Todd R. Hanneken, St. Mary’s University
Coherence-Based Genealogical Method (CBGM) utilizes the power of digital technology to assist scholars reconstructing relationships between textual witnesses. The method originated at the Institut für Neutestamentliche Textforschung (INTF) in Münster and has been used to revise the Editio Critica Maior (ECM) of the New Testament. The reliance on quantitative analysis of large sets of data partly addresses the difficulty in textual criticism of polygenetic (coincidental) agreement. The method does not replace the judgment of a scholar on the genealogical relationship between any two variant readings. The method does, however, aggregate a large number of scholarly judgments and textual agreements and disagreements to render a visualization of the relationships between textual witnesses. The scholarly judgments can then be revised, and the data processed again, in pursuit of a persuasive model of textual history.
The method considers two kinds of coherence (some count stemmatic coherence as a third kind), namely pregenealogical coherence and coherence (without a qualifier, implying genealogical coherence). The first, pregenealogical coherence, is a quantitative analysis of agreements and disagreements in machine-readable texts representing manuscript witnesses. Any one agreement may be the result of independent scribes coincidentally producing the same word, yet one may assume that such coincidences do not recur often in a manuscript. Overall agreement across large spans of text, easily calculated by machines, may be more reliable than scholarly intuition focused on a smaller set of data. The second kind of coherence, without the qualifier “pre,” attempts to not just count but explain disagreements as one prior to the other. For any one disagreement, the computer makes no judgment. A scholar must judge which variant is more likely an ancestor of the other, or in many cases simply “uncertain.” The utility of the computational method is in identifying patterns across large sets of judgments. Even if many relationships are uncertain, a majority of individual judgments may lead to a conclusion that one witness is more potentially the textual ancestor of the other. This tentative conclusion becomes a data point in the next stage of constructing genealogical stemmata on increasingly broader scales (which could be described as stemmatic coherence).
A single textual variant is considered a local stemma. A local stemma does not imply a relationship between witnesses but serves as a point of data along with many others as the number of witnesses and local stemmata increase. With three or more witnesses, the possible combinations of potential ancestors become increasingly complex. If one can imagine an ideal scenario in which there are no coincidental agreements, no cross-contamination readings from one branch influencing another, and no missing witnesses, then the construction of a global stemma would be trivial, not even requiring a computer. If a group of witnesses agree on one variant, we would call those witnesses a branch, and within that branch a more limited agreement would be called a sub-branch. Because our evidence does not follow the ideal scenario, any one point of data can be misleading. Coherence-Based Genealogical Method relies on many points of data to identify patterns emerging from the noise of flawed evidence. A scholar can hypothesize a global stemma with one witness as ancestrally prior to others and test how smoothly the local stemmata flow.
The method is controversial, limited, and open to improvement in several ways. One critique common to computational methodologies is the “black box” critique. There can be a perception that a computer is making more decisions than it is and dictating changes to the critical edition of the New Testament. It would be more precise to say that scholars are making changes using CBGM the way an accountant might use a calculator. The method is a tool that helps scholars quickly visualize the aggregate implications of their judgments. Perhaps more importantly, it helps scholars consider alternative possibilities without having to redo a tremendous amount of tedious work. Another set of questions common to computational methodologies pertain to the creation of digital source data for the method to process. In this case, for example, whether a digital version of a manuscript should normalize orthographic variants and other scribal practices. Also, the method addresses but does not eliminate the endemic problems of textual criticism, such as polygenetic (coincidental) agreements and cross-contamination. The method currently works with extant witnesses only, with no hypothetical lost nodes in the stemmata (hyparchetypes). It would be possible to generate unattested hypothetical witnesses, but to do so could be considered a hybrid or confusion the distinction between texts and manuscripts. That is, the method evaluates the relationships between texts, whereas manuscripts (attested, lost, or reconstructed) witness a state of a text. The method evaluates the coherence of human hypotheses about flow from ancestor to descendant texts, it does not reconstruct a hypothetical text or witness. It may also be noted that Occam’s razor, the pursuit of the simplest possible model that best explains the evidence, is not the same as the more complex actual textual history. The method may support the text-critical goal of determining the most prior, most ancestral Ausgangstext, but at the same time dismiss as “noise” the complex activities of individual human scribes.
The software and its implementation for the New Testament are freely available. The New Testament Virtual Manuscript Reading Room (NTVMR; ntvmr.uni-muenster.de), associated with the Institut für Neutestamentliche Textforschung (INTF) in Münster, allows users to attempt their own critical editions. Users can also see the images and models used to support the Editio Critica Maior, along with a number of philological tools. Although the most significant implementation of CBGM is applied to the New Testament, the method could apply to other projects with large numbers of witnesses with complex relationships. The Virtual Manuscript Room Collaborative Research Environment (VMR CRE; vmrcre.org) is open source. It is being used by other projects, such as the Coptic-Sahidic Old Testament Project (coptot.manuscriptroom.com). Joey McCollum’s Open CBGM (github.com/jjmccollum/open-cbgm) can also be freely downloaded and implemented on new projects.
Stephen C. Carlson, “A Bias at the Heart of the Coherence-Based Genealogical Method (CBGM),” Journal of Biblical Literature 139 (2020): 319.
Peter J. Gurry, A Critical Examination of the Coherence-Based Genealogical Method in New Testament Textual Criticism (New Testament Tools, Studies, and Documents 55; Leiden: Brill, 2017).
Gerd Mink, “Eine umfassende Genealogie der neutestamentlichen Überlieferung,” New Testament Studies 39 (1993): 481-99.
Gregory S. Paulson, “Improving the CBGM: Recent Interactions,” in The New Testament in Antiquity and Byzantium (ed. H.A.G. Houghton et al.; Berlin: De Gruyter, 2019) 295-308.
W. Andrew Smith, “Examining A Critical Examination of the CBGM: A Review Article,” TC: A Journal of Biblical Textual Criticism 24 (2019): 1-13.
Tommy Wasserman and Peter J. Gurry, A New Approach to Textual Criticism: An Introduction to the Coherence-Based Genealogical Method (Atlanta: Society of Biblical Literature, 2017).
Todd R. Hanneken, “Coherence-Based Genealogical Method (CBGM) for Brill, Textual History of the Bible.” San Antonio, Texas: St. Mary’s University, 2021.