Sentence stops delimit a full grammatical sentence.

Writing systems vary as to whether they also terminate a standalone sentence, and therefore act as a sentence terminator. That is the case in formal English (though often not in texting), but it is not the case in Japanese.

• Latin and Cyrillic require space after sentence stops as delimiters between sentences, as does Korean.

• Writing systems do not require space after the final sentence stop in a paragraph.

• Chinese and Japanese by default do not use space after sentence stops.

• None of the writing systems in scope of this framework has space before sentence stops.

• In French in France, Switzerland and Belgium, some punctuation marks, including some sentence stops, are preceded by a non-breaking thin space (U+202F) (“French spacing”). In common practice, a full non-breaking space (U+00A0) is used instead. French spacing is not used for most punctuation marks in Canada for French.

• Korean sentence stops are half-width.

• Chinese and Japanese sentence stops are full-width.

• In text mixed between Latin and Chinese or Japanese, usual practice is to follow the document main language. Fine typography will apply kerning in such contexts so that the switch in width does not look obtrusive.

  Script switch with Chinese as main language

  现在很多人都在用 iPhone。(with full-width punctuation in a Chinese text after English words)

  Script switch with English as main language

  This app 很好用.

• Some desktop publishing applications in Japanese and CJK OpenType fonts allow switching contextually between full-width and half-width punctuation (欧文混在時の約物処理 “treat punctuation when mixed with Western text”).

Mark that indicates the end of a complete declarative grammatical statement. This function is represented in Metanorma i18n files as punct.period.

• Latin, Cyrillic: FULL STOP <.> (U+002E)

• Traditional Chinese: IDEOGRAPHIC FULL STOP <。> (U+3002)

• Simplified Chinese: IDEOGRAPHIC FULL STOP <。> (U+3002)

• Japanese: IDEOGRAPHIC FULL STOP <。> (U+3002)

• Korean: FULL STOP <.> (U+002E)

• French spacing does NOT apply ().

• The placement of period in CJK is different by directionality, but this does not involve different Unicode codepoints, so the browser/operating system is assumed to handle positioning appropriately. (While there is a PRESENTATION FORM FOR VERTICAL IDEOGRAPHIC FULL STOP <︒> (U+FE12), this is only included in Unicode as a compatibility character.)

  • In Japanese and Simplified Chinese horizontal text, the period appears at the bottom right. In Japanese and Simplified Chinese vertical text, the period appears below and to the right of the character.

  • In Traditional Chinese, the period appears at mid-height in both horizontal and vertical text orientations.

Mark that indicates this grammatical statement is a question. This function is represented in Metanorma i18n files as punct.question-mark.

• Latin, Cyrillic: QUESTION MARK <?> (U+003F)

• Traditional Chinese: FULLWIDTH QUESTION MARK <？> (U+FF1F)

• Simplified Chinese: FULLWIDTH QUESTION MARK <？> (U+FF1F)

• Japanese: FULLWIDTH QUESTION MARK <？> (U+FF1F)

• Korean: QUESTION MARK <?> (U+003F)

• French spacing applies ().

• In Spanish and languages under Spanish cultural influence (e.g. Catalan), INVERTED QUESTION MARK <¿> (U+00BF) is used at the start of an interrogative sentence or phrase. If a declarative sentence contains an interrogative phrase, only the interrogative phrase is so delimited:

  Si no puedes ir con ellos, ¿quieres ir con nosotros? “If you cannot go with them, ¿would you like to go with us?”

  • There is some variation by locale in usage: short questions can omit the inverted question mark in Galician; Catalan in Catalonia does not use inverted question marks; Catalan in Valencia uses them optionally.

Mark that indicates this grammatical statement is an exclamation. This function is represented in Metanorma i18n files as punct.exclamation-mark.

• Latin, Cyrillic: EXCLAMATION MARK <!> (U+0021)

• Traditional Chinese: FULLWIDTH EXCLAMATION MARK <！> (U+FF01)

• Simplified Chinese: FULLWIDTH EXCLAMATION MARK <！> (U+FF01)

• Japanese: FULLWIDTH EXCLAMATION MARK <！> (U+FF01)

• Korean: EXCLAMATION MARK <!> (U+0021)

• French spacing applies ().

• In Spanish and languages under Spanish cultural influence (e.g. Catalan), INVERTED EXCLAMATION MARK <¡> (U+00A1) is used at the start of an exclamatory sentence or phrase.

• The same locale considerations apply for Catalan as for inverted question mark.

Phrase stops delimit phrases within a sentence.

CJK full-width punctuation () applies.

Mark that delimits a minor break between phrases within a sentence. This function is represented in Metanorma i18n files as punct.comma.

• The contexts in which the minor break is marked between phrases vary significantly by language and by style.

• The types of phrase eligible to be so marked also vary significantly by language and by style; it usually includes both clauses (which express a complete predicate, including both a subject and a verb) and smaller units such as noun phrases, or successive adjectives.

  Ich weiß, dass du lügst “I know, that you’re lying”, with comma separating a complement clause from the main clause, is correct punctuation in German, but not English

• In many writing systems, the same mark (comma) is used for minor phrase separators and other separating functions, such as enumerator delimiters (), and decimal point ().

• Latin, Cyrillic: COMMA <,> (U+002C)

• Traditional Chinese: FULLWIDTH COMMA <，> (U+FF0C)

• Simplified Chinese: FULLWIDTH COMMA <，> (U+FF0C)

• Japanese: IDEOGRAPHIC COMMA <、> (U+3001)

  • In mixed Japanese-Western text, FULLWIDTH COMMA <，> (U+FF0C) can be used instead to maintain visual consistency

  • IDEOGRAPHIC SPACE <　> (U+3000) can be used in Japanese corresponding to English comma or colon usage in certain contexts.

• Korean: COMMA <,> (U+002C)

The placement of comma in CJK is different by directionality, but this does not involve different Unicode codepoints, so the browser/operating system is assumed to handle positioning appropriately. (While there is a PRESENTATION FORM FOR VERTICAL COMMA <︐> (U+FE10) and a PRESENTATION FORM FOR VERTICAL IDEOGRAPHIC COMMA <︑> (U+FE11), these are only included in Unicode as a compatibility character.)

• In Japanese and Simplified Chinese horizontal text, the comma appears at the bottom right. In Japanese and Simplified Chinese vertical text, the comma appears below and to the right of the character.

• In Traditional Chinese, the comma appears at mid-height in both horizontal and vertical text orientations.

Mark that delimits a major break between phrases within a sentence. This function is represented in Metanorma i18n files as punct.semicolon.

• The contexts in which the major break is marked between phrases varies significantly by language and by style. In many styles, it is avoided as an overly fine distinction from the minor break (comma).

• The phrases separated by a major phrase separator are typically independent clauses grammatically (i.e. complete predicates, with both a subject and a verb).

• Latin, Cyrillic: SEMICOLON <;> (U+002C)

• Traditional Chinese: FULLWIDTH SEMICOLON <；> (U+FF1B)

• Simplified Chinese: FULLWIDTH SEMICOLON <；> (U+FF1B)

• Japanese: FULLWIDTH SEMICOLON <；> (U+FF1B)

• Korean: SEMICOLON <;> (U+002C)

• French spacing applies ().

The placement of semicolon in CJK is different by directionality, but this does not involve different Unicode codepoints, so the browser/operating system is assumed to handle positioning appropriately. (While there is a PRESENTATION FORM FOR VERTICAL SEMICOLON <︔ > (U+FE14), this is only included in Unicode as a compatibility character.)

Mark that delimits phrases within a sentence, where the first phrase introduces the second. This function is represented in Metanorma i18n files as punct.colon.

• Sometimes punctuation is used to separate a term from a description in a definition list, although the default is to use only indented space. This function can be regarded as an extension of the introductory phrase separator. In Metanorma Presentation XML, this is notated as <span class="fmt-dt-delim">.

• Latin: COLON <:> (U+003A)

• Cyrillic: COLON <:> (U+003A)

• Traditional Chinese: FULLWIDTH COLON <：> (U+FF1A)

• Simplified Chinese: FULLWIDTH COLON <：> (U+FF1A)

• Japanese: FULLWIDTH COLON <：> (U+FF1A)

  • IDEOGRAPHIC SPACE <　> (U+3000) can be used in Japanese corresponding to English comma or colon usage in certain contexts.

• Korean: COLON <:> (U+003A)

The placement of colon in CJK is different by directionality, but this does not involve different Unicode codepoints, so the browser/operating system is assumed to handle positioning appropriately. (While there is a PRESENTATION FORM FOR VERTICAL COLON <︓ > (U+FE13), this is only included in Unicode as a compatibility character.)

