NormForm (v0.1 — Experimental)

NormForm is a Python toolkit and command-line interface for context-aware Norwegian morphological standardization, alternative generation, and style auditing in both Bokmål (nob) and Nynorsk (nno).

Written Norwegian officially permits wide intra-norm morphological and orthographic variation—such as feminine vs. common gender nouns (boka / boken, klokka / klokken), weak preterite endings (kasta / kastet), infinitive endings in Nynorsk (bruka / bruke, gjera / gjere), and lexical spelling variants (fram / frem, sjukehuset / sykehuset, fyrste / første). NormForm lets you:

1. Standardize text or large corpora to a consistent sub-norm style (such as Bokmål Redaksjonsnorm, Nynorsk Nasjonalbiblioteknorm, or your own custom style profile).
2. Generate [alt1|alt2] morphological alternative brackets for speech recognition (ASR) and text generation evaluation with altmetrics and altwer.
3. Audit documents and corpora for internal morphological inconsistencies or rank the exact morphological changes made across a standardized dataset.
4. Extract empirical style prompt words from any reference document using information-theoretic submodular maximization.
5. Share, fork, and version style profiles as lightweight YAML repositories on Hugging Face (1 repo = 1 style).

Table of Contents

1. What NormForm Does — and What It Does NOT Do
2. Installation
3. Quick Start (CLI & Python API)
4. Feature Guide
   • 4.1 Standardizing Text & Documents (normform standardize)
   • 4.2 High-Throughput 1-GPU Batch Corpus Standardization (normform batch)
   • 4.3 Generating [alt1|alt2] Morphological Alternatives (normform alternatives)
   • 4.4 Auditing Inconsistencies & Change Statistics (normform audit)
   • 4.5 Extracting Steering Style Words (normform extract-style)
   • 4.6 Managing & Customizing Styles (normform styles)
5. Style File Format (style.yaml)
6. Official Models & Style Repositories
7. License & Credits

1. What NormForm Does — and What It Does NOT Do

NormForm is built around a strict non-destructive lattice contract (normform.contract). The neural model is never allowed to freely rewrite your text. Instead, every sentence is parsed into a verified template of literal spans and [alt1|alt2] candidate slots (Template.align), and only validated candidate slots can change.

What NormForm DOES:

• Chooses consistently among valid intra-norm alternatives in context: e.g., in Bokmål Redaksjonsnorm, selecting jenta and boka for concrete feminine nouns, klokken for klokka | klokken, kastet for weak preterites (kasta | kastet), and utafor / sjukehuset for lexical variants; or in Nynorsk Nasjonalbiblioteknorm, selecting -a infinitives (bruka, gjera, vera), vi, fyrste, and sundagen.
• Guarantees byte-for-byte preservation of all non-alternative text: punctuation, whitespace, line breaks, numbers, dates, URLs, email addresses, proper names, and out-of-vocabulary words are never altered.
• Preserves source token casing automatically:
  • Lowercase: jenten $\rightarrow$ jenta
  • Titlecase (sentence-initial): Jenten $\rightarrow$ Jenta
  • Uppercase: JENTEN $\rightarrow$ JENTA
  • Mixed internal capitals: vebb-TV $\rightarrow$ web-TV
• Protects homographs and semantic collisions (invalid_pairs): When two surface forms can overlap in inflection tables but differ in meaning or grammatical role (e.g., verken | verket "neither" vs. "the pain/work", bad | badet, slott | slåss), the style profile's invalid_pairs guard forces NormForm to preserve the original token unchanged.
• Maintains document- and corpus-level consistency: Once a candidate group (boka | boken) is resolved—either from a reviewed style cache or by the neural style model—that exact decision is cached and applied consistently across every occurrence.

