Old Korean

The Samguk yusa contains most surviving Silla hyangga

The only Korean-language literature that survives from Silla are vernacular poems now called hyangga (Korean: 향가; Hanja: 鄕歌), literally "local songs".[38]

Hyangga appears to have been a flourishing genre in the Silla period, with a royally commissioned anthology published in 888.[38] That anthology is now lost, and only twenty-five works survive. Fourteen are recorded in the Samguk yusa, a history compiled in the 1280s by the monk Iryeon,[39] along with prose introductions that detail how the poem came to be composed.[40] These introductions date the works to between 600 and 879. The majority of Samguk yusa poems, however, are from the eighth century.[38] Eleven additional hyangga, composed in the 960s by the Buddhist monk Gyunyeo,[38] are preserved in a 1075 biography of the master.[41] Lee Ki-Moon and Ramsey consider Gyunyeo's hyangga to also represent "Silla poetry",[38] although Nam Pung-hyun insists on significant grammatical differences between the works of the Samguk yusa and those of Gyunyeo.[42]

Because centuries passed between the composition of hyangga works and the compilation of the works where they now survive, textual corruption may have occurred.[43][44] Some poems that Iryeon attributes to the Silla period are now believed to be Goryeo-era works.[45][46] Nam Pung-hyun nevertheless considers most of the Samguk yusa poems to be reliable sources for Old Korean because Iryeon would have learned the Buddhist canon through a "very conservative" dialect and thus fully understood the Silla language.[47] Other scholars, such as Park Yongsik, point to thirteenth-century grammatical elements in the poems while acknowledging that the overall framework of the hyangga texts is Old Korean.[48]

The hyangga could no longer be read by the Joseon period (1392–1910).[49] The modern study of Old Korean poetry began with Japanese scholars during the Japanese colonial period (1910–1945), with Shinpei Ogura pioneering the first reconstructions of all twenty-five hyangga in 1929.[50][51] The earliest reconstructions by a Korean scholar were made by Yang Chu-dong in 1942 and corrected many of Ogura's errors, for instance properly identifying 只 as a phonogram for *-k.[52] The analyses of Kim Wan-jin in 1980 established many general principles of hyangga orthography.[53][54] Interpretations of hyangga after the 1990s, such as those of Nam Pung-hyun in the 2010s, draw on new understandings of early Korean grammar provided by newly discovered Goryeo texts.[55][56]

Nevertheless, many poems remain poorly understood, and their phonology is particularly unclear.[57] Due to the opaqueness of data, it has been convention since the earliest Japanese researchers[58] for scholars to transcribe their hyangga reconstructions using the Middle Korean lexicon, and some linguists continue to anachronistically project even non-lexical Middle Korean elements in their analyses.[59]

Silla inscriptions also document Old Korean elements. Idiosyncratic Chinese vocabulary suggestive of vernacular influence is found even in the oldest surviving Silla inscription, a stele in Pohang dated to either 441 or 501.[8] These early inscriptions, however, involved "little more than subtle alterations of Classical Chinese syntax".[9]

The Imsin Vow Stone of 552/612 uses Old Korean syntax.

Inscriptions of the sixth and seventh centuries show more fully developed strategies of representing Korean with Chinese characters. Some inscriptions represent functional morphemes directly through semantic Chinese equivalents.[9] Others use only Classical Chinese vocabulary, but reorder them fully according to Korean syntax. A 551 stele commemorating the construction of a fort in Gyeongju, for instance, writes "begin to build" as 作始 (lit. "build begin") rather than the correct Classical Chinese, 始作 (lit. "begin build"), reflecting the Subject-object-verb word order of Korean.[60] The Imsin Vow Stone, raised in either 552 or 612,[9] is also illustrative:

Other sixth-century epigraphs that arrange Chinese vocabulary using Korean syntax and employ Chinese semantic equivalents for certain Korean functional morphemes have been discovered, including stelae bearing royal edicts or celebrating public works and sixth-century rock inscriptions left at Ulju by royals on tour.[62][63] Some inscriptions of the Unified Silla period continue to use only words from Classical Chinese, even as they order them according to Korean grammar.[64] However, most inscriptions of the period write Old Korean morphemes more explicitly, relying on Chinese semantic and phonetic equivalents.[11] These Unified-era inscriptions are often Buddhist in nature and include material carved on Buddha statues, temple bells, and pagodas.[64]

