Song Lyrics and the Canon:
A Look at Anthologies of Tz'u

Pauline Yu

Writing tz'u is difficult; selecting tz'u is even more difficult^[1]

Poetry Anthologies and Canon-Formation

The intimate connection between anthologies of poetry and the shaping of Chinese literary history and theory is well known; it is one that may follow naturally from the fact that the earliest specimens of the genre appear in a text whose provenance in a sage's act of selection represents one of the first assertions to be made about it. The various accretions of prefatory material to the Shih-ching (Book of Songs), moreover, established other precedents for later collections—and critical discourse as well—by virtue of their interest in the origins, lineage, and contexts of literary works. Following upon this canonical forebear, these collections engaged issues that proved central to the Chinese poetic tradition, including questions concerning the definition and nature of literature, its relationship to history, theories of periodization and genre, and modes of interpretation and evaluation.^[2] And indeed, as this list may suggest, anthologies served, both overtly and covertly, as important loci for attempts at defining the secular poetic canon itself.^[3]

[1] Ch'en T'ing-cho (1853–92), Pai-yü-chai tz'u-hua , in Ch'ü Hsing-kuo, ed., Pai-yü-chai tz'u-hua tsu-pen chiao-chu , 2 vols. (Tsinan: Ch'i Lu shu-she, 1983), 1:761 (hereafter cited as Tsu-pen chiao-chu ). In the Tz'u-hua ts'ung-pien , ed. and printed by T'ang Kuei-chang (24 volumes in four boxes, preface dated 1934; hereafter cited as Tz'u-hua ts'ung-pien ), this statement can be found in vol. 22, 8.8a.

[2] For a discussion of these questions, see Pauline Yu, "Poems in Their Place: Collections and Canons in Early Chinese Literature," Harvard Journal of Asiatic Studies 50 (1990): 163–96.

[3] Although he is drawing on a number of forebears, Kao Ping (1350–1423) offers the most cogent example of this process in his T'ang-shih p'in-hui .

The role played by anthologies in the process of canon-formation has not gone unnoticed by a variety of critics. In her preface to an anthology of English and American women poets, for example, Louise Bernikow observes that "what is commonly called literary history is actually a record of choices. Which authors have survived their time and which have not depends on who noticed them and chose to record the notice. Which works have become part of the 'canon' of literature, read, thought about, discussed, and which have disappeared depends, in the same way, on the process of selection and the power to select along the way."^[4] Although Bernikow is specifically concerned with the gendered nature of literary-historical empowerment in England and America, her argument for the social and historical construction of value by means of editorial selection offers insights into Chinese literary history as well. In the case of the short lyric poem or shih , the primary questions center on the inclusion or exclusion of individual poets, as they do for Bernikow, with choices being made among writers working within, generally speaking, a common social and linguistic milieu and, perhaps more important, employing a genre blessed with an unimpeachable genealogy. This being so, given the variety of disagreements that ensue nonetheless regarding the critical issues mentioned above, it should come as no surprise that the same lack of consensus holds, a fortiori , for anthologies of the song lyric, or tz'u . This is a discourse, moreover, that must concern itself above all with the question of its own legitimacy. As a genre whose own status within the tradition was ineluctably compromised by the questionable nature of its origins, a genre that for centuries did not even possess a fixed name, the song lyric found itself continually in the position of having to justify itself vis-à-vis that tradition. We can find traces of this gesture in the collections that have written—and rewritten—the history of the form for us, which has been colored from the beginning by the polemics of marginality.

One clear index of the marginal status of the song lyric is the slender and evidently ravaged nature of the body of texts that has survived. It is well known that literati of the late T'ang and Sung dynasties did not choose to include their tz'u in the collections of their own works that many of them were beginning to assemble themselves; contemporary accounts of their writings often do not bother to mention that they worked within the genre.^[5] Individual collections of tz'u that were com-

[4] Cited by Elaine Showalter in her review article, "Literary Criticism," Signs 1 (1975): 438.

[5] Ch'i Huai-mei gives three examples of literati whose catalogs of works given in thedynastic histories do not identify them as also having written tz'u . See "Hua-chien chi chih yen-chiu" in Tai-wan sheng-li Shih-fan ta-hsüeh kuo-wen yen-chiu-so chi-k'an 4 (1960): 516.

piled, such as the Chin-ch 'üan chi ascribed to Wen T'ing-yün (812–70), were both few and aleatory. As Lung Mu-hsün observed some time ago, the relatively small number of individual tz'u collections from the T'ang through the Chin dynasties that are known to have existed—only a few hundred in all—suggests the low esteem of a genre that the literati evidently took great pleasure in but did not deem worth their time and effort to collect.^[6] Needless to say, the survival rate has been unimpressive.^[7] And poets whose primary activity consisted of working within the genre of tz'u have been treated rather capriciously by some literary historians. Biographical information on them, for example, is more often than not both scant and riddled with error. Not only were they likely, over time, to be provided with wrong birthplaces or wrong dates, but they were sometimes even placed within wrong dynasties.^[8]

The same holds true for the anthologies of tz'u that are known to have been compiled, even during the genre's period of greatest florescence. Although the count varies slightly from source to source, there appear to be only nine anthologies that have survived from the T'ang and Sung dynasties, with an additional thirteen whose existence has been recorded but that have long since disappeared.^[9] The losses are difficult to imagine, as is the treatment accorded to the survivors. What is even more remarkable is the fact that only two of these anthologies—the Hua-chien chi and the Ts'ao-t'ang shih-yü —appear to have been widely known to readers of the Yüan and Ming dynasties. All of them were subjected to violations of their textual integrity, ranging from revisions, additions, and deletions of individual poems to division into varying numbers of fascicles, reorganization according to entirely different systems of categorization, and reappearance of identical titles for utterly different volumes. And in many cases, with a few notable exceptions, the very notion of textual integrity is arguable, given the lack of any dis-

[6] In his "Hsüan tz'u piao-chun lun," Tz'u-hsüeh chi-k'an 1, no. 2 (August 1933): 1.

[7] Exhaustive research documenting the traces of individual collections from the T'ang through the Chin dynasty has been conducted by Jao Tsung-i in his Tz'u-chi k'ao , vol. 1 (Hong Kong: Hsiang-kang ta-hsüeh ch'u-pan-she, 1963).

[8] As noted by T'ang Kuei-chang and Chin Ch'i-hua in "Li-tai tz'u-hsüeh yen-chiu shu-lüeh," Tz'u-hsüeh 1 (1981): 8.

[9] This is the count given by She Chih in "Li-tai tz'u-hsüan-chi hsü-lu," pt. 2, Tz'u-hsüeh 2 (1982): 234. His list begins with the Yün-yao chi and goes on to include the following: Hua-chien chi, Tsun-ch'ien chi, Yüeh-fu ya-tz'u, Mei yüan, Ts'ao-t'ang shih-yü, Hua-an tz'u-hsüan, Yang-ch'un pai-hsüeh chi , and Chüeh-miao hao-tz'u .

cernible order to or rationale for inclusion of the contents. Editorial claims regarding the desultory manner of compilation are familiar from the shih anthological tradition but more often than not belie a careful and systematic, if unarticulated, program; in the case of tz'u collections, however, there may be good reason to believe them.

Small wonder, then, that the now-familiar slogan (motivated by assumptions of the cyclicality of both dynasties and genres) juxtaposing T'ang shih and Sung tz'u requires some qualification. Already discredited as faulty periodization by those who have unearthed the roots of the song lyric in the T'ang and examined the richness of its reflorescence in the Ch'ing, it is equally susceptible to the critique that the nice balance of the juxtaposition belies a true incommensurability in status between the two elements. At the same time, however, the assumption regarding the evolution of genres underlying this mode of periodization resurfaces continually as one of the most persuasive arguments for validation of the tz'u . In what follows I will examine the discourse of legitimation as it takes shape in the prefaces to the major anthologies of T'ang and Sung song lyrics that have come down to us as well as in other textual material surrounding them. I should clarify from the outset that my discussion will focus more on the theoretical and rhetorical concerns of the various editors, articulated both explicitly and implicitly in their collections, than on the questions of what texts were available to them and how that affected the representation of works in their anthologies. Consideration of the relationship between the accessibility of sources and the actual practice of selection is clearly necessary but must await another study.

The Legitimizing Effort of the Hua-Chien Chi

Ou-yang Chiung (896–971) provides us with the first critical statement on the tz'u in his preface (dated 940) to the Hua-chien chi edited by Chao Ch'ung-tso. Although this volume was not the first collection of song lyrics to be compiled, it was long assumed to be,^[10] and as the first statement on tz'u by and for members of the literati class it was clearly instrumental in establishing precedents for subsequent discourse on the genre. In both language and allusions, the ornate parallel prose of the text evokes the preface to Hsü Ling's (507–83) Yü-t'ai hsin-yung by stressing the virtues of embellishment and craft in a milieu of languorous ex-

[10] The Ssu-k'u ch'üan-shu tsung-mu t'i-y'ao , 5 vols. (rpt., Taipei: Shang-wu yin-shu-kuan, 1976), for example, identifies it as the oldest anthology of tz'u (5:4457).

travagance. After opening, for example, with the declaration that "engraved jade and carved jasper mimic [nature's] transforming craft with a different art; cut flowers and scissored leaves steal spring's allure and vie in freshness,"^[11] it situates itself within a line of love poetry that extends from the legendary encounters between King Mu and the Queen Mother of the West, through the yüeh-fu tradition, to the song lyrics of Li Po and insists that the poets of the present day need not feel ashamed before these predecessors.

This link to yüeh-fu is stressed throughout the later literature on tz'u ; it is a logical one, given the shared connections of the two forms with musical performance. Moreover, just as yüeh-fu had been institutionalized in both governmental structures and literary practice (having become literati practice), so might the tz'u , too, hope for some similar sort of legitimation. Ou-yang Chiung does not allude to the putative origins among the people common to both forms; all of his forebears belong rather to elite circles, even though, ironically enough, the title of the song supposedly sung by the Queen Mother of the West to King Mu, "Pai-yün yao," was given to the collection of anonymous popular tz'u whose very style and sources Ou-yang Chiung wishes to distance himself from, the Yün-yao chi . Having sketched the opulent setting of the poetry of the Ch'i and Liang dynasties to which the tz'u in the Hua-chien chi bear such strong resemblances, he hastens to distinguish the latter from the eventual contamination from the pleasure quarters suffered by the former: "When the palace-style [poetry] of the southern states was riffled with the songs of the northern wards, not only was the language uncultured, but while elegant, it was without substance." To this phenomenon, recalled at the end of the preface as the "ditties of the lotus boat," he opposes the contents of the Hua-chien chi , explicitly labeled as the "song lyrics of poets" (shih-k'o ch'ü-tzu-tz'u ).

