Kanji Data

This is a fork from https://github.com/davidluzgouveia/kanji-data

I added to the kanji-kyouiku.json more information for German kanji learners. The preprocessing code is written in Python. The following changes are made.

• Using https://www.kanji-trainer.org/Merksatz/index.html the German meanings are added. See add_german function.
• Using https://www.wanikani.com/radicals (downloaded to wanikani_radicals.html) we translate radical meaning to German. The translation was done manually with https://deepl.com. The translation can be investigated in radical-translations.csv. This is used to create kanji-kyouiku-de-radicals.json. There are also radical kanjis added. Since some radicals changed in WaniKani, unused ones are sorted out. See radicals_check function.
• Since the Silkworm has missing data, it is removed from the dataset.
• The dataset is a dict. To sort it, it is converted to an array of dicts, see kanji-kyouiku-de-radicals-array.json. It is ordered by grade and kanji unicode. See dict_to_array function.
• Mnemonic sentences are prepared and manually filled. The keys are mnemonic_meaning_de and mnemonic_reading_de. It is written to kanji-kyouiku-de-radicals-array-mnemonics.json. See prepare_mnemonics function.
• The kanji-kyouiku-de-radicals-array-mnemonics-wip.json is the Work-In-Progress (WIP) file that is filled manually. The watch_copy_to_docs.sh is used to copy it to docs.
• In docs folder is kyouiku-de.html which shows a table of the kanjis and mnemonics.
• In img folder are image (*-img.jpg) and kanji (*-kanji.jpg, transparent *-kanji.png) pairs to give a visual clue for those kanji which are radicals. Almost all images are generated (thanks to pollinations, deepai and craiyon), some are manually post-processed. There is now an option "has_radical_img": true to enable the visual clue in kyouiku-de.html. Using CSS, image and kanji are overlapped and blended in an animation.
• The order given by the table in Kyōiku-Kanji is used to sort entries in kanji-kyouiku-de-radicals-array-mnemonics-wip.json. Every entry has an order_wiki key. However, to introduce radicals before they are used in a kanji, some kanjis are sorted accordingly. For example, 弓 occurs before 弱.
• With a crawl from jisho, furigana analysis on common words is performed, to inspect the reading distribution. For example, 村 is read in 3 common words as そん (60%) and in 2 common words as むら (40%).
• Completed entries are provieded as Anki cards (see section below).
• To learn real Japanese words (vocabulary) common words are crawled from jisho. They are translated to German with Wadoku. Only words which contain learned kanjis are selected in anki decks. They are collected in pack of 50: for every 50 learned kanjis more words are accessable. The words are sorted by frequency (using japanese subtitles word kanji frequency lists) to learn the most frequent words first. Sometimes the words use readings of kanjis that were not learned which is indicated with a reading score.
• In mdbook are markdown files and resources which compile a German book (WIP) about learning Japanese (created using mdBook). It is published at http://markus-projects.net/www/strukturiert_japanisch_lernen/.

Progress

• 2024-08-12 Milestone: In kanji-kyouiku-de-radicals-array-mnemonics-wip.json for the first grade Kyōiku-Kanji (80) all mnmemonics and in img all visual clues are completed.
• 2024-10-05 Milestone: 200 kanjis done with mnmemonics and visual clues.
• 2024-11-24 Milestone: 300 Kanjis done with mnmemonics and visual clues.
• Some commit messages show the current progress.

Built Anki decks can be downloaded on the Lernmaterialien (learning material) page.

Lesung Merksatz Leitfaden

Um gute Merksätze für Lesungen zu schreiben, ist hier ein Leitfaden mit Empfehlungen zusammengefasst.

1. Die Lesung sollte ein Prefix eines deutschen Wortes sein. Beispiel: Kamm (か).
2. Ist ein Perfix nicht möglich, präferiere ein Postfix. Beispiel: Attacke (たけ).
3. Wenn Prefix und Postfix nicht möglich sind, präferiere ein Infix. Beispiel: Medaillons (だい).
4. Wenn Prefix, Postfix und Infix nicht möglich sind, erzeuge ein (erfundenes) Komposita. Beispiel: Embryokugel (りょく), Wasserhahnnahrung (はな).
5. Wenn Prefix, Postfix, Infix und Komposita nicht möglich sind, erzeuge ein Satz mit mehreren Wörtern. Beispiel: Hai ja schikaniert (はやし).
6. Gleiche Laute können bei der Verbindung von Wörtern genutzt werden. Beispiel: ich Chips (いち), muss Schicksal (むし).
7. Das Merkwort sollte so kurz wie möglich sein. Beispiel: Sand (さん).
8. Vermeide englische, japanische oder ungebräuchliche Wörter. Beispiel: engl. hatch (はち), jap. Shougun (しょう) oder ionisiert (よ) [Chemie Domäne].
9. Bei lang gezogenen Vokalen in der Lesung möglichst auch ein Wort, das auch ein lang gezogenen Vokal hat. Beispiel: Kuh (くう).
10. Bevorzuge Aussprache vor Aussehen. Beispiel: Fleißigjugend (きゅう), anstatt Fleißikyuugend (きゅう).
11. Vermeide Wörter, die weiter gelesen eine andere Lesung ergeben könnten. Beispiel: Für も nimm nicht Mokka, da es auch die Lesung もか sein könnte.
12. Vermeide Namen von Personen, Organisationen, Orten und anderen Entitäten. Beispiel: John (じょ).
13. Der Merksatz sollte so kurz wie möglich sein.
14. Die Merkwörter, die die Lesung enthalten, sollten am Satzende stehen.

