📄 Monotheism and the Language of Divine Plurality in the Hebrew Bible and the Dead Sea Scrolls

Summary

Most Hebrew Bible scholars hold that Israelite religion moved from polytheism to monotheism, with the shift usually dated to the exile or soon after. Peter Hayman (1991) argued for a much later endpoint, because references to many gods (elohim, elim) gathered under YHWH continue into later Jewish writing. Heiser revisits this claim. He compares divine-plurality vocabulary in the canonical Hebrew Bible with the Qumran sectarian texts, which use such language far more often than the rabbinic material Hayman relied on. Drawing on Jan Assmann's account of what a polytheistic system actually requires, he argues that plural divine terms alone do not show polytheism. He also points out an inconsistency: scholars translate the same words as "gods" in biblical texts but as "angels" at Qumran. His proposal is that elohim marks membership in the unseen, non-human realm, not equal status. On this reading YHWH belongs to that class while remaining unique within it. If so, the vocabulary is no longer evidence for religious evolution, and biblical and Second Temple thinkers shared one outlook.

Key points

• The paper replies to Hayman's 1991 article in JJS 42, which asked whether "monotheism" is a wrongly applied term in the study of Judaism. Hayman pushed the arrival of monotheism to the Middle Ages and cited only two Qumran texts (1QM 15:14, 4Q491 11).
• Assmann lists three features a polytheistic system needs:
  • creation and upkeep of the world shared among several deities;
  • a diversified, politicised cult spread over many sites;
  • mythic biographies in which gods are defined through relations with other gods.
• Assmann also separates "evolutionary monotheism" (henotheism growing out of polytheism) from "revolutionary monotheism," which defines itself by what god is not and how god must not be worshipped.
• Biblical texts give YHWH traits never given to other elohim: he is all-powerful, king over them, their creator, and the only one they should worship (e.g. Ps. 29:1, Ps. 95:3, Neh. 9:6).
• Earlier sense inventories of elohim (Burnett 2001; Wardlaw 2008) list its uses but do not say what unites them. Heiser proposes the unifying feature is "inhabitant of the disembodied realm."

Interesting ideas

• I1. The same term in the same phrases and contexts should get the same reading across corpora; letting a prior theory change the gloss per corpus is a methodological error.
• I2. A word can name a class (which realm an entity belongs to) without implying that all members share attributes or rank. Distinctions among members come from predicates and context, not from the noun itself.
• I3. Judge whether a system is present by its required structural components, not by one surface lexical cue. Heiser uses Assmann's three criteria instead of the bare presence of plural "gods."
• I4. A "member of class X, yet unique within X" pattern resolves an apparent contradiction: X is a kind, and uniqueness is stated by exclusive predicates such as "none like you" or "alone."
• I5. A word's meaning should fit the whole worldview of its users; a sense that contradicts their other stated commitments is the wrong sense.

For Mova

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

• I1 (learn): Add consistency checks that flag when one lemma in matching syntactic frames gets different senses or roles in different text collections, as a negative control against genre-driven bias.
• I2 (world): In the ontology, keep a type label (e.g. "supernatural being") separate from per-entity attributes and rank, so sharing a hypernym never forces sharing properties.
• I3 (answer): When a reader decides whether a text shows some category (e.g. a conflict, a trade), require a checklist of frame-level components instead of keyword triggers, and report which ones were found.
• I4 (frames): Treat comparative-exclusion constructions ("no X like Y," "you alone") as uniqueness assertions over a class, and pass them to the world model as constraints instead of reading them as denials that other X exist.
• I5 (frames): In word-sense selection, score candidate senses for coherence with the other assertions already in the text's world, and prefer the sense that creates no contradiction.