New
Testament Stemmatics New
Testament Stemmatics[1] The New Testament has been preserved in more manuscripts than any other work composed in Western antiquity. These manuscripts have been hand-copied over the course of two millennia, and by this human process they differ from one another. New Testament books have also been translated into several foreign-language versions, the most important of which are Latin, Syriac, and Coptic, and the N.T. text has been quoted by theologians throughout the centuries. (See generally, Aland & Aland [1981] 1989, Metzger [1964] 1992).
[2] The task of the N.T. textual critic, therefore, is to sort through the various and varying witnesses to the text of the N.T. and establish a "critical text" that best explains the state of all extant witness (Robinson, P. 2000), not only in "external" terms of the age, provenance, and affiliation of each witness, but also in light of "internal" transmissional and authorial considerations. This critical text is assumed to be one that is as close as humanly possible to that of the originals.
[3] There have been five basic approaches to the textual criticism of the N.T., which can be placed into a spectrum based on the relative weight given to internal considerations and external evidence. Those approaches that favor external evidence are termed "documentary" because they are based on the identity of the documents that transmit the text, and those approaches that tend to prefer internal criteria are termed "eclectic" because the final text is selected, not so much from an identified document, but reading by reading.
- [4] Thoroughgoing or Rigorous Eclecticism aims to apply strictly the canons of internal evidence, paying particular attention to the author's style (e.g. Elliott 1995). No critical text of the N.T. has been thoroughly edited on the premises of rigorous eclecticism. Ultimately, however, some appeal to external considerations must be invoked since the internal criteria are often indecisive.
- [5] Reasoned Eclecticism, also called rational eclecticism, is the most commonly employed approach in modern N.T. text criticism and has been basis of the United Bible Societies' and Nestle-Aland texts. This approach involves the careful weighing of a variety of external and internal criteria at each variation unit to arrive at the critical text. Because of the indecisiveness of the internal criteria at many variation units, reasoned eclecticism in these cases devolves into selecting the readings of the best "text-type", of which three are known (Alexandrian, Western, and Byzantine) and of which the Alexandrian is usually preferred. Different notions of the history of the text, however, may cause the reasoned eclectic to champion a different text-type with significantly different results. (See Holmes 1995)
- [6] Stemmatics is a historical-documentary approach that organizes the witnesses into a pedigree or "stemma" that represents the historical development of the text, typically by recognition of common errors. (Maas 1958) The archetypal text is then produced by selecting the readings from the closest variant carriers based on internal evidence. The critical may also examine the resulting archetypal text for primitive errors and emend the text if necessary. Although stemmatics is the dominant approach in classical philology, where there are few surviving witnesses for each text, the vast number of witnesses to the N.T. and the presence of pervasive contamination have prevented its successful application to the N.T.
- [7] Majority text (e.g. Hodges & Farstad 1985). This documentary approach simply determines the text by selecting the most attested variant among the extant witnesses. Internal evidences plays no role unless the manuscripts are deeply split. Although the majority text will never reproduce the errors found in only a small number of manuscripts, it will reproduce the errors of a prolific family of documents that happens to overwhelm its competitors in numbers, often for historical reasons unrelated to the quality of the text.
- [8] The copy text is another documentary approach, which follows the readings of a single witness, except when its readings can be shown to be clearly in error. The copy text can be chosen critically (e.g. on internal grounds or a witness closest to the author, as in Shakespearean textual criticism) or uncritically (i.e. for reasons unrelated to the quality of the text, such as out of tradition or from among the most conveniently available manuscripts as in Erasmus's first edition of the N.T., which came to be known as the Textus Receptus).
[9] Reasoned eclecticism has dominated 20th century N.T. textual criticism and has produced a very good text of the N.T., yet one cannot escape the feeling of stagnation. Even in the face of spectacular manuscript discoveries in Egypt of the last century, there has been a discernable lack of progress in the theory and history of the N.T. text, prompting a leading critic to dub this period the "Twentieth-Century Interlude" (Epp [1974] 1993). After thirty years, the field is still mired in the Interlude.
Out of the Twentieth-Century Interlude
[10] It is my opinion that rational eclecticism has taken us as far as we can possibly get without an explicit history of the text, and that recent advances in computer technology, especially among computational biologists, have finally made it possible to apply stemmatics on the scale of a manuscript tradition as complex as the N.T. to produce that history of the text. If so, then we may finally have found a way out of the Interlude.
[11] As early as 1977, researchers have noticed deep analogies between biological systematics and textual stemmatics (Platnick & Cameron 1977 fide Cameron 1987). In both fields, kinship is recognized by the presence of shared innovation, i.e. synapomorphy and community of error, respectively. In biology, this principle has produced a method called "phylogenetic systematics" (Hennig 1966) or, more popularly, "cladistics" to infer the family tree of living organisms. Accordingly, "phylogenetic stemmatics" is the application of cladistic analysis in the preparation of manuscript stemmata.
