Rhoticity in English

Red areas indicate where rural English accents were rhotic in the 1950s.[11]

Red areas are where English dialects of the late 20th century were rhotic.[12]

The earliest traces of a loss of /r/ in English appear in the early 15th century and occur before coronal consonants, especially /s/, giving modern "ass (buttocks)" (Old English ears, Middle English ers or ars), and "bass (fish)" (OE bærs, ME bars).[2] A second phase of /r/-loss began during the 15th century and was characterized by sporadic and lexically variable deletion, such as monyng "morning" and cadenall "cardinal".[2] These /r/-less spellings appeared throughout the 16th and the 17th centuries but are uncommon and are restricted to private documents, especially ones written by women.[2] No English authorities described loss of /r/ in the standard language prior to the mid-18th century, and many did not fully accept it until the 1790s.[2]

During the mid-17th century, a number of sources described /r/ as being weakened but still present.[13] The English playwright Ben Jonson's English Grammar, published posthumously in 1640, recorded that /r/ was "sounded firme in the beginning of words, and more liquid in the middle, and ends."[7] The next major documentation of the pronunciation of /r/ appeared a century later, in 1740, when the British author of a primer for French students of English said that "in many words r before a consonant is greatly softened, almost mute, and slightly lengthens the preceding vowel."[14]

By the 1770s, postvocalic /r/-less pronunciation was becoming common around London even in formal educated speech. The English actor and linguist John Walker used the spelling ar to indicate the long vowel of aunt in his 1775 rhyming dictionary.[4] In his influential Critical Pronouncing Dictionary and Expositor of the English Language (1791), Walker reported, with a strong tone of disapproval, that "the r in lard, bard,... is pronounced so much in the throat as to be little more than the middle or Italian a, lengthened into baa, baad...."[7] Americans returning to England after the 1783 end of the American Revolutionary War reported surprise at the significant changes in the fashionable pronunciation.[15]

By the early 19th century, the southern British standard had been fully transformed into a non-rhotic variety but continued to be variable as late as the 1870s.[7] The extent of rhoticity across England in the mid-19th century is summarized as widespread in the book New Zealand English: its Origins and Evolution:

[T]he only areas of England... for which we have no evidence of rhoticity in the mid-nineteenth century lie in two separate corridors. The first runs south from the North Riding of Yorkshire through the Vale of York into north and central Lincolnshire, nearly all of Nottinghamshire, and adjacent areas of Derbyshire, Leicestershire, and Staffordshire. The second includes all of Norfolk, western Suffolk and Essex, eastern Cambridgeshire and Hertfordshire, Middlesex, and northern Surrey and Kent.[16]

The loss of postvocalic /r/ in the British prestige standard in the late 18th and the early 19th centuries influenced American port cities with close connections to Britain and caused upper-class pronunciation in many eastern and southern port cities, such as New York City, Boston, Alexandria, Charleston, and Savannah, to become non-rhotic.[8] Like regional dialects in England, the accents of other areas in America remained rhotic in a display of linguistic "lag," which preserved the original pronunciation of /r/.[8]

Non-rhotic pronunciation continued to influence American prestige speech until the 1860s, when the American Civil War shifted America's centers of wealth and political power to areas with fewer cultural connections to the old colonial and British elite.[9] This largely removed the prestige associated with non-rhotic pronunciation in America, and when the advent of radio and television in the 20th century established a national standard of American pronunciation, it became a rhotic variety that has preserved historical /r/.[9] That trend seems to have accelerated after the Second World War.[10]

Though most English varieties in England are non-rhotic today, stemming from a trend toward this in southeastern England accelerating in the very late 18th century onwards, rhotic accents are still found in the West Country (south and west of a line from near Shrewsbury to around Portsmouth), the Corby area, some of Lancashire (north and west of the centre of Manchester), some parts of Yorkshire and Lincolnshire, and in the areas that border Scotland. The prestige form, however, exerts a steady pressure toward non-rhoticity. Thus the urban speech of Bristol or Southampton is more accurately described as variably rhotic, the degree of rhoticity being reduced as one moves up the class and formality scales.[29]

