About

Controlled sentence exposure is a core technique in second-language acquisition research. When learners encounter unknown kanji in context, cognitive load spikes and comprehension collapses. This tool solves that problem directly: you declare which kanji you have studied, and it filters a curated corpus to return only sentences composed of those characters. Every sentence uses hiragana, katakana, and punctuation freely, but restricts kanji strictly to your declared set. The corpus covers JLPT N5 through N3 level grammar patterns. Note: the tool approximates natural frequency distribution but cannot guarantee coverage of every reading for a given kanji.

Typical flashcard apps treat kanji in isolation. Reading kanji within grammatical structures activates different neural pathways than single-character recall. If you study 20 kanji but never see them in sentences, your recognition speed under reading conditions will remain low. This generator bridges that gap. Pro tip: start with the N5 preset (80 kanji) and add characters as you learn them. Your known set persists across sessions via local storage.

Formulas

The filtering algorithm determines sentence eligibility using set theory. For a sentence S with extracted kanji set K_s and the user's declared known kanji set K_u:

K_s ⊆ K_u &implies; S is valid

Kanji characters are identified by Unicode code point membership:

c ∈ Kanji ⇔ U+4E00 ≤ c ≤ U+9FFF

Where K_s = the set of all characters in sentence S matching the kanji Unicode range, and K_u = the set of all characters the user has entered. A sentence with zero kanji (pure hiragana/katakana) always passes the filter. The difficulty metric D for each sentence is computed as:

D = |K_s| × len(S)

Where |K_s| is the count of unique kanji in the sentence and len(S) is the total character length. Higher D indicates a more complex sentence.

Reference Data

JLPT Level	Kanji Count	Example Kanji	Approx. Vocab	Reading Ability
N5	80	日月火水木金土人大小	800	Basic signs, simple sentences
N4	170	会社員電車駅病院教室	1,500	Simple articles, daily conversation
N3	370	届届届届届届届届届届	3,750	Newspapers with aid, general topics
N2	1,000	論政経済際環境権制規	6,000	Newspapers, business documents
N1	2,000	鬱璧繭彙顎挨拶曖闇萎	10,000	Academic papers, literature
Jōyō Set	2,136	Full daily-use set	-	Full literacy (newspapers, law)
Grade 1	80	一二三四五六七八九十	-	Age 6-7 school level
Grade 2	160	引羽雲園遠何科夏家歌	-	Age 7-8 school level
Grade 3	200	悪安暗医委意育員院飲	-	Age 8-9 school level
Grade 4	202	愛案以衣位囲胃印英栄	-	Age 9-10 school level
Grade 5	193	圧移因永営衛易益液演	-	Age 10-11 school level
Grade 6	191	異遺域宇映延沿恩我灰	-	Age 11-12 school level
Jinmeiyō	863	Names-only kanji	-	Personal/place name reading
Kyōiku Set	1,026	Grades 1-6 combined	-	Elementary school completion
Newspaper	~3,000	Varies by publisher	-	99% coverage of printed text

Frequently Asked Questions

The corpus contains approximately 500 curated sentences covering common JLPT N5 - N3 vocabulary. If you enter an obscure or N1-level kanji, it may not appear in any sentence in the database. The filter requires that all kanji in a sentence be within your known set - it does not require that every known kanji appear. Think of it as a whitelist, not a shopping list.

The filtering operates purely at the character level, not the reading level. A kanji like 生 has over 10 readings (せい, しょう, い·きる, う·まれる, etc.). The generator will include any sentence containing 生 regardless of which reading it uses. The furigana/reading shown with each sentence indicates the specific reading in that context.

Sentences composed entirely of hiragana, katakana, and punctuation have an empty kanji set K_s. Since the empty set is a subset of every set (∅ ⊆ K_u is always true), these sentences always pass the filter. This is intentional - pure kana sentences are readable by any learner and provide grammar practice.

Yes. The entire sentence corpus is embedded in the JavaScript source. No network requests are made at any point. Your known kanji set is saved to localStorage automatically. Once the page is cached by your browser, it functions fully offline.

The official JLPT does not publish a definitive kanji list. The 80-kanji figure follows the Kyōiku Grade 1 set, which closely overlaps with most N5 study guides. The 103 figure comes from older, unofficial compilations. You can always add or remove individual kanji from any preset to match your specific textbook.

The generator tracks recently shown sentences and avoids repeating them until at least half of the eligible pool has been cycled through. If your known kanji set is very small and only 2-3 sentences match, you will see repeats sooner. Adding more kanji to your set expands the available pool.