Japanese Punctuation and Typographic Conventions

Japanese punctuation uses a small, distinct set of marks. They close sentences, separate items, quote speech, indicate range, and shape the visual rhythm of mixed-script text.¹ A few marks share a shape with English (？ ！), but most do not (。、「」・ー〜…). The typography around them, including full-width cells, no inter-word spaces, and the full-width vs. half-width split in mixed-language text, often trips up learners longer than the marks themselves.¹²

Overview

What Japanese punctuation covers in one paragraph

The core inventory is short: 句点「。」 (full stop), 読点「、」 (comma), 鍵括弧「」 (quotation), 二重鍵括弧『』 (nested quotation and titles), 中黒「・」 (interpunct), 長音符「ー」 (prolonged sound mark), 波ダッシュ「〜」 (wave dash), 三点リーダー「…」 (three-dot leader), 感嘆符「！」 (exclamation), and 疑問符「？」 (question mark).¹

Every one of these marks is "full width" by default. Each occupies a square frame the size of a kanji or kana character, which is why the typography rules that surround them matter as much as the marks themselves.¹

A short history: Classical Japanese had almost none, Meiji-era prose adopted the modern set

Pre-modern Japanese ran without sentence-ending marks. Punctuation was not widely used in Japanese writing until translations from European languages became common in the 19th century.¹

The modern Japanese punctuation system emerged primarily during the Meiji period (1868–1912), when European printing presses and translation work created pressure for structured punctuation in vernacular prose.¹³ The first national digital codification followed much later: JIS C 6226 in 1978, renamed JIS X 0208 in 1987, encoded the punctuation inventory alongside kana and kanji for digital text.³

Official central-government style rules are a 20th-century development. In 1951 (昭和26年), Japan issued its first official style guide, 『公用文作成の要領』. It was superseded by the 2022 (令和4年1月7日) 文化審議会建議 『公用文作成の考え方』, which is the current authority for central-government documents and is cited several times below.⁴⁵

The dates are load-bearing, not decorative

The 1868–1912 Meiji window, the 1978 JIS C 6226 codification, and the 令和4年1月7日 (January 7, 2022) 公用文作成の考え方 reform are anchors for everything else in the article. When a future reform changes the official rules, those three dates are what shifts.⁴³

The marks at a glance

Mark	Japanese name	Romanized	English	Primary function
。	句点	kuten	full stop	ends a sentence1
、	読点	tōten	comma	breaks clauses, separates list items1
「 」	鍵括弧	kagikakko	corner brackets	quotation, dialogue1
『 』	二重鍵括弧	nijū-kagikakko	white corner brackets	nested quote, titles1
・	中黒	nakaguro	interpunct	name divider, list separator1
ー	長音符	chōonpu	prolonged sound mark	lengthens preceding vowel in katakana1
〜	波ダッシュ	nami-dasshu	wave dash	range, drawn-out vowel16
……	三点リーダー	santen rīdā	three-dot leader (used in pairs)	pause, omission, speechlessness17
！	感嘆符	kantanfu	exclamation mark	emphasis, exclamation1
？	疑問符	gimonfu	question mark	question (in casual or permitted formal contexts)1

Each mark below has its own section.

Sentence-boundary marks

句点 。: the period (kuten, colloquially maru)

The kuten is Unicode U+3002 IDEOGRAPHIC FULL STOP. Its shape is a small open circle, not a filled dot, and it marks the end of a sentence.¹

Placement follows the character cell, not the line. In horizontal text, the mark sits at the bottom-right of the preceding character cell. In vertical text, it sits at the top-right of the cell.¹ No inter-word space follows it, because the mark already occupies its own square.

The W3C JLReq specification classes the kuten in cl-06 ("Full stops 句点類") and gives the line-end rules in §3.1.9.⁸⁹

明日あした学校がっこうに行いきます。¹
"I'll go to school tomorrow."

