Nominal Erasure — Adonai and Lilit as Case Studies

Public source text: WLC (Westminster Leningrad Codex) + Nestle 1904. Translation: Belem-2025 Bible translation – literal, rigid, straight from the public códices.

What is nominal erasure?

When you read “Lord” in a Portuguese Bible, which original designation is behind it? It could be Yahweh (יהוה — yhwh; trad. “Jehovah”¹) (יהוה), it could be Adonai (אדני), it could be Adoni (אדני with hiriq). Three ontologically distinct designations compressed into a single word: “Lord.”

When you read “screech owl” in Isaiah 34:14 in the KJV, or “night creatures” in the NIV, or “night ghost” in other versions — what is behind it is a proper feminine name: לִּילִ֔ית — Lilit. Erased. Replaced. Invisible.

This is nominal erasure: the replacement of a proper name or specific designation with a generic term in translation, resulting in loss of referential information. The reader not only receives a different translation — they lose the ability to identify WHO or WHAT the original text names.

The numbers: 441,649 tokens scanned

To investigate this phenomenon, we performed an exhaustive computational scan of the Cloudflare D1 database of the Belem-2025 Bible translation — all 441,649 tokens from the 66 canonical books.

Result: two case studies that reveal the same mechanism operating at radically different scales.

Case Study 1: Adonai — 855 tokens leveled to “Lord”

The Hebrew designation Adonai (אדני) occurs in 855 tokens, distributed across 771 verses and 32 books of the Old Testament. There are at least 6 distinct morphological variants.

The 10 books with the most occurrences

Ezekiel concentrates 27.9% of all occurrences — almost exclusively in the construction Adonai Yahweh (yhwh) (אדני יהוה). This compound form appears ~217 times in the OT, and the forensic question emerges: why does Ezekiel insist on Adonai Yahweh (yhwh) while Isaiah and Jeremiah predominantly use Yahweh (yhwh) alone?

The vowel taxonomy: an editorial decision

The lexicon of Brown, Driver & Briggs (1906) distinguishes two consonantally identical forms:

The critical datum: both share the same consonantal skeleton א-ד-נ-י. The difference lies EXCLUSIVELY in the Masoretic vowels — added in the 7th-10th century AD. The text that the prophets wrote contains only אדני, without vowels. The divine/human classification was ADDED by the Masoretic editors.

The trifusion: three designations, one word

The tripartite confusion becomes visible in Psalm 110:1 (WLC), where Yahweh (yhwh) and Adoni coexist —

נְאֻ֤ם יְהוָ֨ה לַֽאדֹנִ֗י שֵׁ֥ב לִֽימִינִ֑י עַד־אָשִׁ֥ית אֹ֝יְבֶ֗יךָ הֲדֹ֣ם לְרַגְלֶֽיךָ

“Declaration of Yahweh (yhwh) (יְהוָה) to my lord (אדֹנִי): Sit at my right hand, until I make your enemies a footstool for your feet.” — Psalm 110:1

Traditional Portuguese translations do this:

Three ontologically distinct designations → one single Portuguese word. Differentiated only by typographic conventions that the common reader does not decode.

Easter Egg #1: Psalm 110:1 — “Declaration of Yahweh (yhwh) to my adoni: sit at my right hand.” The Masoretic form has Adoni (with hiriq — “human” classification), not Adonay (“divine” classification). However, the NT quotes this verse applying it to Christos (Mt 22:44, Acts 2:34, Heb 1:13) — treating it as a divine reference. The contradiction: the Masoretes classified the referent as human; NT authors as divine. By translating everything as “Lord,” translations conceal this tension.

Case Study 2: Lilit — the absolute hapax legomenon

A complete scan of 441,649 tokens returned exactly 1 match: Isaiah 34:14, position 12 of 15 tokens.

1 occurrence in ~31,100 verses. Maximum rarity. Absolute hapax legomenon.

