Ha’aretz reports the development of software which can do Pentateuchal source criticism.
The software is described in a paper by Moshe Koppel, Navot Akiva, Idan Dershowitz and Nachum Dershowitz, entitled “Unsupervised Decomposition of a Document into Authorial Components”, presented at The 49th Annual Meeting of the Association for Computational Linguistics: Human Language Technologies, Portland, Oregon (Monday 20 June, 2011).
The paper can be read here, on pdf. The abstract includes the delightful boast, “This allows us to automatically recapitulate many conclusions reached by Bible scholars over centuries of research”!
The authors report that their method “corresponds to the expert consensus regarding P and non-P for over 90% of the verses in the Pentateuch for which such consensus exists” (p. 8). The paper provides no details, however, of the verses included in this scholarly consensus (most of the description in the paper concerns their analysis of Ezekiel and Jeremiah, which is their “munged” trial-run). It is also not clear whether the authors have found a common style for, say, P across the entire Pentateuch, or whether the authors have simply managed to find a basis to split the books of the Pentateuch into two, based on stylistic differences. But given the debate over the extent of P in recent decades, the exact nature of such findings would be crucial.
Any findings are, however, dependent on the assumptions and input. The basis for distinguishing between sources is the use of synonyms, which as the authors note has long been used in Pentateuchal scholarship (e.g. Carpenter and Harford). As synonyms only allow the separation of a limited number of passages, the authors then “fill out” the analysis by using a sample of such words. Unfortunately, there is not much detail in the paper on which synonyms were used. A possible danger in grouping words together to analyse the rest of the Pentateuch is that it may provide a bias towards constructing pan-Pentateuchal documentary sources where there are only fragments, or not recognizing differences in the nature of the two sources identified by synonym-analysis between each book of the Pentateuch.
However, it pays not to make any conclusions based on the presented paper. The analysis looks very promising. Presumably fuller details of their assumptions and findings on the Pentateuch will be offered in a published paper, sometime in the future. It will be one to look out for!