The Song of Moses (Deut 32): one of Israel's oldest texts?

What the poem is doing

The Song of Moses opens with a courtroom-style summons to the heavens and the earth to listen, then proceeds through a long account of YHWH's care for Israel, Israel's apostasy, YHWH's anger, and a final note that YHWH will vindicate his people and judge their enemies. Inside the poem, verse 8 sets up the picture that gets the most attention: when the Most High gave the nations their inheritance, when he separated the sons of humanity, he set the boundaries of the peoples 'according to the number of...' something. The text-critical question is what 'something' the original verse named.

The Masoretic Hebrew text, the medieval Hebrew standard, reads 'according to the number of the sons of Israel' (Bene Yisrael). The Septuagint, the Greek translation made in Alexandria around the third century BCE, reads 'according to the number of the angels of God' (angelōn theou) in most manuscripts and 'sons of God' (huiōn theou) in others. The Dead Sea Scroll fragment 4QDeut-j, copied in the late second or early first century BCE, reads 'sons of God' (Bene Elohim). The difference is consequential. The Masoretic reading produces a verse about how God divided the nations to match the number of Jacob's descendants. The Qumran and Septuagint reading produces a verse about how the Most High allotted each nation to a heavenly being, retaining Jacob (Israel) as his own portion (verse 9).

Around the text-critical question sits the broader dating question. Linguists who specialize in Hebrew poetry have argued since the late 1940s that the Song of Moses contains grammatical features (specific verb conjugations, prepositional uses, archaic morphology) that cluster with the Song of the Sea (Exodus 15) and the Song of Deborah (Judges 5), texts widely held to be among the oldest preserved Hebrew. Other readings see those same features as either deliberately archaizing or as part of a late composition that draws on older poetic conventions.

The textual witnesses at 32:8

Deuteronomy 32:8 has six significant textual witnesses, each preserving a slightly different reading. The variants line up in a pattern: earlier and non-Jewish witnesses tend to preserve the 'sons of God' or 'angels of God' reading, while the Masoretic Hebrew tradition and texts dependent on it preserve 'sons of Israel.' Whether the Bene Elohim reading is the original (with Bene Yisrael as a later harmonization) is the central text-critical question. Most modern critical editions (BHS, BHQ) accept Bene Elohim as the older reading.

The textual pattern is clear: the older non-Jewish witnesses (the Septuagint, behind which stands a Hebrew Vorlage earlier than the second century BCE) and the Qumran Hebrew tradition preserve 'sons of God' or 'angels of God.' The Masoretic Hebrew tradition, the Samaritan Pentateuch, the second-century Greek revisions of Aquila and Symmachus, and Jerome's Vulgate all preserve 'sons of Israel.' Most modern critical editions of the Hebrew Bible (BHS, BHQ) print 'sons of Israel' in the text but note the variant prominently. The NRSV (1989) and the ESV (2001, in a footnote) shifted to 'sons of God' or 'sons of the gods' on the basis of the Qumran evidence.

Timeline of the dating debate

The linguistic argument

The case for an archaic date for Deut 32 rests primarily on features of Hebrew morphology, syntax, and vocabulary. Specific verb forms (yiqtol used as preterite, the imperfect with -un endings), prepositional uses ('with' meaning 'from' in some constructions), divine titles (El, Eloah, El Elyon, Rock), and parallelism patterns. These features cluster, in the corpus of preserved Hebrew, in a small number of poems: Genesis 49, Exodus 15, Numbers 23-24 (the Balaam oracles), Deuteronomy 32, Deuteronomy 33, Judges 5, and parts of 2 Samuel 22 / Psalm 18. Cross and Freedman argued these constitute the oldest layer of Hebrew literature.

The counter-argument has two main forms. The first is that ancient features can be deliberately imitated; later poets, drawing on a tradition of antique-sounding poetic Hebrew, could produce work that looks ancient without being ancient. The second is methodological: the corpus of pre-monarchic Hebrew is so small that the markers used to identify it are necessarily defined by the very poems whose dates are being argued. Young, Rezetko, and Ehrensvärd's 2008 two-volume study sets out this case at length. Defenders of the early date answer that the clustering of features is too consistent to be archaizing imitation, and that the comparison with Ugaritic and other second-millennium West Semitic literature provides external controls.

