I've been working on a kanji dependency graph and one of the data products it enables is a combined RTK1 and RTK3 list: the RTK1 ordering is treated as sacrosanct and a simple algorithm finds a sensible place to insert non-RTK1 kanji. Here's a mapping from original RTK1 & RTK3 ordering (per Katsuo's spreadsheet) to the combined ordering:



Since RTK3 contains kanji used as primitives in RTK1, as well as kanji built from primitives introduced in RTK1, all kinds of criss-crossings are attested. Here's the combined list, with RTK3 kanji in full-width parentheses:

一二三四五六七八九十口（叶）日月田目古吾冒朋明唱晶品呂昌早旭世胃旦胆亘凹凸旧自白（畠）百中千舌升昇丸（兎）寸（吋）肘専博占上（卜）下卓朝嘲只貝（貰）唄貞員貼見児元頁頑凡負万句肌旬勺的首乙乱直具真（顛）工（肛）左右有賄貢項刀刃切召昭則副別丁町可頂子孔了女（嬰）（姦）（娼）（姑）好如母貫兄呪克小少大（尖）多夕汐（洵）（泪）（汝）（汀）外名石（硯）（碩）肖（哨）硝砕砂妬削光（洸）（胱）太器臭嗅妙省厚奇川州（洲）順水氷永泉腺原願泳沼沖汎江汰汁沙潮源活消況河泊（胡）湖測土（埴）（坦）（涅）吐圧埼垣填圭（娃）（畦）（奎）封涯寺時均火（灼）（焔）炎煩淡灯畑災灰点照魚（鮭）（鮪）（鮎）漁里（哩）（浬）黒墨鯉量厘埋同洞胴向尚字（宥）（宍）守完宣（喧）宵安（晏）宴寄富貯木（宋）（沐）（梁）（栂）（槙）（杜）（櫓）（杓）（李）（柘）（椙）（楕）（梧）林（淋）（焚）（梵）（埜）森桂柏枠梢棚杏桐植椅枯朴村相机本札暦案燥未末昧沫味妹朱株若（茉）（苺）（菖）（萌）（蒔）（苫）（莞）草苦苛寛薄葉（喋）（莫）模漠墓暮膜苗兆桃眺犬（厭）（吠）状黙然荻（狸）（狽）（狗）（狛）狩猫牛（牢）（牡）特告（浩）（皓）先洗介（芥）界茶脊（皐）合（檜）塔王（琳）（瑚）（珪）玉宝珠現玩狂旺皇呈全栓理主注柱金（釦）（銚）（錨）（釘）銑鉢銅釣針銘鎮道（逼）（逞）導辻迅造迫逃辺巡車（轟）（輻）連（蓮）軌輸喩前（剪）煎各（烙）格賂略客額夏（榎）処条（洛）落冗冥（瞑）（暝）軍（渾）（暈）（暉）輝運冠夢坑（杭）高（烹）（膏）（亨）享（淳）塾熟亭（亮）京（椋）涼景鯨舎周（鯛）週士吉（桔）壮荘売（涜）学覚栄書津牧（孜）（敦）攻敗枚故敬言（這）（詢）（諄）（諒）（謂）（諜）（註）警計詮獄訂訃討訓詔詰話詠詩語読調談諾諭式試弐（或）域（國）賊栽（哉）載