“Graphic Variation, Modification, and Replacement in Medieval Chinese Writing: Case Studies and Resources” Workshop Report

By Laurent Van Cutsem, Ghent University

While an increasing number of texts have been made electronically available to researchers in the past decades, the ability to read and engage with manuscripts, woodblock prints, and epigraphic materials remains an invaluable skill and source of enrichment for scholars of medieval Chinese religions. Held as part of the FROGBEAR Summer 2021 Training Sessions, the workshop “Graphic Variation, Modification, and Replacement in Medieval Chinese Writing: Case Studies and Resources” convened virtually on August 17 and November 30, 2021 to guide young scholars in this task. Based on their respective expertise, Dr. Christoph Anderl (Ghent University), Suzanne Burdorf (Ghent University) and Dr. Lia Wei (Institut National des Langues et Civilisations Orientales) lectured on topics related to variability of character forms and phonetic borrowings in Medieval Chinese sources, together with a series of exercises. The discussions centered on: (1) Dunhuang 敦煌 manuscripts, (2) the Longkan shoujian 龍龕手鑑 (originally titled Longkan shoujing 龍龕手鏡; Hand Mirror of the Dragon Shrine, ca. 997), and (3) rock carvings and epigraphic materials of the sixth century, primarily from An Daoyi 安道壹 (fl. 562–579), also known as Seng An Daoyi 僧安道壹.

Below, I summarize and discuss some of the concepts that were introduced or mentioned during the workshops in the hope that this report article may be of help to researchers who wish to embark in the study of medieval Chinese textual sources.

Christoph Anderl (Ghent Univ.), Suzanne Burdorf (Ghent Univ.) and Lia Wei (Inalco). Screenshot courtesy of Carol Lee (UBC Frogbear). Republished with permission.

Research on graphic variations in the Medieval period is a relatively complex endeavor due to the different processes at play in the formation or adaptation of characters and the importance of preliminary methodological considerations.

First, the concept of variation can be approached from a descriptive or a prescriptive (or normative) point of view. The descriptive approach, as the name suggests, does not presuppose any pre-existing norm and does not aim at establishing a standard against which variant orthographic structures are measured. Its primary objective is to describe characteristics of the Chinese script and characters as they appear in the corpus studied. Therefore, variation is generally understood as an integral part of the Chinese writing system. The prescriptive approach, on the other hand, either proposes a standard based on the observations of the variations in the corpus studied or identifies at the outset one or several sources as the normalized expression of orthographic structure and studies variations in contrast to this norm. Historically, both approaches are to some extent attested, as has been discussed, for example, by Imre Galambos in his study on orthography in early Chinese writing.[1] In this respect, there is a significant difference between, on the one hand, the identification of graphic variants on the basis of various official modern standards, which is the guiding principle of the Database of Medieval Chinese Texts (Ghent University and Chung-Hwa Institute of Buddhist Studies 中華佛學研究所; henceforth DMCT)[2] and the Jiaoyubu yitizi zidian  教育部異體字字典 (Dictionary of Variant Characters of the Ministry of Education; henceforth YTZZD),[3] and, on the other hand, the perspective of historians of the script, art historians, and artists, as rightfully noted by Lia Wei.

Second, variability of character forms can either be studied synchronically or diachronically. The first method generally focuses on variations across regions which can be defined from a historical (e.g., political administrative entity, socio-economically cohesive region) or geographical (e.g., the Zhongyuan 中原 area, the Jiangnan 江南 region) perspective. However, variations also occur within local communities of scribes, the corpus of one copyist, and even within a single historical textual witness (e.g., stone inscription, scroll, printing block). In the Goryeo 高麗 woodblock edition of the Zutang ji 祖堂集 (Collection of the Patriarchal Hall, second half of the tenth century), for example, it is not rare to find both 於 and 扵 on the same printing surface.[4] The diachronic approach, in contrast, usually shifts the focus either to the evolution of the Chinese script in general or to specific aspects of the script (e.g., evolution of character structure, graphic representations of mono- or polysyllabic words, regional developments) over a certain period of time.[5]

Eventually, another important methodological consideration raised during the lectures of Christoph Anderl, Suzanne Burdorf, and Lia Wei is that a preliminary definition of the concept of variation, including its diverse modalities, is needed in order to approach the materials studied in a systematic manner. Naturally, these preliminary definitions will be partly conditioned by the method adopted (descriptive or prescriptive, synchronic or diachronic) and they should remain open to adjustments in accordance with the analysis of the sources. Broadly speaking, two general trends appear to prevail in pre-modern and modern scholarship.

Traditionally, orthographic variability has been understood in light of a presumed “standard” or “proper” character (zhengzi 正字) from which variant forms would depart. For instance, when the Longkan shoujian notes “yan su, yan zheng, yi xian fan 烟俗,煙正,伊賢反[…]。” (“烟 is the demotic form, 煙 is the standard form; having the fan[qie] reading 伊賢”), it is likely that its author, the Liao 遼 monk Xingjun 行均 (d.u.), estimated based on his sources that 煙was the appropriate or standard character form (zheng 正) to write the word /ʔɛn/, “haze; smoke,”[6] and that 烟 was a later demotic or “vulgar” form (su 俗). Interestingly, however, the understanding of a set of variant character forms in a given historical lexicographical source does not necessarily coincide with that of another. For example, the Ganlu Zishu干祿字書 (Character Book for Seeking Official Emolument, 774), originally authored by Yan Yuansun 顏元孫 (d. 732?), regarded both 煙 and 烟 as standard characters (“yan, yan, bing zheng 煙、烟,並正。” (“煙 and 烟 are both standard [forms]”; see YTZZD). This example, by itself, already raises a number of questions: did the standard change between the period of compilation of the Ganlu zishu and that of the Longkan shoujian? Was the notion of standard different in their respective regions of compilation? What does it mean to have two standard written forms of the same word? Is there an underlying norm or reference despite the apparent discrepancy in the terminology of the two works? And so forth.[7]

