Core

A subject-initial parse with berēʾšît as nominal subject is grammatically licit yet distributionally under-attested within the currently surveyed Biblical Hebrew corpus. More specifically, it combines a licensed S-initial frame with a licensed but distributionally under-attested coordination profile. The Core distinguishes grammar from theology; reception-historical and doctrinal implications are summarized separately (Appendix E). Replication notes (Appendix F) and working CSVs allow readers to rerun and refine the searches.

Scope

This study is strictly philological. We do not use gematria, numerology, or hidden-code methods. Our argument rests on publicly checkable evidence from Biblical Hebrew grammar (syntax, morphology, word order; use of ’et; asyndeton), textual history (unpointed vs. Masoretic vocalization; comparison with ancient versions), onomastics (how prepositional/adverbial material becomes proper names), and discourse features in Gen 1:1–3. Reception history (e.g., LXX, Targum, Megillah 9a) is cited descriptively, not as proof of intent. Theological reflections (Word/Wisdom/Logos) are presented after the linguistic case and clearly distinguished from it.

Replication/limits (HB only): No exact parallels to the Gen 1:1 object chain were found; the coordination shape is distributionally under-attested in the present sweep—pending wider corpora and alternative annotation.

Keywords: Genesis 1:1; Biblical Hebrew syntax; ’et-marking; coordination/asyndeton; onomastics; subject-initial clauses.

Terminology note: We use ’et-marked / ’et-unmarked for the presence or absence of the direct-object marker את. When discussing corpus tools (e.g., SHEBANQ/BHSA), we use annotated, tagged, or labeled for features assigned by the database. When discussing Gen 1:1’s syntax, we use distributionally under-attested for patterns that are licensed in Biblical Hebrew but comparatively infrequent in the surviving corpus.

Introduction

Genesis 1:1 opens the Hebrew Bible with a clause whose first word—בראשית (berēʾšît)—is syntactically contested and exegetically consequential. In the unpointed script of the oldest witnesses, the form can be construed in more than one way. Later Masoretic vocalization regularized a particular reading; major ancient translations (e.g., the Septuagint) likewise favored clarity. This study asks a narrow, linguistic question: Does Biblical Hebrew usage allow Genesis 1:1 to be read with berēʾšît as a nominal subject without violating the language’s attested clause patterns and coordination behavior?

Because Genesis 1:1 stands at the head of Scripture, clarifying this grammatical possibility matters for how v.1 relates to vv.2–3 and for how later interpretive traditions are assessed.

Orientation note. We treat the Bible as an ancient Hebrew corpus, not a rulebook: our claims rest on attested patterns, not on what “must be” true in principle. Cf. Delitzsch (on Gen 1 as divine disclosure) and A. B. Davidson (on revelation presupposed); our concern here is grammatical possibility, not intent.

Our approach is descriptive and corpus-based. First, we survey subject-initial clauses in narrative prose—especially waw + NP + qatal and S + participle/copular frames—to establish that clauses beginning with a nominal subject are a licensed but non-default option in Biblical Hebrew (see Appendix A — S-initial Clauses). Second, we analyze object-series behavior: direct-object marking with ’et, distribution of waw within lists (every item vs. final-only), and asyndeton across series, in order to situate the sequence אֱלֹהִים, אֶת־הַשָּׁמַיִם, וְאֵת הָאָרֶץ within broader coordination practice (see Appendix B (v4) — Unified Table: Lists, ’et-marking, and Asyndeton). Third, we note that Hebrew onomastics allows some preposition-initial or phrase-like forms to function nominally (e.g., בְּצַלְאֵל and בְּסוֹדְיָה), which keeps a nominal value for berēʾšît within grammatical plausibility (see Appendix C — Onomastics).

The minimal claim of the paper is modest: Biblical Hebrew grammar does not preclude a clause that begins with a nominal subject; such clauses are amply attested and serve recognizable discourse functions (topic switch, scene entry, evaluation). The testable claim is specific: the object-series in Gen 1:1 exhibits a licensed but distributionally under-attested coordination profile. Its individual components are attested separately—mixed ’et-marking of other kinds, final-only waw, and early asyndeton—even though no exact parallel to the full Gen 1:1 fingerprint has yet been located.

To avoid category mistakes, we separate translation practice from hermeneutical rules in a brief sidebar (LXX/Targum models vs. Hillel/Ishmael). We also bracket theological identification in the core argument. Demonstrating a grammatical possibility does not adjudicate theological identity; reception-historical and interfaith implications are summarized elsewhere and kept distinct from the clause audit itself.

