Skip to main content

Should You Learn Kanji in Frequency Order, School Order, or Pedagogical Order?

"What order should I learn kanji in" is the first strategic decision a learner faces after kana. The answer is not "the order the next textbook on the shelf happens to use." Three ordering schemes dominate adult kanji curricula. The one a learner picks determines which 500 kanji they see in the first six months, and therefore what they can read.123

Overview

The kanji learning order debate is a comparison of ordering criteria, not methods. The Heisig method, WaniKani, and kanji-via-vocab each have their own treatment elsewhere on J-Compass. This article asks the upstream question: should the first 2,000 kanji follow school grade, corpus frequency, or component build-up?

The three orderings, named

The Japanese school order is the MEXT (Ministry of Education) 学年別漢字配当表 (gakunenbetsu kanji haitō hyō), the per-grade kanji assignment appended to the 小学校学習指導要領 (elementary school curriculum guidelines). It distributes the 1,026 kyōiku kanji across grades 1 to 6 of compulsory primary schooling, then assigns the remaining roughly 1,110 jōyō kanji to junior-high and high-school years.145

Frequency order ranks kanji by occurrence count in a reference corpus. The canonical sources are the Asahi and Yomiuri newspaper surveys26 and the National Institute for Japanese Language and Linguistics (NINJAL) register-balanced BCCWJ.78

Pedagogical order arranges kanji by component build-up: a kanji is introduced only after all of its constituent primitives are known. Heisig's Remembering the Kanji Volume 1 is the textbook case39; WaniKani applies the same logic in software, with spaced repetition and 60 levels that gate radicals, kanji, and vocabulary in sequence.1011

A fourth de facto ordering, JLPT order, is in wide pedagogical use but has no official basis. The post-2010 JLPT publishes no kanji list at any level; every "N5 kanji list," "N4 kanji list," and so on is a community reconstruction from past-paper analysis.12131415

JLPT has no official kanji list

The administering bodies, 国際交流基金 (Japan Foundation) and 日本国際教育支援協会 (Japan Educational Exchanges and Services), deliberately stopped publishing the 出題基準 (Test Content Specification) when the test was redesigned in 2010. Their stated reason was that communicative competence, not item memorization, is the goal. Any "N3 kanji list" is therefore a community-reconstructed estimate from past papers, not a syllabus.12

Why this question even matters

The first roughly 500 kanji a learner sees barely overlap across the three orders. Frequency-order corpora put 国, 人, 年, 会, 本 inside the top 10 of newspaper text2. School order's grade 1 (80 kanji) is dominated by numerals, body parts, school-life nouns, and weekday names45. Heisig's first 50 stays in pictographs and oracle-bone-style graphics (一二三四五六七八九十口日月田目古吾冒…).16

The cumulative-coverage curve is steep early. Roughly the top 500 kanji buy about 80% of Asahi 1993 kanji tokens6; the top 1,000 buy about 90% (Kanō, via Kandrac)1718; the top 1,600 buy about 99% of Asahi 19936; and the full jōyō (2,136) buys about 99% of 2006 Asahi plus Yomiuri.2 A frequency-ordered first 500 therefore unlocks much more readable text than a school-ordered or component-ordered first 500.

What this article is not

This is not a method review. The Heisig method, WaniKani, and kanji-via-vocab are evaluated elsewhere on J-Compass as methods in the strategic overview of how to learn kanji; this article compares the orderings those methods imply when a learner picks them. The choice of method comes after the choice of order.

Japanese school order (grade 1 to 6 jōyō)

What the order actually is

The kyōiku kanji list is the set of kanji assigned to Japanese elementary school by MEXT in the 学年別漢字配当表, an appendix to the 小学校学習指導要領.14 As of the 2017 revision (in force from 2020), the list contains 1,026 characters distributed as Grade 1 = 80, Grade 2 = 160, Grade 3 = 200, Grade 4 = 202, Grade 5 = 193, Grade 6 = 191.45

The remaining roughly 1,110 jōyō kanji (the 2,136-character 常用漢字表 set under the 2010 Cabinet Notification) are taught in junior-high (中学校) and high-school (高等学校) years under the same MEXT framework.1920

