Text and edition in early Chinese philosophical literature.

• Author: Roth, Harold D.

INTRODUCTION

During the years I have been working on the textual histories of a number of important sources of early Taoist philosophy such as the Huai-nan Tzu and the Chuang Tzu, I have had to confront some conceptual problems in the analysis of my data and have had to adapt and develop working definitions for a number of important technical terms for which there does not seem to be a consensus among scholars working in the field. In my opinion, the two most important of these problems are: (1) how are we to understand the difference between the terms "text" and "edition"; and (2) once we decide what we mean by "text" and "edition," how are we to identify and discuss the various strata of editions that arise during the transmission of a text, and how do they relate to the idea of a "received text" (textus receptus)? I regard the resolution of these problems to be of the utmost importance for the fields of textual history and textual criticism.

In this paper I will present the working definitions I have developed for the above terms and set them in the context of how I view textual history and its relationship to textual criticism. In the course of presenting the ways in which I have dealt with these problems, I will set forth some old and some new definitions for a number of other terms I view as central to the field of textual history, including "recension," "redaction" and "ancestral redaction," "direct and indirect testimony," and will also touch upon a methodology I've developed for determining the lineages of the editions of a text, which I call "filiation analysis."

TEXT AND EDITION

I would like to begin with the observation that quite often scholars in our field fail to distinguish clearly between text and edition. More specifically, the failure often comes in identifying a particular edition of a text as a text. For example, is the unique version of the Huai-nan Tzu [UNKNOWN TEXT OMITTED] created by the Ch'ing scholar Chuang K'uei-chi [UNKNOWN TEXT OMITTED] a text or an edition? Is the 33-chapter version of the Chuang Tzu established by Kuo Hsiang [UNKNOWN TEXT OMITTED] a text or an edition (or something else again, as I will suggest below)? And are the Mawang-tui [UNKNOWN TEXT OMITTED] manuscripts of the Lao Tzu to be thought of as texts or editions?

That Sinologists should often be unclear in their use of these terms will come as no surprise when we recognize that in many cases the paradigm we use for our own studies, Western textual scholarship in New Testament, Greek and Latin, and English literature, fails to enunciate clearly the distinction between these two terms. I can only speculate on the reasons for this, but my admittedly cursory survey of the relevant literature leads me to believe that it may have something to do with the predominance and remarkable survival of manuscript traditions in the transmission of the major texts of the ancient Western world.(2) There is a tendency to refer to these manuscripts and their copies as "texts" and to their much later printed versions as "editions"(3) perhaps because so many of the former are copies and hence the product of scribes that involved little of the sifting, sorting, and emending work we commonly associate with critical editing. In China, given the relatively early invention of printing, compared to the West, the great majority of extant editions are printed, and so from the Western practices, should not be referred to as "texts." However, the manuscript copies of texts that circulated prior to this are no less "editions" than the later printed versions that now survive, and, in my opinion, in most cases should also not be referred to as "texts" in light of the definition I will propose below. In any case, given the predominant influence of Classical and Biblical textual studies in the field, perhaps their tacit assumptions about "texts" and "editions," drawn from the specific conditions of their source materials, have remained largely, but not entirely, unexamined. Not entirely, I say, because the basis for a definitive distinction between text and edition has been clearly drawn in the scholarship of Vinton Dearing,(4) whose text-critical methodology was first applied to Chinese texts by Paul Thompson in his reconstruction of the Shen Tzu fragments.(5)

For Dearing, who relies on "information theory" for his basic definitions, the "text" is a unique complex and expression of ideas created by an author or authors.(6) According to him, in information theory this is called the "message." It is the stated goal of textual criticism to locate-in the unlikely event it still survives intact-or to re-establish, if it does not, this original text. This is indeed the commonly accepted goal of textual criticism.