What NormForm Does NOT Do:

• Does NOT translate between Bokmål and Nynorsk: Specify --lang nob for Bokmål text and --lang nno for Nynorsk text.
• Does NOT fix general typos, grammar errors, or punctuation: Words that are not valid Norwegian forms with morphological alternatives are passed through untouched.
• Does NOT paraphrase, reorder words, or insert/delete tokens: Every output sentence is guaranteed to align 1-to-1 with the source sentence's literal template.
• Does NOT rewrite syntactic agreement across neighboring tokens: NormForm substitutes individual morphological word slots ([boka|boken]). It does not rewrite surrounding articles or adjectives outside the candidate slot (e.g., if a source phrase has ei lita bok, individual word slots are only changed where each slot's own paradigm allows it).
• Does NOT use brittle suffix-stripping heuristics on compounds: Novel compound words absent from the seed cache are either resolved in full sentential context by the neural model or preserved unchanged when --preserve-unreviewed is enabled.

Concrete Examples: What Changes vs. What Stays Untouched

2. Installation

pip install normform

# Or install with batch dataset extras (Parquet, Hugging Face Datasets, progress bars):
pip install "normform[batch]"

Requirements:

• Python $\ge 3.10$ (tested on Python 3.12 with PyTorch $\ge 2.0$ and Transformers $\ge 4.40$).
• A CUDA-capable GPU is strongly recommended for model inference and corpus batch processing (--device cuda:0 or --device auto). CPU execution is supported for small texts and --raw audits.
• No API keys required: The pinned neural models and official style repositories on Hugging Face are public and downloaded automatically on first use.

3. Quick Start (CLI & Python API)

Command Line (normform)

# Check version
normform --version

# 1. Standardize a Bokmål sentence using the official Redaksjonsnorm style:
normform standardize -t "Jenten kasta boken og så på klokka utenfor sykehuset." --lang nob --style redaksjonsnorm
# -> Jenta kastet boka og så på klokken utafor sjukehuset.

# 2. Standardize a Nynorsk sentence using Nasjonalbiblioteknorm:
normform standardize -t "Me ønskjer å bruke den første søndagen til å gjere dette." --lang nno --style nasjonalbiblioteknorm
# -> Vi ynskjer å bruka den fyrste sundagen til å gjera dette.

# 3. Generate [alt1|alt2] morphological alternative brackets:
normform alternatives -t "Katta ligger på matta." --lang nob
# -> [Katta|Katten] ligger på [matta|matten].

# 4. Check a document for internal sub-norm inconsistencies (without modifying it):
normform audit -i manuscript.txt --lang nob --raw

Python API

from normform import Standardizer

# Initialize with a built-in style alias, Hugging Face repo ID, or local .yaml file:
std = Standardizer(
    lang="nob",
    style="redaksjonsnorm",       # or "NbAiLab/normform-bokmaal-redaksjonsnorm"
    preserve_unreviewed=False,    # True = only apply pre-reviewed cache entries
    device="auto",
)

# 1. Standardize text (preserves casing, punctuation, and layout):
text = std("Jenten kasta boken og så på klokka utenfor sykehuset.")
print(text)
# -> "Jenta kastet boka og så på klokken utafor sjukehuset."

# 2. Generate [alt1|alt2] morphological alternative brackets:
bracketed = std.alternatives("Katta ligger på matta.")
print(bracketed)
# -> "[Katta|Katten] ligger på [matta|matten]."

# 3. Check a text or document for internal sub-norm inconsistencies:
issues = std.audit("Boka lå på bordet. Senere leste jeg boken.", raw=True)
for issue in issues:
    print(issue["candidate_group"], "->", issue["first_occurrence"]["choice"], "vs", issue["inconsistent_occurrence"]["choice"])
# -> ['boka', 'boken'] -> Boka vs boken

4. Feature Guide

4.1 Standardizing Text & Documents (normform standardize)

Use normform standardize for interactive strings, plain-text documents (.txt), or small JSONL files (.jsonl).