In casual writing the kuten is often dropped, especially when a short sentence stands on its own or sits inside quotes; a line break carries the sentence boundary instead.¹⁰ The colloquial spoken name for the mark is マル (maru, "circle"); learners hear that name long before they hear 句点.¹¹⁰

読点 、: the comma (tōten)

The tōten is Unicode U+3001 IDEOGRAPHIC COMMA.¹ It breaks clauses, separates list items, and paces prose. Its placement follows reading rhythm rather than syntax: the writer chooses where the reader should pause.¹

Like the kuten, the tōten sits at the bottom-right of the preceding cell in horizontal text and the top-right of the cell in vertical text, with no following space. It belongs to W3C character class cl-07 ("Commas 読点類") and shares the line-end positioning rules in JLReq §3.1.9.⁸⁹

私わたしはパンを食たべて、コーヒーを飲のみました。¹⁰
"I ate bread and drank coffee."

りんご、みかん、ぶどうを買かいました。¹
"I bought apples, mandarins, and grapes."

Choose 「、」 not 「，」 in horizontal-text official Japanese

The 1951 公用文作成の要領 specified the full-width comma 「，」 for horizontal-text official documents. The 2022 公用文作成の考え方 returned the standard to 「、」 in both writing directions, aligning central-government style with general practice. Authors trained on the older guideline who still use 「，」 are now out of step.⁴¹¹

Why these two together are called 句読点 (kutōten)

句読点 (kutōten) is the compound term that school textbooks and style guides use for the kuten and tōten as a pair.¹ The same compound is the section heading under which the 2022 文化審議会建議 公用文作成の考え方 collects its modern punctuation rules.⁴ When a Japanese teacher says "watch your 句読点," they mean both marks at once.

Quotation and bracketing marks

「 」: kagikakko, the default quotation marks

The kagikakko are Unicode U+300C LEFT CORNER BRACKET and U+300D RIGHT CORNER BRACKET. The W3C JLReq classes them in cl-01 (始め括弧類, "opening brackets") and cl-02 (終わり括弧類, "closing brackets").⁸

They are the everyday quotation brackets for direct speech, quoted phrases, and emphasis on a single word or short phrase.¹

先生せんせいは「おはようございます」と言いいました。¹
"The teacher said, 'Good morning.'"

The same Unicode characters serve both writing directions. The rendering engine rotates the visual placement, so the opening 「 sits at the top-left in horizontal text and at the top-right in vertical text.¹

Period-inside-quote: both forms appear

Conservative literary publishing places sentence-ending 。 inside the closing 」 ("こんにちは。」). Modern web and casual text often drop the period entirely inside quotes ("こんにちは」"). W3C JLReq covers the line-end positioning in §3.1.9 but defers the period-in-quote choice to publisher house style, so the practical advice is to recognize both forms rather than picking a side.⁹¹

『 』: nijū-kagikakko, for inner quotes and titles

The nijū-kagikakko are Unicode U+300E LEFT WHITE CORNER BRACKET and U+300F RIGHT WHITE CORNER BRACKET.¹ They serve two roles. The first is the inner quote nested inside an outer 「 」 quote.

The second role is marking titles. Books, films, manga, newspapers, and songs are wrapped in 『 』 because Japanese lacks italics, so the brackets do that work.¹

彼かれは「先生せんせいが『静しずかに』と言いった」と話はなした。¹
"He said, 'The teacher said, "Quiet."'"

夏目なつめ漱石そうせきの『こころ』を読よみました。¹
"I read Natsume Sōseki's Kokoro."

Learners often miss the title role at first. Once you recognize it, Japanese book covers, film posters, and bibliographies become much easier to parse.

