Phonology

Compared with English as spoken in the United Kingdom, North American English is more homogeneous and any phonologically unremarkable North American accent is known as "General American". This section mostly refers to such General American features.

Conservative phonologyedit

Studies on historical usage of English in both the United States and the United Kingdom suggest that spoken American English did not simply deviate away from period British English, but is conservative in some ways, preserving certain features contemporary British English has since lost.

Full rhoticity (or R-fulness) is typical of American accents, pronouncing the phoneme /r/ (corresponding to the letter ⟨r⟩) in all environments, including after vowels, such as in pearl, car, and court. Non-rhotic American accents, those that do not pronounce ⟨r⟩ except before a vowel, such as some Eastern New England, New York, a specific few (often older) Southern, and African American vernacular accents, are often quickly noticed by General American listeners and perceived to sound especially ethnic, regional, or "old-fashioned".

Rhoticity is common in most American accents (yet nowadays rare in England), because, during the 17th-century British colonization, nearly all dialects of English were rhotic, and most North American English simply remained that way. This preservation of rhoticity in North America was also supported by continuing waves of rhotic-accented Scotch-Irish immigrants, most intensely during the 18th century (and moderately during the following two centuries), when the Scotch-Irish eventually made up one-seventh of the colonial population. Scotch-Irish settlers spread from Delaware and Pennsylvania throughout the larger Mid-Atlantic region, the inland regions of both the South and North and throughout the West, all American dialect areas that consistently resisted upper-class non-rhotic influences and that consequently remain rhotic today. The pronunciation of ⟨r⟩ is a postalveolar approximant ɹ̠ (listen) or retroflex approximant ɻ (listen), though a unique "bunched tongue" variant of the approximant r sound is also associated with the United States, and perhaps mostly in the Midwest and the South.

For those American accents that have not undergone the cot–caught merger (the lexical sets LOT and THOUGHT), they have instead retained a LOT–CLOTH split: a 17th-century split in which certain words (labeled as the CLOTH lexical set) separated away from the LOT set. This split, which has now reversed in most British English, simultaneously shifts this relatively recent CLOTH set into a merger with the THOUGHT (caught) set. Having taken place prior to the unrounding of the cot vowel, this results in lengthening and perhaps raising, merging the more recently separated vowel into the THOUGHT vowel in the following environments: before many instances of /f/, /θ/, and particularly /s/ (as in Austria, cloth, cost, loss, off, often, etc.), a few instances before /ŋ/ (as in strong, long, wrong), and variably by region or speaker in gone, on, and certain other words.

The standard accent of southern England, Received Pronunciation (RP), has evolved in other ways too, compared to which General American English has remained relatively more conservative, for example, regarding today's RP features of a trap–bath split and the fronting of /oʊ/, neither of which is typical of General American accents. Moreover, American dialects do not participate in H-dropping, an innovative feature that now characterizes perhaps a majority of the regional dialects of England.

Innovative phonologyedit

On the other hand, General American is more innovative than the dialects of England, or English elsewhere in the world, in a number of its own ways:

• Unrounded LOT: The American phenomenon of the LOT vowel (often spelled ⟨o⟩ in words like box, don, clock, notch, pot, etc.) being produced without rounded lips, like the PALM vowel, allows father and bother to rhyme, the two vowels now unified as the single phoneme /ɑ/. This father–bother vowel merger is in a transitional or completed stage nearly universally in North American English. Exceptions are in northeastern New England English, such as the Boston accent, as well as variably in some New York accents.
• Cot–caught merger in transition: There is no single American way to pronounce the vowels in words like cot /ɑ/ (the ah vowel) versus caught /ɔ/ (the aw vowel), largely due to a merger occurring between the two sounds in some parts of North America, but not others. American speakers with a completed merger pronounce the two historically separate vowels with the exact same sound (especially in the West, northern New England, West Virginia, western Pennsylvania, and the Upper Midwest), but other speakers have no trace of a merger at all (especially in the South, the Great Lakes region, southern New England, and the Mid-Atlantic and New York metropolitan areas) and so pronounce each vowel with distinct sounds (listen). Among speakers who distinguish between the two, the vowel of cot (usually transcribed in American English as /ɑ/), is often a central ɑ̈ (listen) or advanced back ɑ̟, while /ɔ/ is pronounced with more rounded lips and/or phonetically higher in the mouth, close to ɒ (listen) or ɔ (listen), but with only slight rounding. Among speakers who do not distinguish between the two, thus producing a cot–caught merger, /ɑ/ usually remains a back vowel, ɑ, sometimes showing lip rounding as ɒ. Therefore, even mainstream Americans vary greatly with this speech feature, with possibilities ranging from a full merger to no merger at all. A transitional stage of the merger is also common in scatterings throughout the U.S., most consistently in the American Midlands lying between the historical dialect regions of the North and South, while younger Americans in general tend to be transitioning toward the merger. According to a 2003 dialect survey carried out across the United States, about 61% of participants perceive themselves as keeping the two vowels distinct and 39% do not. A 2009 followup survey put the percentages at 58% non-merging speakers and 41% merging.
• STRUT in special words: The STRUT vowel, rather than the one in LOT or THOUGHT (as in Britain), is used in function words and certain other words like was, of, from, what, everybody, nobody, somebody, anybody, and, for many speakers because and rarely even want, when stressed.
• Vowel mergers before intervocalic /r/: The mergers of certain vowels before /r/ are typical throughout North America; the only exceptions exist primarily along the east coast. Such mergers include:
• Mary–marry–merry merger in transition: According to the 2003 dialect survey, nearly 57% of participants from around the country self-identified as merging the sounds /ær/ (as in the first syllable of parish), /ɛr/ (as in the first syllable of perish), and /ɛər/ (as in pear or pair). The merger is already complete everywhere except along some areas of the Atlantic Coast.
• Hurry–furry merger: The pre-/r/ vowels in words like hurry /ʌ/ and furry /ɜ/ are merged in most American accents to ə~ɚ. Only 10% of English speakers across the U.S. acknowledge the distinct hurry vowel before /r/, according to the same dialect survey aforementioned.
• Mirror–nearer merger in transition: The pre-/r/ vowels in words like mirror /ɪ/ and nearer /i/ are merged or at least very close in most American accents. The quality of the historic mirror vowel in the word miracle is quite variable.
• Americans vary slightly in their pronunciations of R-colored vowels—such as those in /ɛər/ and /ɪər/—sometimes monophthongizing towards ɛɹ and ɪɹ or tensing towards eɪɹ and i(ə)ɹ respectively, causing pronunciations like pʰeɪɹ for pair/pear and pʰiəɹ for peer/pier. Also, /jʊər/ is often reduced to jɚ, so that cure, pure, and mature may all end with the sound ɚ, thus rhyming with blur and sir. The word sure is also part of this rhyming set as it is commonly pronounced ʃɚ.
• Yod-dropping: Dropping of /j/ after a consonant is much more extensive than in most of England. In most North American accents, /j/ is "dropped" or "deleted" after all alveolar and interdental consonants (i.e. everywhere except after /p/, /b/, /f/, /h/, /k/, and /m/) so that new, duke, Tuesday, assume are pronounced nu, duk, ˈtʰuzdeɪ, əˈsum (compare with standard British /nju/, /djuk/, /ˈtjuzdeɪ/, /əˈsjum/).

T-glottalization and flapping mountain (glottalized t) ˈmaʊnʔn̩ partner (glottalized t) ˈpʰɑɹʔnɚ leader (d-flapping) ˈliɾɚ cattle (t-flapping) ˈkʰæɾɫ̩ party (t-flapping) ˈpʰɑɹɾi

Optional flapping in certain contexts relatively without flapping ˈɹɛɫətʰɪvɫi relatively with optional flapping ˈɹɛɫəɾɪvɫi

