# Help:IPA/English

Short description: Help page
Short description: Wikipedia key to pronunciation

Throughout Wikipedia, the pronunciation of words is indicated by means of the International Phonetic Alphabet (IPA). The following tables list the IPA symbols used for English words and pronunciations. Please note that several of these symbols are used in ways that are specific to Wikipedia and differ from those used by dictionaries.

If the IPA symbols are not displayed properly by your browser, see the links below.

If you are adding a pronunciation using this key, such pronunciations should generally be formatted using the template {{IPAc-en}}. The template provides tooltips for each symbol in the pronunciation. See the template page for instructions.

## Key

If the words given as examples for two different symbols sound the same to you (for example, if you pronounce cot and caught the same, or do and dew, or marry and merry), you can pronounce those symbols the same in explanations of all words. The footnotes explain some of these mergers. (See also Dialect variation below.)

If there is an IPA symbol you are looking for that you do not see here, see Help:IPA, which is a more complete list. For a table listing all spellings of the sounds on this page, see English orthography § Sound-to-spelling correspondences. For help converting spelling to pronunciation, see English orthography § Spelling-to-sound correspondences.

Consonants
IPA Examples
dj dew[2]
ð thy, breathe, father
f fan, leaf
ɡ guy, bag
hw whine[3]
j[4] yes, hallelujah
k sky, crack
l [5]
lj lute[2]
m my, smile, cam
n nigh, snide, can
nj new[2]
ŋ sang, sink, singer
p
r[6] rye, try, very
s sigh, mass
sj consume[2]
ʃ shy, cash, emotion
t [1]
tj tune[2]
China, catch
θ thigh, math
θj enthuse[2]
v
w wine, swine
z zoo, has
zj Zeus[2]
ʒ pleasure, beige[7]

Marginal segments
IPA Examples
x ugh, loch, Chanukah[8]
ʔ uh-oh /ˈʔʌʔoʊ/
ɒ̃ bon vivant[9]
æ̃ fin de siècle[9]
ɜː Möbius (UK only)[10]
Vowels
Strong vowels ...followed by R[11]
IPA Examples IPA Examples
ɑː PALM, bra ɑːr START
æ TRAP, tattoo[14] ær marry[15]
PRICE, pie[16] aɪər hire[17]
aɪ.ər higher[18]
MOUTH, how[16] aʊər flour[17]
aʊ.ər flower[18]
ɛ DRESS, prestige[19] ɛr merry[15]
FACE ɛər SQUARE, Mary[15][20]
eɪər player[18]
ɪ KIT, historic[21] ɪr mirror
FLEECE, pedigree, idea[22] ɪər NEAR, serious[20]
iːər freer
GOAT[23][21] oʊər mower
ɔː THOUGHT[24] ɔːr NORTH, FORCE[25]
ɔːər sawer
ɔɪ CHOICE ɔɪər coir[17]
ɔɪ.ər employer[18]
ʊ FOOT ʊr courier
GOOSE, cruel[22] ʊər tour, CURE (/ˈkjʊər/)[20][26]
uːər truer
ʌ STRUT, untidy, justiciable[27][28] ɜːr NURSE, blurry, urbane, foreword[29]
ʌr hurry[30]
Weak vowels
IPA Examples IPA Examples
ə COMMA, about ər LETTER, forward, history[31]
ɪ edition, rabbit, Latin, heating[21][32]
i HAPPY, mediocre[33] California[34]
u fruition[35][33] influence[36]
Syllabic consonants[31]
IPA Examples IPA Examples
əl bottle (either [əl] or [l̩]) ən button (either [ən] or [n̩])
əm rhythm (either [əm] or [m̩])

Stress[37] Syllabification
IPA Examples IPA Examples
ˈ intonation /ˌɪntəˈneɪʃən/ . /ˈhaɪər/ hire, /ˈhaɪ.ər/ higher[38]
/ˈtæks.peɪər/ taxpayer
ˌ

Notes

• Words in SMALL CAPITALS are the standard lexical sets. Words in the lexical sets BATH and CLOTH may be given two transcriptions, the former either with /ɑː/ or /æ/, the latter with /ɒ/ or /ɔː/.
• The length mark ⟨ː⟩ does not mean that the vowels transcribed with it are always longer than those without it. When unstressed, followed by a voiceless consonant, or in a polysyllabic word, a vowel in the former group is frequently shorter than the latter in other environments (see Clipping (phonetics) § English).