More recently, structural variability of character forms has been examined in regard to the word that they stand for.[8] In other terms, historically attested variants are seen as graphic realizations of a “cloud of potential forms” of a given word, to borrow Galambos’ expression,[9] some of them being frequent, others being rare. To take the same (theoretical) example, if the character forms 煙, 烟,  (A02420-003) or  (A02420-005) are attested, as they supposedly were in the Yupian 玉篇 (Jade Chapters, ca. 543), among the manuscripts produced by a local community of scribes, these can be viewed as visual representations of the same word. In this perspective, less emphasis is therefore placed on whether or not one of these character forms was regarded as the appropriate form according to the contemporary orthographic standard. Rather, the centrality of the relationship between the word and its graphic representations is restated, a phenomenon which is attested by the fact that, in orthographic variability, the most stable component of the graphic representation of a word is generally its phonetic or phonophoric component.[10]

Methodological considerations aside, research on graphic variants is further complicated by the profusion of technical terms related to orthographic variability (zhengzi 正字, yitizi 異體字, cuozi 錯字, biezi 別字, suzi 俗字, jianhuazi 簡化字, leihuazi 類化字, guzi 古字, jinzi 今字, houqizi 後起字, etc.) and their nebulous relations to one another. Below, some of these terms are discussed in connection with the ideas exchanged during the workshop.

First, in the Hanyu dacidian 漢語大詞典 (Dictionary of the Chinese language), a zhengzi 正字 is defined as “a character whose form or orthograph (i.e., component-level structure) is in accord with the standard.” The definition continues, indicating that “it is distinguished from yitizi 異體字, cuozi 錯字, biezi 別字, etc.” and that the term zhengzi“can also refer to benzi 本字.”[11] The first part of the definition is not overly problematic, although we may wish to clarify that the notion of “standard” could be understood in very different ways. For instance, a standard can be established by a given social group or individual who acts as the source of authority at a given point in time and space. Alternatively, a standard can be determined by the custom, that is, the tacit norm in use among a community. By definition, the standard is therefore susceptible to change, and it is possible for several standards to coexist historically. Eventually, as pointed out by Christoph Anderl, we do not always know what the standard was in a certain region at a given time and new colloquial words that surfaced during the Tang, for example, could not possibly have an established written standard form. A well-known example is the development of the interrogative pronoun shenme 什麼 (“what?”) and its various graphic representations (e.g., 是物, 是沒, 甚沒, 甚物, 是摩, 甚摩, 什摩).[12]

The interrogative pronoun shenme 什麼 (“what?”) and its various graphic representations. Slide courtesy of Christoph Anderl. Republished with permission.

Traditionally, yitizi 異體字 (variant characters or graphic variants) are defined as “characters whose pronunciation and meaning are identical but whose forms differ; i.e., [character categories] such as suti 俗體, guti 古體, huoti 或體.”[13] Strictly speaking, a variant character therefore shares the exact same usage as its presumed zhengzi. For example, 躰and 体can be regarded as graphic variants of ti 體 because their pronunciation and signification are identical to that of 體.[14] In contrast, as Qiu Xigui 裘錫圭 indicates, characters such as 鵰, 彫, 琱 and 凋, which have occasionally been understood as graphic variants of diao 雕, should in theory be considered variants of 雕 in certain cases only.[15] As evidenced by the definition given above, the term yitizi does not specify the nature of the graphic variation in question and for this reason it is sometimes conceived as a general term that encompasses more specific expressions such as suzi 俗字(demotic or popular character) or guzi 古字 (ancient or archaic character).[16]

According to Zhang Yongquan 張涌泉, a suzi 俗字, also called sutizi 俗體字, is a “popular character form that, in comparison with the standard character, primarily circulates among the common people, in any particular period in the history of Chinese characters.”[17] In line with the description given by Yan Yuansun in his Ganlu zishu, the “popular” nature of these character forms chiefly refers to the fact that they were conceived as being simpler than their standard counterparts and that they were mostly in use in more popular forms of writings (or informal situations).[18] “Simpler,” in this case, does not necessarily mean that a suzi was written with less strokes than its contemporary zhengzi. For instance, Zhang Yongquan indicates that the tai 胎 of luntai 輪胎 (“tyre”) is occasionally written with the suzi [車*太] in Guangdong province 廣東省. In this example, the suzi is in fact graphically more complex but it is considered to be more transparent from a semantic and phonetic point of view.[19] This is why suzi and jianhuazi 簡化字 (simplified characters) should be distinguished, although simplified forms can indeed often be placed in the category of suzi. As for the earlier term biezi 別字 (other or distinct character), it appears to be more or less synonymous with suzi, although their context of use differs.[20]

Eventually, together with zhengzi and suzi, a third category which is frequently encountered in lexicographical sources such as the Ganlu zishu and the Longkan shoujian is that of tong 通 (also tongyong 通用, tongsu 通俗, etc.). As rightfully noted by Suzanne Burdorf, character forms that are classified in this category are not so different from suzi. What her research shows is that these characters can generally be identified as demotic forms that have been in use for a longer time and whose range of use is both more widespread and accepted than the suzi, thereby corroborating Zhang Yongquan’s earlier observations.[21]

In medieval manuscripts, many processes were at play in the formation or adaptation of character forms. Common aspects of orthographic variability include: the addition or omission of the semantic element or classifier (e.g., 皃/貌),[22]rearrangement of the components (e.g., 鑑/鑒), abbreviated forms (e.g., 雖/虽, 聲/声), alteration of a component (e.g., 祕/秘, 珍/珎), and so forth.[23] Interestingly, according to Galambos, “the largest number of new structures seems to originate from handwriting habits related to the cursivized writing of characters.”[24] Two categories which diverge from this, however, were discussed in more detail during the sessions of Christoph Anderl and Suzanne Burdorf: (1) leihuazi 類化字 and (2) huiyizi 會意字.

