Lord Shang Revisited 4

4. Layers Within SJS

So far we have located both ends of SJS, at earliest, in the Chin Dynasty. What about the middle? Duyvendak tried to sort out that material with the aid of Karlgren's 1913 criteria, which are really meant to distinguish one text or one century from another. Strictly speaking, the only clusters you get with that method are the ones in the "Kgn" column of p1 of the handout. They do not seem to me to be notably suggestive. Having added some further words to the Karlgren list, Duyvendak proposed five layers in SJS.

Rather than patch Karlgren in this way, it would be nice to replace that procedure altogether, with one designed to detect style differences in the same language. I don't claim that such a thing exists. I merely suggest that if it did, and if it were applied to SJS, its output might look something like

Handout p3

On that chart, SJS chapter pairings with a score of 10 or higher are presumptively similar in style, and those with 9 or below are dissimilar. We notice at once that SJS 2 resembles nothing else in the work. If we read that chapter, we find, in confirmation, that it is in a very special, highly repetitive, literary form. It is indeed like nothing else in SJS (though there are parallels in the Han Feidz). In this case at least, any reader may confirm the verdict of the stylistic test: SJS 2 is extraneous to the rest of the work as we have it.

Against that background, against the indication that the stylistic test seems to give meaningful results, we may now look briefly at the Duyvendak layers, in the chart available at

Handout p4.

5. The First Duyvendak Layer

Duyvendak was uncertain as to whether SJS 2 and 3 belonged to this layer. The style chart confirms that he was right to hesitate: to the extent that style is a valid criterion for layer membership, they do not belong. Also not uniformly similar is SJS 13A. This leaves SJS 4, 5, 19, and 20A. Duyvendak's notes indicate numerous links between SJS 4, 5, and 20A; not so many with 19. If we take a further look inside the three chapters which are the most closely linked stylistically, we may observe something of great interest. It is this: SJS 4 begins with a series of points made in SJS 5, taken largely in their SJS 5 order, and continues with a series of points made in SJS 20A, largely in their SJS 20A order. SJS 20B, on whose separation from 20A Duyvendak rightly insists, is not represented in SJS 4 at all. There is material in SJS 4 which is not present in either of the other two chapters, and vice versa.

This pattern of dependence suggests two possible hypotheses.

We could probably decide between these views by collating all the passages in question and determining their directionality, a somewhat formidable undertaking in view of their number. As a shortcut, we could also notice that the matter turns on the status of 20B. If chapter 20 was written all at one time, as a commentary on part of SJS 4, and if 20B is merely the final statement of that commentary, then there is no reason for that final statement to differ in style from the earlier part of the chapter. Similarity of style within SJS 20 would thus tend to support Theory 1. But if 20B was added later, after the SJS 4 digest had been made, it might well differ from 20A. Difference of style thus tends to support Theory 2. Referring to the style chart, handout p3, we find that 20A and B are extremely divergent in style. Separately, on the content level, we may notice that the lists of social parasites in 5 and 20A are consistent, but 4 juxtaposes different lists without reconciling them. This suggests a conflation plus update, with directionality

5, 20 > 4

The style evidence and the content evidence thus agree in supporting Theory 2. As a second result, two independent trials have now suggested that the style test gives meaningful results at least some of the time.

Whichever theory is correct, the key fact is that, in one direction or another, some of SJS chapters seem to be commenting on other SJS chapters. This shows, in whoever was in charge of the text during its late period, a concern for updating the text: for putting the doctrinal house in better order. The same update pattern is overt in the Gwandz, where some late chapters are labeled as commentaries (jye) on early chapters with otherwise identical titles. In Gwandz too, as in SJS 20, the commentaries do not simply expound, they sometimes reinterpret. The tension behind both situations is most likely a difference between a past written record of doctrine and a more current doctrinal need. The commentaries arise in order to keep past doctrine and present need close together. There is passage of time, there is change of doctrine, and finally, there is a wish to minimize the gap between old and new doctrine.

6. The Second, Third, and Fourth Duyvendak Layers

It would be tempting to go on to analyze the parts of SJS 5 and 20A that were not brought up into SJS 4 and, conversely, what of its own SJS 4 added to this previous material. The result of that analysis should show in great detail what the group's doctrines were, at one moment in their modification. For lack of time, I push on instead to the second Duyvendak layer, still on handout p4. The style test confirms that SJS 7 and 8 are compatible, so there is nothing against their forming a layer, as Duyvendak suggests. It is noted on the handout that, on mere stylistic grounds, SJS 6 could quite plausibly be added to that layer.

Moving on, we find a complementary situation: SJS 6 seems to be extraneous in Duyvendak's third layer. Of the remaining chapters, it is notable that nearly all are from the latter half of the book, and it will be seen from Handout p1 and

Handout p3

that all high-numbered chapters, whether or not they are structured as memorials with the speaker designated as chvn, tend to resemble each other more or less. The idea that the chvn or "Your Servant" persona defines a consistent layer thus does not quite hold. The use of a common period style may provide a better explanation for these resemblances.

Coming to the fourth Duyvendak layer, still on handout p4, we have already dealt with the fact that SJS 1 and 26 are both late, and, separately, with the fact that higher-numbered SJS chapters tend to resemble each other. Since SJS 26 is late, and since the chapters numbered in the 20's tend to resemble each other, we may suspect that all the 20's chapters are generally late (save for 20A, which as we have seen must predate SJS 4). A later page of the handout will context that suspicion.

 

All lectures and abstracts posted on this site are Copyright © by their authors.

Go to Part #5 of Lord Shang

14 Oct 2003 / Contact The Project / Exit to Lectures Page