Journal of Chinese Literature and Culture

Zang Maoxun’s Yuanqu xuan

The background of Zang Maoxun’s publishing venture is well known. A huge influx of silver from Japan and Latin America into China from the sixteenth century onward had greatly contributed to the late Ming’s economic boom, especially in the Jiangnan area. Not only merchants profited from these developments but also the established gentry families who diversified their investments. Profiting from tax credits based on examination status, these families could amass huge private fortunes and live in great luxury. Combined, such fortunes and luxury created a large market for high-quality leisure readings by the end of the sixteenth century. Following the collapse of his official career, Zang Maoxun sought to profit from this burgeoning market by publishing works in various genres—but none of his ventures seems to have equaled the success of his Yuanqu xuan.²

Zang Maoxun was not the first publisher of Yuan drama in the Wanli (1573–1619) period of the Ming. Several publishers had preceded him in developing the market,³ but it is not clear to me whether they published their plays in large sets or as single plays in identical format. Whatever the nature of their ventures, however, these were all dwarfed by the quantity and quality of Zang Maoxun’s Yuanqu xuan. In two sets of fifty plays each, Zang Maoxun published a single collection made up of one hundred plays, a number not approached by any of his predecessors. Each play was also printed in a far more luxurious format than had ever been attempted: the blocks were carefully prepared, the text printed in large and clear type. The texts of the arias and the dialogues were clearly distinguished, and no attempt was made to save space by printing the dialogues in double columns. The division of the texts into acts and wedges was standardized throughout, as was the assignment of role types to characters. Pronunciation notes were provided at the end of each act, but the sensitivity of the intended, highly educated readers was not offended by the presence of explanatory notes. Each play was, moreover, preceded by two or four beautiful, newly designed woodblock illustrations, greatly enhancing the attractiveness of the publication.

But the most important reason for the enduring success of Zang Maoxun’s Yuanqu xuan is not to be found in the quantity and the quality of the physical production, important as these were. The most important reason for the success of this remarkable publishing venture must have been that it completed the process of moving Yuan drama from the stage to the study, in the process [End Page 176] turning scripts into texts. And these texts, which earlier had circulated only as entertainment, now could be properly appreciated as literature, along with other genres of song from China’s past. In this respect, it is important to stress that the collection was preceded not only by two prefaces in which Zang discoursed on the nature of the plays (now reprinted in standard modern editions) but also, following the table of contents, by a host of earlier writings on zaju (borrowing heavily from Zhu Quan’s 朱權 [1378–1448] Taihe zhengyin pu 太和正音譜 [Formulary for Correct Sounds in an Era of Great Harmony], and perhaps for that reason often not reprinted in modern editions). These latter writings provide a host of materials on modes, arias, performance, and composition, as well as a catalog of zaju authors and their works. This catalog enables Zang Maoxun to assign most of the plays to specific authors, allowing the educated reader, in turn, to read each play as the expression of an authorial, moral mind. In order to further prepare his texts for this reception, Zang Maoxun (as he mentions in one of his letters) at times even extensively edits the plays.

Comparison with earlier editions (available for eighty-five of the one hundred) of the plays included in Zang’s Yuanqu xuan suggests that such changes vary greatly in extent from play to play.⁴ Three major categories of changes may be discerned. First, whereas earlier editions of zaju tended toward open-ended conclusions, Zang Maoxun made sure that each play had a true datuanyuan 大團 圓 (grand reunion) that resolved all contradiction and conflict. If needed, Zang would extensively rewrite and develop the final act of a play for this purpose, often inserting arias of his own making—most of these new arias appear in the final act, though we see throughout the plays evidence of a very free editorial hand, with minor as well as major changes, usually to tighten the plot. Second, this emphasis on harmonious conclusions is accompanied by a subtle strengthening of elite (Confucian) morality in the plays through a host of smaller and larger changes to arias and dialogues leading to the final grand reunion scene. And third, the language and orthography of the plays are standardized and cleaned up, faulty allusions corrected, and cases of imperfect parallelism changed.⁵ We cannot underestimate Zang Maoxun’s contribution: once his Yuanqu xuan was printed, Yuan drama occupied a firm place in Chinese literature.

Because of its superior qualities as a reading text, the Yuanqu xuan quickly supplanted all earlier late Ming editions of Yuan drama. Attempts to cash in on this market by other editors failed.⁶ The Yuanqu xuan continued to be avidly read throughout the Qing dynasty. As we learn from the Honglou meng 紅樓夢 (Dream of the Red Chamber) in a long diatribe by Xue Baochai 薛寶釵 against Lin Daiyu 林黛玉, grown-ups tried to hide the book from youngsters, but boys read the plays behind the backs of their teachers and elders, and girls read them [End Page 177] behind the backs of the boys. Twentieth-century pioneering scholars of Yuan drama such as Wang Guowei 王國維 (1877–1927) and Yoshikawa Kojiro 吉川幸 次郎 (1904–1980) based their research primarily on the Yuanqu xuan texts. It is probably fair to say, even now, that for most scholars of Chinese theater and literature, the image of Yuan drama is shaped by the Yuanqu xuan.