Most Scottish accents are rhotic, but non-rhotic speech has been reported in Edinburgh since the 1970s and Glasgow since the 1980s.[28]

Red dots show major U.S. cities where the 2006 Atlas of North American English found 50% or higher non-rhotic speech in at least one white speaker within their data sample.[10] (Non-rhotic speech may be found in speakers of African-American English throughout the country.)

American English is predominantly rhotic today, but at the end of the 1800s non-rhotic accents were common throughout much of the coastal Eastern and Southern U.S., including along the Gulf Coast. In fact, non-rhotic accents were established in all major U.S. cities along the Atlantic coast except for the Delaware Valley area, with its early Scots-Irish influence, centered around Philadelphia and Baltimore. Since the American Civil War and even more intensely during the early to mid-1900s (presumably correlated with the Second World War),[10] rhotic accents began to gain social prestige nationwide, even in the aforementioned traditionally non-rhotic areas. Thus, non-rhotic accents are increasingly perceived by Americans as sounding foreign or less educated due to an association with working-class or immigrant speakers in Eastern and Southern cities, while rhotic accents are increasingly perceived as sounding more "General American".[30]

Today, non-rhoticity in the American South is found primarily among older speakers, and only in some areas such as central and southern Alabama; Savannah, Georgia; and Norfolk, Virginia,[6] as well as in the Yat accent of New Orleans. The local dialects of eastern New England, especially Boston, Massachusetts, extending into the states of Maine and (less so) New Hampshire, show some non-rhoticity, as well as the traditional Rhode Island dialect; however, this feature has been receding in the recent generations. The New York City dialect is traditionally non-rhotic, though William Labov more precisely classifies its current form as variably rhotic,[31] with many of its sub-varieties now fully rhotic, such as in northeastern New Jersey.

African-American Vernacular English (AAVE) is largely non-rhotic, and in some non-rhotic Southern and AAVE accents, there is no linking r, that is, /r/ at the end of a word is deleted even when the following word starts with a vowel, so that "Mister Adams" is pronounced [mɪstə(ʔ)ˈædəmz].[32] In a few such accents, intervocalic /r/ is deleted before an unstressed syllable even within a word when the following syllable begins with a vowel. In such accents, pronunciations like [kæəˈlaːnə] for Carolina, or [bɛːˈʌp] for "bear up" are heard.[33] This pronunciation also occurs in AAVE[34] and also occurred for many older non-rhotic Southern speakers.[35]

Typically, even non-rhotic modern varieties of American English pronounce the /r/ in /ɜr/ (as in "bird," "work," or "perky") and realize it, as in most rhotic varieties, as [ɚ] (listen) (an r-colored mid central vowel) or [əɹ] (a sequence of a mid central vowel and a postalveolar or retroflex approximant).

Canadian English is entirely rhotic except for small isolated areas in southwestern New Brunswick, parts of Newfoundland, and the Lunenburg English variety spoken in Lunenburg and Shelburne Counties, Nova Scotia, which may be non-rhotic or variably rhotic.[36]

The prestige form of English spoken in Ireland is rhotic and most regional accents are rhotic although some regional accents, particularly in the area around counties Louth and Cavan are notably non-rhotic and many non-prestige accents have touches of non-rhoticity. In Dublin, the traditional local dialect is largely non-rhotic but the more modern varieties, referred to by Hickey as "mainstream Dublin English" and "fashionable Dublin English", are fully rhotic. Hickey used this as an example of how English in Ireland does not follow prestige trends in England.[37]

