Help:IPA conventions for English

The various English dictionaries use different and sometimes conflicting IPA transcriptions for English. For example, the transcription may be used for the vowel of sit, of seat, or at the end of city. A dictionary may not even be consistent between one edition and the next. This table correlates the more widely used dictionaries with the conventions of the Help:IPA for English key.

Most dictionaries transcribe a specific dialect or accent, such as the Received Pronunciation (RP) of the Oxford English Dictionary, or a narrow range of dialects. Wikipedia's Help:IPA for English key, on the other hand, is intended to cover RP, General American, Australian, and other national standards. As such, Wikipedia transcribes /r/ where it is found in rhotic dialects, but also the vowel distinctions found in non-rhotic dialects, without distinct UK and US transcriptions. Specific dialects may also be transcribed—local pronunciations of place names are often useful, for example—but they are normally written in addition to a more universal pronunciation.

When entering IPA in an article, please use the IPA template so that it is formatted properly on all web browsers. /Slashes/ and [brackets] should be included inside the IPA template, so that they display in the same font as the IPA itself. Also, please use proper stress and length  marks (available at the bottom of your edit window) rather than the non-IPA shortcuts of apostrophe ' and colon :. Depending on the reader's font preferences, the latter can be ambiguous.

For a list of those languages other than English which have agreed-upon transcriptions in Wikipedia, see IPAkeys. For a comparison of the non-IPA transcriptions found in many US dictionaries, see Pronunciation respelling for English.

Alternative pronunciations
When dictionaries give alternative pronunciations, they may mean that people disagree. For example, some people pronounce bath, with the vowel of bat, while others with the same accent pronounce it , with the vowel of bra. This is the kind of difference celebrated in "You like to--toes and I like to--toes". We would normally need to transcribe both, unless only one is considered correct, as may be the case for personal and place names.

However, often variant transcriptions reflect distinctions between accents, and these we do not need to transcribe, since our IPA key already covers such distinctions. For example, dictionary.com transcribes horse as "" and hoarse as "". The two transcriptions of hoarse are meant to show that some people pronounce it the same as horse; it does not mean that there are two pronunciations of hoarse among those who either do or do not make this distinction. (See horse-hoarse merger.) It would not be possible to have the song lyric "You say hoarse and I say horse", because only those people who say hoarse would be able to sing it. And indeed in the OED there is only one pronunciation for each word: horse and hoarse. Therefore on Wikipedia we would only have one transcription for each: horse, hoarse. Since the IPA key defines the orthographic conventions of and  according to basic English words, readers who do not make the horse-hoarse distinction will see  and  as being equivalent, much as the spelling pronunciations YOU-clid and EWE-clid for "Euclid" would be seen as equivalent.

Consonants
Consonants vary little between dictionaries. The ones which do are those in the words:
 * rich, char ;
 * which, ;
 * and new,.

Wikipedia editors have decided to go with for these words.

A few dictionaries, such as dictionary.com, use "/y/" for, but is more properly a vowel, and appears as such in transcriptions of French and German, often alongside English IPA transcriptions.

Vowels
Parentheses are not used on Wikipedia, as such transcriptions aren't truly phonemic. In non-rhotic accents such as RP, is not pronounced unless followed by a vowel, and non-rhotic dictionaries generally leave it out. American and Australian dictionaries tend not to indicate vowel length. Other differences are highlighted in red.

Stress
One-syllable words may have stress. Most dictionaries leave it out, but that can be confusing when several such words are strung together. For example, in the name Zack de la Rocha, Zack and Rocha have stress, but de la does not:. It would therefore convey an incorrect pronunciation to leave the stress mark off Zack.

OED2 does not indicate stress on monosyllables, but uses the stress mark to disambiguate disyllables: higher vs. hire. On WP, the distinction is made with the aid of the syllabicity mark:.

Dictionaries also disagree on secondary stress. Generally, any stressed syllable prior to the last is marked as secondary, and that convention is followed here. However, several dictionaries also mark full (unreduced) vowels as having secondary stress when they come after the primary stress, even though they are not actually stressed: cerebrate, dict.com, OED2. This practice is avoided on Wikipedia; if you have a word transcribed, it should probably be :.