Wikipedia:Manual of Style/Pronunciation

Page semi-protected
From Wikipedia, the free encyclopedia

Pronunciation in Wikipedia should be transcribed using the International Phonetic Alphabet (IPA), except in the particular cases noted below.

For English pronunciations, broad transcriptions should be used; these are intended to provide a correct interpretation regardless of the reader's accent. The system for doing this is outlined at Help:IPA/English, and the first instance should include a link to that page; for example: England (/ˈɪŋɡlənd/). This should be done using the template {{IPAc-en}}. The Wikipedia respelling system, using the {{respell}} template, can be used in addition to the IPA.

Phonetic transcriptions are not always the best way to render pronunciation. For brand names which are intended to be respellings of an existing word, it is better to provide that word than a phonetic transcription. Similarly, initialisms are better spelled out than transcribed. In both situations this will generally be unambiguous, and accessible to more of our readers.

For foreign-language pronunciations, a phonetic transcription is normally used, with a link to Help:IPA or to various language-specific IPA keys. If phonemic transcriptions are used, these require a link to a description of the phonology of the language in question, as otherwise the symbols used may be ambiguous.

Other options are to link to the corresponding entry in Wiktionary, or to include an audio file together with the transcription. Consider that Wikipedia is not a dictionary when thinking of adding a pronunciation to an article; also, if the pronunciation is included in the main article, it is best not to repeat it in various sub-articles. Besides the clutter, subsequent edits may result in contradictory pronunciations.

See also:

Appropriate use

Normally, pronunciation is given only for the subject of the article in its lead section. For foreign words and names, use the pronunciation key for the appropriate language. If a common English rendering of the foreign name exists (Venice, Nikita Khrushchev), its pronunciation, if necessary, should be indicated before the foreign one. For English words and names, pronunciation should normally be omitted for common words or when obvious from the spelling; use it only for foreign loanwords (coup d'etat), names with counterintuitive pronunciation (Leicester, Ralph Fiennes), or very unusual words (synecdoche).

Pronunciation should be indicated sparingly, as parenthetical information disturbs the normal flow of the text and introduces clutter. In the article text, it should be indicated only where it is directly relevant to the subject matter, such as describing a word's etymology or explaining a pun. Less important pronunciations should be omitted altogether, relegated to a footnote, or to a dedicated section in the article or infobox. Articles about phonology and other aspects of a language may normally use pronunciation keys throughout the text, as appropriate.

Do not include pronunciations for names of foreign countries whose pronunciations are well known in English (France, Poland). Do not include them for common English words just because they have pronunciations that might be counterintuitive for those learning the English language (laughter, sword). If the name consists of more than one word, include pronunciation only for the words that need it (all of Jean van Heijenoort but only Cholmondeley in Thomas P. G. Cholmondeley).

IPA style

Whenever the IPA appears in an article, it should be contained within the {{IPA}} template. This allows software to tell the text is in IPA, and registered users to assign a different font to display the IPA symbols. The [brackets] should be inside the {{IPA}} template for uniformity of the font.

When using the IPA, provide an explanation for the reader. If there are multiple instances of IPA in an article, you may want to use the template {{IPA notice}} at the top of the page. However, if there are only a few instances of IPA, you may instead wish to use a template to link the first to one of the help keys, such as:

  • {{IPA|fr|ɛɡzɑ̃pl}}

This yields:

  • French pronunciation: [ɛɡzɑ̃pl]

Without the label:

  • {{IPA|fr|ɛɡzɑ̃pl|}}

yields:

If the first parameter is an IETF language tag and the second a transcription, as in the above, {{IPA}} automatically provides a label and, if a dedicated key for the language exists, links the transcription to the key.

Audio samples may be included in the IPA template as |audio=.