茂戚成（晟）城誠威滅減蔑桟（賎）銭浅止歩渉頻（瀕）肯企歴武賦正（柾）（鉦）証政定（淀）（碇）錠走超赴越是題堤建鍵延誕（疋）礎婿衣（裟）裁装裏壊哀遠猿初（袷）（裡）巾（吊）（帖）布帆幅帽幕（晃）幌（滉）錦（棉）市柿姉肺帯滞刺（棘）制製（云）転芸雨（雫）雲曇雷霜冬（柊）天妖（呑）沃（喬）橋（蕎）嬌立（站）泣章競帝諦童瞳鐘商嫡（鏑）適滴敵匕（牝）（匙）叱匂頃北（乖）背比（枇）昆皆楷諧混渇謁（靄）褐喝葛旨（鮨）脂詣壱毎（晦）敏梅海乞（吃）（迄）乾腹複欠（欽）吹炊歌軟次（凋）茨資姿諮賠（菩）培剖音暗韻識（幟）鏡境亡盲妄荒望方（於）妨坊芳肪訪放激脱説鋭曽（噌）増贈東（煉）（諌）棟凍妊廷染燃賓歳県栃（砺）（也）地池虫（蛋）（蛸）（蛙）（蛤）（蛎）蛍蛇（詑）虹蝶独蚕風（凧）（凪）（楓）（颯）己起妃改記包胞砲泡亀（蝿）電竜滝豚（啄）（琢）逐遂家嫁豪（壕）（濠）腸（楊）場湯（蕩）羊美洋詳鮮達羨差（瑳）着唯（碓）（截）（讐）（錐）（雀）（隼）堆椎誰焦（蕉）（樵）礁集准進雑（此）雌（柴）（些）（砦）準奮奪確午（杵）許歓（潅）権観羽（翰）（翔）習（翫）翌曜（燿）（耀）濯曰困（梱）固錮国団因姻咽園（薗）回（廻）壇（檀）店（庖）（庇）（廟）庫庭庁床麻（麿）磨心（蕊）（惹）（恕）忘恣忍認忌志誌芯忠串患思恩応意臆想息憩恵恐惑感憂寡忙（惟）（惇）（恰）（恢）悦恒悼悟怖慌悔憎慣愉惰慎憾憶惧憧憬慕添必泌手看摩我（蛾）義（蟻）議犠抹（撞）（掴）（捌）（按）（撚）（挺）（擾）（擢）（摺）（掠）（撹）（掟）拭拉抱搭抄抗批招拓拍打拘捨拐摘挑指持拶括揮推揚提損拾担拠描操（妾）接掲掛捗研戒弄械鼻刑型才財材存在乃（孕）携及（汲）吸扱丈（杖）史吏更硬梗又（蚤）（叉）双桑隻護獲奴怒友抜投没股設撃殻支（妓）技枝肢茎（頚）怪軽叔督寂淑反坂板返販爪妥乳浮淫将奨采採菜受授愛曖払（眸）（牟）広勾拡鉱弁雄（宏）（肱）台（苔）怠治冶始胎窓去（怯）法会（桧）至（姪）（蛭）室到致互棄育撤充銃硫（琉）流允唆（悛）（竣）出山（峻）（嵩）（辿）（嶽）（峨）（嵯）拙岩炭岐峠崩密（樒）蜜嵐崎崖入込分（扮）貧頒公（頌）松翁訟谷浴容（蓉）（熔）溶欲（慾）裕鉛沿賞党（嘗）堂常裳掌皮（頗）波婆披破被残（夙）（殆）殉殊殖列裂烈死葬（舜）瞬耳（囁）（茸）取（諏）趣最撮恥職聖敢聴懐（曼）慢（蔓）（鰻）漫買（罫）（贖）置罰寧濁（燭）環還夫（芙）扶渓規（槻）替賛（讃）潜失鉄迭臣姫蔵臓賢（竪）腎堅（樫）（鰹）臨覧巨（炬）拒力（劫）男労募劣功勧努勃励加（茄）（袈）（迦）（珈）賀架脇脅協行律復得（碍）従徒待往征径彼役徳徹徴懲微街桁衡稿（蘇）（禿）（穎）稼程税稚和移秒秋（萩）（鍬）（鰍）愁私秩秘称利（莉）梨穫穂稲香季委秀透誘（尤）稽穀菌萎米（粕）（糊）（籾）（粁）（糎）粉粘粒（庄）粧迷粋謎糧菊奥（襖）数（薮）楼類漆膝様求（黍）（黎）球救竹（箔）（箭）（竺）（笈）（筈）笑笠笹箋（肋）筋箱筆筒等算答策簿（筑）築篭人（仄）（箆）（臥）佐（倭）（俄）（佃）（仔）（仇）（伽）（僑）（倶）（侃）（偲）（做）（雁）（贋）（伍）（什）（佑）（俣）（侑）侶但住位仲体悠（脩）（篠）件仕他伏伝仏休仮伎伯俗信佳依例個健側侍停値倣