English phonology is the system of sounds used in spokenEnglish. Like many languages, English has wide variation inpronunciation, bothhistorically and fromdialect to dialect. In general, however, the dialects of English around the world have largely similar (but not identical)phonological systems. Among other things, most dialects havevowel reduction inunstressedsyllables and a complex set ofphonological features that distinguishfortis and lenisconsonants (stops,affricates, andfricatives).

Phonological analysis of English often concentrates onprestige orstandard accents, such asReceived Pronunciation for England,General American for the United States, andGeneral Australian for Australia. Nevertheless, many other dialects of spoken English have developed, different from these standard accents, particularly regional dialects. Descriptions of standard reference accents provide only a limited guide to the phonology of other dialects of English.

Aphoneme of a language or dialect is anabstraction of aspeech sound or of a group of different sounds that are all perceived to have the same function by speakers of that particular language or dialect. For example, the English wordthrough consists of three phonemes: the initial "th" sound, the "r" sound, and a vowel sound. The phonemes in that and many other English words do not always correspond directly to the letters used to spell them (English orthography is not as stronglyphonemic as that of many other languages).

The number and distribution of phonemes in English vary from dialect to dialect, and also depend on the interpretation of the individual researcher. The number of consonant phonemes is generally put at 24 (or slightly more depending on the dialect). The number of vowels is subject to greater variation; in the system presented on this page there are 20–25 vowel phonemes inReceived Pronunciation, 14–16 inGeneral American and 19–21 inAustralian English. The pronunciation keys used in dictionaries generally contain a slightly greater number of symbols than this, to take account of certain sounds used in foreign words and certain noticeable distinctions that may not be – strictly speaking – phonemic.

The following table shows the 24 consonant phonemes found in most dialects of English, plus/x/, whose distribution is more limited.Fortis consonants are alwaysvoiceless,aspirated insyllable onset (except in clusters beginning with/s/ or/ʃ/), and sometimes alsoglottalized to an extent insyllable coda (most likely to occur with/t/, seeT-glottalization), while lenis consonants are alwaysunaspirated and un-glottalized, and generally partially or fullyvoiced. The alveolars are usuallyapical, i.e. pronounced with the tip of the tongue touching or approaching the roof of the mouth, though some speakers produce themlaminally, i.e. with the blade of the tongue.^[1]

The following table shows typical examples of the occurrence of the above consonant phonemes in words, usingminimal pairs where possible.

• The pronunciation of/l/ varies by dialect:
  • Received Pronunciation has two main allophones of/l/: the clear, or plain,[l] (the "light L"), and thedark, orvelarized,[ɫ] (the "dark L"). The clear variant is used before vowels when they are in the same syllable, and the dark variant when the/l/ precedes a consonant or is in syllable-final position before silence.
  • In South Wales, Ireland, and the Caribbean,/l/ is usually clear, and in North Wales, Scotland, Australia, and New Zealand it is usually dark.
  • In General American and Canada,/l/ is generally dark, but to varying degrees: before stressed vowels it is neutral or only slightly velarized.^[17] In southern U.S. accents it is noticeably clear between vowels, and in some other positions.^[18]
  • In urban accents of Southern England, as well as New Zealand and some parts of the United States,/l/ can be pronounced as an approximant orsemivowel ([w],[o],[ʊ]) at the end of a syllable (l-vocalization).
• Depending on dialect,/r/ has at least the following allophones in varieties of English around the world (seePronunciation of English /r/):
  • postalveolar approximant[ɹ̠] (the most common realization of the/r/ phoneme, occurring in most dialects, RP and General American included)
  • retroflex approximant[ɻ] (occurs in most Irish dialects and some American dialects)
  • labiodental approximant[ʋ] (occurs in south-east England and some London accents; known asr-labialization)
  • alveolar flap[ɾ] (occurs in most Scottish, Welsh,^[19]Indian^[10] and some South African dialects, some conservative dialects in England and Ireland; not to be confused withflapping of/t/ and/d/)
  • alveolar trill[r] (occurs in some very conservative Scottish dialects and someIndian,South African and Welsh accents)^[10][9][8]
  • voiced uvular fricative[ʁ] (occurs in northern Northumbria, largely disappeared; known as theNorthumbrian burr)
• In most dialects/r/ islabialized[ɹ̠ʷ] in many positions, as inreed[ɹ̠ʷiːd] andtree[t̠ɹ̠̊ʷiː]; in the latter case, the/t/ may be slightly labialized as well.^[20]
• In somerhotic accents, such as General American,/r/ when not followed by a vowel is realized as anr-coloring of the preceding vowel or its coda:nurse[ˈnɚs],butter[ˈbʌɾɚ].
• The distinctions between the nasals areneutralized in some environments. For example, before a final/p/,/t/ or/k/ there is nearly always only one nasal sound that can appear in each case:[m],[n] or[ŋ] respectively (as in the wordslimp,lint,link – note that then oflink is pronounced[ŋ]). This effect can even occur across syllable or word boundaries, particularly in stressed syllables:synchrony is pronounced[ˈsɪŋkɹəni] whereassynchronic may be pronounced with either[sɪŋ-] or[sɪn-]. For other possible syllable-final combinations, see§ Coda in the Phonotactics section below.