The English spoken in Asia is predominantly rhotic. Many varieties of Indian English are rhotic owing to the underlying phonotactics of the native Indo-Aryan and Dravidian languages[23] whilst some tend to be non-rhotic. In the case of the Philippines, this may be explained because the English that is spoken there is heavily influenced by the American dialect. In addition, many East Asians (in Mainland China, Japan, Korea, and Taiwan) who have a good command of English generally have rhotic accents because of the influence of American English. This excludes Hong Kong, whose RP English dialect is a result of its almost 150-year history as a British Crown colony (and later, a British dependent territory). The lack of consonant /r/ in Cantonese also contributes to the phenomenon (although rhoticity started to exist due to the handover in 1997 and influence by US and East Asian entertainment industry). However, many older (and younger) speakers among South and East Asians have a non-rhotic accent. Speakers of Semitic (Arabic, Hebrew, etc), Turkic (Turkish, Azeri, etc), Iranian languages (Farsi, Kurdish, etc) in West Asia would also speak English with a rhotic pronunciation due to the inherent phonotactics of their native languages.

Other Asian regions with non-rhotic English are Malaysia, Singapore, and Brunei.[38] A typical Malaysian's English would be almost totally non-rhotic due to the nonexistence of rhotic endings in both languages of influence, whereas a more educated Malaysian's English may be non-rhotic due to Standard Malaysian English being based on RP (Received Pronunciation). The classical English spoken in Brunei is non-rhotic. But one current change that seems to be taking place is that Brunei English is becoming rhotic, partly influenced by American English and partly influenced by the rhoticity of Standard Malay, also influenced by languages of Indians in Brunei (Tamil and Punjabi) (rhoticity is also used by Chinese Bruneians), although English in neighboring Malaysia, Singapore, and Australia remains non-rhotic; rhoticity in Brunei English is equal to Philippine and Indian dialects of English and Scottish and Irish dialects. Non-rhoticity is mostly found in older generations, its phenomenon is almost similar to the status of American English, wherein non-rhoticity reduced greatly.[39][40]

A typical teenager's Southeast Asian English would be rhotic,[41] mainly because of prominent influence by American English.[41] Spoken English in Myanmar is non-rhotic, but there are a number of English speakers with a rhotic or partially rhotic pronunciation. Sri Lankan English may be rhotic.

The English spoken in most of Africa is based on RP and is generally non-rhotic. Pronunciation and variation in African English accents are largely affected by native African language influences, level of education and exposure to Western influences. The English accents spoken in the coastal areas of West Africa are primarily non-rhotic as are the underlying varieties of Niger-Congo languages spoken in that part of West Africa. Rhoticity may be present in English spoken in areas where rhotic Afro-Asiatic or Nilo Saharan languages are spoken across northern West Africa and in the Nilotic regions of East Africa. More modern trends show an increasing American influence on African English pronunciation particularly among younger urban affluent populations, where the American rhotic 'r' may be over-stressed in informal communication to create a pseudo-Americanised accent. By and large official spoken English used in post colonial African countries is non-rhotic. Standard Liberian English is also non-rhotic because liquids are lost at the end of words or before consonants.[42] South African English is mostly non-rhotic, especially Cultivated dialect based on RP, except for some Broad varieties spoken in the Cape Province (typically in -er suffixes, as in writer). It appears that postvocalic /r/ is entering the speech of younger people under the influence of American English, and maybe an influence of Scottish dialect brought by Scottish settlers.[43][44]

Standard Australian English is non-rhotic. A degree of rhoticity has been observed in a particular sublect of Australian Aboriginal English spoken on the coast of South Australia, especially in speakers from the Point Pearce and Raukkan settlements. These speakers realise /r/ as [ɹ] in the preconsonantal postvocalic position – after a vowel but before another a consonant – but only within stems. For example: [boːɹd] "board", [tʃɜɹtʃ] "church", [pɜɹθ] "Perth"; but [flæː] "flour", [dɒktə] "doctor", [jɪəz] "years". It has been speculated that this feature may derive from the fact that many of the first settlers in coastal South Australia – including Cornish tin-miners, Scottish missionaries, and American whalers – spoke rhotic varieties.[45]