傲倒偵僧億儀償仙催仁侮使便倍優伐（筏）宿傷（呆）保褒傑付（鮒）符府（腑）任賃代（黛）袋貸化花貨傾何荷俊傍（奄）俺（庵）（掩）久（灸）（玖）畝囚内丙柄肉腐（坐）座挫卒（倅）（鋏）（翠）（巫）傘匁以似併瓦瓶宮営善膳年夜液塚幣（瞥）蔽弊喚換融施旋遊旅勿（吻）（惚）（葱）（忽）物（惣）易（錫）賜尿（犀）（屁）（屑）（屡）（屍）尼尻泥（屏）塀履屋（渥）握屈掘堀居（鋸）据裾層局遅漏刷尺（呎）尽（侭）沢訳択昼戸（芦）肩房扇（煽）炉戻涙雇顧啓示（凛）（蒜）（祟）（祓）礼（祐）（祢）（禎）祥祝福祉社視奈（捺）尉（蔚）慰款禁襟宗崇祭察擦由（迪）（柚）抽油袖宙届笛軸甲押岬挿申（坤）（暢）伸神（榊）捜果（夥）菓課裸斤（芹）（欣）析所祈近折哲逝誓斬暫漸断質斥訴昨詐作雪（鱈）録（禄）剥尋急穏侵浸寝婦掃当彙争浄事唐糖康（糠）逮伊君群（而）耐需（嬬）（濡）儒端（瑞）両満画歯（噛）曲曹（糟）遭漕槽斗（斡）料科図用庸（傭）備昔錯借惜措散廿（燕）庶遮席度渡奔噴墳憤（尭）焼暁半（叛）伴畔判拳券巻（倦）圏勝藤謄片（牒）版之乏芝不（歪）否杯矢（薙）（雉）（矩）矯族知智挨矛（茅）（橘）柔務霧班帰弓（弛）（粥）（夷）引弔弘強弥弱（鰯）溺沸（狒）費第弟（剃）（悌）（梯）巧（聘）号朽誇（樗）（袴）顎（鰐）汚与写身（躾）射謝老（姥）（鰭）考孝教拷者（堵）（屠）煮著箸署（薯）（曙）暑諸（藷）猪渚賭峡（侠）狭挟頬追（槌）（鎚）阜（埠）師（獅）帥官（菅）（舘）棺管父（斧）釜交（狡）（鮫）（佼）効較校足促捉距（躓）（蹄）（跨）（躇）路（蕗）露跳躍践踏踪骨（猾）滑髄禍渦鍋過阪（耶）（隕）（椰）（爺）（陀）（郁）阿際障隙随陪陽陳防附院陣隊墜降階陛隣（憐）（燐）（鱗）隔隠堕陥穴（窺）（竃）空（腔）控突究窒窃窟窪（窄）搾窯窮探深丘岳兵（鋲）浜糸（牽）（纂）（繋）（綜）（紘）（纏）（絢）（紬）（綺）（縞）（綬）（紗）（胤）（轡）織繕縮繁縦緻線綻締維羅練緒続絵統絞給絡結終級紀紅納紡紛紹経紳約細累（螺）索総（聡）綿絹繰継緑縁（毅）網緊紫縛縄幼後幽幾（磯）機畿玄（眩）（絃）畜蓄弦擁滋慈磁系係孫懸遜却脚卸御（禦）服（叩）命令（怜）（苓）（玲）（伶）零（澪）齢冷領（嶺）鈴勇湧通（樋）（桶）踊疑擬凝範犯氾厄危（脆）（跪）宛（婉）（椀）（碗）腕苑怨柳卵（昂）留（劉）（溜）瑠貿印臼（舅）毀興（輿）酉（酋）（楢）（醤）（醍）（醐）（醇）（鄭）酒酌酎酵酷酬酪酢酔配酸猶尊（噂）（樽）（鱒）豆（厨）（逗）頭短豊鼓（嘉）喜（嬉）樹皿血盆盟盗温蓋監濫鑑藍（孟）猛盛塩銀恨根即（卿）爵節（櫛）退限眼良（狼）（莨）朗浪娘食（喰）飯（蝕）（飴）（饅）飲飢餓飾餌館餅養飽既概慨平（秤）呼坪評刈（苅）（肴）（欝）刹希（稀）凶（兇）胸離（禽）（檎）璃殺爽純頓（噸）鈍辛辞梓宰壁（譬）（僻）璧避新（噺）薪親幸（睾）（倖）執（蟄）摯報叫（赳）糾収卑（婢）（稗）（牌）碑陸睦勢（藝）熱菱（凌）（崚）（稜）（綾）陵亥（咳）核刻該骸劾述術寒塞醸（穣）譲壌嬢毒素麦（麹）青（靖）（錆）（鯖）精請情晴清静（瀞）責（蹟）績積債漬表俵潔契喫害轄割憲生（笙）（甥）（甦）星醒姓性牲産隆峰（鋒）（峯）蜂（逢）縫（蓬）拝寿（祷）鋳籍春（榛）（