The interpunct () is used instead of hyphen, dash, or colon in Japanese vertical text.

Mark that delimits phrases within a sentence, where there is a conceptual break of some sort between the first phrase and the second. This function is represented in Metanorma i18n files as punct.em-dash.

• An interrupted ending to a sentence can be indicated by a breaking phrase separator. Similarly, a sentence start which counts as a resumption from a previous interruption can be indicated by a breaking phrase separator.

  He was the miracle ingredient Z-147. He was—
  “Crazy!” Clevinger interrupted, shrieking. “That’s what you are! Crazy!”
  “—immense. I’m a real, slam-bang, honest-to-goodness, three-fisted humdinger. I’m a bona fide supraman.” — Joseph Heller, Catch-22

  • Interrupted and resumed sentences are characteristic of literary prose, but not formal prose, such as is in scope of Metanorma.

• Latin, Cyrillic: EM DASH <—> (U+2014)

  • The EN DASH <–> (U+2013) is also used in this function

  • Double and triple hyphens are typewriter approximations of em-dashes, and are commonly used in word processing as easy ways to data enter em-dashes through auto-text; they are not normally expected to display as such in finished documents, although they remain a convention of comic strips.

  • In some style guides (e.g. the Australian Government Style Manual), interrupted and resumed sentence breaks are notated with two em-dashes, as a distinct function from the breaking phrase separator.

• Traditional Chinese: TWO EM DASH <⸺> (U+2E3A)

• Simplified Chinese: TWO EM DASH <⸺> (U+2E3A)

• Japanese: TWO EM DASH <⸺> (U+2E3A)

  • Informally in CJK, twice em-dash (U+2014 U+2014) is used instead of the two em-dash.

• Korean: EM DASH <—> (U+2014)

• The placement of em-dash in CJK is different by directionality, but this does not involve different Unicode codepoints, so the browser/operating system is assumed to handle positioning appropriately. (While there is a PRESENTATION FORM FOR VERTICAL EM DASH <︱> (U+FE31), this is only included in Unicode as a compatibility character.)

• The interpunct () is used instead of hyphen, dash, or colon in Japanese vertical text.

• An interrupted ending to a sentence can be indicated by a breaking phrase separator. In that case, the breaking phrase separator replaces the sentence stop.

  He was the miracle ingredient Z-147. He was—
  “Crazy!”

Mark that delimits phrases within a sentence, where there is some sort of hesitation between them.

• The hesitancy phrase separator is closely related to the breaking phrase separator, and in past practice was conflated with it.

• The hesitancy phrase separator is routinely conflated with the missing text mark (), but the two interact differently with other punctuation.

• “Hesitancy” is to be broadly understood, and it includes pauses, interruption, speechlessness, deliberate silence, longing, and surprise. Different languages place different expectations on the punctuation mark.

• The hesitancy phrase separator is associated with emotion, and is therefore avoided in formal writing, such as is in scope of Metanorma. The missing text mark, on the other hand, is entirely consistent with formal writing.

• Latin, Cyrillic: in fine typography, a single HORIZONTAL ELLIPSIS <…> (U+2026) is used. In informal practice, three periods in succession are used as well.

  • In some fine typographic practice, the em dash for the breaking phrase separator is used instead.

  • In some fine typographic practice, the three periods are interpolated with non-breaking spaces, or thin non-breaking spaces.

• CJK: the ellipse can be presented as three dots, or six dots.

  • The six dot form is entered as twice the full-width three-dot form.

  • The CJK form is either MIDLINE HORIZONTAL ELLIPSIS <⋯> (U+22EF), as explicit centered markup, or the half-width HORIZONTAL ELLIPSIS (U+2026) where centering is inexplicit, rendered in CJK fonts as full-width.

  • Japanese rarely also uses a two-dot ellipse (rīdā).

• If the hesitation follows the final phrase in a sentence, some practice has the hesitancy mark replace the sentence stop (e.g. common British English practice). Other practice appends the sentence stop to the hesitancy mark, as a fourth dot (e.g. Chicago Manual of Style).

  I like traffic lights…
  I like traffic lights… .

• Russian combines not only declarative but interrogative and exclamatory sentence stops with the hesitancy mark and missing text mark: <?..>, <!..>.

• Unlike other stops, the hesitation phrase separator can appear at the start of a sentence.

• In horizontal directonality in Traditional Chinese and Japanese, the ellipse is vertically centered; in vertical directionality, it is horizontally centered.

• In Simplified Chinese, the ellipse is usually aligned to the baseline in horizontal directionality; in vertical directionality, it is still horizontally centered.

• In vertical directionality in CJK, Unicode has a codepoint for VERTICAL ELLIPSIS <⋮> (U+22EE), and it is not designated as a compatibility character. However as with other punctuation, the browser/operating system is assumed to handle positioning appropriately, so MIDLINE HORIZONTAL ELLIPSIS would still be expected to be entered.

Mark that delimits verses in poetic writing.

• In verses presented as poetry, verses are globally separated by line breaks, which are not regarded as punctuation.

• In prose transcription of poetry, verses are often separated by a slash: SOLIDUS </> (U+002F).

  • There is some usage of VERTICAL LINE <|> (U+007C) instead.

• When slash is used to separate verses, there is a space either side of the slash.

In poetic presentation (one verse per line), separate verses must not be right-justified at the line-break.

The section separator includes a break in a document, at a higher level than a sentence or paragraph. This is distinct from a clause heading, which introduces a new section of text, with a number and/or a title.

• In formal documents of the type considered by Metanorma, the only permitted breaks at a document level higher than a paragraph are clauses, and new clauses are explicitly indicated by clause headings. Section separators are characteristic of extended literary prose, such as novels; even though novels have chapters, they do not have subchapters, and breaks in a chapter are indicated by a section separator instead.

  • In literary use, the section separator is intended to convey a logical or emotional break.

• There is overlap between the section separator and the missing text mark (), when the scope of the missing text mark is at the paragraph level.

• The cover term for the traditional punctuation mark for a section separator, used in literary writing, is the dinkus. The usual contemporary form of the dinkus is three asterisks or three bullets in a row, with space between them.

  • Older practice used other decorative symbols, including ASTERISM <⁂> (U+2042) and various fleurons—such as ROTATED FLORAL HEART BULLET <❧> (U+2767) and NORTH WEST POINTING LEAF <🙐> (U+1F650).

• In word processing and HTML documents, this function is conveyed by the horizontal rule (<hr/>), which is not regarded as punctuation.

• The Japanese PART ALTERNATION MARK <〽> (U+303D) has narrower application, being used in the Renga genre of poetry to indicate the start of a song. The LEFT CORNER BRACKET <「> (U+300C) may also be used.

• The dinkus typically appears centrally aligned, on a line of its own, with vertical spacing before and after it.

• In contemporary practice, the dinkus is mutually exclusive with clause headings; in older usage a dinkus can appear before a new clause heading, especially when it conveys an emotional break.

Listing stops are used to delimit multiple items in a listing, other than phrases.

Because of the conceptual similarity between phrase stops and listing stops, as joining multiple items, punctuation marks for the two are often conflated. In fact, if the definition of “phrase” is liberal enough, listing stops are phrase stops. This framework considers punctuation separating noun or adjective phrases to be listing stops, and not phrase stops.

The enumeration delimiter separates items in a list. This function is represented in Metanorma i18n files as punct.enum-comma, and in Metanorma Presentation XML as <span class="fmt-enum-comma">.

In CJK, both comma-like punctuation marks and interpuncts are used as enumeration delimiters. There is no consistent semantic distinction between the two, so no such distinction is made here.

• There is variation in whether the enumeration delimiter is used in a list of two items.

• There is (linguistic) variation in whether a conjunction is used before the last item in a list, and (punctuation) variation in whether the enumeration delimiter is used before the last item. Thus English allows both A, B and C and A, B, and C as style variants (the latter is known as the “Oxford comma”). Chinese allows both A、B及C and A、B、C.

• The punctuation mark for the enumeration delimiter is often the same as that for the minor phrase separator (comma).

• The punctuation mark for the enumeration delimiter is often the same as that for other separators; for example Japanese uses the interpunct as an enumeration delimiter, as a decimal point (), and to separate professional titles, names, and positions ().

• Latin, Cyrillic: COMMA <,> (U+002C)

• Traditional Chinese: IDEOGRAPHIC COMMA <、> (U+3001)

• Simplified Chinese: IDEOGRAPHIC COMMA <、> (U+3001)

• Japanese: IDEOGRAPHIC COMMA <、> (U+3001)

  • Japanese uses the interpunct for short lists instead of comma in some contexts. This is rendered as KATAKANA MIDDLE DOT <・> (U+30FB)

  • In mixed Japanese-Western text, FULLWIDTH COMMA <，> (U+FF0C) can be used instead to maintain visual consistency

• Korean: COMMA <,> (U+002C)

  • Korean uses the interpunct for short lists instead of comma in some contexts. This is rendered as HANGUL LETTER ARAEA <ㆍ> U+318D.