〈 〉 and 《 》: rarer angle and double-angle brackets

These less common brackets show up in academic prose, classical-text editions, and some literary fiction. Learners mainly need to recognize them, not use them: when 〈 〉 or 《 》 wrap a term, treat them as marked emphasis or technical quotation and move on.¹

Bracket placement flips with writing direction

Bracket characters are direction-neutral at the Unicode layer. Their visual orientation flips depending on whether text runs left-to-right horizontal (yokogaki) or top-to-bottom vertical (tategaki).¹ W3C JLReq §3.1.5 ("Positioning of Opening Brackets at Line Head") and §3.1.9 ("Positioning of Closing Brackets, Full Stops, Commas and Middle Dots at Line End") specify the same line-edge handling for both directions. The rendering engine rotates the glyphs.⁹

The character on disk does not change; the layout engine handles the rotation, which is why copy-pasting a horizontal-text quote into a vertical layout does not require swapping the brackets.

Separator and list marks

中黒 ・: the interpunct (nakaguro, also nakaten, nakapotsu)

The nakaguro is Unicode U+30FB KATAKANA MIDDLE DOT and belongs to W3C character class cl-05 ("Middle dots 中点類"). It is full-width by default and occupies its own square cell, matching the surrounding kana.⁸¹

Two functions dominate everyday use. The first is separating foreign first and last names written in katakana (バラク・オバマ). The second is serving as a list separator where 、 would read awkwardly (小・中学校).¹¹⁰

バラク・オバマは元もとアメリカ大統領だいとうりょうです。¹¹⁰
"Barack Obama is a former U.S. president."

小しょう・中学校ちゅうがっこうで英語えいごを教おしえています。¹¹⁰
"I teach English at elementary and middle schools."

The same mark also writes the decimal point in kanji numerals (三・一四 = 3.14), separates titles and positions in compound labels, and clarifies kana runs that would otherwise blur together.¹ Variant spoken names include nakaten (中点), nakapotsu (中ぽつ), and nakaguro (中黒) itself.¹

円周率えんしゅうりつは三さん・一いち四よんです。¹
"Pi is 3.14."

・ as a list separator vs. ・ as a name divider

The two roles look identical on the page but obey different rules.

For foreign personal names written in katakana, the name-divider role is required in standard typography. Using 、 between first and last name is incorrect, not a style choice.¹

The list-separator role does compete with 、 and depends on register and style. ・ is preferred when items are tightly parallel (小・中・高 = elementary, middle, high school) or single-character abbreviations. 、 is preferred for longer items in running prose.¹

Range, repetition, and length marks

長音符 ー: the chōonpu, the katakana long-vowel bar

The chōonpu is Unicode U+30FC KATAKANA-HIRAGANA PROLONGED SOUND MARK, in W3C character class cl-10 ("長音記号").⁸ It lengthens any preceding vowel by one mora, meaning one beat of sound, and is standard in katakana loanwords and onomatopoeia. It is rare in hiragana outside playful or stylized writing.¹¹⁰

コーヒーを一杯いっぱいください。¹⁰
"Please give me a cup of coffee."

スーパーで買かい物ものをします。¹⁰
"I shop at the supermarket."

In vertical writing, the chōonpu rotates to a vertical bar so it still lengthens the vowel of the preceding character in the column.¹

波ダッシュ 〜: the wave dash

The wave dash is Unicode U+301C WAVE DASH. It was added to Unicode 1.0 in 1991 as a compatibility character for the JIS C 6226-1978 / JIS X 0208 inventory.⁶

Three everyday functions cover most uses. The mark indicates a range (5時〜6時, 東京〜大阪), separates a title from its subtitle, and stretches a vowel for casual or cute effect (おいしい〜).¹¹⁰

営業えいぎょう時間じかんは午前ごぜん9時くじ〜午後ごご6時ろくじです。¹⁰
"Business hours are 9 a.m. to 6 p.m."

東京とうきょう〜大阪おおさかの新幹線しんかんせんに乗のりました。¹
"I took the Shinkansen from Tokyo to Osaka."

このケーキ、おいしい〜！¹⁰
"This cake is sooo good!"

