Towards a New Map of Second Temple Literature: Revelation, Rewriting, and Genre Before the Bible

This forum has as its goal a new cartography, a new process for charting the relationships between points on a map – the paths that connect them and the borders that separate them. The area to be re-mapped is the textual landscape of early Judaism, and in particular, in the context of this forum, those texts marked by particularly intense relationships with existing prestigious texts. In the earlier edition of our metaphorical “atlas,” prototypical examples of such texts were found on the adjacent pages “Rewritten Bible” (Temple Scroll, Jubilees, Genesis Apocryphon) and “Rewritten Bible but also Bible” (Deuteronomy, Chronicles). The old maps have been recognized as obsolete – we recognize that the path from Deuteronomy to the Temple Scroll is much shorter and more direct than the border separating “Bible” from “not-Bible” would seem to allow. But we are only beginning to undertake the process of actually drawing new maps; of figuring out how to talk about these texts absent the distorting framework of later canonical realities.

I have been thinking about these issues for a while, as some readers will know. The cartographical language in the description of this forum struck a chord, as I draw on the same imagery in my new book (Genres of Rewriting in Second Temple Judaism, Cambridge 2020).^[1] There, I attempt to demonstrate that rewriting (defined as the deliberate reproduction, with modifications, of specific existing texts) was a ubiquitous literary technique in the Second Temple period, occurring across genres and, yes, across canonical boundaries (and other boundaries often reified by scholars, such as that between “sectarian” and “non-sectarian” literature from Qumran). Part of the point of the book is to argue that the texts later included in Bibles were part of a broader literary culture in which they did not, as a group, occupy any particularly special place. But tearing down the canonical walls, or erasing the border lines that we drew on earlier maps (or whatever metaphor we choose) is only a prolegomenon, however necessary. We need to draw that new map; chart the relationships between different texts. If categories like “Bible” or “parabiblical” are no longer meaningful, what should they be replaced with? What follows are some preliminary thoughts about how we might approach this project.

A large part of the conceptual problem here has to do with the (perhaps natural human) tendency to emphasize one particular feature of the data, and then use that feature to create the basic categories by which we organize said data. This is particularly unfortunate when the feature chosen is a variable that appears to have been irrelevant in the context in which the texts were actually produced and used, as is the case with the distinction between biblical and non-biblical. But other similarly large-scale binaries (“sectarian” vs. “non-sectarian” comes to mind) have also hampered our efforts. We need to be more flexible when we think about grouping texts, and to appreciate a basic truth that emerges from genre studies: texts will be grouped in different ways depending on what features the observer is paying attention to (Devitt 2004, 7).

I want to play with the idea that each early Jewish text that we know of can be coded or described according to a number of different variables. These variables should be presumed in the first instance to be independent, though further analysis may show correlations between them (more on that below). Any given text will appear similar to certain other texts on the basis of one variable, but to different texts on the basis of other variables. Rather than “pigeonholing” every text in a single slot, which will inevitably facilitate its comparison to a narrow selection of texts deemed most similar (on what grounds?), such a model will allow scholars to more easily observe points of connection and difference across the corpus, creating multiple coexisting configurations of the data. Of course, for practical purposes, such as lists or surveys, texts cannot be categorized according to all variables at the same time. But part of the point here is to challenge ourselves to be more aware of what we are doing when we group texts in particular ways – to realize that such groupings are choices and that alternative groupings are both possible and desirable.

In my preliminary thinking about this model, I have come up with five basic variables that I think would form a good starting point for this type of approach. They are the ones that occurred most naturally to me given my understanding of the texts and of how scholars have tended to approach them, but I assume they are not the only or the best variables scholars might want to delineate.

1. Literary form. Is the text law? Narrative? Prophetic oracle? Parenesis/exhortation? Hymn or poetry? A combination?

2. Language. Is the text composed in Hebrew, Aramaic, or Greek?^[2]

3. The text’s self-presentation or voicing, especially as relates to ideas about revelation or traditions about prestigious figures from the ancient past. Is the text presented as the words of God, or as a revelation from God to a human recipient, or as the speech of an ancient luminary like Noah or Enoch? Is it presented as the teachings of a(n anonymous or named) well-trained sage, or as the result of inspired exegesis? Or as the words of an anonymous narrator?

