In this chapter we deal with the second step in weighing internal evidence — evaluating intrinsic probabilities. The goal of assessing intrinsic probabilities is to discover which variant the author is most likely to have written based upon what we know about his style. The major focus at this level of analysis is upon the author rather than the scribe. Of course, all internal considerations have this focus as their aim in the sense that all variations are variations from the original. But intrinsic probabilities focus specifically on the positive question of what an author is most likely to have written, whereas transcriptional probabilities focus on the more negative question of what a scribe is most likely to have introduced (intentionally or accidentally) into the text. The processes of investigating intrinsic probabilities can be summarized as the second step of weighing internal evidence:
Key Principle: The variant that has the most continuity with the author’s style is likely the original.
The numerous text-critical principles designed to support this step will be discussed below.
Style is a vague term in NT studies, and this is unfortunate as it leads to imprecision. For example, textual critics usually discuss style under the heading “style and vocabulary,” but this confuses the issue since vocabulary is one way in which an author’s style is expressed. They are not two distinct categories — as the definition below makes clear. Andrew Pitts has provided a thorough linguistic treatment of the issue. According to Pitts, a study of an author’s style pertains to the variation in language that occurs as a response to social situations.1 This highlights in some ways how difficult it may be to use style as a criterion since it can vary so dramatically, depending upon the social situation the author is addressing. However, evidence for a variant reflecting an author’s style can be increased if one is assessing an author’s style based on writings that share a similar social setting (thus 1 Timothy will have a much different style than Romans since the social settings radically differ). Therefore, when examining an author’s style in order to determine the intrinsic probability of a variant, the textual critic is concerned with discovering the variant that has the most cohesion with the immediate context, has the most coherence with the author’s theological emphases (though this too may vary with social situations, so we must be careful here), and reflects the author’s use of language and sources. All other things being equal, the variant that most conforms to the author’s style at all of these levels (contextual cohesion, theological coherence, linguistic conformity, and source consistency) is most likely to be the original. This is the major governing principle when weighing intrinsic probabilities and may be referred to as the principle of stylistic continuity. Put negatively, variants that do not conform to patterns of style that can be observed in an author, especially in the immediate context of the discourse, are least likely to be original. For example, liturgical forms (standardized forms of expression common in worship practices, e.g., a potential ancient creed such as we find in Phil 2:6-11) and harmonizations will tend to cause patterns of discontinuity, indicating possible scribal interference in the transmission of the text.
Cohesion is the textual phenomenon that describes the linguistic features and functions that enable texts to “hang together” as texts: it refers to the unity of a text created through the use of syntax and lexis. It provides the foundation for the unity (continuity) and intelligibility (coherence) of a text. Contextual cohesion creates an interpretive framework in which “some element in the discourse is dependent on that of another. The one PRESUPPOSES the other in the sense that it cannot be effectively decoded except by recourse to it.”2 For example, a phrase like “that man” assumes that a particular man has already been introduced in the previous context. This phrase then is cohesive with and dependent upon the surrounding context for its meaning. M. A. K. Halliday and Ruqaiya Hasan discuss four types of cohesion in English: (1) conjunctions and continuity, (2) referential cohesion, (3) substitution and ellipsis, and (4) lexical cohesion.3 We can observe these and other types of cohesion in the Greek of the NT. Indeed, any device or form can be used to create a cohesive pattern, and often this will aid in textual criticism.
For example, in Col 3:5-12 we have three imperatives (νεκρώσατε, “put to death,” 3:5; ἀπόθεσθε, “put off,” 3:8; ἐνδύσασθε, “put on,” 3:12), each followed by a string of indicatives. This is a passage that illustrates the use of many different cohesive devices in this relatively small unit of the NT. This creates a kind of grammatical cohesion through syntactic patterning. How does this help us with textual criticism? We have some important evidence (most significantly 𝔓46, and some internal considerations) for reading the subjunctive (ψεύδησθε) in 3:9 — even though neither the NA nor the UBSGNT apparatus recognizes or includes this evidence.4 This is where the concept of cohesion comes in. Since there is an established pattern of contextual cohesion that is used to structure this text (imperatives in primary clauses, followed by indicatives in secondary clauses), the subjunctive conforms with the author’s style better than the imperative, since using an imperative here would confuse the structure. Therefore, the subjunctive is more intrinsically probable by virtue of contextual cohesion.
