Part 2: Statistics on Writing Styles

So, how does Ehrman attempt to prove forgery in the NT? He uses the traditional arguments that have been debated for centuries: differences in style, conceptual/theological differences, and historical discrepancies from known facts. Arguments on both sides have been made, and continue to be made, in the scholarly literature. There is a ready answer to arguments that the authors of the NT are not those claimed; see, for example, the NT introductions by Carson and Moo; Guthrie; and Köstenberger, Kellum, and Quarles.

Ehrman however ratchets up the discussion with statistical analysis. After discussing only a part of the data (word usage) that makes up an author’s style, Ehrman concludes: “In almost every study done [in the last ninety years], it is clear that the word usage of the Pastorals is different from that in Paul’s other letters” (98). The documentation at this point cites but one author, Armin Baum, who argues, contra Ehrman, that Paul wrote the Pastorals! Further, Ehrman fails to mention the most recent sophisticated computer-assisted researches by Anthony Kenny, A Stylometric Study of the New Testament (NY: Oxford University Press, 1986), and K. J. Neumann, The Authenticity of the Pauline Epistles in the Light of Stylostatistical Analysis (Atlanta: Scholars, 1990). Kenny’s research concludes that, according to computer analysis, only 1 and 2 Timothy of the Pastorals are Pauline, while Titus is not. Yet no scholar, as far as I know, makes this claim on other grounds: the Pastorals are virtually always seen as a unit, written by the same author, whether Paul or someone else (though sometimes 2 Timothy, not Titus, is viewed as written by a different author than 1 Timothy and Titus). And Neumann, in spite of expecting quite different results, notes somberly that “The hopes did not materialize that the greater labor connected with several syntactic-category indices might produce some very significant criteria. … there is more variability within authors than anticipated” (205). In one test, 2 Thessalonians and 1 Peter both lined up with Paul’s writing style perfectly; in another, Revelation, chapters 2 and 3 were considered Pauline! No wonder Neumann concludes, “Christian authors, especially Paul, are not distinguished by the indices chosen” (213). Surely, these are not the modern sophisticated statistical studies that Ehrman is thinking of, but neither does he mention any in support of his views.

C Michael Patton
C Michael Patton

C. Michael Patton is the primary contributor to the Parchment and Pen/Credo House Blog. He has been in ministry for nearly twenty years as a pastor, author, speaker, and blogger. Th.M. Dallas Theological Seminary (2001), president of Credo House Ministries and Credo Courses, author of Now that I'm a Christian (Crossway, 2014) Increase My Faith (Credo House, 2011), and The Theology Program (Reclaiming the Mind Ministries, 2001-2006), host of Theology Unplugged, and primary blogger here at Parchment and Pen. But, most importantly, husband to a beautiful wife and father to four awesome children. Michael is available for speaking engagements. Find him everywhere: Find him everywhere

    14 replies to "Book Review of Bart D. Ehrman’s “Forged” – Part 2"

    • Have any of these statistical methods been tested on modern authors, books where we know who the author really is. The results you mention make them sound highly suspect, but if I had such a method that I wished to use on ancient literature (Biblical or otherwise) the first thing I would do is test it on cases where I knew the answer to gauge its accuracy. Yet I never hear of this being done. Is this ever done and if not, why not?

    • Daniel B. Wallace

      Mike, I believe it has been done on Shakespeare and other English writers. I do not know the degree to which it has been done on known authors with target works though. However, the studies I mentioned above are necessarily restricted to Greek. Thus, they could not equally be applied to an author in a different language.

    • Erlend

      Thanks for this Dan. I was under a completely different understanding on the results of statistical analysis. I will really need to look into this more! I was wondering though does Ehrman try to press theological distinctions between the pastorals and other Pauline letters to make his point, i.e. over ‘charismatic’ vs. developed church order?

      I always though a good study would be to compare the language Cicero uses in his letters. Spanning over three volumes they offer a rare chance to compare a classical writer’s employment of language over several years, to different audiences (from close friends to professional colleagues) and covering numerous different issues (from philosophical, political, and more casual-even pastoral- conversations.) Just a shame that he wrote in Latin! Still it would be interesting.

    • Maybe I am just cynical (or have studied the hard sciences to long) but I would find it hard to apply such a technique to a disputed text before first testing it against texts where I felt confident I know author (even if it had to be in Greek). But maybe this is unreasonable.

    • Daniel B. Wallace

      No, you’re right on target, Mike. And that is what was done: on the assumption that the seven undisputed letters of Paul were the basis for comparison, then the statistical analysis was done.

    • Ounbbl

      In the basic of statistics we learned at the beginning that statistical analysis is not meant prove a thing.

      Let Bart bark as he wishes with a high statistical possibility of bark being fart here 😉

    • X

      Is this the best photo Dan Walaace can get of himself? Is he wearing a towel or his underwear on his head?

    • bethyada

      Being a fan of statistics this does not excite me. People think this is not what I expected but they often do not know what should be expected.

      For example, one can’t compare Paul certain to Paul disputed without first comparing a range of undisputed texts by other authors. Eg. Shakespeare to all of Shakespeare and seeing the range of usage, and how far the most different texts diverge. Then another author, and another. Only then will one get the information to analyse Paul which will not give a yes/ no answer but a probability; eg: assuming the church epistles are by the same author then Titus has a probability of being by that same author of x%.

    • bethyada

      And reasons why the word stats may be different?

      Different topic, different writing style (eg. less formal to individual compared with church perhaps), time delay, the use of an amanuensis some times (or different ones), especially if they are allowed significant freedom in taking dictation!

    • […] – Dan Wallace continues to review Bart D. Ehrman’s Forged. […]

    • […] has written a three-part review that is very helpful, and I encourage you to read it here: part 1, part 2, part 3. Wallace concludes with […]

    • […] Dan Wallace on the Parchment and Paper blog reviews Bart Ehrman’s book, Forged in two parts, with a third projected. (part one, part two). […]

    • Paul

      I think these debates only demonstrate the need for the authority of Church tradition, the problem began to unravel once the Reformers departed company with that tradition. it is no wonder a number of prominent intellectual evangelicals have defected to Eastern Orthodxy – Swinburne, Schaffer and Gilchrist. While i have not gone that way, it does make sense as long as it can be establish Jesus conferred his authority to others after him from undisputed texts.

    • CarsonB

      Other than word usage, what exactly makes up the data of an “author’s style” in the NT?

Leave a Reply

Your email address will not be published.