grass script and then reducing it to a dot.

3.6  Late Old Korean

We can characterize LOK texts by the fact that they have detailed t’o writing. This means that we can describe this period’s language in more detail. T’o became detailed in both idu and hyangch’al, and we confirmed the large number of t’o anew through the discovery of interpretive kugyŏl. Not only was the t’o of interpretive kugyŏl (hereafter ‘kugyŏl’ for short) written relatively complete, we can analyze the linguistic facts of it by comparison with the contents of the Chinese text, which allows us to describe relatively accurately the nature of OK. The Kyerim yusa and Hyangyak kugŭppang are also important materials regarding LOK.

3.6.1  Phonology

The LOK phoneme system is identified by comparing the sounds of the Song dynasty Chinese characters in which Korean words are transcribed in Kyerim yusa with MK (Kang 1980). Plain consonants are p t č k as in MOK; aspirates ph th čh kh, which still had low functional load; fricatives s h and now also z; nasals m n ng. I regard liquids to have distinguished r and l. z had a low functional load, occurring only between voiced sounds. MK β still did not exist. 只 knrlmps were used as syllable codas. ng was used as a coda in Sino-Korean, but as it had no letter of its own this cannot be confirmed. There appears to have been a seven-vowel system:


ɨ (= MK u)

u (= MK wu)

ǝ (= MK e)

ɐ (= MK ó)

o (= MK o)


3.6.2  Grammar

In Tables 3.10 to 3.14 I present the grammatical forms of LOK hyangga (the eleven Pohyǒn sibwǒn-ga, plus the Sǒdong-yo and Ch’ǒyong-ga), idu and kugyŏl.

(i) Noun-following particles (Table 3.10). The nominative had -i and -ik. -ik is used once in Kyunyŏjŏn, i.e. 身靡只 mom-ik body-NOM (P-viii), and is rarely used even in kugyŏl. We may hypothesize a development 亦 -iǝk > 弋 -ik > 伊 -i. The genitive distinguishes sentient and non-sentient possessors, but there are exceptions. We note in kugyŏl, apart from -ɨi and -s, the use of -l as an honorific genitive, especially but not only with Buddhist deities: 菩薩尸法乚 bodhisattva-GEN law-ACC ‘the law of the bodhisattva’ (K-ii). Sometimes in LOK, the accusative was used instead of the locative, and the Chinese locative preposition 於 was interpreted as the accusative: [於]法乚歸丷ヿ乚 law-ACC converting-AUX-NMR-INST ‘by converting to Buddhism’ (K-iii). The -a locative was used generally, while the -ɨi tended to be used for sentient beings. Combining them gave -a(k)ɨi > MK -ay/-ey. There are examples of the locative used as a dative. -iǝkɨi often followed the characters 時 ‘time’ and 第 ‘sequence’. The semantic range of the instrumental was broad, e.g. requisite, source, instrument, means or cause. The connective (kugyŏl only) is normally -iǝ… -iǝ, rarely -koa… -koa. -koa… -koa is linked to a following noun with -s, and -iǝ… -iǝ with : 樂亠利亠衰亠丷尸 joy-CONN profit-CONN weakness-CONN do-NMR ‘being joyful, profiting and being weak’ (K-iii).

Table 3.10   Late Old Korean Case and Discourse Particles




a. Case particles:


伊 -i, -ik

iǝ k, 弋只 -ik

II -i, 弋 -ik



衣 -ɨi

矣 -ɨi

ラ -ɨi, 尸 -l


叱 -s

之 -s

七 -s, 尸 -l


乙/肹 -ɨr

乙 -ɨr

し -ɨr



良/阿/悪 -a

良/中 -akɨi, 亦中 -iǝkɨi

-a, 十 -akɨi, 亠十 -iǝkɨi


衣/矣 -ɨi, 中 -kɨi

中 -kɨi

ラ -ɨi, ラ十 -ɨikɨi, 十 -kɨi


良衣,阿希 -aɨi, 悪中 akɨi


留/乙留 -lo

以/乙以 -lo



人-koa…人 -koa, 亠 -…亠 -



良 -a



也 -ia, 亦 -

亠 -


下 -ha

下 -ha


阿叱 -a-s,

人 -koa…人七 -koa-s

悪之叱 -aǝi-s, 留叱 -lo-s

b. Discourse particles:


