Hebrew OT Information Structure¶

Information structure describes how a clause packages information as given vs. new, topic vs. comment, and foregrounded vs. backgrounded. This notebook quantifies several proxy metrics that are reliably computable from MACULA morphological data:

Metric	What it measures
Parataxis ratio	Proportion of verses beginning with waw-consecutive or wayyiqtol
Hypotaxis/1k	Subordinating connectives (כִּי, אֲשֶׁר, לְמַעַן, etc.) per 1,000 tokens
Fronted ratio	Proportion of verses with non-verb-initial clause type
Nominal clause %	Verbless clauses as % of all verses (proxy for copula-free sentences)
Inf. construct/1k	Infinitive construct density (subordinate purpose/time clauses)

Limitation: Full topic/focus analysis requires syntactic annotation beyond what MACULA provides. These metrics are approximations.

References:

• Heimerdinger, Topic, Focus and Foreground in Ancient Hebrew Narratives (1999)
• Longacre, Joseph: A Story of Divine Providence (1989)
• IBHS Chapters 7–8 (clause types and hierarchy)

In [ ]:

import sys
sys.path.insert(0, '../../../src')

from bible_grammar import (
    ot_information_profile, ot_clause_linking_comparison,
    print_ot_information_profile, print_ot_clause_linking_comparison,
    ot_clause_linking_chart,
)
import pandas as pd

import sys sys.path.insert(0, '../../../src') from bible_grammar import ( ot_information_profile, ot_clause_linking_comparison, print_ot_information_profile, print_ot_clause_linking_comparison, ot_clause_linking_chart, ) import pandas as pd

1. Overview — Single Book Profile¶

In [ ]:

# Genesis: highly paratactic narrative
print_ot_information_profile('Gen')

# Genesis: highly paratactic narrative print_ot_information_profile('Gen')

In [ ]:

# Deuteronomy: legal/homiletical, should be more hypotactic
print_ot_information_profile('Deu')

# Deuteronomy: legal/homiletical, should be more hypotactic print_ot_information_profile('Deu')

In [ ]:

# Psalms: poetry, high nominal clause rate expected
print_ot_information_profile('Psa')

# Psalms: poetry, high nominal clause rate expected print_ot_information_profile('Psa')

2. Parataxis vs. Hypotaxis — Narrative vs. Law vs. Poetry¶

Hebrew narrative (Genesis, Kings) is predominantly paratactic — clauses chained by waw-consecutive. Legal texts (Leviticus, Deuteronomy) and poetry (Psalms, Job) use more subordination.

In [ ]:

# Torah comparison
print_ot_clause_linking_comparison(['Gen', 'Exo', 'Lev', 'Num', 'Deu'])

# Torah comparison print_ot_clause_linking_comparison(['Gen', 'Exo', 'Lev', 'Num', 'Deu'])

In [ ]:

ot_clause_linking_chart(['Gen', 'Exo', 'Lev', 'Num', 'Deu'])

ot_clause_linking_chart(['Gen', 'Exo', 'Lev', 'Num', 'Deu'])

In [ ]:

# Genre comparison: narrative / wisdom / prophecy
genre_sample = ['Gen', '1Sa', '2Ki', 'Psa', 'Pro', 'Job', 'Isa', 'Jer', 'Eze']
print_ot_clause_linking_comparison(genre_sample)

# Genre comparison: narrative / wisdom / prophecy genre_sample = ['Gen', '1Sa', '2Ki', 'Psa', 'Pro', 'Job', 'Isa', 'Jer', 'Eze'] print_ot_clause_linking_comparison(genre_sample)

In [ ]:

ot_clause_linking_chart(genre_sample)

ot_clause_linking_chart(genre_sample)

3. Fronted Elements — Topic and Focus¶

Non-verb-initial clauses signal that something is fronted (pre-posed) for topic or focus. High fronted-ratio suggests contrastive/emphatic discourse. Wisdom literature (Proverbs, Psalms) and legal texts tend to have more fronted nominal elements than fast-moving narrative.