That these are song lyrics of "poets," rather than of "beauties from the southern states" (nan-kuo ch'an-chüan ), reveals Ou-yang Chiung's well-known desire to elevate the social status of the tz'u . One might also note that the poems themselves are organized by their authors, who are arranged—in good canonical fashion—in chronological order of birthdate, a principle that of course implicitly identifies a writer as belonging to a class for whom one would trouble to ascertain such dates. As if to

[11] From Hua Chung-yen, ed., Hua-chien chi chu (Honan: Chung-chou shu-hua-she, 1983), p. 1. Lois Fusek's translation of the preface is complete and annotated, if somewhat loose, in Among the Flowers: The Hua-chien chi (New York: Columbia University Press, 1982), pp. 33–36.

emphasize further the social distinction between these "poets" and other likely composers of tz'u (for example, women, monks, and similar social marginalia), they are referred to by official title or, in the absence of a respectable position, as "scholars in retirement" (ch'u-shih ).

Leading off the collection both in position and, with sixty-six lyrics, in number of poems (and literally referred to later as its "cap") is Wen T'ing-yün. As Wu Hsiung-ho points out, during the Five Dynasties the peripheral state of Shu appears to have offered writers of tz'u a relatively more peaceful setting for their pursuits than would have been possible in the capital on the war-ravaged central plain, a presumption suggested by the unearthing of a variety of musical instruments from the grave of Wang Chien, who established the Western Shu dynasty in Ch'eng-tu (907–25).^[12] It should come as no surprise, then, that almost all of the eighteen poets included were natives of or officials in Shu, the most notable exception being Wen T'ing-yün. Wen is also the only poet mentioned by name in the preface, and his anomalously prominent position suggests that Chao Ch'ung-tso was conducting an effort toward valorization on several fronts: valorization of his region's songs, which are being situated in a line going directly to the center of the empire and an era preceding the political disintegration of the T'ang; of the genre itself, being plied with evident accomplishment by a true shih-k'o , a member of the educated elite, no matter how unsuccessful and profligate he may have been; and of Wen's refined and pedigreed style, which, though by no means the only one included in the anthology, clearly dominates the collection.^[13] Thus, despite the lack of any articulated intent to promulgate a particular school, we can see here the desire to recuperate the tz'u as "literature"; even if situated at best on the fringes, it could certainly partake of the compromised respectability of the Yü-t'ai hsin-yung . At the same time, because of this emphasis on sanctioned authorship, its thematic scope was significantly circumscribed in comparison with the broad range of topics addressed in songs of more popular origin—for example, the texts discovered in the Tun-huang

[12] In T'ang Sung tz'u t'ung-lun (Hangchow: Che-chiang ku-chi ch'u-pan-she, 1985), p. 175.

[13] Thus, for instance, although the tz'u by Huang-fu Sung (mid-9th cent.) selected by Chao Ch'ung-tso for the Hua-chien chi clearly echo Wen's style, Marsha Wagner provides examples of a quite different sort, also composed by Huang-fu, in The Lotus Boat: The Origins of Chinese Tz'u Poetry in T'ang Popular Culture (New York: Columbia University Press, 1984), pp. 146–47. Her book argues persuasively for the mutual infiltration of elite and popular influences in the development of tz'u , but I am focusing rather on the motivations behind the literati discourse on separation.

caves.^[14] The range of tunes employed is also different from that associated with entertainment originating from and intended for a broader social spectrum; the popular tunes whose titles are included in Ts'ui Ling-ch'in's mid-eighth-century Chiao-fang chi , for instance, generally correspond with the tunes to which lyrics are composed in the oldest known anthology, the Yun-yao^[*] chi , compiled before 922 and preserved at Tun-huang.^[15] It may well be that such sacrifices were deemed essential toward defining the new genre as an activity appropriate to members of the literati class.

Additional conclusions follow from considering the Hua-chien chi within the context of other collections. One is the evident murkiness of the line between shih and tz'u that is suggested by the fact that many of the lyrics included in the anthology also appear in T'ang collections of shih .^[16] Another is the curious system by which its contents are arranged; though ordered chronologically by author, the five hundred poems are simply divided into ten fascicles of fifty songs each, which in several instances necessitates breaking up the corpus of selections from an individual poet. Later commentators' displeasure with this system has been virtually unanimous, but the very fact that it appears to be so utterly mechanical suggests the lack of any further programmatic drive motivating the collection. Thus, although the anthology might have hoped to establish the writing of tz'u as an acceptable literati activity, it did not aspire to claim for it the canonical functions of shih . As Wu Hsiung-ho points out, what is therefore notable about the Hua-chien chi lies as much in what it does not say as in what it does. Accordingly, for example, although known for his views on the moral obligations of the shih poet, Ou-yang Chiung breathes no word of them in his preface here, and Niu Hsi-chi (fl. 930), who rails against "lascivious" writing (wen ) elsewhere, is nonetheless represented by eleven lascivious song lyrics in this collection.^[17]

Implicit in these two examples is perhaps one of the major obstacles that confronted any attempt to establish the tz'u within the canon: the fundamentally binary nature of Chinese discourse in general and of literary discourse in particular. It is well known that discussions of liter-

[14] This point is well made by Yang Hai-ming, who notes that the later distinction suggested by Li Ch'ing-chao (tz'u pieh shih i-chia , "tz'u constitutes its own household [distinct from shih ]") is in fact one that pertains specifically to the shih and tz'u of the literati alone. See his T'ang Sung tz'u lun-kao (Hangchow: Che-chiang ku-chi ch'u-pan-she, 1988), p. 52.

[15] Ch'i Huai-mei, "Hua-chien chi chih yen-chiu," p. 515.

[16] As noted in the Ssu-k'u ch'üan-shu tsung-mu t'i-yao , 5:4457.

[17] Wu Hsiung-ho, T'ang Sung tz'u t'ung-lun , p. 284.

ary function and style developed and explored the distinction between wen and pi , patterned and plain writing, respectively, during the Six Dynasties, to be replaced by a pairing of shih and wen , poetry and prose, in the T'ang. To shih , as the classics put it, was accorded the function of "expressing what was intently on one's mind" (yen chih ), the direct manifestation of genuine emotion that could provide access to the fundamental character and political disposition of the poet. And wen —as scholars from Han Yü to Niu Hsi-chi had already claimed, and Sung neo-Confusions would argue further—was to serve as an instrument of pedagogical transformation, a vehicle for the Tao. This schema, with its emphasis on personal expression and moral purpose, left little space for the song lyric, which, given its nature as something in the voice of, and often voiced by someone other than, the poet, was of necessity distanced from an original chih and might at best, as Wu Hsiung-ho points out,^[18] claim a share of what Lu Chi in his "Wen fu" had ascribed to the shih , the ability to "trace sentiments with exquisite ornateness" (yüan ch'ing erh ch'i-mi ). As has been the case with marginalized genres and figures elsewhere, a freedom is hereby made possible, but a vulnerability to frustration and attack as well.

This is nowhere so clear as in colophons to various editions of the Hua-chien chi . At the end of the Shao-hsing period edition of 1148, for example, Ch'ao Ch'ien-chih remarks that these are "all long-and-short verses of talented scholars from the end of the T'ang. The sentiments are sincere and the melodies transcendent, the thoughts profound and the language delicate. Although the writing is ornate and of no salutary [benefit] to the world, it can nonetheless be called skillful."^[19] What he grants with one hand he then just as quickly snatches away with the other. Less equivocal and more controversial are two colophons by Lu Yu (1125–1210) that appear in Mao Chin's (1599–1659) reprinting of his family's Sung dynasty edition of the text. The first observes that the collection "consists entirely of writings of people from the T'ang and Five Dynasties. During precisely this period, when the world was in great peril and the people were struggling ceaselessly to stay alive, scholar-officials degenerated to this point. This is worth sighing over!"^[20] The second offers a somewhat more extensive account of the decline of poetry over the course of the T'ang and likens the introduction of writing songs for the purpose of entertainment to the attitude of reckless

[18] Ibid.

[19] Included in Li I-mang, ed., Hua-chien chi chiao , 2d ed. (Hong Kong: Shang-wu yin-shu-kuan, 1973), p. 221.

[20] Ibid., p. 232.

abandon characteristic of the Six Dynasties, an attitude that Lu Yu feels permeates the Hua-chien chi . Thus, song lyrics represent an inferior genre following only upon the degeneration of the shih , and he finds it difficult to understand how individuals can apply their talents to it: "Brush and ink flow with equal speed, so to be able [to write] one and not the other can never be fathomed by reason."^[21]

Lu Yu's incredulity notwithstanding, poets did write song lyrics, himself included. What his comments share with the praises of Ou-yang Chiung is a tension generated by an impulse toward what Michel Foucault terms a "total history," described as follows:

The project of a total history is one that seeks to reconstitute the overall form of a civilization, the principle—material or spiritual—of a society, the significance common to all the phenomena of a period, the law that accounts for their cohesion—what is called metaphorically the "face" of a period. Such a project is linked to two or three hypotheses; it is supposed that between all the events of a well-defined spatio-temporal area, between all the phenomena of which traces have been found, it must be possible to establish a system of homogeneous relations: a network of causality that makes it possible to derive each of them, relations of analogy that show how they symbolize one another, or how they all express one and the same central core; it is also supposed that one and the same form of historicity operates upon economic structures, social institutions and customs, the inertia of mental attitudes, technological practice, political behaviour, and subjects them all to the same type of transformation; lastly, it is supposed that history itself may be articulated into great units—stages or phases—which contain within themselves their own principle of cohesion.

To this Foucault opposes the notion of a "general history" that allows one to conceive of discontinuity and rupture, that "speaks of series, divi-

[21] Ibid., p. 233. Interestingly enough, the Ssu-k'u ch'üan-shu tsung-mu t'i-yao takes issue with both of these statements. To the first it counters that Lu Yu "has not yet reflected upon the basic reasons behind the imperiled state of the world." And in reference to the second it observes that Lu Yu "does not know that literary genres and styles may be high or low and human learning and vigor may be strong or weak. When one's learning and vigor are incommensurate with genre and style, then there won't be enough to carry it off. If learning and vigor are commensurate with genre and style, then there will be more than enough to carry it off. Regulated verse descended from ancient verse; thus, during the middle and late T'ang much ancient verse was not skillful, and of regulated verse there were at times some fine compositions. Song lyrics in turn descended from regulated verse; thus, during the Five Dynasties poets' shih did not match those of the T'ang, and tz'u alone flourished. This is like being able to lift seventy catties: lift a hundred catties and you'll fall down; lift fifty catties and you can move around as you please. Why is it that this cannot be fathomed by reason?" (5:4457–58).

sions, limits, differences of level, shifts, chronological specificities, particular forms of rehandling, possible types of relations."^[22] Foucault's version of a "total history" aptly characterizes the typically historicist assumptions of Chinese literary history (and hermeneutical practice as well) that motivate the tracing of generic development. In this view change comes not as discontinuity or rupture but as mutation or transformation (pien ) from within, with value connotations—if articulated at all—of decline rather than progress. Thus, a new form like the song lyric possesses an immediate genealogy that is also immediately problematic: on the one hand, the identifiable lineage can ratify the genre by situating it within a network of relations and evolutionary forebears and yet, on the other, it can undermine it as a sheer symptom of decay. Typically, therefore, the introduction to the section on tz'u in the Ssu-k'u ch'üan-shu tsung-mu t'i-yao begins by discussing the unsavory origins and ignoble backgrounds of the first practitioners of the genre but then, because it does not question the notion of the evolution of genres through a succession of transformations (pien ), it cannot deny that the heritage of the song lyric goes back to the Shih-ching . It is, however, but the "aftertone of the yüeh-fu and the last outflow of the poets of the 'Airs.'"^[23] Moreover, as we have already seen, any attempt to valorize the form must be conducted on preexisting terms, those of the canon. In the case of the tz'u , of course, those terms are set by the discourse on shih , and one of the major issues with which critics and anthologists therefore grapple is how—or indeed whether—to work within them.