## Dialect variation

This key represents diaphonemes, abstractions of speech sounds that accommodate General American, Received Pronunciation (RP) and to a large extent also Australian, Canadian, Irish (including Ulster), New Zealand, Scottish, South African and Welsh pronunciations. Therefore, not all of the distinctions shown here are relevant to a particular dialect:

• i⟩ does not represent a phoneme but a variation between /iː/ and /ɪ/ in unstressed positions. Speakers of dialects with happy tensing (Australian English, General American, modern RP) should read it as an unstressed /iː/, whereas speakers of other dialects (e.g. some Northern England English) should treat it the same as /ɪ/. In Scotland, this vowel can be considered the same as the short allophone of /eɪ/, as in take. Before /ə/ within the same word, another possible pronunciation is /j/ as in yet.
• Many speakers of American and Canadian English pronounce cot /ˈkɒt/ and caught /ˈkɔːt/ the same.[lower-alpha 9] You may simply ignore the difference between the symbols /ɒ/ and /ɔː/, just as you ignore the distinction between the written vowels o and au when pronouncing them.
• Speakers of rhotic dialects (Irish English, North American English, Scottish English) do not distinguish between the vowels of near /ˈnɪər/, cure /ˈkjʊər/ and square /ˈskwɛər/ on one hand and freerunning /ˈfriːrʌnɪŋ/, Q-rating /ˈkjuːreɪtɪŋ/ and dayroom /ˈdeɪruːm/ on the other. If you speak such a dialect, read /ɪər, ʊər, ɛər/ as /iːr, uːr, eɪr/.
• In Northern Ireland, Scotland and many North American dialects the distinction between /ʊr/ as in courier and the aforementioned /ʊər/ and /uːr/ does not exist. If you speak such a dialect, ignore the difference between /ʊr/, /ʊər/ and /uːr/.
• In Northern Ireland and Scotland this merger occurs in all environments, which means that foot /ˈfʊt/ and goose /ˈɡuːs/ also have the same vowel.[lower-alpha 10][lower-alpha 11] If you speak such a dialect, ignore the difference between /ʊ/ and /uː/ in all contexts.
• In North America, the /ʊr/ of courier and the /ʊər/ of cure may instead merge with /ɔːr/ as in north or /ɜːr/ as in nurse. No such merger is possible in the case of the sequence which we transcribe as /uːr/ as there is an implied morpheme boundary after the length mark.
• In North American dialects that do not distinguish between /ʊr/, /ʊər/ and /uːr/ there is also no distinction between the /ɪr/ of mirror and the aforementioned /ɪər/ and /iːr/. If you speak such a dialect, ignore the difference between /ɪr/, /ɪər/ and /iːr/.
• In many North American dialects there is also no distinction between the vowels in merry /ˈmɛri/, Mary /ˈmɛəri/ and marry /ˈmæri/. If you speak such a dialect, ignore the difference between /ɛr/, /ɛər/ and /ær/. Some speakers keep marry and/or merry separate from the rest, but in the General American accent all three vowels are the same and may not be distinct from /eɪr/ as in dayroom /ˈdeɪruːm/.
• In rhotic North American English there is no distinction between the vowels in nurse /ˈnɜːrs/ and letter /ˈlɛtər/. If you speak such a dialect, read /ɜːr/ as /ər/. The /ʌr/ of hurry often joins this neutralization; if you have it in your speech, read /ɜːr/, /ər/ and /ʌr/ as /ər/.
• Some speakers from Northern England do not distinguish the vowel of square /ˈskwɛər/ and nurse /ˈnɜːrs/.[lower-alpha 12] If you speak such a dialect, ignore the difference between the symbols /ɛər/ and /ɜːr/.
• In New Zealand English, the vowels of kit /ˈkɪt/ and focus /ˈfoʊkəs/ have the same schwa-like quality.[lower-alpha 13][lower-alpha 14] If you are from New Zealand, ignore the difference between the symbols /ɪ/ and /ə/.
• In contemporary New Zealand English and some other dialects, the vowels of near /ˈnɪər/ and square /ˈskwɛər/ are not distinguished.[lower-alpha 15] If you speak such a dialect, ignore the difference between the symbols /ɪər/ and /ɛər/.
• In Northern England English, the vowels of foot /ˈfʊt/ and strut /ˈstrʌt/ are not distinguished.[lower-alpha 16] If you are from Northern England, ignore the difference between the symbols /ʊ/ and /ʌ/.
• In Welsh English and some other dialects, the vowels of unorthodoxy /ʌnˈɔːrθədɒksi/ and an orthodoxy /ən ˈɔːrθədɒksi/ are not distinguished.[lower-alpha 17] If you speak such a dialect, ignore the difference between the symbols /ʌ/ and /ə/.
• Depending on the dialect, vowels can be subject to various mergers before /l/, so that e.g. fill /ˈfɪl/ and feel /ˈfiːl/ or pull /ˈpʊl/ and pool /ˈpuːl/ may not be distinguished. L-vocalization may trigger even more mergers, so that e.g. cord /ˈkɔːrd/ and called /ˈkɔːld/ may be homophonous as /ˈkɔːd/ in non-rhotic dialects of South East England. See English-language vowel changes before historic /l/ for more information.
• In many dialects, /r/ occurs only before a vowel; if you speak such a dialect, simply ignore /r/ in the pronunciation guides where you would not pronounce it, as in cart /kɑːrt/.
• In other dialects, /j/ (yes) cannot occur after /t, d, n/, etc., within the same syllable; if you speak such a dialect, then ignore the /j/ in transcriptions such as new /njuː/. For example, New York is transcribed /njuː ˈjɔːrk/. For most people from England and for some New Yorkers, the /r/ in /jɔːrk/ is not pronounced; for most people from the United States, including some New Yorkers, the /j/ in /njuː/ is not pronounced and may be ignored. (See yod-dropping.)