• When lists contain mixed scripts in CJK, follow the punctuation convention of the list’s primary language while maintaining readability.

• Japanese also supports HALFWIDTH KATAKANA MIDDLE DOT <･> (U+FF65) for the interpunct.

• In Japanese vertical text, the interpunct appears centered rather than at a specific corner position.

List itemisation marks are used at the start of list items, when they are rendered as separate lines in a list paragraph.

List items may be ordered or unordered; if they are ordered, the list itemisation mark is an ordered numeral or letter, optionally followed by a caption number delimiter (). In the case of unordered lists, a single typographic symbol is used.

• The default list itemisation mark for unordered list items is BULLET <•> (U+2022). Alternatives include the em dash, en dash, and WHITE BULLET <◦> (U+25E6).

• List items of different levels in a list are typically indented to differing degrees, so as to indicate that level.

• The list itemisation mark is typically separated from the list item content by a tab (a horizontal space of consistent width between different list items).

• List items are typically rendered with hanging indentation, so that their list itemisation marks and content lines up across multiple-line list items.

• In word processing and HTML, list itemisation marks are not entered directly by the author, but are indicated to be rendered via list markup. The selection of list itemisation mark in unordered lists is almost never specified in markup, but in the document configuration/document stylesheet.

Quotation markers indicate that a span of text constitutes direct speech, or otherwise attribute it to some third party.

• In French in France, Switzerland and Belgium, for paired quotation delimiters, the opening delimiter (left guillemet, left single guillemet) is followed by a non-breaking thin space (U+202F), as an instance of French spacing (); the closing delimiter (right guillemet, right single guillemet) is preceded by a non-breaking thin space, as the mirror counterpart of French spacing. As with other French spacing, in common practice, a full non-breaking space (U+00A0) is used instead.

Paired quotation delimiters indicate the start and end of quoted text. This function is represented in Metanorma i18n files as punct.open-quote and punct.close-quote.

• Paired quotation markers are often conflated with title marks ().

• Latin, Cyrillic: There is significant variation among Latin and Cyrillic script languages on their choice of paired quotation delimiter and their orientation, and different languages and locales draw from 15 Unicode codepoints.

  • In American English, good typographical practice uses LEFT DOUBLE QUOTATION MARK <“> (U+201C) and RIGHT DOUBLE QUOTATION MARK <”> (U+201D) as opening and closing marks. In British English, some publishers use double quotation marks, and others use single quotation marks, LEFT SINGLE QUOTATION MARK <‘> (U+2018) and RIGHT SINGLE QUOTATION MARK <’> (U+2019).

  • Finnish uses the RIGHT DOUBLE QUOTATION MARK <”> (U+201D) as both opening and closing marks.

  • French uses guillemets, LEFT-POINTING DOUBLE ANGLE QUOTATION MARK <«> (U+00AB) and RIGHT-POINTING DOUBLE ANGLE QUOTATION MARK <»> (U+00BB) as opening and closing marks.

• Traditional Chinese: LEFT CORNER BRACKET <「> (U+300C) and RIGHT CORNER BRACKET <」> (U+300D)

• Simplified Chinese: LEFT DOUBLE QUOTATION MARK <“> (U+201C) and RIGHT DOUBLE QUOTATION MARK <”> (U+201D).

  • The corner brackets of Traditional Chinese are also in common use.

• Japanese: LEFT CORNER BRACKET <「> (U+300C) and RIGHT CORNER BRACKET <」> (U+300D).

  • Japanese also uses lenticular brackets, LEFT BLACK LENTICULAR BRACKET <【> (U+3010) and RIGHT BLACK LENTICULAR BRACKET < 】> (U+3011).

• Korean: South Korea uses the same punctuation marks as English. North Korea uses full-width guillemets, LEFT DOUBLE ANGLE BRACKET <《> (U+300A) and RIGHT DOUBLE ANGLE BRACKET <》> (U+300B).

  • Corner brackets and white corner brackets are often used in practice.

• The placement of quotation marks in CJK is different by directionality, but this does not involve different Unicode codepoints, so the browser/operating system is assumed to handle positioning appropriately. (While there is e.g. a PRESENTATION FORM FOR VERTICAL LEFT CORNER BRACKET <﹁> (U+FE41) and PRESENTATION FORM FOR VERTICAL RIGHT CORNER BRACKET <﹂> (U+FE42), these are only included in Unicode as a compatibility character.)

• Quotation marks in English are routinely typed as straight quotes, QUOTATION MARK <"> (U+0022), with the expectation that software will transform them contextually into paired quotation marks (“smart quotes”).

  Metanorma transforms straight quotes in Asciidoc source into smart quotes following English conventions.

• Simplified Chinese uses the font to make Western quotation marks (U+201C, U+201D) full-width, instead of deploying distinct codepoints: REVERSED DOUBLE PRIME QUOTATION MARK `<〝>, (U+301D) DOUBLE PRIME QUOTATION MARK <〞> (U+301E).

• In vertical directionality, Traditional and Simplified Chinese use presentation variants of the codepoints LEFT WHITE CORNER BRACKET <『> (U+300E) and RIGHT WHITE CORNER BRACKET < 』> (U+300F).

Nested paired quotation delimiters indicate the start and end of quoted text within another delimited quoted text, and attributed to a different speaker from that of the nesting quoted text. This function is represented in Metanorma i18n files as punct.open-nested-quote and punct.close-nested-quote.

• The English closing nested paired quotation delimiter is often conflated with the elision mark ().

• Latin, Cyrillic: There is significant variation among Latin and Cyrillic script languages on their choice of paired quotation delimiter, and different languages and locales draw from 15 Unicode codepoints.

  • In American English, good typographical practice uses LEFT SINGLE QUOTATION MARK <‘> (U+2018) and RIGHT SINGLE QUOTATION MARK <’> (U+2019) as opening and closing marks. In British English, some publishers use single quotation marks, and others use double quotation marks, LEFT DOUBLE QUOTATION MARK <“> (U+201C) and RIGHT DOUBLE QUOTATION MARK <”> (U+201D).

  • Finnish uses the RIGHT SINGLE QUOTATION MARK <’> (U+2019) as both opening and closing marks.

  • French uses single guillemets, LEFT-POINTING SINGLE ANGLE QUOTATION MARK <«> (U+2039) and RIGHT-POINTING SINGLE ANGLE QUOTATION MARK <»> (U+203A).

• Traditional Chinese: LEFT WHITE CORNER BRACKET <『> (U+300E) and RIGHT WHITE CORNER BRACKET <』> (U+300F).

• Simplified Chinese: LEFT SINGLE QUOTATION MARK <‘> (U+2018) and RIGHT SINGLE QUOTATION MARK <’> (U+2019).

  • The white corner brackets of Traditional Chinese are also in common use.

• Japanese: LEFT WHITE CORNER BRACKET <『> (U+300E) and RIGHT WHITE CORNER BRACKET <』> (U+300F).

• Korean: South Korea uses the same punctuation marks as English. North Korea uses full-width guillemets, LEFT ANGLE BRACKET <〈> (U+3008) and RIGHT ANGLE BRACKET <〉> (U+3009).

  • Corner brackets and white corner brackets are often used in practice.

• The placement of quotation marks in CJK is different by directionality, but this does not involve different Unicode codepoints, so the browser/operating system is assumed to handle positioning appropriately. (While there is e.g. a PRESENTATION FORM FOR VERTICAL LEFT WHITE CORNER BRACKET <﹃> (U+FE43) and PRESENTATION FORM FOR VERTICAL RIGHT WHITE CORNER BRACKET <﹄> (U+FE44), these are only included in Unicode as a compatibility character.)

• Nested quotation marks in English are routinely typed as straight quotes, APOSTROPHE <'> (U+0027), with the expectation that software will transform them contextually into paired quotation marks (“smart quotes”).

  Metanorma transforms straight quotes in Asciidoc source into smart quotes following English conventions.

• Simplified Chinese uses the font to make Western quotation marks (U+2018, U+2019) full-width, instead of deploying distinct codepoints

• In vertical directionality, Traditional and Simplified Chinese use presentation variants of the codepoints LEFT CORNER BRACKET <「> (U+300C) and RIGHT CORNER BRACKET <」> (U+300D).

A single initial marker used to represent one turn in an alternation of direct speech.

