Pronunciation model: U.S. English

View the pronunciation key for U.S. English here, or return to the main pronunciation page.

The principles underpinning the OED‘s U.S. English pronunciations are those of Professor William Kretzschmar, co-editor of The Oxford Dictionary of Pronunciation for Current English (Upton, Kretzschmar & Konopka 2001, hereafter CDP), who was also directly consulted on many OED entries. CDP and its descendant, the Routledge Dictionary of Pronunciation for Current English (Upton & Kretzschmar 2017, hereafter RDP) continue to be key sources for U.S. English, along with American sources such as Merriam-Webster, and the U.S. pronunciations in the pronouncing dictionaries of Wells (2008) and Jones et al. (2011). As with most OED pronunciation models, consideration is also given to the Varieties of English descriptions of relevant phonologies, in this case primarily the description of Standard American English (Kretzschmar 2008), which sets the context of colonial settlement, expansion and urbanization spreading west, and the effect of demographic shifts in the 20th century.

A generally agreed standard for spoken U.S. English does not exist in quite the same way as ‘Received Pronunciation’ does for British English. Kretzschmar views ‘General American’ as an erroneous term, with educationally-determined standards of pronunciation varying considerably between different regions of the USA. However, he considers the trend among ‘younger educated speakers’ towards the suppression of strongly region-marked features in more formal contexts, and this is the focus of the U.S. model in RDP and similarly OED. It is beyond our scope to explore the full range of regional U.S. accents and what is included or excluded in relation to each, but several key features are detailed below. ‘[Standard American English] can best be characterized as what is left over after speakers suppress the regional and social features that have risen to salience and become noticeable’ (Kretzschmar 2008, p.43). Salience in this respect is subjective and variable, resulting in ‘a linguistic continuum for American English in which no region or social group has pride of place’ (though a key exception is noted regarding Southern American English).

Subtle developments to the evolving OED U.S. English model build on Kretzschmar’s CDP/RDP approach, influenced both by the specific lexicographical demands and remit of the OED, and the nature of the natural language evidence that editors can now access, particularly in relation to many of the exceptionally low-frequency or more specialist terms. For example, the CDP/RDP difference between the vowels of NORTH and FORCE (/nɔrθ/, /fɔ(ə)rs/) are being merged to /ɔr/ in OED, given the evident variability. (Some entries may legitimately have a weakly diphthongized vowel, but the contrasts are not deemed sufficiently salient or reliable to warrant inclusion).

As in CDP/RDP, OED demonstrates the rhoticity of U.S. English with ‘r’ following a vowel rather than with the International Phonetic Alphabet’s rhoticity diacritic (e.g. showing rhotic schwa/diphthongal glide as ɚ). Words such as hammer and pair are therefore /ˈhæmər/ and /pɛ(ə)r/ rather than /ˈhæmɚ/ and /pɛɚ/. This ‘plus-r’ sequence approach simplifies the range of symbols users are required to understand as well as making broader equivalences between different instances of r-colouring and enabling r-colouring and syllabic /r/ possibilities to be combined.


Only the ‘unmarked’ pronunciation of Kretzschmar’s 2008 description is included for LOT and PALM words, although the patterning and stability of low back vowels is notoriously variable. Following CDP’s treatment, these historical /ɑ/ sets have only /ɑ/, whereas the historical /ɔ/ sets (CLOTH, THOUGHT) are given both. OED’s general pattern of listing /ɔ/ before /ɑ/ variants is for lexicographical consistency and – like all variant pronunciation listings in OED – the order should not be interpreted as a reflection of frequency or popularity.

Although some U.S. English speakers will distinguish between them, words such as marry, merry and Mary are shown in their homophonous form /ˈmɛri/, the former shown as the CARRY set in the table above.

The STRUT vowel is typically not as open as its British English counterpart, nor does it share its slightly dropped back quality, leading to CDP/RDP’s use of /ə/. As a stressed vowel, this includes a range of qualities including those more similar to British English /ʌ/, but highlights the general quality difference. Kretzschmar (2008 p.46) mentions the variety of postvocalic /r/ ‘ranging from fully constricted [r] to different levels of constriction (so-called “r-colouring”) to compensatory lengthening of the vowel to vocalization of the r to create a diphthong’, but resolves for NURSE words the unmarked quality of a stressed rhotic schwa; /ər/ in OED.

Vowel length (as shown in e.g. British English FLEECE) is considered environmentally conditioned and not marked. Potential diphthongal glide weakening in PRICE and CHOICE words is not indicated, but in OED as in RDP the loss of schwa glide in NEAR and SQUARE words is shown in a slightly different pattern versus the 2008 description (which only discussed endings creating intervocalic /r/), such that NEAR and SQUARE words are shown only with the glide when the vowel falls in the final syllable of the entry (hence /ˈskwɛrˌdæns/ and /ˈskwɛrɪŋ/ but /ˌəʊvərˈskwɛ(ə)r/). An equivalent approach is taken with the CURE set.

Historical [o] in words such as hoarse, still retained by some speakers, is commonly not distinguished from horse [ɔ] and so both are represented by /ɔ/.

ADDED and BEAUTIFUL weak vowels are typically not as high as in British English. ADDED vowels tend towards schwa except where the most recent preceding vowel ended high and front (/i, ɪ, eɪ, aɪ, ɔɪ/), in which case /ɪ/ might be found, or more often a quality on the ɪ-ə continuum represented with /ᵻ/. The occurrence of /ᵿ/ in British English for a vowel on the ʊ-ə continuum is less commonly used in U.S. English. Words of the BESIDE set are given extremes of /ə/ and /i/, but an intermediate quality close to /ɪ/ is a feasible unshown variant.

Two nasalized vowels are recognized, /æ̃/ in words such as gratin and /ɒ̃/ in salon.

/r/ covers a wide range of qualities including trilled, tapped, bunched and alveolar articulations. Intervocalic (between-vowel) /t/ is usually tapped (or ‘flapped’) and often with voicing, explained by Kretzschmar (2008) as resulting in the same forms for latter and ladder (note also that for many U.S. English speakers /d/ can also be tapped). Both are transcribed in CDP as /ˈlædər/. Prevocalic voicing of /t/ is also noted by Kretzschmar in -kt-, -pt-, -ft-, and -rt- sequences, but these are not routinely shown in OED. Post-initial /j/ in CURE is shown unbracketed, but /j/ is bracketed in words such as news where it often does not occur. Consonant clusters may also be simplified to a greater extent than in British English, including in contexts such as fact (/fæk(t)/), and is consistently shown in /nt/ when between vowels with no intervening stress marker (e.g. in hunter /ˈhən(t)ər/ but not in centile (/ˈsɛnˌtaɪl/).

The stronger preservation of secondary stress compared with British English results in most syllables with full vowels having at least a secondary stress mark. There is also a general pattern for words with several weak vowels in British English having at least one of those as a strong vowel (e.g. contributory U.S. /kənˈtrɪbjəˌtɔri/ versus British English /kənˈtrɪbjʊt(ə)ri/, /ˌkɒntrɪˈbjuːt(ə)ri/). British and U.S. Englishes are also known for differences in preferred stress placement, with U.S. English often showing preference for primary stress at the beginning of words.

Key sources