Traditionally, considerations of cohesion in biblical studies have involved the consideration of more limited syntactic tendencies (i.e., much smaller units of text) of an author and especially the use of vocabulary. For example, many scholars have rejected the long ending of Mark (16:9-20) because the vocabulary used there is so different from the words he uses elsewhere in the Gospel. The following words occur in the long ending, for instance, but nowhere else in the Gospel: ἀπιστέω (“to disbelieve”), βλάπτω (“to harm”), βεβαιόω (“to confirm”), ἐπακολουθέω (“to follow”), θεάομαι (“to see”), μετὰ ταῦτα (“after these things”), πορεύομαι (“to go”), συνεργέω (“to work together”), ὕστερον (“later”). In other words, there is no cohesion between the lexical items in the long ending of Mark and the Gospel of Mark as a whole.
Theological and literary coherence has to do with the meaning of a text: it is the feature that provides the text with intelligibility and therefore relates to which themes, motifs, and characters are emphasized by an author throughout the text. Authors and books of the NT tend to have particular themes (whether theological, literary, or linguistic) that provide coherence to their discourses. Therefore, one way of assessing the intrinsic probability of a variant is by determining whether it is theologically coherent with the author’s broader theological framework or whether it coheres with their use of literary devices, such as characterization (e.g., characterizing a particular participant [e.g., Peter] in a certain context) or particular linguistic ways of describing certain phenomena. In other words, all other factors being equal, the variant that most coheres with the theology and literary and linguistic structure that is developed by a book or author will be the one that is most likely original. In John 12:32, for example, some good witnesses read πάντα (“all” [neuter]) (𝔓66 א et al.) instead of πάντας (“all” [masculine]). There are no examples in John’s Gospel of the neuter being used to refer to people. Therefore, reading the neuter here would seem to suggest the idea of universal or cosmic redemption (Christ will redeem all of creation to himself). John’s theological emphasis in his Gospel, however, is upon the universal availability of the atonement to Jews and Greeks (not as much the redemption of creation), and so the masculine (which is used to refer to people) coheres better with John’s theological framework than the neuter. Of course, this principle must be used with discretion and in tandem with other more formally grounded criteria, especially the weight of the external evidence, since it may be that a scribe harmonized the text to fit John’s theology or that lack of theological or literary coherence constitutes a more difficult reading. Thus this principle serves more of a confirmatory function.
In order to highlight the confusion over style in textual criticism and NT studies more broadly, we may cite here G. D. Kilpatrick’s definition of stylistic correction: there are essentially “three kinds of stylistic correction of the text: (1) the elimination of the traces of Semitic idiom, (2) the correction from a non-literary Greek to a literary Koine, (3) the correction to standards of Attic Greek.”5 From our perspective, Kilpatrick’s definition is only one very restricted type of style, known as language formality since it deals with the form of the language that is being used. Therefore, we may speak of whether the language used in a possible variant conforms to the language form used by the author. It is this category of style and its relevance to textual criticism that we want to consider in this section. In practice, the principles of linguistic conformity affirm that, all other things being equal, the reading that most closely conforms to the form of language (e.g., Semitizing, vulgar, nonliterary, literary) used by the author is the most likely to be original.