Choosing Your Standardization Mode

NormForm supports three ways to control how morphological choices are made:

Examples

# Standardize a .txt file with Redaksjonsnorm and write to output.txt:
normform standardize -i article.txt -o article_standardized.txt --lang nob --style redaksjonsnorm

# Strict reviewed-only mode (never invoke the style model on unseen compounds;
# only apply the 33,221 human/linguist-reviewed choices in the style cache):
normform standardize -i article.txt -o article_standardized.txt \
  --lang nob --style redaksjonsnorm --preserve-unreviewed

# Standardize toward a conservative Bokmål profile without a style file by supplying steering prompt words:
normform standardize -t "Jenta kasta boka på hylla." \
  --lang nob -p "jenten,kastet,boken,hyllen,frem"
# -> Jenten kastet boken på hyllen.

# Persist choices across multiple runs in a local JSON cache and write a Markdown summary:
normform standardize -i chapter1.txt -o chapter1_norm.txt \
  --lang nob --style redaksjonsnorm \
  --cache project_choices.json --summary-output chapter1_report.md

4.2 High-Throughput 1-GPU Batch Corpus Standardization (normform batch)

For large datasets (tens of thousands to millions of records in .jsonl, .parquet, directory shards, or Hugging Face datasets), use normform batch.

How the 3-Stage Batch Engine Works

1. Background I/O Prefetch & Sentence Splitting: A background producer thread streams records from disk or Hugging Face and splits text into whitespace-preserving sentence spans so disk/network I/O overlaps with GPU compute.
2. Stage 1 — Length-Bucketed GPU Alternative Expansion: Non-blank sentences in each chunk (--chunk-size 4096) are sorted by character length to minimize padding waste, expanded in large GPU batches (--batch-size 512), and validated once with clean_direct + parse + Template.align. Over-capacity outlier sentences are automatically isolated via binary bisection and preserved unchanged.
3. Stage 2 — Greedy Set-Cover & Context-Style Resolution:
   • All candidate groups already present in your --style cache are resolved in $O(1)$ memory lookup with zero extra GPU calls.
   • If unseen candidate groups appear in the chunk (and --preserve-unreviewed is not set), NormForm runs Greedy Set-Cover (greedy_set_cover) to pick the minimum subset of sentences needed to cover all unseen candidate groups, runs adapter.style_batch() on just those sentences, and caches the resulting choices for the rest of the corpus.
4. Stage 3 — Async Streaming Rewrite & Sealed Cache Export: Sentences are rendered using the updated cache, reassembled into records, and written asynchronously to .jsonl, .parquet (Zstandard-compressed), or .txt. At completion, NormForm exports a human-editable *_sealed_cache.yaml top-sorted by actual corpus change impact (# seen: ..., changed: ..., already_correct: ..., change_rate: ...%).

Batch Examples

# 1. Standardize a large JSONL corpus on GPU 0 using Bokmål Redaksjonsnorm:
normform batch \
  -i /path/to/corpus.jsonl \
  -o /path/to/corpus_standardized.jsonl \
  --lang nob \
  --style NbAiLab/normform-bokmaal-redaksjonsnorm \
  --text-field text \
  --output-field normalized \
  --batch-size 512 \
  --chunk-size 4096 \
  --device cuda:0

# 2. Standardize a Parquet file or directory of Parquet shards in strict reviewed-only mode:
normform batch \
  -i /path/to/parquet_shards/ \
  -o /path/to/standardized.parquet \
  --lang nno \
  --style NbAiLab/normform-nynorsk-nasjonalbiblioteknorm \
  --preserve-unreviewed \
  --sealed-cache-output /path/to/nynorsk_run_sealed_cache.yaml

# 3. Stream directly from a Hugging Face dataset repository:
normform batch \
  -i hf://NbAiLab/npsc_orto \
  --dataset-split train \
  -o npsc_standardized.jsonl \
  --lang nob \
  --style redaksjonsnorm \
  --limit 10000