• T-glottalization: /t/ is normally pronounced as unreleased or as a glottal stop ʔ when occurring both (1) after a vowel or /r/ and (2) before a consonant or syllabic n̩, as in button ˈbʌʔn̩ (listen). Following a vowel, /t/ is also glottalized when before a significant pause or when in absolute final position: thus, what wʌʔ or sit sɪʔ. (This innovation of /t/ glottal stopping also may occur in British English, as well as variably between vowels.)
• Flapping: /t/ or /d/ becomes a flap ɾ (listen) when occurring both (1) after a vowel or /r/ and (2) before an unstressed vowel or a syllabic consonant other than n̩, including water ˈwɔɾɚ (listen), party ˈpʰɑɹɾi and model ˈmɑɾɫ̩. This results in pairs such as ladder/latter, metal/medal, and coating/coding being pronounced the same. Flapping of /t/ or /d/ before a full stressed vowel is also possible, but only when that vowel begins a new word or morpheme, as in what is it? wʌɾˈɪzɪʔ and twice in not at all nɑɾəɾˈɔɫ. Other rules apply to flapping too, to such a complex degree in fact that flapping has been analyzed as being required in certain contexts, prohibited in others, and optional in still others. For instance, flapping is prohibited in words like seduce səˈdus, retail ˈɹitʰeɪɫ, and monotone ˈmɑnətʰoʊn, yet optional in impotence ˈɪmpəɾɪns.
• Both intervocalic /nt/ and /n/ may commonly be realized as ɾ̃ (a nasalized alveolar flap) or simply n, making winter and winner homophones in fast or non-careful speech.
• L-velarization: England's typical distinction between a "clear L" (i.e. l (listen)) and a "dark L" (i.e. ɫ (listen)) is much less noticeable in nearly all dialects of American English; it is often altogether absent, with all "L" sounds tending to be "dark," meaning having some degree of velarization, perhaps even as dark as ʟ (listen) (though in initial position, perhaps less dark than elsewhere among some speakers). The only notable exceptions to this velarization are in some Spanish-influenced U.S. English varieties (such as East Coast Latino English, which typically shows a clear "L" in syllable onsets) and in older, moribund Southern speech of the U.S., where "L" is clear in an intervocalic environment between front vowels.
• Weak-vowel merger: The vowel /ɪ/ in unstressed syllables generally merges with /ə/, so effect is pronounced like affect and abbot and rabbit rhyme. The quality of the merged vowels varies considerably, though it is typically closer to ɪ when before a consonant; otherwise it is closer to ə.
• Raising of pre-voiceless /aɪ/: Many speakers split the sound /aɪ/ based on whether it occurs before a voiceless consonant or not, so that in rider it is pronounced äɪ but in writer it is raised to ʌɪ (because t is a voiceless consonant while d is not). Thus, words like bright, hike, price, wipe, etc. with a following voiceless consonant (such as /t, k, θ, s/) use a more raised vowel sound compared to bride, high, prize, wide, etc. Because of this sound change, the words rider and writer (listen), for instance, remain distinct from one another by virtue of their difference in height (and length) of the diphthong's starting point (unrelated to both the letters d and t being pronounced in these words as alveolar flaps ɾ). The sound-change also applies across word boundaries, though the position of a word or phrase's stress may prevent the raising from taking place. For instance, a high school in the sense of "secondary school" is generally pronounced ˈhɐɪskuɫ; however, a high school in the literal sense of "a tall school" would be pronounced ˌhaɪˈskuɫ. This sound change began in the Northern, New England, and Mid-Atlantic regions of the country, and is becoming more common across the nation.
• Many speakers in the Inland North, Upper Midwestern, and Philadelphia dialect areas raise /aɪ/ before voiced consonants in certain words as well, particularly d, g and n. Hence, words like tiny, spider, cider, tiger, dinosaur, beside, idle (but sometimes not idol), and fire may contain a raised nucleus. The use of ʌɪ rather than aɪ in such words is unpredictable from phonetic environment alone, though it may have to do with their acoustic similarity to other words that do contain ʌɪ before a voiceless consonant, per the traditional Canadian-raising system. Hence, some researchers have argued that there has been a phonemic split in these dialects; the distribution of the two sounds is becoming more unpredictable among younger speakers.
• Conditioned /æ/ raising (especially before /n/ and /m/): The raising of the /æ/ or TRAP vowel occurs in specific environments that vary widely from region to region, though nationwide most commonly before /n/ and /m/. With most American speakers, for whom the phoneme /æ/ operates under a somewhat continuous system, /æ/ has both a tense and a lax allophone (with a kind of "continuum" of possible sounds between those two extremes, rather than a definitive split). In these accents, /æ/ is overall realized before nasal stops as more tense (approximately eə̯), while other environments are more lax (approximately the standard æ); for example, note the vowel sound in mæs for mass, but meə̯n for man). In some American accents though, specifically those from Baltimore, Philadelphia, and New York City, æ and eə̯ are indeed entirely separate (or "split") phonemes, for example, in planet pʰlænɪ̈ʔ vs. plan it pʰleənɪ̈ʔ. These are called Mid-Atlantic split-a systems. Note that these vowels move in the opposite direction (high and forward) in the mouth when compared to the backed Standard British "broad a", though the two nation's a systems are probably related phonologically if not phonetically; a British-like phenomenon occurs among some older speakers of the eastern New England (Boston) area for whom /æ/ changes to /a/ before /f/, /s/, /θ/, /ð/, /z/, /v/ alone or when preceded by a homorganic nasal.