The category of leihuazi 類化字 (lit. “analogically formed character”) covers character forms that result from a process of graphic assimilation, that is, in most cases, the adaptation of a character under the influence of another character. As one would expect, this process of character “formation” or adaptation is particularly frequent in disyllabic words. For example, in the Quanzhou Qianfo xinzhu zhuzushi song 泉州千佛新著諸祖師頌 (Eulogies newly composed by Qianfo [Deng] of Quanzhou; Or.8210/S.1635), Caoxi 曹溪 (a toponym which is commonly used to refer to the sixth patriarch) is written 漕溪.[25] In this case, the alteration of the original character form (曹) through the addition of the classifier 水/氵 coincides with another pre-existing character (漕). However, although both characters shared the same Middle Chinese pronunciation (L. tsɦaw, E. dzaw),[26] the nature of this type of variation, as indicated by Galambos,[27]is graphic and should probably be distinguished from the phenomenon of phonetic borrowing (see below). With this in mind, one should add that the formation of leihuazi is not unidirectional (e.g., the preceding character influences the following one) and that it may involve relatively complex processes. These have been discussed, notably, by Zhang Yongquan, Qiu Xigui, and Galambos.[28] Other common examples include: putao 葡萄 (蒲萄, 蒲陶), furong 芙蓉 (夫容), fenghuang 鳳凰 (鳳皇), and so forth.[29] In my own research, I have observed that leihuazi may coexist with their standard counterparts. For instance, in the Zutang ji’s entry of Nanyang Huizhong 南陽慧忠 (trad. 675–775), juan 3, on the second half of zhang 張 (printing surface) no. 7, one finds two occurrences of 曹溪 and two of 漕溪 (cf. table below). More specifically, on column 22, both 曹溪 and 漕溪 appear, separated by only five characters.[30]

Examples of variant character forms of 曹 in the disyllabic word Caoxi 曹溪 in the Zutang ji:

Source: Zutang ji, Kyōdai jinbunken 京大人文研print. In Database of Medieval Chinese Texts (forthcoming).

The second category discussed in more detail during the workshop is that of huiyizi 會意字 (syssemantic characters, syssemantographs or semantic compounds). As pointed out by Imre Galambos, this traditional category, which can be traced back to the liushu 六書 system,[31] has been the subject of much debates among scholars.[32] Briefly put, the formation of characters according to the huiyi principle corresponds to the juxtaposition of two or more components whose semantic value together contribute to the meaning of the morpheme written by the compound.[33] In Dunhuang manuscripts and lexicographical materials, the formation of characters based on the huiyi principle is a relatively well-attested phenomenon. For example, the character form 甦, composed of the semantic graphic elements geng 更 (“again”) and sheng 生 (“to live”) is a demotic form of the word su 蘇 (“to revive; resuscitate”) formed on the basis of the huiyiprinciple. Similarly, mi 覓 (“to seek out, look for”) is occasionally written 覔, composed of the elements bu 不 (“not”) and jian 見 (“to see”).[34]

Some characters in this category overlap with the so-called “portmanteau characters.” According to David P. Branner, these are “a composite of two or more graphs for living words, all of which are to be read (in order) to give the meaning of the word represented by the whole character.”[35] To give an example which is attested both in Dunhuang manuscripts and modern context, the colloquial modal auxiliary 甭 (“to not need to”) is composed of the components bu不 and yong 用 which, read syntactically, give “not need to.” In modern Chinese, the pinyin of 甭 is given as beng(second tone) and conceived as a contraction of the pronunciation of its components bu and yong. In the Longkan shoujian, however, the character is said to be pronounced as qi 弃 (L. kʰji`, E. kʰjiʰ):[36] “甭, 音弃.”[37] Despite a certain similarity with the traditional category of huiyizi, however, Branner suggests that portmanteau characters should be distinguished from these because the semantic value of their components are not “abstract” but should be read “as connected words to form a phrase that defines or denotes the word.”[38] Returning to the examples given above and examined during the workshop, if we follow Branner’s definition, 甦 could perhaps be regarded as a portmanteau character. However, it is doubtful that 覔 should be regarded as such.

One interesting example that scholars who are familiar with the Zutang ji might know of is the complex graph that CBETA gives as [企-止+(山/(虫*見))] and which occurs in the entry of Niutou Farong 牛頭法融 (trad. 594–657) in juan3.[39]

Source: Zutang ji, Kyōdai jinbunken 京大人文研print. In Database of Medieval Chinese Texts. (forthcoming)

This character, whose pronunciation is uncertain, appears to be readable syntactically as ru shan jian chong 入山見虫 (lit. “to enter the mountain and see a tiger”).[40] However, its meaning in the context of the passage is probably close to an interjection (gantan ci 感嘆詞) expressing surprise or fear, perhaps equivalent to “eek!” or “yikes!”[41] In their editions of the Zutang ji, Zhang Hua 張華 refers to it as a hetizi 合體字 (“composite character”), and Sun Changwu 孫昌武 et al. use the term heti 合體, while Zhang Meilan 張美蘭 uses hewen 合文, a relatively ambiguous term because it is also commonly used to refer to “ligatures,” that is, another way of character formation in medieval manuscripts.[42]

Although the examples discussed above are mostly taken from medieval manuscripts of the Tang to Northern Song periods, Lia Wei showed that graphic variants were also a common feature of epigraphic materials of the sixth century, such as steles and epitaphs, but also cliff carvings. Her contribution was most welcomed because epigraphic materials allow us to reflect in a different way on the relation between calligraphic styles, whether widely used or personal, and the concept of orthographic variability.

From the work of An Daoyi (and continuators of his style?), Lia Wei distinguished several formal characteristics of his style, some of which are related to the concept of orthographic variability. These include: (a) character forms of which one or more components are written in seal script; (b) character forms which include figurative elements (e.g., abstract drawings of hands, the moon); (c) character forms that are carved with brush-like effects (e.g., the “flying white script” or fei baishu 飛白書, brush-shaped dots); (d) character forms carved with specific sculptural effects (e.g., deeply carved strokes, “relief strokes”); and (e) character forms which have an unfinished aspect. To this list, one should also add character forms which are written with another semantic element for conscious stylistic reasons (e.g., 凊 for 清).[43]These cases are all useful to consider in order to reflect on our definition of what constitutes “variations” in the Chinese writing system.