• Latin, Cyrillic: Properly HORIZONTAL BAR <―> (U+2015) is used; in practice EM DASH <—> (U+2014) is usually used.

  • The quotation dash is almost unknown in English, but is commonplace in other European languages, such as French.

• The Japanese PART ALTERNATION MARK <〽> (U+303D) has narrower application, being used in Noh drama to indicate the start of a speaker’s or the chorus’ part. The opening square quotation mark <「> may also be used.

There is a thin space after the quotation dash.

The conjunctive mark presents two or more words or morphemes as both applicable. It is a punctuation counterpart to “and”.

• In formal usage, ampersand is restricted to citations of multiple authors, and to disambiguate scope of conjunction (items in a list joined by “and”, some of which are themselves joined by “and”).

  Citation of multiple authors

  Jones & Jones (2005)

  Items in a list joined by “and”, some of which are themselves joined by “and”

  Rock, pop, rhythm & blues and hip hop

• In Latin script, the default conjunctive mark is AMPERSAND <&> (U+0026).

  • In Irish and Scots Gaelic, TIRONIAN SIGN ET <⁊> (U+204A) is used traditionally.

  • In Swedish, underlined o is also used.

  • In informal usage, the plus sign may be used.

• By default the ampersand is consider to stand in for “and” as a word, and is spaced either side as a word. It is printed without spaces when it is part of an acronym, e.g. R&D = “Research and Development”.

The disjunctive mark presents two or more words or morphemes as alternatives. It is a punctuation counterpart to “or”.

• The disjunctive mark can be used to express inclusive “or”, or “and”; in that case it overlaps with the conjunctive mark and the range mark.

• English uses both a disjunctive mark and range mark to express points on an itinerary; e.g. Shanghai/Nanjing/Wuhan/Chongqing or Shanghai–Nanjing–Wuhan–Chongqing

• In English, the expression of the disjunctive mark is the slash, SOLIDUS </> (U+002F).

• Normally there is no space either side of the slash. Some style guides require space when the phrase being joined contains a space:

  and/or
  New Zealand / Western Australia

A marker between two items, used to convey a range between the two. The items can be verbal values (Paris–New York), numerals (3–5), dates (Jan–Oct), quantities, etc. This function is represented in Metanorma i18n files as punct.en-dash.

• Latin script conflates the punctuation marks for verbal and numeric ranges. CJK uses distinct punctuation marks for verbal and numeric ranges. The latter function is represented in Metanorma i18n files as punct.numeric-en-dash.

• The range mark is also used to contrast two items, or convey a relationship between them: Mother–daughter relationship,

• Latin, Cyrillic: In good typography, EN DASH <–> (U+2013) is used. Informal use typically uses the hyphen, and some style guides prescribing formal use prefer the hyphen when conveying a relationship, in its role as a word divider ().

  • In French, TILDE <~> (U+007E) is used: 3~5 m.

• Traditional Chinese, Simplified Chinese, Japanese, Korean: EN DASH <–> (U+2013), WAVE DASH <〜> (U+301C), FULLWIDTH TILDE <～> (U+FF5E).

• Typically in Latin script there is no spacing around the en-dash. This differentiates it from the phrase stop use of the en-dash. Some style guides recommend spacing where it would avoid ambiguity in scope, e.g. 12 June – 3 July; others do not, e.g. 12 June–3 July.

• In French, the range mark can be open on either side of the interval: ~3 means “up to 3”, 100~ means “100 or more”.

• In CJK, verbal range is conveyed through en-dash. Tilde and single em-dash can also be used for verbal range.

• Numeric range in CJK is normally wave dash. Some Japanese academic writing uses colons instead of wave dash, though this is not universal.

Particularly in CJK, components of a name are separated from each other using a punctuation mark where necessary for clarity. This applies to names in the broadest sense, including personal names (particularly foreign names), professional titles and positions next to names, and names of works (i.e. titles, separating the title from the subtitle).

• There is a continuity of function between the name separator and the word divider in CJK, particularly as CJK uses ideographs for words (so words are not consistently broken down), and it uses word dividers only sparingly.

• The following are the more refined classes of name separator; because of the fluidity of classes, they are enumerated here rather than being broken down into separate semantic functions in the framework.

  • Components of foreign names (Chinese, Japanese)

  • Components of an organsiation name

  • Title and subtitle of a work

  • Professional title, professional position, personal name

  • Components of a bibliographic reference entry (e.g. Smith, J. 1989. The passage of time. New York: Wiley.)

• Caption markers () include a subset of name separators, but these are discussed separately, as a core concern of Metanorma configuration.

• Latin, Cyrillic: In Western scripts, this function is conflated, with the normal word separator () (space), with phrase stops (comma by default (), colon for title/subtitle (), both for organisation names), or with word divider () (hyphen, used in personal names).

  • The colon or em-dash is used for title/subtitle: Star Trek V: The Final Frontier, Star Trek V—The Final Frontier

  • The comma, colon, or em-dash is used for organsiation name: Yamato Bank: Osaka Branch

  • The hyphen is used within a single component of a personal name, a surname or a given name: Jean-Jacques, Zeta-Jones

• CJK: Various forms of interpunct (): MIDDLE DOT <·> (U+OOB7) in China, HYPHENATION POINT <‧> (U+2027) in Taiwan, KATAKANA MIDDLE DOT <・> (U+30FB) to separate Katakana words in Japanese.

  • Japanese also uses WAVE DASH <〜> (U+301C), to separate title from subtitle.

  • Japanese also uses the KATAKANA-HIRAGANA DOUBLE HYPHEN <゠> (U+U+30A0) and the FULLWIDTH EQUALS SIGN <＝> <U+FF1D>. This double hyphen is used to render Latin hyphen in transliteration of foreign names, i.e. in the function of separating single components of a Western name. That function occasionally uses the interpunct instead. More rarely, the double hyphen also is used instead of the interpunct to delimit names and titles in Western names, e.g. サー゠アーサー゠コナン゠ドイル “Sir=Arthur=Conan=Doyle”

  • Japanese occasionally uses the interpunct for Japanese names, particularly when there would otherwise be confusion as to where one name ends and another begins.

  • Japanese can use ideographic space to delimit components of an organisation name: 大和銀行　大阪支店 “Yamato Bank, Osaka Branch”.

• The interpunct is half-width in Chinese in print, but full-width in online material.

The attribution mark is used to delimit a quotation from its source.

• The attribution mark is comparable to the caption separator () and the breaking phrase separator (). It accordingly shares punctuation with them.

• The em-dash is typically used. In less formal writing, the colon and the comma are also used.

• When applied to a block quote, the attribution starts on a separate line, with the attribution mark dash. The separate line is typically indented.

• When used inline, the attribution mark dash follows the quotation directly.

The word separator is used to separate words from each other.

• In Latin and Cyrillic, this function is fulfilled by the space, which is not regarded as punctuation at all. The space is applied universally as a word separator.

  • This was not the case in antiquity. Roman inscriptions used an interpunct () as a word separator; Greek and Roman manuscripts did not use word separators. (The latter practice is known as scriptio continua, “continuous writing”)

• CJK by contrast defaults to not using word separators, outside of special contexts (here considered as name separators). CJK therefore has scriptio continua.

• There are circumstances in which a word separator is necessary in careful text in CJK. Japanese specifically uses the interpunct to separate ordinary Japanese words, outside of the special cases of name separators (), where the intended meaning would be unclear if the characters were written side-by-side. It is more commonly used to separate foreign words and names when written in kana: パーソナル・コンピューター (pāsonaru·konpyūtā “personal computer”).

The word divider indicates the compound structure of a word, dividing it into pieces smaller than a word.

• The word divider differs from the name separator in scope: name dividers operate on groupings of words, the word divider operates within a word.

• The grammatical word divider () is a special case of the word divider.

• The hyphen () is a special case of the word divider.

• The divisions that a word divider makes can vary in scope. What is divided up may be a compound consisting of meaningful units: independent words, or morphemes (e.g. prefixes: de-emphasise). Or, what is divided up may be a single meaningful unit, broken up into phonetic units (syllables, or letters: syllabification or spelling out of a word).

  Meaningful units

  post-war
  Indo-European

  Phonetic units

  R-E-S-P-E-C-T (spelling out)
  in-cre-di-ble (syllabification)

• The extent to which the word divider is used in compound words varies by language and by style. The use of hyphen to indicate compound words (e.g. pigeon-hole, craftily-constructed chair) is on the decline in English.

• There may be two levels of division represented, with different dividers used to differentiate between them. This applies because here are two actual levels of grouping, or because a component being connected is a multi-word expression with space in it

  Two levels of division

  Pre–Indo-European
  San Francisco–area

• In Latin and Cyrillic, the default word divider is the hyphen, HYPHEN-MINUS <-&#x200c;> (U+002D). Unicode also has HYPHEN <‐> (U+2010) to disambiguate it from the minus sign (), but this “Unicode hyphen” is little used.