Sixth-century mokgan slips from Haman

Ancient Korean scribes often wrote on bamboo and wooden slips called mokgan.[65] By 2016, archaeologists had discovered 647 mokgan, out of which 431 slips were from Silla.[66] Mokgan are valuable primary sources because they were largely written by and reflect the concerns of low-ranking officials, unlike other texts that are dominated by the high elite.[65] Since the majority of discovered texts are inventories of products, they also provide otherwise rare information about numerals, classifiers, and common nouns.[67]

Modern mokgan research began in 1975.[68] With the development of infrared imaging science in the 1990s, it became possible to read many formerly indecipherable texts,[69] and a comprehensive catalog of hitherto discovered slips was published in 2004. Since its publication, scholars have actively relied on the mokgan data as an important primary source.[70]

Mokgan are classified into two general categories.[71] Most surviving slips are tag mokgan,[72] which were attached to goods during transport and contain quantitative data about the product in question.[72] Document mokgan, on the other hand, contain administrative reports by local officials.[71] Document mokgan of extended length were common prior to Silla's conquest of the other kingdoms, but mokgan of the Unified period are primarily tag mokgan.[73] A small number of texts belong to neither group; these include a fragmentary hyangga poem discovered in 2000[71] and what may be a ritual text associated with Dragon King worship.[74][lower-alpha 4]

The earliest direct attestation of Old Korean comes from a mid-sixth century document mokgan first deciphered in full by Lee Seungjae in 2017.[10] This slip, which contains a report by a village chieftain to a higher-ranking official,[76] is composed according to Korean syntax and includes four uncontroversial examples of Old Korean functional morphemes (given below in bold), as well as several potential content words.[10]

Old Korean glosses have been discovered on eighth-century editions of Chinese-language Buddhist works.[78][79] Similar to the Japanese kanbun tradition,[80] these glosses provide Old Korean noun case markers, inflectional suffixes, and phonograms that would have helped Korean learners read out the Classical Chinese text in their own language.[81] Examples of these three uses of glossing found in a 740 edition of the Avatamsaka Sutra (now preserved in Tōdai-ji, Japan) are given below.[81]

Scroll of a Silla edition of the Avatamsaka Sutra, written in 754–755

Portions of a Silla census register with Old Korean elements, likely from 755 but possibly also 695, 815, or 875, have also been discovered at Tōdai-ji.[86]

Though in Classical Chinese, the Korean histories Samguk sagi and Samguk yusa offer Old Korean etymologies for certain native terms. The reliability of these etymologies remains in dispute.[87]

Non-Korean texts also provide information on Old Korean. A passage of the Book of Liang, a seventh-century Chinese history, transcribes seven Silla words: a term for "fortification", two terms for "village", and four clothing-related terms. Three of the clothing words have Middle Korean cognates, but the other four words remain "uninterpretable".[88] The eighth-century Japanese history Nihon Shoki also preserves a single sentence in the Silla language, apparently some sort of oath, although its meaning can only be guessed from context.[89]

The Samguk sagi, the Samguk yusa, and Chinese and Japanese sources transcribe many proper nouns from Silla, including personal names, place names, and titles. These are often given in two variant forms: one that transcribes the Old Korean phonemes, using Chinese characters as phonograms, and one that translates the Old Korean morphemes, using Chinese characters as logograms. This is especially true for place names; they were standardized by royal decree in 757, but the sources preserve forms from both before and after this date. By comparing the two, linguists can infer the value of many Old Korean morphemes.[90]

The modern Korean language has its own pronunciations for Chinese characters, called Sino-Korean.[92] Although some Sino-Korean forms reflect Old Chinese or Early Mandarin pronunciations, the majority of modern linguists believe that the dominant layer of Sino-Korean descends from the Middle Chinese prestige dialect of Chang'an during the Tang dynasty.[93][94][95][lower-alpha 7]

As Sino-Korean originates in Old Korean speakers' perception of Middle Chinese phones,[97] elements of Old Korean phonology may be inferred from a comparison of Sino-Korean with Middle Chinese.[9] For instance, Middle Chinese, Middle Korean, and Modern Korean all have a phonemic distinction between the non-aspirated velar stop /k/ and its aspirated equivalent, /kʰ/. However, both are regularly reflected in Sino-Korean as /k/. This suggests that /kʰ/ was absent in Old Korean.[98]