Inscription of the character ban 般 at various sites. Slide courtesy of Lia Wei. Republished with permission. (forthcoming)

As emphasized by Lia Wei, orthographic and stylistic variations in epigraphy are intimately connected to the process of carving and the materials onto which a text is carved. In this respect, cliff carvings such as those of An Daoyi are different from stele inscriptions that were carved after a written model. As evidenced by the Jingshiyu 經石峪 (“Stone Sūtra Valley”) site of Mount Tai 泰山, his carvings follow the relief of the cliffs, showing adaptability and conciliation between the act of creation and the natural shapes of the rocks. Far from being anecdotic, this should invite us to reflect on similar processes with regard to manuscript culture (e.g., brushes, inks, papers), and it certainly reinforces earlier observations on the connection between cursivized writing habits and graphic variation in manuscripts.

One last complex but relatively frequent textual phenomenon encountered in Dunhuang manuscripts and epigraphic materials discussed during the workshop is the notion of “phonetic borrowing.”

As Imre Galambos pointed out, the English term “phonetic borrowing” is in fact used to translate two traditional categories: jiajie 假借 and tongjia 通假. The first term, like the category of huiyizi discussed earlier, can be traced back to the liushu system. Generally, it is understood in reference to a character which is borrowed to write a homophonous or nearly homophonous word for which no graphic representation previously existed.[44] For example, the character 耳(“ear”) was borrowed to write the sentence-final modal function word (i.e., particle) and contraction of eryi 而已.[45]

In contrast, tongjia or tongjiazi (commonly translated as phonetic loan character or loangraph) is used in reference to characters that are borrowed to write a homophonous or nearly homophonous word for which a character already existed.[46] A well-known example for scholars of Chan Buddhism is the use of the character 曆 to write the homophonous word whose expected standard character in Literary Sinitic would be 歷. For instance, the title of the Lidai fabao ji 歷代法寶記 (Record of the Dharma Jewel Through the Generations) is originally written with 暦 (A01822-001; YTZZD), a common variant of the graph 曆, in Or.8210/S.516.[47] Naturally, these two characters shared the same Middle Chinese pronunciation (L. liajk, E. lɛjk)[48] and 曆 is elsewhere frequently attested as a phonetic loan for 歷.[49]To give another example, in the praise verse composed by Chan master Jingxiu 淨修禪師 for Mahākāśyapa in his Quanzhou Qianfo xinzhu zhuzushi song, the copyist of the manuscript S.1635 wrote or copied “蜜傳佛心” (S.1635r_11), where 蜜 (in fact, a variant of 蜜) is a phonetic loan character for the word whose expected standard character in Literary Sinitic would be 密, “secretly.” In this case, 蜜 (L. mjit, E. mjit)[50] was likely a near homophone of 密 (L. mit, E. mit).[51]

According to Qiu Xigui, the reasons behind the use of tongjiazi are diverse. First, punctual examples which are rarely encountered may well simply be erroneous homophonous characters.[52] These errors, in turn, can happen for several reasons (e.g., fatigue of the copyist, the speed of the copying process, an involuntary association with another word in the context of the phrase copied). Second, phonetic loan characters may occasionally be used for the purpose of graphic simplification (e.g., 只 for 隻).[53] In several cases, phonetic loans were also used to avoid undesired associations or potential confusions due to (one of) the semantic value(s) of the expected graphs in Literary Sinitic (e.g., 原 for 元 in 元來 or 元由), and so forth.[54]

To conclude this report article, I will briefly mention some of the tools for research on graphic variants and/or phonetic loan characters in Medieval Chinese sources that were introduced during the workshop. Naturally, these resources only represent a small fragment of the scholarship that has been produced on this broad topic to this day, and researchers will gradually have to get familiar with more specialized tools according to their own research needs.

First, the three major online tools mentioned and used throughout the workshop were: (1) the Database of Medieval Chinese Texts (Ghent University and Chung-Hwa Institute of Buddhist Studies 中華佛學研究所), (2) the Jiaoyubu yitizi zidian 教育部異體字字典 (Dictionary of Variant Characters of the Ministry of Education), and (3) the Sekkoku takuhon shiryō 石刻拓本資料 (Resources on Rubbings of Stone Inscriptions) database of the Institute for Research in Humanities of Kyōto University 京都大学人文科学研究所.

In addition to these online tools, the Dunhuang suzidian 敦煌俗字典 (Dictionary of Demotic Characters from Dunhuang) of Huang Zheng 黄征 was introduced as an essential tool for research on graphic variants in the Dunhuang scribal context.[55] One should also add the 681-page-long annotated list of demotic characters (Dunhuang suzi huikao敦煌俗字匯考) contained in the authoritative Dunhuang suzi yanjiu 敦煌俗字研究 (Research on Demotic Characters from Dunhuang) of Zhang Yongquan.[56] In addition, because of the close relationship between graphic variation in Dunhuang manuscripts and cursivized writing habits, dictionaries of calligraphy are occasionally helpful as a general tool as well.[57]

Concerning phonetic loan characters, the Middle Chinese transcriptions or reconstructions of Baxter and Sagart and Pulleyblank are indispensable resources for the identification of homophonous characters.[58] Through its convenient collection of historical lexicographical sources, the YTZZD is also an invaluable tool for the Middle Chinese reconstruction or transcription of characters that are not recorded in the two works mentioned above. However, readers will need to be familiar with Chinese historical phonology.[59] To this, one should naturally add general or specialized dictionaries of phonetic loans.[60]

Eventually, with regard to secondary literature, the reader may wish to consult the annotated bibliography compiled by Imre Galambos,[61] and the items listed in the bibliography below.

