This last year, in an effort to better understand the citation patterns in the Babylonian Talmud (Bavli) and the ways in which the Bavli knew and used the Bible, I developed a spreadsheet that contains each biblical citation in the Bavli(described in more detail here). In my first attempt to analyze this data quantitatively, I sought to identify the number of unique verses cited and their frequency (see here). My statistical skills are relatively basic, so I have since been spending some time thinking about the kinds of analyses that might want to run on this data.
In the table below, I seek to identify the relative citation frequency of individual books of the Bible. The number in the table is the ration of relative frequency of the citation of this book in the Bavli to the books relative size (by number of verses). That is: ((Number of times the Bavli cites the book)/(Number of biblical citations contained in the Bavli))/(Number of verses in the biblical book)/((Number of
verses in the Hebrew Bible)).
A number above 1 means that the book is “overrepresented” as a measure of its quantitative size in the Hebrew Bible, whereas a number less than 1 signifies its underrepresentation:
|
Lev |
5.9 |
|
Deut |
3.35 |
|
Exod |
2.04 |
|
Num |
1.83 |
|
Malachi |
1.5 |
|
Esther |
1.48 |
|
Song |
1.24 |
|
Eccl |
1.13 |
|
Gen |
0.88 |
|
Prov |
0.86 |
|
Hosea |
0.84 |
|
Amos |
0.79 |
|
Zeph |
0.78 |
|
Isaiah |
0.76 |
|
Psalms |
0.66 |
|
Ruth |
0.66 |
|
Hab |
0.66 |
|
Lam |
0.65 |
|
Jonah |
0.55 |
|
1 Sam |
0.53 |
|
Hag |
0.51 |
|
Zech |
0.51 |
|
2Sam |
0.49 |
|
Job |
0.42 |
|
Dan |
0.4 |
|
1Kings |
0.39 |
|
Ezek |
0.37 |
|
Judges |
0.36 |
|
Joel |
0.36 |
|
Ezra |
0.34 |
|
Nahum |
0.34 |
|
Jer |
0.34 |
|
2Kings |
0.3 |
|
Joshua |
0.3 |
|
Neh |
0.25 |
|
1Chr |
0.17 |
|
2Chr |
0.16 |
|
Ovad |
0.08 |
Here are a few preliminary observations and hypotheses:
1. Four of the five books of the Pentateuch (Torah) are overrepresented. Since the Bavli is predominantly concerned with law, this is unsurprising (although the dominant position of Leviticus caught me a little by surprise).
Genesis contains little law, and so is underrepresented.It would be interesting to follow this up to identify whether there are clusters of tractates that account for large chunks of these citations. For example, I wonder if Leviticus is vastly overrepresented in a few tractates but otherwise drops to a more “average” rate of representation.
2. One might expect that books that had liturgical functions, such as Esther, Ruth, Ecclesiastes, Lamentations, Psalms, and Song of Songs, would be overrepresented. This is only partially borne out by the data. Psalms, Lamentations, and Ruth are in fact significantly underrepresented.
3. Some prophetic books are cited far more than others. Why?
4. The historical books are all underrepresented. Perhaps this correlates with the rabbis’ relative disinterest in history.
5. It is hard to figure out why Malachi would be overrepresented and Ovadiah so underrepresented. Might this simply be a statistical fluke since both books are so short?
As always, comments and thoughts welcome!
Interesting project, Prof. Satlow. A few questions:
Would it be useful to break out citations by halakhic vs aggadic or mishna/gemara?
What is it you hope to find out?
If you were going to critique this methodology, what would your points be?
What an interesting and useful project! Some thoughts, hopefully helpful:
Over- and underrepresentation measured by size seems problematic. If a topic is described in great detail it may be more likely to generate more citation and discussion. Likewise, topics might be considered major though their biblical treatment is cursory, so that a highly discursive and expository text like the Talmud might spend “inordinate” amount of time on them. The parameters, in other words, cut in many directions at once. You might also distinguish between Talmudic analysis of a passage and (mere) citation of it. You might also aim to track whether certain biblical passages are cited regularly in particular, recurring contexts or in many contexts.
You both raise excellent points. I see two distinct issues here. The first is dealing with the data that I have vs. the data that we would want. I have a spreadsheet that simply lists the over 13,000 biblical citations in the Bavli and where they appear – that’s about it. This was not inexpensive to compile, but it was managable. What we might like is a far more contextually nuanced spreadsheet. This could potentially be an enormously complex task that ultimately involves contextually tagging the Bavli, and depending on the amount of human judgement we would want to integrate into it. That is a kind of dream project that I have, in which the Bavli can be tagged for all kinds of things and thus enable all kinds of digital analyses.
The second issue is to what extent the data that I already have is useful. I originally compiled it simply out of curiosity to see which verse(s) the Bavli “likes” the most. The results, which I indicated, are somewhat interesting and I’m still figuring out what to do with them. At the same time I now have a dataset on which I can run a variety of quantitative analyses. If you have any in mind that you think might be interesting, let me know!
One idea might to be compare usage of biblical text for drawing or supporting halakhic conclusions vs. philosophical thoughts and opinions. Another idea might be to use the mishnah about the three kinds of linkages between halakhic complexes and scriptural anchors (mountain hanging by a thread, etc.) and see what the citation patterns can teach about the scheme and the processes of halakhic reasoning.
It would be useful, while doing this, to list the various citation formulas that accompany each verse (e.g. dikh’tiv, shene’emar), perhaps with variants of the formulas from MS Munich (Dikdukei Soferim). This might facilitate determining whether there are subtle differences in the way each is used, such as more literal vs. less literal use of the verse in question.
Dear Michael,
Thanks for your continued pursuit of the data and their interpretation. I agree with Jeff Tigay, it would be interesting to see how the passages are introduced, if some taxonomy is forthcoming or whether the introductory tags are simply random.
You write: “One might expect that books that had liturgical functions, such as
Esther, Ruth, Ecclesiastes, Lamentations, Psalms, and Song of Songs,
would be overrepresented. This is only partially borne out by the
data. Psalms, Lamentations, and Ruth are in fact significantly
underrepresented.” Note, however, that (as far as I know) the liturgical use of some of these books (Ruth, Qohelet, in particular) is rather late and is part of Ashkenazi tradition only (or mainly).
Re Malachi: Is it a series of verses? (which would be interesting) — or is perhaps 3:23-24, the closing of the book, which resonates in Jewish tradition, cited repeatedly, thus skewing the data to some extent? Perhaps also Mal 2:7?
Finally, do you intend to perform the same study for the Yerushalmi? It would be interesting to compare results, especially given the much greater focus on halakha only in the Yerushalmi, versus the mix of halakha and aggada in the Bavli.