• v
• t
• e

/æ/ raising in North American English

Following consonant	Example words	New York City, New Orleans	Baltimore, Philadel- phia	General American, New England, Western US	Midland US, Pittsburgh	Southern US	Canada, Northern Mountain US	Minnesota, Wisconsin	Great Lakes US
Non-prevocalic /m, n/	fan, lamb, stand	ɛəAB		ɛə	ɛə	ɛə~ɛjə	ɛə		ɛə
Prevocalic /m, n/	animal, planet, Spanish	æ
/ŋ/	frank, language			ɛː~eɪ	æ	æ~æɛə	ɛː~ɛj	eː~ej
Non-prevocalic /ɡ/	bag, drag	ɛəA	æC	æ
Prevocalic /ɡ/	dragon, magazine	æ
Non-prevocalic /b, d, ʃ/	grab, flash, sad	ɛəA					æ	ɛə
Non-prevocalic /f, θ, s/	ask, bath, half, glass	ɛəA
Otherwise	as, back, happy, locality	æD
^ a b c d In New York City, Philadelphia, and Baltimore, most function words (am, can, had, etc.) and some learned or less common words (alas, carafe, lad, etc.) have æ. ^ In Philadelphia, the irregular verbs began, ran, swam, and wan (a local variant of won) have æ. ^ In Philadelphia, bad, mad, and glad alone have ɛə. ^ In New York City, certain lexical exceptions exist (like avenue being tense) and variability is common before /dʒ/ and /z/ as in imagine, magic, and jazz. In New Orleans, ɛə additionally occurs before /v/ and /z/.

• "Short o" before r before a vowel: In typical North American accents (U.S. and Canada alike), the historical sequence /ɒr/ (a short o sound followed by r and then another vowel, as in orange, forest, moral, and warrant) is realized as oɹ~ɔɹ, thus further merging with the already-merged /ɔr/–/oʊr/ (horse–hoarse) set. In the U.S., four words (tomorrow, sorry, sorrow, borrow and, for some speakers, morrow) usually contain the sound ɑɹ instead, and merge with the /ɑr/ set (thus, sorry and sari become homophones, both rhyming with starry).

• v
• t
• e

General American /ɑr/ and /ɔr/ followed by a vowel, compared with other dialects

	Received Pronunciation	General American	Some New England, NYC, Mid-Atlantic, Southern American	Canadian
Only borrow, sorry, sorrow, (to)morrow	/ɒr/	/ɑːr/	/ɒr/ or /ɑːr/	/ɔːr/
Forest, Florida, historic, moral, orange, etc.		/ɔːr/
Forum, memorial, oral, storage, story, etc.	/ɔːr/		/ɔːr/

Some mergers found in most varieties of both American and British English include:

• Horse–hoarse merger: This merger makes the vowels /ɔ/ and /o/ before /r/ homophones, with homophonous pairs like horse/hoarse, corps/core, for/four, morning/mourning, war/wore, etc. homophones. Many older varieties of American English still keep these sets of words distinct, particularly in the extreme Northeast, the South (especially along the Gulf Coast), and the central Midlands, but the merger is evidently spreading and younger Americans rarely show it.
• Wine–whine merger: This produces pairs like wine/whine, wet/whet, Wales/whales, wear/where, etc. homophones, in most cases eliminating /ʍ/, also transcribed /hw/, the voiceless labiovelar fricative. However, scatterings of older speakers who do not merge these pairs still exist nationwide and perhaps most strongly in the South.