4. The text’s observable relationship to other known texts (i.e., allusion, rewriting, or other types of hypertextuality).^[3] Does the text, as far as we can tell, repeatedly reproduce or engage (a) specific earlier text(s)? Is its structure modeled on an existing text? Does it take its point of departure from a character or event known from other texts? (There are particular methodological challenges with this variable, which I’ll address below.)

5. Social group or setting (where identifiable). Can the work be attributed to the yaḥad? Does it evince a particularly priestly orientation, or a certain type of halakhic stance? Other? None?

An Example

The point of a multivariable system like this for thinking and talking about Second Temple texts is to decenter the kinds of categories or boundaries that have tended to silo our work and to reinforce anachronistic frameworks, while still allowing us to recognize the tremendous diversity present within this corpus.

To see what this approach might allow us to do, consider two of the most prototypical exemplars of the category “Rewritten Bible”: the Genesis Apocryphon, with its expansive narratives of the antediluvian patriarchs and imaginative retelling of the Abraham stories, and the Temple Scroll, with its instructions for a gigantic temple complex and rearranged version of the pentateuchal legal corpus. As the traditional classification of both as “Rewritten Bible” suggests, the two appear quite similar in their strong relationship to earlier texts (Variable 4). Both reproduce certain portions of the Pentateuch with minimal changes (e.g., the account of Abram’s wars with the Canaanite kings in GenAp 21:23–22:34 that corresponds closely to Gen 14:1–15:4; the quite precise redeployment of much of Deut 18–24 in 11Q19 60–66). Other portions are rewritten with more substantial interventions, such as the earlier parts of the Abraham cycle as reflected in GenAp 19–21, with their colorful elaborations of Abram and Sarai’s sojourn in Egypt, or the festival laws in TS (11Q19 13–29), which add several feast days unknown in pentateuchal sources. Finally, both texts also contain large sections which do not show close textual connections to the Pentateuch as we know it (though it is possible that each draws on other sources unknown to us) – for instance, the extended sections pertaining to Lamech and the birth of Noah in GenAp 2; and the instructions for the temple and its courts in 11Q19 3–13; 30–46 as well as the Law of the King in 11Q19 57–59.

In other ways, though, GenAp and TS could hardly be more different: the similarity with regard to Variable 4 does not extend to any of the others. With regard to literary form (Variable 1), GenAp is entirely comprised of narrative, while TS contains no narrative at all – it consists almost entirely of commands. GenAp is written in Aramaic, while TS is in Hebrew (Variable 2). The question of self-presentation (Variable 3) also highlights differences: TS presents itself as the direct words of God, who speaks in the first person. GenAp likewise contains first-person speech, but the speakers are the various patriarchs rather than God. GenAp also juxtaposes the first-person sections with a final section in which Abraham is referred to in the 3d person, the voicing here reminiscent of the anonymous narrator of Genesis itself.^[4] The social context (Variable 5) is somewhat difficult to assess given the differences in content and setting of the two texts. TS clearly represents a priestly ideology that is strongly connected in principle to the temple and the sacrificial cult. GenAp shows no trace of such interests, though it need not necessarily follow that this means the compositions originated in two separate social groups.

