The latest release is v1.2.6 (June 2025). Please cite this version when using the dataset.

This database is a full-text digitization of the Kanchi-in manuscript of the Ruiju Myōgishō (abbreviated as KRM), incorporating location information, textual collation, source studies, and more. It is one of the Hanzi dictionary databases comprising the Integrated Database of Hanzi Dictionaries in Early Japan (abbreviated as HDIC). The terms 'kanji' and 'hanzi' are explained later.

The Kanchi-in manuscript of the Ruiju Myōgishō is a Hanzi dictionary compiled in the twelfth century by a Shingon Buddhist monk. It has been regarded as an important resource for Japanese historical linguistics research due to its extensive collection of wakun indicating accent, detailed annotations on Hanzi pronunciations, and annotations on variant characters. Furthermore, its Chinese annotations on fanqie, meanings, and glyph forms have also garnered attention as materials for Chinese linguistics.

It was first published in March 2022, and in March 2025, a revised edition with specification changes and detailed explanations will be released.

To bridge the gap between 'Kanji' and 'Hanzi,' and to facilitate international academic discourse, the following supplementary explanation may be useful:

Dictionaries of Chinese characters compiled in Japan during the Heian period are invaluable resources not only for the study of Japanese linguistics but also for the study of Chinese linguistics. To promote international accessibility, we propose using the term 'Hanzi.' Researchers specializing in Japanese studies may, without any issue, read this term as 'Kanji.' This approach aims to respect the linguistic diversity and academic traditions of both fields, while encouraging broader scholarly exchange.

This explanation aims to provide clarity and respect for both terminologies, ensuring that researchers from different backgrounds can engage with the material without linguistic barriers.

If you use KRM for academic purposes, please cite the following (Chicago Style):

Ikeda, Shōju. (2025). KRM: Database of the Kanchi-in Manuscript of the Ruijū Myōgishō. Version v1.2.6. Zenodo. https://doi.org/10.5281/zenodo.15481563 (Updated January 18, 2026)

BibTeX

@misc{krm2025,
  author       = {Ikeda, Shōju},
  title        = {{KRM: Database of the Kanchi-in Manuscript of the Ruijū Myōgishō}},
  year         = 2025,
  month        = june,
  version      = {v1.2.6}
  publisher    = {Zenodo},
  doi          = {10.5281/zenodo.15481563},
  url          = {https://doi.org/10.5281/zenodo.15481563}
}

The data from the Kanchi-in manuscript of the Ruiju Myōgishō, published at https://github.com/shikeda/krm, is as follows. This includes some files that are currently being prepared for public release.

• krm_main: Basic data. Includes information about head characters, full definitions, locations, etc. TSV and JSON files are available.
• krm_notes: Annotation data. Categorized into head characters, glyph annotations, pronunciation annotations, meaning annotations, wakun, and others, with collation and source studies. TSV and JSON files are available.
• [krm_headword_chars]: Detailed information about all head characters. Includes location in the Kazama edition, location in the Tenri edition, image file names, etc. Currently under preparation for release.
• krm_wakun: Wakun data. Includes information about variant forms of wakun, variant forms of hanzi (itai-ji), and correspondence with the "Notation" field of the Nihon Kokugo Daijiten (Second Edition). TSV and JSON files are available.
• krm_pronunciations: Data for linking with DHSJR regarding pronunciation annotations.
• krm_ndl: Links to the National Diet Library Digital Collections. TSV file available. Same as the already published KRM_ndl.tsv.

A significant specification change was implemented in March 2025. Previously, the published files were prefixed with "KRM," but the files following this specification change will be prefixed with "krm."

The files incorporating the specification changes have been placed in the "v1.2" folder. Please note that this is a temporary arrangement.

Here are the key points of the specification change:

• The at mark "@", which indicates that kana wakun does not have tone marks, has been changed to an underscore "_".
• The double quotation mark """, which indicates that voiced sound wakun has tone marks, has been changed to a half-width English letter "V".
• The half-width parentheses "()", which indicate the presence of tone marks, have been changed to full-width parentheses "（）".
• The half-width parentheses "()" indicating a correction proposal for a typo have been changed to full-width square brackets "〔〕".
• The half-width square brackets "[]", which indicate missing characters, have been changed to full-width square brackets "［］".

