Lightly-opinionated compendium of vocabulary from Minna no Nihongo Shokyū Ⅰ & Ⅱ books.

  - id: [2, 10]
    edition: [1, 2]
    kanji: "新聞"
    kana: "しんぶん"
    romaji: "shinbun"
    meaning: {
      en: "newspaper",
      fr: "journal",
    }

^{Figure – A word straight from the dataset. More information on the format this way.}

1. Overview
2. Usage
3. Bibliography
4. Opinions
5. Notes
6. Warning

This project aims to serve as a comprehensive vocabulary list for the Minna no Nihongo Shokyū series, packaged into a single YAML file – which will be called the dataset.
To be more specific, it intends to be as close as possible to the source material, in an effort to (hopefully) provide a foundation anyone can use or expand on.

The dataset currently targets two languages for meaning :

languages:
  en: "English"
  fr: "Français"

^{Further information here.}

Rōmaji are provided solely for convenience, and do not correspond to those of the rōmaji edition of the books.

They were made using a mix of pykakasi and readings supplied by Google Translate. As a result, they more or less follow standards set by the Modified Hepburn system (yes, mācrōns inclūdēd !).

Here is a basic example in python.

import yaml

# Load the dataset
with open("minna-no-ds.yaml", "r", encoding="utf-8") as f:
  ds = yaml.load(f, Loader=yaml.FullLoader)


# Extract the keys for all available lessons
lessons: list = [lesson["key"] for lesson in ds["lessons"]] # ['lesson-01', 'lesson-02', ...]


# Go through each lesson and print out its contents
for key in lessons:
  print(f"Contents of {key}") # Outputs : Contents of lesson-01
  print(ds[key]) # Outputs : [{'id': [1, 1], 'edition': [1, 2], 'kanji': None, 'kana': 'わたし', 'romaji': 'watashi', 'meaning': {'en': 'I', 'fr': 'je, moi'}}, ...]

As you may know, Minna no Nihongo Shokyū comes in two books of twenty-five lessons each ; both available in two editions (the second of which is an updated version of the original).

Presented bellow is a table showing the books used in the making of the dataset.

What I call an opinion is any rule I set while creating the dataset that is not directly derived from the source material.

See OPINIONS.md

This file also includes general information about the structure of the dataset.

• When starting out with this project, I used Paul Denisowski's vocabulary lists to generate a blank template for me to fill in. Serious time-saver right there !
• As strings in romaji do not need to be spellchecked, you may use the following config with CSpell.

    "cSpell.ignoreRegExpList": [
        "/romaji:\\s*\"[^\"]*\"/gi"
    ]

• This should have shipped with the set of scripts I used to lint and validate the dataset.
  It didn't, but who knows, I may get to it when (if) I stop being obsessed with that one space dwarves simulator ¯\_(ツ)_/¯
• Adding or removing a word will alter the id of all subsequent words in the same lesson.
  Therefore, any time this has to be done, the version number will have to be bumped to the next major release as this could be considered breaking change for anyone using id as a primary key.
• « This must have taken quite the amount of time to make » well you don't say ! (笑)
  Though I think I'm happy with how it turned out c:
• I haven't yet managed to get my hands on a French version of the first edition of book 1, so words found exclusively in Book 1, edition 1 have no French meaning for now.

#   * The selection of words and their respective translations are the sole property of 3A Corporation.
#     This dataset and subsequent projects that depend on it shall only be used *in conjunction with* – and not *as a substitute for* – of the books ; so as to not cause any financial harm to the IP owners.
#   * As per previous remarks, no commercial use of this file shall be admissible.

^{More here.}

The lack of license is deliberate, as I am uncertain about the appropriate licensing options for this project. Content isn't mine, only the dataset structure and the actual work of filling it in.
If you know a suitable option, feel free to open an issue !

Hopefully that doesn't stop anyone from using the dataset though.