The Newsletter 88 Spring 2021

What kind of history can we write of the Song dynasty?

Christian de Pee

What kind of history can we write of the Song dynasty (960-1279 CE)? The question seems naïve and limiting. Didn’t the spread of printing during the Song preserve more texts from that period than survive from earlier dynasties, and don’t these sources allow us to write almost any history we would like? The problem is that these printed texts are nearly all that remains, and that they are formal, generic, and public. They do not lend themselves to the kinds of narratives that historians of medieval Europe have written based on unique, individual manuscripts in monastic libraries and municipal archives.

A lack of transparency

Texts are never transparent. One sometimes reads that an historical source provides ‘a window onto the past’ or that it ‘offers a glimpse’ of a former person or custom, and many a research proposal today proposes to use a certain archival collection ‘as a lens’ for examining a chosen subject matter. Such visual metaphors are misleading, because they imply the possibility of unmediated access to the past. They elide, however briefly, that which exists—a physical text—in favor of something that has vanished—an event in the past—and thereby suggest that the historical value of a text lies behind it rather than in front of it, that the referent of the text is more substantial, and more historical, than the text itself. 1 Cf. Ricoeur, P. 1981 (edited and translated by John B. Thompson) Hermeneutics and the Human Sciences: Essays on Language, Action, and Interpretation. Cambridge University Press and Éditions de la Maison des Sciences de l’Homme, p.93.  If one wishes to use a visual metaphor, one might better compare a text to a painting. Like painters, writers deliberately choose and compose their scenes, compress time and foreshorten space to fit these scenes within a unifying perspective, and represent them in conformity with the conventions of their period and their genre. A painting is not a window; neither is a text.

The texts that survive from the Song dynasty are particularly resistant to a transparent reading. First, almost all surviving texts from the Song are written in classical Chinese, a written language that perpetuated the grammar and the idiom of the fourth century BCE and that was distant from the language spoken during the Song. To write in classical Chinese during the Song involved not only the transformation of vernacular action into formal writing, but also the translation of colloquial speech into classical diction, embellished with literary allusions. One might compare the effect to that of a medieval European text that describes a contemporary market in the high Latin of Horace or Cicero.

Second, almost all surviving texts from the Song are written in set genres and conform to the conventions of form, style, content, and ideology that defined those genres. When learned men and women wished to set down their thoughts, they selected a genre that suited the subject and the mood of those thoughts. When required by an occasion to write in a particular genre, they chose a subject, images, and sentiments appropriate to that genre. They created original effects and an individual style, not by violating conventions but by obeying them, demonstrating their ability to express themselves coherently and elegantly within the strictures of form and style, with subtle variations on the phrases of earlier writers. Writing in classical Chinese, in other words, required the overt display of literary skill and literary knowledge. The mode of representation was to a large extent also the object of representation.

Third, almost all surviving texts from the Song were preserved for ideological reasons, as examples of good style and proper morals. From the trunk of a deceased man of letters, his descendants or his friends selected the compositions they deemed worthy of him and therefore worthy of being imitated by others. (Writings by women were very rarely published during the Song.) Subsequent generations made their own selections, reprinting authors whose learning and virtue retained their reputation or aroused new enthusiasm, while works that had fallen into disfavor became first rare, then obsolete. The writings of those who supported the controversial economic reforms of Wang Anshi (1021-1086), for example, have nearly all been lost, to the extent that much of the ideology of the reforms must be reconstructed from criticism by their opponents, whose writings later generations admired and preserved.

The linguistic anthropologist William F. Hanks has observed that, “historical texts illustrate discourse under minimal conditions, because of the vast amount that can never be known of the context, and this makes it all the more necessary to be explicit about how we read”. 2 Hanks, W.F. 2000. Intertexts: Writings on Language, Utterance, and Context. Rowman and Littlefield, p.11.  Historical research must begin with the acknowledgment that the textual record is incomplete and that the particular incompleteness of the textual record is itself an historical artifact, the product of historical choices. The acknowledgment of this incompleteness is not an obstacle to historical knowledge; it is its precondition.