Southern Sung Revaluations and a New Orthodoxy Defined

Whether one traces any surviving anthologies to the Northern Sung depends on how one dates the Tsun-ch'ien chi ; it is an argument that has been waged for centuries and about which I offer no new information. This collection of 260 lyrics by thirty-six poets is referred to as a collection from the T'ang in the same breath as is the Hua-chien chi by Chang Yen (1248–1320?) in his Tz'u-yüan .^[24] The authors included date from the T'ang and Five Dynasties period and include emperors and high ministers from T'ang Ming-huang onward, arranged, appropriately,

[22] From The Archeology of Knowledge and the Discourse on Language , trans. A. M. Sheridan Smith (New York: Pantheon, 1972), pp. 9–10.

[23] 5:4418.

[24] In the opening of the second fascicle of the work, included in Tz'u-hua ts'ung-pien 2.1a.

largely in descending order of status. Its compiler had clearly seen the Hua-chien chi and essentially followed the earlier collection's sequence of poets; duplications, supplementations, and revisions (often erroneous) are all clearly evident, accumulated through at least three layers of editorial accretion. Some Sung dynasty sources assumed it was a T'ang collection; Ku Wu-fang's preface to the edition he printed in 1582, on the other hand, suggests that he was responsible for its compilation, and Mao Chin believed that Ku had indeed selected the poems.^[25] Internal evidence, however, seems to indicate that it was compiled during the first half of the eleventh century.^[26] In any event, the Tsun-ch'ien chi —like other collections (no longer extant) that were compiled in the wake of the Hua-chien chi and given similar titles referring to banquet contexts—does not appear to have aspired to do more than serve as a convenient songbook. Moreover, its hitherto disputed textual history renders it of little value for the purposes of this discussion.

Whatever the Hua-chien chi 's success in establishing the writing of tz'u as an acceptable literati activity, Northern Sung poets regarded it essentially as a pastime for or about entertainment, and not as a serious form comparable to shih or wen .^[27] This is true despite the fact that the history of literary criticism tells us that many Sung dynasty shih-hua have tz'u-hua embedded within them, and that what could be regarded as the first critical work on tz'u , the Pen-shih ch'ü of Yang Hui (1017–88), dated to 1071, was written close in time to Ou-yang Hsiu's (1007–72) Liu-i shih-hua , one of the first examples of its genre.^[28] Certainly the dearth of

[25] She Chih, "Li-tai tz'u-hsüan-chi hsü-lu," pt. 1, Tz'u-hsüeh 1 (1981): 282–83, and the Ssu-k'u ch'üan-shu tsung-mu t'i-yao , 5:4458. See Daniel Bryant's essay in this volume for an extensive discussion of the anthology's authenticity.

[26] The Tsun-ch'ien chi includes a poem attributed to Li Yü to the tune of "Tieh lien hua" that three Sung sources identify rather as having been composed by Li Kuan, who lived during Jen-tsung's reign (1023–64). At the other end, there is a colophon to the Yang-ch'un chi by Ts'ui Kung-tu that can be dated to the Yüan-feng period of Shentsung's reign (1068–78). I am relying on Hsia Ch'eng-t'ao's argument that the collection was compiled in the T'ang (based on Sung statements to this effect in the Li-tai shih-yü tz'u-hua , the Hua-an tz'u-hsüan , and Chang Yen's Tz'u-yüan ), his consequent chiding of the Ssu-k'u editors' unwillingness to trust the sources enough to date it definitively, and his subsequent recanting of his position after having been persuaded by Wang Chung-wen that it should indeed be dated to the Northern Sung. See his "Ssu-k'u ch'üan-shu tz'u-chi t'i-yao chiao-i," in idem, T'ang Sung tz'u lun-ts'ung (Hong Kong: Chung-hua shu-chü, 1973), pp. 263–65.

[27] Yang Hai-ming cites examples of this attitude in his T'ang Sung tz'u lun-kao , p. 58.

[28] As noted by Wu Hsiung-ho, T'ang Sung tz'u t'ung-lun , pp. 285–86. Only fragments of the Pen-shih ch'ü , which is mentioned in Hu Tzu's (fl. 1147–67) T'iao-hsi yü-yin ts'ung-hua , have survived, having been pieced together by Chao Wan-li; She Chih speculatesthat it had probably disappeared by the end of the Yüan dynasty ("Li-tai tz'u-hsüan-chi," pt. 2, p. 235).

anthologies from the Northern Sung should not be blamed solely on the ravages of time; it attests as well to a persistent ambivalence, one that is examined by Ronald Egan elsewhere in this volume. It remained for poets in the Southern Sung to make the case for a different attitude to the genre.

These strategies can be seen in the surviving anthologies from that period and take different but related forms. One involves the concerted effort to situate the song lyric within a tradition that could be called ya ("elegant" or "refined," but with distinct classical and canonical overtones as well) and subsequently to identify this quality as in fact the "orthodoxy" (cheng ) of tz'u . This strategy not only aims to valorize the song lyric vis-à-vis the dominant shih form, but, even more important, to establish a distinctive and central tradition within the genre of tz'u itself. Another focuses on the individual authorship of tz'u , to shift attention from the conventional and socially conditioned circumstances of its composition and function to its potential as a vehicle for personal expression. These efforts coalesce—during the Southern Sung itself but even more so during the Ch'ing—in the drive to establish schools (p'ai ) with a model master at the head of each (much as Wen T'ing-yün could be identified as the head of the Hua-chien chi tradition). And what they share is the fact that whether the terms with which they are working are borrowed from prior discourse on shih or represent new coinage of the realm of tz'u , they are removed, significantly, from the former's grounding in history and moral intent, facilitating a consequent obsession with questions of language, form, and style.^[29] To what extent these concerns may have undermined the effort toward legitimation that was, in any event, being conducted on alien turf remains to be seen.

Early efforts in this direction are transparent from the record, for among the handful of Southern Sung anthologies whose existence has been noted, two contain the word ya in their titles. Both were evidently compiled shortly after the move south, a transition that no doubt further stimulated the search for legitimacy. One, entitled the Fu-ya ko-tz'u , was edited by someone who refers to himself simply as the "Retired Scholar of T'ung-yang" (T'ung-yang chü-shih) and has not yet been

[29] These strategies can be seen as ways of formulating three of the four aims Lung Mu-hsün isolates as characteristic of anthologies: to be used for singing (pien-ko ), to transmit the person (ch'uan-jen ), to establish a tradition (k'ai-tsung ), and to dignify a form (tsun-t'i ); "Hsüan-tz'u piao-chun lun," p. 27. These three aims represent, of course, strictly literary—and literati—concerns.

identified. The preface is dated 1142 and has been preserved, but the collection itself, which included over 4,300 song lyrics in fifty fascicles from the T'ang through the Northern Sung, has been lost. Its size would seem to have precluded any really rigorous selectivity, but the preface indicates that its primary concern was rather with establishing a creditable genealogy for the genre.^[30] And, indeed, the "return to elegance" announced by the title attests to an awareness that some canonical center to the culture had been lost but, more important, proclaims that it has successfully been recuperated. Another collection, the Yüeh-fu ya-tz'u compiled by Tseng Tsao (preface dated 1146), is clearly attempting to link the song lyric with a genre that had already acquired the mantle of respectability. It singles out ya-cheng as its standard for selection and voices its determination to exclude works that are erotic or jocular and thus leaves out, for example, song lyrics by Liu Yung. Notable, however, is the distinct lack of any discernible "elegance" or "correctness" to its haphazard method of arranging its contents, which consist of song lyrics by thirty-four identified poets of the Sung and over one hundred anonymous pieces. It is not mentioned in any texts from the Yüan or Ming.^[31]

Yet a third and slightly earlier collection employs a somewhat different strategy toward similar ends. Titled Plum Garden (Mei yüan ), it was compiled, according to its preface, by Huang Ta-yü (cognomen Tsaiwan) in 1128, although the edition now extant includes lyrics of later date as well. Like the Hua-chien chi , it contains five hundred works (of poets from the T'ang through Sung dynasties) arranged in ten fascicles, all of which treat the same subject: the compilation of the collection, Huang tells us, had been inspired by a plum tree planted outside his study. Although this topical focus might suggest simply an interest on his part in producing a volume appropriate to a specific social context, it is important to note as well the particular quality of elegance that the plum blossom represents. As the Ssu-k'u editors observe, the collection contributes to the burgeoning effort during the Sung to single this flower out,^[32] and in such a way, we might add, that allowed it to become an emblem for the virtues of the scholar. The Mei yüan thus simultaneously emphasizes an elegance of style and suggests an interest in attesting to an elegance of the song lyricist's character as well.^[33]

[30] See Wu Hsiung-ho, T'ang Sung tz'u t'ung-lun , pp. 303–5.

[31] She Chih, "Li-tai tz'u-hsüan-chi," pt. 1, p. 283.

[32] 5:4458–59. See also Maggie Bickford's study, Bones of Jade , Soul of Ice: The Flowering Plum in Chinese Art (New Haven: Yale University Art Gallery, 1985).

[33] For further information on this volume, see She Chih, "Li-tai tz'u-hsüan-chi," pt.1, p. 285, who notes that despite the unremarkable and—not surprisingly—repetitive quality of many of the lyrics in the collection, it is extremely useful as a source for works, or versions of them, that do not appear in any other anthology of song lyrics. I should like to thank Professor Yang Hai-ming for stressing the importance of the Mei yüan in his comments during the discussion of my conference paper.