*This report article was supported by the Chiang Ching-kuo Foundation for International Scholarly Exchange 蔣經國國際學術交流基金會 (DD010-U-20).


Adamek, Wendi L. The Mystique of Transmission: On an Early Chan History and Its Context. New York: Columbia University Press, 2007.

Anderl, Christoph. “Medieval Chinese Syntax.” In Encyclopedia of Chinese Language and Linguistics, edited by Rint Sybesma, 2:689–703. Leiden: Brill, 2017.

———. “Some Reflections on the Database of Medieval Chinese Texts as a Multi-Purpose Tool for Research, Teaching, and International Collaboration.” In Corpus-Based Research on Chinese Language and Linguistics, edited by Bianca Basciano, Franco Gatti, and Anna Morbiato, 6:341–60. Venezia: Ca’Foscari, 2020.

———, ed. “Database of Medieval Chinese Texts. Ghent University and Chung-Hwa Institute of Buddhist Studies 中華佛學研究所, 2021. https://www.database-of-medieval-chinese-texts.be/.

Baxter, William H., and Laurent Sagart. “The Baxter-Sagart Reconstruction of Old Chinese (Version 1.1),” September 20, 2014. http://ocbaxtersagart.lsait.lsa.umich.edu/.

Boltz, William G. The Origin and Early Development of the Chinese Writing System. American Oriental Series 78. New Haven: American Oriental Society, 1994.

Bottéro, Françoise. “Les Graphies Énigmatiques de l’Impératrice Wǔ Zétiān 武則天.” Études Chinoises 32, no. 2 (2013): 67–99.

Bottéro, Françoise. Review of The Origin and Early Development of the Chinese Writing System, by William G. Boltz. Journal of the American Oriental Society 116, no. 3 (July-September, 1996): 574–77.

Bottéro, Françoise, and Christoph Harbsmeier. “The Shuowen Jiezi Dictionary and the Human Sciences in China.” Asia Major 21, no. 1 (2008): 249–71.

Branner, David Prager. “Portmanteau Characters in Chinese.” Journal of the American Oriental Society 131, no. 1 (January-March, 2011): 73–82.

Chinese Buddhist Electronic Text Association 中華電子佛典協會. “CBETA Online Reader (Xianshang Yuedu 線上閱讀).” 2021. https://cbetaonline.dila.edu.tw/.

Galambos, Imre. “Medieval Ways of Character Formation in Chinese Manuscript Culture.” Scripta 6 (October, 2014): 49–73.

———. Orthography of Early Chinese Writing: Evidence from Newly Excavated Manuscripts. Budapest: Department of East Asian Studies, Eötvös Loránd University, 2006.

———. “Popular Character Forms (Súzì) and Semantic Compound (Huìyì) Characters in Medieval Chinese Manuscripts.” Journal of the American Oriental Society 131, no. 3 (July-September, 2011): 395–409.

———. “The Chinese Script.” Oxford Bibliographies in Chinese Studies. Oxford University Press, 2014. https://doi.org/10.1093/obo/9780199920082-0066.

———. “The Chinese Writing System.” In The Oxford Handbook of Classical Chinese Literature (1000 BCE–900 CE), edited by Wiebke Denecke, Wai-Yee Li, and Xiaofei Tian, 1:31–45. Oxford: Oxford University Press, 2017.

———. “Variant Characters.” In Encyclopedia of Chinese Language and Linguistics, edited by Rint Sybesma, 4:483–84. Leiden: Brill, 2015.

Guojia jiaoyu yanjiuyuan yuwen jiaoyu ji bianyi yanjiu zhongxin 國家教育研究院語文教育及編譯研究中心 and Guoyu tuixing weiyuanhui 國語推行委員會, eds. “Jiaoyubu yitizi zidian 教育部異體字字典 (6th ed.).” 2017. https://dict.variants.moe.edu.tw/.

Handel, Zev. Sinography: The Borrowing and Adaptation of the Chinese Script. Language, Writing and Literary Culture in the Sinographic Cosmopolis 1. Leiden: Brill, 2019.

Huang Zheng 黄征. Dunhuang suzidian 敦煌俗字典. Shanghai 上海: Shanghai jiaoyu chubanshe 上海教育出版社, 2005.

Jacques, Guillaume. “Traditional Chinese Phonology.” In Encyclopedia of Chinese Language and Linguistics, edited by Rint Sybesma, 4:376–91. Leiden: Brill, 2017.

Kinugawa Kenji 衣川賢次. “Sodōshū satsuki 祖堂集札記.” Zen bunka kenkyūjo kiyō 禅文化研究所紀要 24 (1998): 113–28.

Komazawa daigaku nai zengaku daijiten hensanjo 駒澤大學内禪學大辭典編纂所, ed. Zengaku daijiten 禪學大辭典. Tōkyō 東京: Taishūkan shoten 大修館書店, 1985.

Kyōto daigaku jinbunkagaku kenkyūjo 京都大学人文科学研究所. Sekkoku takuhon shiryō 石刻拓本資料. n.d. http://kanji.zinbun.kyoto-u.ac.jp/db-machine/imgsrv/takuhon/index.html.

Luo Zhufeng 羅竹風, Hanyu dacidian bianji weiyuanhui 漢語大詞典編輯委員會, and Hanyu dacidian bianzhuanchu 漢語大詞典編纂處, eds. Hanyu dacidian 漢語大詞典. 13 vols. Shanghai 上海: Shanghai cishu chubanshe 上海辭書出版社, 1986–1994.

Morohashi Tetsuji 諸橋轍次. Dai Kan-Wa jiten 大漢和辞典. Edited by Kamata Tadashi 鎌田正 and Yoneyama Toratarō 米山寅太郎. Revised ed. 15 vols. Tōkyō 東京: Taishūkan shoten 大修館書店, 1999–2000.

Nishihara Kazuyuki 西原一幸. “The Longkan Shoujing in View of the Tang Orthographic Standard of the Kai Script.” Kinjō gakuin daigaku ronshū 金城学院大学論集 9, no. 2 (2013): 84–95.

