MangaBaka Japanese Romanization Style

Core Romanization Style#↑

This style guide defines how to consistently romanize and style Japanese-language titles for metadata purposes.
The goal is to provide a standard that is easy to read, type, and apply across various databases, while staying true to linguistic norms.

Base Romanization#↑

Modified Hepburn Romanization

At the base we use Modified Hepburn as the romanization, meaning;

• は, へ, and を as particles are always romanized as wa, e, and o.
• ん is always romanized as n, or as n' — with an (straight) apostrophe — when the ん is followed by a vowel or y.
• ず and づ are both romanized both as zu as they are phonetically identical.

Adjustments made to Modified Hepburn

We have made the following adjustments on Modified Hepburn for ease of use, clarity, and community recognition;

• Represent long vowels using full-length spellings (e.g., ou, oo, uu, aa, ee, ii) instead of using macrons (e.g., ō, ū, ā ē, ī).
• Represent the sokuon (っ) with a “ch”-syllable as cch instead of the “strict” Modified Hepburn tch, however the tch romanization shall be used in the alternative title.

Regarding loanwords, non-native names, and native names

• Render Non-assimilated or identifiable loanwords, and non-Japanese names in their established source equivalents whenever identifiable (e.g., スキル → Skill, ダンジョン → Dungeon, クリス → Chris).
• Retain clipped slang, or fully lexicalized loanwords in everyday Japanese, where the Japanese phonology or sociocultural meaning is distinct (e.g., ギャル → Gyaru, バイト → Baito, ラブホ → Rabuho).
• Always romanize native Japanese personal and place names using Japanese romanization rules, even if an English exonym is widely known (e.g., 東京 → Toukyou, 大阪 → Oosaka).

Capitalization#↑

As we are romanizing titles of Japanese literature, we use the following capitalization style:

1. Capitalize all nouns, verbs, adjectives, and adverbs.
2. Keep lowercase all particles, copulas (including their inflections or grammar bridges), and conjunctions, even when they appear at the end of a sentence.

• Example: Ore wa Maou de aru (I am the Demon King).
• Example: Akuyaku Reijou de atta node (Because I was a Villainess).

Word Spacing, Fusions, and Hypenation#↑

Keep all words—including particles, copulas (and their inflections), and conjunctives—separate, except for the following specific cases:

• Conjunctives & Grammar Bridges are kept fused and lowercase when functioning as a single unit rather than a sequence of distinct parts (e.g., なのに → nanoni, でも → demo, ですが → desuga).
• Morpheme ん (from の) is fused with the following element to prevent a loose letter in the title romanization (e.g., んです → ndesu, んだけど → ndakedo).
• Auxilaries are always fused to the main verb or adjective—whether via the -te form or the stem (e.g, 見てください → Mitekudasai, 可愛すぎる → Kawaisugiru, してない → Shitenai).
• Fossilized adverbs are kept fused as a single unit (e.g., いつの間にか → Itsunomanika, なんとなく → Nantonaku, ひょっとして → Hyottoshite, もしかしたら → Moshikashitara)
• Lexicalized Compounds & Generic Groups are fused when they form a single dictionary unit, a profession, or a general category. (e.g., 女の子 → Onnanoko, 文芸部 → Bungeibu, 駄菓子屋 → Dagashiya).
• Thematic Portmanteaus and Coined Terms are treated as single, fused lexical units, even if they are not found in standard dictionaries. (e.g., 異世改活 → Iseikaikatsu, 死に戻り → Shinimodori).
• Reduplications (Echo) are fused into a single word when a word is repeated for emphasis or as onomatopoeia (ムリムリ → Murimuri, ワクワク → Wakuwaku, フワフワ → Fuwafuwa).
• Leading honorifics o- and go- are fused without a hyphen and capitalized (お姫 → Ohime, ご主人 → oshujin).
• Hyphenation is used to protect the identity of a Proper Noun or to mark functional/relational boundaries that are not part of a compound noun’s core identity.

Expanded Explanation and Examples#↑

Auxiliaries#↑

Japanese auxiliary elements are always fused with the preceeding verb or adverb, to modify aspect, voice, or nuance, and can be rougly divided in three groups.

• Auxiliary verbs (e.g. -kuru, -iku, -shimau, -ageru, -morau, -kudasai, etc.) typically follow the -te form of a main verb and express sequence, aspect, completion, or request.
• Contracted auxiliaries (e.g. -iru → -teiru/-teru, -ita → -teita/-teta, -iyou → -you, -mashita, etc.) represent shortened or conjugated forms of auxiliary verbs.
• Adjectival auxiliaries (e.g. -sugiru, -yasui, -nikui, -rashii, etc.) to express degree, desire, or ease/difficulty.