A century later, we find renewed interest in placing the song lyric within the shih tradition of individual expression, an interest opposed to the anonymity of the lyrics in the Yün-yao chi and already evident in the arrangement of the Hua-chien chi —and the Tsun-ch'ien chi as well—by author. Essential to the effort to legitimize the genre, however, was the status of its representatives. As mentioned earlier, the strict chronology of the Hua-chien chi marks it as a collection that includes only the members of society for whom dates—and thus social position—could be ascertained. One Southern Sung anthology that shares this concern is Huang Sheng's Hua-an tz'u-hsüan , which combines two separate collections of ten fascicles each, the T'ang Sung chu-hsien chüeh-miao tz'u-hsüan and the Chung-hsing i-lai chüeh-miao tz'u-hsüan . There are two short prefaces dated 1249, one by Huang Sheng and the other by Hu Te-fang, and the statements by both of them are highly reminiscent of those made in prefaces to T'ang anthologies of shih . Hu's opens by emphasizing the rigorousness with which the editor selected the tz'u in the collection from a pool of extremely high quality:

Ancient yüeh-fu are no longer composed, and long-and-short verses [tz'u ] have emerged in their wake. The venerable officials and superior scholars of our dynasty take pleasure in letters, and many have also turned to these [tz'u ]. They are scattered among various collections, however, and are not easy to examine as a whole. In this anthology Yü-lin [Huang Sheng] has been generous in scope and frugal in selection, sending out marvelous tones from a multitude of musical harmonies and bringing forth perfect gems from the midst of a myriad of treasures all on display. Thus, whoever obtains this volume will be able to see in their entirety the wonders of tz'u poets. His [Huang Sheng's] accomplishments are prodigious indeed!^[34]

Huang Sheng's own preface self-consciously situates what he is doing within the brief history of the genre:

Long-and-short verses began in the T'ang and flourished in the Sung. T'ang tz'u are all included in the Hua-chien chi , and Sung tz'u for the most part can be seen in what Tseng Tuan-po [Tseng Tsao] edited. And the Fu-ya collection selects from both T'ang and Sung; it goes up to the end of

[34] Huang Sheng, Hua-an tz'u-hsüan (rpt., Hong Kong: Chung-hua shu-chü, 1973), p. 4.

Hsüan-ho [1126], with more than 4,300 poems in all. Ah, how complete it is! Now, in addition, since the revival of the dynasty writers have emerged one after the other all the way up to the present day. Every individual has tz'u , and each tz'u has different forms. Those who know this but haven't yet seen them, or who have seen some but not all, are countless. In my leisure time I have collected from among several hundred poets and am calling this Selections of Surpassingly Fine Song Lyrics .^[35]

And he goes on to catalog the variety of styles that he is seeking to represent.

Critics generally agree on the judgment Huang Sheng's title passes on his own powers of discernment (although what might strike the contemporary reader as a balance of representation has also been criticized by some as a lack of discrimination). Notable as well is the attention he devotes to establishing proper social and literary-historical credentials for his poets (and his volumes) by arranging them in chronological order; supplying information as to cognomens, sobriquets, and offices held, titles of individual collections, and relevant background details to the poems; and offering his own critical evaluations as well as those of contemporary readers. Such meticulous editorial concern is particularly noteworthy in view of the fact that no contemporary anthologies of shih provide such an impressive critical apparatus (although collections of individual poets certainly do). It is more extensive for the Southern Sung anthology than for the earlier one, and his standards for inclusion have become more socially stringent as well, for, unlike the earlier volume, it contains no song lyrics by monks or women.

Lung Mu-hsün observes that the Hua-an tz'u-hsüan is clearly directed toward "transmitting the author," as indeed this careful attention to biography would suggest. He notes further that this concern leads Huang at times to make questionable judgments about the authenticity of individual lyrics, guided as he was by his own notions of an author's personal style.^[36] Given this penchant for a directly personal mode of expression, as opposed to one focused on the skillful performance of a role, it should come as no surprise that the poets who are best represented in the anthology as a whole are Su Shih and Hsin Ch'i-chi, with thirty-one and forty-two lyrics, respectively. Yang Hai-ming also notes that fondness for Hsin Ch'i-chi's patriotic tz'u in particular is reflected in the selection of lyrics on similar topics by other authors as well.^[37] This interest in the potential for the song lyric to serve as a means of personal

[35] Ibid., p. 158.

[36] Lung Mu-hsün, "Hsüan-tz'u piao-chun lun," p. 13.

[37] Yang Hai-ming, T'ang Sung tz'u lun-kao , p. 291.

and specifically political expression suggests the powerful temptation to appropriate for the tz'u elements of the critical tradition that had grown up around the shih . Huang Sheng's exploration of these possibilities still takes place, however, within the context of an interest in style more elusively and aesthetically conceived. And it was, in any event, not a particularly influential nor widely appreciated direction; if Huang wished seriously to promulgate a particular mode of composition, he was neither polemical enough in his comments nor one-sided enough in his selection to succeed. Despite the recognized quality of the selection, no Sung edition of the Hua-an tz'u-hsüan survives, the earliest extant version dating from 1574.^[38]

Chou Mi's (1232–98) Chüeh-miao hao-tz'u shares with Huang Sheng's collection(s) a disappearance from the literary record that is even more curious and, most obviously, a remarkably similar title. Indeed, one of the main reasons for putting Huang's collections of Surpassingly Fine Song Lyrics from two dynastic periods together and renaming the volume was to distinguish it from Chou's Surpassingly Fine Excellent Song Lyrics (among whose nearly four hundred tz'u by 132 poets he includes his own, perhaps a bit immodestly, in greatest number; most poets are represented by only a handful of works apiece). From the very beginning, critical esteem for this anthology has been as impressive as the precariousness of its transmission. Chang Yen in his Tz'u-yüan , for example, writes that

In recent times the achievements of tz'u poets have been numerous. For example, the Yang-ch'un pai-hsüeh chi^[39] and the Chüeh-miao tz'u-hsüan [by Huang Sheng] can indeed be looked at, but what [their editors] have selected is neither quintessential nor unified. How can they compare with the quintessential purity of what Chou Ts'ao-ch'uang has selected in the Chüeh-miao hao-tz'u ? It is regrettable that its printing blocks have not been preserved; I'm afraid some meddler has also hidden away the printed editions.^[40]

[38] This edition is rarely seen. See She Chih, "Li-tai tz'u-hsüan-chi," pt. 2, p. 229.

[39] An anthology in eight fascicles edited by Chao Wen-li that includes song lyrics in what would later be characterized as the wan-yüeh style, arranged not by author but apparently by tune (but unsystematically at that). Its title alludes to the two songs mentioned by Sung Yü in "Tui Ch'u-wang wen" whose difficulties elude the talents of most musicians (Wen-hsüan , ch. 45; noted by Lois Fusek, Among the Flowers , p. 33 n. 5). Chao places his volume explicitly in the tradition of the Ts'ao-t'ang shih-yü , claiming to be including the song lyrics omitted by the earlier collection; the two anthologies were aiming primarily to serve as songbooks, and they share a reputation for a certain editorial laxity. For a brief discussion, see She Chih, "Li-tai tz'u-hsüan-chi," pt. 2, pp. 230–31.

[40] Ch. 2; in T'ang Kuei-chang, Tz'u-hua ts'ung-pien , 2:9a.

And this was written at most only two decades after the compilation of the anthology.^[41] Rediscovered during the Ch'ing, the Chüeh-miao hao-tz'u then acquired a useful set of annotations with literary, biographical, and historical information—some of it, typically, arrived at inductively from reading the lyrics themselves—from the hands of Cha Wei-jen and Li O (1692–1752) in 1749. In his colophon to the text Li O makes comments similar to those of Chang Yen: "Sung anthologies of Sung tz'u , like Tseng Tuan-po's Yüeh-fu ya-tz'u and Huang Shu-yang's [Huang Sheng's] Hua-an tz'u-hsüan , yield [to the Chüeh-miao hao-tz'u ] in terms of quintessential purity. It sets the standards for tz'u poets."^[42] This opinion is echoed by the Ssu-k'u editors, who add: "Collections of song lyrics by Sung poets have also for the most part not been passed down to the present day. And even the names and surnames of the writers were not known to the world in their entirety. These scattered pearls and chips of jade have all been preserved thanks to this [the Chüeh-miao hao-tz'u ]. Among anthologies of song lyrics, it is by far the best edition."^[43]

Although the Ch'ing edition of the Chüeh-miao hao-tz'u we have today treats the song lyrics included with a literary-historical eye trained on the reading of shih , these critics also recognize that Chou Mi's primary concern lay rather with defining and promoting a distinctive style of tz'u that could, in effect, be considered canonical. Although no preface by him has come down with the collection, this program was evident to later readers in the poets and song lyrics represented in the volume. Cen-

[41] There is a preface to the Tz'u-yüan by Ch'ien Liang-yu that is dated 1317, and some scholars have therefore assumed that the work was completed within two or three years of that date. Wu Hsiung-ho, however, has argued on the basis of both internal and external references that it was written between 1297 and 1307. He also notes that the latest poem in the Chüeh-miao hao-tz'u can be dated to 1295 and that the collection itself was thus probably compiled shortly thereafter, since Chou Mi died in 1298. See his T'ang Sung tz'u t'ung-lun , pp. 315–16, 344–45.

[42] Included, with other prefatory material, in Chou Mi, Chüeh-miao hao-tz'u chien (Shanghai: Shang-hai ku-chi ch'u-pan-she, 1983), p. [xvix].

[43] Ssu-k'u ch'üan-shu tsung-mu t'i-yao , 5:4460. Anthologies like the Yang-ch'un pai-hsüeh chi mentioned above were notorious for the inconsistency and carelessness with which they treated names of tz'u poets included, thus providing small confirmation for Kao Shih-ch'i's opinion, evidently based on Chou Mi's accomplishments, regarding the relative ease of collecting poets of one's own time: "I have discussed how when anthologies examine the past from the present they suffer from being too far away. This is what the Ku-liang [chuan ] calls listening to a distant tone and hearing what is fast and not hearing what is slow. When compared with how contemporaries can research and learn extensively, consider and examine carefully, eliminating infelicities and correcting what is farfetched, later generations have it ten times more difficult." From his preface to the Chüeh-miao hao-tz'u , included in Chüeh-miao hao-tz'u chien , p. [xvi].

tral to it were the terms being explored by many Southern Sung writers, as noted above, of ya and cheng , with their associations with notions of elegance, classicism, correctness, and orthodoxy.^[44] Chang Yen, for example, opens the second section of his Tz'u-yüan with the statement that ancient forms of music all "come from ya-cheng ," and states somewhat later that "in song lyrics one desires elegance and correctness."^[45] In marked contrast to Huang Sheng, who had considered lyrics of what were later to be characterized as the hao-fang and wan-yüeh styles to be "surpassingly fine" in equal measure, for Chang Yen and Chou Mi only the latter meet the criteria of classical elegance and orthodox correctness. Thus, Chang writes that "among the writings of Hsin Chia-hsüan [Hsin Ch'i-chi] and Liu Kai-chih [Liu Kuo, 1154–1206], the song lyrics with heroic [hao ] spirit are not elegant song lyrics."^[46] And as if in agreement with this opinion, Chou Mi includes only three of Hsin's lyrics in his anthology, none of which is written in the patriotic, "heroic" mode with which the latter is associated, but rather in the wan-yüeh style that was becoming marked as the essence of the genre.^[47]

Conflicting Positions

It would be convenient to be able to say that the effort to gain acceptance for the song lyric into the Chinese literary canon was immediately successful and confirmed by the textual record. Unfortunately, however, the evidence indicates that although it was clearly common practice by the end of the Sung for literati to compose song lyrics, the collections that proved most popular in succeeding dynasties were not those that

[44] For a succinct account of the differing but related poetics within which this concept developed, see Grace Fong, Wu Wenying and the Art of Southern Song Ci Poetry (Princeton: Princeton University Press, 1987), esp. pp. 44–56.