秦）椿泰奏（湊）実奉俸棒謹僅勤漢（歎）嘆難（灘）華（嘩）（樺）垂唾睡錘乗剰今（衿）含貪吟念（稔）捻琴（琵）陰（蔭）予序預野兼嫌鎌謙廉（簾）西（洒）（茜）（晒）（栖）価（粟）要腰票漂標栗慄遷覆煙南楠献門（閃）（悶）（閤）問閲閥間闇簡開閉閣閑聞潤欄闘倉（蒼）（槍）創非（緋）（誹）（琲）俳排悲罪輩扉侯喉候決（抉）（訣）快偉（葦）違緯衛韓干（旱）（栞）（竿）（鼾）肝刊汗軒岸幹芋（迂）宇余除徐叙途斜塗束（漱）頼瀬勅疎（疏）辣速整剣険検倹重（鍾）動腫勲働種衝薫病（痒）（痰）（痔）（癌）（痺）痴痘症瘍痩疾嫉痢痕疲疫痛癖匿（堰）（匡）（嘔）匠医匹区（躯）枢殴欧抑仰迎登（葵）（燈）澄発（溌）廃僚（遼）（燎）瞭寮療彫（厖）（彬）（穆）形影杉彩彰彦（諺）顔須膨参（鯵）惨修珍（滲）（疹）（謬）診文（斐）対紋蚊斑斉剤済斎粛（繍）塁楽薬率渋摂央英（瑛）映赤赦変跡蛮恋湾黄横（巴）把（邑）（芭）（杷）（琶）色絶艶肥甘（嵌）（柑）紺某（煤）謀媒欺（箕）（簸）（斯）棋旗期碁基甚（湛）勘堪貴遺遣（鑓）潰舞無（蕪）組粗租狙祖阻査助（鋤）宜（誼）畳並普譜湿顕繊霊業撲僕共供異（糞）翼戴洪港暴（曝）爆恭（巽）選（撰）殿（澱）井丼囲耕亜（壷）（唖）悪円角触解（蟹）再講購構溝論（綸）倫輪偏（篇）遍編冊（珊）柵典氏（祇）紙（昏）婚低（砥）抵底民眠（甫）（輔）（圃）（鋪）捕（葡）哺浦蒲舗補邸郭（廓）郡郊部都郵邦那郷（饗）響郎廊盾（遁）（楯）循派脈衆（聚）逓段鍛后（垢）幻司（笥）（覗）伺詞飼嗣舟（舵）（艘）舶航舷般（槃）（磐）盤搬船艦艇瓜（狐）（瓢）弧孤（菰）繭益（溢）暇（蝦）（霞）敷来気汽飛沈（耽）枕妻（棲）凄衰（蓑）衷面麺革（鞍）（鞭）（鞘）（鞄）（靭）（鞠）靴覇声（馨）眉呉娯誤（丞）蒸承函極牙（冴）（穿）（訝）芽邪雅釈（悉）番（幡）（播）（蕃）審翻藩毛（橇）（毬）（粍）耗尾（梶）宅（托）（詫）託為偽畏（猥）（隈）長張帳脹髪（髭）展喪巣単（箪）（蝉）戦禅弾桜獣脳悩厳（巌）鎖挙誉猟（蝋）鳥（鷹）（鴻）（鴎）（鵬）（鸚）（鵡）（鵜）（鷺）（鴨）（鳶）（梟）（嶋）鳴鶴烏蔦鳩鶏島暖媛援緩属嘱偶（寓）（萬）（邁）遇愚隅逆（蕨）（朔）塑遡岡鋼綱剛缶陶（萄）揺謡鬱（鹵）（鹸）就（鷲）蹴懇墾貌（豹）免（挽）逸晩勉象像馬（駿）（憑）（駕）（騨）（馳）（騙）（馴）（駁）（駈）（驢）駒験騎駐駆駅騒（掻）駄驚篤罵騰虎（虔）（彪）虜膚虚（嘘）戯虞慮（濾）劇虐鹿（麟）（漉）（塵）（麒）麓薦慶麗熊能態寅演辰（晨）（賑）辱震振娠唇農（膿）濃送（渕）関咲鬼（蒐）（魁）醜魂魔魅塊（龍）襲（寵）（聾）（赫）嚇朕雰箇錬遵罷屯（沌）且（姐）藻隷（愈）癒（爾）璽潟丹丑羞卯（昴）巳（祀）（畷）（綴）（鎧）（凱）（韮）（籤）（懺）（芻）（雛）（趨）（菫）（儲）（鳳）（剽）（咎）（彊）（捏）（揃）（揖）（捧）（撫）（撒）（捷）（蘭）（蒙）（葺）（藁）（遥）（晋）（胚）（橿）（燦）（畢）（碧）（絨）（絆）（紐）（聯）（訊）（輯）（餐）（鼠）（鑿）（曳）（洩）（彗）（慧）（兜）（鼎）（齟）（齬）（嚢）（肇）（叢）（套）（叡）（鴬）（已）（薩）（淵）（脛）（勁）（壽）（躊）（鼈）（瑶）（亙）（煕）

