📄 Mette Bundvad – Kasper Siegismund (eds.), Vision, Narrative… , and Wisdom in the Aramaic Texts from Qumran. Essays from the Copenhagen Symposium, 14–15 August, 2017 (with the collaboration of Melissa Sayyad Bach – Søren Holst – Jespers Høgenhaven)

Summary

This is Henryk Drawnel's review, in The Biblical Annals, of a Brill volume (STDJ 131, 2020) edited by Mette Bundvad and Kasper Siegismund. The volume collects twelve essays from a 2017 Copenhagen symposium on the Aramaic manuscripts found at Qumran. These texts are scarce and badly fragmented, yet they are among the few windows onto Judean history in the third and early second centuries BCE.

The essays fall into three blocks:

• memory and eschatological expectation in single texts,
• the Visions of Amram, covering its characters, its geography, a reconstruction of fragment order and a disputed verb form,
• social setting, links to Luke, and modern forgeries.

Drawnel summarizes each contribution and then critiques two of them. He questions Antonissen's reading of New Jerusalem through Greco-Roman banquet culture and suggests Israel's sacrificial system and Hatra epigraphy would be better comparisons. He also argues that Goldman's two possible themes for Visions of Amram, priestly line versus exodus, do not compete with each other. Overall he judges the book a valuable step for a rarely convened field. He also warns that broad social reconstructions rest on very thin evidence.

Key points

• The book has 289 pages and twelve papers in three groups of four. The third group covers socio-historical context, ties to the New Testament, and fake fragments.
• The background cites 4Q208, dated by palaeography to roughly the end of the third century BCE. It holds a lunar calculation more coherent than the later Ethiopic 1 Enoch 73:4-8, and so pushes the Astronomical Book back at least to the third century BCE.
• Søren Holst used textual overlaps to order the Visions of Amram fragments: 4Q545 cols. I–IV, then 4Q544 col. I, then 4Q543 cols. I–II and V, then 4Q547 col. III.
• Kasper Siegismund rejects Robert Duke's reading of a Qumran Aramaic verb meaning "to give" as a 3rd sg. suffix-conjugation form. That root appears only in the prefix conjugation, with a suppletive root covering the perfect. He therefore parses it as 1st pl. "we will give".
• Årstein Justnes examines eight unprovenanced "Dead Sea" fragments traced to dealer William Kando:
  • holders: four at Schøyen, two at Southwestern Baptist Theological Seminary, one at Azusa Pacific;
  • contents: two from Tobit, three from Daniel, three from 1 Enoch.

Interesting ideas

• I1. Siegismund rules out a morphological parse because the corpus never attests that root in that paradigm slot, where a suppletive stem fills the gap; corpus attestation acts as a hard constraint on analysis.
• I2. Holst rebuilds the order of a fragmentary text purely from overlapping material across copies, chaining shared stretches into one sequence.
• I3. Høgenhaven reads a sequence of time periods out of place names alone, and notices that the hero's journey inverts the surrounding background story's route.
• I4. Drawnel faults an essay for reaching for an outside parallel before interpreting a phrase within its own composition.
• I5. Justnes shows that unprovenanced items entered collections and publications through buyer naiveté, so provenance itself becomes evidence.

For Mova

How these ideas could serve Mova, our small open language model — its files, stages and tests:

• I1 (tokens): The lemmatizer/tagger could record which lemma–paradigm cells a corpus actually fills, including suppletive pairs like go/went, and down-weight analyses that land in empty or suppletively occupied cells.
• I2 (world): Overlapping mentions across paraphrased or partial retellings could be aligned to reconstruct a single event order for the world of a text, rather than trusting any one witness's sequence.
• I3 (world): Location changes of entities can serve as a proxy timeline, and comparing a character's path with a framing story's path can flag deliberate inversions or parallels worth explaining.
• I4 (frames): Coreference and frame interpretation should first exhaust evidence inside the document before falling back to external knowledge, and the explanation trace should show which source decided.
• I5 (learn): Every silver or teacher-generated training example should carry a documented origin, and examples with missing provenance could be quarantined or used as negative controls.