Old Korean phonology can also be examined via Old Korean loanwords in other languages, including Middle Mongol[99] and especially Old Japanese.[100]

Fifteenth-century Middle Korean was a tonal or pitch accent language whose orthography distinguished between three tones: high, rising, and low.[147] The rising tone is analyzed as a low tone followed by a high tone within a bimoraic syllable.[148]

Middle Chinese was also a tonal language, with four tones: level, rising, departing, and entering. The tones of fifteenth-century Sino-Korean partially correspond to Middle Chinese ones. Chinese syllables with level tone have low tone in Middle Korean; those with rising or departing tones, rising tone; and those with entering tone, high tone. These correspondences suggest that Old Korean had some form of suprasegmentals consistent with those of Middle Chinese, perhaps a tonal system similar to that of Middle Korean.[149] Phonetic glosses in Silla Buddhist texts show that as early as the eighth century, Sino-Korean involved three tonal categories and failed to distinguish rising and departing tones.[150]

On the other hand, linguists such as Lee Ki-Moon and S. Roberts Ramsey argue that Old Korean originally had a simpler prosody than Middle Korean, and that influence from Chinese tones was among the reasons for Korean tonogenesis.[151] The hypothesis that Old Korean originally lacked phonemic tone is supported by the fact that most Middle Korean nouns conform to a tonal pattern,[151] the tendency for ancient Korean scribes to transcribe Old Korean proper nouns with Chinese level-tone characters,[152] and the accent marks on Korean proper nouns given by the Japanese history Nihon Shoki, which suggest that ancient Koreans distinguished only the entering tone among the four Chinese tones.[153]

Middle Korean had a complex syllable structure that allowed clusters of up to three consonants in initial[154] and two consonants in terminal position,[155] as well as vowel triphthongs.[156] But many syllables with complex structures arose from the merger of multiple syllables, as seen below.

Middle Korean closed syllables with bimoraic "rising tone" reflect an originally bisyllabic CVCV form in which the final vowel was reduced,[161] and some linguists propose that Old Korean or its precursor originally had a CV syllable structure like that of Japanese, with all clusters and coda consonants forming due to vowel reduction later on.[162] However, there is strong evidence for the existence of coda consonants in even the earliest attestations of Korean, especially in mareum cheomgi orthography.[163]

On the other hand, Middle Korean consonant clusters are believed not to have existed in Old Korean and to have formed after the twelfth century with the loss of intervening vowels.[158] Old Korean thus had a simpler syllable structure than Middle Korean.

The consonant inventory of fifteenth-century Middle Korean is given here to help readers understand the following sections on Old Korean consonants. These are not the consonants of Old Korean, but of its fifteenth-century descendant.

Three of the nineteen Middle Korean consonants could not occur in word-initial position: /ŋ/, /β/, and /ɣ/.[167] Only nine consonants were permitted in the syllable coda. Aspiration was lost in coda position; coda /ts/ merged with /s/;[168] and /β/, /ɣ/, /h/, and the reinforced consonants could not occur as the coda.[169] Coda /z/ was preserved only word-internally and when followed by another voiced fricative; otherwise it merged with /s/.[168]

Sino-Korean evidence suggests that there were no major differences between Old Korean and Middle Korean nasals.[170]

Middle Chinese onset *ŋ is reflected in Sino-Korean as a null initial, while both Chinese and Korean transcriptions of Old Korean terms systematically avoid characters with onset *ŋ. Middle Korean phonotactic restrictions on /ŋ/ thus seem to have held true for Old Korean as well.[171]

The Samguk Sagi sometimes alternates between nasal-initial and liquid-initial characters in transcribing the same syllable of the same proper noun. This suggests that Old Korean may have had a sandhi rule in which nasals could become liquids, or vice versa, under certain circumstances.[172]

The Middle Korean series of aspirated stops and affricates developed from mergers of consonant clusters involving /h/ or a velar obstruent, which in turn had formed from the loss of intervening vowels.[173] To what extent this process had occurred in the Old Korean period is still debated.

Middle Chinese had a phonemic distinction between aspirated and unaspirated stops. This is reflected somewhat irregularly in Sino-Korean.