I've spot-checked the results and fixed a few problems, but no doubt more exist. Corrections and suggestions and enhancements are as always most welcome. (One way to improve the algorithm is, instead of inserting RTK3 kanji into the list as soon as its primitives are available, delay them based on stroke order/kanji complexity: I'm looking at 穴（窺）（竃）空... And it's valuable, to me at least, to improve the reordering algorithm instead of manually reordering, since you could imagine many people wanting to do this with many sets of non-RTK1 kanji.)

I'll be spinning up a webapp that uses the graph to generate this and other kanji data products in a few days, よろしく。

Wow, looks great! I'd definitely be interested in something like this.

I don't want to be a party pooper, but what exactly is the point of this? I did RTK1, I'm now working on RTK3...what value does this graph add?

[ファブリス]

I guess aldebrn is proposing an optimized index (based on introduction of primitives) for someone who is quite confident they want to go through Vol 1 and Vol 3.

It is interesting that RTK 3 characters are marked. If I understand correctly, a user may be able to abandon RTK 3 partway through, by skipping over the RTK 3 kanji till the end of the character list.

However it could be argued it's better to start with vanilla RTK 1, so you get more of the crucial characters memorized first. The main value may be for non-RTK sets as aldebrn mentioned.

john555 Wrote:I don't want to be a party pooper, but what exactly is the point of this? I did RTK1, I'm now working on RTK3...what value does this graph add?

A person who decides to do RTK1+3 will spend less time if they study the kanji in order presented by aldebrn than if they study RTK1 followed by RTK3 (like yourself...).

Primarily this was intended as a tech demonstration, and a resource to people who want to incorporate some non-RTK1 kanji into their studies. I started learning the most frequently used kanji (from Core5000) and didn't have a good way to 'RTK-ify' that list, so I've been just learning them in the unnaturally-split RTK1-then-3 ordering. This list as is might help others in that situation.

The ultimate goal is a way to RTK-ify a list (or lists!) of arbitrary kanji to produce a list that builds on primitives, while also trying to stay close to the ordering of the original list/lists (or satisfy some mathematical constraint). This is done by reordering the primitives introduced in RTK: make 人 come before 隹, etc.

I am not expecting it to be any kind of breakthrough ? Just smoothing down one speed bump in the road of life!

ファブリス Wrote:I guess aldebrn is proposing an optimized index (based on introduction of primitives) for someone who is quite confident they want to go through Vol 1 and Vol 3.

It is interesting that RTK 3 characters are marked. If I understand correctly, a user may be able to abandon RTK 3 partway through, by skipping over the RTK 3 kanji till the end of the character list.

However it could be argued it's better to start with vanilla RTK 1, so you get more of the crucial characters memorized first. The main value may be for non-RTK sets as aldebrn mentioned.

In general, I advocate annotating ALL kanji that one is learning with meta-data like frequency (I prefer Core5000's ordering) to help you decide whether to skip it or not. If a user has a corpus of literature that he or she is interested in (in the form of Japanese subtitles, ebooks, etc.), even better: annotate kanji with how frequently (and if!) they are used in that corpus. There are many 常用 kanji that are far from 'crucial' while several non-常用 kanji could be said to be 'crucial'. RTK1's use of the 常用 list is kind of artificial.

As another use of this dependency graph, consider: if you have a list of kanji from your favorite manga and have resolved to learn those kanji before hitting page 1, this dependency graph will help you "fill out" this kanji list, i.e., extend the list with additional kanji that can be used as primitives to build the ones on it. In the film 耳をすませば, "絶" is used. But a very simple algorithm can suggest you first learn 巴 (#2237, in RTK3) and 色 (#1890, in RTK1), even though neither of these is used in the film. Of course one would hope you've learned 色 already, since it is 常用 and in Core5000, i.e., top 5000 words in Japanese = top 1503 kanji. But there are numerous other examples like this. In this way, the 629 kanji in the film's subtitles can be "filled out" to 773 kanji to include kanji used as components by the original 629.

Something that might be more useful to the larger community of RTK1-learners is the following: the RTK1 list, sensibly interleaved with only those few (forty-four) RTK3 kanji that are used as primitives by RTK1 kanji. Again, RTK3 kanji are in parentheses.