In most dialects, the fortis stops and affricate/p,t,tʃ,k/ have various different allophones, and are distinguished from the lenis stops and affricate/b,d,dʒ,ɡ/ by several phonetic features.^[21]

• The allophones of the fortes/p,t,tʃ,k/ include:
  • aspirated[pʰ,tʰ,kʰ] when they occur in the onset of astressed syllable, as inpotato. In clusters involving a following liquid, the aspiration typically manifests as the devoicing of this liquid. These sounds are unaspirated[p,t,k] after/s/ within the same syllable, as instan, span, scan, and at the ends of syllables, as inmat,map,mac.^[22] The voiceless fricatives are nearly always unaspirated, but a notable exception is English-speaking areas of Wales, where they are often aspirated.^[23]
  • In many accents of English, fortis stops/p,t,k,tʃ/ areglottalized in some positions. That may be heard either as a glottal stop preceding the oral closure ("pre-glottalization" or "glottal reinforcement") or as a substitution of the glottal stop[ʔ] for the oral stop (glottal replacement)./tʃ/ can be only pre-glottalized. Pre-glottalization normally occurs in British and American English when the fortis consonant phoneme is followed by another consonant or when the consonant is in final position. Thusfootball andcatching are often pronounced[ˈfʊʔtbɔːl] and[ˈkæʔtʃɪŋ], respectively. Even more frequently, glottal replacement happens in such cases involving/t/, so thatfootball is pronounced[ˈfʊʔbɔːl]. In addition, however, glottal replacement is increasingly common in British English when/t/ occurs between vowels if the preceding vowel is stressed; thusbetter is often pronounced by younger speakers as[ˈbeʔə].^[24] Sucht-glottalization also occurs in many British regional accents, includingCockney, where it can also occur at the end of words, and where/p/ and/k/ are sometimes treated the same way.^[25]
  • For some RP-speakers, final voiceless stops, especially/k/, may become ejectives.^[26]
• Among stops, both fortes and lenes:
  • May haveno audible release[p̚,b̚,t̚,d̚,k̚,ɡ̚] in the word-final position.^[27][28] These allophones are more common in North America than Great Britain.^[27]
  • Almost always have a masked release before another plosive or affricate (as inrubbed[ˈɹʌˑb̚d̥]), i.e. the release of the first stop is made after the closure of the second stop. This also applies when the following stop ishomorganic (articulated in the same place), as intop player.^[29] A notable exception isWelsh English in which stops are usually released in that environment.^[23]
  • The affricates/tʃ,dʒ/ have a mandatory fricative release in all environments.^[30]
• Very often in the United States and Canada and less frequently in Australia^[31] and New Zealand,^[32] both/t/and/d/ can bepronounced as a voiced flap[ɾ] in certain positions: when they come between a preceding stressed vowel (possibly with intervening/r/) and precede an unstressed vowel orsyllabic/l/. Examples includewater,bottle,petal,peddle (the last two words sound alike when flapped). The flap may even appear at word boundaries, as input it on. When the combination/nt/ appears in such positions, some American speakers pronounce it as anasalized flap that may become indistinguishable from/n/, sowinter[ˈwɪɾ̃ɚ] may be pronounced similarly or identically towinner[ˈwɪnɚ].^[33]
• Yod-coalescence is a process thatpalatalizes theclusters/dj/,/tj/,/sj/ and/zj/ into[dʒ],[tʃ],[ʃ] and[ʒ] respectively, frequently occurring with clusters that would be considered to span a syllable boundary.^[34]
  • Yod-coalescence in stressed syllables, such as intune anddune, occurs inAustralian,Cockney,Estuary English,Hiberno-English (some speakers),Newfoundland English,South African English, and to a certain extent inNew Zealand English andScottish English (many speakers). This can lead to additional homophony; for instance,dew anddue come to be pronounced the same asJew.^[35]
  • In certain varieties such asAustralian English, South African English, and New Zealand English,/sj/ and/zj/ in stressed syllables can coalesce into[ʃ] and[ʒ], respectively. In Australian English for example,assume is pronounced[əˈʃʉːm] by some speakers.^[36] Furthermore, some British, Canadian, American, New Zealand and Australian speakers may change the/s/ sound to/ʃ/ before/tr/;^[37] for example, these speakers pronouncestrewn as[ʃtruːn].^[38]
• Thepostalveolar consonants/tʃ,dʒ,ʃ,ʒ/ are stronglylabialized:[tʃʷdʒʷʃʷʒʷ].^[39]
• In addition to/tʃ,dʒ/, the sequences/ts,dz,tr,dr,tθ,dð,pf,bv/ also have affricate-like realizations in certain positions (as incats, roads, tram, dram, eighth, behind them, cupful, obvious; see also§ Onset), but usually only/tʃ,dʒ/ are considered to constitute the monophonemic affricates of English because (among other reasons) only they are found in all of morpheme-initial, -‍internal, and -‍final positions, and native speakers typically perceive them as single units.^[40][41][42]

English, much like other Germanic languages, has a particularly large number of vowel phonemes, and in addition thevowels of English differ considerably between dialects. Consequently, corresponding vowels may be transcribed with various symbols depending on the dialect under consideration. When considering English as a whole,lexical sets are often used, each named by a word containing the vowel or vowels in question. For example, theLOT set consists of words which, likelot, have/ɒ/ in BritishReceived Pronunciation (RP) and/ɑ/ inGeneral American (GA). The "LOT vowel" then refers to the vowel that appears in those words in whichever dialect is being considered, or (at a greater level ofabstraction) to adiaphoneme, which represents this interdialectal correspondence. A commonly used system of lexical sets, devised byJohn C. Wells, is presented below; for each set, the corresponding phonemes are given for RP and General American, using the notation that will be used on this page.

For a table that shows the pronunciations of these vowels in a wider range of English accents, seeSound correspondences between English accents.

The following tables show the vowel phonemes of three standard varieties of English. The notation system used here for Received Pronunciation (RP) is fairly standard; the others less so. The feature descriptions given here (back, open, etc.) are abstracted somewhat; the actual pronunciations of these vowels are somewhat more accurately conveyed by theIPA symbols used (seeVowel for a chart indicating the meanings of these symbols; though note also the points listed below the following tables). The symbols given in the table are traditional but redirect to their modern implementation.