Thematic Portmanteaus and Coined Terms (Zougo)#↑

Many Japanese titles feature unique, coined compounds (造語, zougo) created by blending or clipping multiple existing words into a single keyword. These function as a specific "brand name" or thematic identity for the series.

For romanization, these are to be treated as single, fused lexical units, even if they are not found in standard dictionaries. This preserves the "wordplay" and distinguishes the term from standard descriptive modifiers. If the component parts are clipped or intended to be read as a single concept, they should always be fused.

Hyphenation#↑

Honorifics, Titles, and Household/Clan names

• 直美さん is romanized as Naomi-san (name + honorific)
• お兄ちゃん is romanized as Onii-chan (kinship + “honorific”)
• 田中先生 is romanized as Tanaka-sensei (name + title)
• 魔王様 is romanized as Maou-sama (demon king + honorific)
• 天使様 is romanized as Tenshi-sama (rank + honorific)
• 聖騎士さま is romanized as Seikishi-sama (rank + honorific)
• 京兼家 is romanized as Kyougane-ke (household/clan name)
• 篝家 as Kagari-ke (household/clan name)

Titles that are a standalone noun title, such as 陛下, are always romanized separately and without a hyphen; 国王陛下 → Kokuou Heika.

• お姉さん / おねえさん is romanized as Oneesan (elder sister)
• お兄さん / おにいさん is romanized as Oniisan (elder brother)
• 叔父さん / おじさん is romanized as Ojisan (uncle or middle-aged man)
• 叔母さん / おばさん is romanized as Obasan (aunt or middle-aged woman)
• 奥さん is romanized as Okusan (wife)
• お嫁さん is romanized as Oyomesan (bride + polite prefix)
• おっさん is romanized as Ossan (old man)
• 皆さん is romanized as Minasan (everyone)

• 神様 is romanized as Kamisama (god)
• 姫様 is romanized as Himesama (princess)
• 旦那様 is romanized as Dannasama (husband or master of x)
• お嬢様 is romanized as Ojousama (young lady)
• ご主人様 is romanized as Goshujinsama (master)
• おひとり様 is romanized as Ohitorisama (a person alone)

• れい姉ちゃん is romanized as Rei-neechan

Multi-word Expressions

These words function as a single semantic or grammatical unit, particularly modifier–noun phrases, semi-lexicalized constructions, or commonly recognized set expressions. This is especially common when the first element is a katakana loanword or roman-letter term or abrivation, where the hyphen helps clarify word boundaries and preserves the perception of a single semantic unit

• Ｓランク is romanized as S-Rank (Considered as a fully loaned word, so it is capitalized as such).
• バイト先 is romanized as Baito-saki (先 is fused with a hyphen due to the katakana loanword; location).
• ポーション師 is romanized as Potion-shi (師 is fused with a hyphen due to the katanana loanword; master).
• Ｓ級 is romanized as S-kyuu (級 is fused with a hyphen due to the Roman letter term; class).
• 僕たち is romanized as Boku-tachi (たち is always fused with a hyphen; pluralization).
• 環状戦 is romanized as Kanjou-sen (級 is always fused with a hyphen; battle/war/fight).
• 契約婚 is romanized as Keyaku-kon (“contract marriage” hyphenated due to the clipped 婚約)

Arabic numerals and Their Counters and Modifiers

Arabic numerals followed by counters are always fused with a hyphen to the counter.
The counter uses its regular numeral-attached reading; the standard form used in dictionaries and this is most often On’yomi, but not always.

• ２人 is romanized as 2-nin (and not as 2-ri using the -ri from “futari” special reading).
• １０００体 is romanized as 1000-tai.
• 10年 is romanized as 10-nen.
• 8月31日 is romanized as 8-gatsu 31-nichi.
• ８号 is romanized as 8-gou.

• １０年間 is romanized as 10-nenkan as number + counter (年) + fused modifier (間).
• ２２２日目 is romanized as 222-nichime as number + counter (日) + fused modifier (目).
• ３１番目 is romanized as 31-banme as number + counter (番) + fused modifier (目).
• ６歳上 is romanized as 4-saijou (number + counter (歳) + fused lexicalized (上)).
• １０歳下 is romanized as 10-saishita as number + counter (歳) + fused lexicalized (下).
• ８万枚 is romanized as 8-manmai as number + number unit (万) + fused counter (枚).
• ３億円 is romanized as 3-okuen as number + number unit (億) + fused counter (円).
• １０万年 is romanized as 10-mannen as number + number unit (万) + fused counter (年).