Okayama Hajime 丘山新, Kinugawa Kenji 衣川賢次, and Ogawa Takashi 小川隆. “Sodōshū Gozu Hōyū shō soshō 『祖堂集』牛頭法融章疏證.” Tōyō bunka kenkyūjo kiyō 東洋文化研究所紀要 139 (2000): 39–83.

Pulleyblank, Edwin G. Lexicon of Reconstructed Pronunciation: In Early Middle Chinese, Late Middle Chinese and Early Mandarin. Vancouver: UBC Press, 1991.

———. Outline of Classical Chinese Grammar. Vancouver: UBC Press, 1995.

Qiu Xigui 裘錫圭. Chinese Writing. Translated by Gilbert Louis Mattos and Jerry Norman. Berkeley: Society for the Study of Early China, 2000.

———. Wenzixue gaiyao 文字學概要. Edited by Xu Tanhui 許錟輝. Zaiban 再版. Taibei 臺北: Wanjuanlou tushu gufen youxian gongsi 萬卷樓圖書股份有限公司, 1995.

Shangwu yinshuguan bianjibu 商務印書館編輯部, ed. Ciyuan 辭源. Xiuding ben 修訂本. 4 vols. Beijing 北京: Shangwu yinshuguan 商務印書館, 1998.

Sun Changwu 孫昌武, Kinugawa Kenji 衣川賢次, and Nishiguchi Yoshio 西口芳男, eds. Zutang ji 祖堂集. 2 vols. Beijing 北京: Zhonghua shuju 中華書局, 2007.

Tian Qishi 田其湜, ed. Liuti shufa dazidian 六體書法大字典. 2 vols. Changsha 長沙: Hunan renmin chubanshe 湖南人民出版社, 2004.

Van Cutsem, Laurent. “The Quánzhōu Qiānfó xīnzhù zhūzǔshī sòng 泉州千佛新著諸祖師頌 (S.1635): A TEI-Based Edition.” In Database of Medieval Chinese Texts. Ghent University and Chung-Hwa Institute of Buddhist Studies 中華佛學研究所, 2021. https://www.database-of-medieval-chinese-texts.be/.

Van Cutsem, Laurent, and Christoph Anderl. “A Translation and Study of Chán Master Jìngxiū’s 淨修禪師 Preface to the Zǔtáng jí 祖堂集.” Religions 12, no. 11 (974) (November 2021): 39 pp. https://doi.org/10.3390/rel12110974.

Wang Haigen 王海根, ed. Gudai Hanyu tongjiazidian 古代漢語通假字大字典. Fuzhou 福州: Fujian renmin chubanshe 福建人民出版社, 2006.

Wang Hui 王輝, ed. Guwenzi tongjiazidian 古文字通假字典. Beijing 北京: Zhonghua shuju 中華書局, 2008.

Wu Chengyuan 吳澄淵, ed. Xinbian Zhongguo shufa dazidian 新編中國書法大字典. Beijing 北京: Shijie tushu chuban gongsi 世界圖書出版公司, 2001.

Xu Han 徐寒, ed. Shufa dazihai 書法大字海. Xiuding ban 修訂版. 2 vols. Beijing 北京: Shangwu yinshuguan guoji youxian gongsi 商務印書館國際有限公司, 2009.

Zen bunka kenkyūjo 禅文化研究所, ed. Sodōshū 祖堂集. Kihon tenseki sōkan 基本典籍叢刊. Kyōto 京都: Zen bunka kenkyūjo 禅文化研究所, 1994.

Zhang Hua 張華, ed. Zutang ji 祖堂集. Zhengzhou 鄭州: Zhongzhou guji chubanshe 中州古籍出版社, 2001.

Zhang Meilan 張美蘭, ed. ‘Zutang ji’ jiaozhu 祖堂集校注. Beijing 北京: Shangwu yinshuguan 商務印書館, 2009.

Zhang Qiang 張强 and Wei Liya 魏離雅 (Lia Wei). “‘Qing’ heyi bian ‘qing’: Beichao Seng An Daoyi shuke yanjiu zhi yi ‘清’何以變‘凊’: 北朝僧安道壹書刻研究之一.” Shufa yanjiu 書法研究, no. 1 (2020): 48–60.

Zhang Yongquan 張涌泉. Dunhuang suzi yanjiu 敦煌俗字研究. Di er ban 第二版. Shanghai 上海: Shanghai jiaoyu chubanshe 上海教育出版社, 2015.

———. Dunhuang suzi yanjiu daolun 敦煌俗字研究導論. Taibei shi 臺北市: Xinwenfeng chuban gongsi yinxing 新文豐出版公司印行, 1996.

———. “Dunhuang wenshu leihuazi yanjiu 敦煌文書類化字研究.” Dunhuang yanjiu 敦煌研究 4 (1995): 71–79.

[1] See Galambos, Orthography of Early Chinese Writing, 11–30.

[2] On the DMCT, see, for example, Anderl, “Some Reflections on the Database.”

[3] See Bianji shuoming 編輯說明 (“Editorial policies”), ‘Yitizi zidian’ jianjie 《異體字字典》簡介 in Guojia jiaoyu yanjiuyuan yuwen jiaoyu ji bianyi yanjiu zhongxin and Guoyu tuixing weiyuanhui, Jiaoyubu yitizi zidian (6th ed.).

[4] See, for example, the fourth character on line 9 (於) and 10 (扵), juxtaposed to each other, on the second printing surface of the third fascicle (Zen bunka kenkyūjo, Sodōshū, 101). On 於 and 扵, see Zhang Yongquan, Dunhuang suzi yanjiu, 359.

[5] For a good recent overview of the development of the Chinese writing system, see Handel, Sinography, 30–52.

[6] Late Middle Chinese (henceforth L.): ʔjian; Early Middle Chinese (henceforth E.): ʔɛn (Pulleyblank, Lexicon of Reconstructed Pronunciation, 355); Baxter and Sagart, “The Baxter-Sagart reconstruction,” give ‘en for their Middle Chinese transcription; see also YTZZD.