In [ ]:

df = ot_clause_linking_comparison(genre_sample)
df[['total_verses', 'fronted_ratio', 'nominal_clause_pct']].sort_values(
    'fronted_ratio', ascending=False
)

df = ot_clause_linking_comparison(genre_sample) df[['total_verses', 'fronted_ratio', 'nominal_clause_pct']].sort_values( 'fronted_ratio', ascending=False )

4. Nominal Clauses — Verbless Sentences¶

Biblical Hebrew regularly omits the copula (to be) in present-tense assertions. Wisdom, praise (Psalms), and instructional genres have high nominal-clause rates; narrative has low.

In [ ]:

all_books = [
    'Gen', 'Exo', 'Lev', 'Num', 'Deu', 'Jos', 'Jdg',
    '1Sa', '2Sa', '1Ki', '2Ki', 'Job', 'Psa', 'Pro',
    'Ecc', 'Isa', 'Jer', 'Eze', 'Dan', 'Amo', 'Jon'
]
df_all = ot_clause_linking_comparison(all_books)
df_all[['total_verses', 'parataxis_ratio', 'nominal_clause_pct', 'hypotaxis_per1k']].sort_values(
    'nominal_clause_pct', ascending=False
).head(15)

all_books = [ 'Gen', 'Exo', 'Lev', 'Num', 'Deu', 'Jos', 'Jdg', '1Sa', '2Sa', '1Ki', '2Ki', 'Job', 'Psa', 'Pro', 'Ecc', 'Isa', 'Jer', 'Eze', 'Dan', 'Amo', 'Jon' ] df_all = ot_clause_linking_comparison(all_books) df_all[['total_verses', 'parataxis_ratio', 'nominal_clause_pct', 'hypotaxis_per1k']].sort_values( 'nominal_clause_pct', ascending=False ).head(15)

5. Isaiah 1–39 vs. 40–66 — Clause-Linking Structure¶

Isaiah 40–66 (Deutero-Isaiah) is known for its elevated style. Does it also differ in parataxis/hypotaxis ratios from Proto-Isaiah?

In [ ]:

from bible_grammar._utils import load_ot_data
from bible_grammar.information_structure import ot_information_profile

isa_df = load_ot_data()
isa_h = isa_df[(isa_df['book'] == 'Isa') & (isa_df['lang'] == 'H')]

# Temporarily monkey-patch to profile each half
isa1_39 = isa_h[isa_h['chapter'] <= 39].copy()
isa40_66 = isa_h[isa_h['chapter'] >= 40].copy()

def _profile_raw(df, label):
    from bible_grammar.information_structure import (
        _SUBORDINATING_LEMMAS, _NON_VERB_INITIAL_TYPES
    )
    total = len(df)
    verse_firsts = df.groupby(['book', 'chapter', 'verse']).first().reset_index()
    total_verses = len(verse_firsts)
    parataxis = int(
        (verse_firsts['type_'] == 'wayyiqtol').sum() +
        verse_firsts['lemma'].isin({'וְ', 'וּ', 'וַ'}).sum()
    )
    subordinating = df['lemma'].isin(_SUBORDINATING_LEMMAS).sum()
    fronted = int(verse_firsts['type_'].isin(_NON_VERB_INITIAL_TYPES).sum())
    verbal_types = {'wayyiqtol', 'qatal', 'yiqtol', 'imperative', 'cohortative',
                    'jussive', 'participle active', 'participle passive',
                    'infinitive construct', 'infinitive absolute'}
    verse_has_verb = df.groupby(['book', 'chapter', 'verse'])['type_'].apply(
        lambda s: s.isin(verbal_types).any()
    )
    nominal = int((~verse_has_verb).sum())
    return {
        'label': label, 'total_verses': total_verses,
        'parataxis_ratio': round(parataxis / max(total_verses, 1), 4),
        'hypotaxis_per1k': round(subordinating / total * 1000, 2),
        'fronted_ratio': round(fronted / max(total_verses, 1), 4),
        'nominal_pct': round(nominal / max(total_verses, 1) * 100, 2),
    }