When 後 or 前 follow an Arabic-numeral counter construction, they are treated as unit-final temporal suffixes and read as -go and -mae respectively.

• １００日後 is romanized as 100-nichigo
• １００年後 is romanized as 100-nengo
• １００日前 is romanized as 100-nichimae
• １００年前 is romanized as 100-nenmae

Note: While -zen is technically a Sino-Japanese reading (On-yomi) that matches -go, in modern Japanese, -zen as a suffix is almost exclusively formal, historical, or academic.

• 5キロ is romanized as 5 Kilo (Capitalized because it is used in a title).
• 5キロ減 is romanized as 5 Kilo-ge (number + counter + modifier).
• 2メートル is romanized as 2 Meter (Capitalized because it is used in a title).
• 2メートル越え is romanized as 2 Meter-koe (number + counter + modifier).

Romanization of Native Kanji Numerals and Their Counters#↑

Native anji numerals followed by counters are always fused without a hyphen to the counter, any modifier present is also fused directly to the counter.
This ensures that special numeral–counter readings or suppletive readings are reflecting natural lexicalization and are correctly preserved in romanization.

• 二人 → Futari (and not Ninin)
• 一年 → Ichinen
• 四年生 → Yonensei
• 八男 → Hachinan

• 二日前 → Futsuka-mae
• 数日前 → Suujitsu-mae
• 三年後 → Sannen-go

Romanization of Quotations#↑

1. Japanese quotation marks 「文章」, and 『文章』 are to be romanized to double straight quotation marks "text".
2. Any straight or smart quotation marks “文章” are to be romanized to double straight quotation marks "text".
3. If the lenticular brackets 【文章】 is clearly used as quotation then it also should be romanized to double straight quotation marks "text".

Romanization of Other Special Characters#↑

1. Full-width square brackets ［文章］ or lenticular brackets 【文章】 when unclear if they are for quotation, are to be romanized to normal square brackets [text].
2. Double Angle Brackets 《文章》 or the Much Less/Greater-Than ≪文章≫ are to be romanized to double straight quotation marks "text"
3. Any other (unicode)symbol present (×,♥,♡,★,☆,○,♂,♀,ect.) shall be copied as-is but there shall nearly always (overruled by author/cover styling) be a space around these symbols!

Suffixes#↑

Suffixes are handled based on their function. We distinguish between suffixes that create a new "Identity" (Internal) and those that denote a "Relationship or Time" (External).
Regarless, any suffix after a personal- or placename, or non-embedded identifiable loanword (e.g., they are romanized to their loaned origin) are always hypenated.

Conceptual & Identity Suffixes (Direct Fusion)

Use this for suffixes that “transform” the base noun into a new concept, class, or trope. These create a single Identity Pillar.

For example but not limited to;

• 婚 (-kon, “marriage suffix”): 契約婚 → Keiyakukon (Contract Marriage), 転生婚 → Tenseikon (Reincarnated Marriage), or 速婚 → Sokkon (Quick Marriage).
• 化 (-ka, “-ification suffix”): 猛獣化 → Moujuuka (Beastification), or 悪役化 → Moujuuka (Villainization).
• 屋 (-ya, “shop suffix”) 本屋 → Hon'ya (Book Store), パン屋 → Panya (Bakery), or ゲーム屋 → Game-ya (Game Store).

Attribute & Bound Suffixes (Hyphenated)

Use this for suffixes that describe a specific attribute, lifestyle, or action performed upon/by the noun. These often undergo Rendaku (voicing), which must be reflected in the romanization.

Suffix-like Content Designators

These suffix-like content designators (編, 譚, 章, 録, etc.) have their grammatical behavior shifts depending on what precedes them;

• When directly attached to a single noun or compound, they act as suffixes and are hyphenated to the preceding word (冬編 → Fuyu-hen, 妖怪譚 → Youkai-tan).
• When following a phrase or clause-like structure, they function as independent nouns and are separated and capitalized (俺が最強になった編 → Ore ga Saikyou ni Natta Hen, 斬吸血鬼譚 → Zan Kyuuketsuki Tan).

Hyphens mark tight morphological fusion, while separation marks syntactic or semantic independence.

Ordinal Number (第) Prefix#↑

The prefix Dai (第) transforms a cardinal number into an ordinal (e.g., from “seven” to “seventh”). To maintain consistency and scannability, we use two mirrored formats depending on the numeral type.