隱/焉- (ɨ/ɐ)n, 恨 a

c t-a-n,

ヿ -n,

hɐ-n, 等 b tɐ-n

乙良 c r-a-n

丨十ヿ c -t-akɨi-n


沙 -sa

沙 -sa

氵 -sa


置 -tu, 刀 -to

刀 -to, 斗 -tu


每如如 -mata,

每如 -mata

馬洛 -marak


亇 -ma


火(七) -pɐs


巳只 -iki


()ホ -akɨm


乃- na

In idu t-a-n is a fusion of the bound noun + locative + topic (as also is kugyŏl t-akɨi-n), and -ra-n is accusative -r + -a (? locative) + topic -n. The topic particle was also used for the conditional as in EOK. -sa was emphatic, but also conjunctively for Chinese 乃 ‘then, indeed’, attached to nouns, locative particles, adverbs, and honorific auxiliary roots as i-si-sa COP-HON-EMPH. -mata ‘every’ developed from an adverb meaning ‘universally’ examples of the adverb occurring in kugyŏl. Deliminative pɐs occurs in kugyŏl, but rarely. Distributive -akɨm and alternative -ina were also rare. -iki (idu only) expressed a temporal endpoint.

(ii) Verb enders (Table 3.11). LOK verb roots retained a high degree of independence. In Kyerim yusa, the 鋪 po of “讀書曰乞鋪” ‘reading books is called kɨr po’ is the root of MK po- ‘to see’, the 索 so of “射曰活索” ‘shooting is called hoar so’ is the root of MK sso-/pso- ‘to shoot’, the 耻 ti of “雪下曰嫩耻” ‘snowing is called nun ti’ is the root of MK ci- ‘to fall’, and the 打里 tari of “煖酒曰蘇孛打里” ‘heating rice-wine is called sopɐr tari’ is the root of MK tali- ‘to boil, infuse’. Such independence led to great numbers of root + root compounds, and agglutination especially with auxiliary verbs in threes, fours or more. Verb enders, however, could not be used without a verb root.

Table 3.11   Late Old Korean Verb Enders




a. Sentence enders:

Declarative 1:

如/多 -ta

之/如 -ta

丨 -ta

Declarative 2:

也/耶 -ia

亦 -, 丁 b -t-iǝ, 齊 -čiǝi

亠/〻 -, 丁 b -t-iǝ, -čiǝi


良羅 -ara, 兮 -hiǝi



ㄔ -siǝ

Interrogative (polar):

去 -


Interrogative (wh):

古 -ko, 焉古 -ǝn.ko, 叱過 -s.ko.a

口 -ko

Interrogative (rhetor.):

下呂/下里 -hari


去良 -kǝa, 齊/制 -čiǝi

立 -siǝ



(Kyerim yusa) 良 -ra

立 -ɐsiǝ


少時/受勞 siosiǝ

b. Converbs:


良 -a, 可 a k-a

良/ -a

-a, 下 -ha


良只/悪只 -ak, 良厼 -akɨm

良尒 -akɨm

ハ -ak, ホ -akɨm


古/遣 -ko, 昆 -ko-n

遣 -ko

口 -ko


旀 -miǝ

彌/旀/ -miǝ

ち、 -miǝ


乃/奈 -na

乃 -ma


而也 -mariiǝ

馬於隱 -marǝm


如可 -taka


乙 -r

し -r, 斤-nsr


色只/所只 -torok

巴 -torok


欲 -koa

人 -koa


乎矣 -oltɐi

刁ム -ntɐi, しム -rtɐi

Causal 1:

米 -mai

Causal 2:

等以 -tɐlo, 等用 -tɐlsɨa

入 -tɐlo, 罒 -ra

c. Nominalizers:

Established fact:

隱/焉 -n,

c -o-n

Ί -n

a ko-n, 斤 a k-ɐn,

a p-ɐn, 昆 a ko-n,

a so-n, 仁 a i-n,

a cč-ɨn


尸 -l, 乙 -r

尸 -l, し -r

Declarative t-iǝ was a contraction of the bound noun with declarative 2 - and was emphatic. - displays wider functions in kugyŏl, including conjunctive function, also seen with -čiǝi; one could regard each clause listed as its own respective sentence. In hyangga, -čiǝi is used as a hortative: 禮 爲白齊 greeting do-HUM-DEC2 ‘let’s greet’ (P-i). Plain imperative -ra is seen in Kyerim yusa in “來曰烏.” ‘come! is called o-ra’ Lamentory 良羅 -ara and 兮 -hiǝi express the speaker’s lament, e.g. 脚烏伊 四是良羅 leg-NOM four-COP-LAMENT ‘the legs are four in number’ (H-v), while exclamatory -siǝ seems more objective.

a expresses the premise for another action. In LOK idu abbreviated kugyŏl was sometimes used in writing t’o, shown by 僧 人乙 合 monk layman-ACC bring.together-INF ‘bringing together monks and laymen’, hinting to us that kugyŏl letters were widely used for writing Korean. Concessive 而也 -mariiɐ and 馬於隱 -marɐn are not thought to have differed functionally. Interrupted 如可 -taka had the same function as modern -taka: 夜 入伊遊 行如可 night enter-ADV gallivant-INTERRUPT ‘as I was gallivanting into the night, …’ (H-v). Discontinued -r/-nɐr were used when a situation was suspended and a new situation started following it, e.g. example (3) 遷世 爲去在乙 pass.away do-PAST-CONT-NMR-DISC ‘even though he passed away’. -ol/-r/-n + tɐ-i introduces a quotation like MK -ótój, which dropped the nominalizer l/n within the form. Causal 2 -tɐlo consists of the bound noun plus instrumental -lo; -tɐlsɨa substitutes accusative -l + sɨ-a ‘using’ for -lo, developing from Chinese text interpretive style.

The nominalizer for an established fact was -n, used with various ch’aja reflecting fusion with the final sound of (auxiliary) verb roots in hyangga: 見根 po-ko-n see-CONVIC-NMR, 明斤 pɐlk-ɐn bright-NMR, 迷反 ip-ɨn be.confused-NMR, 惡寸 kuč-ɨn bad-NMR. There was no distinct letter in idu for -n, but 乎 -o-n was used for intentional -o + -n. The functions of nominalizers -n and -l were diverse in kugyŏl, acting as nominalizations with particles (e.g. hɐ-n-ɐn do-NMR-TOP) and the copula (e.g. hɐ-n-i-miǝ do-NMR-COP-LINK) as well as noun modifiers.

(iii) Auxiliary roots (Table 3.12). Honorific - was used in kugyŏl till (K-ii), but from (K-iii) it changed to -si. In contrast with confirmatory 去 - which confirmed an objective fact, -a is thought to have involved the speaker’s will, and ǝ-n has the nominalizer added. In kugyŏl - was not only an auxiliary root but was also widely used word-finally. -ha was used in rhetorical -hari above, e.g. 不冬 應爲賜下呂 NEG respond-AUX-HON-Q ‘might X not respond?’ (P-vii), and is thought to be honorific. Convictive 遣 -ko was a contraction of - with intentional -o. We can tell the function of essential -ms because it is often used in kugyŏl to gloss Chinese 應 ‘should’: ノ應七丨 ho-ms-ta ‘must certainly do’. There are various views on the reading and the function of volitional 將來 riǝ, and kugyŏl 吕 only appears in (K-iv) and is an unknown form whose function is not clear.

Table 3.12   Late Old Korean Auxiliary Roots





賜 -

賜 -

-, 二 -si


乎/屋 -o (留 -ru)

乎 -o

乊/ろ -o

Confirmatory 1:

去 -

去 -

古 -

Confirmatory 2:

良/於 -a, 焉 c -ǝ-n, 下 -ha

良/ -a



遣/遣只 -ko

口 -ko


音叱 -ms

七 -ms


理 -ri

ォ/未 -ri, 仒 -li


臥 -nu, 飛 -m

臥 -nu

ト -nu, 又 -no, -


如 -

如 -

! -


叱 -s, 孫 c -s-o-n

叱 -s

七 -s


將來 -riǝ

吕 ?

