Notes on the languages of the Three Kingdoms

The following notes are taken from Song Gi-jung’s Historical Comparative Linguistics and Theories on the Genealogy of the Korean Language (2004).  Song offers an insightful and critical summary into what is known and what opinions are held by Koreans regarding the languages of Goguryeo, Baekje and Silla.  I’ve made these notes a mix of translation and general paraphrasing with very occasional supplementation.  They hopefully give a useful overview but should not be quoted from without referring to the original book!  (The numbers in brackets indicate the page of Song’s book where the equivalent information can be found.)

  • It may forever be impossible to accurately reconstruct the contemporary pronunciation of any vocabulary from the Three Kingdoms period because the phonetic reading (讀音) of hanja (Sino-Korean characters), in which all known sources are recorded, has differed throughout periods and regions. (174)
  • It is not unreasonable to surmise that written Chinese (漢文) and Chinese characters (漢字) would have been introduced alongside Chinese culture (漢文化) to the Three Kingdoms which suddenly rose up both within and adjacent to the territory occupied by the Han Commanderies. (175)
  • It is presumed, therefore that Sino-Korean vocabulary (한문어식 단어) already existed to a considerable degree at the beginning of the Three Kingdoms period. (175)

Relying on the language sources currently known, it cannot be estimated to what degree the languages of the three countries differed, nor to what extent the language of each kingdom changed according to period.  Consequently, whilst it cannot be said whether the difference was one between that of language families or dialects, on the premise that the languages did at least differ to some degree, they can be distinguished as ‘Goguryeoic’ (고구려어), ‘Baekjeic’ (백제어) and ‘Sillaic’ (신라어).


  • Goguryeo is recorded as only ever having used the Sino-Korean term wang (王) ‘king’ to designate their ruler.  This is in contrast to both Silla in the south which early on used the terms geoseogan (居西干), chacha’ung (次次雄) and isageum (尼師今), and the peoples who neighbored to the north of Goguryeo such as the Xiongnu (匈奴) and Göktürks (突厥).  It is thought therefore that Chinese vocabulary was introduced to Goguryeo particularly early on. (177)
  • Official ranks and titles such as mangniji (幕離支), daeryeo (對廬) and toesal (褪薩); personal names such as Yeon Gaesomun (淵蓋蘇文), Ondal (溫達) and Eulpaso (乙巴素); and toponyms such as Dunul (杜訥), Maehol (買忽) and Buji (夫只) are examples of indigenous words which were rendered into hanja using the characters only for their phonetic value, regardless of their original meaning.  They are clearly built from Goguryeoic word roots but for most, neither the root, nor the exact meaning or pronunciation of the word can be conjectured.  Only in the few cases where old or alternate names have been recorded alongside using hanja for their meaning, can the meaning of these Goguryeoic words be deduced. (177)
  • Examples where this is possible are principally found in books (권) 35 and 37 of the geography section (지리지) of the Samguk Sagi.  An example from Book 35, is where a former Goguryeo place name was subsequently revised during the Unified Silla period: “Sujeong-gun (水城郡 ‘water fortress county’) was originally Goguryeo’s Maehol-gun (買忽郡); King Gyeongdeok revised the name.  It is now Su-ju (水州 ‘water province’).”[1]  In Book 37 there is the example, “Namcheon-hyeon (南川縣 ‘south river county’) is also known as Nammae (南買).”[2]
  • In total there are around 80 such Goguryeoic words for which there are also hanja “translations” of their meaning; of these there are around 20 which occur in more than one toponym and thus can be confirmed with greater certainty.[3]  They include the following:

gosa                           古斯                            jewel                            (玉)
geumhol                    今忽                            black                            (黑)
nae, no, noe              內-奴-惱                       land                             (壤 ‘땅’)
naemi                         內米                             pond                            (池)
dan, tan, don             旦-呑-頓                       valley                           (谷 ‘골짜기’)
dal                               達                                mountain, high            (山,高)
mae                             買                                water, river                  (水, 川)
sabok                          沙伏                            red                               (赤)
somun                         蘇文                            gold (metal?)              (金)
sur’i, doni                  述爾-道尼                     mountain peak            (峯 ‘봉우리’)
sik                               息                                earth                             (土)
eosa                            於斯                            horizontal(?)                (橫)
eo’eul                          於乙                            water spring                 (泉)
to                                 吐                               embankment, dyke       (堤 ‘뚝방’)
pa’ui, pa’ui, pahye    巴衣-波衣-波兮          rock                                (巖 ‘바위’)
pa                               波                                sea                                 (海)
hol                              忽                                fortress                          (城)