  • The EN DASH <–> (U+2013) is used as a higher level of division in careful typesetting (Pre–Indo-European, San Francisco–area residents).

  • In lexicography, interpunct is also used for syllabification.

  • In some transcription practice, e.g. for Arabic, Japanese and Chinese into English. the apostrophe is used to indicate syllabification, as disambiguation (مصحف transcribed as mus’haf, to avoid the pronunciation “moo-shaf”; しんいち transcribed as Shin’ichi to indicate the syllabification is “Shi-n-i-chi” rather than “Shi-ni-chi”).

  • This is also done in Pinyin: Xi’an is two syllables, Xī-ān, whereas xian would be read as a single syllable.

Space is not meant to appear after word dividers normally, as any word separators would contradict the intent of the word divider as showing divisions within a single word.

Some instances of word divider are not meant to be conflated with the hyphen, and a line break between such word divisions is inappropriate. For such cases, NON-BREAKING HYPHEN <‑> (U+2011) is used.

The grammatical word divider, like the word divider, indicates the compound structure of a word, dividing it into pieces smaller than a word.

The word divider can indicate the compound structure of any compound word, and is not a normal part of conventional orthography. The grammatical word divider, by contrast, is used to designate specific grammatical functions of morphemes, and is part of the conventional orthography of a language.

• The use of a grammatical word divider distinct from the general word divider is idiosyncratic, and inconsistent between languages.

• The use of the grammatical word divider is also less widespread than the elision mark, with which English conflates it.

  • Wikipedia lists instances in Danish, Estonian, Finnish, Polish, Turkish, and Welsh. In the first four it is restricted to foreign words, where the inflection would be hard to parse as distinct from the unfamiliar word; in Turkish it is restricted to proper nouns, for similar reasons; and in Welsh it is used to disambiguate infixed pronouns. None of them use it regularly for every instance of the grammatical function, the way English does for possessive -’s.

  • English attaches possessive -s to nouns using a grammatical word divider. No other Germanic language does.

    English: Julia’s
    German: Julias

  • Current practice in English uses a grammatical word divider for apostrophes, but a normal word divider for unexpected verb inflections:

    Julia’s
    to-ing and fro-ing

  • Practice in English is in flux for particular functions: unexpected plurals, and verb inflections after vowels, used to be separated by a grammatical word divider; increasingly, that is not used at all.

    Older practice using grammatical word dividers

    bastinado’d
    KO’d
    B’s and C’s

    Newer practice avoiding grammatical word dividers

    bastinadoed
    KOed
    Bs and Cs

  • Grammatical word dividers are also increasingly avoided in formal labels, such as geographical place names

    Coffs Harbour (town in Australia, originally Korff’s Harbour)
    Earl’s Court, Barons Court (adjacent London Underground stations)

• In Latin and Cyrillic, the grammatical word divider is usually the apostrophe.

  • In fine typography, RIGHT SINGLE QUOTATION MARK <’> (U+2019) is used as the apostrophe. In typewriter usage, which is inherited in online usage, APOSTROPHE <'> (U+0027) is used.

The grammatical word divider function of the apostrophe has the same special handling as the elision mark function ().

The hyphen is a special case of word divider (). It is used to divide a word at a line-break, so that the amount of empty space in a line is reduced.

• The hyphen is in routine use in Latin and Cyrillic.

• The word delimiter is alien to CJK; accordingly, the hyphen is also alien to CJK.

The hyphen is only meaningful in that function at the end of a line, and no characters are meant to appear after it.

• An optional hyphen, SOFT HYPHEN (U+00AD) is often used in word processing, which is only rendered when near a line break.

The identifier divider is used to break up a token that is not a linguistic expression.

• The function of the identifier divider and the word divider are quite similar: they both are breaking up a token and showing its internal structure. They differ in that one divides a linguistic word, and the other divides a non-linguistic token.

• The function of the identifier divider is also similar to the decimal point, as a number token divider; however decimal points are dealt with separetely, as numeric formatting.

• Identifier dividers can apply to numbers as identifiers, but are distinct from decimal points in that they don’t have an arithmetic function. So when separating groups of digits in a phone number or a date, the hyphen is distinct from a decimal point.

• The components of a numeric date are also split by identifier divider.

• In the case of hierarchical identifiers of entities in a document, such as clauses, Metanorma Presentation XML notates this as <span class="fmt-autonum-delim">.

• The usual identifier divider is the hyphen. In Unicode, the identifier divider function is properly indicated with FIGURE DASH <‒> (U+2012), as that symbol is designed to have the same width as Arabic numerals.

• Period, slash, colon and parentheses are also used idiosyncratically, through analogy with their core meanings (hyphen as a word divider; sentence stop as delimiter of a meaningful unit; slash as a disjunctive mark; colon as an introductory stop; parentheses as additional material).

  • Period is the most common for hierarchical identifiers of document entities; colon is the most common for delimiting a following date.

The identifier divider is used to separate an identifier from surrounding text, whether at the start, the end, or both.

• Identifier delimiters distinct from word delimiters are very rare. The most common use of them is in structured documents such as under Metanorma, for subfigures and formulas. Whether an identifier delimiter is used is a matter of style convention.

• The identifier delimiter is similar in function to title marks () and emphasis marks ().

• Metanorma Presentation XML currently notates this function as <span class="fmt-autonum-delim">, although this is properly the notation for identifier dividers.

• The usual identifier delimiter is the close parenthesis, or surrounding parentheses.

In formal documents (particularly the standards in scope of Metanorma), titles of cross-referencable entities within the document (figures, tables, lists, list items, footnotes, clauses, etc.) are labelled in such a way as to isolate the identifier of the entity. This is done both in labelling the entity, and in cross-referencing that entity.

The identifier of the entity may be signalled with punctuation acting as a delimiter. In Metanorma Presentation XML, this is notated as <span class="fmt-label-delim">.

• Different kinds of entity can take different punctuation. Ordered list items are commonly delimited with a closing parenthesis instead of a period.

• Latin, Cyrillic: Period is the most common caption number delimiter, if a delimiter is used at all. Closing parentheses are common for ordered list items and footnotes.

If the number of the entity appears together with a caption or title, giving more information about the entity, punctuation is usually used to separate the two. In Metanorma Presentation XML, this is notated as <span class="fmt-caption-delim">

• Different kinds of entity can take different punctuation. For example, a document may use period as a caption separator for clauses, but dash for tables and figures.

• This function overlaps with the caption separator. If a caption number delimiter is used, a distinct separator is usually not used after it. The following illustrates different possible configurations.

  No caption number delimiter, colon as caption separator

  Table 1
  Table 1: Distribution of rice yields

  No caption number delimiter, period as caption separator

  Table 1
  Table 1. Distribution of rice yields

  Period as caption number delimiter, caption delimiter blocks distinct caption separator

  Table 1.
  Table 1. Distribution of rice yields
  NOT: Table 1.: Distribution of rice yields

• Latin, Cyrillic: All of the following may appear in this function: space (i.e. no special punctuation), colon, comma, em-dash, period

The caption or title of an entity in a document may be terminated with a punctuation mark. In Metanorma Presentation XML, this is notated as <span class="fmt-label-delim"> (conflating it with the Caption number delimiter.)

• This function is an extension of the sentence stop, since captions can be seen as sentences. Whether it is used or not depends on the publisher’s style conventions.

• Latin, Cyrillic: if a caption stop is provided, it is a period.

The hierarchical caption separator separates hierarchical components of a cross-reference to entity in a document (e.g. Clause 5, Note 1). This is used when the entity cross-reference by itself is ambiguous (e.g. note numbering restarts each clause, so “Note 1” on its own is ambiguous between the first note of clause 4 and of clause 5.) In Metanorma Presentation XML, this is notated as <span class="fmt-comma">.

• This is a special case of a name separator (), and shares punctuation with it: comma in Latin script.

• Latin, Cyrillic: if a caption stop is provided, it is a period.

• In some languages and styles, this function is conveyed by linguistic words instead of punctuation; e.g. in Japanese, the particle の “of” is used instead.

Annotation markers provide supplementary information or references related to the main text content.

• Annotation markers are routinely conflated with grouping markers (), which enclose, separate, or highlight specific text elements.

• Different kinds of annotation are indicated by different punctuation conventions, as described below, but these are usually not rigorously differentiated.

CJK full-width punctuation () applies.

Parenthetical annotation marks enclose supplementary or explanatory information. This function is represented in Metanorma i18n files as punct.open-paren and punct.close-paren.