Translations from the Yuanqu xuan

In view of the Yuanqu xuan’s popularity with Chinese readers and its position in Chinese scholarship (resulting in wide availability), it should come as no surprise that Western translations of Yuan drama have been based almost exclusively on the Yuanqu xuan edition. The selection of titles as well as the manner of translation, however, varied according to the purpose of the translation as much as the personality of the translator. The first translation of a play from the Yuanqu xuan dates from as early as the eighteenth century, when Joseph Henri Marie de Prémare (1666–1736), the most accomplished linguist among Jesuit missionaries then at Beijing, completed his French translation of Ji Junxiang’s 紀 君祥 (c. 1264–c. 1294) Zhaoshi gu’er 趙氏孤兒 (The Orphan of Zhao) as L’orphelin de la maison de Tchao, tragédie chinoise (The Orphan of the House of Zhao: A Chinese tragedy). Published in 1734, this play would have great influence on European theater in the eighteenth century. It should be pointed out, however, that Prémare’s translation was limited to the dialogues, as it was not originally intended as an illustration of Chinese literature but as a study aid for modern Chinese grammar. In that context, there was no point in translating the arias; in addition, Prémare may have found them too difficult and was at a loss as to how to render them. Ironically, the fourteenth-century edition of Zhaoshi gu’er provides only the texts of the arias and not a single word of the dialogue or stage directions.

The arias continued to baffle translators well into the nineteenth century. Abel Rémusat (1788–1832), the first professor of Chinese at the Collège de France in Paris, once even declared that Chinese poetry was “most of the time unintelligible and at all times untranslatable.”⁷ John Francis Davis (1795–1890), working in Macao and Canton as a servant of the British East India Company, was the first to produce an English translation of Yuan plays based directly on the Chinese text. When in 1827 he translated Ma Zhiyuan’s 馬致遠 (1250?–after 1321) Hangong qiu 漢宮秋 as Han Koong Tsew; or, The Sorrows of Han: A Chinese Tragedy, he also, like Prémare before him, omitted the arias. This was, however, not his first translation of a play from the Yuanqu xuan because more than ten years earlier, his Laou-seng-urh; or, “An Heir in His Old Age,” A Chinese Drama was published in London by J. Murray in 1817. In this rendition of Wu Hanchen’s 武漢臣 Laosheng’er 老生兒, a more youthful Davis, who was a pioneer [End Page 178] translator of Chinese poetry, had provided (rhymed) translation of the arias, even though he confessed to having often been baffled by their meaning. In a brief preface, Davis stated his intention to offer a complete rendition but noted that “a few passages, and they are but few, which were either grossly indecent, or insufferably tedious, have been purposely omitted from the translation.”⁸

Traditional Chinese fiction and drama remained popular with Europe-based sinologists as a window on daily life in China for much of the nineteenth century when China was still closed to foreigners—and after, when China began to open up but continued to seem remote. In mid-nineteenth-century Paris, the most famous specialists on Yuanqu xuan drama were Stanislas Julien (1797–1873) and his student Antoine-Pierre-Louis Bazin (1799–1863). Julien’s first translation of a Yuan drama was his 1832 rendition of Li Xingdao’s 李行道 (1279?) Huilan ji 灰闌記 (The Chalk Circle).⁹ This was followed in 1834 by his complete translation of Zhaoshi gu’er, this time including the arias. In his Théatre chinois, ou choix des pièces de théatre composées sous les empereurs mongols (Chinese Theater; or, An Anthology of Plays Written under the Mongol Emperors) of 1838, Bazin published full translations of Guan Hanqing’s 關漢卿 (1219–1301) Dou E Yuan 竇娥冤 (Injustice to Dou E), Zheng Dehui’s 鄭德輝 Zhou meixiang 梅香 (The Headstrong Maid), Zhang Guobin’s 張國賓 He hanshan 合汗衫 (The Untied Shirt), and the anonymous Huolang dan 貨郎旦 (The Peddler, a Female Lead), all based on the Yuanqu xuan. In his 1850 Le siècle des Youên, ou tableau historique de littérature chinoise depuis l’avènement des empereurs mongols jusqu’à la restauration des Ming (The Century of the Yuan: A Historical Survey of Chinese Literature from the Establishment of the Mongol Dynasty to the Ming Restoration), he presented summaries of all one hundred plays in the Yuanqu xuan, often with translated excerpts. Many of the translations by Julien and Bazin were retranslated and adapted in other European languages well into the twentieth century, and some of these adaptations (for instance the German editions of Klabund and Bertold Brecht) also enjoyed considerable success onstage—though none of course achieved the éclat of Voltaire’s L’orphelin de la Chine (The Orphan of China) in the eighteenth century. Original translations of Yuan drama became rare in the second half of the nineteenth century as Westerners more freely traveled in China: drama, like fiction, was no longer a unique window on daily and private life in China.