The differences between these tables can be explained as follows:

• RP and General American divide words among thePALM,LOT,CLOTH, andTHOUGHT sets separately from one another; RP has the phonemes/ɑː/ (PALM),/ɒ/ (LOT andCLOTH), and/ɔː/ (THOUGHT); whereas General American has the phonemes/ɑ/ (LOT andPALM) and/ɔ/ (CLOTH andTHOUGHT). In a few North American accents, namely inEastern New England (Boston)LOT words do not have the vowel ofPALM (thefather–bother merger has not occurred) but instead merge withCLOTH/THOUGHT.^{[49][50][51][52]}
• Although the notation/ʌ/ is used for the vowel ofSTRUT in RP and General American, the actual pronunciation in RP may be closer to a near-open central vowel[ɐ], especially among older speakers. In modern RP, this vowel is increasingly realized as[ʌ] to avoid the clash with the lowered variety of/æ/ in the[a] region (thetrap–strut merger). In General American,/ʌ/ is realized as[ʌ̟].^{[53][54][55][56]}
• RP transcriptions use ⟨e⟩ rather than ⟨ɛ⟩ largely for convenience and historical tradition; it does not necessarily represent a different sound from the General American phoneme, as theDRESS vowel is generally realized as[ɛ] in modern RP.^[57]
• The different notations used for the vowel ofGOAT in RP and General American (/əʊ/ and/oʊ/) reflect a difference in the most common phonetic realizations of that vowel.
• The triphthongs given in the RP table are usually regarded as sequences of two phonemes (a diphthong plus/ə/); however, in RP, these sequences frequently undergosmoothing into single diphthongs or even monophthongs.
• The different notations used here for some of the Australian vowels reflect the phonetic realization of those vowels in Australian: a central[ʉː] rather than[uː] inGOOSE, a more closed[e] rather than[ɛ] inDRESS, a close-mid[oː] rather than traditional RP's[ɔː] inTHOUGHT, an open-mid[ɔ] rather than traditional RP's[ɒ] inLOT, an opener[a] rather than somewhat closer[ʌ] inSTRUT, a central[aː] rather a back[ɑː] inCALM andSTART, and somewhat different pronunciations of most of the diphthongs. Note that central[ʉː] inGOOSE, close-mid[oː] inTHOUGHT and open-mid[ɔ] inLOT are standard realizations in modern RP and the difference between modern RP and Australian English in these vowels lies almost only in transcription, rather than pronunciation.^[58][44][59]
• Both Australian/eː/ and RP/ɛː/ are long monophthongs, the difference between them being in tongue height: Australian/eː/ is close-mid[eː], whereas the corresponding RP vowel is open-mid[ɛː].^[60][59]

Other points to be noted are these:

• The vowel/æ/ is generally pronounced more open, approaching[a], by modern RP speakers.^[61][62][57] In American speech, however, there is a tendency for it to become more closed, tenser and even diphthongized (to something like[eə]), particularly in certain environments, such as before anasal consonant,^[63] though younger speakers of some varieties are lowering/æ/ like RP speakers (seeCanadian shift). Some American accents, for example those ofNew York City,Philadelphia andBaltimore, make a marginal phonemic distinction between/æ/ and/eə/, although the two occur largely in mutually exclusive environments. See/æ/ raising.
• A significant number of words (theBATH group) have/æ/ in General American, but/ɑː/ in RP. The pronunciation varies between/æ/ and/aː/ in Australia, with speakers fromSouth Australia using/aː/ more extensively than speakers from other regions.
• In General American and Canadian (which arerhotic accents, where/r/ is pronounced in positions where it does not precede a vowel), many of the vowels can ber-colored by way of realization of a following/r/. This is often transcribed phonetically using a vowel symbol with an added retroflexiondiacritic ⟨˞⟩; thus the symbol[ɚ] has been created for an r-coloredschwa (sometimes called schwar) as inLETTER, and the vowel ofSTART can be modified to make[ɑ˞] so that the wordstart may be transcribed[stɑ˞t]. Alternatively, theSTART sequence might be written[stɑɚt] to indicate an r-colored offglide. The vowel ofNURSE is generally always r-colored in these dialects, and this can be written[ɚ] (or as a syllabic[ɹ̩]).
• In modern RP and other dialects, many words from theCURE group are coming to be pronounced by an increasing number of speakers with theNORTH vowel (sosure is often pronounced likeshore).^[64]
• The vowels ofFLEECE andGOOSE are commonly pronounced as narrow diphthongs, approaching[ɪi] and[ʊu], in RP. Near-RP speakers may have particularly marked diphthongization of the type[əi] and[əu~əʉ], respectively. In General American, the pronunciation varies between a monophthong and a diphthong.^[43]

Listed here are some of the significant cases ofallophony of vowels found within standard English dialects.

• Vowels are shortened when followed in a syllable by a voiceless (fortis) consonant.^[65] This is known aspre-fortis clipping. Thus in the following word pairs the first item has a shortened vowel while the second has a normal length vowel: 'right'/raɪt/ – 'ride'/raɪd/; 'face'/feɪs/ – 'phase'/feɪz/; 'advice'/ədvaɪs/ – 'advise'/ədvaɪz/.
• In many accents of English,tense vowels undergobreaking before/l/, resulting in pronunciations like[pʰiəɫ] forpeel,[pʰuəɫ] forpool,[pʰeəɫ] forpail, and[pʰoəɫ] forpole.^{[citation needed]}
• In RP, the vowel/əʊ/ may be pronounced more back, as[ɒʊ~ɔʊ], before syllable-final/l/, as ingoal. In standard Australian English the vowel/əʉ/ is similarly backed to[ɔʊ] before/l/. A similar phenomenon occurs inSouthern American English.^[66]
• The vowel/ə/ is often pronounced[ɐ] in open syllables.^[67]
• ThePRICE andMOUTH diphthongs may be pronounced with a less open starting point when followed by avoiceless consonant;^[68] this is chiefly a feature of Canadian speech (Canadian raising), but is also found in parts of the United States.^[69] Thuswriter may be distinguished fromrider even whenflapping causes the/t/ and/d/ to be pronounced identically.