The argument is genuinely technical and most non-specialists find it hard to evaluate directly. What can be said is that the linguistic case for archaic poetry has been on the table for seventy years, has been repeatedly defended (Sanders 1996, Pat-El and Wilson-Wright 2013), and remains contested. No consensus has emerged. The most that can be said with confidence is that the Song of Moses uses Hebrew that does not look like the Hebrew of Jeremiah or Ezra. Whether that points to genuinely older composition or to deliberately archaic style is the open question.

How the three positions read the divine council in 32:8-9

The archaic-dating reading (Cross, Freedman, Sanders, Smith) takes the Bene Elohim text at 32:8 as preserving a stage of Israelite thought in which YHWH was conceived as head of a divine council, with each nation under the supervision of a heavenly being while Israel was YHWH's own portion. On this reading, the poem reflects an early Israelite theology that later texts (Psalm 82, Daniel 10) preserve in similar form, and that later editing in the Masoretic tradition smoothed toward a strict monotheism. The Bene Yisrael reading would be a later harmonization, removing the picture of subordinate heavenly beings.

The pre-exilic monarchic reading (Mayes, Tigay) accepts the Bene Elohim reading as text-critically prior but treats its theology as compatible with monarchic-period Yahwism. The divine-council framework was a live category in 8th-century prophetic literature (Isa 6, 1 Kgs 22:19-22), so a monarchic-period poem could use the same vocabulary without being pre-monarchic. The poem's covenant-lawsuit form, on this reading, places it alongside Hosea and Micah rather than alongside the Song of the Sea.

The post-exilic codification reading (Otto, Römer) takes the Bene Elohim reading as preserving an older tradition incorporated into the late composition. The Persian-period editors did not invent the divine-council vocabulary; they preserved it from earlier material they were integrating. The poem in its present form belongs to the post-exilic period, but the divine-council layer it preserves is older. The Masoretic Bene Yisrael reading is then either a still-later harmonization or, in some versions of the position, the original reading that the divine-council tradition glossed.

What each position has to account for

The archaic-dating position has to explain how a 12th- or 11th-century poem reached the prose framework of Deuteronomy, how it was preserved across centuries, and why its theological vocabulary fits comfortably alongside later canonical material (Psalm 82, Daniel 10) without obvious updating. Defenders generally answer that liturgical use preserved the poem, that the Deuteronomic editors deliberately incorporated old material to give the book canonical depth, and that the canonical compatibility is evidence the poem stood within a continuous Yahwistic tradition.

The pre-exilic monarchic position has to explain why the linguistic features of the poem cluster with the Song of the Sea and the Song of Deborah more than with monarchic-period prophetic literature, and why the Bene Elohim reading would have been incorporated into a monarchic Hebrew text that, on independent grounds, was already moving toward stricter Yahwism. Defenders generally answer that prophetic and poetic registers retain conservative features and that the divine-council framework remained available in formal poetry even when prose theology had moved on.

The post-exilic codification position has to explain how the poem's language could imitate or preserve so consistently the features of pre-monarchic Hebrew poetry, and why other passages in the same period (e.g., parts of Daniel, Esther, Chronicles) do not show the same density of archaic features. Defenders generally answer that the Persian-period editors were preserving an older substrate rather than composing freshly, and that the archaic features are integral to that substrate.

Reading the Song with the question open

The Song of Moses is at the center of Deuteronomy's farewell scene. Moses, on the last day of his life, teaches Israel a poem they are to learn by heart, so that when the predicted apostasy comes, the poem will testify against them. The poem itself reads as both indictment and promise: Israel will fail, the LORD's anger will fall, but the LORD will not finally abandon his portion. That structure is in the poem regardless of which dating position the reader holds.

What changes with the dating position is the texture of the voice. On the archaic reading, the poem comes from the earliest stage of Israelite poetry, embedded in the late material of Deuteronomy as an inheritance from the period of the judges or earlier. The voice is ancient and the framework that holds it is later. On the monarchic reading, the poem belongs to the prophetic chorus of the 9th-7th centuries, speaking the same indictment Hosea and Micah speak. On the post-exilic reading, the poem belongs to a community in the Persian period that preserved older divine-council vocabulary as part of its account of how Israel ended up scattered among the nations. Three different historical contexts produce three different ways of hearing the same words. The words themselves are unchanged.