• Parenthetical annotations can appear either within a sentence, or as a group of one or more sentences, between sentences.

• Parenthetical annotations can be conflated with grouping markers (), although they are less common in that function than for other paired punctuation marks.

• As an extension of their role, parenthetical annotations can also occur within a word, to indicate that a part of the word is optional in some sense. This is a special case of a word divider. Examples include indication of singular and plural, or masculine and plural; both give the letter distinguishing those grammatical categories in parentheses, as optional:

  (s)he
  plan(s)

• Within a sentence, paired breaking phrase (em-dashes) separators () can overlap in function with parenthetical annotations: the parenthetical information is presented as a break from the main sentence, and another breaking phrase separators resumes the main sentence.

  The red-nosed reindeer (Rudolph was his name) had a very shiny nose.

  The red-nosed reindeer—Rudolph was his name—had a very shiny nose.

• Latin, Cyrillic: By default, parenthetical annotations are rendered as LEFT PARENTHESIS <(> (U+0028) and RIGHT PARENTHESIS <)> (U+0029)

• Paired em dashes are used within a sentence, when the parenthetic annotation can be presented as a sentence break and resumption. This is associated with less formal style.

• CJK: FULLWIDTH LEFT PARENTHESIS <（> (U+FF08) and FULLWIDTH RIGHT PARENTHESIS <）> (U+FF09)

  • Use of a paired em dash equivalent is rare in CJK.

  • Japanese: Japanese in addition uses double wave dash, WAVE DASH <〜> (U+301C)

    〜〜答え〜〜

• Languages differ as to whether parentheses enclosing italicised text should themselves be in italics. In German, they are expected to; in English, they are expected not to.

• Parenthetical annotations can be nested within other parenthetical annotations. In informal writing, parentheses are used for both nesting and nested annotations. In formal writing, concern for ambiguity drives many style guides to require that the nested annotation be in brackets instead of parentheses, with any third level of nesting in curly brackets.

  From Wikipedia:

  Parentheses may be nested (generally with one set (such as this) inside another set). This is not commonly used in formal writing (though sometimes other brackets [especially square brackets] will be used for one or more inner set of parentheses [in other words, secondary {or even tertiary} phrases can be found within the main parenthetical sentence]).

Footnote annotations are text annotation marks that provide supplementary information or references related to the main text content, like parenthetical annotations. Unlike parenthetical annotations, the annotation does not appear inline with the text it is annotating, but in a separate place: either the bottom of a page (footnote), the end of a chapter or book (endnote), or, in older practice, the margin of a text (marginalia). A footnote mark is used to cross-reference the place annotated in the text (where it is a footnote reference), to the annotation content (where it is a footnote label).

• Latin, Cyrillic: there are two repertoires of marks drawn on for footnote marks

  • Where a small repertoire of footnote marks is required (e.g. footnotes are cycled through once for each page), and in older practice, footnote marks are drawn from a set of typographical symbols, in the traditional order <* † ‡ § ‖ ¶>, supplemented in ad hoc function by other symbols.

  • Where a large repertoire of footnote marks is required (e.g. footnotes are cycled through once per chapter or document), and in newer practice, footnote marks are drawn from an ordered sequence of numerals or letters. Arabic numbers are the most commonly used, but Roman numbers and letters of the alphabet also appear.

• There is normally no space between text being annotated and the footnote reference.

• There may be space between the footnote label and the annotation.

• Footnote marks are normally superscripts, both as footnote references and as footnote labels.

• Footnote references normally appear after sentence and phrase stops.

• There are different conventions on whether the repertoire of footnote symbols is cycled through (i.e. restarted from 1 or *) each page, each chapter, or once in a document.

• There may be a caption number delimiter after the footnote mark. Documents may differ in whether they place a caption number delimiter after footnote references and after footnote labels.

• In some documents, a semantic distinction is made between different footnote mark sequences.

• Grouping markers overlap with parenthetical annotations ().

• Grouping markers are used to indicate editorial interventions in text; the Leiden conventions () are a very formalised equivalent of the less granular conventions expressed by grouping markers. Editorial interventions are not here treated as a distinct semantic function.

• Brackets can be used to insert explanatory material, instead of parentheses. Typically the brackets represent an editorial intervention, rather than the authorial voice.

• Latin, Cyrillic: LEFT SQUARE BRACKET <[> (U+005B), RIGHT SQUARE BRACKET <]> (U+005D)

• CJK: FULLWIDTH RIGHT SQUARE BRACKET <［> (U+FF3B), FULLWIDTH RIGHT SQUARE BRACKET <］> (U+FF3D)

• The main use of braces in English normal text is to indicate editorial additions and interpolations. They are overwhelmingly used in formal notation systems instead.

• Latin, Cyrillic: LEFT CURLY BRACKET <{> (U+007B), RIGHT CURLY BRACKET <}> (U+007D)

• CJK: FULLWIDTH RIGHT CURLY BRACKET <｛ > (U+FF5B), FULLWIDTH RIGHT CURLY BRACKET <｝> (U+FF5D)

• Angle brackets have limited use to indicate editorial interpolation, or to indicate that a text was thought by a character instead of spoken. They are overwhelmingly used in formal notation systems instead.

• Latin, Cyrillic: Good typography expects MATHEMATICAL LEFT ANGLE BRACKET <⟨> (U+27E8), MATHEMATICAL RIGHT ANGLE BRACKET < ⟩ > (U+27E9) for mathematical and normal text usage. In informal usage, and in many formal notation schemes such as computer programming, LESS-THAN SIGN <&#x3c;> (U+003C) and GREATER-THAN SIGN <&#x3e;> (U+003E) are used instead.

• CJK: LEFT ANGLE BRACKET <〈 > (U+3008), RIGHT ANGLE BRACKET < ⟩ > (U+3009).

Semantic identification punctuation identifies a span of text as conveying a specific semantics, as distinct from the structural information conveyed by punctuation marks in the foregoing usage categories.

• Semantic identification overlaps routinely with emphasis marks (), and with quotation markers ().

• Latin, Cyrillic: Punctuation marks for most semantic identification, other than abbreviation marks, are not used in contemporary Latin and Cyrillic practice—although number marks were commonplace in ancient and mediaeval versions of those writing systems.

Abbreviation marks are used to indicate that a word is an abbreviation of another word, and is to be understood as such rather than as a phonetically intact word.

• Latin, Cyrillic: the usual abbreviation mark is the period.

  • The ambiguity of abbreviation mark and declarative sentence stop () is particularly pernicious, and is a pressing issue for Metanorma converting between Latin and CJK punctuation.

  • The slash, SOLIDUS </> (U+002F), is used for some abbreviations in English, especially involving initials of separate words or morphemes (w/o = “without”)

  • The apostrophe, as an elision mark (), is also used in some abbreviations: gov’t < government.

• CJK: while CJK languages do form truncated expressions (e.g. Běidà 北大 for Běijīng Dàxué 北京大学 “Peking University”), abbreviation marks are not used explicitly.

• There is variation as to whether an abbreviation of a multi-word phrase retains the spaces of its source phrase (U. S. vs U.S.).

• There is variation by language, style, and instance as to when to use the abbreviation mark, and whether to use abbreviation marks within an acronym (N.A.T.O. vs NATO, etc. vs etc).

The elision mark is used to indicate that a word is derived from a word of the same meaning in an older, more formal or more standard version of the language, through the deletion of letters or numbers in the base form.

• Letters may be deleted from the beginning, the middle, or the end of the base form.

  ’tis < it is
  shootin’ < shooting
  ’n’ < and
  bo’sun < boatswain

• Letters may be deleted from a single word derived from a phrase; the resulting word is called a contraction.

  isn’t < is not
  won’t < woll not (Middle English variant of will not)

• Numbers may also be deleted, specifically in expressions for decades (this is an English-specific convention):

  ’70s < 1970s

• Elision marks often differentiate colloquial forms from formal forms, and therefore words using elision marks are often avoided in formal style.

• Elision marks prioritise the formal form of a language, deriving elided forms from it; that can make its use politically contentious, if the prioritisation of that form becomes controversial.

  In the 19th century, Scots forms were derived from Standard English forms in spelling, using elision marks (e.g. a’ < all, gi’e < give); this was rejected in the 20th century, as Scots was increasingly regarded as a distinct language from English (with the words now spelled aw, gie), rather than as a corrupt form of London English; the older use of the elision mark is dismissed as the “apologetic apostrophe”.

  A similar backlash has occured in the Anglosphere against other such transcription of non-standard English (“eye dialect”).