The 2017 revision (平成29年告示, fully implemented in 2020) added 20 prefectural-name kanji to grade 4 only, so that students can write all 47 prefectural names by the end of fourth grade. The 20 newly added kanji are 茨, 媛, 岡, 潟, 岐, 熊, 香, 佐, 埼, 崎, 滋, 鹿, 縄, 井, 沖, 栃, 奈, 梨, 阪, 阜.121 The same revision shifted four kanji from grade 5 down to grade 4 (賀, 群, 徳, 富) and one kanji from grade 6 down to grade 4 (城), giving grade 4 a full set of 25 prefecture-related kanji on top of those already in earlier grades.121

Ordering inside a grade is driven by teaching needs and children's life context, not by corpus frequency. Grade 1's 80 kanji include numerals 一 through 十, body parts 口・耳・手・足, weekday and nature words 日・月・火・水・木・金・土, and school-life nouns 校・学・先・生.45

四年生よねんせい都道府県とどうふけん漢字かんじならいます。121
"In fourth grade, students learn the kanji for the prefectures."

Who it actually fits

School order fits learners who have a practical reason to mirror native schooling. Parents reading children's books with their child track the kyōiku list directly, because children's-book kanji follow it by convention. JET ALTs supporting elementary classes need the same sequence the classroom is teaching. Adult learners using Japanese-school edutainment material (くもん workbooks, NHK children's programming captions) benefit from the same alignment.45

The order also fits learners preparing for the 漢字検定 (Kanken) at levels 10 through 5, which test the kyōiku kanji per grade as their assignment.5

What it gets wrong for adult learners

The order optimizes for a Japanese child's life context, not for an adult learner's reading. Grade 1's 80-kanji set burns early slots on 校 (school), 円 (yen, anchored by the coin), 王 (king), 玉 (jewel or ball), 犬 (dog), 虫 (insect), and 貝 (shellfish). All of these are useful, but none are top-frequency in newspaper or general adult text.467

The 私 problem

私 is classified grade 6 by MEXT45, but it ranks among the highest-frequency kanji in adult Japanese prose. Both BCCWJ jōyō analyses and Asahi rankings place it in the very top tier.67 A learner who follows school order strictly does not write or fluently read 私 until late in the kyōiku sequence, even though it appears constantly in any adult text.

The 2017 prefectural-kanji addition is itself an example of school order's life-context bias. The 20 new kanji (茨, 媛, 岡, 潟, …) were added not because they rose in adult-prose frequency but because Japanese fourth-graders need to write Japan's 47 prefectures.121 For a non-resident adult learner with no need to write Japan's prefectures, those 20 sit in grade 4 ahead of higher-frequency kanji that appear later.

Frequency order (corpus rank)

What the order actually is

Frequency order ranks each kanji by its number of occurrences, or tokens, in a reference corpus. The canonical corpora are Asahi newspaper full-year 1993 (Chikamatsu et al. 2000; about 23 million kanji tokens, 4,476 distinct types)6; the Bunkachō 2007 newspaper survey of Asahi plus Yomiuri 2006 sampling2; the 2000 Bunkachō 385-book corpus (33.3 million kanji tokens, 8,474 types)22; and NINJAL's BCCWJ analyzed by Joyce, Masuda & Ogawa 2014.78

Different corpora produce different orderings, especially in the long tail of less-common kanji. Kandrac 2022 compares six major frequency databases and finds inter-database scatter, with "almost a quarter of all kanji from [Yatskov's Wikipedia frequency report] and almost one-fifth from [the Kanji Database]" deviating by more than 300 rank positions from the database average; the divergence is concentrated in less-frequent characters.17

The top 10 of the 2006 Asahi plus Yomiuri newspaper corpus (Bunkachō 2007) begins with 日, 年, 大, 人, 国, 会, 本, 中. At rank 9, Asahi has 者 and Yomiuri has 月; at rank 10, Asahi has 一 and Yomiuri has 事.2 日 is the top kanji in both corpora, not 一.

日本にほん国会こっかいおおきな会議場かいぎじょうです。2
"Japan's National Diet is a large assembly hall."

Frequency depends on register, meaning the type of language being sampled. Newspapers cluster in formal-register kanji and concentrate on news topics, so jōyō (2,136) buys about 99% of newspaper kanji tokens.2 BCCWJ's register-balanced mix (books, magazines, web, white papers, law) puts jōyō at 96.12% of kanji tokens.7 Aozora Bunko (pre-1946-reform literature, much of it pre-jōyō) raises the rank-to-coverage requirement at every threshold.23

Where the order pays off fast

The cumulative-coverage curve rises steeply at the beginning. The top roughly 500 buy about 80% of Asahi 1993 kanji tokens (Chikamatsu)6. The top roughly 1,000 buy about 90% (Kanō, via Kandrac)1718. The top roughly 1,600 buy about 99% of Asahi 1993.6 The full jōyō (2,136) buys about 99% of 2006 Asahi plus Yomiuri.2 A frequency-ordered first 500 unlocks substantially more readable text per kanji than any other order.