In the twentieth century too the Yuanqu xuan continued to be the most important source text for translation and study of Yuan drama in the West. But by the beginning of the twentieth century, European sinology had completed its philological turn, and few academic sinologists continued to work on the vernacular literature of late imperial China.¹⁰ An important exception was German scholar Alfred Forke (1867–1944), best known for his work on Chinese [End Page 179] philosophy, who also produced a considerable number of translations of Chinese plays from the Yuan, Ming, and Qing. While a few of these translations were published during his lifetime, many came out only later (1978–93), when they were edited by Martin Gimm in three fat volumes.¹¹ The most important English translation of Yuan drama before World War II was S. I. Hsiung’s 1935 rendition of Wang Shifu’s 王實甫 (1260?–1316?) Xixiang ji 西廂記 as The Romance of the Western Chamber (His Hsiang Chi): A Chinese Play Written in the Thirteenth Century.¹²

As Chinese studies began to flourish in postwar United States, many students of Chinese literature were inspired by developments from the May Fourth generation in the study of Chinese literary history, and they came to work on vernacular literature of the Yuan and later. Employed by “departments of language and literature,” they taught courses on “Chinese literature in translation” to undergraduates, including students with no knowledge of Chinese. This created a market for general and specialized English-language anthologies of Chinese literature in English translation. One of the earliest successes dates to 1965, the Anthology of Chinese Literature, from Early Times to the Fourteenth Century, edited by Cyril Birch: it included translations of two zaju, Ma Zhiyuan’s Hangong qiu, and Kang Jinzhi’s 康進之 Li Kui fujing 李逵負荊 (Li Kui Carries Thorns). The first play was translated by Donald Keene (best known as a translator of Japanese) who rendered its arias into blank verse. This experiment in translation may well have been inspired by an early article of James J. Y. Liu, “Elizabethan and Yuan: A Brief Comparison of Some Conventions in Poetic Drama” (1958), in which Liu contrasted the alternation of prose and verse in Shakespearean drama with the alternation of spoken dialogue and sung arias in zaju. Keene’s experiment has, to the best of my knowledge, never been imitated.

Kang Jinzhi’s Li Kui fujing was translated by James I. Crump Jr. (1921–2002), who chose to render the play’s arias in free verse, a decision followed by most English-language translators of traditional Chinese drama. Based at the University of Michigan, Crump was not only a productive translator and scholar of Yuan-dynasty zaju (and sanqu 散曲),¹³ he also trained a number of students who wrote dissertations on early Chinese drama, such as Dale R. Johnson, Paul Kroll, Perng Chi-hsing, and Stephen H. West.¹⁴ For his own translations, Crump continued to rely on the Yuanqu xuan texts throughout his life.¹⁵ This applies as well to the 1970s English-language anthologies of Yuan zaju prepared by George Hayden,¹⁶ Liu Jung’en,¹⁷ and Richard Yang.¹⁸ In her Injustice to Tou O (Tou O Yüan): A Study and Translation (1972), Chung-wen Shih based her translation (intended as an aid for familiarizing the reader with the language of Yuan drama) on the Yuanqu xuan edition of Guan Hanqing’s play, “as it is a fuller and more readable one,” but she includes notes to an earlier printed edition of 1588.¹⁹ The [End Page 180] only English-language anthology of Yuan zaju that incorporates several plays from sources outside the Yuanqu xuan was Selected Plays of Kuan Han-ching (1958), translated by Yang Hsien-yi and Gladys Yang, but the issue of editions is not raised in the introduction at all. In Europe during the 1960s and 1970s, the most productive British scholar on traditional Chinese drama was William Dolby, but he published only a few translations.²⁰ In France, Yuan drama never regained its mid-nineteenth-century position of preeminence in Chinese studies,²¹ but in Germany, the study of Yuan drama was greatly stimulated by Alfred Hoffmann (1911–1997). Best known for his studies of Chinese bird names and the lyrics of Li Yu 李煜 (937–978), he supervised a number of doctoral students who worked on Yuan drama.²²

The central position that the Yuanqu xuan enjoyed in drama studies of the second part of the twentieth century can also be discerned in The Columbia Anthology of Yuan Drama, initiated some decades ago by C. T. Hsia (1921–2013) and George Kao (1912–2008), and finally brought to publication in the spring of 2014 by Wai-yee Li. Most of the translations in this anthology are based on Yuanqu xuan texts, but reflecting recent interest in other editions of Yuan plays, Wai-yee Li decided to include a translation of the fourteenth-century printing of Zhaoshi gu’er alongside a rendition of the Yuanqu xuan text, dramatically demonstrating the distance that can exist between editions. In the fourteenth century edition, the play consists of four acts (not five, as in the Yuanqu xuan), only the arias are printed, and the text of nearly every song in the play shows major discrepancies when compared with the later text of the Yuanqu xuan.The dialogues translated by Prémare are totally absent from this edition. Zhang Guangqian’s translations in his Selected Plays from the Yuan Dynasty (2010) are also based on the texts in Zang Maoxun’s Yuanqu xuan.

Zaju at the Ming Court

The fourteenth-century printings of Yuan zaju (yuankan zaju 元刊雜劇) most likely were produced for members of the audience unable to follow the words sung in performance.²³ Actors and actresses, meanwhile, worked with zhangji 掌記 (palm notes), manuscript copies of the texts small enough to hide in their sleeves. Some of those who bought the printed version of the arias must have preserved them after the show, either because they wanted to learn to sing these songs or because they appreciated them as works of literature. We currently have access to thirty such printings (now collectively known as Yuankan zaju sanshizhong 元刊雜劇三十種), which by the Qing dynasty had come to be prized and preserved as rare books.²⁴ We know that a few more must have been in circulation in the sixteenth century, but since commercial theaters where zaju were performed in the Yuan and early Ming had disappeared by then, it is [End Page 181] unlikely that such printings, now without a ready-made market, continued to circulate in the sixteenth century in large numbers, despite the proud claims of some Ming bibliophiles about their large collections of plays. With the exception of an uncertain number of plays that have reached us through the intermediary of Li Kaixian’s 李開先 (1502–1568) Gaiding Yuanxian chuanqi 改定元賢傳奇 (Revised Plays by Yuan Authors, to be discussed below) and a few others, the majority of Yuan-dynasty zaju that have been preserved (including those in the Yuanqu xuan) are the plays selected for inclusion in the Ming court repertoire.²⁵ Even though the Ming imperial family originated in the South, it adopted Northern drama for court performance because it enjoyed much higher prestige at the time than Southern drama (xiwen 戲文).