• Latin, Cyrillic: in fine typography, RIGHT SINGLE QUOTATION MARK <’> (U+2019) is used as the elision mark. In typewriter usage, which is inherited in online usage, APOSTROPHE <'> (+0027) is used.

  • Both forms of the punctuation mark are called apostrophe.

• There is variation as to whether an abbreviation of a multi-word phrase retains the spaces of its source phrase (U. S. vs U.S.).

• There is no space before an elision mark in a contraction, even if the elision mark denotes the start of a new day (it’s < it is.)

• Data-entry of the straight APOSTROPHE character is automatically corrected by word processors to the curved RIGHT SINGLE QUOTATION MARK character. The conversion uses the identical mechanism as smart quotes for quotation marks ().

• The RIGHT SINGLE QUOTATION MARK punctuation mark is ambiguous between a quotation mark and an elision mark, as indeed its Unicode name indicates. That means that initial elision marks are incorrectly corrected from the straight quote: it is treated as an initial quotation mark, and changed to LEFT SINGLE QUOTATION MARK, whereas initial elision marks are always RIGHT SINGLE QUOTATION MARK.

  'n' (data entry)
  ’n’ (correct rendering)
  ‘n’ (incorrect smart quotes conversion, treating elision mark as quotation mark)

In writing systems where number symbols are ambiguous with letters of the script, it is routine to highlight the numeric use of letters to prevent ambiguity, through a cardinal number mark.

• Hebrew and Greek have distinct cardinal number marks (gershayim and geresh in Hebrew, keraia in Greek), including disambiguating marks for counts of thousands (quote mark in Hebrew, lower keraia in Greek) and for reciprocal fractions (double keraia in Greek).

  Greek numerals

  δʹ “4” (Greek letter delta, used as a number)
  ͵δ “4000”
  δ″ “1/4”

• Roman numbers used an overbar in antiquity to indicate that they were numbers; as it happens, an overbar (vinculum) was also used in the same period, to indicate that the number was to be multiplied by a thousand (so V͞I was used for both “6” and “6,000”). The number mark is rare in contemporary usage of Roman numbers.

In some languages, ordinal numbers are preceded by a mark indicating that this is an ordinal, and which is often derived from an abbreviation of the spoken word for “number”.

• In Romance languages such as French and Spanish, an abbreviation of a word for “number” is used, usually an abbreviation of Italian numero. In French and Spanish, this is n^o, with the o superscript. In British English, No. is used. In German, Nr. is used. Such abbreviations are not to be regarded as punctuation.

  • In informal use in the French- and Spanish-speaking world, DEGREE SIGN <°> (U+00B0) is used, as a truncation of n^o.

  • Some languages, such as Portuguese and Italian, use the masculine ordinal indicator <º> (U+00BA) instead of a superscript o.

• The number sign NUMERO SIGN <№> (U+2116) is very common in use in Russian and Bulgarian; it is derived from the Latin script abbrevation.

• In American English, NUMBER SIGN <#> (U+0023) (hash) is used.

• The abbreviations of “number” and their derivatives, including the Numero sign, are followed by space.

• The number sign (hash) is not followed by a space.

The title mark is used to indicate the title of a document. This is a core function in bibliographic rendering, but it is also applied outside of formal bibliographic rendering, in running text. This function is represented in Metanorma i18n files as punct.open-title and punct.close-title.

• Latin, Cyrillic: this is by default handled through italics rather than a punctuation mark. Quotation marks are often used instead, although in more formal referencing, quotation marks are used as secondary title marks instead.

• Traditional Chinese: WAVY LOW LINE (U+FE4F) is used as official title markup, particularly in texts that also use the proper name mark.

  • Square brackets 【】and double quotation marks『』are also used.

• Simplified Chinese: uses《…​》for titles.

  • Square brackets【】and double quotation marks『』are also used for song titles.

• Japanese: 『…​』(double quotes) are used for book titles.

  • Western quotes are unofficially used in press for titles in Traditional Chinese, Japanese, and Korean.

• The title mark is associated with the name separator (), to separate components of a bibliographic reference entry.

The subsidiary title mark is used to indicate the subtitle of a document.

• The title mark is associated with the name separator (), to delimit titles from subtitles.

• Latin, Cyrillic: subtitles are not marked up as a separate span from titles, and Western script referencing relies instead on the name separator to differentiate titles from subtitles.

• Japanese: WAVE DASH <〜> (U+301C) can be used to mark subtitles: 〜概要〜

The secondary title mark is used to indicate the title of a subsidiary document in bibliographic referencing. This applies to article and chapter titles, as distinct from book and journal titles. This function is represented in Metanorma i18n files as punct.open-secondary-title and punct.close-secondary-title.

• Latin, Cyrillic: this is by default handled through quotation marks.

• Traditional Chinese, Simplified Chinese: Single title marks 〈…​〉() are used for article and chapter titles (LEFT ANGLE BRACKET <〈> (U+3008) and RIGHT ANGLE BRACKET <〉> (U+3009).)

The proper name mark is used to indicate a proper name within a document.

• Writing systems can use the name separator instead (), to differentiate names from surrounding text.

• Traditional Chinese: an underline is occasionally used, in didactic and ambiguous contexts (teaching materials, movie subtitles).

A clause mark is used to indicate that an identifier is a cross-reference to a clause in a document.

• Use of the clause mark is now infrequent outside of legal documents, with the word for “clause” typically supplied instead. In ISO documents, a dot-delimited numeral on its own (i.e. featuring an <identifier-divider>>) is understood to be a clause reference, without an explicit indication of the cross-reference scope (“clause”).

• Latin: SECTION SIGN <§> (U+00A7)

• The clause mark is linked to the following identifier with a non-breaking space.

• When multiple clauses are referenced, the section sign may be duplicated: “§§ 13–21”. This is analogous to cross-reference abbreviations of reference locality types, such as “pp.” for “pages”.

A paragraph mark is used to indicate that an identifier is a cross-reference to a paragraph in a document.

• Use of the paragraph mark is infrequent outside of legal documents, with the word for “paragraph” typically supplied instead.

• There is limited use of the paragraph mark to represent carriage return at the end of a paragraph, or the concept of a paragraph: that use is not properly punctuation.

• Latin: PILCROW SIGN <¶> (U+00B6).

• The paragraph mark is linked to the following identifier with a non-breaking space.

Highlight markers highlight and draw attention to specific text elements through visual modification or annotation, for a range of functions.

• The range of functions for which highlight markers are used is open-ended, and often overlaps with other semantic functions. This includes emphasis on the part of the author; emphasis on the part of a speaker in direct speech; use–mention distinction (so citation of words as subjects of discussion, as opposed to use of words in language); titles and names of entities; and foreign words.

• The extent of use of highlight markers varies by language and style.

• Latin and Cyrillic do not use punctuation in this function, and resort instead to typographical styling, including italics, boldface, and underlining. These are not in scope of this framework, except where Metanorma needs to alternate between typographical styling in Latin/Cyrillic, and punctuation marks in CJK.

  • While some documents make meaningful distinctions between italics, boldface, and underlining, they are idiosyncratic to the document, and cannot be generalised readily. Italics is the default highlight marking device in English.

• CJK does not use italics, although it does use boldface.

  • Instead of italics, CJK uses a range of explicit punctuation, including single or double quotes, brackets, lenticular brackets, or emphasis marks.

  • For example, Japanese uses lenticular brackets in dictionaries for quoting Chinese characters and Sino-Japanese loanwords.

• CJK emphasis marks are a diacritic applied to every character in the text span to be emphasised. Chinese is usually restricted to using an underdot. Japanese uses a much wider range of emphasis marks: filled or hollow dot, circle, double circle, triangle, and “sesame” marks.

• While the underdot emphasis mark could be realised as a distinct Unicode character, CJK emphasis marks are treated in computer typesetting as markup. They are realised in CSS using the text-emphasis-style attribute. This means that Western and CJK emphasis marks are both treated as markup rather than as punctuation.

• CJK emphasis marks are rare online and unsupported in many Word processors. (Microsoft Word also supports a smaller range of emphasis marks in Japanese than does HTML CSS or XSL:FO.)

The decimal point delimits integer and decimal portions of a number.

• The decimal point is a special case of an identifier divider ().

• Latin, Cyrillic: there is variation among Western script languages: the Anglosphere uses a period, while Continental Europe uses a comma. In the case of ISO, even English-language texts adhere to Continental norms in using comma as a decimal point.

• Traditional Chinese: the HYPHENATION POINT <‧> (U+2027) (interpunct) is used as a decimal point in Chinese numbers: 三‧五 “3.5”.

• Japanese: the same practice is followed, using the Japanese version of the interpunct, KATAKANA MIDDLE DOT <・> (U+30FB): 三・一四 “3.14”.