When a specific phonetic pronunciation is indicated, as in foreign names, this is marked by square brackets. Normally a reader will not know the structure of the language in question well enough for a phonemic transcription in slashes to be useful. The use of slashes is only permitted in cases where the pronunciation represents phonemes, as in broad transcriptions of English. However, phonetic transcriptions of English may be useful to represent a specific accent, local or historical pronunciations, or how a person pronounces their own name. For example, the English name Florence would normally be given the generic transcription /ˈflɒrəns/, but in the case of Florence Nightingale we have a recording of her saying her name, and she pronounces it [ˈflɒɾəns], with a flapped ar [ɾ] that is no longer the norm in English. Non-universal pronunciations such as these should be clearly marked for what they are.

Distinction between varieties of English

It is often possible to transcribe a word in a generic way that is not specific to any one accent, e.g. Oxford as /ˈɒksfərd/. Speakers of non-rhotic accents, as in much of Australia, England, New Zealand, and Wales, will pronounce the second syllable [fəd], those with the fatherbother merger, as in much of the US and Canada, will pronounce the first syllable [ˈɑːks], and those with the cotcaught merger but without the fatherbother merger, as in Scotland and Boston, will pronounce the first syllable [ˈɔːks], but since such variation on the part of the speaker is automatic, it need not be spelled out in a simple pronunciation guide to a key word in an article. Indeed, the Help:IPA/English key, designed for readers who are unfamiliar with the IPA, simply defines the sequence /ər/ as the sound at the end of letter, and warns that it may not be distinct from /ə/ for many people. That is, there is little point in transcribing Oxford as [ˈɒksfərd], [ˈɒksfəd], [ˈɑːksfərd], [ˈɑːksfəd], [ˈɔːksfərd], or [ˈɔːksfəd], depending on accent, and this would add a considerable amount of clutter to the article.

If the pronunciation in a specific accent is desired, square brackets may be used, perhaps with a link to IPA chart for English dialects, which describes several national standards, or with a comment that the pronunciation is General American, Received Pronunciation, Australian English, etc. Local pronunciations are of particular interest in the case of place names. If there are both local and national or international standards, it may be beneficial to list both. As long as a transcription is linked to the Help:IPA/English key, the conventions of the key should be used—even if the pronunciation itself is only found in certain locations—so that it can be pronounced in other accents as well.

Foreign names

When a foreign name has a set English pronunciation (or pronunciations), include both the English and foreign-language pronunciations; the English transcription must always be first. If the native name is different from the English name, the native transcription must appear after the native name.

For example:

'''Venezuela''' ({{IPAc-en|ˌ|v|ɛ|n|ə|ˈ|z|w|eɪ|l|ə}};
 {{lang-es|República Bolivariana de Venezuela}},
 {{IPA|es|reˈpuβlika βoliβaˈɾjana ðe βeneˈswela|pron}})

which gives: Venezuela (/ˌvɛnəˈzwlə/; Spanish: República Bolivariana de Venezuela, pronounced [reˈpuβlika βoliβaˈɾjana ðe βeneˈswela])

Similarly,

'''Nikita Khrushchev''' ({{IPAc-en|n|ɪ|ˈ|k|iː|t|ə|_|ˈ|k|r|ʊ|ʃ|tʃ|ɛ|f}};
 {{lang-rus|Никита Хрущёв}} {{IPA|ru|nʲɪˈkʲitə xruˈɕːof|}})

gives: Nikita Khrushchev (/nɪˈktə ˈkrʊʃɛf/; Russian: Никита Хрущёв [nʲɪˈkʲitə xruˈɕːof])

Transcriptions should always have a label identifying which language they are transcribing, with the exception of English. They are normally given in the national or international standard of the language in question, unless there is a reason to give a more local pronunciation. For example, the Help:IPA/Spanish key generally uses Castilian Spanish as its standard, for Venezuela [beneˈθwela], but the local pronunciation of [beneˈswela] may be considered more relevant. If a local pronunciation is transcribed, it should be marked as such:

'''Venezuela''' ({{IPA|es|beneˈswela|local}})
Venezuela (locally [beneˈswela])

Language-specific templates should not be used for local pronunciations unless the keys they link to cover the IPA symbols that are used. (In this case, either IPA|es or IPA|all would work.)