Since the notion of cheng was traditionally paired with that of pien in critical discourse going back to the "Great Preface," it should come as no surprise that one persistent tangent pursued by tz'u critics was the determination of the moments in which the history of poetry, or that of the song lyric, witnessed transformations as well. They commonly defined the tz'u itself as a transformation of the regulated verse or of the chüeh-chü and also located various subsequent "mutations" within the genre itself. These included, for example, those identified as having occurred after Wen T'ing-yün, after Su Shih, after the Northern Sung, and after the Southern Sung.

[45] Tz'u-hua ts'ung-pien , 2:1a and 9b.

[46] Ibid., 2:10a. Emphasis mine. The distinction between the hao-fang and wan-yüeh styles was made by the Ming dynasty Chang Yen in his Shih-yü t'u-p'u of 1594–95, in which he states in the introduction that in general the latter is the orthodox style for tz'u . As noted by Wu Hsiung-ho, T'ang Sung tz'u t'ung-lun , pp. 157–58.

[47] As noted by Yang Hai-ming, T'ang Sung tz'u lun-kao , p. 291.

promulgated the most literary, elegant, or classical vision of tz'u . As mentioned above, the only two anthologies that seem to have circulated at all during the Yüan and Ming dynasties, each in multiple editions, were the Hua-chien chi and the Ts'ao-t'ang shih-yü ,^[48] with the others literally disappearing from sight. Nor does the handful of collections produced during this period deviate noticeably from the model set by the latter. The principles of selection are similarly unclear,^[49] and so are the methods of organizing the lyrics, although they tend to be arranged either by topic or by tune length. In the case of the Tz'u-lin wan-hsüan of Yang Shen (1488–1559), for example, the tz'u selected vary widely in quality and appear in no discernible order, with songs by the same author scattered widely throughout the volume. There is no system to the nomenclature, with individuals referred to variously by given name, cognomen, or sobriquet, and there are frequent errors in attribution of authorship.^[50]

[48] The full title of this volume as it appears on the earliest extant edition, dating from 1392, is Tseng-hsiu chien-chu miao-hsüan ch'ün-ying ts'ao-t'ang shih-yü . Broad in coverage and of unknown editorial hand, this Sung collection has been criticized roundly for the lack of rigor in its selection. It is notable both for its arrangement by topic, which suggests close ties to an entertainment context and function as a source for songs appropriate to a particular situation, and for the remarkable disregard for this aspect of textual integrity over time. On the first point, Yang Hai-ming cites the Ch'ing scholar Sung Hsiang-feng's (1776–1860) observation that the versions of many lyrics as they appear in the anthology often differ from those in the poets' individual collections, disclosing revisions that were evidently made to render them more suitable for singing (T'ang Sung tz'u lun-kao , pp. 289–90). And as for the second, one Ming edition of 1538 tinkered with the text by reducing the number of topics, placing the songs in a different order, and revising many of the annotations. Another volume, edited in 1550 by Ku Ts'ung-ching and titled Lei-pien ts'ao-t'ang shih-yü (which enjoyed the widest circulation but was itself subject to further emendations), completely rearranged the songs from categorization by topic into categorization by tune pattern and length, reflecting, among other things, a loss of almost all of the actual music by the Ming and the consequent need to have the songs' prosodic forms presented in a more marked fashion. As the Ssu-k'u editors point out, the versions of the songs in this edition were used as bases for character counts of tunes in later prosodic registers of tz'u (Ssu-k'u ch'üan-shu tsung-mu t'i-yao , 5:4459). For further information on the Ts'ao-t'ang shih-yü , see She Chih, "Li-tai tz'u-hsüan-chi," pt. 2, pp. 226–27.

[49] One notable exception, however, is the Ming-hung yü-yin compiled during the Yüan by P'eng Chih-chung (Hsien-yu-shan tao-shih), an anthology of "Taoist" tz'u written by thirty-six poets from the T'ang dynasty on. Many of the song lyrics included have to do with meditation, fasting, and alchemy, and a fair number of the tunes employed appear to be sui generis, having been devised for a particular occasion or topic. See Chao Tsun-yüeh, "Tz'u-chi t'i-yao," pt. 4, in Tz'u-hsüeh chi-k'an 3, no. 1 (March 1936): 41–44.

[50] She Chih, "Li-tai tz'u-hsüan-chi hsü-lu," pt. 3, Tz'u-hsüeh 3 (1985): 276. The Ssu-k'u ch'üan-shu tsung-mu t'i-yao (5:4480) also takes issue with both the claims and the qualityof Yang Shen's collection, criticizing the credibility of his assertion in the preface that he had gone through the collections of five hundred T'ang and Sung tz'u poets; the fact that the anthology turns out to include poets from the Chin, Yüan, and Ming dynasties as well resolves this problem of numbers but then opens the way for another critique regarding the lack of correlation between the preface and the contents of the volume. The editors also object to what they consider to be the crudeness of Yang's comments on the lyrics themselves.

Yang Hai-ming observes that this phenomenon reveals what kind of tz'u —those with no loftier aspirations than to suit an entertainment context—were valued throughout this period.^[51] There seems obviously, at the same time, to have been a remarkable fall-off in the composition and preservation of all types of tz'u in general; the introduction to a seventeenth-century collection of tz'u by women, for example, laments the loss of what it estimates to have been half of all song lyrics ever written and bewails the difficulty ("ten times greater") of collecting tz'u when compared to shih , for which individual collections are both of higher quality and more accessible.^[52] The familiar theory of the evolution of genres would account for the virtual disappearance of the song lyric by pointing to the supersession of ch'ü , but that, of course, tells us very little.

What does seem clear, first, is that the movement whose development we have been tracing is one whose outlines are visible probably only in hindsight. We have in truth but scattered bits of evidence: two anthologies with the word ya in their titles; later critics' approbation and discussion of a collection about whose aims its own editor, Chou Mi, was silent; and a few isolated critical statements, however articulate. Second, the discourse on tz'u seems as yet unable to resolve its potentially contradictory aims, either of accommodating itself to the legitimating terms established by the shih tradition or of insisting on its own distinctiveness as a genre, heralded by Li Ch'ing-chao's oft-cited statement in her "Tz'u-lun" that the song lyric was of a different species altogether—an attitude evident in Chang Yen's Tz'u-yüan as well. The accommodation was as yet halfhearted, for the attractions of a space and mode within which to speak those things that could not be spoken in shih were powerful ones, yet the insistence on absolute differentiation ran the risk of utterly marginalizing that mode. Third, the lack of a systematic attempt at this point to canonize either the genre itself or its

[51] Yang Hai-ming, T'ang Sung tz'u lun-kao , p. 289.

[52] As cited by She Chih, "Li-tai tz'u-hsüan-chi hsü-lu," pt. 4, Tz'u-hsüeh 4 (1986): 246–47. The volume is the Lin-hsia tz'u-hsüan , edited by Chou Ming; three prefaces to the collection are dated 1671.

specific practitioners should hardly surprise us, given the fact that that process was just seriously getting under way for shih . Yen Yü's Ts'ang-lang shih-hua , which ground the historical, theoretical, and critical lenses within which T'ang poetry was to be examined, was itself but a product of the Southern Sung, and Kao Ping's T'ang shih p'in-hui , the first anthology to develop Yen Yü's model at length, did not appear until the beginning of the fifteenth century. Indeed, one might partially attribute the lack of attention paid to tz'u during the Yüan and Ming dynasties to the obsession with both defining the canon of shih and, implicitly, defending the latter's traditionally venerated position against the increasingly powerful encroachments of the claims of wen , or prose, as the truly serious vehicle of expression (demonstrated quite graphically, for example, in the elimination of a test of shih composition from the civil service examination until the middle of the eighteenth century). As suggested earlier, the tendency for discussions to set their terms in binary relationships—in this case, shih and wen —left little space for tz'u , the tertiary other. Thus, even in the Ch'ing dynasty we find an anthologist resorting to an involved argument that first creates a legitimate space for tz'u but then consigns it to the margins of respectability.

This text, Fu Hsieh-t'ung's Tz'u-kou ch'u-p'ien , which exists only in manuscript, displays the potentially contradictory attitudes just outlined. In the middle of his preface, for example, Fu presents a history of the song lyric from the Hua-chien chi that catalogs the variety of styles developing throughout the Sung and asserts, in a gesture designed to accommodate difference, that "[Huang] Hua-an's Chüeh-miao tz'u-hsüan both continues the Hua-chien chi and also paves the way for the Ts'ao-t'ang . Now alluring thoughts and intricately beautiful language are stimulated and proceed in many ways," and just as the ancients regaled themselves on delicacies of many different sorts—leeks, mullet, broth, and wine, all served at the same meal—so can tz'u "be joined together with shih and wen to please all tastes." He then reviews poetic forms of the past, from the Shih-ching through the yüeh-fu of the Han and up to the T'ang chüeh-chü , defining each as the song lyric of its respective era. In doing so, he provides the tz'u with nearly hegemonic validity.

Fu makes clear that his primary interest is in the music of tz'u , whose vanishing he can only lament. "Therefore," he continues, "since what is included in this cannot seek to accord with the ancients by means of their music, it must use the syntax and diction of contemporary tz'u to accord with the syntax and diction of ancient tz'u . As for the melodies and pitches, they will await someone who knows the tones." In his "General Principles" for the anthology, however, Fu invokes this con-

cern with rhyme and pronunciation to downplay what might have been the literary aspirations of the collection. He offers it as the reason behind his relatively casual manner of compilation, which, he claims, simply sets the song lyrics down as he read them, in no particular order and without regard for, or indication of, an author's official status or biography. And he concludes with a remarkably straightforward assessment of the prevailing disesteem for the genre—"tz'u is certainly the Lesser Path [hsiao tao ]"—and explains: "Writers of shih and wen for the most part regard filling in song lyrics as an inferior form, and students of the classics also see it as an activity that is not pressing. Therefore, while those who have found it interesting to dip their writing brushes in it are numerous, specialists in this path are few."^[53]

Canons and Canonicity in the Ch'ing

Given the lack of a canonical definition of the origin and function of tz'u analogous to that possessed by shih (however restrictive the latter may have proved for both practice and interpretation), it is small wonder that so much critical energy came to be expended on providing some self-definition for the song lyric. Nor, furthermore, that the urgency—if not insecurity—endemic to its situation ran the risk of ossifying those definitions into rigid, exclusionary "schools" for whom only certain individuals could serve as exemplary models or "patriarchs." Nor, finally, that this discussion developed into a preoccupation with aspects of prosody, language, and style that marked a domain to which the other genres did not care to lay first claim.^[54]

[53] Included in Chao Tsun-yüeh, "Tz'u-chi t'i-yao," pt. 3, Tz'u-hsüeh chi-k'an 2, no. 3 (April 1935): 83, 88–90. Li O, who together with Cha Wei-jen had provided annotations to Chou Mi's Chüeh-miao hao-tz'u , comes to a similar conclusion in his preface to the Ch'ün-ya tz'u-chi . He argues there that because the song lyric is intrinsically a low form, if it does not possess elegance as its woof (wei ), it will lose its correctness (cheng ). As cited by Lung Mu-hsün, "Hsüan-tz'u piao-chun lun," p. 20.