[12] By 1987, cladistics had been applied to manuscripts traditions (Lee 1989), but four years later, cladistics managed to pass a significant test. Peter Robinson, who had laboriously constructed a stemma of the Old Norse narrative Svipdagsmal based, in part, on direct external evidence, posted an on-line challenge in 1991 to "prove Housman wrong" by reconstructing his stemma solely by mathematical techniques. One of the entrants was a biologist, Robert J. O'Hara, who ran the data through a cladistics computer program and successfully found all of Robinson's textual groups (Robinson & O'Hara 1992). Cladistics was later applied by Robinson in the editing of the Canterbury Tales (O'Hara & Robinson 1993).
Cladistics Primer
[13] A more detailed explanation of how cladistics works can be found elsewhere (see generally Kitching et al. 1998, and O'Hara & Robinson 1993 for textual applications), but the following should present at least a flavor of the method.
[14] Cladistics proceeds by examining each point of variation to find the "optimum tree" with the fewest changes. This can be illustrated with a New Testament example in the Gospel of Mark using the following witnesses: A B D E L. At Mark 1:2, there is a variation unit in which an Old Testament quotation is introduced by either "in the prophet Isaiah" (en [tw] Hsaia tw profhth) or "in the prophets" (en toij profhtaij), in which B D L read "Isaiah", and A E read "prophets." Although cladistics does not require prior knowledge of the original reading, this example assumes that the UBS reading ("Isaiah") is original to simplify this discussion. This variation unit results in the following tree:
Isaiah | ------------- | | | | | | | Isaiah->prophets | | | | | | | --- | | | | | B D L A E[15] In this tree, the archetype reads "Isaiah" and only one change is needed to explain the A-E reading of "prophets", by supposing that A and E descend from a common archetype whose scribe changed "Isaiah" to "prophets." In stemmatic terms, A and E's agreement in error (i.e. "prophets" instead of "Isaiah") implies a common origin. However, the following tree requires two changes (with "Isaiah" represented by "I" and "prophets" by "P"):
I | ----------- | | | | I>P | | ---- | | | | | | | | | I>P | | | | | A D L B E[16] Since the first tree has a lower cost in terms of the number of changes, it is preferred as a more parsimonious explanation of the textual evidence than the second tree.
[17] This process is repeated over every variation unit to find the most parsimonious tree. In the example, the next variant is at Mark 1:4, "John the baptizer" (Iwannhj o baptizwn) vs. "John was baptizing" [W] (Iwannhj baptizwn). Here, B and L read "the baptizer" (code: B), but A D E have "was baptizing" (code:W). If this change is fit into the first tree assuming that the UBS reading of "the baptizer" (B) is original, we would have the following tree:
I+B | ------- | | | | | B>W | | | | | ---- | | | | | | | I>P | | | | | | | --- | | | | | B L D A E[18] This stemma requires only two changes. There are actually thirty-five possible trees for five witnesses capable of explaining the above data, but others are sub-optimal. For example, the following tree, in which the position of D and L are swapped, results in a tree that requires three changes:
I+B | --------- | | | | | ---- | | | | | | | I>P | | | | | B>W | B>W | | | | | | | --- | | | | | B D L A E[19] If this second tree were the real tree, then there would have been two independent changes for the Mark 1:4 variation unit, one on the lineage to D and another one on the lineage to the most recent common archetype of A and E. Because the second tree requires three changes and the first tree requires only two changes, cladistics prefers the first tree over the second tree. The first tree also corresponds to our sense of what is actually happening, in that B and L are Alexandrian, A and E are Byzantine, and D is "Western."
Unsolved Problems with Cladistics
[20] Unfortunately, cladistics is not a completely ready-made solution to stemmatics. In fact, there are two major problems that require further research before stemmatics can fully come of age in N.T. textual criticism: contamination and bifurcation. (See generally Robinson, P, 1996).
- [21] Contamination. Cladistics assumes that there is only one line of descent and that horizontal transfer between different lineages does not occur. However, contamination, in which readings from multiple exemplars are mixed, is a common occurrence in manuscript traditions and any stemmatic method must be able to handle it. This is especially true in the N.T., where, for example, Westcott & Hort thought the Byzantine text-type to be a mixture of the Western and Alexandrian, and where most of the medieval "Caesareans" are suspected to be Byzantine mixed.
- [22] Bifurcation. Most modern cladistic analyses produce strictly bifurcating trees, in which each ancestor has exactly two descendents. This assumption is somewhat artificial, but techniques, such as consensus trees and bootstrapping, have been developed to produce polychotomies (multiple direct descendents). The problem is more on the stemmatics end, because stemmatics treats such multiple lineages by a simple majority vote (taken to an extreme, this becomes the majority text approach), but this approach may not be fully appropriate when the polychotomy is at the level of the archetype.
