Scribal Notation in Medieval Chinese Manuscripts: The hewen ...

GALAMBOS | SCRIBAL NOTATION5ArticleScribal Notation in Medieval Chinese Manuscripts:The hewen (Ligature) and the chongwen (Duplication)MarksImre Galambos | LondonEarly Chinese manuscripts and inscriptions often make useof two devices referred to by modern researchers as hewen (ligature) and chongwen (duplication). Both ofdashes which are placed below the lower right corner of thecharacter. The mark resembles the character written in asmall script, similar to what we would today call a subscript. 1Since the notation is identical in both cases, it is the contextthat determines whether it marks a joint character or a repetition.cle-bonerecords but their heyday was during the centuriesBC 8th–3rd. While their use in inscriptional material up tothe Han is relatively well-studied, there is almost no treatmentof it with regard to paper manuscripts, especially onesfrom the post-Han period. 2 In this article, I would like touse the Chinese manuscripts from Dunhuang and Turfanto demonstrate the application of this notation during themedieval period. This has added relevance because, althoughthe continuity of orthography and its transitions fromearly China to the medieval period has been fairly well researched,the secondary or peripheral aspects of writing, suchas the marking of repetitions or the notation used in editingand correcting mistakes, have received little attention.Hewen (ligature)Hewen is what modern researchers call a scribal device usedon early manuscripts where two or more adjoining charactersare united into a single composite graph. A parallel phenomenonin Western manuscript studies is the ligature, which iswhen two consecutive letters are combined in such a manner1 Whether this notation is actually related to the character is open to debate.Since such a connection is yet to be proven, I am, at this point, hesitantto make a definite identification of this mark with any particular characterin the Chinese script.2 For examples of hewen on oracle-bone inscriptions, see Qiu 1992a and1992b. On the same phenomenon on Warring States seals, see Wu 1989; onbronze inscriptions, see Shen 2002.that one or both lose their normal form to a greater or lesserdegree. 3The joint graph appears in the text as a single entity and isunpacked into its original components by the reader, whoreads and pronounces it as a multisyllabic string. Strictlyspeaking, hewen is a graphical device without any direct indicationof phonetic changes; it is read as the combination ofits original component graphs, and is pronounced as if thesewere written out in full. Of course, it is also possible that, atleast in some cases, the hewen also represented a phoneticabbreviation but we do not currently have any evidence forthis. 4Hewen was relatively common in pre-Qin times but almostcompletely disappeared in later periods. It used to bemarked with two short parallel strokes added below the lowerright corner of the graph. Generally speaking, this device wasused for characters that commonly occurred together, even ifthe words they represented did not form a grammatical unit.For example, the characters appear in the Houma covenanttexts (ca. 490 BC) as , while the characters inthe Baoshan bamboo strips (ca. 320 BC) as . In the Kongzishilun manuscript (ca. 300 BC) in the ShanghaiMuseum collection, the characters are written togethersharing their horizontal stroke as a single constellation of .In each of these cases, the reader is alerted with the hewenmark at the lower right corner of the graph. Technicallyspeaking, writing the two characters this way was not an abbreviation,since even if the scribe economized one strokein the characters themselves, he still had to write two moreto indicate the omission. Instead, in continuous text, it wasperhaps more of an indication that these characters appearedtogether frequently, even if the words they stood for did not3 Lowe 1980, vol. 1:158.4 The use of ligatures in Western liturgical traditions (e.g. Church Slavic)also points to the predominantly graphical nature of this device.NEWSLETTER N O 2manuscript cultures