(iv) Derivative suffixes (Table 3.13). tɐr was used to express plurality: 善陵等沙 good.deed-PL-EMPH ‘good deeds’ (P-v). We can confirm transivitizers 是 i and 乎 o from idu, e.g. 成是 ir-i ‘make’. Deadjectival nominalizer ɐi in 明期 pɐrk-ɐi ‘bright’ (H-v) shows a function close to that of nominalizer n. As adverbializers, we find -u/o and -i (逐好/追于 čoč-o ‘following’, 加于 tǝ-u ‘more’, 幷以 apɐl-o ‘together’, 具乊 kač-o ‘ready’, 夜 入伊 tɨr-i night enter-ADV ‘into the night’, 深以 kip-i ‘deeply’, 密只 kɨsɨk-i ‘secretly’, 普〢 ‘widely’, 常〢 ‘always’), and -s which was not used in MK (e.g. 然叱 爲賜隱 that-ADV do-HON-NMR ‘did it like that’ (P-viii)) and is thought to express a mental situation like the -ehó- does in MK phelehóta ‘be blue’, kulehóta ‘be so’, -miǝ occurs solely in kugyŏl, used in 善 ‘excellently’, 能 ‘skillfully’. It is the same as the conjunctive suffix miǝ and includes a present-tense meaning. -ti was used in e.g. example (6) 能矢 and expressed factual mood. ハ is read as ki or k and was used as in 當ハ pɐntɐ-ki ‘without fail’, 今ハ iǝt-ki ‘now’, 最ハ anči-k ‘most’.

Table 3.13   Late Old Korean Derivative Suffixes





等/冬 -tɐr


伊 -i-

是 -i-, 乎-o-

〢 -i-, 乊-o-




h-o, 以l-o

于 -u, 以 l-o

乊 -o

伊/以 -i, 只 k-i

兮 -hi

i, 國 -hi

叱 -s

叱 -s

七 -s


矢/支 -ti, 古 -, ハ-ki

(v) Auxiliary verbs (Table 3.14). was used just as in MOK. 好/乎 ho, a contraction of and intentional -o, does not appear in idu texts, probably because of the limitations of the texts. In kugyŏl it is often found attached to Chinese verb roots, e.g. 減ノ尸 ho-l ‘will decrease’ (K-iii). Causative 海 hɐi of hyangga is used as 海伊 hɐi in 願海伊過 ‘I want to make wish’ (P-xi). This is a combination of auxiliary and a causative derivative suffix i. Considerate 內/良/於 is used in hyangga as in 拜內乎隱 ‘I bow (in reverence)’ (P-i), 直体良焉 ‘I set right’ (P-iii), and examples occur in idu. In kugyŏl it is in the 白 sɐrp-a used in 佛乚 見白 ヿ丨十ヿ Buddha-ACC see-HUM-CONSID-NMR fact-LOC-TOP ‘when he saw the Buddha’ (K-iii). Humble sɐrp occurs frequently in hyangga, idu and kugyŏl and was ubiquitous in the tenth century. Idu 白內 sɐrpa was also written 白於 and in kugyŏl as 白. This expressed the humble mood strongly by combining two humble auxiliary verbs. Factual 支 ti was used in hyangga and kugyŏl, but it is not known in idu. There are already examples in the twelfth-century (K-iii) of it changing its rules, and in the thirteenth century it was only used as the hun of 如, which shows us that its duties had vanished. Perfect aspect 置 tu expressed something completed in the past that continues into the present, as in 嫁良置古 ‘had married, and’ (H-vi) and 爲白置在敎如乙 do-HUM-PERF-CONT-HON[-NMR] fact-ACC ‘the fact that I had done’ (LT-iv). There are no examples in kugyŏl. Continuative kiǝ is frequent in idu and kugyŏl, but is completely absent in hyangga. 向 -a was used in idu in 向事 -a-n ir, as in 右塔 重修向 事之 段 right.pagoda repair-EXEC[-NMR] matter-GEN fact-LOC-TOP ‘at the time of the repairs to the right-hand pagoda’ (LT-vi). I treated 只 ki ‘instruct’ as a quasi-grammatical form in MOK. In LOK it was used in the form 只賜 ki-sɐ in 十方叱 佛体 閼遣只賜立 ten.directions-GEN Buddha know-CONV-HON-IMV ‘please know the Buddhas of all the directions’ (P-iv). This corresponds to kugyŏl kisɐ/kisi, and where hyangga kisɐ showed greatest respect to Buddhas, kugyŏl kisɐ did so to the king. This 只賜 kisɐ corresponds to the idu superhonorific auxiliary kisi.