An intellectual history of social life

I began thinking about such historiographical problems as a Ph.D. student in the early 1990s. I noticed that the Song Empire described in the scholarly books and articles I read looked different from the Song Empire that I imagined when I read primary sources from the period. The prose style and the mode of reasoning in the block quotations and paraphrases of historical monographs often did not sound like the style of the prose and reasoning I thought I heard in texts from the Song dynasty. It seemed to me that the translations and paraphrases inflected the language and manner of the sources in order to fit them for analysis by social-scientific methods. I did not think that the translations were wrong or that the arguments were inaccurate, but I felt that the translations and the arguments diminished the distinctiveness of the sources and that a subtle variety of cultural knowledge was lost in the process. Instead of conforming the language and reasoning of the sources to existing analytical categories, I wondered whether it might not be more interesting to use the language and reasoning of the sources to generate new analytical categories.

Text as practice. The editors of the thirteenth-century Ceremonies and Rites Illustrated (Yili tu) reconstructed the choreographies of ancient rituals and represented these choreographies in maps, as in this map of a betrothal ceremony.


In my dissertation I attempted this latter approach. Having originally intended an anthropological analysis of wedding ritual during the Song and Yuan (1272-1368) dynasties, I found that the wedding ceremonies recorded in the sources did not add up to a coherent sequence and that they had to be understood in the first place as scripts for the performance of the ideology of the genres in which they were written. Instead of reconstructing an average wedding sequence and subjecting that sequence to a structuralist analysis, I placed wedding ceremonies within their generic context, showing that the ceremonies inscribed in each genre—ritual manuals, writing manuals, mantic texts, and legal texts—instantiated the ideological notions of time, space, and body of that genre:

"the symmetrical, centered, porous time and space of exegetical discourse, where the groom and the bride merge with sacred Antiquity; the linear time and space of literary discourse, where the written bodies of the groom and the bride proceed through an anterior, metaphorical time and space; the cyclical time and space of cosmological discourse, in which calculations and diagrams chart a safe path for the liminal groom and bride through liminal time and space; and the imperial time and space of legal discourse, in which codes and verdicts carefully inscribe the groom and the bride into a transparent hierarchy of imperial subjects." 3 de Pee, C. 2007. The Writing of Weddings in Middle-Period China: Text and Ritual Practice in the Eighth through Fourteenth Centuries. State University of New York Press, p.13.

This post-structuralist approach to text and ritual practice revealed, among other things, that the authors of the ritual manuals had themselves a structuralist understanding of ritual.

Since in these texts the inscription of the wedding ceremony constituted the primary historical act—its eventual performance in reading or in ritual embodiment being unknown—I argued that we know most about the historical practice of Song and Yuan weddings when writing itself was a ritual act (e.g., the calculation of a horoscope) and when the text itself was a ritual object (e.g., the letters exchanged during betrothals). Critics of my work have objected that these arguments deny the existence of an historical reality beyond the text, but in fact the arguments extend historical reality to include the text, and demonstrate that any knowledge of the world behind the text depends on our understanding of the world in front of it. This enlarges the potential evidence and the potential subject matter of history instead of diminishing them. The analysis of letters of betrothal illustrates this most vividly. Although these letters present direct evidence of the practice of Song and Yuan weddings, previous historians of weddings and marriage had omitted them from their monographs, presumably because they are literary exercises that do not provide specific material details. Read as performances in their own right, however, these letters can be placed among the written and material exchanges by which two families confirmed to one another that they had firm possession of the learning and the wealth that they had advertised to each other through the matchmaker.

The performativity of the act of writing that dominates the surviving texts from the Song dynasty may prohibit the kind of social history from the bottom up that historians of medieval and early modern Europe have written, but it does allow a lively intellectual history of social life, and a detailed knowledge of historical ways of seeing and thinking among the literate elite. As a Fellow at the International Institute for Asian Studies during the 2019-2020 academic year, I began writing a general history of eleventh-century China based on these historiographical insights. I hope to show that the history of the Song dynasty can be made accessible and interesting to a general reading public when it is told as intellectual history.


Christian de Pee, University of Michigan; former IIAS Fellow,