  • Some grammatical particles and endings can be deduced from mid 5th century stone inscriptions (石刻文) that record the construction of Pyeongyang-seong (平壤城).  They represent an early form of idu where certain characters are borrowed to represent grammatical features of the vernacular Goguryeoic, e.g. ‘-jung (-中)’ is used for the equivalent to the modern Koreanic dative particle -e (-에) and ‘-ji (-之)’ for the verb ending -da (-다).  An example sentence is:

丙戌十二月 漢城 下後卩 小兄文達 自此西北行涉之  (卩=部)

해석: 병술 12월 한성 하후부 소형 문잘 감독관(?)[이] 여기서부터 서북방(?)[공사를] 치르(?).[4]  (178)


  • Compared to Goguryeoic, the introduction of hanja to Baekje was significantly later, however there are far fewer confirmable Baekjeic words.
  • Although the Sino-Korean wang (王) is exclusively used in the Samguk Sagi and Samguk Yusa, in the Chinese Zhoushu (周書) earlier terms are recorded.
  • The first 22 listed Baekje rulers have names thought to be Baekjeic (as opposed to carrying a meaning in Chinese).  This is in contrast to Goguryeo where only the name of the 2nd king, Yuri (琉璃) , appears to be Goguryeoic.
  • Identifiable Baekjeic vocabulary is similarly found in the Geography chapters of the Samguk Sagi, Book 36.  In Baekjeic han (翰) means “big” (大) and is equivalent to Sillaic han (韓).  Han meaning “big” is attested still in 15th century Middle Korean. (179)  Other examples include:

sa                           沙              new                             (新, 현대국어 ‘새’)
buri                        夫里           field, plain                  (平野, ‘벌’; Sillaic beol (伐))
ji                             只              fortress                        (城)
o                            烏              alone, lonely (single child?)  (孤, ‘외’)
bi                            比              rain                              (雨)
sobi                       所比           red     (赤; Goguryeoic sabok (沙伏), sabi (沙非))
eum                       陰  fang, tusk, molar (牙, Middle Korean ‘엄’, modern ‘어금니’)
mulgeo                 勿居           clear                          (淸, 현대국어 ‘맑-‘)
maro                     馬老           dry                              (乾, ‘마르-‘)


  • There are many more sources available for Sillaic than Goguryeoic or Baekjeic.  They include inscriptions in which the grammar does not match Classical Chinese but rather Old Korean in the manner of the idu system; there are also the 14 hyangga songs recorded in the Samguk Yusa. (179)
  • From very early on in the Silla period, there are examples of hanja characters being used in names for their Chinese meaning, but at the same time the overall influence of Chinese remained weaker than in Goguryeoic or Baekjeic.  Examples of hanja being utilized for their meaning include the title of queens/consorts (王妃) as bu’in (夫人) e.g. in Alyeong-bu’in (閼英夫人) and Unje-bu’in (雲帝夫人); gong (公) used for princes, e.g. Ho-gong (瓠公) and Sobeol-gong (蘇伐公); wang (王) used for kings, e.g. Galmun-wang (葛文王) and Heoru-wang (許婁王); ju (主) used to designate commanders and lords e.g. gun-ju (郡主 “governor”), seong-ju (城主 “fortress commander; castle lord”), gun-ju (軍主 “chief; military governor”), jin-ju (鎭主 “garrison chief”). (179)
  • Until King Jijung (智證王) in the 6th century, the titles for and names of kings, together with official positions created during the reign of the 3rd monarch Isageum Yuri (r.24-57), were all non-Chinese, i.e. Sillaic.  Titles of the Silla king include geoseogan (居西干), chacha’ung (次次雄) , isageum (尼沙今) and maripgan (麻立干); names of kings includes Hyeokgeose (赫居世), Namhae (南解), Yuri (儒理), Talhae (脫解), Pasa (婆娑) and Jima (祗麻); official positions ibeolson (伊伐飡), icheokson (伊尺飡), pajinson (波珍飡) and ason (阿飡). (179-80)
  • In most cases, still, the meaning of identifiable Sillaic words cannot be deduced, only when Chinese translations are given beside the phonogram (character used for its phonetic value).  In total, there are more than 30 Sillaic words which can be reconstructed; in contrast to Goguryeoic or Baekjeic many of them can be compared with words found in 15th century Middle Korean and modern Korean.  In the 1st Book of the Samguk Sagi, it is recorded that, “The people of Jinhan (辰韓) call gourds (瓠) bak (朴).” (180)  Other examples include:

gawi         嘉俳         Han’gawi (Chuseok)   (秋夕, ‘한가위 < ᄀᆞᄇᆞㅣ’)
pajin          破珍           sea                            (海 ‘바다 < 바ᄅᆞᆯ’)
han            韓              big                              (大, 중세국어 ‘한’)
na              那              river                             (川, ‘내’)
bulgeo      弗炬           red, light                     (赤, 光明 ‘붉-’)
eul             乙              water well                    (井)
mul            勿              water                           (水 ‘물’)
alji             閼知           child                            (小兒 ‘아지’)
gil              吉              long                             (永 ‘길-’)
geochil       居柒        rough                           (萊, 荒, ‘거츨’)
icha, icheo   異次-異處   hate(?)                   (厭, 중세국어 ‘잋-’)
mil             密              push                            (推, ‘밀-’)

  • From the 14 recorded hyangga (鄕歌) songs it is possible to deduce grammatical particles:

subject (주격)                                 i, shi                            伊-是 (‘-이/가’)
genitive (속격)                                ui, ui                            矣-衣(‘-의’)
dative (처격)                             jung, yangjung, yajung   中-良中-也中(‘-에’)
accusative (대격)                             eul                              乙(‘-을/를’)
instrumental (조격)                          [r]yu                             留(‘-로’)
comitative (공동격)                          gwa                             果(‘-와/과’)
topic marker (주제지칭사)              eun                              隱(‘-은/는’)

The relationship between the languages of the Three Kingdoms

  • Those who profess that the languages of the Three Kingdoms were the same are at most able to suggest 30 words they believe are cognates, but all of them are problematic. (181)

For example North Korean scholar Kim Su-gyeong quotes a list from another North Korean consisting of 29 Three Kingdoms’s cognates.[5]  In contrast to this, South Korean scholar Kim Bang-han claims there are no cognates for all three kingdoms and lists only a small number of possible cognates between Goguryeo and Baekje.[6]  (Note 7, page 181.)

As an example, the first on the list of Kim Su-gyeong’s 29 cognates (given p45 of his book) may be assumed to be the one he and his colleagues were most confident about.  It comprises Goguryeoic su’eul (首乙) and mak-ri (莫離); Baekjeic moryang(go) (毛良(高)) and Sillaic suro/sureung (首露/首陵), marip/masu (麻立/麻袖) and mal(sang) (末(上).  He surmises their shared reading (공통적인 독음) was *mara/mari (마라/마리), and their meaning was meori (머리”head”).  In the Goguryeo and Silla cognates, the character su (首 “head”) is used for its meaning whilst all the remaining hanja are interpreted for their sound value, that is as phonograms.  The only point in common observable about these characters is that when read with Korean pronunciation, they all begin with an “m” and the second consonant is an “l/r” but this, simply, is not sufficient to confidently posit them as cognates.  (Note 8, p181).

In a separate list in Kim Su-gyeong’s book (p48-54) he suggests 17 Goguryeoic-Baekjeic cognates, 37 Goguryeoic-Sillaic cognates and just 10 Baekjeic-Silla cognates.  By contrast, Kim Bang-han gives only 2, 6 and none respectively.  (Note 9, p181).