Two codepoints that look almost identical

U+301C WAVE DASH and U+FF5E FULLWIDTH TILDE (～) render almost identically. Microsoft's CP932 variant of Shift_JIS maps the byte 0x8160 to U+FF5E rather than U+301C. As a result, legacy Windows-Japanese text and modern Unicode text disagree on which codepoint is "right," and Windows users in Japan typically type U+FF5E even when intending the wave dash.²⁶¹² Unicode's original specimen glyph was even printed inverted (falling-then-rising rather than rising-then-falling). It was corrected through errata fixed in Unicode 8.0.0 (2015).⁶

々: the kanji repetition mark (noma, dō-no-jiten)

々 is Unicode U+3005 IDEOGRAPHIC ITERATION MARK. It repeats the previous kanji: 人人 is written 人々 (hitobito, "people"), 時時 is written 時々 (tokidoki, "sometimes").¹ It is technically an iteration mark rather than a punctuation mark, but learners meet it alongside 。 and 「」 in basic script-literacy material.¹

人々ひとびとが公園こうえんに集あつまりました。¹
"People gathered at the park."

時々ときどき日本語にほんごの本ほんを読よみます。¹
"I sometimes read Japanese-language books."

Pause, omission, and emphasis marks

三点リーダー …: the three-dot leader, used in pairs as ……

The three-dot leader is Unicode U+2026 HORIZONTAL ELLIPSIS, with a close relative at U+2025 TWO DOT LEADER.¹ The character is its own glyph, not three period characters typed in a row.⁷

Standard formal-print convention uses six dots, written as two characters: ……. A single three-dot character is acceptable in tight space.¹ In horizontal text, placement is along the baseline or centered between the baseline and ascender. In vertical text, it is centered horizontally.¹

彼かれは何なにも言いわなかった……。⁷
"He didn't say a word..."

あの……ちょっといいですか。¹
"Um... do you have a moment?"

Manga ellipses are louder than they look

A manga or visual-novel panel containing only "……" (or a longer run) is a full beat of silence, embarrassment, or speechlessness. That visual weight does not survive translation, so English editions routinely render the beat as a full sentence rather than three dots.¹⁷ When a learner reads native text and sees a paragraph that is just dots, the dots are the dialogue.

感嘆符 ! and 疑問符 ？: exclamation and question marks

！ is Unicode U+FF01 FULLWIDTH EXCLAMATION MARK and ？ is U+FF1F FULLWIDTH QUESTION MARK; the half-width forms U+0021 and U+003F also appear in mixed-language contexts. Both belong to W3C character class cl-04 ("区切り約物 / dividing punctuation marks," punctuation that separates units of text).⁸²

Neither mark was part of the classical-Japanese inventory. Both were borrowed from Western punctuation during the Meiji-era import of European print conventions.¹

They are common in casual writing, fiction, manga, and advertising. In formal prose they are traditionally absent, because sentence-final か already marks a question and sentence-ending tone carries the exclamation weight.¹

危あぶない！　車くるまが来くるよ。¹
"Look out! A car is coming."

本当ほんとうですか？　信しんじられない。¹
"Really? I can't believe it."

The official-document rule is recent and has a fixed date. The 文化審議会建議 『公用文作成の考え方』 (令和4年1月7日, January 7, 2022) formally permits ？ and ！ in central-government documents for promotional and explanatory materials, and in transcripts that preserve direct statements verbatim.⁴⁵ The same document adds a spacing rule: when ？ or ！ is followed by more text on the same line, leave one space after the mark. That space may be full-width or half-width, as long as the document is consistent.⁴¹¹ W3C JLReq formalizes the same line-internal spacing convention in §3.1.6 ("Positioning of Dividing Punctuation Marks and Hyphens"), specifying a full-width space after cl-04 characters when they are not at line end.⁹

Casual nicknames: マル, びっくりマーク, はてなマーク

Learners hear these colloquial labels in everyday conversations about texting and writing, alongside the formal kanji names. 句点 is マル (maru, "circle"); 感嘆符 is びっくりマーク (bikkuri māku, "surprise mark"); 疑問符 is はてなマーク (hatena māku, "question/what mark") or クエスチョンマーク (kuesuchon māku).¹¹⁰ A Japanese speaker who says "put a びっくり here" is asking for an exclamation mark, not a noise.

