Pronunciation model: British English

Review the pronunciation key for British English here, or return to the main pronunciation page.

The Third Edition of the OED transcribes British English according to a system rooted in the model of ‘Received Pronunciation’ (RP) developed by Professor Clive Upton for Oxford Dictionaries in the 1990s.

The term RP is much debated. Many linguists argue that it is outdated, reflecting the prestigious standard of the socially and economically elite; a socially acceptable (i.e. ‘received’) form based on the speech of privately-educated men and their families (the ‘Public School Pronunciation’ described by Jones 1917). In his analysis and deconstruction of RP, Wells (1982) highlighted the variation within it, distinguishing between ‘mainstream RP’, ‘U-RP’ (‘upper crust’ RP), ‘adoptive RP’ (for speakers who did not acquire RP as children) and even ‘quasi-RP’ (varying from adoptive RP in certain allophonic respects), and ‘near-RP’ (not falling with the definition of RP but with few clearly regional features).

In attempts to steer towards a broader and more contemporary conceptualization free of the more objectionable social connotations of ‘RP’, many alternative names have been proposed, including ‘Modern Received Pronunciation’, ‘Standard Southern British English’, ‘BBC English’ (see e.g. Roach 2004), and ‘Non-Regional Pronunciation’. The last of these is described by Collins & Mees (2013, p.4) as reflecting ‘a more encompassing neutral type of modern British English but one which nevertheless lacks obvious local accent features’, one which allows ‘for the present-day range of variation to be heard from educated middle and younger generation speakers in England who have a pronunciation which cannot be pinned down to a specific area’. Yet another term experiencing some revival is that of ‘General British’, first proposed by Windsor Lewis (1972) but more recently adopted by Cruttenden (2014) and Carley, Mees & Collins (2018). Maidment’s (2016) Speech Internet Dictionary focuses its definition of General British on it being the ‘British accent whose varieties are least associated with any specific areas of Great Britain’ and notes it is the variety which is the ‘most frequent model employed in the teaching of British English as an additional language’. There is a locus of General British speakers in south-east England, but GB can readily be found in smaller numbers across Great Britain, including in Wales and Scotland (Cruttenden 2014), which offers some justification for referring to OED’s variety as British rather than English. It is the lack of specifically regional accent features which perhaps best characterizes the variety of British English represented in OED.

All of these attempts to redefine and re-encapsulate this British variety have merit, but an alternative is to view RP as a more flexible concept which evolves with societal views of standard or regionally neutral forms. Upton’s view of RP followed the lead of A.C. Gimson in looking to a redefined, ‘diluted’ RP which is not so narrow in social and geographical representation (see Upton 2008, which is also a useful summary of much of the OED model). His description of vowel qualities reflects the symbols he adopted for OED, for The Oxford Dictionary of Pronunciation for Current English (Upton, Kretzschmar & Konopka 2001, hereafter CDP), and its second edition The Routledge Dictionary of Pronunciation for Current English (Upton & Kretzschmar 2017, hereafter RDP).

For an OED user to fully appreciate the variety being described by OED’s British English pronunciation model, several points need to be made clear:

Perhaps most importantly, RP for OED is not concerned with prescribing ‘correctness’, nor is it an intentionally ‘prestigious’ or class-linked variety but, as Upton describes in RDP, ‘that accent which will be most widely acceptable, as well as most intelligible’ to native British English speakers, ‘and to which the speech of very many of them will in turn approximate closely’ (p.xiv). Both in the model and in the range of variant pronunciations provided, OED holds to the same principle as Jones (1917, p.viii-ix), who concisely explained: ‘I should like here to state that I have no intention of becoming either a reformer of pronunciation or a judge who decides what pronunciations are “good” and what are “bad”. The proper function of the phonetician is to observe and record accurately, to be, in fact, a kind of living phonograph. It may be as well to add that I am not one of those who believe in the desirability or the feasibility of setting up any one form of pronunciation as a standard for the English-speaking world.’ Upton’s model as described in RDP is similarly broad in this respect: ‘the criterion for inclusion being what is heard to be used by educated, non-regionally-marked speakers rather than what is “allowed” by a preconceived model.’

It is also vital to appreciate that transcriptions are phonological or ‘phonemic’, rather than phonetic per se. Although the adopted symbols for each sound attempt to reflect something of their phonetic (articulatory) nature in correspondence with the International Phonetic Alphabet (IPA), they are inherently phonemes: the minimally contrastive speech sound units in a variety, highlighting where the substitution of one sound for another may make a difference to the meaning of the word. Each phoneme has a range of possible allophones (ways of realizing or articulating it) depending on context and/or speaker preference. Consequently, RP transcriptions should not be viewed as representing one individual’s pronunciation. Even though there will be a single sound file attached to each, the sound file demonstrates just one feasible way of realizing the string of phonemes which constitute the transcription.