comparison = pd.DataFrame([
    _profile_raw(isa1_39, 'Isa 1–39'),
    _profile_raw(isa40_66, 'Isa 40–66'),
]).set_index('label')
comparison

from bible_grammar._utils import load_ot_data from bible_grammar.information_structure import ot_information_profile isa_df = load_ot_data() isa_h = isa_df[(isa_df['book'] == 'Isa') & (isa_df['lang'] == 'H')] # Temporarily monkey-patch to profile each half isa1_39 = isa_h[isa_h['chapter'] <= 39].copy() isa40_66 = isa_h[isa_h['chapter'] >= 40].copy() def _profile_raw(df, label): from bible_grammar.information_structure import ( _SUBORDINATING_LEMMAS, _NON_VERB_INITIAL_TYPES ) total = len(df) verse_firsts = df.groupby(['book', 'chapter', 'verse']).first().reset_index() total_verses = len(verse_firsts) parataxis = int( (verse_firsts['type_'] == 'wayyiqtol').sum() + verse_firsts['lemma'].isin({'וְ', 'וּ', 'וַ'}).sum() ) subordinating = df['lemma'].isin(_SUBORDINATING_LEMMAS).sum() fronted = int(verse_firsts['type_'].isin(_NON_VERB_INITIAL_TYPES).sum()) verbal_types = {'wayyiqtol', 'qatal', 'yiqtol', 'imperative', 'cohortative', 'jussive', 'participle active', 'participle passive', 'infinitive construct', 'infinitive absolute'} verse_has_verb = df.groupby(['book', 'chapter', 'verse'])['type_'].apply( lambda s: s.isin(verbal_types).any() ) nominal = int((~verse_has_verb).sum()) return { 'label': label, 'total_verses': total_verses, 'parataxis_ratio': round(parataxis / max(total_verses, 1), 4), 'hypotaxis_per1k': round(subordinating / total * 1000, 2), 'fronted_ratio': round(fronted / max(total_verses, 1), 4), 'nominal_pct': round(nominal / max(total_verses, 1) * 100, 2), } comparison = pd.DataFrame([ _profile_raw(isa1_39, 'Isa 1–39'), _profile_raw(isa40_66, 'Isa 40–66'), ]).set_index('label') comparison

6. Pentateuch — Law vs. Narrative Sections¶

Compare all five Torah books. Leviticus and Deuteronomy should show lower parataxis (less narrative) and higher hypotaxis (more subordinate clauses in legal formulas).

In [ ]:

df_torah = ot_clause_linking_comparison(['Gen', 'Exo', 'Lev', 'Num', 'Deu'])
df_torah[['total_verses', 'parataxis_ratio', 'hypotaxis_per1k', 'nominal_clause_pct', 'inf_construct_per1k']]

df_torah = ot_clause_linking_comparison(['Gen', 'Exo', 'Lev', 'Num', 'Deu']) df_torah[['total_verses', 'parataxis_ratio', 'hypotaxis_per1k', 'nominal_clause_pct', 'inf_construct_per1k']]

7. Pedagogical Note — LXX Translation Tendencies¶

The LXX often renders Hebrew parataxis (waw-consecutive) as Greek hypotaxis (participle constructions, subordinate clauses). The high parataxis ratios in narrative books explain why direct translation into Greek required significant syntactic restructuring.

In [ ]:

# Books with highest parataxis ratio — these are the ones most affected by LXX restructuring
df_all.sort_values('parataxis_ratio', ascending=False).head(10)[[
    'total_verses', 'parataxis_ratio', 'hypotaxis_per1k'
]]

# Books with highest parataxis ratio — these are the ones most affected by LXX restructuring df_all.sort_values('parataxis_ratio', ascending=False).head(10)[[ 'total_verses', 'parataxis_ratio', 'hypotaxis_per1k' ]]