The following ER diagram shows the relationship between the three tables: krm_main, krm_notes, and krm_wakun.

Moreover, krm_notes.json has a nested structure as shown in the following diagram.

This section describes information common to all data files published here.

Date published: March 30, 2025
Last modified: January 18, 2026

• Version: 1.2.8 (latest)
  Date published: March 28, 2025
  Last modified: January 18, 2026
• Version: 1.1.347
  Date: March 11, 2022
  Last modified : March 17, 2025

• Version: 1.2.10 (latest, Newly released after the specification change)
  Date published: March 28, 2025
  Last modified: January 18, 2026

• Version: 1.2.2 (latest, Newly released after the specification change)
  Date published: May 11, 2025
  Last modified: May 21, 2025

• Version: 1.2.2 (latest, DHSJR format)
  Date published: May 20, 2025
  Last modified: June 13, 2025

• Version: 1.2.4 (latest)
  Date published: March 28, 2025
  Last modified: January 18, 2026

• Version: 1.1.97
  Date published: June 11, 2024
  Last modified: March 17, 2025

• Version: 1.1.5 (latest) Date published: October 20, 2015 Last modified : June 14, 2025

HDIC Project Representative: Shoju Ikeda (Professor Emeritus, Hokkaido University) Copyright (c) 2022-2026 HDIC project, IKEDA Shoju (Chair, Professor Emeritus, Hokkaido University)

Contact: [email protected], [email protected], [email protected], [email protected]

The contents of this site are provided under the CC BY-NC-SA 4.0 license.
This is open access data.

For each cited reference, the original notation is given first, followed by the English translation. The Romanized notation is provided in parentheses where necessary.

• 正宗索引: 正宗敦夫編, 類聚名義抄 仮名索引, 日本古典全集刊行会, 1939-1940
  Masamune Index: Edited by Masamune Atsuo, Kana Index to the Ruiju Myogisho, Nihon Koten Zenshu Kankokai, 1939-1940
  (Romanization: Masamune Sakuin: Masamune Atsuo hen, Ruiju Myōgishō Kana Sakuin, Nihon Koten Zenshū Kankōkai, 1939-1940)
• 岡田研究: 岡田希雄, 類聚名義抄の研究, 一条書房, 1944
  Okada Research: Okada Mareo, Research on the Ruiju Myogisho, Ichijo Shobo, 1944
  (Romanization: Okada Kenkyū: Okada Mareo, Ruiju Myōgishō no Kenkyū, Ichijō Shobō, 1944)
• 望月和訓集成: 望月郁子編, 類聚名義抄: 四種声点付和訓集成, 笠間書院, 1974
  Mochizuki Wakun Collection: Edited by Mochizuki Ikuko, Ruiju Myogisho: Collection of Four Types of Wakun with Tone Marks, Kasama Shoin, 1974
  (Romanization: Mochizuki Wakun Shūsei: Mochizuki Ikuko hen, Ruiju Myōgishō: Shishu Shōten-tsuki Wakun Shūsei, Kasama Shoin, 1974)
• 中村文選: 中村宗彦, 九条本文選古訓集, 風間書房, 1983
  Nakamura Monzen: Nakamura Munehiko, Old Japanese Readings of the Kujo Text of the Wen Xuan, Kazama Shobo, 1983
  (Romanization: Nakamura Monzen: Nakamura Munehiko, Kujō-bon Monzen Kokunshū, Kazama Shobō, 1983)
• 草川和訓集成: 草川昇編, 五本対照類聚名義抄和訓集成, 汲古書院, 2000
  Kusakawa Wakun Collection: Edited by Kusakawa Noboru, Comparative Collection of Wakun from Five Texts of the Ruiju Myogisho, Kyuko Shoin, 2000
  (Romanization: Kusakawa Wakun Shūsei: Kusakawa Noboru hen, Gohon Taishō Ruiju Myōgishō Wakun Shūsei, Kyūko Shoin, 2000)
• 西端誤写考察: 西端幸雄, 類聚名義抄における誤写の考察, 訓点語と訓点資料45, 1971
  Nishihata Miscopy Study: Nishihata Yukio, A Study on Miscopies in the Ruiju Myogisho, Kunten-go to Kunten Shiryo 45, 1971
  (Romanization: Nishihata Gosha Kōsatsu: Nishihata Yukio, Ruiju Myōgishō ni okeru Gosha no Kōsatsu, Kunten-go to Kunten Shiryō 45, 1971)
• 西端誤写諸例: 西端幸雄, 類聚名義抄における誤写の諸例, 訓点語と訓点資料52，1973
  Nishihata Miscopy Examples: Nishihata Yukio, Examples of Miscopies in the Ruiju Myogisho, Kunten-go to Kunten Shiryo 52, 1973
  (Romanization: Nishihata Gosha Shorei: Nishihata Yukio, Ruiju Myōgishō ni okeru Gosha no Shorei, Kunten-go to Kunten Shiryō 52, 1973)
• 略注: 佐藤喜代治，色葉字類抄略注，明治書院，1995
  Brief Notes: Sato Kiyoji, Brief Notes on the Iroha Jirui Sho, Meiji Shoin, 1995
  (Romanization: Ryakuchū: Satō Kiyoji, Iroha Jirui Shō Ryakuchū, Meiji Shoin, 1995)
• 群書治要: 小林芳規・原卓志・山本秀人・山本真吾・佐々木勇編, 宮内庁書陵部蔵本群書治要経部語彙索引, 汲古書院, 1996
  Gunsho Chiyo: Edited by Kobayashi Yoshinori, Hara Takushi, Yamamoto Hideto, Yamamoto Shingo, Sasaki Isamu, Index to the Vocabulary of the Classics Section of the Gunsho Chiyo, Imperial Household Agency Archives Collection, Kyuko Shoin, 1996
• 毛詩鄭箋: 毛詩鄭箋（一）（二）（三）, 古典研究会叢書漢籍之部１～３, 原本所蔵静嘉堂文庫, 汲古書院, 1992
  Mao Shi Zheng Jian: Mao Shi Zheng Jian (1) (2) (3), Series of the Classical Studies Association, Chinese Classics Section 1-3, Original Texts in the Seikado Bunko, Kyuko Shoin, 1992