Spacing and width conventions

No inter-word spaces in standard Japanese prose

Japanese prose does not separate words with spaces. The script handles word boundaries by alternating kanji, hiragana, and katakana. That mixed-script rhythm does the work that whitespace does in English.¹¹⁰

Spaces do appear in a narrow set of contexts: children's books, beginner-learner materials, some early-grade textbooks, and the genkō yōshi convention, where each punctuation mark fills its own grid cell.¹ A learner moving from textbook spacing to native prose loses the visual word-boundary crutch and has to read by script alternation instead.

Full-width spaces around proper nouns in some genres

The 全角スペース (full-width space, Unicode U+3000 IDEOGRAPHIC SPACE) inserts a square-cell blank to frame personal names in formal letters, official documents, and some newspaper styles.¹ The same character commonly follows 。 in genkō yōshi paragraph indentation. It also sits between family and given names in form-input layouts that ask for both fields concatenated.¹

全角 (zenkaku) vs. 半角 (hankaku) punctuation in mixed-language text

Unicode block U+FF00–U+FFEF ("Halfwidth and Fullwidth Forms") was added so that legacy East Asian encodings (JIS X 0201 from 1969, JIS X 0208 from 1978, and Shift_JIS) could round-trip into Unicode without loss.²¹³³¹⁴ U+FF01–U+FF5E reproduces ASCII 0x21–0x7E as full-width forms. U+FF61–U+FF9F encodes half-width katakana and related punctuation (the JIS X 0201 transposition of bytes 0xA1–0xDF).²¹⁴

In Japanese terminology, 全角 (zenkaku, "full width") characters occupy a square-cell width matching kanji, while 半角 (hankaku, "half width") characters occupy roughly half that width.²

The practical rule follows the same pattern as the diagram. Japanese prose uses full-width 。、「」 alongside Japanese characters. Latin-script blocks (URLs, code, scientific names, English quotations) use half-width . , " "; and pure-English sentences embedded in Japanese take half-width Western punctuation. The Unicode block makes every Western punctuation mark available in both widths so authors can choose by context.²

The form-field corner: why your address form rejects your input

Japanese web forms often require 半角 numerals and 全角 katakana, or sometimes the reverse. The rule varies by site and is rarely explained on the form itself.² The 2022 公用文作成の考え方 directly addresses input width for official documents, defaulting numerals to half-width in horizontal text alongside the 「、」 comma rule.⁴¹¹

Half-width katakana is legacy, not style

Half-width katakana (ｶﾀｶﾅ) survives in receipts, ATM displays, station LED signage, and some legacy retail and banking systems. The major Japanese funds-transfer network used half-width kana until 2018. Modern web text avoids it.¹³ When a form demands 全角 katakana for a name field, the half-width set is the wrong character set, not a smaller font.

Good to know

Using 、 between a foreign first and last name written in katakana

The interpunct ・ is the standard name divider for foreign personal names in katakana. Using 、 reads as a list of two separate people rather than one person's full name.¹ The correct form is:

バラク・オバマは元もとアメリカ大統領だいとうりょうです。¹¹⁰
"Barack Obama is a former U.S. president."

Treating a single … as the formal-print ellipsis

Formal-print convention uses six dots, written as two U+2026 characters. A single three-dot character is acceptable only in tight space. A single … at the end of a literary sentence reads as informal.¹ The expected form is:

彼かれは何なにも言いわなかった……。¹⁷
"He didn't say a word..."