I chose this example because it’s a case where it is quite easy to see that the connection between these two texts is based largely on a single variable, relationship to earlier texts (Variable 4), and that if we started looking for connections between these and other texts based on other variables, we would be led in quite different directions. On the basis of literary form (Variable 1), GenAp would be grouped with other narrative texts – the book of Genesis, but also (depending on how broad or narrow we wanted to be) parts of Numbers, Joshua-Judges, Samuel-Kings, Jubilees, 1 Enoch, Tobit, Daniel 1–6, etc. TS on the other hand most resembles (of course) the legal portions of Exodus, Leviticus, Numbers, and Deuteronomy, but also could be compared to a variety of halakhic texts: the legal sections of the Community Rule, the Damascus Document, and 4QMMT, as well as (in the architectural details) Ezekiel 40–48 and New Jerusalem. With regard to Variable 3 (self-presentation), the Temple Scroll’s insistence on God speaking in the first person is somewhat distinctive, but the Sinaitic provenance puts TS in conversation with (of course) Exodus–Numbers and Jubilees, as well as with some fragmentary “para-Torah” texts such as 4Q368. GenAp, on the other hand, in presenting the first-person speech of Israel’s earliest ancestors, resembles 1 Enoch and other texts like Aramaic Levi and Visions of Amram, as well as parts of Tobit, Daniel’s Visions, and 4 Ezra. It is likely not a coincidence that the texts that most resemble TS in self-presentation are also in Hebrew, while most of those that resemble GenAp are also in Aramaic (Variable 2; see below on the issue of correlation of variables). Finally, in terms of Variable 5, possible social setting, the differences in content and apparent ideology would probably start us looking in quite different directions in each case.

Now, it’s not that these differences between TS and GenAp have not been observed, or that no one has compared TS to other legal texts or GenAp to other narratives about the patriarchs.^[5] But somehow this breadth of connections must be foregrounded even in more casual considerations or mentions of these texts. When their primary identifier is a label like “Rewritten Bible” (or even something less anachronistic like “centripetal rewriting”^[6]), and this is seen as “what they are” in some sense, all these other differences and alternate connections are missed. Why is the fact that TS is a legal text, or the fact that it records direct divine revelation, not seen as its primary identifier, rather than the fact that it rewrites the Pentateuch? Of course, the answer has to do with our own biases; our well-documented tendency to see everything about this period in relation to the Bible (Mroczek 2016, 118–27). Resisting the urge to put every text into one box (and keep it there) seems to me to be one potent way to actively oppose these biases, which as we have seen do not go down without a fight.

Comments and Clarifications

All of the above, needless to say, is a very basic suggestion about how we could begin to draw new, multidimensional maps of the literary landscape of early Judaism. Actually drawing and refining the maps will be an ongoing task of future scholarship. For now, a few further observations must suffice.

First, some of these variables (which, again, I offer only as a starting point for discussion) pose serious methodological challenges. Many Second Temple texts, for instance, resist an easy connection to a specific social setting (Variable 5). Fragmentary preservation can obscure a text’s voicing or self-presentation (Variable 3), or its literary form (Variable 1). A composition may be preserved in multiple languages, or the original language of composition may be uncertain (Variable 2). Variable 4 presents even more vexing problems. Using rewriting (or other forms of hypertextuality), on its own, as a feature for classifying texts is dangerous because rewriting can only be confidently recognized and described when the earlier text that is being rewritten is available to us. This means we are much more likely to recognize rewriting of texts that ended up in the Bible, precisely because they have been more extensively preserved for us than other texts. We are lucky that enough of the Qumran collection survived to allow us to see connections and overlaps between texts that did not end up in the Bible – that is, we can see that nonbiblical texts, including sectarian texts, were also rewritten.^[7] But the frequency of such rewriting is hard to judge with confidence: if a text rewrites an earlier text that is not (fully) preserved, we won’t recognize it. So it is easy to see what the Temple Scroll has done to the text of Deuteronomy (all caveats, such as fluid texts of both Deuteronomy and TS, notwithstanding; see below). But if other parts of TS (say, the Temple Plan, or the Law of the King) existed in prior forms, we do not know what kinds of changes the composer of TS might have made in the course of reuse. As a result, we need to be careful about how we use rewriting as a mapping tool – because we can never actually be confident that a given text does not employ rewriting. We can group texts based on similar modes of rewriting, but we have to be aware that other texts that seem different (i.e., “non-rewritten”) might in fact be placed in the group if we had a fuller picture of Second Temple literary culture. Similarly, the gaps in our evidence make it difficult to assess accurately the point at which certain rewritings enter the manuscript record. Are the distinctive rewritings we see, for instance, in the 11Q19 version of TS to be viewed as characteristic of that manuscript? Or did some of them originate in earlier versions of TS, or in other versions of the Pentateuch, which we no longer have access to?^[8] Finally, we may need to be careful about characterizing rewriting as tied to some particular aspect of Second Temple literary culture in contrast to earlier periods (e.g., an apparently increased interest in texts as loci of divine wisdom) – I suspect that we simply don’t know enough about how texts were composed in earlier periods to make a viable comparison.^[9]