Some scholars, such as Kilpatrick, believe that scribes Atticized the NT language between A.D. 100 and 200. (This theory is often put forward as evidence for the need of thoroughgoing eclecticism; see ch. 7.) Attic Greek was a literary style of Greek popularized in ancient Athens, representing a high, literary form of Greek. Therefore, to Atticize the NT means that scribes attempted to bring the language of the NT into conformity with the high literary standards found in a select few ancient Athenian authors. However, the theory of Atticizing the early texts does not provide a consistent framework since much of the NT does not reflect Atticizing tendencies.6 Some scribes, for example, may have used Attic word order instead of nonliterary Greek word order, in which case nonliterary word order likely represents the original. For example, in Attic Greek verbs tend to occur at the end of their clauses whereas in nonliterary (esp. NT) Greek they tend to occur at the beginning. Much of this too will depend upon the author. Luke-Acts is composed in a more literary style, whereas John’s writing comes closer to vulgar nonliterary Greek in places. We would expect less Atticizing (if this does rarely exist in the NT) in John than in Luke-Acts, in other words.
Another significant element of linguistic conformity involves Semitic influence, which may take the form of Hebrew or (esp.) Aramaic interference with syntax or vocabulary. The scribal tendency was to smooth out Semitized language. Put positively, Semitic forms of expression can often be taken to indicate the originality of a reading. This is especially important in Gospel analysis, where John and Matthew write from a Semitic background. But since we also have good evidence for some of Jesus’ teaching in Greek, the Greek style may be original too if the passage was likely originally transmitted in Greek (e.g., the Sermon on the Mount or Jesus’ dialogue with Pilate, which cannot have transpired in Aramaic).7 Therefore, it will be important to assess whether a text was likely originally produced in Aramaic or Greek. If a teaching of Jesus, for example, was originally delivered in Aramaic, the possibility that the original author retained Semitic elements is more likely than the hypothesis that a later scribe added them.
That being said, we must note that there is a great deal of debate among scholars concerning what counts as a Semitism. For example, Mark’s Gospel has a tendency to begin its clauses with καί (“and”). At one point, many scholars said this reflected the common Semitic tendency of using the equivalent Semitic conjunction at the beginning of its clauses in order to carry the narrative forward. But now numerous examples have been illustrated from Greek papyri (which have no reason to be influenced by Semitic languages!) in which narratives are carried forward by καί in a way that is similar to what is found in Mark. So while continuity of linguistic form can be factored into the discussion at times, not all scholars will find this body of evidence equally compelling.
A final canon that is especially relevant in Synoptic Gospel studies — and could factor in at other places where sources might be involved (e.g., 2 Peter/Jude) — is source consistency. Most scholars believe that Mark was written first and was used as a source by Matthew and Luke when they composed their Gospels (the view known as Markan priority). Some have viewed Matthew as the first Gospel, with Luke and Mark using Matthew as a source. Still other (mostly) more recent scholars have sought to explain Gospel similarities and differences primarily by reference to the fact that they all draw upon the same body of oral tradition. So depending upon one’s theory of Gospel relationships, the issue of sources will factor into the text-critical discussion. For example, assuming Markan priority for now, at Mark 1:40 NA27/28/UBSGNT4/5 has καὶ γονυπετῶν (“and kneeling down”) in brackets. Several manuscripts (e.g., D W) omit it, but it is present in others (e.g., א). In addition to the weight of the external evidence being on the side of retaining it, similar readings (i.e., words for kneeling) are found in the parallel passages in Matt 8:2 and Luke 5:12, indicating that it was probably original in Mark since Mark functioned as Matthew and Luke’s source — or, at least, so the theory goes.
In this chapter we have highlighted the role of internal evidence as a productive secondary tool when external evidence is not decisive and have pointed to another set of criteria that textual critics often use when weighing internal evidence: intrinsic probabilities. Whereas transcriptional probabilities assess the probability that particular variants arose as a result of scribal errors and/or practices, intrinsic probabilities deal with the level of probability that a particular variant was written by the author. Transcriptional probabilities, the first step in weighing internal evidence (covered in ch. 9), deal with assessing which variants conform to known scribal tendencies. Intrinsic probabilities, the second step in weighing internal evidence, deal with assessing which variants conform to what is known of an author’s style. A number of principles of textual criticism have been discussed that support step 2 when weighing internal evidence, including stylistic continuity, contextual cohesion, theological coherence, linguistic conformity, and source consistency.