Inspecting the Sealed Cache (*_sealed_cache.yaml) & Iterative Refinement

Every normform batch run automatically writes a *_sealed_cache.yaml file organized by grammatical word-class category (and by root stem family for lexical/stem variants), and sorted within each category/family by actual corpus change impact (changed descending, then seen descending):

_metadata:
  language: nob
  sort_order: grammatical_category_and_stem_family_then_frequency
choices:
  # =========================================================================
  # 2. VERB — Weak Preterite & Past Participle (-a / -et / -d / -de / -te)
  # =========================================================================
  skada | skadd | skadet: skadet  # [VERB.Past/PerfPart] seen: 1563, changed: 1420 (skadd: 1410, skada: 10), already_correct: 143, change_rate: 90.85%
  kasta | kastet: kastet  # [VERB.Past/PerfPart] seen: 739, changed: 20 (kasta: 20), already_correct: 719, change_rate: 2.71%

  # =========================================================================
  # 4. NOUN — Feminine / Common Definite Singular (-a / -en)
  # =========================================================================
  boka | boken: boka  # [NOUN.Fem.Sing.Def] seen: 1842, changed: 1410, already_correct: 432, change_rate: 76.5%
  klokka | klokken: klokken  # [NOUN.Fem.Sing.Def] seen: 540, changed: 488, already_correct: 52, change_rate: 90.4%

  # =========================================================================
  # 12. LEXICAL & STEM VARIANTS (Grouped by Root Family, then Frequency)
  # =========================================================================
  # --- Stem family: fram / frem (525 entries, changed: 6,100, seen: 31,332) ---
  fram | frem: fram  # [STEM:fram/frem] seen: 20464, changed: 1445 (frem: 1323, Frem: 122), already_correct: 19019, change_rate: 7.06%
  framtida | framtiden | fremtida | fremtiden: framtida  # [STEM:fram/frem + NOUN.Fem.Sing.Def] seen: 1707, changed: 1180, already_correct: 527, change_rate: 69.13%

  # --- Stem family: sjuk / syk (631 entries, changed: 1,975, seen: 17,008) ---
  sjuk | syk: syk  # [STEM:sjuk/syk] seen: 520, changed: 45, already_correct: 475, change_rate: 8.65%
  sjukehuset | sykehuset: sjukehuset  # [STEM:sjuk/syk] seen: 1502, changed: 562 (sykehuset: 552, Sykehuset: 10), already_correct: 940, change_rate: 37.42%

Recommended Corpus Curation Workflow ("Run $\rightarrow$ Inspect $\rightarrow$ Lock"):

1. Run normform batch on your corpus (or a 10,000-record sample with --limit 10000).
2. Open *_sealed_cache.yaml and review each grammatical category section (VERB — Infinitive, VERB — Weak Preterite & Past Participle, NOUN — Feminine / Common Definite Singular, NOUN — Neuter Indefinite Plural, LEXICAL & STEM VARIANTS) where high-impact choices sit at the top of each section.
3. Edit any choice in *_sealed_cache.yaml (or move a domain-specific collision into invalid_pairs), and re-run with --style my_reviewed_sealed_cache.yaml --preserve-unreviewed for 100% deterministic corpus standardization.

Python API for Batch Standardization

from pathlib import Path
from normform import run_batch_standardization

run_batch_standardization(
    input_source="/path/to/corpus.jsonl",
    output_path=Path("/path/to/corpus_standardized.jsonl"),
    sealed_cache_output=Path("/path/to/sealed_cache.yaml"),
    lang="nob",
    style="redaksjonsnorm",
    preserve_unreviewed=False,
    text_field="text",
    output_field="normalized",
    batch_size=512,
    chunk_size=4096,
    device="cuda:0",
)

4.3 Generating [alt1|alt2] Morphological Alternatives (normform alternatives)

normform alternatives expands each sentence into alphabetically sorted [alt1|alt2] brackets while keeping all non-alternative text unchanged. This output is directly compatible with altmetrics and altwer for ASR/TTS evaluation where any valid spelling in the bracket is accepted as correct.

