Richard A. O’Keefe, Critical Remarks on Houk’s ‘Statistical Analysis of Genesis Sources,’ JSOT 29.4 (2005) 409-437.
This essay critiques the methodology of Cornelius B. Houk’s article which appeared in JSOT in 2002. Houk argues on the basis of statistical analysis of syllable-word frequency that eleven distinct authors contributed to the composition of Genesis.
One critique of Houk’s methodology that O’Keefe gives is the process of determining the sections of the text of Genesis that are compared to one another. Keefe states:
There is a cirularity here. The artile begins with a partitioning of the text according to ‘the judgments of source criticism.’ The questions put to the data do not seek alternative groupings; the test results correlate with source criticism because of the way the tests are constructed, not because the data have nothing suprising to tell us…If we are given a group of students and separate them into tall ones and short ones, it comes as no surprise if the tall ones turn out to be significantly heavier than the short ones (412).
Another critique that Keefe presents is Houk’s assumption that statistically significant differences among text groups indicates different authorship and vice versa. One must take into account several factors that may lead to differences between passages–some of these differences can arise from a single author. For instance, Keefe states:
Houk’s own tables show between-genre differences comparable to between-author differences.
Any authorship study needs to be sensitive to at least statistical, linguistic, and literary issues. Before attributing any effect to authorship, one must be sure that the effect is present, is not attributable to straightforward linguistic causes, and is not attributable to literary considerations (424).
Direct speech is particularly prone to variations from the expected norm of linguistic usage. O’Keefe asserts that direct speech “should be expected to vary with the age, sex, ethnicity, and emotional state of the character speaking (424).” Therefore, if one encounters a speech by Abimelech and it seems different from other passages one cannot automatically assume that this speech was composed by a different author. Maybe the author was making Abimelech sound like a foreigner.
Finally, O’Keefe states that Houk failed to adequately double-check his research methodology. After Houk attained his results–results that confirmed his hypothesis–Houk did not rigorously test his methodology to make sure that his analysis was correct. O’Keefe advises:
The hardest thing of all is to remember that when we do get the right answers, it’s still true that we may have made a mistake. We still need to probe the data for evidence that we might have got it wrong. We must put questions to our data that give nature the chance to tell us about the unexpected (434).
O’Keefe provides us with valuable advice and warnings as we comb the biblical text, or any text for that matter, for statistically significant differences that might lead to conclusions concerning authorship (I might add also syntatical, genre, literary, dating, and stylistic conclusions as well). Statistical analysis is a wonderful complement to traditional biblical research. Statistics yield hard and fast data along with a methodology that can be replicated and checked. Statistics moves us away from vague feelings or hunches, but statistical analysis is only as good as the data, the method of analysis, and the interpretation of the results. There is still plenty of room for ambiguity even within statistics.
O’Keefe also warn us against circular reasoning as we construct research methodologies and opinions. He brings a sharp critique of Houk because Houk set out to prove multiple authorship of Genesis. As Houk segmented the text of Genesis, he did so according to the conventional thought of source criticism, which itself asserts multiple authorship of Genesis. You prove multiple authorship of Genesis by assuming multiple authorship of Genesis and working that assumption into your research structure. This circular reasoning resides not only with Houk but with many other researchers and even possibly within entire disciplines of thought such as source criticism itself.
As we reflect upon O’Keefe’s advice and warnings, the most valuable aspect of his work may not be the additional scrutiny with which we judge other scholarship with–but in fact, the additional scrutiny with which we judge our own conclusions and methodologies. As O’Keefe states, we are only human and are prone to error. Therefore, we must constantly re-evaluate our own positions and the way that we arrived at them to ensure that we are not employing faulty data, methodologies, or conclusions.