The verse: Isaiah 34:14

Masoretic Text:

וּפָגְשׁ֤וּ צִיִּים֙ אֶת־ אִיִּ֔ים וְשָׂעִ֖יר עַל־ רֵעֵ֣הוּ יִקְרָ֑א אַךְ־ שָׁם֙ הִרְגִּ֣יעָה לִּילִ֔ית וּמָצְאָ֥ה לָ֖הּ מָנֽוֹחַ

Rigid literal translation (Belem AnC):

“And tsiim met with iyyim; and a sa’ir upon his companion called; indeed, there Lilit rested and found for herself repose.”

Quadruple evidence of feminine gender

The morphology leaves no room for doubt:

Lilit is a feminine being. The verbal, pronominal, and nominal agreement is unequivocal.

The erasure: no Portuguese translation preserved the name until 2025

Easter Egg #2: ACF and NIV translate in the PLURAL (“night animals,” “night creatures”) — erasing the morphological singularity. The Hebrew has a singular form. A singular feminine entity becomes a neutral plural concept. The LXX, already in the 3rd-2nd century BC, did not recognize the name: by translating as ονοκενταυρος (onocentaur), the Alexandrian translators reveal that Lilit’s meaning was already obscure — or deliberately avoided — two centuries before Christ.

The sa’ir network: quantified context

Lilit does not appear alone in Isaiah 34:14. In the same sentence is the sa’ir (שָׂעִ֖יר). We mapped the entire network:

The 6 domains

The ENTITIES domain, although representing only 4% of tokens, concentrates all the critical forensic verses. And the most disturbing datum:

Easter Egg #3: All 4 verses in the ENTITIES domain present translation errors in the database — an error rate of 100%. Offset errors (pt_literal contains the next Hebrew word instead of the translation) and lexical errors (שָׁם/sham = “there” confused with שֵׁם/shem = “name”). Automated translation fails systematically in precisely the most critical contexts.

The intertextual pattern: ruins inhabited by entities

The pattern “empire falls → entities inhabit ruins” appears three times in the corpus, forming an OT-OT-NT chain:

Lilit appears exclusively in the Edomite setting, not in the Babylonian. The se’irim appear in both. This territorial exclusivity is forensic data: why is Lilit specific to Edom/Seir?

Easter Egg #4: DES 18:2 replicates exactly the structure of Isaiah 13 and 34: a destroyed city/empire becomes the habitation of spiritual entities. The same formula, separated by ~700 years of composition. Sa’ir is translated as “goat” or “hairy one.” Lilit as “owl.” Daimonion as “demon.” When you translate everything generically, the intertextual connection breaks.

The counter-argument — and its failure

The traditional argument for erasure is accessibility: translating “Lilit” as “owl” makes the text more understandable. The same for “Adonai” → “Lord.”

This argument fails for two reasons:

1. Presupposition of meaning. Translating “Lilit” as “owl” implies that the translators KNOW that Lilit = owl. But the LXX translates as “onocentaur,” the Vulgate as “lamia,” the ACF as “night animals” (plural). The disagreement demonstrates that nobody knows what Lilit is — and replacing the unknown with a generic term is not translation, it is concealment.

2. Asymmetry of treatment. Proper names like “Jerusalem,” “Moses,” and “Elijah” are systematically transliterated. Nobody translates “Jerusalem” as “the holy city” or “Moses” as “the one drawn from the waters.” The principle should be the same for Lilit and Adonai.

Conclusion: the erasure is not accidental — it is a pattern

The data from the computational scan sustain:

1. Adonai (855 tokens, 32 books): designation with complex vowel taxonomy, uniformized to “Lord” in all traditional translations, merged with Yahweh (yhwh) and deprived of its referential identity.

2. Lilit (1 token, 1 verse): proper feminine name with quadruple morphological evidence of gender, erased throughout the entire history of Bible translation in Portuguese until 2025.

3. Nominal erasure is not the exception — it is the pattern. It has operated since the LXX (3rd-2nd century BC) and persists in all contemporary translations.

4. The sa’ir network (100 tokens, 6 domains): verses in the ENTITIES domain — the most forensically significant — present a 100% error rate, suggesting systemic failure in the translation pipeline.

5. Rigid literalness returns to the reader the information that the original text contains. The Belem-2025 Bible translation is the first translation in the Portuguese language to adopt systematic transliteration for Adonai and Lilit.

The underlying philosophy:

“You read. And the interpretation is yours.”