1. Native (Kanji) Numeral Sequences:

• Rule: Fuse Dai directly to the number word. The following “counter” or “noun” is always separated with a space and capitalized.
• Logic: This treats the rank as a single descriptor (e.g., Dainana = Seventh).

1. Arabic (Digits) Numeral Sequences:

• Rule: Attach Dai to the digit using a hyphen. The following “counter” or “noun” is always separated with a space and capitalized.
• Logic: The hyphen acts as a visual anchor between the Japanese text and the Western digit, preventing the number from “floating.”

Descriptor (Intensifier and Status) Prefixes#↑

Many Japanese titles utilize prefixes to act as Intensifiers (Power Words) or Status Modifiers. These “say” something about the subject that follows. To keep the romanization modular and clean, these are always to be romanized separately and capitalized.

This approach prevents “clutter” when prefixes are stacked and ensures that auxiliary elements (like verbs) remain attached to the subject without creating an unreadable “wall of text” or using excessive hyphenation.

1. Intensifier Prefixes: These amplify the following subject.,

• Examples are but not limited to; 超 (Chou), 激 (Geki), 鬼 (Oni), 神 (Kami), and 極 (Kyoku).

2. Status Prefixes: These define the current state of the subject.

• Examples are but not limited to; 元 (Moto), 新 (Shin), 生 (Nama).

Lexicalization: If a word with one of these prefixes is a fixed dictionary entry where the meaning has shifted away from the literal parts (e.g., 新聞 → Shinbun meaning Newspaper and not “New News”) or are, it is a single compound and must be fused. The clipped compounds 元カレ → Motokare and 元カノ → Motokano also fall under here, and are always romanized fused.

The 大 Prefix

The prefix 大 (Dai) requires extra attention. Many words using Dai are fully lexicalized (the “Great” is invisible), while others are used specifically as Titles or Epithets for emphasis in fiction.

1. Fuse if the term is a common dictionary noun or a shifted compound.
   If the Dai acts as a permanent part of the word's definition, keep it fused.

2. Separate if Dai is a descriptive Title or Fantasy Epithet.
   In Light Novel and Manga titles, Dai is often used as a rank (like “Arch-” or “High-”). In these cases, separate it to highlight the character's status.

3. With Tie-Breakers
   In extreme edge cases where both fused and split versions carry meaning (e.g., 大天使), and the author's specific intent is unknown, always default to the fused dictionary version.

Dual-Role Connector 兼 (Ken)#↑

兼 (ken) is a conjuntion meaning “and” or to act as a the “/” or “-cum-” in English so it is always romanized in lowercase as ken with spaces around it (e.g., Ａ兼Ｂ → A ken B).

Onomatopoeia and Onomatopoeic Sequences#↑

Onomatopoeic words (words representing a sound effect) are always romanized literally. They must never be replaced with English equivalents or approximations (e.g., use Pan for パン, not Clap).

• Sequences: In a sequence of different sound effects, each individual sound is treated as a standalone, capitalized word and is not fused (e.g., ぺチパンパチン → Pechi Pan Pachin).
• Reduplication: If a single sound is repeated to form a lexicalized rhythmic word, the sounds are fused as one word (e.g., ぺチぺチ → Pechipechi).

Particles, Copulas, and Conjunctions#↑

Particles and Copula#↑

Conjuntions#↑

Copula Grammar Bridges#↑

Note: These are only some examples, be aware of any possible inflections of aru like “de arou” and suru like “shimashita”!
Note: Any auxiliary that follows a bridge "sticks" to the vocal part (e.g., “to shitekuru”)

As we always romanize as da to two separate particles, proper chaining can occur.

Separte or Conjunction Special Cases#↑

ので

Romanize ので single conjunctive node when:

• It functions as a single conjunctive particle meaning “because / so / since”.
• Commonly follows verbs, adjectives, or copulas to connect cause and result.
• Very often written as ので、 — with a comma — when starting a new clause.
• Example: 困ったので帰ります → Komatta node Kaerimasu (“Because I got into trouble, I’ll go home”).

Romanize ので as two particles no de — in extremely rare cases for titles — when:

• の no is the genitive or nominalizing particle, and で de marks the location or instrument (with / by / at).
• The meaning is literal “of X + with / by / at” rather than “because / so / since”.
• If you can replace の no with な na or remove の no entirely and the phrase still makes sense, it is two particles.
• Example: 本の写真で説明する → Hon no de Shashin de Setsumei Suru (“Explain with pictures of the book”).