  • Among existing sources there are no reliable examples of shared vocabulary; there are however examples of vocabulary from each of the Three Kingdoms to be found in Middle Korean (중세국어). (182) For example:

pawi, pawi, pahye    巴衣-波衣-巴兮       rock                     (巖 ‘바회/바위’)
hol                              忽                   administrative region   (城 ‘골/고을’)
sul’i, suni                   述爾-首泥     mountain peak              (峯  ‘수늙’ 嶺)
su                               首                   cow                               (牛 ‘쇼/소’)
gosa                           古斯              jewel, bead                   (玉, ‘구슬’)

sa                               沙                 new                             (新 ‘새’)
buri                             夫里             field, plain                    (‘平野, 벌’; 신라어 ‘伐’)
o                                 烏                 alone, lonely                (孤, ‘외’)
bi                                比                  rain                              (遇 ‘비’)
eum                             陰              fang, tusk, molar  (牙; 중세국어 ‘암’, 현대극어 ‘어금니’)
mulgeo                        勿居           clear                              (淸 ‘맑-’)
maro                            馬老           dry                                 (乾, ‘마르-’);

Sillaic – same as the above list.

  • There are three possible theories as to why this is so:
  1. The Three Kingdoms all had the same language which was thus inherited through Goguryeo to Joseon.  If this were the case, however, it needs to be explained why the individual languages share no vocabulary.
  2. The languages of the Three Kingdoms all came from a common ancestor; this recognizes a difference between the languages but fails to explain still the lack of shared vocabulary.
  3. Korean developed into a unified language subsequent to the Three Kingdoms period and was thus influenced by historical vocabulary absorbed as Silla expanded into the former territories of Baekje and Goguryeo, and further with the establishment of the Goryeo dynasty and the return of remnant population from Balhae: in the absence of convincing evidence of shared vocabulary during the Three Kingdoms period this appears to be the most appropriate explanation. (182)
  • Some scholars point to the fact that no mention of different languages is made in historical sources and that therefore the Three Kingdoms must have shared the same language; but this is not positive proof and in the histories of many other countries reference is not always give to the fact that foreign countries spoke foreign languages. (183)
  • Another observation made in support of a single Three Kingdoms’ Korean language is that the founders of Goryeo named their state after Goguryeo whilst being the successors to Silla.  Together with the grouping of the three kingdoms in the Samguk Sagi and Samguk Yusa, this implies they considered themselves to be the same minjok  (민족 “ethnic people”) and therefore must have had the same language: but this is still not strong positive evidence. (183)

Records in Chinese histories and the relationship between the Three Kingdoms’ languages

  • Opinions on the relationship between the Three Kingdoms’ languages have been chiefly based on the Chinese histories. (183)  There are four main theories:
  1. That the language of Goguryeo was Buyeoic (夫餘系) whilst the languages of Silla and Baekje were Koreanic (韓系).[7]  This theory was professed by South Korean scholar Lee Gi-mun (李基文) from the early 1960s and has remained influential.
  2. That a distinction can be made in the language of Goguryeo before and after its move south.  Before hand it was Tungusic and after it was the same as Silla and Baekje.  This theory was professed by Kim Bang-hang (金芳漢) similarly from the 1960s onwards.
  3. That the Three Kingdoms all spoke the same language differing only to the extent of dialects.  This is the North Korean stance as professed by Kim Su-gyeong.
  4. That the language of Goguryeo was originally Paleo-Asiatic but through contact with Tungus tribes it became Tungusic.  This is the hypothesis of Japanese scholar Kōno Rokurō (河野六郞).  In 1945 he described the languages of ancient Northeast Asia falling in two principle groups: Japonic (日本語系), consisting of Japanese and Sillaic (formerly the dialects of the Three Han); and Buyeoic, consisting of Goguryeoic (formed from the languages of Yemaek, Okjeo and Buyeo).  From this time on, Japanese scholars began to treat Goguryeoic as a branch of the Tungusic language family. (184)
  • The main Chinese sources upon which opinions are based are: Book (卷) 85 in the Dongyi-liezhuan (東夷列傳) section of the Hou Hanshu (後漢書); Book 30 in the Wuwan-xianbei-dongyi-zhuan (烏丸鮮卑東夷傳) in the Weishu (魏書) section of the Sanguozhi (三國志); Book 54 in the Zhuyi Donyi (諸夷 東夷) section of the Liangshu (梁書); Book 100 in the Liezhuan section of the Weishu (魏書); and Book 49 in the Yicheng-zhuan (異城傳) section of the Zhoushu (周書). (184)

From these there are thirteen key passages:[8]