At court, two agencies were entrusted with the performance of plays: the Court Entertainment Bureau (Jiaofangsi 教坊司), for court functions attended by officials, and the Office of Bell and Drum (Zhonggusi 鍾鼓司), a eunuch agency that performed inside the inner palace for the entertainment of the imperial family. A late Ming source indicates that the scripts of plays performed on such occasions were to be submitted to a court censor for approval before performance. If this was standard practice throughout the dynasty, then it would explain why palace manuscripts (or their copies) tend to be so long, including extensive sections of repeated dialogue and every (lame) joke. It would also explain the plays’ uniformity of prose style. Unfortunately, our information on court performance and the organization of court theatricals during the Ming is very limited, even despite recent research that seems to have mined every possible source.²⁶ Many palace scripts, preserved in manuscript copies in the Maiwangguan 脈望館 collection, became available to the scholarly world only on the eve of the Anti-Japanese War. Those scripts in that collection that would appear to have been preserved through the intermediary of Yu Xiaogu 于小榖 mostly likely derive from the Court Entertainment Office and may reflect performances dating to 1460–80, whereas the scripts originating from the Office of Bell and Drum mostly likely date from the Jiajing period and represent inner-palace performances of the mid-sixteenth century. These two sets of play scripts differ significantly in a variety of ways, not least in the formal aspects of their endings.

From the late fourteenth to the late sixteenth century, court zaju went through many transformations and interventions.²⁷ Like in other forms of drama worldwide, the plays that remained in the repertoire changed in response to new theatrical companies and performance venues, developments in technology, and the ever-shifting tastes of patrons and audiences. The two most powerful forces on the zaju during this time were the size and nature of the theatrical companies and political interference. For Yuan-dynasty performance, most [End Page 182] theatrical companies appear to have been rather small commercial ventures, often organized around a male or female star—thus, most plays could be performed with a very small cast. The imperial palace could, of course, summon much larger numbers of performers, whether through the Court Entertainment Office or the Office of Bell and Drum. Zaju may have originated as a highly asymmetrical dramatic form focused on the zhengmo 正末 (male lead) or zhengdan 正旦 (female lead), but at court an attempt was made to downplay these central roles. Comparisons of fourteenth-century printings and palace manuscripts of the same play often show the palace scripts drastically reducing the number of arias assigned to the lead performer. Several arias might be cut from each of the four sets in a play. Since in at least some fourteenth-century printings long stretches of a play would have consisted of songs by the main performer with little action, such cuts may also have served to enhance the liveliness of the stage business. Comparisons between original printed texts of plays by Zhu Youdun 朱有燉 (1379–1439) and later palace scripts of the same plays show a simultaneous increase in the dialogue of the other performers, including newly assigned recitation pieces like poems, letters, and ballads to fill out their role.

While the decrease in arias and expansion of dialogues is the most conspicuous change in the texts related to their new function, other changes are also evident. For instance, plays dealing with episodes from longer story cycles are sometimes given more open endings so they may be performed as part of a series. Theatrical fashion may have dictated the disappearance of the bamboo horse (zhuma 竹馬) from the stage in Yuan zaju. The most conspicuous aspect of ideological interference was the well-known prohibition of any portrayal of an emperor onstage, whereas the role-type of emperor (jia 駕) had been quite popular on the Yuan stage. Fourteen of the thirty plays from the fourteenth century feature a jia in their cast: such so-called jiatouju 駕頭劇 could hardly be adopted into the palace repertoire, but efforts were made anyway, with revisions that expunged the jia from the cast.²⁸ When the emperor played only a minor role, such a change required only minor surgery, but in other cases, a far more drastic rewriting was called for. In some cases, it seemed that completely new versions of old stories were composed. The disappearance of an emperor from a script is, of course, easy to spot if we have materials for comparison,²⁹ but the rewriting of plays often went much further in order to make them more suitable for performance before the emperor. Even with the disappearance of the figure of the emperor, the ultimate authority of the state (in such matters as revenge) was stressed in many rewrites. A fine example is provided by Zhaoshi gu’er. In the fourth (and final) act of the fourteenth-century printing, the orphan enters describing his eager participation in an anticipated revolt against the duke of Jin, [End Page 183] and by the end of the act he anticipates with gusto his private revenge on Tu’an Gu 屠岸賈, the man who murdered his entire family. In the Yuanqu xuan version—with its extensive and repetitive dialogues, most likely based on a palace script—a fifth act appears in which the orphan asks permission from the duke of Jin to arrest Tu’an Gu for punishment and acts only after he has received permission. In other later versions of plays for which we have fourteenth-century printings, one notes the songs that have been cut are often precisely the ones detailing the social injustice suffered by the poor at the hands of the rich and powerful.³⁰