Although New Zealand English is predominantly non-rhotic, Southland and parts of Otago in the far south of New Zealand's South Island are rhotic from apparent Scottish influence.[46] Older Southland speakers use /ɹ/ variably after vowels, but today younger speakers use /ɹ/ only with the NURSE vowel and occasionally with the LETTER vowel. Younger Southland speakers pronounce /ɹ/ in third term /ˌθɵːɹd ˈtɵːɹm/ (General NZE pronunciation: /ˌθɵːd ˈtɵːm/) but sometimes in farm cart /fɐːm kɐːt/ (same as in General NZE).[47] However, non-prevocalic /ɹ/ among non-rhotic speakers is sometimes pronounced in a few words, including Ireland /ˈɑɪɹlənd/, merely /ˈmiəɹli/, err /ɵːɹ/, and the name of the letter R /ɐːɹ/ (General NZE pronunciations: /ˈɑɪlənd, ˈmiəli, ɵː, ɐː/).[48] The Māori accent varies from the European-origin New Zealand accent; some Māori speakers are semi-rhotic like most European New Zealand speakers, although it is not clearly identified to any particular region or attributed to any defined language shift. The Māori language itself tends in most cases to use an r with an alveolar tap [ɾ], like Scottish dialect.[49]

This merger is present in non-rhotic accents which have undergone the weak vowel merger. Such accents include Australian, New Zealand, most South African speech, and some non-rhotic English speech (e.g. Norfolk, Sheffield). The third edition of Longman Pronunciation Dictionary lists /əd/ (and /əz/ mentioned below) as possible (though less common than /ɪd/ and /ɪz/) British pronunciations, which means that the merger is an option even in RP.

A large number of homophonous pairs involve the syllabic -es and agentive -ers suffixes, such as merges-mergers and bleaches-bleachers. Because there are so many, they are excluded from the list of homophonous pairs below.

A merger of /ɜː(r)/ and /ʌ/ occurring for some speakers of Jamaican English making bud and bird homophones as /bʌd/.[50] The conversion of /ɜː/ to [ʌ] or [ə] is also found in places scattered around England and Scotland. Some speakers, mostly rural, in the area from London to Norfolk exhibit this conversion, mainly before voiceless fricatives. This gives pronunciation like first [fʌst] and worse [wʌs]. The word cuss appears to derive from the application of this sound change to the word curse. Similarly, lurve is coined from love.

In the terminology of John C. Wells, this consists of the merger of the lexical sets comma and letter. It is found in all or nearly all non-rhotic accents[51] and is present even in some accents that are in other respects rhotic, such as those of some speakers in Jamaica and the Bahamas.[51]

In some accents, syllabification may interact with rhoticity and result in homophones for which non-rhotic accents have centering diphthongs. Possibilities include Korea–career,[52] Shi'a–sheer, and Maia–mire,[53] and skua may be identical with the second syllable of obscure.[54]

In Wells' terminology, this consists of the merger of the lexical sets GOAT and FORCE. It may be found in some southern U.S. non-rhotic speech, some speakers of African-American English, some speakers in Guyana and some Welsh speech.[51]

The merger of the lexical sets FACE, SQUARE and NEAR is possible in Jamaican English and partially also in Northern East Anglian English.