However, in language articles such as Spanish phonology, where the phonology is made explicit, examples may be given in either phonetic or phonemic notation, depending on the point being made, as the reader will have the information available to make sense of either. If for some reason it is desired to indicate the pronunciation of a foreign word phonemically in a non-linguistic article, a link should be provided to the phonology of the language in question.

Tone

Tone should always be included in the transcriptions of tonal languages. Because tone numbers are ambiguous—the reader may not know whether [ma4] is supposed to be high tone, low tone, or a tone number, for example—IPA transcriptions should use diacritic marks ([má]) or tone letters ([ma˦]), unless the article explains the numbering system.

Other transcription systems

If a language is not usually written in the Latin alphabet, an official romanization may exist for it. For example, pinyin for Standard Chinese and the Royal Thai General System of Transcription. In such cases, both the romanisation and the IPA rendering may be given.

For English words, transcriptions based on English spelling ("pronunciation respellings") such as prə-NUN-see-AY-shən (using {{respell}}) may be used, but only in addition to the IPA ({{IPAc-en}}). Whatever system is used, any transcription should link to an explanation of its symbols, since such symbols are not universally understood.

For other languages, only the IPA is normally used. Respelling foreign pronunciations into English is inadequate and misleading. If an English respelling is given for a Welsh or Māori name, not only would it be bad Welsh or Māori but the implication would be that it's the English pronunciation. Nonetheless, an ad hoc description of a foreign-language word in that language is permitted. An example is Renault 4CV. This is called the Quatre chevaux in French, but the re is elided, so the French pronunciation can be clarified with "pronounced quat'chevaux" (this appearing with the French IPA transcription [kat.ʃəvo], assuming the editor finds that worth including).

Ad hoc descriptions such as "rhymes with both" or "rhymes with paid" may be useful for describing English sounds, but many such descriptions (e.g. "rhymes with bath", "rhymes with caught") will be interpreted differently depending on the reader's accent, so caution is advised, and this approach should not be used alone.

However, when a name is intended to be a homonym of an existing English word or phrase, as is the case with many brand names and entertainers' names, giving that word or phrase may be all that is needed:
     The '''Motorola Razr''' (styled '''RAZR''', pronounced "razor") produces: The Motorola Razr (styled RAZR, pronounced "razor")

Nonetheless, even here some cases may be ambiguous. For example:
     '''Kioti Tractor''' ({{IPAc-en|k|aɪ|'|oʊ|t|i}} "coyote") produces: Kioti Tractor (/kˈti/ "coyote")

If the IPA was omitted, it would not be clear which of the pronunciations of coyote is intended. In such cases, combining the IPA with the intended homonym may be needed. These should not be formatted with the {{respell}} template, as they do not follow that format. For example:
     '''Peter John Coke''' ({{IPAc-en|'|k|ʊ|k}} "cook") produces: Peter John Coke (/ˈkʊk/ "cook")
There, the "cook" should not be formatted with the {{respell}} template, because then it would need to be {{respell|KUUK}} (giving KUUK) to match the key it is linked to.

Places where other systems without the IPA are often appropriate are initialisms and names composed of numbers or symbols. This is because the names of the letters, numbers, and symbols can be spelled out in normal English orthography in a way that makes the pronunciation unambiguous across dialects. For example, Dead on arrival (DOA) may be better explained as "(an initialism: D-O-A)" rather than as the equally correct but less accessible /ˌdˌˈ/. Similarly:
     '''C++''' ({{IPAc-en|ˌ|s|iː|ˌ|p|l|ʌ|s|ˈ|p|l|ʌ|s}} "cee plus plus"<!-- "cee" is the name of letter "c" -->)
     produces: C++ (/ˌsˌplʌsˈplʌs/ "cee plus plus")

Rather than solely using the less accessible /ˌsˌplʌsˈplʌs/. Similarly:
     '''javac''' (pronounced "java-cee") produces: javac (pronounced "java-cee")

See English alphabet#Letters for how the names of the letters of the alphabet are spelled.