のに

Romanize のに as single conjunctive noni when:

• It functions as a single conjunctive particle meaning “despite / even though”.
• Commonly follows i-adjectives, verbs, or noun+particle structures.
• Very often written as のに、 — with a comma — when starting a new clause.
• Example: みんな疲れているのに頑張っている → Minna Tsukareteiru noni Ganbatte Iru (“Even though everyone’s tired, they’re pushing on”).

Romanize のに as two particles no ni — in extremely rare cases for titles — when:

• の no is the genitive or nominalizing particle, and に ni marks location or direction (at / in / on).
• The meaning is literal “of X” + “at / in / on” rather than “despite / even though”.
• Example: 君のに夢を置いた → Kimi no ni Yume o Oita (“I placed a dream in yours”).

なので and なのに

The sequences なので and なのに are handled differently depending on what precedes them.

Romanize なので or なのに as single conjunctive nanode or nanoni when they follow a “Concrete Noun” or a “Na-Adjective”:

• Functioning as a bridge meaning “because / since” or “even though / despite that” respectively.
• Concrete Nouns: Sensei nanode, Manga nanoni, Himitsu nanode.
• Na-Adjectives: Suki nanoni, Kirei nanode, Shizuka nanode.

Romanize なので or なのに as a copula + conjunctive structure na node or na noni when they follow “Formal Noun”:

• This prevents these “weighty” Formal Nouns from being swallowed by the grammar.
• See the list down below for the these Formal Nouns (abstract concepts) and are generally written in kana only.
• In this structure, the sequence is a complete bridge so these Formal Nouns are lowercased!

だって

Romanize だって as datte (conjunctive particle) when:

• It functions as a single conjunctive particle meaning “even / after all / because”.
• It usually appears before nouns or at the beginning of a sentence.
• It can often be replaced with も mo without changing the meaning.
• Example: 女の子だって遊びたい → Onnanoko datte Asobitai (“Even girls want to play.”)
• Test: Onnanoko mo Asobitai

Romanize だって as da tte (copula + partcile) when:

• It is literally da (copula “is/was”) followed by tte (quotative/colloquial topic particle).
• It usually follows a noun or adjective to state “X is” + quoted/topic element.
• You can replace だ da with です desu or だった datta and the meaning stays the same.
• Example: 「犯人は君だ」って言った → Hannin wa Kimi da tte Itta (“I said ‘The culprit is you’”)
• Test: Kimi desu tte itta

では

If では follows a noun or phrase and directly governs a predicate (especially before negation, contrast, or evaluation), they are always as two separate elements de wa.
Any standalone adjective or adjectival noun following では, are always separated and capitalized as independent words;

• ではない → de wa Nai
• ではなく → de wa Nakku
• ではいられない → de wa Irarenai
• では済まない → de wa Sumanai

If では appears sentence-initially or functions as a discourse marker meaning “well then / in that case”, treat it as lexicalized conjunctive dewa.

な + ん + Copula Sequence#↑

The な + ん + copula sequence is treated as a single grammatical bridge:

1. The morpheme ん (from の) fuses directly with the following copula or conjunctive as per the core style (e.g., んです → ndesu).
2. The initial な is then fused to create a single lowercase unit which reflects the function of a fixed explanatory phrase (“it is that…” / “the reason is…”).

Ensure that it nan comes from the particles な + の, and not from 何 (“What”) as it would be otherwise separated and capitialized (e.g, 何ですか → Nan desu ka as for “What is it?”).

じ + Adjective Sequences#↑

When the copula ja is followed by a standalone adjective or adjectival noun, they are kept as separate and capitalized as independent words. This preserves the semantic weight of the negative or prohibitive word.

1. Ja is a contraction of the cluster “de wa”, so it is treated as a copula and as per the core style copula are always written separately and in lowercase.
2. The negative adjective or adjectival noun (e.g. nai, dame) is not fused to ja and capitalized as independent words.

Note that じゃん is always romanized separate and in lowercase as jan, as it is a colloquial contraction of ja Nai (and not from ja + no).

Noun + Suru Romanization#↑

When Suru or Suru inflected forms (Shita, Shite, Sareta, Saseru, etc.) attaches productively to a (productive) noun write it separately and capitalize it.

These are simply nouns that gain verbal force through the addition of suru; dictionaries may list them as so called “suru”-verbs but morphologically they are still a noun + suru.

By contrast, fixed or lexicalized suru-verbs such as 恋する (Koisuru), 愛する (Aisuru), or 画する (Kakusuru) are romanized fused, since suru is no longer detachable but part of the verb stem.

Dictionaries may list them as normal suru-verbs, found by its suru base form, and they should be treated as a verb.