Dearing further maintains that the "states" of a text are the various "forms" of the original "message" as determined in the process of "textual analysis," Dearing's name for his particular methodology for sorting out the various states of a text and their unique textual variants and determining the state from which they have descended in preparation for the final emendations that are the last stage in the critical process. The actual physical objects in which the forms of the message or states of the text are embodied are called the "records" of the text. In information theory these are called the "transmitters.(7)" In my own analysis, in the Chinese context at least, these "records" constitute the "editions" of a text.

A recent and very perceptive article by G. Thomas Tanselle places Dearing's work in the context of the entire Western tradition of textual criticism, and it would be instructive to take a moment to examine this.(8) Tanselle talks about two general approaches to preparing a new edition, one that has been favored by scholars who work with ancient and medieval texts, and one that has been favored by scholars who work with modern texts. In the former, due to the antiquity of the texts, where there is little hope of recovering an author's original manuscript, scholarly efforts have been devoted to establishing a critical edition that is the best approximation of the authorial original by careful analysis of the extant testimony to that text. In the latter, due to the existence of various editions or autograph manuscripts of a text that are often produced during an author's lifetime, scholarly efforts have turned more to deciding upon an authoritative "copytext" to use as the basis for the new edition.(9)

Dearing falls squarely in the former tradition, which he himself claims began with the development of the "genealogical method" by the New Testament scholar, Karl Lachmann (1793-1851).(10) He further maintains that his work has gone beyond Lachmann's reliance on common textual errors (that is, outstanding textual variants) by incorporating the formal logic and statistical analysis pioneered by Sir Walter Greg and Dom Henri Quentin.(11) According to Tanselle, Dearing's work still falls within "classical textual criticism," which involves two fundamental stages, recensio and emendatio.(12) In the former, after surveying the extant sources of textual testimony and choosing only the most important, one next analyzes their patterns of textual variants and constructs a genealogical tree, technically called a stemma codicum, which one then uses to decide among the variants and to construct the textual archetype of this testimony. In the second stage, wherever the stemma fails to decide among the variants, one must make the decision oneself, on the basis of what the famous textual scholar, Paul Maas, calls "conjecture" (divinatio).(13) Dearing's methodology of "textual analysis" is a procedure of the recensio stage. However, he differs from such classical statements of the methodology of this stage, as found in Maas, in attempting to clarify the formal logic that underlies it, incorporating statistical analysis of textual variants, and in making a new distinction that he regards as extremely important.

In Dearing's definition, "textual analysis" determines the genealogy of the variant "states" of the text; "bibliography" determines the genealogy of the "records" of the text, and it is absolutely imperative not to confuse the two. (14) He regards this as one of the central achievements of his methodology, freeing "textual analysis" from "bibliographical thinking," which must consider the dates and circumstances of creation of particular records. This ultimately leads him to be able essentially to do away with the problem of conflation, which bedeviled earlier "genealogical" or "stemmatic"" approaches such as that of Maas. When two states of a text are freed from their bibliographical records, one needn't worry that a third state is a conflation of them; it is simply a third state and may be either a combination of the two or their common ancestor. However, as Tanselle points out, Dearing's mechanical solution (the formation and breaking of textual rings at their weakest statistical link, uninfluenced by dating) does not really do away with the problem of conflation: it simply addresses it from another point of view.(15) Furthermore, I agree with Tanselle in objecting to the exclusion of "bibliographical" evidence. Such evidence is often extremely valuable in establishing the filiation of editions necessary for the construction of the stemma and can be extremely valuable in delimiting the textual testimony that one must consider in order to establish a critical edition. Also, it often provides valuable information as to how certain states of a text arose. As Tanselle rightly says, "physical details sometimes explain textual variants," and that without knowing how the variants in a particular record of a text arose, " . . . one may postulate relationships that are shown by physical evidence to be incorrect."(6)

Let me give an example from my work on the Huainan Tzu. In my text-historical analysis of the Chung-li ssu-tzu chi...