6GALAMBOS | SCRIBAL NOTATIONFig. 1: A fragment from the Houma covenant texts. Even within this short text,the hewen with the word occurs three times, most clearly in the lastline as in the phrase (‘down to his descendants’). Another hewencombination on the same fragment is the name of the city of Handan appearing asFig. 2: Two sections from manuscript Or.8210/S.238, showing the compositecharacters and the empty space left before them. On the left image we can seethe name written as a single unit; and on the right one, the name form a compound. In other words, the words written by jointcharacters do not always form a semantic unit and their relationshipis simply that of collocations.An interesting type of hewen was when one of the twooriginal characters structurally already included the other.For example, in the Houma covenant texts (ca. 490 BC) weoften see the form , which stood for the characters (Fig. 1). From a structural point of view, this was only thecharacter with the hewen sign underneath, alerting thereader that some sort of duplication was at play. Since the orthographyof the character already incorporated the character as a component, was enough to represent bothof them. Examples of similar usage from the Warring Statesperiod are the characters written as on seals, or written as on so-called Chu manuscripts.In Western Zhou and Warring States periods, when theuse of hewen and chongwen was most common, the notationfor both of these devices was identical: a small double-notchsign placed underneath the lower right corner of the character.In both cases, the mark indicated a doubling: either thattwo characters have been joined together, or that one was tobe read twice.In medieval Chinese manuscript culture, the use of hewendiffered markedly from that seen in pre-Qin manuscripts.-being grouped together. On manuscript Or.8210/S.529, aseries of letters of introduction dated from 9th-10th centuries,the name of monks Guiwen and Dequan arejoined into single graphs as and , respectively. Whilethe other characters in the manuscript are distinctly separatedfrom each other, these names appear written togetheras single entities. The obvious reason behind writing nameslike this would be to treat them as a whole, lending them anemblematic quality.Manuscript Or.8210/S.238 5 with a Taoist text called bearsa colophon dating the document to 692 AD. There are twois the name Yuanfu written as , appearing in thestring The interesting phenomenon is thatan empty space stands before the composite character on themanuscript, as if indicating that, if not for the name taboo,part of the graph to follow would have actually been writtenin that space (Fig. 2). In other words, there is enough physicalspace left for unpacking the joint character. The sametext has also been preserved in the Taoist Canon, only there5 All manuscripts from the Stein collection in the British Library, beginningwith ‘Or.8210’, are taken from the IDP website: http://idp.bl.uk.manuscript cultures NEWSLETTER N O 2

GALAMBOS | SCRIBAL NOTATION7the characters appear without observing the taboo. The secondexample within the same manuscript is very similar. In thiscase, it is the name Yinjing in the string written as . In the new character, the character isfully present but is missing its radical, thus in this case thefusion also involves an abbreviation.It is clear from the context that, in this case, the compositecharacter was used as a means of observing a name taboo fordeceased masters. This was very similar in nature to the nametaboo of imperial names during a given dynasty. The characters and in the Tang dynasty, for example, were routinelywritten without their last stroke due to the fact that they occurredin the personal name of Li Shimin , the founderof the dynasty. It was also a common practice to replace thesetwo characters with and writing more or less synonymouswords. In the case of the two composite characters seenin the Taoist manuscript Or.8210/S.238 above, the name taboowas observed by writing the names of the late masters togetheras single units, and by leaving an empty space before the jointcharacter. Needless to say that while the joint graphs functionas a hewen, their use and application are very unlike thoseseen on pre-Qin manuscripts. 6At the same time, there is a small number of words thatoccasionally appear in Buddhist manuscripts from Dunhuangin the form of hewen. The most common of these is the graphstanding for the word pusa (). 7 Anotherjoint graph is for the word puti (), written as ,or . The last form here overlaps in structure with howthe word pusa was abbreviated and can be distinguishedonly with the help of the context. A somewhat less frequentexample of hewen in the Dunhuang material is the wordniepan () which was sometimes written as ,or . 8 Now it is apparent that all three examples are Buddhisttechnical terms and in this sense their usage is closelyreminiscent of Western ligatures. It is perhaps significant thateach of these three words was a transliteration of a Sanskritterm and because of this their individual component charactershad no semantic significance. Another important aspect is thatthese forms never appear in sutras but only in non-canonical6 It is also worth mentioning here the Daoist tradition of combining charactersinto elaborate talismans (fu ). Such graphic constellations, however, arestrictly speaking not part of writing habits, and shall not be considered here.7 The following examples are from Huang 2005.8 Interestingly, the late 10th century dictionary ,compiled on the basis of Tang-Song Buddhist manuscripts, identified thesehewen graphs as consisting of two separate sounds. For the graph , forexample, it says, ‘pronounced mang, meaning the abundant growth of vegetation;also pronounced as the two characters pusa’ , revealing that it identifies the same graph both as avariant form of the character and as a hewen for the word pusa.texts, such as commentaries or transformation texts. 