Li Kaixian and his Gaiding Yuanxian chuanqi

Ever since Sun Kaidi 孫楷第 and others have argued that the overwhelming majority of extant late Ming manuscript and print editions of Yuan zaju derive directly or indirectly from Ming court scripts, many scholars have voiced their doubts, unwilling to abandon the Yuanqu xuan texts as a reliable source for the study of Yuan drama of Yuan-dynasty times. One of the most vociferous scholars in this debate, Xu Shuofang 徐朔方, argued that if Zang Maoxun’s texts departed from earlier versions, not all the differences should be blamed on Zang’s textual interference. Zang, Xu argued, may have had access to alternative versions dating back to Yuan times.³¹ Though Xu Shuofang himself was not able to produce these alternative versions, he may have been closer to the truth than he imagined, even if these alternative versions on which Zang Maoxun (indirectly), in some cases, based his editions may not go back to the Yuan. We already noted the differences between texts deriving from the Court Entertainment Bureau and the Office of Bell and Drum, suggesting that alternative versions of palace scripts might have circulated.³² However, since the rediscovery two decades ago of six surviving plays from Li Kaixian’s Gaiding Yuanxian chuanqi,³³ we have a better idea of a second channel of transmission of Yuan zaju. The plays that survived in this collection underwent a process of editing that produced quite different results from the court adaptations.

Zang Maoxun may have completed the process of turning Yuan zaju from script into text, and from entertainment into literature, but he certainly did not initiate the process. Half a century before him, Li Kaixian already had taken a major step by publishing his Gaiding yuanxian chuanqi.³⁴ For most of the twentieth century, scholars were aware that plays from this project had been preserved, but their location remained a mystery until six plays were redis-covered on the shelves of Nanjing Tushuguan (Nanjing Library). In his preface and postface, Li Kaixian has left us a detailed description of his motives in producing the collection, as well as his selection criteria and editorial process. While priding himself on the size of his personal holdings of texts, Li Kaixian [End Page 184] deplored the absence of easily readable editions of Yuan drama, by his time already recognized as the representative literary genre of the Yuan dynasty. Li’s editorial committee, headed by Zhang Zishen 張自慎,³⁵ prepared fifty texts for publication, but because of financial limitations, only sixteen plays were printed. Li Kaixian explained his selection as follows: “The plays we have selected here have been chosen because of their lofty and classical diction, their agreeable and harmonious melodies, and because they have the power to stir and convert the human heart and to contribute to moral education by their emotional appeal. Moreover, the basis is provided by those works that display a lofty talent and accomplished learning, a penetrating insight, and a correct structure” 今所選傳 奇, 取其辭意高古, 音調協和, 與人心風教倶有激勸感移之功. 尤以天分高而學力 到, 悟入深而體裁正者, 為之本.³⁶ With such selection criteria, it is no accident that two of the preserved plays are by Ma Zhiyuan, two by Qiao Ji 喬吉 (c. 1280–1345), and one each by Bai Pu 白樸 (1226–c. 1306) and Wang Ziyi 王子一 (fl. 1360s, an early Ming playwright). Later editions of these plays show only minor changes, suggesting that these later editions, including those in the Yuanqu xuan, derive from the Gaiding Yuanxian chuanqi, but for how many others this would be true is difficult to ascertain.

Li Kaixian has provided us with a detailed description of his editorial strategies in preparing the plays for print: “We pared away the superfluous and opted for conciseness, changed the rhymes and corrected the sounds: whenever a song did not fit the melody, whenever a sentence was not appropriate, and whenever the dialogues contained passages that were irrelevant or too nonsensical, we edited and corrected them. It also happened that we created [songs or dialogues] on behalf [of the original authors]. This is why we have called the printed work Revised Plays by Yuan Masters” 刪繁歸約, 改韻正音, 調有不協, 句 有不穩, 白有不切及太泛者, 悉訂正之, 且有代作者.³⁷ The emphasis on paring away the superfluous and doing away with irrelevant and nonsensical dialogues would seem to suggest that Li Kaixian and his collaborators mostly worked on palace scripts, but for none of the six extant plays have palace manuscripts been preserved. As four of the six plays feature an emperor in their cast, it is unlikely that these derive from the palace repertoire.³⁸ The one play of the six, then, that might have come from palace texts is Wang Ziyi’s Wuru Tiantai 誤入天台 (Entering Mt. Tiantai Inadvertently). Explicit reference, however, to “substitute creation” (daizuo 代作) points in a different direction—that of the fourteenth-century editions of zaju as a major source for the Gaiding Yuanxian chuanqi.

In one case, Ma Zhiyuan’s Chen Tuan gaowo 陳摶高臥 (Chen Tuan’s Undisturbed Sleep), we can compare the Gaiding Yuanxian chuanqi edition with a fourteenth-century text of the same play. The latter consists of stage direction, important lines of dialogue, and the complete arias for the play’s lead [End Page 185] performer, in this case, a zhengmo.³⁹ In contrast to palace editors who cut many songs from the plays they adapted for the court stage, Li Kaixian and his collaborators retained all the arias of the fourteenth-century edition and made few (but significant) changes in the aria’s texts. They regularized the stage directions and with the information culled from them, as well as from arias and cue lines, created simple dialogues and stage direction for the secondary characters in the play. Compared to the elaborate dialogues in the palace scripts, these dialogues are quite lean.