[54] It should go without saying that this emphasis on the stylistic distinctiveness of tz'u was shared by all critics, whatever their aspirations for the position of the genre. One Ch'ing anthology, for example, that explicitly places itself in the tradition of the Ts'ao-t'ang shih-yü —against which scholars promoting a more elegant notion of the form railed bitterly—provides one of the most succinct statements to this effect. This is the Ts'ao-t'ang ssu-hsiang in four fascicles, edited during the K'ang-hsi period by Ku Ts'ai. The title of the volume, of course, acknowledges its model, as does its specification of a certain number of topics and organization into chüan by length of tune (with hsiao-ling defined as consisting of any lyric up to 59 words, chung-tiao as 60 to 92 words, and ch'ang-tiao as anything longer; Ku observes that Chu I-tsun criticized these limits, but Ku defends himself by saying that "since these names exist, it can be done"). At the same time, however, theintroductory principles to the collection offer interesting refinements on some of the standard war-horses of the critical tradition. On the theory of the evolution of genres, for example, Ku writes that "when shih died out and then tz'u were composed, it was not that shih died out, but that the means of singing shih died out. And when tz'u died out and Northern and Southern ch'ü were composed, it was not that tz'u died out, but that the means of singing tz'u died out." He also presents a rhetorically effective argument for the specificity of the song lyric as a genre:

If one can use T'ang shih to write tz'u , then one's tz'u will be excellent; however, tz'u are definitely not shih . If one can use Sung tz'u to write ch'ü , then one's ch'ü will be excellent; however, tz'u are definitely not ch'ü . For tz'u has a form and structure peculiar to tz'u , and a music and sentiment peculiar to tz'u . If it resembles shih then it will be too literary, and if it resembles ch'ü then it will be too unpolished, and both of these are defects. As a comparison, if one uses the Shih [-chi ], the Han [shu ], and the Eight Great Masters to write pa-ku-wen , it will certainly be excellent, but will pa-ku-wen thereupon become ku-wen ?

Included in Chao Tsun-yüeh, "Tz'u-chi t'i-yao," pt. 1, Tz'u-hsüeh chi-k'an 1, no. 1 (April 1933): 96, 95.

Wu Hsiung-ho, in discussing the aftermath of the Ming dynasty Chang Yen's distinction between wan-yüeh and hao-fang , argues that it was made both with a very personal sense of style in mind—as opposed to a more generalizable doctrine—and also on the basis of a limited acquaintance with tz'u poets of the Sung, and that it was only later that critics like Wang Shih-chen (1634–1711) transformed features originally intended merely to characterize particular individuals into schools with identifiable "patriarchs."^[55] Even more central to this new discourse on the song lyric was Wang's contemporary Chu I-tsun (1629–1709), who has traditionally been credited with the resurgence of tz'u to the critical and literary-historical arena after several centuries of relative slumber.^[56]

[55] In his Hua-ts'ao meng-shih , for example, Wang Shih-chen identifies his fellow natives of Chi-nan, Li Ch'ing-chao and Hsin Ch'i-chi, as the exemplars of the two styles, respectively. Wu's discussion also touches on the various ways in which other critics disagreed with or modified the distinction (T'ang Sung tz'u t'ung-lun , pp. 158–64).

As Wu points out earlier, the notion of "school" goes back to Lü Pen-chung's (1136 chin-shih ) grouping of twenty-five poets into the Chiang-hsi shih-p'ai with Huang T'ing-chien as its patriarch, based on similar practices already common in Ch'an hagiography (p. 152). A version of this can be seen in Wang Cho's linking of tz'u poets with either Su Shih or Liu Yung in Wang's Pi-chi man-chih (ca. 1145–49) (ch. 2, included in Tz'u-hua ts'ung-pien , 1.1b; also noted by Wu Hsiung-ho, T'ang Sung tz'u t'ung-lun , p. 155). But the precedent for establishing genealogies is, of course, of much older vintage, having been set early in the sixth century by Chung Jung in his Shih-p'in , which identified poets being ranked as belonging to the tradition of either the Shih-ching or the Ch'u-tz'u .

[56] Impressive evidence to this effect strikes the eye immediately when one opens T'ang Kuei-chang's Tz'u-hua ts'ung-pien . Of the eighty-five tz'u-hua included, eleven date from the Sung, two from the Yüan, four from the Ming, and all of the rest from the Ch'ing.

The reasons behind this renaissance, and Chu's particular contributions toward it, are too numerous and complex to be examined here.^[57] Among others, one might cite the widespread fascination during the early Ch'ing with antiquity in general and the Sung dynasty in particular; the regional interest of Chu, a native of western Chekiang, in the many Southern Sung poets from that same geographical area;^[58] his discovery and reprinting of Chou Mi's Chüeh-miao hao-tz'u ; and the contemporary preoccupation, profoundly conservative at its core, with collecting, defining, and mastering the entire literary tradition. This effort shapes his Tz'u-tsung , compiled by 1678.

It is well known that Chu compiled this anthology in twenty-six fascicles (with four more added by Wang Sen [1653–1726]) of song lyrics from the T'ang through the Yüan dynasties on the basis of a limited number of individual collections, and even fewer anthologies, that were circulating at the time. His introduction comments on the losses that he knows to have occurred (to which later scholars have added), lists the volumes he has been able to peruse, and catalogs the various options other editors have employed in recording the name, title, place of birth, and so on, of the poets included. Also well known is Chu's desire to revise significantly the image of the tz'u prevalent at the beginning of the Ch'ing, requiring a small, profound, and oft-cited revision of literary history: "People now say that one must praise the tz'u of the Northern Sung; however, only in the Southern Sung did tz'u reach the ultimate craft, and only at the end of the Sung did it reach the ultimate transformation. Chiang Yao-chang [Chiang K'uei] is the most outstanding [poet of the genre]. What a pity it is that of Pai-shih's [Chiang K'uei's] yüeh-fu in five fascicles today only twenty-odd pieces have survived."^[59]

Wang Sen's preface to the Tz'u-tsung articulates the theoretical program of the anthology at somewhat greater length. He begins by offering a famous attack on the view of literary history that had characterized the song lyric as the "remnant" of the shih , an attitude encapsulated in the very title of the much-reviled Ts'ao-t'ang shih-yü . Long-and-short verses, he opens, have been around as long as shih , so the tz'u can boast a lineage as venerable as that of its better-placed cousin. Moreover, the

[57] Madeline Chu touches on some of these issues in her article "Interplay between Tradition and Innovation: The Seventeenth-Century Tz'u Revival," in Chinese Literature: Essays, Articles, Reviews 9 (1987): 71–88.

[58] Chu articulates this interest clearly in his preface to Meng Yen-lin tz'u , as cited by Lung Mu-hsün, "Hsüan-tz'u piao-chun lun," pp. 18–19.

[59] Chu I-tsun, Tz'u-tsung (1691; rpt., n.p.: Chung-hua shu-chü, 1973) 1.5b.

familiar evolution from ancient verse to yüeh-fu to regulated verse to tz'u , he declares, has produced forms that are like horses with "separate bits but galloping side-by-side, not one in front and the other behind. To say that shih degenerated to become tz'u or to consider tz'u as the remnant of shih is not an argument that penetrates the whole."^[60] After touching on the vibrant but qualitatively mixed early history of the genre, he identifies his patriarch: "Chiang K'uei of Hsiang-yang emerged, with lapidary verses and refined words that return to purity and elegance. There-upon Shih Ta-tsu and Kao Kuan-kuo flanked him on either side. Chang Chi and Wu Wen-ying took him as master first, and Chao I-fu, Chiang Chieh, Chou Mi, Ch'en Yün-heng, Wang I-sun, Chang Yen, and Chang Chu studied him afterward."^[61] Whatever disagreements one might have with the hierarchy Wang provides here—and they have been numerous—the point remains that he is singling out tz'u poets from the Southern Sung who embody in different ways ideals of elegance and refinement that he and Chu are seeking to establish as the preeminent style for the genre.^[62] Their instrument will be this anthology, which he hopes "may eliminate at once the vulgarity of the Ts'ao-t'ang ; those who write to music will then know the orthodox tradition [tsung ]."^[63]

The language Wang employs here and the critical order of evaluation he presents are reminiscent not only of descriptions of the Kiangsi poetry group,^[64] but of Kao Ping's categorization of the T'ang poets as well, which in turn was heavily influenced by Yen Yü's example. And the selection of song lyrics for the volume itself, however hampered by the limited number of sources and deficient in text-critical oversight,^[65] also reflects the priorities that the two compilers established. Whereas some poets only have one song lyric each included, those from the

[60] Chu I-tsun, Tz'u-tsung 1.2a–2b.

[61] Ibid. 1.3a–3b.

[62] Chiang K'uei was a particularly appealing model because, as Lin Shuen-fu puts it, his life was one "almost entirely devoted to the cultivation of art." See The Transformation of the Chinese Lyrical Tradition: Chiang K'uei and Southern Sung Tz'u Poetry (Princeton: Princeton University Press, 1978), p. 58.

[63] Chu I-tsun, Tz'u-tsung 1.5a–5b.

[64] As noted by Lung Mu-hsün, "Hsüan-tz'u piao-chun lun," p. 17.

[65] Ting Shao-i, who compiled a Ch'ing tz'u-tsung pu in 1894 (rpt., Peking: Chung-hua shu-chü, 1986), says that he has corrected many—but not all—of the errors in the Tz'u-tsung , which he attributes both to the unavailability of good editions of Sung and Yüan tz'u and to Chu's failure to compare versions from one edition to the next. In his T'ing-ch'iu-sheng-kuan tz'u-hua , as cited by She Chih, "Li-tai tz'u-hsüan-chi hsü-lu," pt. 5, Tz'u-hsüeh 5 (1986): 257.