Unstressed syllables in English may contain almost any vowel, but in practice vowels in stressed and unstressed syllables tend to use different inventories of phonemes. In particular, long vowels are used less often in unstressed syllables than stressed syllables. Additionally there are certain sounds – characterized bycentral position and weakness – that are particularly often found as the nuclei of unstressed syllables. These include:

• schwa,[ə], as inCOMMA and (in non-rhotic dialects)LETTER (COMMA–LETTER merger); also in many other positions such asabout,photograph,paddock, etc. This sound is essentially restricted to unstressed syllables exclusively. In the approach presented here it is identified as a phoneme/ə/, although other analyses do not have a separate phoneme for schwa and regard it as a reduction or neutralization of other vowels in syllables with the lowest degree of stress.
• r-colored schwa,[ɚ], as inLETTER in General American and some other rhotic dialects, which can be identified with the underlying sequence/ər/.
• syllabic consonants:[l̩] as inbottle,[n̩] as inbutton,[m̩] as inrhythm. These may be phonemized either as a plain consonant or as a schwa followed by a consonant; for examplebutton may be represented as/ˈbʌtn̩/ or/ˈbʌtən/ (see above underConsonants).
• [ɨ̞], as inroses andmaking. This can be identified with the phoneme/ɪ/, although in unstressed syllables it may be pronounced more centrally, and for some speakers (particularly in Australian and New Zealand and some American English) it is merged with/ə/ in these syllables (weak vowel merger). Among speakers who retain the distinction there are many cases wherefree variation between/ɪ/ and/ə/ is found, as in the second syllable oftypical. (TheOED has recently adopted the symbol⟨ᵻ⟩ to indicate such cases.)
• [ʉ̞], as inargument,today, for which similar considerations apply as in the case of[ɨ̞]. (The symbol⟨ᵿ⟩ is sometimes used in these cases, similarly to⟨ᵻ⟩.) Some speakers may also have a rounded schwa,[ɵ̞], used in words likeomission[ɵ̞ˈmɪʃən].^[70]
• [i], as inhappy,coffee, in many dialects (others have[ɪ] in this position).^[71] The phonemic status of this[i] is not easy to establish. Some authors consider it to correspond phonemically with a close front vowel that is neither the vowel ofKIT nor that ofFLEECE; it occurs chiefly in contexts where the contrast between these vowels is neutralized,^[72][73][74] implying that it represents anarchiphoneme, which may be written/i/. Many speakers, however, do have a contrast in pairs of words likestudied andstudded ortaxis andtaxes; the contrast may be[i] vs.[ɪ],[ɪ] vs.[ə] or[i] vs.[ə], and thus some authors consider that thehappY-vowel should be identified phonemically either with the vowel ofKIT or that ofFLEECE, depending on the speaker.^[75] See alsohappy-tensing.
• [u], as ininfluence,to each. This is the back rounded counterpart to[i] described above; its phonemic status is treated in the same works as cited there.

Vowel reduction in unstressed syllables is a significant feature of English. Syllables of the types listed above often correspond to a syllable containing a different vowel ("full vowel") used in other forms of the samemorpheme where that syllable is stressed. For example, the firsto inphotograph, being stressed, is pronounced with theGOAT vowel, but inphotography, where it is unstressed, it is reduced to schwa. Also, certain common words (a,an,of,for, etc.) are pronounced with a schwa when they are unstressed, although they have different vowels when they are in a stressed position (seeWeak and strong forms in English).

Some unstressed syllables, however, retain full (unreduced) vowels, i.e. vowels other than those listed above. Examples are the/æ/ inambition and the/aɪ/ infinite. Some phonologists regard such syllables as not being fully unstressed (they may describe them as havingtertiary stress); some dictionaries have marked such syllables as havingsecondary stress. However linguists such as Ladefoged^[76] andBolinger (1986) regard this as a difference purely of vowel quality and not of stress,^[77] and thus argue that vowel reduction itself is phonemic in English. Examples of words where vowel reduction seems to be distinctive for some speakers^[78] includechickaree vs.chicory (the latter has the reduced vowel ofHAPPY, whereas the former has theFLEECE vowel without reduction), andPharaoh vs.farrow (both have theGOAT vowel, but in the latter word it may reduce to[ɵ]).

Lexical stress is phonemic in English. For example, the nounincrease and the verbincrease are distinguished by the positioning of the stress on the first syllable in the former, and on the second syllable in the latter. (Seeinitial-stress-derived noun.) Stressed syllables in English are louder than non-stressed syllables, as well as being longer and having a higher pitch.

In traditional approaches, in any English word consisting of more than onesyllable, each syllable is ascribed one of three degrees of stress:primary,secondary orunstressed. Ordinarily, in each such word there will be exactly one syllable with primary stress, possibly one syllable having secondary stress, and the remainder are unstressed (unusually-long words may have multiple syllables with secondary stress). For example, the wordamazing has primary stress on the second syllable, while the first and third syllables are unstressed, whereas the wordorganization has primary stress on the fourth syllable, secondary stress on the first, and the second, third, and fifth unstressed. This is often shown in pronunciation keys using theIPA symbols for primary and secondary stress (which are ˈ and ˌ respectively), placed before the syllables to which they apply. The two words just given may therefore be represented (inRP) as/əˈmeɪzɪŋ/ and/ˌɔːɡənaɪˈzeɪʃən/.

Some analysts identify an additional level of stress (tertiary stress). This is generally ascribed to syllables that are pronounced with less force than those with secondary stress, but nonetheless contain a "full" or "unreduced" vowel (vowels that are considered to be reduced are listed underEnglish phonology § Unstressed syllables above). Hence the third syllable oforganization, if pronounced with/aɪ/ as shown above (rather than being reduced to/ɪ/ or/ə/), might be said to have tertiary stress. (The precise identification of secondary and tertiary stress differs between analyses; dictionaries do not generally show tertiary stress, although some have taken the approach of marking all syllables with unreduced vowels as having at least secondary stress.)

In some analyses, then, the concept of lexical stress may become conflated with that of vowel reduction. An approach that attempts to separate both is provided byPeter Ladefoged, who states that it is possible to describe English with only one degree of stress, as long asunstressed syllables are phonemically distinguished forvowel reduction.^[79][80] In this approach, the distinction between primary and secondary stress is regarded as a phonetic or prosodic detail rather than a phonemic feature – primary stress is seen as an example of the predictable "tonic" stress that falls on thefinal stressed syllable of aprosodic unit. For more details of this analysis, seeStress and vowel reduction in English.