[23] Much of my research in phylogenetic systematics has been focused on solving these two problems, and I am in the process of preparing my ideas for publication.
Projects
- A Glossary of Phylogenetic Stemmatics
- A Phylogeny of Galatians: prolegomena, stemma, analysis.
References
- Aland & Aland [1981] 1989
- Kurt Aland and Barbara Aland, The Text of the New Testament: An Introduction to the Critical Editions and to the Theory and Practice of Modern Textual Criticism (2d. ed.; English trans., Erroll F. Rhodes from 2d German ed. 1987; Grand Rapids, Mich.: Eerdmans, 1989).
- Cameron 1987
- H. Don Cameron, "Problems in Manuscript Affiliation," in Henry M. Hoenigswald & Linda F. Wiener, eds., Biological Metaphor and Cladistic Classification: An Interdisciplinary Perspective (Philadephia: U. Penn., 1987): 227-242.
- Elliott 1995
- J. Keith Elliott, "Thoroughgoing Eclecticism in New Testament Textual Criticism," in Bart D. Ehrman & Michael W. Holmes, eds., The Text of the New Testament in Contemporary Research: Essays on the Status Quaestionis (Studies & Documents 46; Grand Rapids, Mich.: Eerdmans, 1995): 321-335.
- Epp [1974] 1993
- Eldon Jay Epp, "The Twentieth-Century Interlude in New Testament Textual Criticism" in Studies in the Theory and Method of New Testament Textual Criticism (Studies & Documents 45; Grand Rapids, Mich.: Eerdmans, 1993): 83-108, orig. deliv. as the W.H.P. Hatch Memorial Lecture (Chicago, Ill.: SBL, 11 Nov. 1973) and pub. in JBL 93 (1974): 386-414.
- Hennig 1966
- Willi Hennig, Phylogenetic Systematics (English trans. and extensively rev., D. Dwight Davis & Rainer Zangerl; Urbana: U. Ill. Press, 1966)
- Hodges & Farstad 1985
- Zane C. Hodges & Arthur L. Farstad, The Greek New Testament According to the Majority Text (2d ed.; Nashville, Tenn.: Nelson, 1985).
- Holmes 1995
- Michael W. Holmes, "Reasoned Eclecticism in New Testament Textual Criticism," in Bart D. Ehrman & Michael W. Holmes, eds., The Text of the New Testament in Contemporary Research: Essays on the Status Quaestionis (Studies & Documents 46; Grand Rapids, Mich.: Eerdmans, 1995): 336-360.
- Kitching et al. 1998
- Ian J. Kitching, Peter L. Forey, Christopher Humphries, & David M. Williams, Cladistics: The Theory and Practice of Parsimony Analysis (Systematics Assoc. Pub. 11; Oxford: Oxford, 1998).
- Lee 1989
- Arthur R. Lee, III, "Numerical Taxonomy Revisited: John Griffith, Cladistic Analysis and St. Augustine's Quaestiones in Heptateuchum," Studia Patristica 20 (1989): 24-32, orig. deliv. (Oxford: 10th Int'l Conf. of Patristic Studies, 1987).
- Maas 1958
- Paul Mass, Textual Criticism (English trans. Barbara Flowers from 3d German ed., 1957; Oxford: Oxford, 1958)
- Metzger [1964] 1992
- Bruce M. Metzger, The Text of the New Testament: Its Transmission, Corruptions, and Restoration (3d ed.; New York: Oxford, 1992).
- O'Hara & Robinson 1993
- Robert J. O'Hara & Peter M.W. Robinson, "Computer-Assisted Methods of Stemmatic Analysis," Occasional Papers of the Canterbury Tales Project 1 (1993): 53-74. Online: Computer-assisted methods of stemmatic analysis.
- Platnick & Cameron 1977
- Nelson I. Platnick & H. Don Cameron, "Cladistic Methods in Textual, Linguistic, and Phylogenetic Analysis," Sys. Zool. 26 (1977): 380-385.
- Robinson, P. 1996
- Peter M. W. Robinson, "Computer-Assisted Stemmatic Analysis and 'Best-Text' Historical Editing," in Pieter van Reenen & Margot van Mulken, eds., Studies in Stemmatology (Amsterdam: Benjamins, 1996).
- Robinson, P. 2000
- Peter Robinson, "The One Text and the Many Texts," Literary and Linguistic Computing 15 (2000): 5-14.
- Robinson & O'Hara 1992
- Peter M.W. Robinson & Robert J. O'Hara, "Report on the Textual Criticism Challenge 1991," Bryn Mawr Classical Review 3 (1992): 331-337. Online: Report on the Textual Criticism Challenge 1991.