Based on this variable reflection of Middle Chinese aspirates, Korean is thought to have developed the dental aspirates *tsʰ and *tʰ first, followed by *pʰ and finally *kʰ.[175] *kʰ is often believed to have been absent when Sino-Korean was established.[175]

Silla scribes used the aspirate-initial characters only infrequently.[176] When they did, the aspirates were often replaced with unaspirated equivalents. For instance, the 757 standardization of place names sometimes involved changing aspirated phonograms for unaspirated ones, or vice versa.[177] This suggests that aspiration of any sort may have been absent in Old Korean,[176] or that aspirate stops may have existed in free variation with unaspirated ones but were not distinct phonemes.[177] On the other hand, Ki-Moon Lee and S. Roberts Ramsey argue that Silla orthography confirms the existence in Old Korean of at least the dental aspirates as phonemes.[175]

Meanwhile, Nam Pung-hyun believes that Old Korean had *kʰ and *tsʰ but may have lacked *pʰ and *tʰ, while noting that the functional load of the aspirates was "extremely low".[146]

Some characters with initial *k in Middle Chinese are reflected in Sino-Korean as /h/. Conversely, some instances of the Middle Chinese initial *ɣ~*ɦ, usually loaned as Sino-Korean /h/, are found as Sino-Korean /k/. This may be because Koreans mistakenly assigned the same initial consonant to characters which do share a phonetic radical but in practice had different Middle Chinese initials.[178] On the other hand, this may reflect a velar value for the Old Korean ancestor of Middle Korean /h/. Korean scholars often propose an Old Korean velar fricative *x as ancestral to Middle Korean /h/.[179][180]

Some orthographic alternations suggest that Silla writers did not distinguish between Middle Chinese initial *k and initial *ɣ~*ɦ, although linguist Marc Miyake is skeptical of the evidence,[181] while some Middle Korean allomorphs alternate between /h/ and a velar. Linguist Wei Guofeng suggests that the Old Korean phonemes *k and *h had overlapping distributions, with allophones such as *x being shared by both phonemes.[182] Alexander Vovin also argues via internal reconstruction that intervocalic *k in earlier Korean lenited to Middle Korean /h/.[183]

In some pre-Unified Silla transcriptions of Korean proper nouns, Chinese affricate and fricative sibilants appear interchangeable. This has been interpreted as some stage of Old Korean having lacked the Middle Korean distinction between /ts/ and /s/. The hyangga poems, however, differentiate affricates and fricatives consistently, while the Chinese distinction between the two is faithfully preserved in Sino-Korean phonology. Koreans thus clearly distinguished /ts/ from /s/ by the eighth century, and Marc Miyake casts doubt on the notion that Korean ever had a stage where affricates and fricatives were not distinct.[184]

Middle Korean had only one liquid phoneme, which varied between [l] and [ɾ]. Old Korean, however, had two separate liquid phonemes. In Old Korean orthography, this first liquid phoneme was represented by the PAP 尸, whose Old Chinese value was *l̥[ə]j, and the second phoneme by the PAP 乙, whose Old Chinese value was *qrət.[125][185] Besides this orthographic variation, the distinction between the liquid phonemes is also suggested by the tonal behavior of Middle Korean verb stems ending in /l/.[186]

According to Alexander Vovin, Lee Ki-Moon asserts that 尸 represented /r/ and that 乙 represented /l/. Vovin considers this claim "unacceptable" and "counterintuitive", especially given the reconstructed Old Chinese pronunciations of both characters, and suggests instead that 尸 represented /l/ while 乙 stood for a rhotic.[187] Ramsey and Nam Pung-hyun both agree with Vovin's hypothesis.[186][188]

Three Old Korean numerals are attested in the hyangga texts: those for one, two, and thousand. All three are found in the Docheonsugwaneum-ga, while the word for one is also attested identically in the Jemangmae-ga.[189] The Cheoyong-ga uses a somewhat different form for "two",[189] though it is unlikely to be authentically Silla.[190] The mokgan data, discussed in Lee Seungjae 2017, suggests that multiples of ten may have been referred to with Chinese loanwords but that indigenous terms were used for single-digit numbers.[191] Lee's work on mokgan yields Silla words for four of the latter: one, three, four, and five.[192] The orthography of Old Korean numerals, in both hyangga and the mokgan texts, is marked by the hunju eumjong principle typical of Silla.[193]

The Old Korean single-digit numerals are given with fifteenth-century and Modern Korean equivalents below. The Modern Korean terms used to refer to the ages of cattle, which Lee Seungjae considers more closely related to Old Korean forms, are also provided.[194]