Table 3.14   Late Old Korean Auxiliary Verbs





爲 --, 好 -ho-

爲 --

丷 --, ノ -ho-


海 -hɐi-

令是 -hɐi-

亽 -hɐi-


內/良 -a-

內/於 -a-


Humble 1:

白 -sɐrp-

白 -sɐrp-

白 -sɐrp-

Humble 2:

白內 -sɐrpa-

白 -sɐrpa-


支 -ti-

支 -ti-


置 -tu-

置 -tu-


在 -kiǝ-

ナ -kiǝ-


向 -a-


只 -ki-, 敎 -ki.sɐ-

敎 -ki.sɐ-

ハ -ki-, ハ -ki.sɐ-

As auxiliary roots and auxiliary verbs had the independence that verb roots had, they were attached without an intermediary suffix. 安爲白置在敎 ‘once he had enshrined…’ is analyzed as five verbs combined in their root form: 安 ‘enshrining’ + auxiliary 爲 + humble 白 sɐrp + perfect 置 tu + continuative 在 kiǝ + superhonorific 敎 kisi + converb miǝ.

(vi) Quasi-grammatical forms. 等 was used as the bound noun of the factual mood in hyangga and in idu, and 入 was used consistently in kugyŏl. 段 ta-n, frequently found in idu, was explained above under discourse particles. The bound nouns 所 pa and 事 ir were used in the sense of ‘thing, matter’. The 下 ha in 吾下於叱古 ‘they are mine’ and 誰支下焉古 ‘whose are they?’ (H-v) is a bound noun used in the sense of ‘possession’. a is used as a sentence-nominalizing bound noun in 十四王ラ 乚說口ハ二ヿ 14.kings-GEN matter-ACC explain-HON-NMR ‘he explains the matter of the 14 Kings’ (旧仁 11) means matter, confirmed also in dot-type kugyŏl. 爲 is read in context as sam, meaning either ‘for the purpose of’ or ‘because of’, used with locative -ɨi in the purpose sense. In kugyŏl sam is indicated with 三/一/氵 on the characters 爲, 以, 由 or 因 in the Chinese text which express a causal relationship, and the use of sam in a causal sense is considered grammatical influence from Chinese texts. Very commonly seen in LOK idu is 元 in an ablative sense, read pirɨs. 念丁 used in MOK in the sense of ‘across’ space, is used in LOK in the sense of ‘through’ time: 同年 春秋冬 念丁 ‘through spring, autumn and winter of that year’ (LT-v). 仍于 čičɨru is common in LOK idu meaning ‘in accordance with’.

(vii) Negation. The distinction between nominal and verbal sentence patterns of negation is clearly seen in interpretive kugyŏl (18): nominal sentence pattern (a) noun + anti + copula i; nominal sentence pattern (b) verb root + nominalizer -n + anti + copula i; verbal sentence pattern (c) verb root + nominalizer -l + antɐr + auxiliary .