For stress as a prosodic feature (emphasis of particular words within utterances), see§ Prosodic stress below.

Phonotactics is the study of the sequences of phonemes that occur in languages and the sound structures that they form. In this study it is usual to represent consonants in general with the letter C and vowels with the letter V, so that a syllable such as 'be' is described as having CV structure. TheIPA symbol used to show a division between syllables is the full stop ⟨.⟩. Syllabification is the process of dividing continuous speech into discrete syllables, a process in which the position of a syllable division is not always easy to decide upon.

Most languages of the world syllabifyCVCV andCVCCV sequences as/CV.CV/ and/CVC.CV/ or/CV.CCV/, with consonants preferentially acting as the onset of a syllable containing the following vowel. According to one view, English is unusual in this regard, in that stressed syllables attract following consonants, so thatˈCVCV andˈCVCCV syllabify as/ˈCVC.V/ and/ˈCVCC.V/, as long as the consonant clusterCC is a possible syllable coda; in addition,/r/ preferentially syllabifies with the preceding vowel even when both syllables are unstressed, so thatCVrV occurs as/CVr.V/. This is the analysis used in theLongman Pronunciation Dictionary.^[81] However, this view is not widely accepted, as explained in the following section.

English allows clusters of up to three consonants in the syllable onset and up to four consonants in the syllable coda,^[82][83] giving a general syllable structure of (C)³V(C)⁴, a potential example beingstrengths/strɛŋkθs/ (although this word has variant pronunciations with only 3 coda consonants, such as/strɛŋθs/). A five-consonant coda may occur in the wordangsts, but this is a highly exceptional case, as the word is both infrequent and not always pronounced with five final segments^[83] (it can be analyzed as a VC⁴ syllable^[82]/æŋsts/ rather than as VC⁵/æŋksts/). From the phonetic point of view, the analysis of syllable structures is a complex task: because of widespread occurrences of articulatory overlap, English speakers rarely produce an audible release of individual consonants in consonant clusters.^[84] This coarticulation can lead to articulatory gestures that seem very much like deletions or complete assimilations. For example,hundred pounds may sound like[hʌndɹɪbpaʊndz] andjumped back (in slow speech,[dʒʌmptbæk]) may sound like[dʒʌmpbæk], but X-ray^[85] andelectropalatographic^[86][87][88] studies demonstrate that inaudible and possibly weakened contacts or lingual gestures may still be made. Thus the second/d/ inhundred pounds does not entirely assimilate to a labial place of articulation, rather the labial gesture co-occurs with the alveolar one; the "missing"[t] injumped back may still be articulated, though not heard.

Division into syllables is a difficult area, and different theories have been proposed. A widely accepted approach is the maximal onset principle:^[89] this states that, subject to certain constraints, any consonants in between vowels should be assigned to the following syllable. Thus the wordleaving should be divided/ˈliː.vɪŋ/ rather than */ˈliːv.ɪŋ/, andhasty is/ˈheɪ.sti/ rather than */ˈheɪs.ti/ or */ˈheɪst.i/. However, when such a division results in an onset cluster that is not allowed in English, the division must respect this. Thus if the wordextra were divided */ˈɛ.kstrə/ the resulting onset of the second syllable would be/kstr/, a cluster that does not occur initially in English. The division/ˈɛk.strə/ is therefore preferred. If assigning a consonant or consonants to the following syllable would result in the preceding syllable ending in an unreduced short vowel, this is avoided. Thus the wordlemma should be divided/ˈlɛm.ə/ and not */ˈlɛ.mə/, even though the latter division gives the maximal onset to the following syllable.

In some cases, no solution is completely satisfactory: for example, in British English (RP) the wordhurry could be divided/ˈhʌ.ri/ or/ˈhʌr.i/, but the former would result in an analysis with a syllable-final/ʌ/ (which is held to be non-occurring) while the latter would result in a syllable-final/r/ (which is said not to occur in this accent). Some phonologists have suggested a compromise analysis where the consonant in the middle belongs to both syllables, and is described asambisyllabic.^[90][91] In this way, it is possible to suggest an analysis ofhurry that comprises the syllables/hʌr/ and/ri/, the medial/r/ being ambisyllabic. Where the division coincides with a word boundary, or the boundary between elements of a compound word, it is not usual in the case of dictionaries to insist on the maximal onset principle in a way that divides words in a counter-intuitive way; thus the wordhardware would be divided/ˈhɑː.dweə/ by the maximal onset principle, but dictionaries prefer the division/ˈhɑːd.weə/.^[92][93][94]

In the approach used by theLongman Pronunciation Dictionary, Wells^[81] claims that consonants syllabify with the preceding rather than following vowel when the preceding vowel is the nucleus of a more salient syllable, with stressed syllables being the most salient, reduced syllables the least, and full unstressed vowels ("secondary stress") intermediate. But there are lexical differences as well, frequently but not exclusively with compound words. For example, indolphin andselfish, Wells argues that the stressed syllable ends in/lf/, but inshellfish, the/f/ belongs with the following syllable:/ˈdɒlf.ɪn,ˈself.ɪʃ/ →[ˈdɒlfɪ̈n,ˈselfɪ̈ʃ], but/ˈʃel.fɪʃ/ →[ˈʃelˑfɪʃ], where the/l/ is a little longer and the/ɪ/ is not reduced. Similarly, intoe-strap Wells argues that the second/t/ is a full plosive, as usual in syllable onset, whereas intoast-rack the second/t/ is in many dialects reduced to the unreleased allophone it takes in syllable codas, or even elided:/ˈtoʊ.stræp/,/ˈtoʊst.ræk/ →[ˈtoˑʊstɹæp,ˈtoʊs(t̚)ɹæk]; likewisenitrate/ˈnaɪtr.eɪt/ →[ˈnaɪtɹ̥eɪt] with a voiceless/r/ (and for some people an affricatedtr as intree), vsnight-rate/ˈnaɪt.reɪt/ →[ˈnaɪt̚ɹeɪt] with a voiced/r/. Cues of syllable boundaries include aspiration of syllable onsets and (in the US) flapping of coda/t,d/(a tease/ə.ˈtiːz/ →[əˈtʰiːz] vs.at ease/ət.ˈiːz/ →[əɾˈiːz]), epenthetic stops like[t] in syllable codas (fence/ˈfens/ →[ˈfents] butinside/ɪn.ˈsaɪd/ →[ɪnˈsaɪd]), and r-colored vowels when the/r/ is in the coda vs. labialization when it is in the onset(key-ring/ˈkiː.rɪŋ/ →[ˈkiːɹʷɪŋ] butfearing/ˈfiːr.ɪŋ/ →[ˈfɪəɹɪŋ]).