Except for the final six keywords in the table above, the British vowels correspond to Upton 2008’s vowel qualities. Several choices are worthy of comment here. DRESS and TRAP reflect a general lowering of RP’s short front vowels, particularly the use of /a/ rather than /æ/ (as historically and even often contemporarily used, and the symbol chosen for OED’s U.S. English model). BATH words have been addressed briefly above, though Upton further details the increasing centralization and shortening of the /ɑː/ variant as well as the influence of otherwise RP speakers who diverge only in use of bath /a/. Upton et al.’s justification for including both within CDP is that the variety is not exclusively southern-British, and explains that its inclusion fits the ‘diluted’ RP form being described. CLOTH entries are losing their more traditional /ɔː/ transcriptions as words such as cross and soft are now typically /ɒ/ for the vast majority of RP speakers. Upton adopts the broader central mid quality of /əː/ for NURSE rather than the /ɜː/ common in many RP transcriptions.

For many RP speakers, SQUARE is not diphthongal as traditionally regarded but a tense monophthong in the region of the DRESS vowel. PRICE begins with a quality between centralized front and centralized back, better reflected as /ʌɪ/ than the traditional and still widely-used /aɪ/. Although retraction of the start quality is noted in MOUTH words too (as Cruttenden 2014, p.149 explains, ‘at a point between the back and front open positions, more fronted than the position for General British /ɑː/’), the extent of retraction from front is not necessarily very significant, and no alternative representation to /aʊ/ has gained widespread attraction. CURE, traditionally and still not infrequently /ʊə/, is joined by monophthongal /ɔː/, leading to homophones such as moor and more for some RP speakers.

Supplementing the strong vowels are a range of weak vowel qualities, found in the rightmost column of the table. Weak vowels in RP have traditionally included /ɪ/, /ə/, and /ʊ/. Word-final /ɪ/ in HAPPY is now more often /i/, though in non-utterance-final flowing speech may be less close and front, a possibility covered by the use of /ɪ/ in such cases (e.g. Betty Crocker). It should not be assumed that the use of /ɪ/ here implies the same as a ‘trad-RP’ HAPPY vowel. CDP’s novel approach to vowel qualities falling on the continuum between /ɪ/ and /ə/ and between /ə/ and /ʊ/ has been adopted into OED, using the symbols /ɪ/ and /ʊ/ respectively.

Not included in the above table are the two accepted nasalized vowels in the British English model, /ã/ (as in gratin) and /ɒ̃/ (as in mignon). Nasalized vowels are exclusively found within loan words that have not become fully naturalized in English, but where anglicization is evident and speakers are not code-switching. The common RP pronunciation retains a nod to its etymology in the form of a nasalized vowel quality (i.e. airflow directed through oral and nasal cavities simultaneously).

Several consonant features warrant discussion here: ‘non-English’ sounds, glottal stops, /r/, /j/, nasal assimilation, wh, and syllabic consonants. RP consonants are consistently described as a set of 24: six plosives (/p, t, k, b, d, ɡ/), nine fricatives (/f, v, θ, ð, s, z, ʃ, ʒ, h/), two affricates (/tʃ, dʒ/), three nasals (/m, n, ŋ/), and four approximants of which one is lateral (/r, w, j/ and /l/). To these OED adds two other consonants which, like nasalized vowels, are routinely accepted in otherwise anglicized loans depending on the root. The voiceless velar fricative /x/ is preserved by many RP speakers in Irish and Scottish Gaelic loans such as loch/lough, and the voiceless alveolar lateral fricative /ɬ/ may similarly appear in Welsh loans such as penillion.

The use of a glottal stop [ʔ] as an allophone of /t/ between vowels in RP is still not widespread (e.g. butter is usually [ˈbʌtə] not [ˈbʌʔə]). However, the glottal stop is often found as a /t/ allophone before nonsyllabic consonants such as in Gatwick, Rottweiler (Upton 2008). There are can also be glottallyreinforced /p, t, k, tʃ/ sounds in which a glottal stop supports the closure in the oral cavity, as in stop, wait, hike, much. Glottal closures may also support vowels in some contexts. OED does not represent glottal stops in transcriptions, even though they may even be present in the soundfiles, because they are not phonemically distinctive. That is, they never make a difference to the meaning of a word in and of themselves ([ˈbʌʔə] and [ˈbʌtə] both always mean butter, just as [ˈneɪtʃə] and [ˈneɪʔtʃə] both mean nature), because glottal stops or glottal reinforcement simply contribute to the range of allophones of one of the English phonemes.

British English in OED is non-rhotic, so /r/ does not pattern as it does in, say, U.S. English. ‘Linking’ /r/, i.e. ‘retained historical post-vocalic word-final /r/ occurring before a vowel in the following word’ (Upton 2008, p.249) or more broadly where there is an orthographic ‘r’ not pronounced when the word is pronounced in isolation, is routinely included in OED. ‘Intrusive’ /r/, where no ‘underlying’ /r/ or orthographic ‘r’ exists and which does not factor in traditional RP, is nonetheless commonplace and is represented by /(r)/.