The minus sign indicates negative numbers.

• The same sign is used for negative numbers and for the arithmetic subtraction operator; the former is in scope of this framework, as negative numbers can appear outside the context of mathematical typesetting, inserted into normal text.

• Fine typography prefers to use the distinct MINUS SIGN <−> (U+2212) in mathematical use. Informal use uses HYPHEN-MINUS <-> (U+002D).

Indicates a ratio or proportion of two numbers.

• The ratio function is a generalisation of both the fraction sign and the division sign in mathematics.

• Proportions can include a proportion of completion, e.g. Page 17/35, indicating how many pages have been traversed out of a total of 35.

• In English, both colon and slash are used: 1:7, 1/7.

  • Unicode differentiates the FRACTION SLASH <⁄> (U+2044), the division sign DIVISION SLASH <∕> (U+2215), and the non-mathmetical use of slash.

Indicates that a numeric value is approximate.

• TILDE <~> (U+007E) is used by default

  • Some languages such as French also use ALMOST EQUAL TO <≈> (U+2248).

The missing text mark represents omitted text. This function is represented in Metanorma i18n files as punct.ellipse.

• The ellipse is conflated with the hesitancy phrase separator (), but the two interact differently with other punctuation.

• The same mark, the elllipse, is used as for the hesitancy phrase separator.

  • Occasionally triple asterisks, <* * \*> or <***>, are used as an explicit missing text mark. This is done in some legal writing in English. In this regard, the missing text mark usually has scope at paragraph-level rather than sentence-level, and it is a variant of the section separator ().

• When the omitted material is part of a word, e.g. out of religious or social taboo, or anonymisation, one or two em dashes are used traditionally. For taboo deletion, asterisk is also used; for more expressive or jocular deletion, a range of symbols is used.

  Anonymisation

  It was alleged that D—— had been threatened with blackmail.

  Religious taboo

  In the name of G–d

  Social taboo

  F—— you!
  F*ck you!
  F*@# you!

• When the missing text mark represents one or more missing words, it is usually treated for spacing as a separate word, rather than as punctuation. Therefore unlike the hesitancy phrase separator, the missing text mark is preceded by space as a word separator, just as a word would. This extends to treating the missing text mark as a distinct sentence.

  So British practice, as reflected in the Oxford Style Guide, has the hesitancy mark replace the sentence stop, but the missing text mark appear after a sentence stop. It also has space before both the missing text mark and the hesitancy mark.

  The …​ fox jumps …​
  _The quick brown fox jumps over the lazy dog. …​ And if they have not died, they are still alive today.
  It is not cold …​ it is freezing cold.

• But as an illustration of the flux around spacing rules: the University of Oxford Style Guide (which applies to the university rather than to the commercial printer) requires no space around the missing text mark, and space after and not before the hesitancy mark:

  The…​fox jumps…​
  _The quick brown fox jumps over the lazy dog…​And if they have not died, they are still alive today.
  It is not cold…​ it is freezing cold.

• The same special handling applies as for the hesitancy phrase separator.

• As a sign of editorial intervention, the missing text mark is often put inside brackets () in English, to make the editorial intervention explicit, and to disambiguate it from the hesitancy phrase separator. In Spanish, parentheses are preferred in that role.

  Original text: The President said that, for as long as this situation continued, he would not be satisfied
  Reported text: The President said that […​] he would not be satisfied

The repetition mark indicates a place where text is repeated from a previous passage, and it is obvious visually what the repeated text is.

• Repetition marks are infrequent in contemporary use outside of specific contexts, to avoid ambiguity. Space saving is not now as pressing a concern as it was in the past in texts.

• English traditionally uses a ditto mark under each word to be repeated from a previous line; these can take the form of straight quotes, right smart quotes, or apostrophe.

  Black pens, box of twenty ... $2.10 +
  Blue  "     "   "  "      ... $2.35

  • In Quebec French and Greek, a right guillemet is used. In other European languages (e.g. Italian), a double prime is used; the English use of apostrophes is an approximation of this. In yet others (e.g. Swedish), the double prime ditto mark is set on the baseline, and surrounded by em-dashes.

  • In CJK, DITTO MARK <〃> (U+3003) is used.

  • These are clearly visual variants of each other, that have been conflated historically with other punctuation in the same local writing systems.

• In some citation conventions, if two consecutive reference entries in a bibliography have the same author(s), a repetition mark is used instead of repeating the author, consisting of two or three em-dashes.

  Smith, J. & J. Doe. 1990. The elements of style. New York: Wiley.
  ——— 1991. The rudiments of style. New York: Wiley.

The iteration mark indicates the repetition of a word or a component of a word.

• In Chinese, IDEOGRAPHIC ITERATION MARK <々> (U+3005) is used in casual usage to repeat a character, but is no longer used in formal usage.

• In Japanese, IDEOGRAPHIC ITERATION MARK <々> (U+3005) remains in common use for repeating a single kanji, and follows different norms from the Chinese instance of the punctuation mark.

  • Japanese also has repeat marks for repeating the preceding word or phrase of two or more characters: VERTICAL KANA REPEAT MARK <〱> (U+3031) for exact repetition, VERTICAL KANA REPEAT WITH VOICED SOUND MARK <〲> (U+3032) for repetition with the first letter voiced. The repeat marks are restricted to vertical directionality, and are no longer in common use.

    ところ tokoro “place”
    ところ〲; tokorodokoro “in places”

  • Hiragana and Katakana have distinct iteration marks. They are no longer in common use, but they still appear in names.

    Isuzu in Japanese is いすゞ, using a voiced iteration mark: I-su-{repeat, voiced}

    Japanese Wikipedia refers to “heart” in Kanji: 心. Its Hiragana entry こころ includes a dozen literary works titled kokoro “heart”. There is a distinct reference こゝろ to the 1914 novel by Soseki; being a much older work than the others, it has retained the Hiragana repetition mark, “ko-{repeat}-ro”.

• Among Latin-script languages, a normal or superscript number <2> is used as an iteration mark in Filipino, Malay, and Indonesian, although in Indonesian this is no longer official practice.

The internationalisation configuration files for Metanorma, managed in YAML format, include configuration values for punctuation marks, under the punct: heading. These values can be invoked directly in the templates for automatically generated text, and filled in with language-specific values, matching the requested semantic function.

The configuration files are set globally per language in the isodoc gem; their values are overridden in the configuration for a Metanorma flavour (and/or a Metanorma taste), and can optionally be overridden further in a configuration file supplied with the document. There is no mechanism for providing multiple configuration values for a document, and selecting which of the values to apply for a particular instance: once the overrides have been worked out, there is one configuration value applied per punctuation mark per document.

The overrides per flavour allow Metanorma to deal with SDO-specific alterations in punctuation; JIS for example uses punctuation more closely aligned with Western practice, whereas Plateau uses traditional Japanese punctuation.

As of this writing, the following punctuation values can be configured in internationalisation files. While the names correspond to English punctuation marks, the intention of the configuration is to support punctuation semantic functions; the specific semantic functions involves are named in the foregoing.

• colon ()

• comma ()

• enum-comma ()

• semicolon ()

• period ()

• close-paren ()

• open-paren ()

• close-bracket ()

• open-bracket ()

• question-mark ()

• exclamation-mark ()

• emphasis-mark ()

• em-dash ()

• en-dash ()

• number-en-dash ()

• open-quote ()

• close-quote ()

• open-nested-quote ()

• close-nested-quote ()

• ellipse ()

• open-title ()

• close-title ()

• open-secondary-title ()

• close-secondary-title ()

A subset of the foregoing is be translated from Latin punctuation in automatically generated text, via l10n():

• colon: :

• comma: ,

• semicolon: ;

• period: .

• close-paren: )

• open-paren: (

• close-bracket: ]

• open-bracket: [

• question-mark: ?

• exclamation-mark: !

• em-dash: —

• en-dash: –

• number-en-dash: – (invoked depending on context — both surrounding characters need to be Unicode numbers)

• open-quote: "

• close-quote: "

• ellipse: …

In this list,

• Quotation marks are out of scope, because of the ambiguity of apostrophe, and the high variability of quotation marks between languages of l10n() translation.

• Enumeration delimiters are not supported directly in l10n(), because of the ambiguity of the comma. They are handled instead through the multiple_and i18n YAML value.

• Title marks require explicit semantics to drive them, which are not available in the l10n() context, and introducing them is deferred to bibliographic processing.

Values in the internationalisation configuration files can reference other entries in the configuration (including values overridden downstream), using the Ruby-derived syntax #{ self.label }, where label is the key of the value to be referenced. These values can include punctuation. For example, the following is how a YAML configuration value references the current rendering of the enumeration comma: