Claude-skill-registry example-sentences

Requirements for creating and revising example sentences in je-dict-1. Covers minimum counts, progressive length, vocabulary restrictions by tier, and quality standards.

install
source · Clone the upstream repo
git clone https://github.com/majiayu000/claude-skill-registry
Claude Code · Install into ~/.claude/skills/
T=$(mktemp -d) && git clone --depth=1 https://github.com/majiayu000/claude-skill-registry "$T" && mkdir -p ~/.claude/skills && cp -r "$T/skills/data/example-sentences" ~/.claude/skills/majiayu000-claude-skill-registry-example-sentences && rm -rf "$T"
manifest: skills/data/example-sentences/SKILL.md
source content

Example Sentence Guidelines

This skill defines the standards for example sentences across all dictionary entries. All entry-creation and revision activities must follow these guidelines.


Minimum Example Counts by Tier

Basic and Core Tiers

Every sense of every entry in the basic and core tiers must have at least 5 example sentences.

TierMinimum Examples per Sense
Basic5
Core5

General Tier

Every sense of every entry in the general tier must have at least 3 example sentences.

TierMinimum Examples per Sense
General3

Multi-sense entries: The requirements apply per sense. An entry with 3 senses in the basic tier needs at least 15 examples total (5 per sense).


Progressive Length Requirement

Examples within each sense should progress from shorter to longer:

  1. Example 1: Short and simple - demonstrates the word clearly
  2. Example 2: Slightly longer - shows basic context
  3. Example 3: Medium length - natural usage with fuller context
  4. Example 4 (basic/core): Longer sentence with more complex structure
  5. Example 5 (basic/core): Longest - may consist of 2-3 short sentences or one longer sentence with relative clauses, compound structures, etc.

Key principle: Early examples help learners recognize the word; later examples show how the word functions in natural, flowing Japanese.

Length Guidelines

ExampleTarget Length (Japanese)
15-15 characters
210-20 characters
315-30 characters
425-45 characters
5+35-70 characters (or 2-3 short sentences)

These are guidelines, not strict rules. Natural expression takes priority over hitting exact character counts.


Vocabulary Restrictions by Tier

Basic Tier Entries

For entries in the basic vocabulary tier:

ExamplesVocabulary Restriction
1-2Basic tier only - Use only words from the basic tier
3-5Basic + Core tiers - May use basic and core vocabulary

Critical: Basic tier examples must contain no general-tier or unlisted vocabulary. This ensures learners at the basic level can fully understand all examples without encountering unknown words.

Core Tier Entries

For entries in the core vocabulary tier:

ExamplesVocabulary Restriction
1-2Basic + Core tiers - Use only basic or core vocabulary
3-5No restriction - May use any vocabulary

Recommendation: Even for unrestricted examples (3-5), avoid vocabulary that is not in the dictionary. Learners should be able to look up any unfamiliar word.

General Tier Entries

For entries in the general vocabulary tier:

ExamplesVocabulary Restriction
1-3+No restriction - May use any vocabulary

Recommendation: Avoid obscure vocabulary not in the dictionary. Use common, natural expressions that learners are likely to encounter.


Vocabulary Tier Verification

When creating examples for basic and core tier entries, verify vocabulary tier compliance:

# Check if a word is in the dictionary and its tier
python3 build/check_duplicate.py "word" "reading"

For systematic verification during polishing:

  • Read the example sentence
  • Identify all content words (nouns, verbs, adjectives, adverbs)
  • Verify each word is in the appropriate tier
  • Replace non-compliant words with tier-appropriate alternatives

Common substitution patterns:

Non-compliantBasic alternativeCore alternative
{購入こうにゅう}する{買
{使用しよう}する{使
{非常ひじょう}にとても
{困難こんなん}{難

Sense Numbers Requirement

Every example must include a

sense_numbers
field linking it to the definition(s) it illustrates.