# Single sentence:
normform alternatives -t "Jenta kasta ballen." --lang nob
# -> [Jenta|Jenten] [kasta|kastet] ballen.

# Whole file (.txt or .jsonl):
normform alternatives -i references.jsonl -o references_with_alternatives.jsonl --lang nob

4.4 Auditing Internal Sub-Norm Consistency & Change Statistics (normform audit)

A very common use case for NormForm is checking whether a document, book, or corpus is internally consistent in its sub-norm choices—without rewriting the text.

For example, an author or transcription dataset might unintentionally mix boka and boken, fram and frem, or kasta and kastet across different paragraphs. Because normform audit compares morphological choices case-insensitively across contextually validated [alt1|alt2] slots, sentence-initial capitalization (Boka vs. boka) is treated as consistent, while actual sub-norm switches (Boka vs. boken) are flagged with exact record and sentence locations.

A. Check a Raw Document or Corpus for Internal Sub-Norm Inconsistencies (--raw)

# Human-readable terminal report of all internal sub-norm inconsistencies:
normform audit -i manuscript.txt --lang nob --raw

Example terminal output:

Found 2 internal sub-norm inconsistencies in manuscript.txt:

  1. [boka | boken]
     - First choice : 'Boka' (record 0, sentence 0: "Boka lå på bordet hele dagen.")
     - Conflict     : 'boken' (record 0, sentence 2: "Senere på kvelden leste jeg boken.")

  2. [fram | frem]
     - First choice : 'fram' (record 1, sentence 0: "Vi gikk fram til scenen.")
     - Conflict     : 'frem' (record 4, sentence 1: "Hun kom frem i tide.")

Export a structured JSON report for automated QA pipelines:

normform audit -i corpus.jsonl --lang nob --raw --json -o inconsistencies.json

Example JSON structure (inconsistencies.json):

[
  {
    "candidate_group": ["boka", "boken"],
    "first_occurrence": {
      "record_index": 0,
      "sentence_index": 0,
      "choice": "Boka",
      "sentence": "Boka lå på bordet hele dagen."
    },
    "inconsistent_occurrence": {
      "record_index": 0,
      "sentence_index": 2,
      "choice": "boken",
      "sentence": "Senere på kvelden leste jeg boken."
    }
  }
]

In Python:

from normform import Standardizer

std = Standardizer(lang="nob")
issues = std.audit("Boka lå på bordet. Jeg leste boka. Senere tok jeg boken.", raw=True)
# Returns the same list of dicts with 'candidate_group', 'first_occurrence', and 'inconsistent_occurrence'

B. Model & Style Consistency Audit (--uncached)

Tests whether the neural model (under a given --style or --prompt) makes conflicting choices for the same candidate group across different sentences when cross-sentence caching is disabled:

normform audit -i evaluation_set.jsonl --lang nob -p "boka,jenta,kasta" --uncached -o audit.json

C. Post-Standardization Corpus Change Statistics (--stats)

Compares the source --text-field and standardized --normalized-field in a processed .jsonl or .parquet corpus and outputs a ranked YAML report of the top morphological groups that caused changes:

normform audit -i corpus_standardized.jsonl --stats --top 50 -o change_report.yaml

Example output:

corpus: corpus_standardized.jsonl
total_records_scanned: 50000
total_records_changed: 14210 (28.42%)
unique_alternative_groups_changed: 842
top_alternatives:
  boka | boken: {chosen: boka, total_changes: 1410, changed_from: {boken: 1410}}
  kasta | kastet: {chosen: kastet, total_changes: 615, changed_from: {kasta: 615}}
  klokka | klokken: {chosen: klokken, total_changes: 488, changed_from: {klokka: 488}}

4.5 Extracting Steering Style Words (normform extract-style)

Given an existing reference document (for example, an editorial style sample or an author's manuscript), normform extract-style discovers the top $N$ single-word morphological forms that best explain the document's empirical style choices via Information-Theoretic Log-Likelihood Submodular Maximization:

# Output full JSON report with Explained Style Variance (%):
normform extract-style -i reference_article.txt --lang nob -n 10

# Output only the comma-separated prompt string for direct piping:
PROMPTS=$(normform extract-style -i reference_article.txt --lang nob -n 10 --prompt-only)
normform standardize -i new_article.txt -p "$PROMPTS" --lang nob

4.6 Managing & Customizing Styles (normform styles)

NormForm decouples code from linguistic styles. Every style profile is a self-contained YAML file (style.yaml) hosted in its own Hugging Face dataset repository (1 repo = 1 style).

# 1. List built-in official styles and short aliases:
normform styles list

# 2. Download an official style to a local YAML file to inspect or customize:
normform styles pull redaksjonsnorm -o my_style.yaml

# 3. Edit my_style.yaml in any text editor, then push to your own HF dataset repo:
#    (Creates a private HF dataset repo by default; add --public to make it public)
normform styles push -f my_style.yaml --repo username/my-normform-style --lang nob

# 4. Use your custom style from anywhere (or pin a specific commit with @<sha>):
normform standardize -i document.txt --style username/my-normform-style
normform batch -i corpus.jsonl --style username/my-normform-style@main

5. Style File Format (style.yaml)

A NormForm style.yaml file has three human-readable sections:

_metadata:
  language: nob
  style_name: my-custom-style
  prompt_words:
    - jenta
    - boka
    - kastet
    - fram
    - sjukehuset

choices:
  # Mapping from "alt1 | alt2" (lowercase, alphabetical) to the chosen lowercase form:
  boka | boken: boka
  fram | frem: fram
  jenta | jenten: jenta
  kasta | kastet: kastet
  klokka | klokken: klokken
  "no | nå": nå

invalid_pairs:
  # Ambiguous or homographic candidate groups that must NEVER be altered:
  bad | badet: Homographic noun/verb inflection; preserve source token.
  verken | verket: Distinct lemmas (conjunction 'verken' vs. noun 'verket').

6. Official Models & Style Repositories

All official models and style repositories are publicly available on Hugging Face under NbAiLab and pinned by commit SHA for reproducibility:

Linguistic Summary of the Official Style Profiles

• Bokmål — NbAiLab/normform-bokmaal-redaksjonsnorm (76,979 reviewed choices, 3,857 invalid_pairs):
  • Feminine nouns: Uses -a for concrete/everyday feminine nouns (boka, jenta, hylla, katta, tida, uka, lista, helga), while selecting -en for specific editorial nouns (klokken, avisen, gruppen, grensen).
  • Verbs: Uses -et in the preterite and past participle of weak -e verbs (kastet, laget, skadet, snakket) and stod (sto | stod).
  • Neuter plurals: Uses short indefinite plural (produkt, resultat, prosjekt, dokument) and -ene in definite plural (husene, årene, barna).
  • Lexical variants: Uses fram / framover / framtida, utafor, innenfor, sju / tjuende, sjukehuset, and syk.
• Nynorsk — NbAiLab/normform-nynorsk-nasjonalbiblioteknorm (14,665 reviewed choices, 1,293 invalid_pairs):
  • Infinitives: Consistent -a infinitive (vera, bruka, gjera, halda, koma, setja, skriva).
  • Pronouns & function words: vi (me | vi), no (no | nå), difor (derfor | difor), blant (blant | millom), ikkje.
  • Nouns, adjectives & verbs: fyrst / fyrste, sundag / sundagen, ynskja / ynskjer, gje / gjev, vart (blei | vart), definite feminine plural -ene (bøkene, jentene, hendene), and vowel alternation in strong/irregular verbs (kjem, fer, held, tek).

7. License & Credits

Licensed under the Apache License 2.0 (see LICENSE).

Created by Per Egil Kummervold and Magnus Breder Birkenes at the National Library of Norway (Nasjonalbiblioteket).