The following can occur as theonset:^{[citation needed]}

Notes:

Certain English onsets appear only in contractions: e.g./zbl/ ('sblood), and/zw/ or/dzw/ ('swounds or'dswounds). Some, such as/pʃ/ (pshaw),/fw/ (fwoosh), or/vr/ (vroom), can occur ininterjections. An archaic voiceless fricative plus nasal exists,/fn/ (fnese), as does an archaic/snj/ (snew).

Several additional onsets occur inloan words (with varying degrees of anglicization) such as/bw/ (bwana),/mw/ (moiré),/nw/ (noire),/tsw/ (zwitterion),/zw/ (zwieback),/dv/ (Dvorak),/kv/ (kvetch),/ʃv/ (schvartze),/tv/ (Tver),/tsv/ (Zwickau),/kʃ/ (Kshatriya),/tl/ (Tlaloc),/vl/ (Vladimir),/zl/ (zloty),/tsk/ (Tskhinvali),/hm/ (Hmong),/km/ (Khmer), and/ŋ/ (Nganasan).

Some clusters of this type can be converted to regular English phonotactics by simplifying the cluster: e.g./(d)z/ (dziggetai),/(h)r/ (Hrolf),/kr(w)/ (croissant),/(ŋ)w/ (Nguyen),/(p)f/ (pfennig),/(f)θ/ (phthalic),/(t)s/ (tsunami),/(ǃ)k/ (!kung), and/k(ǁ)/ (Xhosa).

Others can be replaced by native clusters differing only invoice:/zb~sp/ (sbirro), and/zɡr~skr/ (sgraffito).

The following can occur as thenucleus:

• All vowel sounds
• /m/,/n/ and/l/ in certain situations (see below underword-level patterns)
• /r/ inrhotic varieties of English (e.g.General American) in certain situations (see below underword-level patterns)

Most (in theory, all) of the following except those that end with/s/,/z/,/ʃ/,/ʒ/,/tʃ/ or/dʒ/ can be extended with/s/ or/z/ representing themorpheme -s/-z. Similarly, most (in theory, all) of the following except those that end with/t/ or/d/ can be extended with/t/ or/d/ representing the morpheme -t/-d.

Wells (1990) argues that a variety of syllable codas are possible in English, even/ntr,ndr/ in words likeentry/ˈɛntr.i/ andsundry/ˈsʌndr.i/, with/tr,dr/ being treated as affricates along the lines of/tʃ,dʒ/. He argues that the traditional assumption that pre-vocalic consonants form a syllable with the following vowel is due to the influence of languages like French and Latin, where syllable structure is CVC.CVC regardless of stress placement. Disregarding such contentious cases, which do not occur at the ends of words, the following sequences can occur as thecoda^{[citation needed]}:

• Notes:

For some speakers, a fricative before/θ/ is elided so that these never appear phonetically:/fɪfθ/ becomes[fɪθ],/sɪksθ/ becomes[sɪkθ],/twɛlfθ/ becomes[twɛlθ].

• Syllables may consist of a single vowel, meaning that onset and coda are not mandatory.
• The consonant/ŋ/ does not occur in syllable-initial position (most speakers do not maintain it even in loans likeNgorongoro andNguyen).
• The consonant/h/ does not occur in syllable-final position.
• Onset clusters ending in/j/ are followed by/uː/ or its variants (see§ Onset note e above).
• Long vowels and diphthongs are not found before/ŋ/, except for the mimetic wordsboing andoink, unassimilated foreign words such as Burmeseaung and proper names such asTaung, and American-type pronunciations of words likestrong (which have/ɔŋ/ or/ɑŋ/). The short vowels/ɛ,ʊ/ occur before/ŋ/ only in assimilated non-native words such asginseng andSong (name of a Chinese dynasty) or non-finally in some dialects in words likestrength andlength as well as in varieties without thefoot-strut split.
• /ʊ/ is rare in syllable-initial position (although in the northern half of England,[ʊ] is used for/ʌ/ and is common at the start of syllables).
• Stop +/w/ before/uː,ʊ,ʌ,aʊ/ (all presently or historically/u(ː)/) are excluded.^[102]
• Sequences of/s/ + C₁ + V̆ + C₁, where C₁ is a consonant other than/t/ and V̆ is a short vowel, are virtually nonexistent.^[103]

• /ə/ does not occur in stressed syllables, unless it is merged with another vowel as in some varieties.
• /ʒ/ does not occur in word-initial position in native English words, although it can occur syllable-initially as inluxurious/lʌɡˈʒʊəriəs/ in American English, and at the start of borrowed words such asgenre.
• /m/,/n/,/l/ and, inrhotic varieties,/r/ can be the syllable nucleus (i.e. asyllabic consonant) in an unstressed syllable following another consonant, especially/t/,/d/,/s/ or/z/. Such syllables are often analyzed phonemically as having an underlying/ə/ as the nucleus. See above underConsonants.
• The short vowels arechecked vowels, in that they cannot occur without a coda in a word-final stressed syllable. (This does not apply to/ə/, which does not occur in stressed syllables as mentioned above.)

Theprosodic features of English – stress, rhythm, and intonation – can be described as follows.

Prosodic stress is extra stress given to words or syllables when they appear in certain positions in an utterance, or when they receive special emphasis.