Southern Sung mentioned in the introduction and preface are heavily represented in the anthology; the thirty-two tz'u of Chiang K'uei, for example, probably constituted a significant proportion of the total number of works by the poet that Chu had seen, and the lesser members of the group are present in similar depth. The collection's insistence on the stylistic distinctiveness of the song lyric as a genre confirms what John Guillory, drawing on Bakhtin, has argued to be fundamental to the process of canon-formation, the marking of a language that can be defined as the specifically "literary" coin of a hermetic, privileged community and the final expunging of the popular or "vulgar" traces of the form.^[66]

Arguments waged, however, concerning the precise nature of that language. However unsuccessful Chu I-tsun may ultimately have been in eliminating "heterodox"—whether "vulgar" or "heroic"—styles of tz'u and controlling the quality of lyrics written according to "orthodox" models, his influence on the subsequent history of the genre is undeniable. Although critical opinion agrees that followers of the Che school eventually declined into a vacuous or, worse, "lascivious," preciosity, Chu's framing of the discourse was adopted by later scholars who were adamantly opposed to his aims. The very fact that critical discussion coalesced into schools with distinct programs may reflect to a certain extent the power of Chu's example. Moreover, many of his key assumptions remained intact, albeit developed in very different ways. Thus, scholars associated with the Ch'ang-chou school that flourished a century later might have rejected his notion of what was orthodox but not the presumption that some orthodox or canonical style existed and should be promulgated. Chou Chi (1781–1839) in his Tz'u-pien , for example, clearly identifies a tradition beginning with Wen T'ing-yün that he considers cheng and one starting with Li Yü that is pien , and he meticulously lists the poets that fall under either category. Other critics, following a model developed earlier for discussing and classifying painters, which itself was borrowing a precedent established in Ch'an Buddhism, divided tz'u poets into a Northern and a Southern tradition (tsung ) and, much as had been the case in the visual arts, found the latter decidedly superior.^[67]

The Ch'ang-chou critics differed significantly from the earlier Che school, however, in the manner by which they sought to establish the canon of the song lyric. Rather than insisting on the peculiar distinctiveness of tz'u , scholars like Chang Hui-yen (1761–1833) and Chou Chi

[66] See his "Canonical and Non-canonical: A Critique of the Current Debate," ELH 54 (1987): 483–527.

[67] As noted by Wu Hsiung-ho, T'ang Sung tz'u t'ung-lun , p. 163.

sought to rehabilitate the genre by appropriating for it the hermeneutics that had developed in connection with the Shih-ching , thereby accommodating it to the established shih tradition. Thus, the inscription of the male gaze and the ascription of male desire to the languishing woman in lyrics by poets like Wen T'ing-yün could be re-viewed as the figuration of political concern about the state of the empire and the frustration of a loyal official at his inability to do anything about it. The details of this program have been discussed at length by Chia-ying Yeh Chao^[68] and need not be recapitulated here. Suffice it to say at this point that although Chang Hui-yen has been faulted both for his outrageously improbable allegorical interpretations and for assuming that a valid comparison could be made between the song lyric and the Book of Songs at all, the former had already been accepted practice for several centuries and the latter could be supported, as we have seen, by several textual precedents within the literature on tz'u as well.

Later Redefinitions

Other tz'u scholars in the nineteenth century followed Chang's and Chou's leads in different ways. The rich literature on and of the song lyric in the nineteenth century attests to the transforming influence of the Ch'ang-chou school, and I shall mention only two examples. The first is a general discussion introducing an unpublished anthology in eight fascicles, the Tz'u-kuei (dated 1863), compiled by Yang Hsi-min, who makes a number of points that suggest his decision to situate the song lyric unambiguously within the shih historical and critical tradition. First, he interprets the shih-yü epithet in such a way as to insist on the connection, rather than the differentiation, between the two forms: "Long-and-short verses are remnants of shih . This being so, shih then is the source and tz'u the tributary. If the source is not distant, how can the tributary be long?" And he goes on to stress the fact that tz'u poets like Wen T'ing-yün, Wei Chuang, Yen Shu, Yen Chi-tao, Ch'in Kuan, and Ho Chu could all write shih poetry as well; that Su Shih and Huang T'ing-chien were especially renowned for the latter; and that tz'u poets whose other writings are not worth reading are "like tributaries without the source." Second, he de-emphasizes the distinguishing feature of the song lyric, its musical performance, by arguing that "even though the ancient poems all were written to music, later poems without music are

[68] In "The Ch'ang-chou School of Tz'u Criticism," in Chinese Approaches to Literature from Confucius to Liang Ch'i-ch'ao , ed. Adele Austin Rickett (Princeton: Princeton University Press, 1978), pp. 151–88.

numerous and are not therefore prevented from being good poems. But a tone and rhythm that come naturally cannot be lost. With tz'u the situation is the same. If you can sing, you will certainly be good; but if you cannot sing, are you therefore not a tz'u poet?" He then observes that few people can understand the musical notation Chiang K'uei provided for his song lyrics.

Yang Hsi-min also argues that critical concepts apply with equal validity to the two genres. For example, he writes that "some people think that in tz'u one esteems having the meaning within and the words on the surface [i nei yen wai ], but those who understand this are few. They fail to realize that having meaning within words is true of all writing of subtlety, and not just the song lyric." The most compelling incorporation of the shih tradition can be seen in a statement that appears more than once in his discussion, a statement to the effect that "in studying tz'u one should begin with yüeh-fu of the Han, Wei, and Six Dynasties, and then take Wen [T'ing-yün], Wei [Chuang], the two Yens, Ch'in [Kuan], and Ho [Chu] as the canonical orthodox tradition [cheng tsung ]."^[69] Chao Tsun-yüeh notes that this is modeled on Kao Ping's discussion of T'ang poetry, but the precedent in fact goes back to Yen Yü, who was particularly concerned with the question of systematic study of past models.^[70]

As the admiring reference above to Su Shih and Huang T'ing-chien should suggest, Yang Hsi-min's tastes run precisely to the Northern Sung tz'u poets whose standing Chu I-tsun and his school had been anx-

[69] Chao Tsun-yüeh, "Tz'u-chi t'i-yao," pt. 2, Tz'u-hsüeh chi-k'an 1, no. 2 (August 1933): 83–84.

[70] Yen Yü's prescription, of course, appears in different forms and is considerably longer. One version recommends the following:

First, one must thoroughly recite the Ch'u-tz'u and sing them morning and night so as to make them his basis. When he recites the "Nineteen Old Poems," the "Yüeh-fu in Four Sections," the five-syllabic poetry of Li Ling and Su Wu and of the Han and the Wei, he must do them all thoroughly. Afterward, he will take up the collected poetry of Li [Po] and Tu [Fu] and read them [the poems] in dovetail fashion as people of today study the classics. Next, he will take up comprehensively the famous masters of the High T'ang. Having allowed all this to ferment in his bosom for a long time, he will be enlightened spontaneously [tzu-jan wu-ju ]. Although he might not attain the ultimate of study, still he will not go off the correct road.

From Ts'ang-lang shih-hua chiao-shih , ed. Kuo Shao-yü (Peking: Jen-min wen-hsüeh ch'u-pan-she, 1962), p. 1. Trans. (with slight revisions) Richard John Lynn, "Orthodoxy and Enlightenment: Wang Shih-chen's Theory of Poetry and Its Antecedents," in The Unfolding of Neo-Confucianism , ed. W. T. de Bary, Studies in Oriental Culture, no. 10 (New York: Columbia University Press, 1975), p. 220.

ious to diminish. Ironically enough, Yang employs the same verb hsi , "wash away," "eliminate," that Wang Sen had used in speaking of the Tz'u-tsung's accomplishment vis-à-vis the Ts'ao-t'ang shih-yü , to describe the contribution made by major Northern Sung poets: "Ou[-yang Hsiu], Su [Shih] and Huang [T'ing-chien] are unrestrained and unconventional, genial and liberal, heroic and strange. They eliminated at once the practice of writing about latticework and gauze; this is how the orthodox was transformed and became [a new] orthodoxy." He also takes Tsou Hsü-shih (Tsou Chih-mo, 1658 chin-shih ) to task for criticizing Northern Sung poets' failure to write long tz'u in great numbers. This reflects a poor understanding of literary history, Yang argues, for all forms of a genre are not always simultaneously to a poet's hand at any given moment in time. To say that Su and Huang could not write long tz'u makes little sense if Liu Yung was the first to develop the form. Like Chu and other early Ch'ing critics, Yang concludes, Tsou is too one-sided: "When shopping for food, he only knows how to buy what is inexpensive; he has never heard the subtleties of tz'u ."^[71]

An earlier passage of the text articulates in a different manner Yang's rejection of the narrow, doctrinaire exclusivity that he finds characteristic of critics at the beginning of the dynasty:

Now writing has its roots in inborn emotions and is assisted by learning; when both are present, then as soon as one sets the brush down to send forth words, the workings of heaven will start of themselves. Nothing has been determined with regard to long composition or short piece, nor between the "clear and empty" [ch'ing-k'ung ] and the "solid and substantial" [chih-shih ] styles. The Shih [-chi ] is not the same as the "[Li-]sao"; the "Sao" is not the same as the Chuang-tzu ; the Kung [-yang ] and Ku [-liang ] are not the same as the Tso [-chuan ] and the Kuo [-yü ]—how can one therefore settle on one explanation?^[72]

Although considerably more eclectic and flexible in his judgments, Yang Hsi-min has clearly been influenced by the views of Chang Hui-yen; he nonetheless feels that the latter's school has still not paid suf-

[71] In Chao Tsun-yüeh, "Tz'u-chi t'i-yao," pt. 2, p. 85. Tsou Chih-mo's position is also presented by Grace Fong in her Wu Wenying and the Art of Southern Song Ci Poetry , p. 164. One might, of course, take Yang to task in turn for assuming that genres have life histories of their own that evolve independently of human agency, but given the established discourse, it would have been difficult for him to think otherwise.

[72] Chao Tsun-yüeh, "Tz'u-chi t'i-yao," pt. 2, p. 84. Ch'ing-k'ung and chih-shih were terms first used by Chang Yen in his Tz'u-yüan to establish the superiority of Chiang K'uei as the exemplar of the former style. Along with the Ming dynasty Chang Yen's distinction between wan-yüeh and hao-fang , they stand at the center of the critical terminology developed to refer specifically to the song lyric.

ficient attention to a significant dimension of tz'u history. Thus, on the one hand he notes that in selecting songs for his anthology he has looked for those with pi-hsing , "comparison and evocative images," in other words, metaphorical imagery that would lend itself to the interpretive operations favored by the followers of Chang Hui-yen's method. He also credits Chang for having liberated scholars of tz'u from the fetters of Chu's Tz'u-tsung by "going against the current" to bring Wen T'ing-yün to prominence, thereby "opening up a new realm" in which tz'u incorporates pi-hsing . On the other hand, however, he then chooses to focus his critical attention on a group of poets different from that favored by the earlier critic: "But the path of [Ou-yang] Liu-i, [Su Tung-]p'o, and [Huang Shan-]ku has not yet been traversed by many wooden clogs. For an era to have brave heroes, it must not shrink from asking where the ford is."^[73] And there is no question that this relatively broad-minded attitude has indeed been evident throughout his entire discussion.^[74]

Finally, we can see a balance between the positions of Chu I-tsun and Chang Hui-yen explored quite literally in the writings of the late Ch'ing critic Ch'en T'ing-cho. Ch'en is best known as a proponent of the theories associated with the Ch'ang-chou school, which he discussed in his Pai-yü-chai tz'u-hua of 1892 and illustrated in his anthology Tz'u-tse . Less than twenty years earlier, however, in 1874, Ch'en had edited a collection entitled Yün-shao chi and explained the principles behind it, which at that time were wholeheartedly those of Chu I-tsun, in a work called the Tz'u-t'an ts'ung-hua . A recent reprinting of the Pai-yü-chai tz'u-hua provides extensive documentation on these two efforts, so I shall not discuss them in detail here. The two pairs of texts do, however, not only articulate quite clearly the general principles of both the Chu and the Chang schools, but also present us with positions that are somewhat more supple than those of either faction.