一二三四五六七八九十口日月田目古吾冒朋明唱晶品呂昌早旭世胃旦胆亘凹凸旧自白百中千舌升昇丸寸肘専博占上（卜）下卓朝嘲只貝唄貞員貼見児元頁頑凡負万句肌旬勺的首乙乱直具真工左右有賄貢項刀刃切召昭則副別丁町可頂子孔了女好如母貫兄呪克小少大多夕汐外名石肖硝砕砂妬削光太器臭嗅妙省厚奇川州順水氷永泉腺原願泳沼沖汎江汰汁沙潮源活消況河泊（胡）湖測土吐圧埼垣填圭封涯寺時均火炎煩淡灯畑災灰点照魚漁里黒墨鯉量厘埋同洞胴向尚字守完宣宵安宴寄富貯木林森桂柏枠梢棚杏桐植椅枯朴村相机本札暦案燥未末昧沫味妹朱株若草苦苛寛薄葉（莫）模漠墓暮膜苗兆桃眺犬状黙然荻狩猫牛特告先洗介界茶脊合塔王玉宝珠現玩狂旺皇呈全栓理主注柱金銑鉢銅釣針銘鎮道導辻迅造迫逃辺巡車連軌輸喩前煎各格賂略客額夏処条（洛）落冗冥軍輝運冠夢坑高享塾熟亭京涼景鯨舎周週士吉壮荘売学覚栄書津牧攻敗枚故敬言警計詮獄訂訃討訓詔詰話詠詩語読調談諾諭式試弐（或）域賊栽載戊茂戚成城誠威滅減蔑桟銭浅止歩渉頻肯企歴武賦正証政定錠走超赴越是題堤建鍵延誕（疋）礎婿衣裁装裏壊哀遠猿初巾布帆幅帽幕（晃）幌錦市柿姉肺帯滞刺制製（云）転芸雨雲曇雷霜冬天妖（呑）沃（喬）橋嬌立泣章競帝諦童瞳鐘商嫡適滴敵匕叱匂頃北背比昆皆楷諧混渇謁褐喝葛旨脂詣壱毎敏梅海乞乾腹複欠吹炊歌軟次茨資姿諮賠培剖音暗韻識鏡境亡盲妄荒望方妨坊芳肪訪放激脱説鋭曽増贈東棟凍壬妊廷染燃賓歳県栃（也）地池虫蛍蛇虹蝶独蚕風己起妃改記包胞砲泡亀電竜滝豚逐遂家嫁豪腸場湯羊美洋詳鮮達羨差着唯堆椎誰焦礁集准進雑（此）雌準奮奪確午許歓権観羽習翌曜濯曰困固錮国団因姻咽園回壇店庫庭庁床麻磨心忘恣忍認忌志誌芯忠串患思恩応意臆想息憩恵恐惑感憂寡忙悦恒悼悟怖慌悔憎慣愉惰慎憾憶惧憧憬慕添必泌手看摩我義議犠抹拭拉抱搭抄抗批招拓拍打拘捨拐摘挑指持拶括揮推揚提損拾担拠描操（妾）接掲掛捗研戒弄械鼻刑型才財材存在乃携及吸扱丈史吏更硬梗又双桑隻護獲奴怒友抜投没股設撃殻支技枝肢茎怪軽叔督寂淑反坂板返販爪妥乳浮淫将奨采採菜受授愛曖払広勾拡鉱弁雄台怠治冶始胎窓去法会至室到致互棄育撤充銃硫流允唆出山拙岩炭岐峠崩密蜜嵐崎崖入込分貧頒公松翁訟谷浴容溶欲裕鉛沿賞党堂常裳掌皮波婆披破被残殉殊殖列裂烈死葬（舜）瞬耳取趣最撮恥職聖敢聴懐（曼）慢漫買置罰寧濁環還夫扶渓規替賛潜失鉄迭臣姫蔵臓賢腎堅臨覧巨拒力男労募劣功勧努勃励加賀架脇脅協行律復得従徒待往征径彼役徳徹徴懲微街桁衡禾稿稼程税稚和移秒秋愁私秩秘称利梨穫穂稲香季委秀透誘（尤）稽穀菌萎米粉粘粒（庄）粧迷粋謎糧菊奥数楼類漆膝様求球救竹笑笠笹箋（肋）筋箱筆筒等算答策簿（筑）築篭人佐侶但住位仲体悠件仕他伏伝仏休仮伎伯俗信佳依例個健側侍停値倣傲倒偵僧億儀償仙催仁侮使便倍優伐宿傷（呆）保褒傑付符府任賃代袋貸化花貨傾何荷俊傍（奄）俺久畝囚内丙