"examples": [
  {
    "id": "00001_word_ex1",
    "japanese": "...",
    "english": "...",
    "sense_numbers": [1],
    "has_audio": false,
    "notes": null
  }
]

Rules

  • Single-sense entries: Use
    [1]
    for all examples
  • Multi-sense entries: Each example must specify which sense(s) it demonstrates
  • Examples illustrating multiple senses: Use
    [1, 2]
    format
  • Must reference valid senses: Numbers must match
    sense_number
    values in definitions

Distribution for Multi-Sense Entries

For entries with multiple senses in basic/core tiers:

  • Each sense needs at least 5 examples
  • Distribute examples to cover all senses adequately
  • Some examples may illustrate multiple senses (use
    [1, 2]
    notation)

Quality Standards

Essential Requirements

  1. Natural Japanese: Examples should sound like something a native speaker would say
  2. Clear context: The usage should be understandable from the example
  3. Complete furigana: All kanji must have
    {kanji|reading}
    markup
  4. Accurate translation: English captures meaning without being overly literal

Content Quality

  1. First example should be simple - Demonstrate the word clearly without complexity
  2. Progress to natural complexity - Later examples show real-world usage
  3. Include fixed phrases - High-frequency collocations aid memory (at least one per sense)
  4. Show grammatical connections - Demonstrate how words connect with particles, etc.
  5. Annotate non-obvious grammar - Use
    notes
    field for explanations when needed

What to Demonstrate

Examples should illustrate:

  • Common collocations and set phrases
  • Typical particle patterns (for verbs)
  • Predicate vs. modifier usage (for adjectives)
  • Register-appropriate contexts
  • Both literal and figurative meanings (if applicable)

Example Format Reference

{
  "id": "00396_taberu_ex1",
  "japanese": "{朝|あさ}ごはんを{食|た}べる。",
  "english": "To eat breakfast.",
  "sense_numbers": [1],
  "has_audio": false,
  "notes": null
}

ID Format

Pattern:

{entry_id}_ex{N}

  • entry_id
    : The entry's ID (e.g.,
    00396_taberu
    )
  • N
    : Sequential number starting from 1

Fields

FieldRequiredDescription
id
YesUnique example ID
japanese
YesJapanese sentence with furigana
english
YesNatural English translation
sense_numbers
YesArray of sense numbers illustrated
has_audio
YesBoolean (usually
false
)
notes
YesExplanatory note or
null

Common Issues to Avoid

  1. Adding inline links: Do NOT add inline word links (⟦...⟧) when creating or revising examples. Links are added in a separate polishing step via
    prompts/polish_add_inline_links.md
    .
  2. Missing furigana:
    お金が余った
    instead of
    お{金|かね}が{余|あま}った
  3. Overly literal translations: "As for me, I eat sushi" instead of "I eat sushi"
  4. Wrong sense_numbers: Example linked to wrong definition
  5. Unnatural constructions: Grammatically correct but rarely used patterns
  6. Vocabulary tier violations: Using general-tier words in basic-tier examples 1-2
  7. Insufficient examples: Not meeting minimum counts per sense
  8. No length progression: All examples being similar length

Polishing Checklist for Examples

When reviewing or revising examples, verify:

Count and Distribution

  • Minimum examples per sense met (5 for basic/core, 3 for general)
  • All senses have adequate example coverage
  • Examples are properly distributed across senses

Progressive Length

  • Examples progress from shorter to longer within each sense
  • Final examples are substantially longer or multi-sentence

Vocabulary Restrictions

  • Basic tier: Examples 1-2 use only basic vocabulary
  • Basic tier: Examples 3-5 use only basic+core vocabulary
  • Core tier: Examples 1-2 use only basic+core vocabulary
  • No unlisted vocabulary in restricted examples

Quality

  • All examples sound natural
  • All kanji have complete furigana
  • Translations are accurate and natural
  • sense_numbers are correct for each example
  • At least one example shows a common collocation
  • Grammar patterns are demonstrated clearly

Format

  • IDs follow
    {entry_id}_ex{N}
    pattern
  • IDs are sequential (ex1, ex2, ex3...)
  • All required fields present
  • has_audio
    is accurate
  • notes
    field present (null if not needed)

Adding Examples to Existing Entries

When adding examples to bring an entry into compliance:

  1. Read the existing entry to understand current examples
  2. Identify gaps: Which senses need more examples? What lengths are missing?
  3. Check vocabulary tier: Determine what vocabulary can be used
  4. Write new examples following progressive length guidelines
  5. Verify compliance with all requirements
  6. Update
    modified
    timestamp
    in metadata

Example Addition Template

For a basic-tier entry with one sense needing 5 examples:

Example 1: [short, basic vocab only]
Example 2: [short-medium, basic vocab only]
Example 3: [medium, basic+core vocab]
Example 4: [medium-long, basic+core vocab]
Example 5: [long or multi-sentence, basic+core vocab]

Summary Table

TierMin Examples/SenseEx 1-2 VocabEx 3+ Vocab
Basic5Basic onlyBasic + Core
Core5Basic + CoreAny (prefer dictionary words)
General3AnyAny (prefer dictionary words)