  1. As a separate group (별종) of Buyeo, the language and customs (제반사) of Goguryeo were very similar to Buyeo.  <三國志>, <後漢書>
  2. The language of East Okjeo (東沃沮) was basically the same as Goguryeo, or just slightly different. <三國志>
  3. The language, food, housing and clothes of East Okjeo were similar to Goguryeo. <後漢書>
  4. The language and customs (법속) of Ye (濊) was basically similar to Goguryeo. <三國志>
  5. The people of Eumnu (挹婁) appear similar to Buyeo, but their language is different to Buyeo and Goguryeo.  It is the old country of Suksin-ssi (肅愼씨). <三國志>
  6. Eumnu later became Suksin (肅愼).  The people appear similar to Buyeo but their language is distinct.  <後漢書>
  7. Mulgilguk (勿吉國) is to the north of Goguryeo and was formerly Suksin.  Its language is uniquely different. <魏書>
  8. Jinhan (辰韓) is to the east of Mahan (馬韓).  According to what their elders have passed down through generations, to avoid labour duty under the Qin (秦), many people came to Han’guk (韓國).  Mahan shared its eastern borderlands.  There is a fortified fence (城柵) and their language is not the same as Mahan’s.  They call country (國) as na ‘那’[9], bow (弓) as ho ‘弧’, and robber/bandit (賊) as gu ‘寇’… So their language is similar to the people of Qin (秦). <三國志>, <後漢書>
  9. The Byeonjin (弁辰) live amongst the Jinhan (辰韓).  They have their own fortress (성곽).  Their clothes and housing are similar to Jinhan and their languages and customs resemble one another. <三國志>
  10. The Byeonjin (弁辰) live amongst the Jinhan (辰韓); their fortress and clothing are all the same.  There is a difference in their languages and customs. <後漢書>
  11. The ancestors of Baekje were Dong’i (東夷).  Today their language and clothing is roughly similar to Goguryeo. <梁書>
  12. The ancestors of Silla were originally of Jinhan (辰韓) stock.  Jinhan (辰韓) is also called Jinhan (秦韓 i.e. Qinhan in Chinese).  According to legends passed down, during the Qin (秦) people seeking to avoid corvèe duty came to Mahan and settled there; Mahan allowed them to live in its eastern border region.  Because they were people of Qin (秦) they called their territory Jinhan (秦韓).  Their language and words for things (사물) resembles those of the Chinese.  They call country (國) as na ‘那’, bow (弓) as ho ‘弧’, and robber/bandit (賊) as gu ‘寇’, which is different to Mahan…  Originally Jinhan was divided into six countries (나라), and even smaller as 12 statelets (나라); Silla was one amongst these…  The manners (拜禮), behaviour and lifestyle (행동거지) of Silla is the same type as Goguryeo.  They have no writing but carve on wood and use them as tokens (信表).  Their language can be interpreted by people from Baekje (언어는 백제 사람을 중간에 넣고서 통할 수 있다). <梁書>
  13. The ancestors of Baekje were made up of those from Mahan (마한의 속국) and separately those from Buyeo…  The surname of the king was Buyeo-ssi (夫餘氏)  and his title (호) was Eoraha (於羅瑕) whilst the Baekje people (백성) called him Geon’gilji (鞬吉支) which in Chinese both mean king (王).  The queen (왕비) was called Eoryuk (於陸) which in Chinese means queen (妃). <周書>

These thirteen points and scholarly opinion can be summarized as follows:

Points 1-4: Offer evidence that the languages of Buyeo, Goguryeo, Eastern Okjeo and Ye were the same.  These languages are typically referred to as “Buyeoic” (부여계 언어) but it is not known whether these languages were related to Koreanic (韓系) or the Tungusic family (Lee Gi-mun thinks the former, Kōno Rokurō the latter).

Points 5-7: Offer evidence that the languages of Suksin (肅愼), Eumnu (挹婁) and Mulgilguk (勿吉國) were the same and they are typically referred to as “Suksinic” (숙신계 언어).  Mulgil (Malgal) is a tribal name that was used significantly later than Suksin or Eumnu.  Suksinic languages are thought either to be Tungusic or Paleo-Asiatic, but either way not directly related to Koreanic.

Points 8-13: Deal with the countries that existed in the Three Han but they do not all agree with one another and are slightly confusing.