柄肉腐（坐）座挫卒傘匁以似併瓦瓶宮営善膳年夜液塚幣蔽弊喚換融施旋遊旅勿物易賜尿尼尻泥（屏）塀履屋握屈掘堀居据裾層局遅漏刷尺尽沢訳択昼戸肩房扇炉戻涙雇顧啓示礼祥祝福祉社視奈尉慰款禁襟宗崇祭察擦由抽油袖宙届笛軸甲押岬挿申伸神捜果菓課裸斤析所祈近折哲逝誓斬暫漸断質斥訴昨詐作雪録剥尋急穏侵浸寝婦掃当彙争浄事唐糖康逮伊君群（而）耐需儒端両満画歯曲曹遭漕槽斗料科図用庸備昔錯借惜措散廿庶遮席度渡奔噴墳憤（尭）焼暁半伴畔判拳券巻圏勝藤謄片版之乏芝不否杯矢矯族知智挨矛柔務霧班帰弓引弔弘強弥弱溺沸費第弟巧号朽誇顎汚与写身射謝老考孝教拷者煮著箸署暑諸猪渚賭峡狭挟頬追阜師帥官棺管父釜交効較校足促捉距路露跳躍践踏踪骨滑髄禍渦鍋過阪阿際障隙随陪陽陳防附院陣隊墜降階陛隣隔隠堕陥穴空控突究窒窃窟窪（窄）搾窯窮探深丘岳兵浜糸織繕縮繁縦緻線綻締維羅練緒続絵統絞給絡結終級紀紅納紡紛紹経紳約細累索総綿絹繰継緑縁網緊紫縛縄幼後幽幾機畿玄畜蓄弦擁滋慈磁系係孫懸遜却脚卸御服（叩）命令零齢冷領鈴勇湧通踊疑擬凝範犯氾厄危宛腕苑怨柳卵留瑠貿印臼毀興酉酒酌酎酵酷酬酪酢酔配酸猶尊豆頭短豊鼓喜樹皿血盆盟盗温蓋監濫鑑藍（孟）猛盛塩銀恨根即爵節退限眼良朗浪娘食飯飲飢餓飾餌館餅養飽既概慨平乎呼坪評刈刹希凶胸離璃殺爽純頓鈍辛辞梓宰壁璧避新薪親幸執摯報叫糾収卑碑陸睦勢熱菱陵亥核刻該骸劾述術寒塞醸譲壌嬢毒素麦青精請情晴清静責績積債漬表俵潔契喫害轄割憲生星醒姓性牲産隆峰蜂（逢）縫拝寿鋳籍春椿泰奏実奉俸棒謹僅勤漢嘆難華垂唾睡錘乗剰今含貪吟念捻琴陰予序預野兼嫌鎌謙廉西価要腰票漂標栗慄遷覆煙南楠献門問閲閥間闇簡開閉閣閑聞潤欄闘倉創非俳排悲罪輩扉侯喉候決快偉違緯衛韓干肝刊汗軒岸幹芋宇余除徐叙途斜塗束頼瀬勅疎辣速整剣険検倹重動腫勲働種衝薫病痴痘症瘍痩疾嫉痢痕疲疫痛癖匿匠医匹区枢殴欧抑仰迎登澄発廃僚瞭寮療彫形影杉彩彰彦顔須膨参惨修珍診文対紋蚊斑斉剤済斎粛塁楽薬率渋摂央英映赤赦変跡蛮恋湾黄横（巴）把色絶艶肥甘紺某謀媒其欺棋旗期碁基甚勘堪貴遺遣潰舞無組粗租狙祖阻査助宜畳並普譜湿顕繊霊業撲僕共供異翼戴洪港暴爆恭（巽）選殿井丼囲耕亜悪円角触解再講購構溝論倫輪偏遍編冊柵典氏紙（昏）婚低抵底民眠（甫）捕哺浦蒲舗補邸郭郡郊部都郵邦那郷響郎廊盾循派脈衆逓段鍛后幻司伺詞飼嗣舟舶航舷般盤搬船艦艇瓜弧孤繭益暇敷来気汽飛沈枕妻凄衰衷面麺革靴覇声眉呉娯誤（丞）蒸承函極牙芽邪雅釈番審翻藩毛耗尾宅託為偽畏長張帳脹髪展喪巣単戦禅弾桜獣脳悩厳鎖挙誉猟鳥鳴鶴烏蔦鳩鶏島暖媛援緩属嘱偶遇愚隅逆（朔）塑遡岡鋼綱剛缶陶揺謡鬱就蹴懇墾貌免逸晩勉象像馬駒験騎駐駆駅騒駄驚篤罵騰虎虜膚虚戯虞慮劇虐鹿麓薦慶麗熊能態寅演辰辱震振娠唇農濃送関咲鬼醜魂魔魅塊（龍）襲（赫）嚇朕雰箇錬遵罷屯且藻隷（愈）癒（爾）璽潟丹丑羞卯巳

(NB. that this could be readily built using a resource that I've seen somewhere on this forum: a list of primitives useful for RTK1 but not in it.)

(NB2. I notice an unfortunate imperfection in the algorithm. Consider this substring: ...妖（呑）沃（喬）橋... 呑 is placed where it is because 妖 is the first RTK1 kanji using 夭, and because the algorithm places RTK3 kanji as soon as all its components have been attested in the RTK1 sequence. For similar reasons, it placed 喬 just before its first RTK1 use, 橋. If 沃 wasn't so unfortunately placed, we'd have the very felicitous ordering of "（呑）（喬）橋", two RTK3 kanji building on each other, without any interruption between them. I have made a note of this.)

Inny Jan Wrote:

john555 Wrote:I don't want to be a party pooper, but what exactly is the point of this? I did RTK1, I'm now working on RTK3...what value does this graph add?

A person who decides to do RTK1+3 will spend less time if they study the kanji in order presented by aldebrn than if they study RTK1 followed by RTK3 (like yourself...).

I guess the negative thing about interfiling the RTK3 kanji into the RTK1 list and making one monster list is that it will delay/prolong the completion of the RTK1 jouyou kanji. (Not only that, people burn out part way through RTK1 as it is...)

If you're pressed for time it's still better (at least it was for me) to focus totally on the RTK1 jouyou kanji first.

@john555, you're 100% right. The full combined RTK1 & 3 lists aren't something I advocate people use, but was rather intended as a technology demonstration of the kanji dependency walker (which I have yet to publicly release...).

The second list, of RTK1 kanji with just a handful of simplifying RTK3 kanji interleaved (forty-four) posted yesterday, may have broader usefulness to RTK1 learners: I personally find it quite aggravating when a "primitive" in RTK1 is a bona fide kanji that happens to be outside RTK3. I don't see why Heisig included some non-常用 kanji like 勺 in RTK1 but omitted these other forty-four.

Doing these tricks with RTK1 is really a tech demo though, and the real benefit of all this is going to be when people bring their own kanji lists, that have nothing to do with 常用 kanji and everything to do with their own personalized regimen, and can reorder them sensibly for learning quickly. I personally feel RTK's reliance on 常用 kanji is a historic accident, and and that several equally valid approaches exist (e.g., learn the top 500 kanji in RTK-ized order, then then next 500, and so on: if you're really pressed for time and want immediate gratification, this is the way to go).

How about: the kanji that appear in the first x words of optimized Core? Plus all other Core2k kanji that can be built from the same primitives.

aldebrn Wrote:I personally feel RTK's reliance on 常用 kanji is a historic accident, and and that several equally valid approaches exist (e.g., learn the top 500 kanji in RTK-ized order, then then next 500, and so on: if you're really pressed for time and want immediate gratification, this is the way to go).

The downside to learning less then the full RTK set is that you don't get as much repetition of primitives in subsequent kanji. For instance, I'm doing a set of ~1000 kanji(RTK lite) there's more than a few primitives that only show up once or twice. For a smaller set, like 500 i'll bet there's not much primitive repetition at all and more akin to random memorization. Maybe if you run the numbers, it's not as bad as all that, but that's how it feels to me doing RTKlite.

yogert909 Wrote:The downside to learning less then the full RTK set is that you don't get as much repetition of primitives in subsequent kanji. For instance, I'm doing a set of ~1000 kanji(RTK lite) there's more than a few primitives that only show up once or twice. For a smaller set, like 500 i'll bet there's not much primitive repetition at all and more akin to random memorization. Maybe if you run the numbers, it's not as bad as all that, but that's how it feels to me doing RTKlite.

You're absolutely right. Repetition of primitives is very valuable in cementing stroke order and building awareness of the diversity of combinations possible, and is sacrificed whenever you decide to not study some kanji. Some primitives are going to be hurt more than others, and it'll be just chance that decides which. It's an important soft-criterion to keep in mind.

We'd love for you to release the tool, or the script/code.