9 Thisshows that the hewen forms were not accepted as standardforms and were banned from canonical usage.Beside the above examples, there are also the cases of thegraphs (twenty), (thirty), and (forty) which werecommonly used in medieval Chinese manuscripts, althoughnot limited to them. While some researchers believe thatthese were read as two-syllable words in medieval times, andthus represented true cases of hewen, there is also evidenceto the contrary. For example, the celebrated Song dynastyscholar Hong Mai (1123–1202) described how, in transmittedsources, the odes on the First Emperor’s steles werecomposed in four-character units, except when a date was). 10 When a fragmentof one of the steles was discovered, it became clear thatthe numbers in the dates had originally been written with thejoint form (e.g. ) and thus did not violate the tetrasyllabicprinciple. Of course, this also means that these jointcharacters were read as a single syllable, at least during theQin, and because of this they should be considered charactersin their own right, rather than hewen combinations.The above cases are the types of hewen that occur in me -dieval Chinese manuscripts. An important difference fromearly usage is that hewen in Dunhuang and Turfan is nevermarked. Although there are many different kinds of notationsfor repetition, deletion or insertion of characters, it wasnot considered necessary to indicate composite characters inwriting. The obvious reason for this was that, in contrast withthe use of hewen in early Chinese manuscripts, combinedcharacters in medieval practice made up meaningful units(e.g. words, names). Accordingly, there seems to be no directevolutionary connection between the use of hewen in pre-Qinand in medieval times.ChongwenIn pre-Qin manuscripts, the hewen mark was identical to thatof chongwencharacters were fused into a single unit, in the second it indicatedthat a character or a string of characters was to be readtwice. The reader had to rely on the context to determine howhewen, thechongwen device in medieval manuscripts suggests a directconnection with the pre-Qin one. In both its function and ap-9 A similar pattern can be observed with regard to the use of the graph , anon-standard form of the character (Buddha). This form, coinciding withthe modern Japanese way of writing the same character, was never used insutras, only in non-canonical Buddhist texts.10 Hong 1978:69–70.NEWSLETTER N O 2manuscript cultures

8GALAMBOS | SCRIBAL NOTATIONFig. 3: Section of manuscriptOr.8210/S.1547 showing the use ofthe repetition mark. The characters in the first line and inthe second are repeatedFig. 4: Repetition marks for multi-character chongwenA: Or.8210/S.2067 B: 80TBI:009 C: Or.8210/S.116pearance, it remained practically unchanged, as abundantlymanifested in the Dunhuang and Turfan corpora.Principally speaking, there are two kinds of chongwen:single and multiple ones. In the first type, only one characteris repeated, whereas in the latter two or more. Whilethis may seem a trivial distinction, the notation for these inactual usage was somewhat different. The single characterrepetition is simply marked by a small mark in place ofthe second character. This mark was sometimes written asor , and probably derived from the pre-Qin chongwenmark. Nevertheless, the form is by far the most commonin Dunhuang and Turfan. An important difference betweenearly and medieval usage was that, in the latter, the mark wasplaced within the main text, in place of the omitted secondcharacter. In this way, the repetition mark occupied a fullcharacter space.Or.8210/S.1547, for example, is a manuscript of theChengshilun (*) dated to 512 AD.At the very end of the scroll, we find the following two sentences(given below in modern punctuation):Such a man is like a fire burning the firewood: once the firewoodis exhausted, it will become extinguished; this man isalso like this; because he receives no more, he becomes extinguished.If he extinguishes the three minds, he will attaineternal liberation from all sufferings.chongwen:first the character and then a bit later the character . Ineach case, the second characte r is omitted and a mark isplaced in its stead (Fig. 3). Although the characters do notform a single unit in the text grammatically (i.e. , or ) and, in a