Concatenating three period characters to form an ellipsis

A row of three U+002E full-stops (...) is not the Japanese ellipsis. The Japanese ellipsis is its own font character with its own spacing metrics. Three periods produce visibly wrong spacing and break copyfitting, the process of fitting text into a set layout, in typeset text.⁷ The correct form uses U+2026, doubled:

彼かれは何なにも言いわなかった……。¹⁷
"He didn't say a word..."

Using ！ or ？ in a legislative or academic document

The 2022 公用文作成の考え方 permits ！ and ？ only in promotional or explanatory central-government materials and in transcripts of direct speech. Legislation, regulations, and formal academic reports still rely on か and on declarative sentence-ending tone to do the work these marks would do in English.⁴

Comma placement is rhythm, not syntax

読点 placement is a writer's choice, not a syntactic rule. Too many commas read as belabored; too few read as breathless.

The standard editorial advice is to read the sentence aloud and place a 、 where a natural breath falls. There is no English-style "comma before the conjunction" rule to apply.¹

Period inside or outside the closing 」

Conservative literary publishing places sentence-ending 。 inside the closing 」 ("こんにちは。」). Modern web and casual text often drop the period entirely inside quotes ("こんにちは」"). Both forms appear in published Japanese, and publisher-by-publisher variation means recognizing both is more useful than memorizing one.⁹¹

The wave dash 〜 has two Unicode codepoints

U+301C WAVE DASH and U+FF5E FULLWIDTH TILDE render almost identically. Microsoft's CP932 Shift_JIS variant maps the legacy byte to U+FF5E, so old text and modern text disagree on which codepoint is "right." The keyboard usually picks one for the user, so this is a recognition issue rather than a production rule for learners.²⁶¹²

Why classical Japanese had no punctuation

Pre-modern Japanese ran without sentence-ending marks. Readers relied on the kakari-musubi system, sentence-final particles, and verb endings to mark grammatical boundaries instead. The modern punctuation set was a Meiji-era (1868–1912) import absorbed alongside Western printing, solving a problem the language had not yet felt.¹

Half-width katakana is a legacy form, not a current style

The half-width katakana set (ｶﾀｶﾅ) exists because the 1969 JIS X 0201 single-byte encoding could not fit full-width characters in 8-bit columns. It survives in receipts, ATM displays, and station signage but is avoided in modern web text. A learner who sees it should treat it as a historical artifact rather than a style option.¹³

See also

• Mixed Script: How Japanese Combines Kanji, Hiragana, and Katakana
• Vertical Writing (Tategaki) vs. Horizontal (Yokogaki)
• Furigana: Reading Aids Above Kanji
• Okurigana: When Kanji Bleeds Into Hiragana
• Long Vowels in Katakana: How the Chōonpu ー Works and Why Hiragana Doesn't Use It
• How to Write Your Name in Katakana: Foreign-Name Transcription Rules with Examples

References

Footnotes

1. Wikipedia contributors. "Japanese punctuation." Wikipedia. https://en.wikipedia.org/wiki/Japanese_punctuation ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸ ↩⁹ ↩¹⁰ ↩¹¹ ↩¹² ↩¹³ ↩¹⁴ ↩¹⁵ ↩¹⁶ ↩¹⁷ ↩¹⁸ ↩¹⁹ ↩²⁰ ↩²¹ ↩²² ↩²³ ↩²⁴ ↩²⁵ ↩²⁶ ↩²⁷ ↩²⁸ ↩²⁹ ↩³⁰ ↩³¹ ↩³² ↩³³ ↩³⁴ ↩³⁵ ↩³⁶ ↩³⁷ ↩³⁸ ↩³⁹ ↩⁴⁰ ↩⁴¹ ↩⁴² ↩⁴³ ↩⁴⁴ ↩⁴⁵ ↩⁴⁶ ↩⁴⁷ ↩⁴⁸ ↩⁴⁹ ↩⁵⁰ ↩⁵¹ ↩⁵² ↩⁵³ ↩⁵⁴ ↩⁵⁵ ↩⁵⁶ ↩⁵⁷ ↩⁵⁸ ↩⁵⁹ ↩⁶⁰ ↩⁶¹ ↩⁶² ↩⁶³ ↩⁶⁴ ↩⁶⁵ ↩⁶⁶ ↩⁶⁷ ↩⁶⁸ ↩⁶⁹ ↩⁷⁰ ↩⁷¹ ↩⁷² ↩⁷³