The Yün-shao chi , which exists only in draft form, includes 3,434 songs by over 1,100 poets from the T'ang through the Ch'ing dynasties. Interestingly enough, given Ch'en's repeated insistence in the Tz'u-t'an ts'ung-hua on the distinctiveness of tz'u , the anthology is modeled on Shen Te-ch'ien's (1673–1769) Ku-shih yüan in being organized chronologically by dynasty and, within each section, by the social status

[73] Ibid. The last allusion, of course, is to Lun-yü 18/6.

[74] As has been noted by others, the more supple and balanced rethinking of Chang Hui-yen's ideas had already been evident in the critical work of Chou Chi. See Chia-ying Yeh Chao, "The Ch'ang-chou School," pp. 176–83, and Grace Fong, Wu Wen-ying and the Art of Southern Song Ci Poetry , pp. 167–73.

of each author.^[75] (Women come at the end, preceding only a fascicle of "miscellaneous forms" [tsa-t'i ] that includes some lusty "mountain songs" [shan-ko ] that Ch'en evidently could not bear to leave out.) The preface to the collection borrows shamelessly from Chu I-tsun's introduction to his anthology,^[76] although it does admit that its model is the Tz'u-tsung . There is the same recounting of the evolution of tz'u , an argument similar to Wang Sen's against the notion of tz'u as the remnant of shih ("tz'u is that by which one remedies the deficiencies of shih ; it is not the remnant of shih "),^[77] and the declaration that he "considers ya-cheng to be the orthodox tradition."^[78] Song lyrics from the Southern Sung are represented in greatest number within the collection; Ch'en's Tz'u-tan ts'ung-hua in fact announces that he regards the Southern Sung as the "orthodox tradition,"^[79] and there are references throughout to the importance of prosodic rules and to his esteem for the wan-yüeh and marginalization of the hao-fang styles, all of which stems from Li Ch'ing-chao's critical dictum regarding the distinctiveness of tz'u and the theories of Chang Yen and Chu I-tsun. At the same time, however, Ch'en cannot resist confessing to a fondness for the songs of Wen T'ing-yün, whom Chu had not valorized but whose style Ch'en feels many critics of his time have failed to understand, and for the song lyrics of the Northern Sung. The Che school, of course, had given pride of place to the Southern Sung, but Ch'en writes that he personally prefers the loftiness and naturalness of the earlier tz'u , which are, admittedly, more "vulgar" and "impure" than those of the following era. If one thinks of the former as analogous to the "Airs" and the latter as the "Elegances," he argues, then there is room for both.^[80]

Less than two decades later Ch'en T'ing-cho declared that Chang Hui-yen's Tz'u-hsüan was ten times finer than Chu I-tsun's Tz'u-tsung .^[81]

[75] Tsu-pen chiao-chu , 2:855. This system had an earlier precedent within the tz'u anthological tradition, for the Chung-hsiang chi , a collection of song lyrics by women compiled by Ch'ien Yüeh in the last decades of the seventeenth century, arranged its over four hundred works by author in descending order of social status, from wives of high officials to singing girls. As noted by She Chih, "Li-tai tz'u-hsüan-chi," pt. 4, pp. 249–50. The same ordering principle can be found, of course, in certain anthologies of shih .

[76] This is graphically presented in a chart in the recent reprinting; see Tsu-pen chiao-chu , 2:894 n. 10.

[77] Ibid., p. 805.

[78] Ibid., p. 806.

[79] Ibid., p. 846.

[80] Ibid., p. 816.

[81] Ibid., 1:11 (Tz'u-hua ts'ung-pien , vol. 1/2a), and 2:533 (Tz'u-hua tsung-pien , vol. 5/11b). Ch'ü Hsing-kuo discusses the reasons behind Ch'en's change of views—key amongwhich may have been his fondness for the song lyrics of Wen T'ing-yün, whom Chang Hui-yen also esteemed—in the appendix to Pai-yü-chai tz'u-hua tsu-pen chiao-chu , pp. 896–97.

The influence of the Ch'ang-chou school is evident throughout this work in its insistence on treating tz'u and shih in the same critical terms. Both genres, for example, share a list of traits to be avoided:

No matter whether one is composing shih or composing tz'u , one cannot have the style of a rotten Confucian, nor the style of a common person, nor the style of a talented scholar. Though people realize that the rotten Confucian and the common styles are impermissible, they don't realize that the talented scholar's style is also impermissible. . . . As soon as people see the rotten Confucian and the common person's styles, they detest them; but when it comes to the talented scholar's style, there's no one who sees it who doesn't take pleasure in it. Thus, the defect therein is even more profound.^[82]

In addition to linking shih and tz'u here from a negative perspective, Ch'en T'ing-cho also insists on their relationship and comparability for other purposes as well. More than once in the Pai-yü-chai tz'u-hua , therefore, Ch'en remarks on the analogies between their stages of evolution: "If we compare tz'u to shih , the T'ang is like the Han and Wei; the Five Dynasties is like the Western and Eastern Chin and Six Dynasties; the Northern and Southern Sung are like the T'ang; the Yüan and Ming are like the Northern and Southern Sung; and Ch'ing tz'u are like Ch'ing shih ."^[83] As the two can be periodized, so can they be collected in similar manners. The Pai-yü-chai tz'u-hua includes the general preface to Ch'en's Tz'u-tse that describes the selection and organization of its 2,600 songs into four categories: "great elegances" (ta-ya ), "heroic songs" (fang-ko ), "calmed emotions" (hsien-ch'ing ), and "distinctive modes" (pieh-tiao ), with the first group defined as the orthodox and the other three as its subordinates.^[84] Not only are the titles of the groups drawn largely from specific texts in the shih tradition, but the arrangement and the discussion that frames it echo those found in Po Chü-i's famous letter to Yüan Chen concerning his collected poems.^[85]

Running throughout Ch'en's text is the assumption that "shih and tz'u have the same form and different functions"; this in fact, as he con-

[82] Tsu-pen chiao-chu , 2:561; Tz'u-hua ts'ung-pien , vol. 5/18a.

[83] Tsu-pen chiao-chu , 2:576; Tz'u-hua ts'ung-pien , vol. 5/19b.

[84] Tsu-pen chiao-chu , 2:538–39; Tz'u-hua ts'ung-pien , vol. 5/12b–13a.

[85] "Yü Yüan Chiu shu," in Po Chü-i hsüan-chi , ed. Wang Ju-pi (Shanghai: Shang-hai ku-chi ch'u-pan-she, 1980), p. 359.

tinues in this passage, links them more closely than tz'u and ch'ü , which "have different functions, and forms that are also slightly different. This cannot fail to be discriminated."^[86] Ultimately, however, there is a recognizable priority assigned to shih , and on several grounds. The diction of the shih , for example, is inviolable: "Within a shih one cannot write in the language of tz'u , but within a tz'u there is no prohibition against the language of shih ."^[87] The mastery of one must also precede that of the other: "Shih and tz'u have one principle; nevertheless, someone who is not skilled at tz'u can be skilled at shih , but someone who is not skilled at shih decidedly cannot be skilled at tz'u . Therefore, in studying tz'u it is important first to master shih . If in writing shih one has not yet planted one's feet firmly anywhere and yet rushes off to study tz'u —I've never seen anyone manage both."^[88] And finally, there is the intractable fact of temporal precedence: "Shih has its own realm, and tz'u has its own realm: the two share one principle. There are realms that have been opened up by shih poets, however, that tz'u poets have not yet seen, owing to the fact that one came before the other in time." Ch'en goes on to single out various shih poets whose "realms" have in fact been explored by counterparts in the tz'u tradition, but he insists that none has matched the achievements of T'ao Ch'ien and Tu Fu; that possibility, however, may certainly yet be fulfilled.^[89]

Conclusion

The discussions that take place in these anthologies encapsulate the conflicts endemic to discourse on the song lyric up through the Ch'ing dynasty. On the one hand, the tz'u could be located squarely within, indeed at the very heart of, the lyric tradition and legitimated through treatment analogous to that accorded to texts by shih poets and through the appropriation of terms borrowed from the shih tradition. On the other, however, critics were on the whole neither able nor inclined to deny the special roots of the tz'u in a history of performance and song that would not necessarily wish to lay claim to the political and moral seriousness of other written forms. We have seen how persistent was the effort to eradicate, as much as possible, the questionable elements of

[86] Tsu-pen chiao-chu , 2:778; Tz'u-hua ts'ung-pien , vol. 8/11a.

[87] Tsu-pen chiao-chu , 2:573; Tz'u-hua ts'ung-pien , vol. 5/20b. This position runs directly counter to the dominant opinion, mentioned above, against importing the diction and methods of shih into tz'u .

[88] Tsu-pen chiao-chu , 2:673; Tz'u-hua ts'ung-pien , vol. 7/1a.

[89] Tsu-pen chiao-chu , 2:781–82; Tz'u-hua ts'ung-pien , vol. 8/11b–12a.

the history of tz'u and establish—though not without considerable disagreement—a canon within its own tradition. But much as contemporary critiques of dominant traditions have discovered the fineness of the line between difference and marginalization, so the emphasis on the distinctive aesthetics of tz'u ran the risk of consigning it to the privacy—and the isolation—of a room of its own. That the process of legitimation eventually explored more accommodating paths as well may account for what, by the end of the Ch'ing, could be recognized as its success. Early in the twentieth century, May Fourth rewritings of Chinese literary history that discovered and valorized what could be seen as popular genres within the written tradition consolidated this rehabilitation. What should be kept firmly in mind, finally, is the recognition of this entire history as a discursive process, one that does not necessarily reflect the actual dimensions of the interest in and composition of tz'u . And in the end, what may have appealed most powerfully was precisely what resisted appropriation and could never be acknowledged: the song lyric's claim on a private space of the affections within which to speak that which could not be spoken in the shih , whose canons of discourse were inexorably framed by public and political values.