Auxiliary verbs or contracted auxilaries are directly fused to the suru-form without a hyphen

Title Examples (for test cases)#↑

Title 1 MB:359750#↑

異世界グルメで成り上がり無双　～山に追放されたので、のんびりキャンプを楽しんでいたらいつの間にか強くなっていて、王侯貴族や実力者たちが俺を放っておいてくれません。一方、俺を追放した貴族たちは破滅が始まる～
Isekai Gourmet de Nariagari Musou: Yama ni Tsuihou Sareta node, Nonbiri Camp o Tanoshindeitara Itsunomanika Tsuyoku Natteite, Oukou Kizoku ya Jitsuryokusha-tachi ga Ore o Houtte Oitekuremasen. Ippou, Ore o Tsuihou Shita Kizoku-tachi wa Hametsu ga Hajimaru

Title 2 MB:8670#↑

信じていた仲間達にダンジョン奥地で殺されかけたがギフト『無限ガチャ』でレベル9999の仲間達を手に入れて元パーティーメンバーと世界に復讐＆『ざまぁ！』します！
Shinjiteita Nakama-tachi ni Dungeon Okuchi de Korosarekaketa ga Gift “Mugen Gacha” de Level 9999 no Nakama-tachi o Te ni Irete Moto Party Member to Sekai ni Fukushuu & “Zamaa!” Shimasu!

Title 3 MB:30392#↑

俺が好きなのは妹だけど妹じゃない
Ore ga Suki na no wa Imouto dakedo Imouto ja Nai

Title 4 MB:46802#↑

異世界転生して魔女になったのでスローライフを送りたいのに魔王が逃がしてくれません
Isekai Tensei Shite Majo ni Natta node Slow Life o Okuritai noni Maou ga Nigashitekuremasen

Title 5 MB:28009#↑

農民関連のスキルばっか上げてたら何故か強くなった。
Noumin Kanren no Skill bakka Agetetara Nazeka Tsuyoku Natta.

Title 6 MB:9333#↑

真の仲間じゃないと勇者のパーティーを追い出されたので、辺境でスローライフすることにしました
Shin no Nakama ja Nai to Yuusha no Party o Oidasareta node, Henkyou de Slow Life Suru Koto ni Shimashita

LLM Ruleset and prompts#↑

romanization_style:
  name: "MangaBaka Romanization Style"
  version: "1.2"
  base_system: "Modified Hepburn (customized)"
  macrons: false
  long_vowels: "Always written fully (ou, oo, uu, aa, ee, ii)"
  sokuon_rule: "Represent っ before ch as cch (e.g., Maccha instead of Matcha)"
  n_rule: "ん is romanized as n. Use n' when ん is followed by a vowel or y to avoid ambiguity (e.g., Ten'i, Kon'yaku)."

  capitalization: >
    Capitalize all nouns, verbs, adjectives, adverbs, and auxiliary verbs when written separately.
    Keep all particles, copulas, and copula inflections or contractions lowercase. Copula forms
    (da or desu and all inflections) are never capitalized, even when sentence-final.

  suru_taxonomy: >
    Distinguish suru-forms as follows:
    (1) Productive noun + Suru constructions are written separately, with Suru
        and its inflections capitalized (e.g., Benkyou Suru, Tensei Shita).
    (2) Verbs ending in -su are true verbs, not noun + Suru; their inflected forms
        fuse normally with auxiliaries (e.g., Mezashite, Otoshita).
    (3) Fixed or lexicalized suru-verbs listed as single verbs in dictionaries are
        always fused, since Suru is part of the verb stem (e.g., Koisuru, Aisuru, Kakusuru).

  auxiliaries: >
    All auxiliary verbs and auxiliary-like forms are fused to the main verb
    or adjective, whether attached via the -te form, the stem, or as
    contracted forms (e.g., Natteshimatta, Kawaisugiru, Shiteita).

  te_form_chains: >
    When two or more verbs appear consecutively in the -te form, each lexical verb
    is written separately. Any auxiliaries that follow are fused to the immediately
    preceding final -te form. Examples are:
    - 放っておく → Houtteoku
    - 放っておいてくれ → Houtte Oitekure
    - 放っておいてくれません → Houtte Oitekuremasen

  particles: >
    Particles are always written separately and in lowercase
    (e.g., no, ga, o, de, mo, ni, wa).

  conjunctive_particles: >
    Conjunctive particles are fused when functioning as a single lexical unit
    (e.g., nanoni, demo, dakedo, desuga). Node and noni are fused unless they are
    clearly separate particles in a grammatical construction.