Similarly, the dispute over how to pronounce the X in Mac OS X may be better described as ten versus ex rather than as /ˈtɛn/ versus /ˈɛks/. In the case of Z, spelling out the letter as zee or zed is sufficient, if only one is considered correct.

Respelling should also be avoided when a respelled syllable would be the same as an existing word that is pronounced differently. "Maui" /ˈmi/ respelled as MOW-ee, "metonymy" /mɛˈtɒnɪmi/ as meh-TON-im-ee, and "cobalt" /ˈkbɒlt/ as KOH-bolt are susceptible to being misinterpreted as /ˈmi/, /mɛˈtʌnɪmi/, and /ˈkblt/, because of the existing words "mow", "ton", and "bolt", so only IPA should be provided for such words, if any.

Placement

Opening of lead

Transcriptions are frequently placed immediately after the head words of the article in dictionary format:

London (/ˈlʌndən/ LUN-dən) is the...

or even,

London /ˈlʌndən/ is the...

This method is good when the pronunciation can be adequately covered in a short parenthetic string. However, this method can become distracting, especially when alternate transcription conventions are used, if there are regional differences, or if the pronunciation is otherwise not straightforward. In such cases, other options are better. Some articles, such as Halley's Comet, have a naming or pronunciation section that covers pronunciation explicitly. In other cases the pronunciation is given in the lead, but in a second dedicated sentence, leaving the first sentence unencumbered to define the term.[example needed]

Footnote

Another possibility is to relegate everything beyond the most common pronunciation to a footnote, or to remove all of it to a footnote. Examples:

☒NGreenwich Village (/ˈɡrɛnɪ/ GREN-itch, /ˈɡrɛnɪ/ GREN-ij, /ˈɡrɪnɪ/ GRIN-itch, /ˈɡrɪnɪ/ GRIN-ij),[1] in New York often simply called "the Village"...

checkYGreenwich Village,[pron 1] in New York often simply called "the Village"...

☒NFranklin Delano Roosevelt (/ˈrzəvɛlt/ ROH-zə-velt or /ˈrzəvəlt/ ROH-zə-vəlt; January 30, 1882 – April 12, 1945), also known by his initials, FDR, was the 32nd President of the United States (1933–1945).

checkYFranklin Delano Roosevelt[a] (January 30, 1882 – April 12, 1945), also known by his initials, FDR, was the 32nd President of the United States (1933–1945).

In the latter of each example, the pronunciation can be read in a footnote. It needs to be listed in the notes or references section with the appropriate wikitext immediately after the heading, in a new line, using {{notelist}}, <references/>, or <references group=pron/> if it's a named reference (changing "pron" to the relevant text). If this is not done an error notice would appear when previewing the edits or after the fact when accessing the editing window. When this happens, the footnote cannot be accessed at its location.

Notes
References

See WP:REFNOTE for help with embedding references within footnotes like this.

Infobox

Several infobox templates provide for a pronunciation entry. In such articles there is generally no need to repeat it in the lead. Examples:

IPA templates on Wikipedia

The IPA should always be enclosed within an IPA template. This ensures proper formatting across browsers; it also enables editors to more easily find and review IPA transcriptions.

There are several types of IPA template. The simplest is {{IPA}}, which merely formats the enclosed text. It is normally used when the reader can be expected to follow the IPA, either because it's found in an article on phonology where the symbols are defined, because the article is tagged with {{IPA notice}}, or because an earlier instance of IPA in the text was enclosed with one of the following templates, which are designed to be more useful to the reader than the generic IPA notice.

English

For more aid to the reader, there are multiple templates specific to various languages. For generic English, {{IPAc-en}} should be used, which will link the notation to Help:IPA/English, which is a key of established conventions for transcription of English on Wikipedia, and automates conversion to IPA and provides mouse-over keys. A description of the various parameters of these templates, such as marking a pronunciation as American, British, or local, or adding sound files, can be found on the template page. When using any key-linking IPA template such as these, English or foreign, an editor should transcribe using the conventions of the key it links to; for example, the generic English ar sound is transcribed /r/ in Wikipedia articles, not */ɹ/, and is used where speakers of rhotic dialects would pronounce it, even in personal and place names. (These diaphonemic conventions, which are not specific to any one dialect or national standard, are covered at the top of the Help:IPA/English key.)