Aland, Kurt, 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. Trans. Erroll F. Rhodes. 2nd ed. Grand Rapids: Eerdmans, 1989.
Epp, Eldon Jay, and Gordon D. Fee. Studies in the Theory and Method of New Testament Textual Criticism. SD 45. Grand Rapids: Eerdmans, 1993.
Greenlee, J. Harold. Scribes, Scrolls, and Scripture: A Student’s Guide to New Testament Textual Criticism. Grand Rapids: Eerdmans, 1985.
———. The Text of the New Testament: From Manuscript to Modern Edition. Peabody, MA: Hendrickson, 2008. [Revision of Scribes, Scrolls, and Scripture.]
Halliday, M. A. K. Halliday’s Introduction to Functional Grammar. Rev. C. M. I. M. Matthiessen. 4th ed. London: Routledge, 2014.
Halliday, M. A. K., and Ruqaiya Hasan. Cohesion in English. English Language Series 9. London: Longman, 1976.
Kilpatrick, George D. “Style and Text in the Greek New Testament.” Pages 153-60 in Studies in the History and Text of the New Testament in Honor of Kenneth Willis Clark. Ed. Boyd L. Daniel and M. Jack Suggs. SD 29. Salt Lake City: University of Utah, 1967.
———. “Atticism and the Future of ΖΗΝ.” NovT 25 (1983): 146-51.
Landon, Charles. A Text-Critical Study of the Epistle of Jude. JSNTSup 135. Sheffield: Sheffield Academic Press, 1996.
Metzger, Bruce M. The Text of the New Testament: Its Transmission, Corruption, and Restoration. 4th ed. Rev. Bart D. Ehrman. Oxford: Oxford University Press, 2005.
Moulton, James Hope, Wilbert Francis Howard, and Nigel Turner. A Grammar of New Testament Greek. 4 vols. Edinburgh: T&T Clark, 1906-1976.
Pitts, Andrew W. “Style and Pseudonymity in Pauline Scholarship: A Register Based Configuration.” Pages 113-52 in Paul and Pseudepigraphy. Ed. Stanley E. Porter and Gregory P. Fewster. PAST 6. Leiden: Brill, 2013.
Porter, Stanley E. “P.Oxy. 744.4 and Colossians 3:9.” Bib 73 (1992): 565-67.
1. Andrew W. Pitts, “Style and Pseudonymity in Pauline Scholarship: A Register Based Configuration,” in Paul and Pseudepigraphy (ed. Stanley E. Porter and Gregory P. Fewster; PAST 6; Leiden: Brill, 2013), 113-52.
2. M. A. K. Halliday and Ruqaiya Hasan, Cohesion in English (English Language Series 9; London: Longman, 1976), 2, 4.
3. M. A. K. Halliday, Halliday’s Introduction to Functional Grammar (rev. C. M. I. M. Matthiessen; 4th ed.; London: Routledge, 2014), 603-7; cf. Halliday and Hasan, Cohesion, 13.
4. See Stanley E. Porter, “P.Oxy. 744.4 and Colossians 3:9,” Bib 73 (1992): 565-67.
5. George D. Kilpatrick, “Style and Text in the Greek New Testament,” in Studies in the History and Text of the New Testament in Honor of Kenneth Willis Clark (ed. Boyd L. Daniel and M. Jack Suggs; SD 29; Salt Lake City: University of Utah, 1967), 153. On stylistic correction, see also, e.g., J. Keith Elliott, The Greek Text of the Epistles to Timothy and Titus (Salt Lake City: University of Utah Press, 1968), passim.
6. See J. A. L. Lee, “The Atticist Grammarians,” in The Language of the New Testament: Context, History, and Development (ed. Stanley E. Porter and Andrew W. Pitts; ECHC 3; LBS 6; Leiden: Brill, 2013), 283-308.
7. See Stanley E. Porter, “Jesus and the Use of Greek in Galilee,” in Studying the Historical Jesus: Evaluations of the State of Current Research (ed. Bruce Chilton and Craig A. Evans; NTTS 19; Leiden: Brill, 1994), 123-54.