  ja_auxiliary: >
    Forms like じゃない and じゃだめ are romanized as ja + Auxiliary
    (e.g., ja Nai, ja Dame). Ja is treated as a particle (from de wa) and
    remains separate and lowercase; auxiliaries are capitalized when
    written separately. Contractions like じゃん are romanized as jan.

  compounds_semantic_binomial: >
    Semantic binomial compounds (X and Y relationships) may be split for
    clarity and consistency (e.g., Oukou Kizoku, Shinou Shoukou).

  compounds_determinative: >
    Determinative compounds (modifier + head noun) are written as separate
    words unless strongly lexicalized (e.g., Kokugai Tsuihou, Kokunai Ryokou).

  lexicalized_idioms: >
    Strongly lexicalized idioms and fixed expressions, including yojijukugo,
    remain fused (e.g., Yuuyuujiteki, Issekinichou, Jakunikukyoushoku).

  lexicalized_adverbs: >
    Lexicalized adverbs and set expressions are written as single words
    (e.g., Itsunomanika, Nazeka, Nanto, Nidoto, Doushitemo).

  native_personal_and_place_names: >
    Always romanize native Japanese personal and place names using Japanese romanization
    rules, even if an English exonym is widely known (e.g., Toukyou, Oosaka).

  loanwords: >
    Render Non-assimilated or uncommon loanwords, and non-native personal or place names
    in their established native-language equivalents whenever identifiable (e.g., Skill, Dungeon).
    Retain loanwords that are clipped slang, phonologically adapted, or fully lexicalized in everyday Japanese,
    where the foreign source is no longer treated as a recoverable identit (e.g., Baito, Rabuho, Gyaru).

  hyphens: >
    Use hyphens with honorifics, personal titles, household or clan names,
    Arabic numbers with counters, and multi-word expressions functioning as
    one unit. Honorifics are hyphenated except for lexicalized kinship terms
    (e.g., oneesan, ojisan) or established dictionary nouns (e.g., kamisama).
    Arabic numbers + counter + modifier use a single hyphen with the modifier
    fused to the counter (e.g., 31-banme) including 後 or 前 which are treated
    as Sino-Japanese unit-final temporal suffixes and read as -go and -zen
    respectively. Counters use their numeral-attached readings. Fully native
    numbers + counter + modifier are fully fused to preserve special readings
    (e.g., nidome, futari).

  titles_and_compound_title_nouns: >
    Distinguish between true titles/honorifics and compound title nouns.
    True titles and honorifics that attach to a person or name are written
    with a hyphen (e.g., Ou-sama).
    Compound title nouns that function as a single lexical noun are fused
    without a hyphen (e.g., Ningyouhime, kamisama, Maoujou).

  bound_suffixes: >
    Bound suffixes that modify the preceding element or when cannot stand alone as noun, verb, or adjective
    without the loss of meaning are fused with a hyphen for clarity (e.g., Oshare-suki, Kami-goroshi, Inaka-gurashi)

  copula_vs_conjunctive_dewa: >
    Distinguish grammatical de wa (copula + particle) from lexicalized
    conjunctive dewa. Grammatical de wa is written separately (de wa),
    while conjunctive dewa meaning “well then / in that case” is written
    fused (dewa), usually at the start of a clause.

  example_validation:
    input: "異世界グルメで成り上がり無双　～山に追放されたので、のんびりキャンプを楽しんでいたらいつの間にか強くなっていて、王侯貴族や実力者たちが俺を放っておいてくれません。一方、俺を追放した貴族たちは破滅が始まる～"
    output: "Isekai Gourmet de Nariagari Musou: Yama ni Tsuihou Sareta node, Nonbiri Camp o Tanoshindeitara Itsunomanika Tsuyoku Natteite, Oukou Kizoku ya Jitsuryokusha-tachi ga Ore o Houtte Oitekuremasen. Ippou, Ore o Tsuihou Shita Kizoku-tachi wa Hametsu ga Hajimaru"

YAML Ruleset (2.0) WIP!!#↑

romanization_style:
  name: "MangaBaka Romanization"
  version: "2.0"
  response_language: "English"
  response_style: "Romanized title in plain text and a short breakdown of the romanization and applied rules, with additional notes on important areas"
  base_system: "Modified Heburn"
  macrons: false
  n_rule: "ん is romanized as n. Use n' when ん is followed by a vowel or y to avoid ambiguity (Ten'i, Kon'yaku, Onna)"
  zu_rule: "ず and づ are always both romanized as zu, no exceptions"
  long_vowels: "Always written out fully (ou, oo, uu, aa, ee, ii) so correct romanization occurs without macrons"
  sokuon_rule: "Represent っ before ch as cch (e.g., Maccha instead of Matcha) essentially always doubling the following letter"

  capitalization: >
    Capitalize all nouns, verbs, adjectives, adverbs. All particles, copulas, copula inflections, 
    and the vocal bridge elements of grammatical bridges are always lowercase, even when sentence-final.