According to Ladefoged's analysis (as referred to under§ Lexical stress above), English normally has prosodic stress on the final stressed syllable in anintonation unit. This is said to be the origin of the distinction traditionally made at the lexical level between primary and secondary stress: when a word likeadmiration (traditionally transcribed as something like/ˌædmɪˈreɪʃən/) is spoken in isolation, or at the end of a sentence, the syllablera (the final stressed syllable) is pronounced with greater force than the syllablead, although when the word is not pronounced with this final intonation there may be no difference between the levels of stress of these two syllables.

Prosodic stress can shift for variouspragmatic functions, such as focus or contrast. For instance, in the dialogueIs it brunch tomorrow? No, it'sdinner tomorrow, the extra stress shifts from the last stressed syllable of the sentence,tomorrow, to the last stressed syllable of the emphasized word,dinner.

Grammaticalfunction words are usually prosodically unstressed, although they can acquire stress when emphasized (as inDid you find the cat? Well, I founda cat). Many English function words have distinct strong and weak pronunciations; for example, the worda in the last example is pronounced/eɪ/, while the more common unstresseda is pronounced/ə/. SeeWeak and strong forms in English.

English is claimed to be astress-timed language. That is, stressed syllables tend to appear with a more or less regular rhythm, while non-stressed syllables are shortened to accommodate this. For example, in the sentenceOne make of car is better than another, the syllablesone,make,car,bett- and-noth- will be stressed and relatively long, while the other syllables will be considerably shorter. The theory of stress-timing predicts that each of the three unstressed syllables in betweenbett- and-noth- will be shorter than the syllableof betweenmake andcar, because three syllables must fit into the same amount of time as that available forof. However, it should not be assumed that all varieties of English are stress-timed in this way. The English spoken in the West Indies,^[104] in Africa^[105] and in India^[106] are probably better characterized assyllable-timed, though the lack of an agreed scientific test for categorizing an accent or language as stress-timed or syllable-timed may lead one to doubt the value of such a characterization.^[107]

Phonological contrasts in intonation can be said to be found in three different and independent domains. In the work of Halliday^[108] the following names are proposed:

• Tonality for the distribution of continuous speech into tone groups.
• Tonicity for the placing of the principal accent on a particular syllable of a word, making it thetonic syllable. This is the domain also referred to asprosodic stress orsentence stress.
• Tone for the choice of pitch movement on the tonic syllable. (The use of the termtone in this sense should not be confused with thetone of tone languages, such as Chinese.)

These terms ("the Three Ts") have been used in more recent work,^[109][110] though they have been criticized for being difficult to remember.^[111] American systems such asToBI also identify contrasts involving boundaries between intonation phrases (Halliday'stonality), placement of pitch accent (tonicity), and choice of tone or tones associated with the pitch accent (tone).

Example of phonological contrast involving placement of intonation unit boundaries (boundary marked by comma):

Example of phonological contrast involving placement of tonic syllable (marked by capital letters):

Example of phonological contrast (British English) involving choice of tone (\ = falling tone, \/ = fall-rise tone)

There is typically a contrast involving tone betweenwh-questions andyes/no questions, the former having a falling tone (e.g. "Where did you \PUT it?") and the latter a rising tone (e.g. "Are you going /OUT?"), though studies of spontaneous speech have shown frequent exceptions to this rule.^[112]Tag questions asking for information are said to carry rising tones (e.g. "They are coming on Tuesday, /AREN'T they?") while those asking for confirmation have falling tone (e.g. "Your name's John, \ISN'T it.").

The pronunciation system of English has undergone many changes throughout the history of the language, from thephonological system of Old English, tothat of Middle English, through to that of the present day. Variation betweendialects has always been significant. Former pronunciations of many words are reflected in their spellings, asEnglish orthography has generally not kept pace with phonological changes since the Middle English period.

The English consonant system has been relatively stable over time, althougha number of significant changes have occurred. Examples include the loss (in most dialects) of the[ç] and[x] sounds still reflected by the⟨gh⟩ in words likenight andtaught, and the splitting of voiced and voiceless allophones of fricatives into separate phonemes (such as the two differentphonemes represented by⟨th⟩). There have also been manychanges in consonant clusters, mostly reductions, for instance those that produced the usual modern pronunciations of such letter combinations as⟨wr-⟩,⟨kn-⟩ and⟨wh-⟩.

Thedevelopment of vowels has been much more complex. One of the most notable series of changes is that known as theGreat Vowel Shift, which began around the late 14th century. Here the[iː] and[uː] in words likeprice andmouth became diphthongized, and other long vowels became higher:[eː] became[iː] (as inmeet),[aː] became[eː] and later[eɪ] (as inname),[oː] became[uː] (as ingoose), and[ɔː] became[oː] and later[oʊ] (in RP now[əʊ]; as inbone). These shifts are responsible for the modern pronunciations of many written vowel combinations, including those involving asilent final⟨e⟩.

Many other changes in vowels have taken place over the centuries (see the separate articles on thelow back,high back andhigh front vowels,short A, anddiphthongs). These various changes mean that many words that formerly rhymed (and may be expected to rhyme based on their spelling) no longer do.^[113] For example, inShakespeare's time, following the Great Vowel Shift,food,good andblood all had the vowel[uː], but in modern pronunciationgood has been shortened to[ʊ], whileblood has been shortened and lowered to[ʌ] in most accents. In other cases, words that were formerly distinct have come to be pronounced the same – examples of such mergers includemeet–meat,pane–pain andtoe–tow.