On the other hand, there are some distinctions which you might make but which this key does not encode, as they are seldom reflected in the dictionaries used as sources for Wikipedia articles:

• The vowels of kit and bit, distinguished in South Africa.[lower-alpha 18] Both of them are transcribed as /ɪ/ in stressed syllables and as /ɪ/ or /ə/ in unstressed syllables.
• The difference between the vowels of fir, fur and fern, maintained in some Scottish and Irish English but lost elsewhere.[lower-alpha 19] All of them are transcribed as /ɜːr/.
• The vowels of north and force, distinguished in Scottish English, Irish English and by a minority of American speakers.[lower-alpha 19] Both of them are transcribed as /ɔːr/.
• The vowels of pause and paws, distinguished in Cockney and by some Estuary English speakers.[lower-alpha 20] Both of them are transcribed as /ɔː/ when the spelling does not contain ⟨r⟩ and /ɔːr/ or /ʊər/ (depending on the word) when it does.
• The vowels of manning and Manning, distinguished in some parts of the United States (see /æ/ raising). Both of them are transcribed as /æ/.
• The difference between the vowels of pain and pane found in some English, Welsh, and Newfoundland dialects. Both of them are transcribed as /eɪ/.
• The difference between the vowels of holy and wholly found in Cockney and many Estuary English speakers.[lower-alpha 21] Both of them are transcribed as /oʊ/.
• Any allophonic distinctions, such as:
• The vowels of bad and lad, distinguished in many parts of Australia. Both of them are transcribed as /æ/.
• The vowels of spider and spied her, distinguished in many parts of Scotland,[lower-alpha 22] plus many parts of North America. Both of them are transcribed as /aɪ/.
• The vowels of rider and writer, distinguished in most parts of Canada and many parts of the United States. Both of them are transcribed as /aɪ/.
• The vowels of powder and pouter distinguished in most parts of Canada and some parts of the United States. Both of them are transcribed as /aʊ/.
• Allophonic vowel length (including the Scottish vowel length rule), as in knife /ˈnaɪf/ vs. knives /ˈnaɪvz/. Phonemic vowel length, which exists in some dialects and involves pairs such as /ɛ/ vs. /ɛər/ and /ə/ vs. /ɜːr/ is also not marked explicitly. /i/ and /u/ do not represent phonemes; see above.
• Flapping in words such as better, which we write /ˈbɛtər/, rather than /ˈbɛdər/.
• Glottalization in words such as jetlag and, in some accents, daughter, which we write /ˈdʒɛtlæɡ/ and /ˈdɔːtər/, rather than /ˈdʒɛʔlæɡ/ and /ˈdɔːʔər/. In this system, /ʔ/ is used only for paralanguage or in loanwords where it occurs phonemically in the original language.
• L-vocalization in words such as bottle and Alps, which we write /ˈbɒtəl/ and /ˈælps/, rather than /ˈbɒtʊ/ and /ˈæwps/.
• The difference between allophones of /ə/ in balance ([ə]) vs. the ones in about and Russia (and, in non-rhotic dialects, better), both of which may be closer to /ʌ/ in dialects with the foot-strut split (that is, [ɐ]) vs. the one in button (the syllabicity of the following consonant). All are transcribed as /ə/ in our system.
• The difference between the phonetic realization of English sounds (mostly vowels) in various dialects. Let's pick some grapes for Betty should be transcribed /lɛts ˈpɪk səm ˈɡreɪps fər ˈbɛti/ regardless of the variety of English and everyone should interpret that transcription according to their own dialect. Thus, a person from South East England will read it as something like [lɛʔs ˈpʰɪk səm ˈɡɹɛɪps fə ˈbɛtˢɪi], a Scot as [ɫɛts ˈpʰɪk səm ˈɡɾeps fɚ ˈbɛte], whereas someone from New Zealand will interpret that transcription as [ɫɪts ˈpʰək səm ˈɡɹæɪps fə ˈbɪɾi]. Because we are transcribing diaphonemes rather than phones (actual sounds), it is irrelevant that, for example, the vowel in let's as pronounced by someone from New Zealand overlaps with how people with England and Scotland typically pronounce the first vowel in pick, or that the Scottish realization of /r/ after /ɡ/ overlaps with the New Zealand realization of /t/ between vowels. In other words, the symbol ⟨ɛ⟩ does not stand specifically for the open-mid front unrounded vowel in our system but any vowel that can be identified as the vowel in let's, depending on the accent. This is also why we use the simple symbol ⟨r⟩ for the second sound in grapes.