Points 8 and 12 (which was likely referenced on 8): State that the language of Jinhan was that of people who had migrated from the Qin dynasty and was different to Mahan.  Some scholars such as Lee Gi-mun believe the theory that 辰 was derived from 秦 was invented by the Chinese authors, whilst other scholars such as Kōno Rokurō have interpreted it as historical fact.  Either way, as long as it is understood as having been a language brought by migrants from Qin and not the indigenous language of the local population, it does not cause a problem to the debate on the relationship between the languages of the Three Kingdoms.

Point 9: States that the languages of Byeonjin and Jinhan were similar whilst Point 10 says they were different.  Some, such as Lee Gi-mun, believe this is a matter of interpretation of dialectical differences whilst others think that it must simply be a scribal error because the information for Point 10 from the Hou Hanshu (後漢書) was based on the Sanguozhi (三國志).  Point 9 is consequently taken as evidence by Kōno Rokurō and others that the people of Byeonjin and Jinhan were the same race (同族).

Point 11: Is the only record claiming the languages of Baekje and Goguryeo were the same.  Taken together with Point 13, it is conjectured to be referring to Goguryeoic and the language of the ruling class of Baekje.  Because the Liangshu (梁書) was compiled in the early 7th century, it is interpreted by some (An Byeong-hui and Kim Bang-han) to be referring to the language of Goguryeo after Goguryeo’s movement southward; whilst by others such as Kim Su-gyeong it is taken as evidence that the language of Baekje and Goguryeo were the same.

Point 12: That Sillaic could be understood through a Baekje interpreter is generally taken as evidence by scholars (represented by An Byeong-hui) that the two languages were similar whilst Point 13 has been noted (by Kōno Rokurō and subsequently Lee Gi-mun and An Byeong-hui) to infer that the language of the ruling class and those below was different.  Other scholars (Kim Bang-han and Kim Su-gyeong), however, have stressed that there is no evidence the language of the ruling Baekje class was Goguryeoic and they are unwilling to take these points as evidence that Baekjeic and Goguryeoic could have been different languages.


An Byeong-hui 安秉禧. 1987: ‘어학편’ (Linguistics) in 韓國學基礎資料選集古代篇 (A selection of basic sources materials for Korean Studies: ancient period), pages 1019-22.  Seoul: 韓國精神文化硏究員 (Academy of Korean Studies).

Kim Bang-han 金芳漢. 1983: 韓國語系統 (The Korean language family). Seoul: 民音社 (Mineumsa).

Kim Su-gyeong 김수경. 1989: 세나라시기 언어력사에 관한 남조선학계의 견해에 대한 비판적 고찰 (A critical examination into the opinions of South Korean academia concerning the linguistic history of the Three Kingdoms period).  Pyongyang: 평양출판사 (Pyeongyang chulpansa).

Kōno Rokurō (河野六郞). 1945: 朝鮮方言學試攷 (Experimental study on Korean dialects). Seoul: Tokyo Shosekirin (東都書籍林).

1993: 三国誌に記された東アジアの言語および民族に関する基礎的研究 (Basic research concerning the languages and peoples of East Asia recorded in the Sanguozhi ). 平成2-3-4年度 科學硏究費補助金 一般硏究(B) 硏究成果報告書, 1993)

Lee Gi-mun 李基文. 1981: 韓國語形成史 (Evolution of the Korean language). Seoul: Samseong Munhago (三星文化文庫 )

Ryu Ryeol 류렬. 1983: 세나라시기의 리두에 대한 연구 (Research on the ridu of the Three Kingdoms period). Pyongyang.

Song Gi-jung 송기중. 2004: 역사비교언어학과 국어계통론 (Historical comparative linguistics and theories on the genealogy of the Korean language). Seoul: 집문당 (Jimmundang).

[1]水城郡 本高句麗買忽郡 景德王改名 今水州.

[2] 南川縣 一云南買.

[3] Taken from Lee 1981:72

[4] Taken from An 1987:1019-22.

See also:

[5] Kim 1989:45-48 and Ryu 1983.

[6] Kim 1983: 114

[7] See Lee 1981

[8] My tenses are inaccurate here.

[9] These are only the modern Korean readings of the hanja.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s