We would like to express our gratitude to Tenri Central Library and Yagi Bookstore for granting permission to publish the decipherment text of the Kanchi-in manuscript of the Ruiju Myōgishō.

This research is partly supported by JSPS KAKENHI Grant Numbers 16H03422, 19H00526, 23K17500 and 25K00466. We gratefully acknowledge this support.

This section describes the core files of the database for the Kanchi-in manuscript of the Ruiju Myōgishō (hereinafter "Myōgishō").

Previously, the released file was a TSV file named KRM.tsv.

It contains information regarding Headwords, the full content of the Definition (Original Glosses), volume, radical, and the locations in the Kazama Shobō edition and the Tenri Central Library/Yōtokusha (Tenri Zenhon Sōsho) edition.

In March 2025, the specifications for column names and the display method for Tone marks (*shōten*) were updated. To clearly indicate that it is the file with these updated specifications, it was renamed krm_main.tsv. A JSON version of this file has also been made available.

The correspondence between the old and new column names is as follows:

The KR2ID column was omitted, and the kazama_location column was aligned with the KRID column.

The Remarks column was omitted; this information is now consolidated in the krm_notes file (which contains data for the Compiler's Remarks).

The content of each column (v1.2.5) is explained below.

A new file, krm_notes.tsv, has been created, containing detailed annotation information added to the KRM_definitions.tsv file.

This is available in both TSV and JSON formats. To explicitly indicate that these are the filenames after the specification change in March 2025, lowercase "krm" was used instead of uppercase "KRM," resulting in the names krm_notes.tsv and krm_notes.json.

This section compares the column names in the new krm_notes.tsv (v1.2.6) with those in the previous files it replaces or incorporates data from.

The following table shows the correspondence between columns primarily related to definition details derived from the previous definitions file:

The new krm_notes.tsv also incorporates information previously stored in KRM.tsv. The corresponding column names are compared below:

The JSON representation of krm_notes utilizes a nested format, as detailed below.

In the ER diagram, the krm_notes table is shown as a child table linked to krm_main (as detailed in the krm_main section) by entry_id. However, in the actual JSON data, the equivalent of the krm_notes table is not flat: instead, it is implemented as a nested array of objects under the key "definitions" within each top-level record (referred to as a krm_main conceptual record).