Other words may have different vowels depending on the speaker.

The pronunciation of the /æ/ vowel in most dialects of Scotland, Northern Ireland, Northern England and Wales has always been closer to [a]. Received Pronunciation has moved away from the traditional near-open front realization [æ] towards almost fully open front realization [a], and both the Oxford English Dictionary and the 2014 edition of Gimson's Pronunciation of English transcribe the vowel in lad, bad, cat, trap with /a/.[lower-alpha 23]

For more extensive information on dialect variations, you may wish to see the IPA chart for English dialects.

Note that place names are not generally exempted from being transcribed in this abstracted system, so rules such as the above must be applied in order to recover the local pronunciation. Examples include place names in much of England ending ‑ford, which although locally pronounced [‑fəd] are transcribed /‑fərd/. This is best practice for editors. However, readers should be aware that not all editors may have followed this consistently, so for example if /‑fəd/ is encountered for such a place name, it should not be interpreted as a claim that the /r/ would be absent even in a rhotic dialect.

## Other transcriptions

If you feel it is necessary to add a pronunciation respelling using another convention, then please use the conventions of Wikipedia's pronunciation respelling key.

• To compare the following IPA symbols with non-IPA American dictionary conventions that may be more familiar, see Pronunciation respelling for English, which lists the pronunciation guides of fourteen English dictionaries published in the United States.
• To compare the following IPA symbols with other IPA conventions that may be more familiar, see Help:IPA/Conventions for English, which lists the conventions of eight English dictionaries published in Britain, Australia, and the United States.