Falsifiability. The proposal would be undermined by: (i) a robust demonstration that subject-initial clauses of the relevant shapes are vanishingly rare or illusory; (ii) evidence that Gen 1:1’s coordination profile has no parallels in BH prose; or (iii) strong onomastic arguments that prepositional/adverbial bases cannot function nominally. See Appendix F, “Box F.1 — Falsifiable Counter-Reads (How this proposal could fail).” Conversely, additional corpus matches for the coordination pattern or further S-initial attestations increase plausibility, not certainty.

Method — Substitution / Minimal-Pair Diagnostic

Purpose. Isolate the syntactic question by replacing the contested first NP with an uncontroversial proper name.

Control clause (grammatical frame):

יַעֲקֹב בָּרָא אֱלֹהִים אֵת הַשָּׁמַיִם וְאֵת הָאָרֶץ

“Jacob created elohim, the heavens, and the earth.”

Predictions & checks

1. S-initial licensing. BH narrative allows subject-initial clauses at topic/scene entry → see Appendix A.
2. Object-series behavior. Category אֱלֹהִים may be ’et-unmarked; the definite pair takes ’et with final-only waw → see Appendix B.
3. Coordination profile. Mixed ’et + final-only waw + an asyndetic break between items 1–2 → attested variants (Appendix B).

Result. The frame is fully grammatical. What remains is onomastic/philological: can בְּרֵאשִׁית function as a name/nominal subject in that frame?

Replication status (updated). Two-pass HB sweep: Pass 1 cataloged 329 verses (348 total instances) of the template ’et X … wᵉ’et Y. Pass 2 tightened the apposition detector (head-term + repeated ’et names/titles) and annotated a Definiteness basis (see Appendix B, “Legend”) per item (article / suffix / proper-name / quantifier / bare) by checking wider clause context. Result: still no exact parallel to INDEF → DEF(’et) → DEF(’et) with final-only waw and asyndeton 1→2 in a single simple DO chain. Borderline cases and negatives are logged in Appendix B (v4), “Extended Diagnostics & Sweeps” (with tables, near-misses, and notes). Working CSVs (extended_sweep_working_table.csv; extended_sweep_second_pass.csv) and supporting scripts are available via the website Downloads menu.

SHEBANQ replication note. A SHEBANQ-based replication workflow, extended through export-to-CSV and custom clause/verse aggregation (Appendix F), yielded a manageable review set of candidate verses in Genesis. Only a small hand-reviewed subset proved materially relevant to Appendix B’s object-chain question.

Replication/limits. Under our current corpus and criteria we find no exact parallels for the Gen 1:1 object-chain; we therefore characterize its coordination profile as distributionally under-attested in the present corpus, while avoiding absolute claims pending wider corpora and alternative annotation.

Falsifier. If S-initial frames of the relevant type or mixed-marking lists prove vanishingly rare/illusory, the diagnostic fails.

Distributional status. At present, the specific coordination shape (INDEF → DEF[’et] → DEF[’et] with final-only waw in a single DO chain) appears distributionally under-attested in the present corpus. Near-misses exist (e.g., Exod 1:11 apposition; Deut 12:6 mixed ’et with suffixal definites; Ps 146:6 predicate list with final ’et-cluster), but no exact parallel has yet been located in the present sweep.

Onomastic note. For focused preposition-initial name evidence, see Appendix C — Onomastics: בְּצַלְאֵל and בְּסוֹדְיָה. The point is nominal plausibility in principle, not a claim about historical segmentation.

Word terminology (pointer). For a brief, non-doctrinal clarification of “Word” terminology, see Appendix E, “Promise Spoken, Written, Enacted — A Working Analogy for ‘Word’” (illustrative only; not part of the grammatical evidence).

Methodological Posture: Grammar as Descriptive, Not Prescriptive

We do not treat the HB as an infallible rulebook of grammar; we treat it as a corpus whose usage constrains plausible analysis. Absence of a pattern from the surviving corpus does not, by itself, render that pattern “ungrammatical,” but positive attestation raises confidence. Our procedure is therefore: (1) state the pattern precisely; (2) search for independent parallels; (3) weight by genre/register; (4) keep claims calibrated to the distribution actually observed.

Minimal Semantics & Discourse Guardrails (non-theological)

Purpose. Keep only the constraints needed to rule out impossible readings; leave theology to Appendix E.