{{IPAc-en}} accepts both IPA and SAMPA. Each phoneme, including rhotic vowels, should receive its own cell (such as the 'ou' here), and not broken up, or the wrong popups will appear. See the template page for details.

{{IPAc-en|ˈ|m|ɛ|k|s|ɪ|k|oʊ}}/ˈmɛksɪk/
{{IPAc-en|'|m|E|k|s|I|k|oU}}/ˈmɛksɪk/

Technical Latin or Greco-Latin words in fields such as biology, astronomy, mythology and medicine cause frequent problems. If there is one generally accepted pronunciation in the field, use that. However, there are often multiple pronunciations heard, along a cline from highly anglicized pronunciations, as found in Shakespeare, to attempts to remain faithful to the Latin or Greek pronunciation. For example, Io may be pronounced either /ˈ/ or /ˈ/. Both are "correct". However, it may be impractical to list all possible pronunciations. In such cases, the traditional (literary) pronunciation is the most difficult as well as the most anglicized and is therefore the one that should be transcribed. Other conventions are generally straightforward and can all be covered, simultaneously, by including the Latin or Greek orthography, or the Greek in Latin transcription. As long as Latin long vowels ā ē ī ō ū ȳ are indicated, readers will be able to pronounce the word according to the convention of their choice. (Note that both the Latin and Greek alphabets are defective when it comes to vowel length, which determines the location of English stress in these words.)

To transcribe the pronunciation of a particular individual or dialect, or to use transcription conventions other than those of the IPA-for-English key, |generic=yes can be used, which links to a generic IPA key that is not restricted to any one dialect or language. It is often useful to add a link to a phonological description of the dialect being transcribed; if this is done, there will normally not be any need for the generic IPA key. Such links may be done manually, as in:

{{IPA|[[Australian English phonology|[ˈmɛɫbn̩]]]}}[ˈmɛɫbn̩]

The distinction between /slashes/ used for generic English and [brackets] for individual or dialectical pronunciations: A phonemic transcription (between slashes) is not meaningful without a description of the phonology of the speech variant, whereas a phonetic transcription (between brackets) can stand on its own, and allowing a choice between slashes or brackets in a template leads to frequent misuse of the symbols. As long as the phonology of the speech variety is accessible to the reader, as with the [ˈmælbən] example above, slashes may be used (/ˈmælbən/), but this will need to be done manually. Place any brackets or slashes within the IPA template, so that they are formatted in the same size and font as the enclosed transcription. (This may be individuated at your CSS code by adding a line such as .IPA { font-family: Gentium, Charis SIL !important; }.)

Other languages

In the first parameter of {{IPA}}, enter the 2-letter ISO 639-1 code or the 3-letter ISO 639-3 code for the language in question, as in {{IPA|el|...}} for Greek or {{IPA|fa|...}} for Persian. A few languages also have dedicated templates that automatically convert ordinary letters (or conventional ASCII equivalents) to IPA characters that are used to transcribe the language in question, such as {{IPAc-cmn}} for Mandarin Chinese and {{IPAc-pl}} for Polish. These languages and templates are listed at {{IPA}}. Again, if the language you're transcribing has such an IPA key, use the conventions of that key. If you wish to change those conventions, bring it up for discussion on the key's talk page. Creating transcriptions unsupported by the key or changing the key so that it no longer conforms to existing transcriptions will confuse readers.

If the template is used by a sufficient number of articles, a dedicated key for that language can be created under Help:IPA/. Once the key is ready, add it to {{IPA keys}}.

To link a single-phone transcription to an article on that phone, use {{IPA link}}; for example, . The template {{IPAblink}} provides brackets for phonetic transcriptions, and {{IPAslink}} provides slashes, for phonemic transcriptions. For an extended transcription, the template {{WikiIPA}} may be used at the beginning of the text, with the transcription itself formatted only with {{IPA}}.