Scriptin's interactive Wikinews cumulative curve corroborates the shape: about 45% at rank 100, about 72% at rank 300, about 96% at rank 1,000.23

The coverage payoff is front-loaded

The diminishing-returns shape means each added high-frequency kanji gives less new coverage than the one before it. Hitting jōyō completion adds the final 1% of newspaper coverage at roughly thirty times the per-kanji effort of the first 500. Frequency order is the only ordering that respects this curve from the start.2617

The Kodansha Kanji Learner's Course as the named instance

The Kodansha Kanji Learner's Course (KKLC), by Andrew Scott Conning (Kodansha USA, 2013), is the cleanest in-print example of a frequency-aware, learner-curated ordering. The course indexes 2,300 characters: all 2,136 jōyō (including the 196 added by the 2010 revision) plus 164 high-frequency non-jōyō entries.2425

Conning describes the KLC sequence as a "rational learning sequence" that introduces kanji "in rough order of importance" while also "presenting graphically related characters one after the other to help learners give significance to their contrastive features as they learn them, and thereby avoid having to relearn them later."25 The order is led by frequency, with a graphical-similarity overlay. It is not pure rank.

KLC entry #1 is 日 (sun, day), entry #2 is 一 (one), and the early entries are dominated by high-frequency adult-prose kanji rather than the numeral-first opening of Heisig and school order.2526

Pure-frequency decks (Anki "Kanji 2k" community decks, JPDB's frequency-ranked exposure) implement the same axis without KLC's editorial graphical-similarity tweaks.27

What it gets wrong

Frequency order does not account for component logic. Rank #45 may share no radicals with rank #44 or #46. Each new kanji can become an isolated visual puzzle, and the learner cannot build on a previously learned component.259

Frequency also varies by corpus (news vs. literature vs. web). A learner whose target is literary fiction but who trains on a news-corpus frequency list will hit a long tail of fiction-specific hyōgaiji, or non-jōyō kanji, not covered by that ranking.177

The first roughly 50 kanji in a pure frequency list include visually complex adult-vocabulary kanji (国, 会, 議, 経, 政, 義). They demand stroke-count and component effort the learner has not yet built up, a load that pedagogical order explicitly avoids.29

Pedagogical order (component build-up)

What the order actually is

Pedagogical order presents kanji so that every component (primitive or radical) needed for kanji N has already been introduced before N. Heisig's Remembering the Kanji Volume 1 (6th edition, 2011) is the canonical textbook implementation: it orders about 2,200 kanji so that "a kanji is introduced only after all of its primitives have been," with primitives including traditional Kangxi radicals, kanji repurposed as components, and stroke clusters Heisig names himself.39

WaniKani implements the same logic in software, with spaced repetition and 60 levels. Each level teaches radicals first, then the kanji built from those radicals, then the vocabulary using those kanji; a kanji is not studyable until all radicals it depends on are at Guru stage in the spaced-repetition system.101128

Heisig's first 50 frames (per the published RTK Volume 1 6th-edition order, retransmitted by community indices)163 are:

FrameKanjiHeisig keywordFrameKanjiHeisig keyword
1one26early
2two27rising sun
3three28generation
4four29stomach
5five30nightbreak
6six31gall bladder
7seven32span
8eight33concave
9nine34convex
10ten35olden days
11mouth36oneself
12day37white
13month38hundred
14rice field39in
15eye40thousand
16old41tongue
17I42measuring box
18risk43rise up
19companion44round
20bright45measurement
21chant46specialty
22sparkle47Dr.
23goods48fortune telling
24spine49above
25prosperous50below

WaniKani's level 1 radical-then-kanji set starts with single-stroke and two-stroke radicals (ground, fins, gun, sticks, leaf) and the kanji 一, 二, 三, 上, 下, 大, 工, 八, 入, and so on. WaniKani's own gating rules document the radical-to-kanji unlocks.1011 About 481 radicals (community-counted) compose into roughly 2,048 WaniKani kanji and roughly 6,352 vocabulary items. The per-kanji story-writing burden stays small because the building blocks recur.10

Where it pays off

New kanji become combinations of pieces the learner already knows, so per-card learning time drops as primitives accumulate. Migaku and Tofugu both document this as the central pedagogical payoff of Heisig and WaniKani.2910

Heisig's preface relates that he personally learned the writing of "some 1,900 characters" in roughly a month of intensive study, the seed of the popular "three months to 2,000" claim.3 The pace assumes the building-block payoff has materialized.

Pedagogical order strongly favors learners with no prior CJK (Chinese, Japanese, or Korean) exposure, because it teaches the visual grammar of kanji first and the meanings second.3929

What it gets wrong

In the first three months, a learner cannot read much native text, because the kanji learned are graphically convenient but textually rare or absent. Heisig's frames 18 to 35 (冒, 朋, 唱, 晶, 品, 呂, 昌, 旭, 旦, 胆, 亘, 凹, 凸, 旧) contain several items that do not appear in the Asahi top tier (晶, 呂, 旭, 旦, 胆, 亘, 凹, 凸 in particular).162

The order ignores meaning groups. Family kanji (父 母 兄 姉 弟 妹) and weekday kanji (月 火 水 木 金 土 日) get scattered across hundreds of cards in Heisig's frame index, where school order groups them.164

Drop-off in month 2 to 3 is the canonical failure mode

Both RTK and WaniKani lose a substantial fraction of new users between month two and month three. The cause is typically review-load arithmetic, not memory failure: spaced-repetition queues compound, the learner falls behind, and the visible payoff (reading native text) has not yet arrived to motivate catching up.2928

WaniKani's locked spaced-repetition pacing means a learner with prior CJK exposure cannot speed past easy early levels: the Apprentice 4-hour, 8-hour, 1-day, and 2-day floors apply regardless of prior knowledge.28

Side-by-side: the first 50 kanji each order produces

Three columns, same starting line

The first-50 slice across the three orderings makes the differences concrete. School order is taken from the MEXT 学年別漢字配当表 grade-1 list (80 kanji), using the conventional sequence found in reference publications and the kanjicards.org grade-list view.4527 Frequency order is taken from the Bunkachō 2007 Asahi plus Yomiuri ranking. Only the top 10 are individually published in the retransmitted source, so the rest of the column is drawn from the Asahi-based ranking surfaced on the kanjicards.org frequency-list view.272 Pedagogical order is the Heisig RTK Volume 1 6th-edition frame order, frames 1 to 50.163

SlotSchool order (Grade 1)4527Frequency order (Asahi-based)227Pedagogical order (Heisig)163
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50

The frequency-order column past rank 10 is reconstructed from the kanjicards.org Asahi-based newspaper ranking. The primary Bunkachō 2007 publication directly cites only the top 10.227 The internal sequence within grade 1 is a convention, not a MEXT prescription. MEXT publishes the set per grade, but not an order within a grade.45

What the table makes obvious

School order and Heisig order are identical for slots 1 to 10: the numerals 一 through 十. They diverge at slot 11. School order's slot 11 is 日 (a high-utility everyday kanji); Heisig's slot 11 is 口 (chosen because it is a primitive component reused in dozens of subsequent frames including 古, 吾, 品, 呂, 占).416 Slot 11 is the clearest "Heisig vs. school order" worked example in the entire first 50.

Frequency order does not even share the numeral opening. Slot 1 is 日 (rank 1 in both Asahi and Yomiuri 2006)2, slot 2 is 一, slot 3 is 大. High-frequency adult-vocabulary kanji (国, 会, 業, 議, 経, 政, 関, 委) appear inside the first 50. School order does not reach those slots until grades 4 to 6, and Heisig does not reach them until late in Volume 1.2416

KKLC's first slot is 日, matching frequency order at slot 1, but its slot 2 is 一. This is consistent with Conning's stated principle of "rough order of importance" with graphical-similarity grouping.2526

How to choose

The three questions that pick your order

Question 1: prior CJK exposure? Speakers of Mandarin or other languages that use Chinese characters start with most components already familiar as shapes. Heisig's primitive-build-up payoff therefore compresses. The slow climb through pictograph decomposition is unnecessary. Frequency or school orders waste less time for these learners.3929

Question 2: what do you want to read this year? Adult prose (news, novels, manga for adults) rewards frequency order. The cumulative-coverage curve is steepest for the top 500 newspaper-rank kanji.61718 Children's material (folk tales, picture books, primary-school readers) rewards school order, because the readers are written against the kyōiku list.45 Genre fiction with a kanji-heavy style rewards a hybrid: jōyō recognition (any order) plus author-specific hyōgaiji as encountered.7

Question 3: are you committing to a method that ships a fixed order? Heisig RTK, WaniKani, and KKLC each provide a method whose mnemonic system is built around its sequence. Once committed, fighting the order forfeits the mnemonic system's payoff. Switching orders mid-method costs the partial training already invested.3102529

The hybrid that most experienced learners end up at

A common hybrid is "Heisig or WaniKani fast-pass to install visual literacy, then immersion plus mining in a kanji-via-vocab style for actual reading order." Khatzumoto (AJATT) explicitly advocates the pattern. This includes the option to cut Heisig off early once the primitive-decomposition payoff has been internalized.3029

Migaku documents the same pattern as the typical completion route for RTK users: Volume 1 complete or partially complete, then vocabulary cards mined from native text rather than Volume 2.29

When order genuinely does not matter

Past roughly 1,500 kanji, every ordering converges and the learner is filling in long-tail jōyō. The differences between school grade 6, frequency ranks 1500 to 2136, and Heisig's late-volume frames are small and largely irrelevant. Past this point, the dominant kanji-acquisition mechanism is reading-mined exposure (kanji-via-vocab), regardless of the order the learner started with.6717

Tool support for each order

School / grade order

Genki Volume 1 introduces kanji in a loosely school-aligned order, biased toward what a beginner adult learner is reading in the textbook's dialogues.27 The kyōiku list underwrites the entire elementary-school workbook ecosystem (くもん, 学研, ベネッセ pre-gymnasium series). Kanken drill apps target the kyōiku-by-grade Kanken levels 10 to 5.5

kanjicards.org publishes the kyōiku-by-grade list as a printable PDF and online searchable table, alongside the frequency-rank and JLPT-level views.27

Frequency order

The Kodansha Kanji Learner's Course (Conning, 2013) is the canonical in-print frequency-led ordering, about 2,300 kanji with a frequency-plus-graphical-similarity arrangement.2425 Anki "Kanji 2k" and similar community decks expose pure frequency rank without KKLC's graphical overlay.27

JPDB and Migaku frequency-driven exposure both implement frequency order indirectly through vocabulary frequency. A learner sees high-frequency words first, and high-frequency words contain high-frequency kanji, so the kanji-order side effect is frequency-led.29

Pedagogical order

Heisig's Remembering the Kanji Volume 1 (6th edition, 2011, University of Hawai'i Press) is the textbook implementation, with about 2,200 frames in primitive build-up order including the 196 kanji added by the 2010 jōyō revision.39

kanji.koohii is the community-built spaced-repetition application keyed to Heisig's frames, operating since 2006 with explicit permission from Heisig to use the RTK keyword index.31

WaniKani delivers the same logic with built-in spaced repetition and gamification: 60 levels, radical-then-kanji-then-vocabulary gating, and about 481 radicals composing into 2,048 kanji and 6,352 vocabulary items.101128

JLPT-order as a practical hybrid

JLPT order is widely used even though it has no official basis. Tools that ship pre-built N5 to N1 decks let a learner adopt the community-reconstructed JLPT ordering without curating it personally. J-Compass recommends Amenokori for it: its pre-built collections are labeled N5 to N1, with the per-level entry counts visible on the collection cards as N5 (801), N4 (750), N3 (3,355), N2 (1,477 plus 855 extended), and N1 (3,239 plus 803 extended). The per-level decks bundle kanji, vocabulary, and grammar together, so the ordering is adopted in one move.3233 Renshuu also ships pre-built JLPT-level kanji collections,34 and Migii follows the same pattern with N5 to N1 mock-test material that exposes kanji in JLPT-aligned order. None of these tools is a JLPT-authoritative syllabus. Each is a community reconstruction made convenient by tooling.1415

Good to know

Treating a school-order "first 1,000" as a frequency-led "first 1,000"

The kyōiku list is selected for a Japanese child's reading environment, not for adult-prose frequency. A learner who says "I'll learn the first 1,000 kanji a Japanese child learns; those are the most useful 1,000" is mixing up two different selection criteria. The accurate framing is: "those optimize for a child's life context, not for adult corpus frequency." 私 lands at grade 6 in school order but inside the top 100 in adult prose; 必, 経, 政, 義 sit in grades 5 to 6 or in the junior-high tier yet appear constantly in adult writing.467

Picking pure frequency order without prior CJK exposure

Frequency rank 45 may share no components with rank 44 or 46. A learner with no CJK background who starts with a pure frequency deck (Anki "Kanji 2k") and tries to learn ranks 1 to 50 in week 1 gives up the cognitive-load advantage of component reuse, because there is none. The better move is either a frequency-with-graphical-similarity course such as KKLC, where component recurrence is built into the sequence,2425 or a frontloaded month of pure-pictograph build-up (Heisig frames 1 to 50, WaniKani level 1) before joining the frequency curve.925

Switching order mid-method

Each method's mnemonic system depends on its own component vocabulary. A learner who completes 800 RTK frames and then abandons Heisig for a frequency deck loses the primitive-decomposition payoff that the next 1,200 frames were meant to build on. The two better moves are either finishing Volume 1 to capture the payoff, then mining vocabulary, or cutting RTK off explicitly at a chosen frame and switching deliberately, the AJATT-style early-cut pattern.3029 Migrating partial Heisig keywords into a WaniKani or KKLC workflow doubles the keyword work, because the systems use different primitive names.291025

"JLPT N3 kanji" is a community list, not an entitlement

The JLPT has published no kanji list at any level since 2010. The administering bodies' stated rationale is that communicative competence, rather than item memorization, is the goal.12 A learner using "N3 kanji" as a fixed target is using a community reconstruction (Tanos, jlptstudy, JLPTsensei), not an official syllabus. Even the pre-2010 Wikipedia summary notes that "about 20% of the kanji in any one exam may have been drawn from outside the prescribed lists." In other words, even the older official list never fully predicted the paper.1415

The 私 problem as the canonical school-vs-frequency anchor

私 is grade 6 in MEXT's 学年別漢字配当表 and sits inside the top 100 in adult-prose frequency.467 It is the single example that makes the school-vs-frequency split sharpest. A learner who can describe 私 as "the kanji I would meet in week one if I followed Asahi rank, and in year six if I followed MEXT" has internalized the central trade-off this article is built around.

The 2017 prefectural-kanji addition is life-context-driven

The 20 kanji added in 2017 (茨, 媛, 岡, 潟, 岐, 熊, 香, 佐, 埼, 崎, 滋, 鹿, 縄, 井, 沖, 栃, 奈, 梨, 阪, 阜) were not added because they rose in adult-prose frequency. They were added so Japanese fourth-graders can write all 47 prefectural names by the end of grade 4, and they were assigned to grade 4 only.121 A learner who sees a "kyōiku list" published before 2020 will find these 20 kanji at the older grades. A learner who sees a list from 2020 onward finds them concentrated in grade 4.

The order the tool delivers is the order you actually follow

A theoretically optimal frequency-ordered curriculum that the learner abandons in six weeks delivers fewer kanji than a pedagogically ordered curriculum the learner finishes. In practice, tool stickiness matters more than abstract ordering optimality. Picking a tool the learner will actually use beats picking the theoretically optimal order.1029

See also

References

Footnotes

  1. 文部科学省. 『小学校学習指導要領(平成29年告示)解説 国語編』. 平成29年(2017)告示, 解説書 平成29年7月. PDF. https://www.mext.go.jp/component/a_menu/education/micro_detail/__icsFiles/afieldfile/2019/03/18/1387017_002.pdf 2 3 4 5 6 7 8

  2. 文化庁文化部国語課. 『漢字出現頻度数調査(新聞)』. 平成19年(2007). Corpus: Asahi Shimbun + Yomiuri Shimbun, 2006 sampling window. Top-10 frequency table retransmitted via 漢字カフェ (Taishukan Publishing). https://www.kanjicafe.jp/detail/8195.html 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

  3. Heisig, James W. Remembering the Kanji 1: A Complete Course on How Not to Forget the Meaning and Writing of Japanese Characters. 6th edition, University of Hawai'i Press, 2011. ISBN 978-0-8248-3592-7. https://uhpress.hawaii.edu/title/remembering-the-kanji-1-a-complete-course-on-how-not-to-forget-the-meaning-and-writing-of-japanese-characters/ 2 3 4 5 6 7 8 9 10 11