‘it is not the inside, it is not the outside …’ (K-ii, 13)






‘is not firm and …’ (K-ii, 18)






‘does not abandon’ (K-ii, 14)

Only the verbal pattern is seen in hyangga: 不冬 喜好尸 ‘shall we not rejoice?’ (P-v). In kugyŏl 不ハ antɐk occurs as the potential negator (知尸不ハノ仒七過失〢 know-NMR NEG-AUX-PROSP-GEN fault-COP-LINK ‘being a fault that one cannot know, …’ (K-vi, 13)), while in contrast it is 毛冬/毛等 motɐr that was used in hyangga (毛冬 長乙隱 ‘[where] we cannot cultivate’ (P-vi)). Potential negator 不得 motsir is only known in idu: 成是 不得爲乎 make[-REL] NEG-AUX-INTENT ‘could not bring it to fruition’ (LT-v). It was a calque from Chinese 不得, so motɐr is found in later idu texts too.

3.7  Conclusion

I personally feel that the description of OK remains problematic. Firstly, OK covers a long period of more than 900 years. Each sub-period has scarce materials, and there is much that we cannot decipher or analyze by relying on discovered texts. There are many divergent views regarding Three Kingdoms toponyms and hyangga, and amongst interpretive kugyŏl there are undecipherable forms too. However, the discovery and research of newly discovered interpretive kugyŏl and idu texts have provided new views on OK. In this chapter I have outlined based on such achievements what has been researched up to very recently. As I prepare this chapter, there is newly edited data, and I have abstained from including what I cannot yet verify concretely.

I shall briefly conclude with some words about differences between OK and MK. OK nominalizers -n and -l have more or less vanished as such by MK, remaining as relativizers. The functions of MK nominalizer -(ó/u)m, whose origins are obscure, were much reduced compared with OK. The independent nature of verb roots was weakened by MK, and they were used as bound roots before endings; root + root sequences were fossilized in compound verbs. Auxiliary verb 白 sɐrp changed into the auxiliary root -sóp/cóp/zóp-. 在 kiǝ ceded its function to is- and vanished. Factual mood 支 -ti vanished in the twelfth century, and considerate 內 -a and superhonorific 敎 -kisi also vanished by the fifteenth century. The nominal vs verbal sentence distinction of negation vanished, and the -ti and -tɐr of anti and antɐr developed into adverbializers on the verb and the functional difference between them vanished. The OK conditional was expressed through the topic particle, but -myen and -ketun developed as a new grammatical category in MK. Sentence enders -čiǝi and - and auxiliary root -ms virtually vanished and were used only in an extremely restricted way. -oltɐi and -ontɐi developed into productive -otɐi (LMK -otój) through loss of the component nominalizer -l/-n.


This chapter was translated from the original Korean by Nicolas Tranter, with great help from a Japanese version prepared by Yun-jin Nam. Metalinguistic – and metagraphic – issues in making this chapter accessible to readers with no knowledge of Korean or of Chinese characters meant a number of concessions, particularly a severe reduction in the number of examples from primary texts.


Kang, Sinhang (1980) Kyeylim Yusa Kolye Pangen Yenkwu [A study of Koryŏ dialect in Kyerim yusa], Seoul: Sungkyunkwan University Press.
Kono, Rokuro (ed.) (1996) Gengogaku Daijiten 6-kan: Jutsugo-hen [Encyclopedia of linguistics: Predicates], Tokyo: Sanseido.
Lee, Ki-Moon (1972) Kwukesa Kaysel [Introduction to the history of the Korean Language], Seoul: Thap Chwulphansa.
Lee, Ki-Moon and Ramsey, S. Robert (2011) A History of the Korean Language, Cambridge: Cambridge University Press.

Further Reading (All in Korean)

權仁瀚 (2006), 武寧王陵 出土 銘文들에 대한 語學的 考察, 『口訣硏究』 17, 口訣學會.

金永萬 (1997), 석독구결 ‘皆七’, ‘悉’. 와 고려향찰 ‘頓部七’, ‘盡’ 의 비교 고찰, 『口訣硏究』 2, 口訣學會.

金永旭 (2003), 百濟 吏讀에 대하여, 『口訣硏究』 11, 口訣學會.

金完鎭 (1980), 『鄕歌解讀法硏究』, 서울大學校出版部.

김완진 (2000), 『향가와 고려가요』, 서울대학교출판부.

김지오 (2010), <懺悔業障歌>의 國語學的 解讀, 『口訣硏究』 24, 口訣學會.