Coalescence of /tj/, /dj/, /sj/, and /zj/ to /tʃ/, /dʒ/, /ʃ/, and /ʒ/ respectively is represented by the inclusion of both forms, e.g. traduce as both /trəˈdjuːs/ and /trəˈdʒuːs/. Historical /lj/ and /sj/ syllable onsets, such as in lucid and suit, are increasingly without /j/ in RP, and for this reason are either bracketed or presented as separate variants (in part with view to the sound files, which by default record all bracketed segments as articulated; separate entries permit the more common pronunciation to have its own sound file).

Segments that are frequently elided in natural speech are bracketed (e.g. softness /ˈsɒf(t)nəs/). Representation of assimilation (the tendency for certain consonants to assume features of neighbouring ones in particular contexts) is limited, as these are often predictable by rule (see e.g. Roach 2009) and to do so would introduce considerable numbers of additional and potentially unnecessary variants. Consequently, despite light bulb being frequently [ˈlʌɪp bʌlb] and handbag being [ˈhambaɡ], they remain /ˈlʌɪt bʌlb/ and /ˈhan(d)baɡ/ respectively. One form of assimilation that is explicitly recognized is the assimilation of /n/ to /ŋ/, in words such as ungallant and include. Factors of stress and the salience of any morpheme boundary are evaluated in determining between /n/ only, /ŋ/ only, or both.

The sequence wh in words such as when and which is historically more of a voiceless labial-velar fricative, often transcribed lexicographically as /hw/. Though this remains a feature of Scottish English and a variable feature of U.S. English, it has dropped rapidly as a feature of RP, and all such OED entries are purely /w/.

Syllabic consonants are extremely common in RP, though generally limited in British English to /n/, /l/ and /m/ (very rarely /ŋ/). OED indicates these with a subscript diacritic syllabic marker, not shown if the only logical articulation would be to syllabify such a consonant, hence little as /ˈlɪtl/ (the final /l/ must be syllabic).

Prosodically, British English is described with one primary stress in each entry, barring some interjections which may have multiple primary stresses where more than one syllable is equally weighted (and monosyllables, where no stress mark is appropriate). Compared with some of the other OED Englishes, the primary stressed syllable often has considerably greater prominence than other syllables and so secondary stresses are less frequently marked (e.g. British English blackbird /ˈblakbəːd/ is marked with a primary only, compared with U.S. /ˈblækˌbərd/). Affixes and combining forms do not have their ‘partial’ pronunciations transcribed, but instead supply a statement of their typical effect on primary stress placement. Adjectives are given their predicative stress pattern, rather than with attributive stress-shift (hence overpaid as in the manager was /ˌəʊvəˈpeɪd/, not with the stress on the first syllable as in the /ˌəʊvəpeɪd ˈmanᵻdʒə/).

Key sources

British English pronunciations throughout the OED’s history

Transcribed pronunciations were a feature of James Murray’s original A New English Dictionary on Historical Principles (NED), known since the 1890s as the first edition of the OED. To Murray, such transcriptions were an essential inclusion, since they sought to represent ‘the actual living form or forms of a word, that is, the word itself, of which the current spelling is only a symbolization’ (Vol. I, p.xxiv). Pronunciations have consequently retained a special status in OED, such that the aim is not to encompass an evolution of pronunciations nor prescribe pronunciations faithful to entries’ etymologies, but to give the current form(s) of each ‘living word’; ‘the latest fact in the form-history of the word’.

Pronunciations in OED are therefore reflective of the principle that language is first and foremost spoken. However, new words may be formed on orthographic bases, leading to challenges such as those Murray describes: ‘On several occasions, the Editor has applied directly to the introducer of a word, to know how he pronounces it, or means it to be pronounced, and has received the answer, that he has never thought of its pronunciation, does not presume to say how it ought to be pronounced, and leaves it to people to pronounce as they like, or to the Dictionary to say what is the right pronunciation. This, of course, reverses the natural order of language, in which speech comes first, and writing is only its symbolization; for here the first thing is the written symbol addressed to the eye […], while, for ‘pronunciation’, anything passes muster which suffices to recall the written symbol in question, just as any reading of a mathematical formula passes muster, if it enables an auditor to write down the formula again.’ (Vol. I, p.xi)

The transcriptions of NED would not be fully recognizable to modern OED users. Murray’s attempt to represent all ‘simple’ sounds by one symbol apiece remains very much part of modern segmental phonetics and phonology, but accepted notation has evolved, largely in light of the International Phonetic Alphabet (IPA) which came along just too late for the beginnings of the dictionary’s publication in 1884 (the first version of the IPA was published in 1888). A key to NED’s pronunciations is reproduced here. The Second Edition of OED converted Murray’s notation to a more IPA-based system (a key to these unrevised transcriptions can be found here), and the model for the Third Edition is broadened and reformulated as discussed above. Minor refinements continue to be introduced as the Dictionary evolves. Since OED does not set out to trace the pronunciation developments of each entry in great detail but primarily to provide current pronunciations, not every alteration is systematically noted. However, where there is a significant change between editions, editorial notes in the etymology of an entry record the earlier pronunciation.