  4. 学習指導要領LOD. 「別表 学年別漢字配当表」. Linked Open Data rendering of the MEXT 学年別漢字配当表 (current revision: 平成29年告示, in force from 2020). https://jp-cos.github.io/821/0000100000000 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17

  5. Wikipedia. "Kyōiku kanji." Summary of the MEXT 学年別漢字配当表 totals after the 2017 revision (1,026 characters; per-grade split 80/160/200/202/193/191). https://en.wikipedia.org/wiki/Ky%C5%8Diku_kanji 2 3 4 5 6 7 8 9 10 11 12

  6. Chikamatsu, Nobuko, Shoichi Yokoyama, Hironari Nozaki, Eric Long, and Sachio Fukuda. "A Japanese logographic character frequency list for cognitive science research." Behavior Research Methods, Instruments, & Computers, vol. 32, no. 3, 2000, pp. 482–500. Corpus: one full year (1993) of Asahi Shimbun morning and evening editions; 56.6 million character tokens. https://link.springer.com/article/10.3758/BF03200819 2 3 4 5 6 7 8 9 10 11 12 13

  7. Joyce, Terry, Hisashi Masuda, and Taeko Ogawa. "Jōyō kanji as core building blocks of the Japanese writing system: Some observations from database construction." Written Language & Literacy, vol. 17, no. 2, 2014, pp. 173–194. Corpus: NINJAL BCCWJ. https://www.jbe-platform.com/content/journals/10.1075/wll.17.2.01joy 2 3 4 5 6 7 8 9 10

  8. 国立国語研究所 (NINJAL). 『現代日本語書き言葉均衡コーパス』(BCCWJ), 2011. 104.3 million words across published books, magazines, newspapers, white papers, blogs, bulletin boards, textbooks, and law. https://clrd.ninjal.ac.jp/bccwj/ 2

  9. "Remembering the Kanji." Wikipedia. Summary of Heisig's primitive-element method and ordering rationale. https://en.wikipedia.org/wiki/Remembering_the_Kanji 2 3 4 5 6 7 8

  10. WaniKani. Landing page. Tofugu LLC. https://www.wanikani.com/ 2 3 4 5 6 7 8 9

  11. WaniKani Knowledge. "How Do I Level Up?" Tofugu LLC. https://knowledge.wanikani.com/wanikani/getting-started/level-up/ 2 3 4

  12. 日本語能力試験 (JLPT) 公式 FAQ. 国際交流基金 (Japan Foundation) and 日本国際教育支援協会 (JEES). On the post-2010 non-publication of the 出題基準 (Test Content Specification). https://www.jlpt.jp/e/faq/index.html 2 3

  13. 日本語能力試験 (JLPT). 「N1〜N5:認定の目安」(Summary of Linguistic Competence Required for Each Level). 国際交流基金・日本国際教育支援協会. Level descriptors only; no kanji counts. https://www.jlpt.jp/e/about/levelsummary.html

  14. Tanos (Jonathan Waller). "JLPT Kanji" study resource. Community-compiled per-level kanji lists derived from past-paper analysis. http://www.tanos.co.uk/jlpt/skills/kanji/ (limitation: community resource, methodology not formally published) 2 3

  15. Wikipedia. "Japanese-Language Proficiency Test." Pre-2010 four-level kanji-count specifications drawn from the JLPT 出題基準, first published 1994, revised 2004. https://en.wikipedia.org/wiki/Japanese-Language_Proficiency_Test (limitation: Wikipedia secondary summary; primary 出題基準 documents are out of print) 2 3

  16. LentoMan wiki. "Remembering The Kanji List." Community-compiled frame-by-frame index of RTK Volume 1 (6th edition), used here as the source for ordered frames 1–50. http://www.lentoman.net/wiki/index.php?title=Remembering_The_Kanji_List (limitation: community-compiled; primary book required for definitive frame order) 2 3 4 5 6 7 8

  17. Kandrac, Patrick. "How Reliable and Consistent Are Kanji Frequency Databases?" Electronic Journal of Contemporary Japanese Studies, vol. 22, no. 2, article 5, 2022. https://www.japanesestudies.org.uk/ejcjs/vol22/iss2/kandrac.html 2 3 4 5 6 7