In Jamaica, the merger occurs after deletion of the postvocalic /r/ in a preconsonantal position, so that fade can be homophonous with feared as [feːd], but day [deː] is normally distinct from dear [deːɹ], though vowels in both words can be analyzed as belonging to the same phoneme (followed by /r/ in the latter case, so that the merger of FACE and SQUARE/NEAR does not occur). In Jamaican Patois, the merged vowel is an opening diphthong [iɛ] and that realization can also be heard in Jamaican English, mostly before a sounded /r/ (so that fare and fear can be both [feːɹ] and [fiɛɹ]), but sometimes also in other positions. Alternatively, /eː/ can be laxed to [ɛ] before a sounded /r/, which produces a variable Mary-merry merger: [fɛɹ].[55]

It is possible in northern East Anglian varieties (to [e̞ː]), but only in the case of items descended from ME /aː/, such as daze. Those descended from ME /ai/ (such as days) have a distinctive /æi/ vowel. It appears to be receding, as items descended from ME /aː/ are being transferred to the /æi/ class; in other words, a pane-pain merger is taking place. In the southern dialect area, the pane-pain merger is complete and all three vowels are distinct: FACE is [æi], SQUARE is [ɛː] and NEAR is [ɪə].[56]

A near-merger of FACE and SQUARE is possible in General South African English, but the vowels typically remain distinct as [eɪ] (for FACE) and [eː] (for SQUARE). The difference between the two phonemes is so subtle that they're [ðeː] can be misheard as they [ðeɪ] (see zero copula). In other varieties the difference can be greater, e.g. [ðeː] vs. [ðʌɪ] in Broad SAE and [ðɛə] vs. [ðeɪ] in the Cultivated variety. Even in General SAE, SQUARE can be [ɛə] or [ɛː], strongly distinguished from FACE [eɪ]. NEAR remains distinct in all varieties, typically as [ɪə].[57][58]

In the Cardiff dialect SQUARE can also be similar to cardinal [e] (though long [eː], as in South Africa), but FACE typically has a fully close ending point [ei] and thus the vowels are more distinct than in the General South African accent. An alternative realization of the former is an open-mid monophthong [ɛː]. Formerly, FACE was sometimes realized as a narrow diphthong [eɪ], but this has virtually disappeared by the 1990s. NEAR is phonemically distinct, normally as [iː] before any /r/ (a near-fleece merger) and a disyllabic [iːə] elsewhere.[59]

In Geordie, the merger of FACE and NEAR is recessive and has never been categorical (SQUARE [ɛː] has always been a distinct vowel), as FACE can instead be pronounced as the closing diphthong [eɪ] or, more commonly, the close-mid front monophthong [eː]. The latter is the most common choice for younger speakers who tend to reject the centering diphthongs for FACE, which categorically undoes the merger for those speakers. Even when FACE is realized as an opening-centering diphthong, it may be distinguished from NEAR by the openness of the first element: [ɪə] or [eə] for FACE vs. [iə] for NEAR.[60][61][62]

Some of the words listed below may have different forms in traditional Geordie. For the sake of simplicity, the merged vowel is transcribed with ⟨eː⟩.

In Wells' terminology, the father–farther merger consists of the merger of the lexical sets PALM and START. It is found in the speech of the great majority of non-rhotic speakers, including those of England, Wales, the United States, the Caribbean, Australia, New Zealand and South Africa. It may be absent in some non-rhotic speakers in the Bahamas.[51]

Minimal pairs are rare in accents without the father-bother merger. In non-rhotic British English (especially the varieties without the trap-bath split) and, to a lesser extent, Australian English, /ɑː/ most commonly corresponds to /ɑːr/ in American English, therefore it is most commonly spelled with ⟨ar⟩. In most non-rhotic American English (that includes non-rhotic Rhode Island, New York City,[63] some Southern U.S.,[64] and some African-American accents),[65] the spelling ⟨o⟩ is equally common in non-word-final positions due to the aforementioned father-bother merger. Those accents have the god-guard merger (a merger of LOT and START) in addition to the father–farther merger, yielding a three-way homophony between calmer (when pronounced without /l/), comma and karma, though minimal triplets like this are scarce.

The middle column (which only contains words with a preconsonantal vowel) participates in the merger only in the case of North American accents. In other accents, those words have a distinct vowel /ɒ/.