Cleanup

If you're not sure how to fix a messed-up, provincial, or ambiguous transcription, there are several cleanup templates, which are monitored by editors who do. These are:

  • {{Needs IPA}}, an inline tag for pronunciations which do not include the IPA, for which an attempt at the IPA is inadequate (this includes correct IPA for what appears to be an incorrect pronunciation) or articles with IPA problems;
  • {{Pronunciation needed}} if you wish to have a pronunciation and none is provided.

Note that a simple pronunciation hint, such as an unambiguous rhyme ("rhymes with kind"), an unambiguous homonym ("pronounced like Smith"), or a spelled-out acronym ("pronounced S-O-S"), does not generally need an IPA transcription. However, more elaborate transcriptions, such as ad hoc respellings and US-dictionary formats, should be replaced by the IPA.

Entering IPA characters

Many IPA characters cannot be typed with a regular keyboard layout, but there are various ways to enter them.

Beneath the edit box on Wikipedia is a character map (see Help:CharInsert § IPA for more information).
Choose IPA in the drop down box on the left, then just click on the symbol you want, and it will be added at the insertion point in the edit box.
Copy and paste from an online IPA keyboard
Utilities like the IPA i-charts, IPA character picker 19, TypeIt, or IPA Chart keyboard cover the complete range of IPA symbols and diacritics which are not available from the Wikipedia IPA character map.
Copy and paste them from elsewhere (other articles or websites, for example)
This method sometimes fails when copying characters with diacritics.
Many of the familiar Latin letters can be typed with a normal English keyboard layout. However,
( ː ) note the length mark is a different character from a colon
( ˈ ) note the primary stress mark is a different character from a typewriter apostrophe
( ɡ ) note the symbol for the voiced velar stop is a different character from the lowercase English letter g
On Mac English keyboard layouts ("ABC"), a few special characters can be typed:
option-c for ç
option-o for ø
option-q for œ
option-' for æ
On Mac English extended keyboard layouts ("ABC — Extended"), most special characters, including composed diacritical symbols used in IPA, can be typed using option or shift-option. To see the live mapping from keys to (compositions of) characters, turn on the Keyboard Viewer which is a virtual keyboard (also known as the soft keyboard).
Enter them using a special character utility
On macOS, use the Character Viewer (formerly Character Palette): choose the "Phonetic Symbols" category and double-click on a character to enter it
Type them with a custom keyboard layout
For macOS, download the IPA-SIL keyboard layout

Numeric entry

The following methods require you to know the Unicode code point of the character you wish to enter: for example, the IPA symbol [ɒ] is represented by the hexadecimal value U+0252. See the resources links below for reference charts.

Enter them using a numeric keyboard input method
On most Windows keyboard layouts, type alt-0–2–5–2 for ɒ (Note: Requires you to first set the registry key HKCU\Control Panel\Input Method\EnableHexNumpad to type REG_SZ and value 1 and reboot.)
On macOS, use the Unicode Hex Input keyboard layout, type option-0–2–5–2 for ɒ
On macOS Yosemite 10.10.5 you can hold down a key for a second and a number of diacritics will appear above the cursor as clickable options. 'a' for example offers à á â ä æ ã å and ā.
Enter them into wikitext as HTML character entities
List of XML and HTML character entity references: enter &aelig; for æ
Numeric character reference (decimal): enter &#0594; for ɒ
Numeric character reference (hexadecimal): enter &#x0252; for ɒ

Resources

Related templates

  • {{IPA}} allows proper display of IPA characters.
    • {{IPAc-en}} has slashes and a link to Help:IPA/English; it accepts ASCII input and displays a mouse-over key to the IPA. This is the preferred template for generic English.
  • {{IPA notice}} is a message that warns the reader of the presence of IPA characters
  • {{Needs IPA}} is an inline template noting need for a single instance of transcription.

See also