  auxiliaries: >
    All auxiliary elements are fused to the main verb or adjective (via stem or te-form), this includes the vocal element of 
    a gramatical bridge ('Natteshimatta', 'Kawaisugiru', 'Shiteita', 'te shiteiru'). Kudasai is considered a request auxilary 
    for -te form verbs, so is also always fused ('Mitekudasai').

  formal_nouns: ['koto', 'hazu', 'wake', 'mono', 'tokoro', 'tame', 'you']
  formal_nouns_description: "Nouns that are abstract concepts rather than concrete 'things'"
  formal_noun_bridges:
    description: >
      Lowercase these specific visual patterns to ensure grammar stays in the valley. These patterns overrule general 
      noun capitalization and should also prevent lowercasing where these formal nouns are a pillar ('Sonna Koto ni').
    patterns:
      - sequence: "no <formal noun> [na/ni/no/o/ga/wa/desu/da]"
        logic: "Always lowercase the formal noun and the particle/copula that follows."
        example: "'no koto ni', 'no you na', 'no tame no', 'no koto desu'"
      - sequence: "[Verb/Adjective] <formal noun> [na/ni/desu/da/o/ga/wa]"
        logic: "Always lowercase the formal noun and following bridge element."
        example: "'Shiteita you desu', 'Iku wake nai', 'kawaii koto ni'"
      - sequence: "<formal noun> na [node/noni]"
        logic: "Lowercase the formal noun and the conjunction as a bridge."
        example: "'koto na node', 'wake na noni'"
  grammatical_bridges:
    description: >
      In addition to the formal noun bridges, we want to apply the lowercase bridge logic to the following 
      selected bridges, inflections of these vocal elements also count. Any auxilary elements after the 
      bridge are fused to the vocal part of the bridge normally ('to shitekuru').
    accepted_bridges: ['de aru', 'to suru', 'ni suru', 'ni tsuite', 'ni taishite', 'ni yotte', 'no aru', 'no nai']
    inflection_note: "Aru and Suru inflections ('de atta', 'de arou', 'to shite', 'ni shita') follow the same lowercase split rule."

  particles_and_conjunctions:
    particles: "Always separate and lowercase (no, ga, o, de, mo, ni, wa)"
    conjunctions: "Are kept fused as one single lexical unit and lowercase (でも, だけど, ですが, だったら)"
    always_separate: "Elements that gramatically are split and don't have a specific new meaning are always separate (だと, だに, なの, のか)"
    conditional: >
      Conjuctions such as ので, のに, なので, なのに, だって, では that have meaning both when fused as a lexical unit or as 
      individual elements need to be romanized accordingly depending on the sentence context, meaning 'node' vs 'no de', 
      'noni' vs 'no ni', 'datte' vs 'da tte', and 'dewa' vs 'de wa'. なので and なのに specifically, they are romanized as 
      a single conjuntion 'nanode' or 'nanoni' when they follow a 'concrete noun' or a 'na-adjective'. If they are part of 
      a formal noun bridge sequence they are romanized as 'na node' or 'na noni', see formal_noun_bridges patterns.

  ja_sequence: >
    'ja' (from 'de wa') is a conjunction so always lowercase and separate. Any element following 'ja' or 'de wa' is always capitalized,
    especially with negation, for emphasis. This overrules the lowercase bridge rule ('ja Nai', 'ja Nakatta', 'ja Dame', 'de wa Jimi').
  jan_contraction: "じゃん is always romanized as 'jan', lowercase and separate"

  explanatory_sequences: >
    The explanatory 'no' or 'n' (derived from 'no') are always fused to the following copula or conjunction 
    to avoid loose characters and semantic ambiguity with 何 when preceeded by 'na' which in turn is also
    directly fused to the already fused to the sequence to create one identifiable unit.
    (1) n + copula fuses ('nda', 'ndesu', 'ndatte', 'ndesukara').
    (2) na + n + copula fuses into a single unit ('nandesu', 'nanoda', 'nandakara').
    (3) na + no + copula also fuses into a single unit ('nanodesu', 'nanoda', 'nanodakara').