modern punctuated transcription, are separatedfrom each other by a comma or a period, this did not stop themedieval scribe from applying the chongwen device purelybased on their physical adjacency.In multi-character repetition, the chongwen mark is placedeither underneath the character or at its lower right corner. Anexample of the former usage is Or.8210/S.2067 (Fig. 4/A)where the characters (indescribable or unspeakable)are repeated in the phrase indescribable and indescribablemyriads of sentient beings . What makesthis case different from the single chongwen seen above is thatthe three characters are to be read together and only then repeatedas a string. At least theoretically, it would be possibleto read them repeated one by one as but thiswould produce a meaningless string of characters. Therefore,the context is used by the reader for disambiguation.manuscript cultures NEWSLETTER N O 2

GALAMBOS | SCRIBAL NOTATION9The other way of marking multi-character chongwen canbe seen on manuscript 80TBI:009 discovered in Turfan (Fig.4/B), where the device is marked with a slanted double dashunderneath the lower right corner of the character. The sectionshown on the picture contains two such cases: in the firstline, we find Buddha’s habitual exclamation (Excellent, excellent!’); and in the third line, thewords (‘monks and nuns’) arewritten with the characters marked as having to be readtwice. This latter case is a wonderful example to show thatthe chongwen device is completely unrelated to the grammaticalstructure of the text and it relies solely on the physicalposition (i.e. adjacency) of characters. In other words,chongwen appears to be concerned only with characters, notwords or sentences.In addition, the doubled chongwen mark is sometimeswritten as a single slanted stroke, as can be seen in manuscriptOr.8210/S.116 (Fig. 4/C), where the word niepan () is marked in this way. In this particular case,the repeated word occurs at the end of one sentence andthe beginning of another: ‘This is why it is called the Great. In there is no pleasure...’ . In the first sentence, the word is actually ‘Great’is semantically not completely parallel. In this respect, thisusage is similar to that of the words (‘monks and nuns’).Looking through concrete examples of chongwen, it is apparentthat the device was optional in medieval manuscripts.Even in documents where it occurs, there are places of oftenidentical context where it is not used and the characters are‘spelled out’ in their full form. In fact, there are fewer caseswhere chongwen is used than where it is not—it is rather anexception than the norm.REFERENCESHong Mai (1978), (Shanghai:Shanghai guji chubanshe).Huang Zheng (2005), (Shanghai: Shanghai jiaoyu chubanshe).Lowe, Elias Avery (1980), South Italian miniscule (Roma: Edizioni di storia e letteratura).Qiu Xigui (1992a), ‘Jiaguwen zhong chongwen he hewenchongfu pianpang de shenglue’ ‘’, in Qiu Xigui, . (Beijing:Zhonghua shuju), 141–146.–––– (1992b), ‘Zai tan jiaguwen zhong chongwen de shenglue’ ‘, in Qiu Xigui, (Beijing: Zhonghua shuju), 147–150.Shanxi sheng wenwu gongzuo weiyuanhui (1976) (ed.), Houma mengshu (Shanghai: Wenwu).Shen, Baochun (2002), ‘Xi Zhou jinwen chongwen xianxiangtanjiu: yi “Yinxu jinwen jicheng” guilei chongwen wei li’: “”’ , 24: 307–11.Wu Zhenwu (1989), ‘Guxi hewen kao (shiba pian)’ ' 17: 268–281.Xinjiang Weiwu’er Zizhiqu Tulufanxue yanjiuyuan , Wuhan daxue zhongguo san zhijiu shiji yanjiusuo (2007)(eds.), (Beijing: Wenwu).SummaryThe use of hewen and chongwen devices in medieval manuscriptsis interesting from the point of view of the transmissionof scribal practices in Chinese history. While the hewenin the medieval corpus shows no similarity to Warring Statesusage, chongwen remains a common phenomenon and ismarked in a similar way as it was fifteen hundred years earlier.This observation is significant because we do not haveany evidence that such scribal techniques would have beentaught. They are certainly absent from the linguistic treatises,dictionaries and primers describing some of the more obviousfeatures of the script (correct character forms, variants,etc). Therefore, the analogous use of the chongwen deviceand its similar notation arguably demonstrate a direct continuitybetween pre-Qin and Tang-Song manuscript culture.Picture credits:Fig. 1: Tracing from Shanxi 1976: 264.Fig. 2: International Dunhuang Project (http://idp.bl.uk),Or.8210/S.238. Copyright of the British Library.Fig. 3: International Dunhuang Project (http://idp.bl.uk),Or.8210/S.1547. Copyright of the British Library.Fig. 4: Images A and C are from International Dunhuang Project(http://idp.bl.uk), shelfmarks Or.8210/S.2067 and Or.8210/S.116.Copyright of the British Library. Image C is from the publicationXinjiang 2007: 11.NEWSLETTER N O 2manuscript cultures