[7] For an overview of the orthographic terminology of the Ganlu zishu, the Longkan shoujian, and other ziyang 字樣 (“Model of Characters”) works, see Nishihara, “The Longkan shoujing.” According to Nishihara Kazuyuki 西原一幸, the Tang orthographic standard was elaborated primarily on the basis of the Shuo wen jie zi 說文解字 (Nishihara, “The Longkan shoujing,” 89–90). On the Shuo wen jie zi, see, for example, Bottéro and Harbsmeier, “The Shuowen Jiezi Dictionary.”

[8] See, for example, Galambos, “Variant Characters.”

[9] Galambos, Orthography of Early Chinese Writing, 2.

[10] Galambos, Orthography of Early Chinese Writing, 3, 141; Boltz, The Origin and Early Development of the Chinese Writing System, 72.

[11] The original is as follows: “字形或拼法符合標準的字。區別於異體字、錯字、別字等。亦指本字。” (Luo Zhufeng et al., Hanyu dacidian, Vol. 5, 310, No. 1). Compare with Morohashi, Dai Kan-Wa jiten, Vol. 6, 667b, No. 1.

[12] See, for example, Kinugawa, “Sodōshū satsuki,” 118; Anderl, “Medieval Chinese Syntax,” 690–91.

[13] The original is as follows: “音同義同而形體不同的字。即俗體、古體、或體之類。” (Luo Zhufeng et al., Hanyu dacidian, Vol. 7, 1354). See also Qiu Xigui, Wenzixue gaiyao, 233; cf. Qiu Xigui, Chinese Writing, 297. Note that in Gilbert L. Mattos and Jerry Norman’s translation of Qiu Xigui’s work, yitizi is rendered as “allograph.”

[14] Qiu Xigui, Wenzixue gaiyao, 235.

[15] See Qiu Xigui, Wenzixue gaiyao, 233.

[16] See, for example, the brief definition of suzi in the Ciyuan 辭源 (Source of Words): “在民間流行的異體字。別於正體字而言。” (“An yitizi(i.e., variant character) that circulates among the common people and differs from the standard character”; Shangwu yinshuguan bianjibu, Ciyuan, Vol. 1, 221). Alternatively, the term yitizi is occasionally understood as an equivalent of suzi.

[17] The original is as follows: “漢字史上各個時期與正字相對而言的主要流行於民間的通俗字體稱為俗字。” (Zhang Yongquan, Dunhuang suzi yanjiu daolun, 2; cf. Zhang Yongquan, Dunhuang suzi yanjiu, 2). Note that Zhang Yongquan explains each element of his definition in the following pages (see Zhang Yongquan, Dunhuang suzi yanjiu daolun, 2–4; cf. Zhang Yongquan, Dunhuang suzi yanjiu, 2–5). Compare with Morohashi, Dai Kan-Wa jiten, Vol. 1, 785a-b.

[18] Zhang Yongquan, Dunhuang suzi yanjiu daolun, 3–4. See Nishihara, “The Longkan shoujing,” 87.

[19] Zhang Yongquan, Dunhuang suzi yanjiu daolun, 2, 4.

[20] Zhang Yongquan, Dunhuang suzi yanjiu daolun, 5–6. Galambos notes that the term biezi is generally used with regard to stone inscriptions and that suzi is used with reference to medieval manuscripts (Galambos, “Variant Characters,” 483–84). In addition, in the past biezi was occasionally used to designate erroneous characters (Zhang Yongquan, Dunhuang suzi yanjiu daolun, 8).

[21] Zhang Yongquan, Dunhuang suzi yanjiu daolun, 7–8.

[22] Also known as “semantic element,” “signific,” “determinative,” “taxogram,” and so forth. On this terminology, see Handel, Sinography, 43.

[23] Qiu Xigui, Wenzixue gaiyao, 235–37. On this topic, see, for example: Zhang Yongquan, Dunhuang suzi yanjiu and Galambos, “Medieval Ways of Character Formation.”

[24] Galambos, “Medieval Ways of Character Formation,” 58. See also Handel, Sinography, 17.

[25] To be more accurate, S.1635 has [氵*曺]. See S.1635r_03.18 or S.1635r_82 (Van Cutsem, “The Quanzhou Qianfo xinzhu zhuzushi song”).

[26] Pulleyblank 1991, 45.

[27] Galambos, “Medieval Ways of Character Formation,” 69; see also Zhang Yongquan, “Dunhuang wenshu leihuazi yanjiu” and  Zhang Yongquan, Dunhuang suzi yanjiu, 259–60.

[28] See, respectively: Zhang Yongquan, “Dunhuang wenshu leihuazi yanjiu” and Zhang Yongquan, Dunhuang suzi yanjiu, 259–69; Qiu Xigui, Wenzixue gaiyao, 265–66; and Galambos, “Medieval Ways of Character Formation,” 67–69.

[29] See Qiu Xigui, Wenzixue gaiyao, 266.

[30] See Zen bunka kenkyūjo, Sodōshū, 112. Like S.1635, the Goryeo edition of the Zutang ji in fact has 曺 (A01836-003; YTZZD), a common variant of 曹. On the Goryeo edition of the Zutang ji, see Van Cutsem and Anderl, “A Translation and Study of Chán Master Jìngxiū’s 淨修禪師Preface,” 3–6.

[31] On the liushu system, see Galambos, “The Chinese Writing System,” 33–34; Handel, Sinography, 48–50.

[32] Galambos, “Popular Character Forms,” 396-99; See, for example, Bottéro, “Review of The Origin and Early Development of the Chinese Writing System,” 576; Handel, Sinography, 49–50, n. 42.

[33] See Galambos, “Medieval Ways of Character Formation,” 62; Handel, Sinography, 49–50.