The phonemic status of the velar nasal consonant[ŋ] is disputed; one analysis claims that the only nasal phonemes in English are/m/ and/n/, while[ŋ] is an allophone of/n/ found before velar consonants. Evidence in support of this analysis is found in accents of the north-west Midlands of England where[ŋ] is found only before/k/ or/ɡ/, withsung being pronounced as[sʌŋɡ]. However, in most other accents of Englishsung is pronounced[sʌŋ], producing a three-way phonemic contrastsum – sun – sung/sʌmsʌnsʌŋ/ and supporting the analysis of the phonemic status of/ŋ/. In support of treating the velar nasal as an allophone of/n/,Sapir (1925) claims on psychological grounds that[ŋ] did not form part of a series of three nasal consonants: "no naïve English-speaking person can be made to feel in his bones that it belongs to a single series withm andn. ... It stillfeels likeƞg."^[114] More recent writers have indicated that analyses of[ŋ] as an allophone of/n/ may still have merit, even though[ŋ] may appear both with and without a following velar consonant; in such analyses, an underlying/ɡ/ that isdeleted by aphonological rule would account for occurrences of[ŋ] not followed by a velar consonant.^{[115][116][117]} Thus the phonemic representation ofsing would be/sɪnɡ/ and that ofsinger is/ˈsɪnɡə/; in order to reach the phonetic form[sɪŋ] and[ˈsɪŋə], it is necessary to apply a rule that changes/n/ to[ŋ] before/k/ or/ɡ/, then a second rule that deletes/ɡ/ when it follows[ŋ].

• 1./n/ →[ŋ] / ____ velar consonant
• 2./ɡ/ → ∅ /[ŋ] _____

These produce the following results:

However, these rules do not predict the following phonetic forms:

In the above cases, the/ɡ/ is not deleted. The words are all singlemorphemes, unlikesinger andsinging which are composed of two morphemes,sing plus-er or-ing. Rule 2 can be amended to include a symbol # for a morpheme boundary (including word boundary):

2./ɡ/→∅/[ŋ]___#

This rule then applies tosing,singer andsinging but not toanger,finger, orhunger.

According to this rule, the wordshangar ('shed for aircraft'), which contains no internal morpheme boundary, andhanger ('object for hanging clothes'), which comprises two morphemes, are expected to constitute a minimal pair ashangar[ˈhæŋɡə] versushanger[ˈhæŋə]; in actuality, their pronunciations are not consistently distinguished in this manner, ashangar is frequently pronounced[ˈhæŋə].

Additionally, there are exceptions in the form of comparative and superlative forms of adjectives, where Rule 2 must be prevented from applying. The ending-ish is another possible exception.

As a result, there is, in theory, a minimal pair consisting oflonger ([ˈlɒŋɡər] 'more long') andlonger ([ˈlɒŋər] 'person who longs'), though it is doubtful that native speakers make this distinction regularly.^[118] Names of persons and places, and loanwords, are less predictable.Singapore may be pronounced with or without[ɡ];bungalow usually has[ɡ]; andInge may or may not have[ɡ].^[119]

It is often stated that English has a particularly large number of vowel phonemes and that there are 20 vowel phonemes in Received Pronunciation,^[120] 14–16 in General American, and 20–21 in Australian English. These numbers, however, reflect just one of many possible phonological analyses. A number of "biphonemic" analyses have proposed that English has a basic set of short (sometimes called "simple" or "checked") vowels, each of which can be shown to be a phoneme and can be combined with another phoneme to form long vowels and diphthongs. One of these biphonemic analyses asserts that diphthongs and long vowels may be interpreted as comprising a short vowel linked to a consonant. The fullest exposition of this approach is found inTrager & Smith (1951), where all long vowels and diphthongs ("complex nuclei") are made up of a short vowel combined with either/j/ (for which the authors use the symbol⟨y⟩),/w/ or/h/ (plus/r/ forrhotic accents), each thus comprising two phonemes.^[121] Using this system, the wordbite would be transcribed/bajt/,bout as/bawt/,bar as/bar/ andbra as/brah/. One attraction that the authors claim for this analysis is that it regularizes the distribution of the consonants/j/,/w/, and/h/ (as well as/r/ in non-rhotic accents), which would otherwise not be found in syllable-final position.Trager & Smith (1951) suggest nine simple vowel phonemes to allow them to represent all the accents of American and British English they surveyed, symbolized/i,e,æ/ (front vowels);/ᵻ,ə,a/ (central vowels); and/u,o,ɔ/ (back vowels).

The analysis fromTrager & Smith (1951) came out of a desire to build an "overall system" to accommodate all English dialects, with dialectal distinctions arising from differences in the ordering of phonological rules,^[122][123] as well as in the presence or absence of such rules.^[124] Another category of biphonemic analyses of English treats long vowels and diphthongs as conjunctions of two vowels. Such analyses, as found inSweet (1877) orKreidler (2004) for example, are less concerned with dialectal variation. InMacCarthy (1967:4, 36–37), for example, there are seven basic vowels and these may be doubled (geminated) to represent long vowels, as shown in the table below:

Some of the short vowels may also be combined with/i/ (/ei/bay,/ai/buy,/oi/boy), with/u/ (/au/bough,/ou/beau) or with/ə/ (/iə/peer,/eə/pair,/uə/poor). The vowel inventory of English RP in MacCarthy's system therefore totals only seven phonemes. (Analyses such as these could also posit six vowel phonemes, if the vowel of the final syllable incomma is considered to be an unstressed allophone of that ofstrut.) These seven vowels might be symbolized/i/,/e/,/a/,/o/,/u/,/ʌ/ and/ə/. Six or seven vowels is a figure that would put English much closer to the average number of vowel phonemes in other languages.^[125]

A radically different approach to the English vowel system was proposed byChomsky andHalle. TheirSound Pattern of English (Chomsky & Halle 1968) proposed that English has lax and tense vowel phonemes, which are operated on by a complex set of phonological rules to transform underlying phonological forms into surface phonetic representations. This generative analysis is not easily comparable with conventional analyses, but the total number of vowel phonemes proposed falls well short of the figure of 20 often claimed as the number of English vowel phonemes.

Wikimedia Commons has media related toEnglish phonology.

• University of Glasgow Dynamic Dialects Accent Map
• IPA Phonetics Chart of English

• English phonology

• Pages with plain IPA
• Articles with short description
• Short description matches Wikidata
• All articles with unsourced statements
• Articles with unsourced statements from April 2018
• Articles with unsourced statements from June 2025
• CS1 maint: DOI inactive as of January 2026
• Commons category link from Wikidata