  18. Kanō, Chieko, cited in Kandrac 2022. The "top 1,000 kanji ≈ 90% coverage" figure as conventionally cited in Japanese pedagogy. (limitation: secondary citation; primary Kanō reference accessed via 17) 2 3

  19. 文部科学省. 「音訓の小・中・高等学校段階別割り振り表(平成29年3月)」. 平成29年(2017)3月. https://www.mext.go.jp/a_menu/shotou/new-cs/1385768.htm

  20. 文化庁. 『常用漢字表』. 平成22年11月30日内閣告示第2号(2010). 2,136 characters. https://www.bunka.go.jp/kokugo_nihongo/sisaku/joho/joho/kijun/naikaku/kanji/

  21. id.fnshr.info / Colorless Green Ideas. "平成29年(2017年)告示小学校学習指導要領における学年別漢字配当表の変更点." Detailed enumeration of the 20 prefectural-name kanji added by the 2017 revision and the four-plus-one kanji shifted between grades. https://id.fnshr.info/2017/02/18/kyo-kan-2017/ (limitation: independent commentator; primary MEXT 解説 in 1 confirms the 20-kanji directive) 2 3 4 5

  22. 文化庁文化部国語課. 『漢字出現頻度数調査』. 平成12年(2000). Corpus of 385 books, 33.3 million kanji tokens, 8,474 distinct kanji types. Summary statistics retransmitted via 漢字文化資料館 (Taishukan Publishing), Q0006. https://kanjibunka.com/kanji-faq/history/q0006/

  23. Scriptin (Lewdwig). "Kanji usage frequency." Interactive cumulative-frequency visualization across Aozora Bunko, Japanese Wikipedia, and Japanese Wikinews. https://scriptin.github.io/kanji-frequency/ (limitation: indie-built; raw data is open-source) 2

  24. Conning, Andrew Scott. The Kodansha Kanji Learner's Course: A Step-by-Step Guide to Mastering 2300 Characters. Kodansha USA, 2013. ISBN 978-1-56836-526-8. https://kodansha.us/book/the-kodansha-kanji-learners-course/ 2 3

  25. Conning, Andrew Scott. "The Kodansha Kanji Learner's Course (KLC)." Keys to Japanese / Kanji Learner's Course Official Site. Course-author description of the ordering principle. https://keystojapanese.com/klc/ 2 3 4 5 6 7 8 9 10

  26. Conning, Andrew Scott. "KLC entry numbers for kanji used in Genki textbooks" (cross-reference PDF). Keys to Japanese. Establishes the first-entry slot mapping for KLC. https://keystojapanese.com/wp-content/uploads/2020/03/Genki-KKLC.pdf 2

  27. kanjicards.org. "Kanji lists ordered by JLPT-level, Grade or Frequency of use." Aggregator surfacing the kyōiku-by-grade list, the per-level community JLPT lists, and a newspaper-frequency rank list side by side. https://kanjicards.org/kanji-lists.html (limitation: aggregator; underlying source per column is named on the page) 2 3 4 5 6 7 8 9

  28. WaniKani Knowledge. "WaniKani's SRS Stages." Tofugu LLC. https://knowledge.wanikani.com/wanikani/srs-stages/ 2 3 4

  29. Migaku. "Remember the Kanji! A Review of the Heisig RTK Method." Cited for explicit articulation of frequency-versus-pedagogical-order trade-offs. https://migaku.com/blog/japanese/heisig-remembering-the-kanji-review (limitation: vendor-adjacent blog; used only where it triangulates with primary sources) 2 3 4 5 6 7 8 9 10 11

  30. Khatzumoto. "What If RTK Isn't Working For Me? Early Cut-Off Kanji." AJATT: All Japanese All The Time. Cited for the named hybrid strategy of "Heisig fast-pass, then immersion mining." https://alljapanesealltheti.me/what-if-heisig-isnt-working-for-me-early-cut-off-kanji/index.html (limitation: independent learner-method blog, cited only for the named position) 2

  31. kanji.koohii. Fabrice Denis. Web SRS application built around the Heisig RTK keyword index, operating since 2006 with explicit permission from Heisig. https://kanji.koohii.com/

  32. Amenokori. Product landing page (pre-built N5–N1 collections). https://amenokori.com/

  33. Amenokori. Mobile app page. https://amenokori.com/mobile-app/

  34. renshuu.org. Official site, listing pre-built JLPT-level kanji decks. https://www.renshuu.org/