南廣祐編著 (1995), 『古今漢韓字典』, 仁荷大學校出版部.

南廣祐編著 (1999), 『古語辭典』, 敎學社.

남기심·고영근 (1985), 『표준국어문법론』, 塔出版社.

南豊鉉 (1981), 『借字表記法硏究』, 檀大出版部.

南豊鉉 (1999), 『國語史를 위한 口訣 硏究』, 太學社.

南豊鉉 (2000), 『吏讀硏究』, 太學社.

南豊鉉 (2009), 『古代韓國語硏究』, 시간의 물레 .

南豊鉉 (2010a), 獻花歌의 解讀, 『口訣硏究』24. 口訣學會.

南豊鉉 (2010b), 韓國語史 硏究에 있어 口訣資料의 寄與에 대하여, 『口訣硏究』 25. 口訣學會.

南豊鉉 (2011), 古代韓國語의 謙讓法 助動詞 ‘白/Ↄ’과 ‘內/아’의 發達, 『口訣硏究』 26. 口訣學會.

노명호·이승재 (2009), 釋迦塔에서 나온 重修文書의 判讀 및 譯註, 『重修文書』, 국 립중앙박물관·대한불교조계종.

都守熙 (2005), 『百濟語 語彙 硏究』, 제이앤씨.

박진호 (1998), 고대 국어 문법, 『국어의 시대별 변천 연구 3』, 국립국어연구원.

박진호 (2008), 鄕歌解讀과 國語文法史, 『國語學』51, 國語學會.

박창원 (2002), 『고대국어 음운 (1)』, 태학사.

백두현 (1997), 高麗時代 釋讀口訣에 나타난 선어말어미의 계열관계와 통합관계, 『口訣硏究』2, 口訣學會.

徐在克 (1975), 『新羅 鄕歌의 語彙 硏究』, 啓明大學 韓國學硏究所 .

서종학 (2004), 借字 ‘令是-’攷, 『口訣硏究』12, 口訣學會.

梁柱東 (1965), 『增訂 古歌硏究』, 一潮閣.

李基文 (1981), 『韓國語形成史』, 三星美術文化財團

李基文 (1991), 『國語 語彙史 硏究』, 東亞出版社.

李崇寧 (1961), 『中世國語文法』, 乙酉文化史.

李丞宰 (1992), 『高麗時代의 吏讀』, 國語學叢書 17, 國語學會, 太學社.

李丞宰外 (2005–2006), 『角筆口訣의 解讀과 飜譯 1–3』, 태학사.

장경준 (2007), 『「瑜伽師地論」点吐釋讀口訣의 解讀 方法 硏究』, 國語學叢書 58, 國語學會, 太學社.

장경준 (2009), 점투구결 자료의 문법 형태에 대하여, 『國語學』56, 國語學會.

장윤희 (1998), 석독구결 자료의 감탄법 종결어미, 『口訣硏究』4. 口訣學會.

鄭在永 (2003), 百濟의 文字生活, 『口訣硏究』11, 口訣學會.

鄭在永 (2004), 鷄林類事의 高麗方言에 나타난 文法形態에 대한 硏究, 『口訣硏究』12, 口訣學會.

鄭在永외 (2003), 『韓國 角筆 符號口訣 資料와 日本 訓點 資料 연구』, 태학사.

황선엽 (2008), 三國遺事와 均如傳의 鄕札 表記字 비교 , 『國語學』51. 國語學會.

황선엽 (2010), 향가의 연결어미 ‘-아’ 표기에 대하여 , 『口訣硏究』25, 口訣學會.

小林芳規 (2008), 角筆による新羅語加点の華嚴經, 『南都佛敎』91. 南都佛敎硏究會

東大寺.小林芳規 (2010), 日本의 오코토点의 起源과 古代 韓國語의 點吐와의 關係 , 『口訣硏究』25, 口訣學會.

Search for more...
Back to top

Use of cookies on this website

We are using cookies to provide statistics that help us give you the best experience of our site. You can find out more in our Privacy Policy. By continuing to use the site you are agreeing to our use of cookies.