• If your browser does not display IPA symbols, you probably need to install a font that includes the IPA (for good, free IPA fonts, see the download links in the articles for Gentium and the more complete Charis SIL; for a monospaced font, see the complete Everson Mono)
• To add IPA pronunciations to Wikipedia articles, see the {{IPA}} template
• For a guide to adding IPA characters to Wikipedia articles, see Wikipedia:Manual of Style/Pronunciation § Entering IPA characters.
• Help:IPA/Conventions for English
• Help:Pronunciation respelling key
• Pronunciation respelling for English

## Notes

1. In varieties with flapping, /t/ and sometimes also /d/ between a vowel and a weak or word-initial vowel may be pronounced with a voiced tap [ɾ], making latter sound similar or identical to ladder. Some dictionaries transcribe /t/ subject to this process as ⟨d⟩ or ⟨⟩, but they are not distinguished in this transcription system. In those varieties, the sequence /nt/ in the same environment may also be realized as a nasalized tap [ɾ̃], making winter sound similar or identical to winner. This is also not distinguished in this system.
2. In dialects with yod dropping, /j/ in /juː/, /ju/, or /jʊər/ is not pronounced after coronal consonants (/t/, /d/, /s/, /z/, /n/, /θ/, and /l/) in the same syllable, so that dew /djuː/ is pronounced the same as do /duː/. In dialects with yod coalescence, /tj/ and /dj/ mostly merge with /tʃ/ and /dʒ/, so that the first syllable in Tuesday is pronounced the same as choose. In some dialects /sj/ and /zj/ are also affected and frequently merge with /ʃ/ and /ʒ/. Where /j/ in /juː/, /ju/, or /jʊər/ following a coronal is still pronounced in yod-dropping accents, place a syllable break before it: menu /ˈmɛn.juː/.
3. The phoneme /hw/ is not distinguished from /w/ in the many dialects with the winewhine merger, such as RP and most varieties of General American. For more information on this sound, see voiceless labialized velar approximant.
4. The IPA value of the letter ⟨j⟩ may be counterintuitive to English speakers, but the spelling is found even in some common English words like hallelujah and fjord.
5. /l/ in the syllable coda, as in the words all, cold, or bottle, is pronounced as [o], [u], [w] or a similar sound in many dialects through L-vocalization.
6. In most varieties of English, /r/ is pronounced as an approximant [ɹ]. Although the IPA symbol ⟨r⟩ represents the alveolar trill, ⟨r⟩ is widely used instead of ⟨ɹ⟩ in broad transcriptions of English.
7. A number of English words, such as genre and garage, may be pronounced with either /ʒ/ or /dʒ/.
8. In most dialects, /x/ can also be replaced by /k/ in most words, including loch. It is also replaced with /h/ in some words, particularly of Yiddish origin, such as Chanukah.
9. /ɒ̃, æ̃/ are only found in French loanwords and often replaced by another vowel and a nasal consonant: bon vivant /ˌbɒn viːˈvɒnt/, ensemble /ɒnˈsɒmbəl/, etc.[lower-alpha 1]
10. /ɜː/ is only found in loanwords and represents a situation where such an r-less vowel is used only in British or Southern Hemisphere accents, and therefore a transcription that includes it must always be prefaced with a label indicating the variety of English. It is to be used only when a reliable source shows that General American has a different vowel in the same position. If r-ful NURSE is used even in GA, even if spelled without ⟨r⟩, as in Goethe and hors d'oeuvre, use /ɜːr/. /ɜː/ is also not the same as ⟨œ⟩ seen in some American dictionaries. ⟨œ⟩ in those dictionaries is merely a notational convention and does not correspond to any vowel in any accent of English, so a transcription containing ⟨œ⟩ cannot be converted to one that uses this key.
11. In non-rhotic accents like RP, /r/ is not pronounced unless followed by a vowel.
12. In dialects with the fatherbother merger such as General American, /ɒ/ is not distinguished from /ɑː/.
13. In most of the United States, /ɒr/ is merged with /ɔːr/, except for a handful of words such as borrow, tomorrow and sorry, which instead have /ɑːr/. In some parts of the Southern and Northeastern US, it is always merged with /ɑːr/. In Canada, it is always merged with /ɔːr/.