Each object inside the definitions array corresponds to a definition note and contains the following fields:

• definition_seq_id
• definition_elements
• definition_type_code
• definition_type_name
• remarks

This structure can be conceptually represented in the ER diagram as follows:

• The krm_main table has a one-to-many relationship with a conceptual definitions or notes table.
• In the JSON representation, the definitions are embedded as an array of objects within each krm_main object, rather than being stored in a separate flat table.

Example JSON:

{
  "entry_id": "F00001",
  ...
  "definitions": [
    {
      "definition_seq_id": "F00001_01",
      "definition_elements": "音仁（LV）「ニン」",
      "definition_type_code": 215,
      "definition_type_name": "音注声点有_類音注等",
      "remarks": "広韻「如鄰切」..."
    },
    ...
  ]
}

Next, the content of the column names will be explained.

Please note that this remarks column stores the Compiler's Remarks (annotations by the database creators).

The remarks column provides the following types of information:

• Additional context: Supplementary background or related information that aids in understanding the Myōgishō's entries.
• Scholarly observations: Philological, linguistic, or other expert perspectives on specific descriptions, including references to previous research.
• Results of textual collation: Findings from comparisons with variant manuscripts or related materials, and textual interpretations based on these collations.
• Source investigations: Results and considerations regarding the textual sources of the Myōgishō's entries, including references to findings from previous studies.

These remarks are each associated with one of the following specific parts of a Myōgishō Entry:

• A specific definition_element (an individual component of the Definition (Original Glosses)): This refers to a distinct element within the Myōgishō's original annotation for an Entry (such as a particular Note on Character Form, Phonetic Gloss, Semantic Gloss in Chinese, or Japanese Native Reading (*wakun*)), as itemized in the krm_notes file.
• Or the Headword: The main character(s) of the Entry.

In essence, the remarks column serves to provide specialized, supplementary information from the database compilers, enabling a deeper understanding and facilitating further research that goes beyond what can be gleaned from the Myōgishō's main text and original glosses alone.

Headwords in the Myōgishō can consist of single Chinese characters or multiple Chinese characters (multi-character compounds). The krm_headword_chars data file provides a list of all constituent characters that form these Headwords from the Myōgishō. The characters are ordered according to the sequence of Entries (items) in the Myōgishō and then by the order of appearance of characters within each Headword.

In the Myōgishō database, the primary data file krm_main, the krm_notes file (containing data for Compiler's Remarks), and the krm_wakun file (Japanese Native Reading (*wakun*) data) are all structured on an Entry-by-Entry basis. Consequently, for Headwords composed of multiple Chinese characters, any character subsequent to the first cannot be directly referenced from these particular data files.

To search for Headwords from the Myōgishō character by character, display their original manuscript images, or perform analyses at the individual Chinese character level, a complete list of all constituent characters of the Headwords, including those from the second character onwards in multi-character compounds, is necessary.

The krm_headword_chars data file was created for this purpose. This data is provided in TSV and JSON formats. Each row (or record) corresponds to a single constituent character of a Headword and includes information such as: the sequential ID of the Headword (single or multi-character) in the Myōgishō to which the constituent character belongs (hanzi_id); the ID of the Myōgishō Entry to which this character's Headword belongs (entry_id); the order of the character within its Headword (character_order); the character itself (constituent_char); the file name of the individually cropped image for the character (img_file_name); and location information for that character in both the Kazama and Tenri editions (kazama_location_id, tenri_location_id). This enables information access at the individual character level while allowing linkage with Entry-based data files such as krm_main.

The column names and their descriptions for krm_headword_chars are as follows:

This data file is derived by extracting Japanese Native Readings (*wakun*) from the KRM.tsv file (an older version of krm_main) of the Myōgishō database, organizing variant forms of these wakun, and adjusting their correspondence with variant characters (*itaiji*).

Collation notes and source investigations related to wakun are documented in the krm_notes file (which contains data for Compiler's Remarks), so they are omitted here.

In some wakun entries, different phonetic readings are presented side-by-side as annotations. For example, the wakun "マサル" (masaru) is assigned to the Hanzi (Chinese character) "倍" (bai), but "ス" (su) is written in small katakana to the right of "ル" (ru) as an additional note. This indicates that the wakun "マス" (masu) is also noted in addition to "マサル" (masaru).

Since information from the JapanKnowledge version of the Nihon Kokugo Daijiten (Second Edition) will be added to the wakun data, it is necessary to accommodate cases where variant forms of wakun are presented together.

The correspondence with variant characters (*itaiji*) has been adjusted because Headwords in the Myōgishō sometimes indicate such variants. For example, the wakun "ヤツカレ" (yatsukare) appears in the Definition (Original Glosses) for the Headword(s) "㒒／僕". The wakun "ヤツカレ" is a Japanese Native Reading for "僕" (boku) and simultaneously for "㒒". The relationship between standard and variant forms such as "爲" and "為", or "來" and "来" is handled similarly.

The JapanKnowledge version of the Nihon Kokugo Daijiten (Second Edition) has a "Notation" (表記) field that includes Hanzi (Chinese character) notations from the Myōgishō; this adjustment is a measure to ensure correspondence with that resource.

To explicitly indicate that these are the filenames after the specification change in March 2025, lowercase "krm" was used instead of uppercase "KRM," resulting in the names krm_wakun.tsv and krm_wakun.json.

The comparison of the new and old column names is as follows:

Remarks have been omitted as this type of information is now consolidated in the krm_notes file (data for Compiler's Remarks).

Next, the content of the column names will be explained.

The Definition (Original Glosses) in the Myōgishō is composed not only of Japanese Native Readings (*wakun*) but also other elements such as Notes on Character Form, Phonetic Glosses, and Semantic Glosses in Chinese.

The KRM_definitions.tsv file, which was previously released, provided these elements of the Definition (Original Glosses) categorized by type and ordered by their appearance. This dataset was created to facilitate collation and source investigation.

Following the specification changes implemented in March 2025, the data and functionalities previously found in KRM_definitions.tsv have been integrated into the krm_notes file (which contains data for Compiler's Remarks and detailed analyses of these original gloss elements). Consequently, a separate detailed explanation for krm_definitions is omitted here.

The Phonetic Glosses in the Kanchi-in manuscript of the Ruiju Myōgishō (hereafter Myōgishō) include Fanqie spellings (反切), Similar sound notes (類音注, ruion-chū), and Kana glosses (仮名注, kana-chū). These are often accompanied by Tone marks (*shōten*). As a database for Sino-Japanese character pronunciations, the "Database of Historical Sino-Japanese Readings" (abbreviated as DHSJR), developed by Professor Katō Taitsuru and others, offers exceptionally rich content. Its specifications are also publicly available in detail. We are currently considering releasing data aligned with the DHSJR specifications.

The DHSJR defines a data structure with 23 column names.

To facilitate linkage with the Myōgishō data included in HDIC, it is necessary for HDIC to assign unique column names to its own data files and to establish Primary Keys and Foreign Keys for interoperability between HDIC's internal data files.

For this purpose, pronunciation_id (音注ID) has been set as the Primary Key, and definition_seq_id (注文ID) as the Foreign Key.

Since the Myōgishō features diverse formats for its Phonetic Glosses, a classification field named annotation_format (音注型) has been established to categorize them.

While DHSJR uses Japanese column names, HDIC employs English ones. Therefore, for data processing convenience within HDIC, English column names have been adopted.

The current draft, with English and Japanese explanations side-by-side, is as follows. The Japanese explanations are those stipulated by DHSJR. The English explanations are formulated to facilitate correspondence with HDIC. This is a provisional measure until official English explanations are released by DHSJR.

HDIC's original column names are indicated in bold.

The material_location is indicated in the format: K + Volume (2 digits) + Kazama Edition Page (3 digits) + Line (1 digit) + Segment (1 digit). For example, K0201474 indicates an appearance in Volume 2, Page 14, Line 7, Segment 4.

Currently, this is under consideration in the case study "Linkage with DHSJR," which should also be consulted.

This file compiles links to the images of the Kanchi-in manuscript of the Ruiju Myōgishō that are publicly available in the National Diet Library Digital Collections.

This data cross-references the locations within the Kanchi-in manuscript of the Ruiju Myōgishō with their corresponding URLs in the National Diet Library Digital Collections. The file name is krm_ndl.tsv.

This data indicates the volume name, radical character, Kazama edition page number, Tenri edition page number, and the URL for the National Diet Library Digital Collections.

Next, a sample of the beginning of the data is shown below.

We plan to make improvements to facilitate easier integration with krm_main.