By this daizuo Li Kaixian and his editorial team created a readable text, one that consisted of arias likely written in the late thirteenth or early fourteenth century, and dialogues mainly composed around the middle of the sixteenth century. From the case of Chen Tuan gaowo and other plays exhibiting similarly sparse dialogues and full song sets, we can infer that Li’s team likely approached other source texts in the same way. But as our reading of these arias is unavoidably impacted by our sense of a play’s plot—gleaned, as it is, primarily from dialogue—we cannot claim that we are reading a Yuan play when reading Gaiding Yuanxian chuanqi. And Li Kaixian does not claim that his editorial team is presenting the reader with plays as performed in the Yuan: he honestly called these texts gaiding chuanqi (revised plays) to draw attention to the extent of the changes made.⁴⁰

For most of their plays, Li Kaixian and his editorial team mostly likely relied on fourteenth-century editions of Yuan zaju, but it’s unclear to what extent they might have employed the early fifteenth-century printed editions by Zhu Youdun and Liu Dui 劉兌 (c. 1383) for their model.⁴¹ It is also unclear whether they were influenced by editions of Wang Shifu’s Xixiang ji. We have a few pages of that text that must have come out in the early decades of the Ming, as well as a sumptuous edition of 1498, fully illustrated and annotated, with a large number of paratextual materials. This latter edition is the earliest known printing of a Yuan zaju to divide its texts formally into acts, a procedure that well may have been inspired by its length.⁴² Most likely, those fifteenth-century publications enjoyed only a very limited circulation in their own time and by the mid-sixteenth century had dropped from sight following “the black hole of Ming history” in the middle to the late fifteenth century. A more likely model for the editorial conventions of the Gaiding Yuanxian chuanqi may have been the Shiduanjin 十段 錦 (Ten Pieces of Brocade). This collection, printed in 1558, consists of ten zaju of the early Ming dynasty (eight by Zhu Youdun in updated versions and two anonymous works) and is mentioned by Li Kaixian in his own anthology’s preface. The influence of sixteenth-century editions of zaju can be clearly observed in various features of the plays preserved in the Gaiding Yuanxian chuanqi. Xie Yufeng already pointed out that the plays from the Gaiding Yuanxian [End Page 186] chuanqi tend to be divided into acts and introduce role types not found in fourteenth-century editions, such as the chou 丑 (clown), tiedan 貼旦 (additional female), and chongmo 冲末 (opening male). But as it was the earliest of the collective editions of Yuan zaju in the late Ming, the similarities between the Gaiding Yuanxian chuanqi and the later Wanli editions may well be due to the fact that those later anthologies borrowed its plays and adopted its conventions.⁴³ Li Kaixian appears to have initiated the late Ming fashion of Yuan drama anthologies.

The Earliest Printed Copies of Yuan Drama and Their Translations

The zaju of the Yuan and first century of the Ming that have come down to us in printed or manuscript late Ming versions are quite numerous. With a few exceptions, these plays appear to have come down to us by one of two channels: either as part of the editorial activities of Li Kaixian and his assistants or through the Ming court repertoire. In either case, we are not dealing with Yuan-dynasty texts but, rather, texts heavily revised by Ming editors and censors, actors and literati, even before they were reedited by Zang Maoxun and incorporated into his Yuanqu xuan.⁴⁴ Such a conclusion is not intended to diminish or deny the literary value of the Yuanqu xuan but ought to serve as an invitation to read the Yuanqu xuan neither as reliable texts of Yuan-dynasty works nor as Ming-dynasty compositions, but very much as the reflections of late Ming Jiangnan literati on the essence of Yuan zaju. Thus, these texts are perhaps better treated in the context of Ming rather than Yuan literary history. Some recent histories of Chinese literature have drawn this conclusion, for instance, Gong Pengcheng’s 龔鵬程 Zhongguo wenxue shi 中國文學史 (2010) and The Cambridge History of Chinese Literature (2010).⁴⁵ But while these two works readily write zaju out of the literary history of the Yuan, they are less successful incorporating Yun zaju into Ming literary history.⁴⁶

If late Ming editions of Yuan zaju must remain under suspicion of heavy editorial intervention by Ming actors, literati, editors, and censors, whether at court or in private studies, then we are left for the study of Yuan drama with only the thirty zaju printings from the fourteenth century. That may not be much, but it is still a considerable number of texts on a wide variety of subjects by known and anonymous authors from both Northern and Southern China in the late thirteenth into the fourteenth century, exhibiting a remarkable variety of linguistic registers and literary styles. We cannot assume these fourteenth-century editions in every aspect faithfully reflect the texts as they left their authors’ hands. Indeed, there are indications that texts printed in Hangzhou but derived from the North were adapted to the local pronunciation, and we know that early fourteenth-century songs sometimes circulated in slightly different versions. But [End Page 187] we can be reasonably sure that these texts derive from a theatrical milieu in which zaju was the dominant genre of drama and performed both at court and in the best public theaters of the big cities. If we are interested in theater as performed in the century and a half from 1250 to 1400 (and beyond, until c. 1450), then we have to base ourselves first of all on these materials.⁴⁷

If one tries to forget for a moment the later editions of Yuan zaju and considers only these thirty plays preserved in their fourteenth-century editions, then one encounters a genre that in many ways is different from its later incarnations. The Yuan playwrights may not have conceived of a play in terms of acts and wedges, but of suites and scenes. Their plays probably relied more on song and less on action than the later revisions of their works.⁴⁸ While they often used the emperor in their cast, their protagonists just as often show an absolute disregard for state power and delight in personal revenge, reveling in bloody violence. Without the constant interruptions by secondary characters of later editions, the lead performers’ songs often exhibit a strong forward drive. No wonder the aristocratic playwrights of the early Ming tried to dissociate themselves from the brash emotions and crude violence of the Yuan stage and highly praised the playwrights who had written on enlightenment and the joys of retirement.