14. In North America, /æ/ is often pronounced like a diphthong [eə~ɛə] before nasal consonants and, in some particular regional dialects, other environments. See /æ/ raising.
15. /ær/, /ɛr/ and /ɛər/ are not distinguished in many North American accents (Marymarrymerry merger). Some speakers merge only two of the sounds (most typically /ɛər/ with one of the short vowels) and less than a fifth of speakers of American English make a full three-way distinction, like RP and similar accents.[lower-alpha 2]
16. In much of North America, /aɪ/ or /aʊ/ may have a slightly different quality when it precedes a voiceless consonant, as in price or mouth, from that in ride/pie or loud/how, a phenomenon known as Canadian raising. Since this occurs in a predictable fashion, it is not distinguished in this transcription system.
17. In some dialects, especially in the UK, the second segment in a diphthong followed by /ə/ is often omitted. This process or lack thereof may help choose between /aɪər, aʊər, ɔɪər/ in some words (diary, admirer) and /aɪr, aʊr, ɔɪr/ in others (pirate, siren), a distinction not always clear.
18. Some speakers pronounce higher, flower, mayor and coyer ("more coy") with two syllables, and hire, flour, mare and coir with one. Others pronounce them the same.
19. /ɛ/ is transcribed with ⟨e⟩ in many dictionaries. However, /eɪ/ is also sometimes transcribed with ⟨e⟩, especially in North American literature, so ⟨ɛ⟩ is chosen here.
20. /ɛə/, /ɪə/, or /ʊə/ may be separated from /r/ only when a stress follows it. The IPAc-en template supports /ɛəˈr/, /ɪəˈr/, /ʊəˈr/, /ɛəˌr/, /ɪəˌr/, and /ʊəˌr/ as distinct diaphonemes for such occasions.
21. /ɪ/ and /oʊ/ may be strong or weak depending on context.[lower-alpha 3] Whether an instance of unstressed /ɪ/ is strong or weak may not be clear in some circumstances.[lower-alpha 4]
22. Words like idea, real, theatre, and cruel may be pronounced with /ɪə/ or /ʊə/ in non-rhotic accents such as Received Pronunciation, and some dictionaries transcribe them with /ɪə, ʊə/,[lower-alpha 5] but since they do not stem from historical /r/ and are not pronounced with /r/ in rhotic accents, they should be transcribed with /iːə, uːə/, not with /ɪə, ʊə/, in this transcription system.
23. /oʊ/ is often transcribed with ⟨əʊ⟩, particularly in British literature, based on its modern realization in Received Pronunciation. It is also transcribed with ⟨o⟩, particularly in North American literature.
24. /ɔː/ is not distinguished from /ɒ/ in dialects with the cotcaught merger such as Scottish English, Canadian English and many varieties of General American. In North America, the two vowels most often fall together with /ɑː/.
25. Some conservative dialects make a distinction between the vowels in horse and hoarse, but the number of speakers who make this distinction any longer is very small and many dictionaries do not differentiate between them (horsehoarse merger). The vowel in hoarse was formerly represented as /ɔər/ on Wikipedia, but is now represented as /ɔːr/, identical to horse.
26. /ʊər/ is not distinguished from /ɔːr/ in dialects with the cureforce merger, including many younger speakers. In England, the merger may not be fully consistent and may only apply to more common words. In conservative RP and Northern England English /ʊər/ is much more commonly preserved than in modern RP and Southern England English. In Australia and New Zealand, /ʊər/ does not exist as a separate phoneme and is replaced either by the sequence /uːər/ (/uːr/ before vowels within the same word, save for some compounds) or the monophthong /ɔːr/.
27. Some, particularly North American, dictionaries notate /ʌ/ with the same symbol as /ə/, which is found only in unstressed syllables, and distinguish it from /ə/ by marking the syllable as stressed. Also note that although ⟨ʌ⟩, the IPA symbol for the open-mid back vowel, is used, the typical modern pronunciation is rather close to the near-open central vowel [ɐ] in most dialects, including Received Pronunciation and General American.
28. /ʌ/ is not used in the dialects of the northern half of England, some bordering parts of Wales, and some broad eastern Ireland accents. These words would take the /ʊ/ vowel: there is no footstrut split.
