Japonic — The Japanologist

The Japanologist — Old Japanese song to the Sino-Japanese dictionary tradition.

The Japanese lane: the ONCOJ corpus and lexicon, EDRDG’s dictionaries, HDIC and Unihan shared with the Sinologist, and the kaikki ojp extract riding wiktionary-recon.

The desk spans two eras of Japanese. Old Japanese (ojp) is live through ONCOJ and its lexicon; the modern lane is Aozora Bunko (青空文庫) — the public-domain library at paragraph grain, 17,121 works live below. Its scope is PD text ONLY: discovery excludes the in-copyright works before any file is touched. Ruby (振り仮名) readings ride as annotations over the base text, never spliced into it; the 底本 colophon is carried as document metadata.

New here? The Quickstart sets up the library in minutes.

The shelves

A source wears every desk it serves — these eight answer this desk. Holdings are read live from the catalog and dated; a shelf with nothing synced yet says so.

Languages on this desk (live doc-or-entry counts as of 19 September 2026): jpn 287,405 · zho 102,998 · lzh 63,807 · ojp 11,392 · sga 6,564 · gem-pro 5,717 · gmw-pro 5,551 · sla-pro 5,431 · ine-pro 1,905 · iir-pro 799 … and 6 more (nabu axis japonic lists all).

The desk’s instruments

• The Japanese lane: the ONCOJ Old Japanese corpus and its lexicon, EDRDG’s dictionaries (JMdict / KANJIDIC family — these feed nabu char’s kun/on readings and the Jōyō/JLPT/frequency lines), HDIC and Unihan shared with the Sinologist, KRADFILE’s radical-component index, and the kaikki ojp extract on wiktionary-recon.
• The kyūjitai↔shinjitai reform fold (P38-4 + P38-r1): modern (jpn) search folds old-form, new-form and merged spellings onto one skeleton — matching modern reading habits, onto the SAME skeleton the Sinologist’s Han fold uses, so Aozora and kanripo meet. Two lanes: the held Unihan kJinmeiyoKanji name-kanji pairs, PLUS a KANJIDIC2-jōyō lane (744 fold entries) that lands the high-frequency non-name pairs 學/学, 體/体, 醫/医, 觀/観 and admits the famous merges (辨/瓣/辯 → 弁). nabu search --exact is the glyph-literal escape hatch when you need 弁 apart from 辨/瓣/辯. nabu char cross-references each authoritative jinmeiyō pair’s kyūjitai/shinjitai (conventions §9).

Working the japonic desk

The generic axis surfaces — every desk answers to these, in working order (enable once, sync, then query):

nabu enable japonic               # first time: put this desk's shelves in this box's profile
nabu sync japonic                 # fetch/refresh the desk's enabled members
nabu list --axis japonic          # the shelf census, this desk only
nabu axis japonic                 # the desk card: members, holdings, gold coverage
nabu search WORD --axis japonic   # a query scoped to this desk's shelves

This desk’s own surfaces:

nabu char 天                           # the character card, with KANJIDIC2 readings and desk codes
nabu char 國                           # the reform cross-reference — a kyūjitai names its shinjitai 国
nabu search 学                         # folds to 學 (and 弁 finds 辨/瓣/辯) — modern reading habits across jpn/lzh
nabu search 弁 --exact                 # the glyph-literal escape hatch — only the stored 弁, not the merged olds
nabu search --char-component 木 --strokes 8-12  # KRADFILE component containment on the Han corpus
nabu search --radical 75 --axis japonic  # the KangXi-radical filter
nabu show urn:nabu:oncoj:MYS.1.1      # Man'yōshū 1.1 — romanization layer per ONCOJ's design
nabu search 吾輩は猫である --lang jpn        # Sōseki's opening line, top hit on the Aozora reading shelf

Ask your model

With the MCP server connected, this desk answers conversational research questions. Each example ran live against this library:

• “Old Japanese lines about mountains?” → nabu_search yama (lang: ojp) — Kojiki kayō and Man’yōshū lines (unebwi yama — Mount Unebi) — the ONCOJ shelf stores romanized Old Japanese, so type romaji.

Terminal setup

• Kana / kanji (ojp): ONCOJ carries romanization and original layers by its own design (man’yōgana rides the annotations, not the romanized KWIC). Install the Noto CJK casks plus Jigmo, and keep iTerm2’s ambiguous-width toggle off. nabu models CJK cell width, so aligned columns stay aligned.
• Gaiji (Aozora): Aozora’s not-yet-encoded characters resolve at parse — JIS X 0213 kuten and explicit U+XXXX notations map mechanically through config/jis0213; component-description-only notations stay verbatim loud sentinels. On the render side --display reading runs the four-rung gaiji ladder (faithful glyph → IDS composition → marked ⌈substitute⌉ → ⬚ box), and --display diplomatic keeps the refs byte-honest; the IDS rung, empty for kanripo, is live machinery here since Aozora’s gaiji are largely IDS compositions (docs/display.md §1a).

The full guidance, per script, is on the display page.

One of the twenty-five research desks; the flat shelf map is The Library and the reasoning is docs/axes.md.