Anki

Anki is a program which makes remembering things easy. Because it's a lot more efficient than traditional study methods, you can either greatly decrease your time spent studying, or greatly increase the amount you learn.

Using data from kanji-kyouiku-de-radicals-array-mnemonics-wip.json Anki decks are generated. They are published at Lernmaterialien as *.apkg files. Instead of creating sub-decks for each grade, one deck named "Unterrichtsschriftzeichen" is created with all finished kanjis in them.

Two card types are available:

• Merksatz Karte - learn the meaning and reading of kanjis by mnemonic sentences.
• Merkbild Karte - learn the meaning and reading of kanjis by visual clues (meaning) and a mnemonic sentence (reading).

This way, meaning and reading(s) are learned together with one card.

For reading input there are two versions: hiragana or romaji. The latter proves to be useful when the learners do not have a Japanese input method. For multiple reading possibilities, a separator is used (by default space " ").

Screenshots

JSON Example

An entry in kanji-kyouiku-de-radicals-array-mnemonics-wip.json has the following structure:

{
    "strokes": 3,
    "grade": 1,
    "freq": 97,
    "jlpt_old": 4,
    "jlpt_new": 5,
    "meanings": [
        "Below",
        "Down",
        "Descend",
        "Give",
        "Low",
        "Inferior"
    ],
    "readings_on": [
        "か",
        "げ"
    ],
    "readings_kun": [
        "した",
        "しも",
        "もと",
        "さ.げる",
        "さ.がる",
        "くだ.る",
        "くだ.り",
        "くだ.す",
        "-くだ.す",
        "くだ.さる",
        "お.ろす",
        "お.りる"
    ],
    "wk_level": 1,
    "wk_meanings": [
        "Below",
        "^Down",
        "^Under",
        "^Beneath"
    ],
    "wk_readings_on": [
        "か",
        "げ"
    ],
    "wk_readings_kun": [
        "!した",
        "!さ",
        "!くだ",
        "!お"
    ],
    "wk_radicals": [
        "Ground",
        "Toe"
    ],
    "meanings_de": [
        "unten",
        "abwärts",
        "hinabsteigen"
    ],
    "wk_radicals_new": [
        "Ground",
        "Toe"
    ],
    "wk_radicals_missing": [],
    "wk_radicals_de": [
        "Boden",
        "Zehe"
    ],
    "wk_radicals_kanji": [
        "一",
        "ト"
    ],
    "kanji": "下",
    "kanji_ord": 19979,
    "is_radical": false,
    "mnemonic_meaning_de": "Im <span class='radical' data-kanji='一'>Boden</span> <span class='radical_kanji'>(一)</span> steckt meine <span class='radical' data-kanji='ト'>Zehe</span> <span class='radical_kanji'>(ト)</span> und zeigt nach <span class='meaning' data-kanji='下'>unten</span> <span class='meaning_kanji_meaning'>(下)</span>.",
    "mnemonic_meaning_de_done": true,
    "mnemonic_reading_de": "Haare hängen nach <span class='meaning' data-kanji='下'>unten</span> <span class='meaning_kanji_reading'>(下)</span>, also mit <span class='reading onyomi' data-hiragana='か'>Ka</span>mm <span class='hiragana'>(か)</span> und <span class='reading onyomi' data-hiragana='げ'>Ge</span>l <span class='hiragana'>(げ)</span> stylen.",
    "mnemonic_reading_de_done": true,
    "order_wiki": 14,
    "reading_dist": [
        {
            "reading": "した",
            "count": 15,
            "prop": 0.29
        },
        {
            "reading": "か",
            "count": 11,
            "prop": 0.21
        },
        {
            "reading": "げ",
            "count": 11,
            "prop": 0.21
        },
        {
            "reading": "くだ",
            "count": 7,
            "prop": 0.13
        },
        {
            "reading": "さ",
            "count": 3,
            "prop": 0.06
        },
        {
            "reading": "しも",
            "count": 2,
            "prop": 0.04
        },
        {
            "reading": "へた",
            "count": 2,
            "prop": 0.04
        },
        {
            "reading": "お",
            "count": 1,
            "prop": 0.02
        }
    ],
    "order_wiki_radical_corrected": 16,
    "grade_corrected": 1
}

Radical -> Meaning Mnmemonic

A German sentence that combines the radicals to form the meaning.

Im Boden (一) steckt meine Zehe (ト) und zeigt nach unten (下).

Meaning -> Reading Mnmemonic

A German sentence that uses markup annotation to emphasize how to read the kanji.

Haare hängen nach unten (下), also mit Kamm (か) und Gel (げ) stylen.