29. In Received Pronunciation, /ɜːr/ is pronounced as a lengthened schwa, [əː]. In General American, it is phonetically identical to /ər/. Some dictionaries therefore use ⟨əː, ər⟩ instead of the conventional notations ⟨ɜː, ɜr⟩. When ⟨ər⟩ is used for /ɜːr/, it is distinguished from /ər/ by marking the syllable as stressed. Word-initially, /ər/ never occurs, giving way to /ɜːr/. Where there is a free variation between /ɜːr/ and /ər/ in RP, it is acceptable to transcribe only the more common variant (e.g. /ər/ for persona).
30. /ʌr/ is not distinguished from /ɜːr/ in dialects with the hurryfurry merger such as General American.
31. In a number of contexts, /ə/ in /ər/, /əl/, /ən/, or /əm/ is often omitted, resulting in a syllable with no vowel. Some dictionaries show /ə/ in those contexts in parentheses, superscript, or italics to indicate this possibility, or simply omit /ə/. When followed by a weak vowel, the syllable may be lost altogether, with the consonant moving to the next syllable, so that doubling /ˈdʌb.əl.ɪŋ/ may alternatively be pronounced as [ˈdʌb.lɪŋ], and Edinburgh /ˈɛd.ɪn.bər.ə/ as [ˈɛd.ɪn.brə].[lower-alpha 8] When not followed by a vowel, /ər/ merges with /ə/ in non-rhotic accents.
32. In accents with the weak vowel merger such as most Australian and American accents, /ɪ/ in unstressed positions is not distinguished from /ə/, making rabbit and abbot rhyme and Lenin and Lennon homophonous. Pairs like roses and Rosa's are kept distinct in American accents because of the difference in morphological structure,[lower-alpha 6] but may be homophonous in Australian.[lower-alpha 7] In these accents, unstressed /ɪl, ɪn, ɪm/ merge with /əl, ən, əm/, so that the second vowel in Latin may be lost and cabinet may be disyllabic (see the previous note).
33. i⟩ represents variation between /iː/ and /ɪ/ in unstressed positions. It is realized with a quality closer to /iː/ in accents with happy tensing, such as Australian English, General American, and modern RP, and to /ɪ/ in others. ⟨u⟩ likewise represents variation between /uː/ and /ʊ/, but is restricted to not only unstressed but prevocalic positions.
34. The sequence ⟨⟩ may be pronounced as two syllables, [i.ə] or [ɪ.ə], or as one, [jə] or [ɪə̯]. When pronounced as one syllable in a non-rhotic accent, it may be indistinguishable from, and identified as, the NEAR vowel (/ɪər/).[lower-alpha 5] This transcription system uses ⟨⟩, not ⟨i.ə⟩, ⟨ɪə⟩, etc., to cover all these possibilities.
35. Cite error: Invalid <ref> tag; no text was provided for refs named schwa-w
36. The sequence ⟨⟩ may be pronounced as two syllables, [u.ə] or [ʊ.ə], or as one, [wə] or [ʊə̯]. When pronounced as one syllable in a non-rhotic accent, it may be indistinguishable from, and identified as, the CURE vowel (/ʊər/).[lower-alpha 5] This transcription system uses ⟨⟩, not ⟨u.ə⟩, ⟨ʊə⟩, etc., to cover all these possibilities.
37. The IPA stress mark ⟨ˈ⟩ comes before the syllable that has the stress, in contrast to stress marking in pronunciation keys of some dictionaries published in the United States.
38. Syllable divisions are not usually marked, but the IPA dot ⟨.⟩ may be used when it is wished to make explicit where a division between syllables is (or may be) made.

## References

1. Vaux, Bert; Golder, Scott (2003). "How do you pronounce Mary/merry/marry?". Harvard Dialect Survey. Harvard University Linguistics Department.
2. Flemming & Johnson (2007), pp. 91–2.
3. Wells, John (25 March 2011). "strong and weak". John Wells's phonetic blog.
4. Wells (1982), p. 240.
5. Flemming & Johnson (2007), pp. 94–5.
6. Wells (1982), p. 601.
7. Wells (2008), pp. 173, 799.
8. Wells (1982), pp. 473–6, 493, 499.
9. Stuart-Smith (2004), p. 58.
10. Corrigan (2010), pp. 33–5.
11. Wells (1982), pp. 361, 372.
12. Wells (1982), pp. 605–7.
13. Bauer et al. (2007), pp. 98–9.
14. Bauer et al. (2007), p. 98.
15. Wells (1982), pp. 351–3, 363–4.
16. Wells (1982), pp. 380–1.
17. Wells (1982), pp. 612–3.
18. Stuart-Smith (2004), p. 56.
19. Wells (1982), pp. 304, 310–1.
20. Wells (1982), pp. 304, 312–3.
21. Stuart-Smith (2004), p. 57.
22. Cruttenden (2014), pp. 119–20.