Second, although I think it is important to imagine these variables as independent from one another, it seems clear that certain variables may correlate (or reverse-correlate) with one another under certain conditions. Rewriting of a given text, especially extensive rewriting, makes likely that the new text will be of the same basic literary form as the older one. But extensive rewriting (at least of texts seen as ancient at the time of their rewriting) might also correlate to certain modes of self-presentation, such as pseudepigraphy – or at least inversely correlate with modes like identification of the author as a present-day figure (see Ben Sira). Or to give an example that correlates Variables 3 and 5: the fact that a text like the Community Rule stems from a community that traces its knowledge of proper Torah to special revelations to its founding figure in the relatively recent past may explain why none of the core yaḥad texts are voiced as revelations from figures in the ancient past (like Enoch or Noah, or even Moses). From a different perspective, recent studies have suggested that the Aramaic texts found at Qumran share a particular social context and ideology (thus correlating Variables 2 and 5).^[10] Teasing out these patterns, to my mind, is a key aspect of the work that needs to be done to construct new models of Second Temple textuality.

Third, on a theoretical or heuristic level, I can imagine some sort of large-scale project that actually codes each text according to each of these variables (or others). Texts could be assigned a value A, B, C, D, E etc. based on literary form (Variable 1); a value i, ii, iii, iv, v etc. based on self-presentation (Variable 3); etc. One could then actually search or filter a database according to each variable or multiple variables (all texts that have A for Variable 1 and iii for Variable 3, for instance), which could show at a glance the varying sorts of relationships or connections between texts.^[11] In fact, a similar project has already been conducted, resulting in the Manchester/Durham Database for the Analysis of Anonymous and Pseudepigraphic Jewish Texts of Antiquity.^[12] That database, as exhaustive as it is (over 500 discrete literary features applied to nearly 100 different texts), presents several difficulties in relation to the Second Temple literary landscape. For instance, it excludes biblical texts, thus assuming the very boundary that we seek to move beyond; and the group of features that pertains to rewriting is restricted to rewriting of biblical texts. This project deserves more discussion than it appears to have received from scholars of Bible and Second Temple Judaism, but at the same time, it illustrates the difficulty with such endeavors: So much of the end result depends on the interpretive choices made at every step of the project. Nevertheless, whether or not something like the Manchester/Durham database could or should be extended or adapted for our purposes, we would be well-served by thinking as if we were doing it – by trying consciously to introduce more multidimensionality into our discussions of textual categories.

Fourth and finally, I want to close by saying a little more about genre. I deliberately opted for the term “literary form” instead of “genre” for Variable 1, though many would understand the kinds of things I offered as examples (law, narrative, etc.) as genres. Of course, they are, but we need to operate with a more flexible and nuanced understanding of genre: genre cannot be reduced to literary form. For one thing, many (most?) texts are made out of a mix of literary forms (e.g. Jubilees, containing both law and narrative, or the Community Rule, containing paranesis, command/halakhah, and poetic material). For another, there is a real sense in which other variables (especially Variable 3, self-presentation, but also Variable 4, relationship to prior texts, and perhaps in some cases also Variable 2, language) can themselves contribute to a text’s generic identity. Thus, a term like Najman’s “Mosaic Discourse” can certainly describe a genre.^[13] At the same time, it is worthwhile to be aware that texts that might be seen as belonging to this group can exhibit different formal features (Variable 1) as well as differences in other variables (easily Variable 4, different relationships to earlier texts; likely Variable 3 and 5 as well). Genre is not about fixed formal features, but about the way various aspects of the text and its framing coalesce to give a certain impression of what kind of text this is. The perception of what is significant in this interplay of features inevitably varies from one reader/auditor of the text to another. As I’ve said in other contexts, I have found thinking about the fluidity and multidimensionality of genres an excellent tool for us perplexed cartographers, as we struggle to make sense of and accurately represent the complex web of interrelationships that characterizes Second Temple literary culture.^[14]