1. Verb selection (ברא). baraʾ selects a volitional agent as subject. A proper name/title (Bereshit) can satisfy this; a mere temporal phrase (“in-the-beginning”) cannot.
2. Object series fit. Treat אֱלֹהִים in Gen 1:1—if object—as a category/common NP (“gods/powers”), and הַשָּׁמַיִם… הָאָרֶץ as a definite pair. Do not read the pair as an appositive gloss on אֱלֹהִים (“elohim, namely the heavens and the earth”); those sets are not co-extensive.
3. Coordination plausibility. Mixed ’et-marking + final-only waw + an asyndetic break between items 1–2 is licensed but distributionally under-attested, with individual components attested separately (see Appendix B). Under-attested ≠ ungrammatical.
4. Discourse coherence (Gen 1:1–3). A subject shift from v. 1 to v. 3 (from Bereshit → Elohim) is normal in BH narrative (topic switch/scene entry); nothing forces v. 1’s subject to persist.

Outcome. If a reading clears (1)–(4), it is semantically admissible; adjudicating intent remains outside the Core.

Homography Note: אלהים across Gen 1 (object vs. subject)

Purpose. Flag how the same written form אֱלֹהִים functions with two semantic values in our analysis—(a) a common plural (“gods/powers”) as a direct object in Gen 1:1, and (b) the proper singular Deity (“Elohim”) as subject from 1:3 onward—without violating Biblical Hebrew morpho-syntax.

1) The lexeme.
אֱלֹהִים is morphologically m.pl.. In BH it is homographic:

• Plural common noun “gods/powers” (takes plural agreement/adjectives when it’s the subject; as an object, agreement is irrelevant).

• Singular proper “God” (the Deity), typically with singular verbal agreement when it’s the subject.

2) Diagnostics used here.

• Subject-side agreement. Where אֱלֹהִים is subject in Gen 1 (e.g., וַיֹּאמֶר אֱלֹהִים, וַיַּרְא אֱלֹהִים), the verbs are 3ms, signaling proper-singular Deity usage.

• Object-side marking. In Gen 1:1, our test parse treats אֱלֹהִים as the first item in a three-object coordination: (INDEF) אֱלֹהִים · אֶת הַשָּׁמַיִם · וְאֵת הָאָרֶץ. Lack of article/suffix/’et on the first item supports a category/common reading (“gods/powers”), not the proper singular.

• Apposition check. We do not read “אֱלֹהִים, namely the heavens and the earth.” Semantically, “gods/powers” ≠ cosmic domains, so the pair cannot be appositive renamings of אֱלֹהִים (see Appendix B, Apposition exemplars).

3) Discourse shift (v.1 → v.3).

A shift from object in v.1 (category gods/powers, created) to subject in v.3 (proper Elohim, the Deity acting) is ordinary at the discourse level: BH narratives frequently pivot topics between adjacent clauses. Agreement patterns (3ms subject verbs from v.3 on) cue the reader that the Deity is now the acting subject, distinct from any created “powers.”

4) Construct phrases (e.g., רוח אלהים).

Constructs with אֱלֹהִים are resolved by context: when surrounding verbal agreement and discourse mark Deity, X-of-Elohim is read as “of God.” This is independent of the object-reading in 1:1; the construct itself does not force plural “gods.”

5) Bottom line.

The homograph is resolved locally by agreement (on subjects) and by definiteness/’et-marking (on objects). Thus, אֱלֹהִים can be a created category object in 1:1 and the creating Deity from 1:3 onward, without grammatical contradiction or theological claims inside the Core.

Why this proposal is easy to overlook

Genesis 1:1 has long been received through stabilized public forms—ancient translation, Masoretic vocalization, and inherited interpretive habits—that foreground Elohim as subject. In addition, Biblical Hebrew subject-initial clauses are licensed but non-default, and the Gen 1:1 coordination profile remains distributionally under-attested in the presently surveyed corpus. For those reasons, the reading tested here is easily passed over even if it is not grammatically excluded. That historical and interpretive background helps explain the proposal’s neglect; it does not decide the syntax. The grammatical question therefore remains the same: does the clause, read on its own linguistic terms, permit berēʾšît as a nominal subject? On the evidence assembled here, the answer is yes—cautiously, and with explicit distributional limits.

Answer to the Guiding Question (concise)

Permissible but distributionally under-attested. A subject-initial parse with berēʾšît as nominal subject does not violate Biblical Hebrew clause licensing (see Appendix A on S-initial frames). Its coordination profile—an ’et-unmarked/category first object followed by two definite ’et-marked objects with final-only waw and an asyndetic 1→2 break—remains distributionally under-attested in BH prose: we found no exact triad match in two corpus sweeps (Appendix B), only a set of near-misses. Apposition cues (“X—namely Y and Z”) are absent in Gen 1:1, and the semantics of אֱלֹהִים (“gods/powers”) vs. הַשָּׁמַיִם / הָאָרֶץ (cosmic domains) argues against apposition. Conclusion: the reading is grammatically licit yet distributionally under-attested in the present corpus. This report confines itself to grammatical possibility and observed frequency; adjudicating among competing interpretations lies outside its scope.