2. Wikipedia contributors. "Halfwidth and fullwidth forms." Wikipedia. https://en.wikipedia.org/wiki/Halfwidth_and_fullwidth_forms ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸ ↩⁹

3. Wikipedia contributors. "JIS X 0208." Wikipedia. https://en.wikipedia.org/wiki/JIS_X_0208 ↩ ↩² ↩³ ↩⁴

4. 文化審議会. 『公用文作成の考え方（建議）』（付）「公用文作成の考え方（文化審議会建議）」解説. 令和4年1月7日. 文化庁. https://www.bunka.go.jp/seisaku/bunkashingikai/kokugo/hokoku/93657201.html ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸

5. 文化庁 報道発表. 「文化審議会建議『公用文作成の考え方』について」. https://www.bunka.go.jp/koho_hodo_oshirase/hodohappyo/93650001.html ↩ ↩²

6. Wikipedia contributors. "Wave dash." Wikipedia. https://en.wikipedia.org/wiki/Wave_dash ↩ ↩² ↩³ ↩⁴ ↩⁵

7. Mandelin, Clyde. "Encountering Japanese Ellipses in English Translations." Legends of Localization, 8 September 2013. https://legendsoflocalization.com/articles/japanese-ellipsis-usage/ (limitation: localization industry source; used here for the manga-ellipsis weight convention) ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷

8. W3C. Requirements for Japanese Text Layout, Appendix A, "Character Classes" (cl-01 through cl-10). https://www.w3.org/TR/jlreq/#character_classes ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶

9. W3C. Requirements for Japanese Text Layout, §3.1, "Line Composition Rules for Punctuation Marks." https://www.w3.org/TR/jlreq/#line_composition_rules_for_punctuation_marks ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶

10. Croes, Jaered Koichi, and Kristen Dexter. "Japanese Punctuation: The Definitive Guide." Tofugu, 21 March 2016. https://www.tofugu.com/japanese/japanese-punctuation/ (limitation: language-learning publisher, used here only for natural example sentences also attested elsewhere) ↩ ↩² ↩³ ↩⁴ ↩⁵ ↩⁶ ↩⁷ ↩⁸ ↩⁹ ↩¹⁰ ↩¹¹ ↩¹² ↩¹³ ↩¹⁴ ↩¹⁵

11. 神楽坂編集室. 「公用文における数字・記号の書き方」 (summary of the 2022 文化審議会建議 公用文作成の考え方 rules on 句読点 and 符号). https://kagurazaka-editors.jp/official-document-number/ (limitation: secondary summary; used only where it corroborates the primary 文化庁 PDF on the ?/! spacing rule) ↩ ↩² ↩³

12. Microsoft Corporation (Shawn Steele). cp932 to Unicode table (Table format A, Table version 2.01). Published by The Unicode Consortium under WindowsBestFit vendor mappings. https://www.unicode.org/Public/MAPPINGS/VENDORS/MICSFT/WINDOWS/CP932.TXT (entry 0x8160 0xFF5E #FULLWIDTH TILDE). ↩ ↩²

13. Wikipedia contributors. "Half-width kana." Wikipedia. https://en.wikipedia.org/wiki/Half-width_kana ↩ ↩² ↩³

14. The Unicode Consortium. The Unicode Standard, "Halfwidth and Fullwidth Forms" (block U+FF00–U+FFEF). Code chart and block introduction. https://www.unicode.org/charts/PDF/UFF00.pdf ↩ ↩²