The foot–goose-thought-north-force merger occurs in cockney in fast speech, so that hood /ˈhʊd/ can sound the same as who'd /ˈhʉːd/ and horde /ˈhoːd/, as [ˈʊd]. The distinction is always recoverable, and the vowels are readily distinguished by length (or length and quality) in more deliberate speech: [ˈʊd] vs. [ˈʊʉd ~ ˈəʉd ~ ˈɨːd] vs. [ˈoːd ~ ˈoʊd] (/h/ is commonly dropped in cockney). Because of the cure-force merger, some of the CURE words also join this neutralization, resulting in the foot-cure merger. The neutralization with THOUGHT/NORTH/FORCE occurs only in morphologically closed syllables, so that cored /ˈkɔəd/ cannot fall together with could /ˈkʊd/, unlike cord and chord (both /ˈkoːd/).[66]

There are not many possible homophones due to the restricted distribution of /ʊ/; see foot-strut split. This neutralization is not fully restricted to (morphologically) closed syllables, as the morpheme- and word-final GOOSE can also have an /ʊ/-like quality. More neutralizations are possible due to l-vocalization; see Merger of non-prevocalic /ʊl/, /ʉːl/, /əl/, /oːl/ with /oː/.[67]

The CURE vowel often occurs in morphologically open syllables, so that assured [əˈʃuəd ~ əˈʃʊːd ~ əˈʃɔəd ~ əˈʃɔːd ~ əˈʃɔd] is always distinct from a should [əˈʃʊd]. While there are very few (if any) minimal pairs with FOOT, CURE does occur in the morpheme-internal position, as in curious /ˈkjuərɪiəs/, [ˈkjuəɹɪiəs ~ ˈkjʊːɹɪiəs], where it can have the quality and length of /ʊ/ due to the merger with /oː/: [ˈkjoːɹɪiəs ~ ˈkjʊɹɪiəs].[68]

The lot-thought-north-force merger occurs in cockney in fast speech (though only in the morpheme-final position in the case of THOUGHT/NORTH/FORCE; in the morpheme-internal position [oː~oʊ] is used instead - see thought split), so that ignored /ɪɡˈnɔəd/ may rhyme with nod /ˈnɒd/ as [ɪgˈnɔd] vs. [ˈnɔd]. The distinction is always recoverable, and the vowels are readily distinguished by length (or length and quality) in more deliberate speech: [ɪgˈnɔːd] or [ɪgˈnɔəd] vs. [ˈnɔd] or [ˈnɒd]. Because of the cure-force merger, some of the CURE words also join this neutralization, resulting in the lot-cure merger. The lot-thought-north merger (with a distinct FORCE vowel /oə/) may be also present in some Eastern New England accents.[69][70]

The lot-thought-north-force merger is also present in Singapore English.

A complete merger of LOT with NORTH can be alternatively called the shot-short merger. The name is inappropriate in the case of cockney, where short [ʃoːʔ ~ ʃoʊʔ] is always distinct from shot [ʃɔʔ ~ ʃɒʔ]. Therefore, the columns labelled as morpheme-internal always have a distinct /oː/ vowel in cockney. Unlike the LOT vowel itself, this neutralization is not restricted to closed syllables; in open syllables, THOUGHT/NORTH/FORCE and CURE can also have an /ɒ/-like quality, merge to /ɔə/ or stay distinct as /ɔə/ vs. /uə/. Morpheme-internal /oː/ and /uə/ and any /ʉː/ can neutralize with /ʊ/ in fast speech.[71]

In Wells' terminology, the pawn–porn merger consists of the merger of the lexical sets THOUGHT and NORTH. It is found in most of the same accents as the father–farther merger described above, but is absent from the Bahamas and Guyana.[51]

Labov et al. suggest that, in New York City English, this merger is present in perception not production. As in, although even locals perceive themselves using the same vowel in both cases, they tend to produce the NORTH/FORCE vowel higher and more retracted than the vowel of THOUGHT.[72]

Most speakers with the pawn-porn merger also have the same vowels in caught and court (a merger of THOUGHT and FORCE), yielding a three-way merger of awe-or-ore/oar (see horse-hoarse merger). These include the accents of Southern England, non-rhotic New York City speakers, Trinidad and the Southern hemisphere.