We’re very lucky these days to have at our disposal not only fine reproductions of these Yuankan zaju sanshizhong 元刊雜劇三十種 (the common way to refer to the fourteenth-century zaju editions as a set) but also three high-quality critical editions of the whole collection. Individual editions of single plays had been provided earlier, but the first scholar to offer a complete critical edition was Zheng Qian 鄭騫 (1906–1991) in 1962 with his Jiaoding Yuankan zaju sanshizhong 校訂元刊雜劇三十種. This work is by all accounts a superior achievement in vernacular philology, but its impact outside Taiwan has been hindered by political factors for decades. Once the Cultural Revolution was a thing of the past, Xu Qinjun 徐沁君 published his critical edition of the thirty plays as Xinjiao yuankan zaju sanshizhong 新校元刊雜劇三十種 in 1980, followed by Ning Xiyuan’s 寧希元 Yuankan zaju sanshizhong xinjiao 元刊雜劇三十 種新校 of 1988. Each of these works has its own strengths, and scholars will, of course, want to use the three editions side by side.⁴⁹ But if one wants to be nitpicky, one has to note that all three editors occasionally succumb to the temptation to impose their image of a homogeneous, clearly regulated genre (derived, no doubt, from late editions like the Yuanqu xuan) on their texts; as a result, they may not always show enough awareness of the possibility that the genre’s conventions, both onstage and on the page, had not in the fourteenth century yet become standardized to the same degree as they would later at court and in the scholar’s study.⁵⁰ This point is increasingly realized by a number of younger scholars who have started to take a new look at these early printings.⁵¹ [End Page 188]

But despite the availability of the wonderful critical editions mentioned above and, in many cases, of annotated editions too, these zaju in their fourteenth-century versions do not make for easy reading. If this already applies to modern Chinese readers of such texts, then it applies even more to foreign students of early Chinese vernacular literature, both in Japan and in Europe and North America. But there, too, one can notice a growing interest in these earliest editions of Yuan zaju. Japan has a long tradition of Chinese drama studies, but for most of the twentieth century the Yuanqu xuan had been the preferred source for translations of zaju.⁵² More recently, however, a group of scholars, many of them students of Tanaka Kenji 田中謙二, have devoted themselves over many years to the study of the earliest fourteenth-century editions. This has resulted not only in a series of critical editions of individual plays but also in the publication of two volumes by Akamatsu Norihiko 赤松紀彥 and his collaborators titled Genkan zatsugeki no kenkyū 元刊雜劇の研究 (Studies of Yuan Plays in Early Printings), consisting of heavily annotated Japanese translations of these early editions. The first volume (2007) provided translations of four plays (two by Shang Zhongxian 尚仲賢, viz., San duoshuo 三奪槊 [The Thrice-Stolen Lance] and Qi Ying Bu 氣英布 [Enraging Ying Bu]; and two by Guan Hanqing 關漢卿, viz., Xi Shu meng 西蜀夢 [A Dream of Western Shu] and Dandaohui 單刀會 [The Single-Sword Meeting]); the second volume (2011) included only two (Wang Bocheng’s 王伯成 Bian Yelang 貶夜郎 ([Banished to Yelang]) and Di Junhou’s 狄君厚 Jie Zi Tui 介子推).

In the English-speaking world, the scholars most consistently involved in the translation of these fourteenth-century editions of Yuan zaju have been Stephen H. West and Wilt L. Idema. West, as mentioned before, did his PhD with James Crump Jr. The topic of his dissertation was the literature of the Jin dynasty, resulting in a monograph focused on the yuanben 院本 (farce) of the Jin. Idema studied at Leiden University, but he also spent one year at Kyoto University, where he read Yuan drama with Tanaka Kenji. Idema’s dissertation was devoted to the earliest examples of China’s vernacular fiction.⁵³ The first joint publication of Idema and West was Chinese Theater 1100–1450: A Sour-cebook (1982), which provided translations and discussions of the contemporary Chinese materials on the theatrical activities and lives of actors and actresses in that period. As part of their documentation of theatrical life, they also collected and translated the plays featuring performers in their cast. As luck would have it, these five plays illustrated the various stages in the printing history of zaju from the fourteenth to the early seventeenth century, from the Yuankan zaju sanshizhong to the late Ming printings. The play from the Yuankan zaju sanshizhong translated here was Shi Junbao’s Ziyunting 紫雲庭 (Purple-Cloud Pavilion), on the love between an “all-keys-and-modes” performer and a young [End Page 189] Jurched nobleman, preserved only in a fourteenth-century edition. The collection also included (apart from one xiwen from the Yongle dadian 永樂大典 [Great Canon of Yongle Era]) two plays by Zhu Youdun translated on the basis of early fifteenth-century editions, and one play (Lan Caihe 藍采和) in a Gu mingjia 古名家 printing (c. 1588), while the Yuanqu xuan was represented by a translation of the final act of Huolang dan. When they later (1991) published their translation of Wang Shifu’s Xixiang ji, they did not base their translation on the seventeenth-century edition by Jin Shengtan (already available in an English translation by S. I. Hsiung), but on the earliest available text, that of 1498.⁵⁴