[34] Galambos, “Popular Character Forms,” 402–03. Some of the characters created by Wu Zetian 武則天 (624–705) follow the huiyi principle (see Bottéro, “Les Graphies Énigmatiques,” especially pp. 94–96). In contrast to what Xu Shen 許慎 (c. 58–147?) noted in the postface to his Shuo wen jie zi, however, the character 信, for example, is not a huiyizi because the component 亻/人 in fact corresponds to the phonophoric (see Qiu Xigui, Wenzixue gaiyao, 121; cf. Qiu Xigui, Chinese Writing, 155).

[35] Branner, “Portmanteau Characters,” 73.

[36] Pulleyblank, Lexicon of Reconstructed Pronunciation, 247; Baxter and Sagart, “The Baxter-Sagart reconstruction,” give khjijH.

[37] YTZZD; see also Branner, “Portmanteau Characters,” 73.

[38] Branner, “Portmanteau Characters,” 73.

[39] See Zen bunka kenkyūjo, Sodōshū, 100.

[40] The passage is as follows: 「[…] 師遂見虎狼遶庵,麈鹿縱橫四畔。師乃兩手作怕勢云:「[屳/(虫*見))]。」」 (see Zen bunka kenkyūjo, Sodōshū, 100; Sun Changwu et al., Zutang ji, 136; CBETA 2021.Q3, B25, no. 144, p. 349b9-10); “[…] The master (i.e., Daoxin 道信) thereupon saw that tigers and wolves were circling around the thatched hut [of Fǎróng] and that deer were running wild in all directions. The master, then, with his two hands, made a gesture expressing fear and exclaimed: ‘Yikes!’” Note that in this passage, chong 虫 is likely equivalent to chong 蟲 and probably refers to dachong 大蟲 which itself is commonly used to refer to tigers (laohu 老虎) (see Luo Zhufeng et al., Hanyu dacidian, Vol. 2, 1398; Komazawa daigaku nai zengaku daijiten hensanjo, Zengaku daijiten, 807b). Alternatively, this term could also point, more generally, to ferocious beasts. Interestingly, in the Tiansheng guangdenglu’s 天聖廣燈錄 (Extensive Lamp Record of the Tiansheng [era]) entry for Chan master Yanzhao 延昭禪師, juan 15, a passage reads as follows: 「問:「古人入山見大蟲,意旨如何?」師云:「木人嶺上高聲呌,石女谿邊側耳聽。」」(CBETA 2021.Q4, X78, no. 1553, p. 489b15-17); “Someone asked: ‘[As for] the person of old (i.e., Daoxin?) who entered the mountain and saw tigers,’ what is the purport [of this]?’ The master replied: ‘The wood/dull-witted man, on the top of the mountain, shouts loudly; the stone/barren woman, on the stream bank, bends the ear and listens.’” See also Okayama et al., “Sodōshū Gozu Hōyū,” 52.

[41] According to Zhang Meilan 張美蘭, Ōta Tatsuo 太田辰夫 (1916–1999) suggested that this composite character was a demotic form of xian 嶮. Probably rightfully, however, she noted her reservations about this (Zhang Meilan, Zutang ji jiaozhu, 84, n. 3).

[42] Zhang Hua, Zutang ji, 94, n. 5; Sun Changwu et al., Zutang ji, 136, n. 2; Zhang Meilan, Zutang ji jiaozhu, 84, n. 3. On ligatures, see, for example, Galambos, “Popular Character Forms,” 66–67.

[43] See Zhang Qiang and Wei Liya, “‘Qing’ heyi bian ‘qing’”.

[44] The original passage is as follows: “[…] 假借就是借用同音或音近的字來表示一個詞” (Qiu Xigui, Wenzixue gaiyao, 203); Galambos, Orthography of Early Chinese Writing, 80. This phenomenon is also known as paronomasia or, more commonly, as rebus usage (see Handel, Sinography, 48).

[45] Qiu Xigui, Wenzixue gaiyao, 205; see Pulleyblank, Outline of Classical Chinese Grammar, 134.

[46] Galambos, Orthography of Early Chinese Writing, 80.


[47] For example, S.516r_01.01; see also Adamek, The Mystique of Transmission, 463, n. 2.

[48] Pulleyblank, Lexicon of Reconstructed Pronunciation, 189–90; Baxter and Sagart, “The Baxter-Sagart reconstruction,” give lek.

[49] See, for example, Wang Haigen, Gudai Hanyu tongjiazi dazidian, 417.

[50] Pulleyblank, Lexicon of Reconstructed Pronunciation, 213; Baxter and Sagart, “The Baxter-Sagart reconstruction”, give mjit.

[51] Pulleyblank, Lexicon of Reconstructed Pronunciation, 213; Baxter and Sagart, “The Baxter-Sagart reconstruction”, give mit.

[52] Qiu Xigui, Wenzixue gaiyao, 209.

[53] Qiu Xigui, Wenzixue gaiyao, 210.

[54] See Qiu Xigui, Wenzixue gaiyao, 209–13.

[55] See Huang Zheng, Dunhuang suzidian. In addition to the dictionary of graphic variants itself, Huang Zheng also describes in a lengthy introduction his understanding of demotic variants in the Dunhuang context, including a proposed classification (see Huang Zheng, Dunhuang suzidian, 1–35). Note that an expanded second edition was manifestly published in 2019 by the same press.

[56] Zhang Yongquan, Dunhuang suzi yanjiu.

[57] See, for example, Tian Qishi, Liuti shufa dazidian; Wu Chengyuan, Xinbian Zhongguo shufa dazidian; Xu Han, Shufa dazihai.

[58] Pulleyblank, Lexicon of Reconstructed Pronunciation; Baxter and Sagart, “The Baxter-Sagart reconstruction.”

[59] For a good introduction, see, for example, Jacques, “Traditional Chinese Phonology.”

[60] For example, Wang Haigen, Gudai Hanyu tongjiazi dazidian; Wang Hui, Guwenzi tongjiazidian.

[61] Galambos, “The Chinese Script.”

Click here to the original posting