The lot-cloth split coupled with those mergers produces a few more homophones, such as boss–bourse. Specifically, the phonemic merger of the words often and orphan was a running gag in the Gilbert and Sullivan musical, The Pirates of Penzance.

In cockney and some Estuary English the merged THOUGHT-NORTH-FORCE vowel has split into what can be analyzed as two separate phonemes which Wells writes /oː/ and /ɔə/ - see THOUGHT split.

In Wells' terminology, this consists of the merger of the lexical sets THOUGHT and CURE. It is found in those non-rhotic accents containing the caught–court merger that have also undergone the pour–poor merger. Wells lists it unequivocally only for the accent of Trinidad, but it is an option for non-rhotic speakers in England, Australia and New Zealand. Such speakers have a potential four-way merger taw–tor–tore–tour.[74]

In Wells' terminology, this consists of the merger of the lexical sets GOAT and CURE. It may be present in those speakers who have both the dough–door merger described above, and also the pour–poor merger. These include some southern U.S. non-rhotic speakers, some speakers of African-American English (in both cases towards /oʊ/) and some speakers in Guyana.[51] It can be seen in the term "Fo Sho", an imitation of "for sure".

In Geordie, the merger (towards /ʊə/, phonetically [uə]) is variable and recessive. It is also not categorical, as GOAT can instead be pronounced as the close-mid monophthongs [oː] and [ɵː]. The central [ɵː] is as stereotypically Geordie as the merger itself, though it is still used alongside [oː] by young, middle-class males who, as younger speakers in general, reject the centering diphthongs for /oː/ (females often merge /oː/ with /ɔː/ instead, see thought-goat merger). This categorically undoes the merger for those speakers. Even when GOAT is realized as an opening-centering diphthong, it may be distinguished from CURE by the openness of the first element: [ʊə] or [oə] vs. [uə].[60][61][75]

Some of the words listed below may have different forms in traditional Geordie.

In some cases, particularly in New York City, the NURSE sound gliding from a schwa upwards even led to a phonemic merger of the vowel classes associated with the General American phonemes /ɔɪ/ as in CHOICE with the /ɜr/ of NURSE; thus, words like coil and curl, as well as voice and verse, were homophones. The merged vowel was typically a diphthong [əɪ], with a mid central starting point, rather than the back rounded starting point of /ɔɪ/ of CHOICE in most other accents of English. The merger is responsible for the "Brooklynese" stereotypes of bird sounding like boid and thirty-third sounding like toity-toid. This merger is known for the word soitanly, used often by the Three Stooges comedian Curly Howard as a variant of certainly in comedy shorts of the 1930s and 1940s. The songwriter Sam M. Lewis, a native New Yorker, rhymed returning with joining in the lyrics of the English-language version of Gloomy Sunday. Except for New Orleans English,[79][80][81] this merger did not occur in the South, despite up-gliding NURSE existing in some older Southern accents; instead, a distinction between the two phonemes was maintained due to a down-gliding CHOICE sound: something like [ɔɛ].

In 1966, according to a survey that was done by William Labov in New York City, 100% of the people over 60 used [əɪ] for bird. With each younger age group, however, the percentage got progressively lower: 59% of 50- to 59-year-olds, 33% of 40- to 49-year-olds, 24% of 20- to 39-year-olds, and finally, only 4% of people 8–19 years old used [əɪ]. Nearly all native New Yorkers born since 1950, even those whose speech is otherwise non-rhotic, now pronounce bird as [bɝd].[82] However, Labov reports this vowel to be slightly raised compared to other dialects.[83]