In more recent translations of Yuan drama, West and Idema have increasingly relied on fourteenth-century editions. In Monks, Bandits, Battles, and Immortals: Eleven Early Chinese Plays (2010), their translation for each play is based on the earliest available edition, again ranging from Yuankan zaju sanshizhong to Yuanqu xuan. This anthology, compiled at the request of the publisher as a gentle first introduction to the genre for undergraduate students, includes only one play in a fourteenth-century edition, viz., Guan Hanqing’s Baiyueting 拜月亭 (Pavilion for Praying to the Moon). The collection further includes two Shuihu 水滸 (water margin) plays by Zhu Youdun, translated on the basis of early fifteenth-century editions. Their translation of Bai Pu’s Wutongyu (Rain on the Wutong Tree) 梧桐雨 is based on the edition in Li Kaixinan’s Gaiding Yuanxian chuanqi. The other plays in the anthology were based on either Gu mingjia or Yuanqu xuan editions. Their Battles, Betrayals, and Brotherhood: Early Chinese Plays on the Three Kingdoms (2012) consists of seven early zaju: one by Zhu Youdun, two palace scripts, and one play in a late Ming printing (and its version in the Yuanq uxuan) together with three plays from fourteenth-century editions—Guan Hanqing’s Dandaohui and Xi Shu meng, and the anonymous Bowang Shaotun 博望燒屯 (Burning the Sores at Bowang).⁵⁵

The latest collective production of West and Idema is The Orphan of Zhao and Other Early Plays: The Earliest Known Versions (2015). This volume consists of seven plays in fourteenth-century editions (accompanied where available by later editions). Included in this collection are Ji Junxiang’s Zhaoshi gu’er (and the Yuanqu xuan version of the same play); Yang Zi’s 楊梓 (?–1327) Huo Guang guijian 霍光鬼諫 (Huo Guang Remonstrates as a Ghost); Zhang Guobin’s Xue Rengui 薛仁貴 (and its Yuanqu xuan version); Fan Kang’s 范康 (c. 1294) Zhuyezhou 竹葉舟 (with its Yuanqu xuan version); Gao Wenxiu’s 高文秀 Haojiu Zhao Yuan 好酒趙元 (and the manuscript copy in the Maiwangguan collection deriving from the Yu Xiaogu collection); Kong Wenqing’s 孔文卿 Dongchuang shifan 東窗事犯 (The Affair of the Eastern Window Exposed); and the anonymous Fen’er jiumu 焚兒救母 (Immolating His Child to Save His Mother). [End Page 190]

Conclusion

Already more than half a century ago, Zheng Qian made the argument that the many different editions of Yuan drama were produced for very different purposes, all had their own integrity and value, and therefore the collation of different editions of zaju in an effort to produce the original or best text was a misguided enterprise.⁵⁶ Zheng Qian reached this drastic conclusion after having compared all the texts in the Yuanqu xuan with available earlier editions.⁵⁷ Such comparisons between editions of the same play alert us to the different functions these texts had to serve, as Yuan zaju changed from an elite genre of commercial theater in the Yuan to court theater of the early centuries of the Ming, and the closet drama of Jiangnan literati. As the function of the plays changed, so did the nature of the texts at our disposition: from “hearing aids” of a listening audience to inspection copies for court censors to entertaining reading for gentlemen (and precocious teenagers of both sexes). To meet such different requirements, authors, censors, actors, and editors felt free to adapt these texts to changing needs.

But if the texts of Yuan zaju were produced for different purposes, then translations, too, have served different aims: from language study aids and social documentation to undergraduate textbooks in North America, these aims have shaped both the selection and presentation of the plays, as well as the literary and academic proclivities of the translators concerned. Some translators at least have felt free to omit those passages they disliked or failed to grasp, and they have often adapted the presentation conventions to the assumed expectations of a Western audience. We should also point out that none of these translations was produced with performance in mind; while some of the plays have been performed, they have always required versions specifically edited for the stage.

Once zaju had ceded the stage to southern drama and the many varieties of regional opera, Zang Maoxun’s Yuanqu xuan turned out to be the perfect edition to provide readers with reading copies of the plays. The enduring popularity of his edition over four centuries speaks to the quality of his edition in view of its purpose. It also comes as no surprise that it served as the almost unique source of translations of Yuan drama until the end of the twentieth century. But Yuan drama went through three centuries of development onstage and on the page before it became perfect literary reading material. If we want to understand the function and meaning of Yuan zaju during those centuries, we will have to base our study on the editions that reflect these earlier modes of its existence. In doing so, we discover a world much more variegated and diverse than is represented solely in the Yuanqu xuan. Chinese readers, of course, have had access to the full richness of this tradition for many decades now, but in [End Page 191] recent years, Japanese and Western readers, too, have been provided with translations that allow them to form a more complete image of Yuan drama in its origins and development.