The DARE Pronunciation Key

This section has three major divisions: I. Alternations between Sound Units, II. Variation within Sound Units, III. Summary of Major Regional Variants.

The purpose of section 3.I is to describe particular phonological contexts and to show how more than one sound unit can be used in such contexts without changing the meaning of a word. For example, in the words near, cheer, and beer, two vowels, /i/ and /ɪ/, occur frequently in AE. The words mean the same thing no matter which vowel sound is used. In other contexts, however, the two vowels are not interchangeable: in such word pairs as /bit/ beat and /bɪt/ bit, /sip/ seep and /sɪp/ sip, /liv/ leave and /lɪv/ live, /pič/ peach and /pɪč/ pitch, the vowel is the sound unit that signals a difference in meaning, and to use a different vowel is to confuse the listener. Section 3.I describes many contexts in which more than one vowel can occur without altering the meaning of a word. This kind of alternation between vowels can vary either by geographic region or by social group.

Section 3.II describes the kinds of variation that can occur within individual sound units. For example, the diphthong /aᴜ/ of about occurs as [aᴜ], [ɑᴜ], or [æᴜ] in most parts of the country but is often [ʌᴜ] in parts of Virginia, Maryland, and North Carolina and the coastal areas of South Carolina, Georgia, and Florida. All three variants, however, are understood as signifying the same word—that is, though they are recognizably different, they all represent the same sound unit. This kind of alternation of sounds within individual units can also vary by geographic region, social group, or individual speaker.

Because both variation between units and variation within units can be the result of the conditioning effects of neighboring sounds, the distinctions may seem somewhat artificial. For example, the phonological context of a vowel followed by /š/ yields variation between [æš] and [æɪš] in the word ash, with both [æ] and [æɪ] being considered variants of the same sound unit. But before /š/ in the word push, many speakers who have [æɪ] in ash will have [puš] rather than [pᴜš]—/u/ and /ᴜ/ being considered two different sound units. Since the conditioning effect of the following /š/ determines the vowel in both cases, to consider one variation to be within a unit and the other to be between units may seem inconsistent. But it is the behavior of the sounds in all of their environments—not just one—that determines the classification of sound units; and while the divisions used here may overlap on occasion, the basic framework reflects as accurately as possible the functioning of the sound units in AE as a whole.

Section 3.III describes some of the major alternations considered in 3.I and 3.II in terms of their broad regional patterns.

Even though geographical areas where variants are used frequently are listed, the reader should keep in mind the following caveats: (1) Not all speakers within the given area can be expected to use any given form; social and localized restrictions are numerous and complex. (2) The frequency of occurrence of any form relative to its variants may be very high in some geographical locations but low in others. For example, the tensing of all lax vowels before /š/ and /ž/ occurs in the speech of a diminishing minority of speakers from Delaware through the Ohio River Valley. But the use of the vowel /e/ rather than /ε/, as in measure [mežɚ], treasure [trežɚ], is much more widespread geographically and in terms of the number of speakers using it than alternations between other vowels before /š/ and /ž/. (3) The sound described may well occur outside the designated geographical area; these areas should be taken as broad designators of relatively frequent occurrence of a variant. (4) No study is likely to discover all possible pronunciation variants. Especially in a wide area survey such as the DARE survey, done largely to collect lexical variants, there are limitations on both the number and the social types of speakers interviewed. The complexities become greater but more identifiable in intensive local studies. In brief, the following sections describe the most notable variants and their major geographical limitations. They are to be used as a pronunciation key to the dictionary entries and for general interest and information.

3.I Alternations between Sound Units

3.I.0 Vowels.

3.I.1 Vowels before retroflex /r/ in monosyllabic words.

a. fear, beard, near, etc. Both /i/ and /ɪ/ are found nationwide, varying in complex ways. /i/ occurs most strongly in the Southeast and as far west as Texas, less strongly in the Northeast, North, and urban West Coast. /ɪ/, more general geographically, appears to predominate in the West and eastward through the center of the country into Pennsylvania. Infrequently the sounds /e/, /ε/, /ɜ/ occur, particularly among older speakers in parts of the South Midland, such as the Ozark region.

b. bear, chair, hair, etc. /ε/ predominates, with /e/, /æ/, /ɪ/, /ɑ/, and /ɜ/ as alternates. Word incidence (chair, care, dare, etc.) is highly variable. /e/ is most frequent in the northeastern and southeastern quarters of the country. /æ/ occurs, especially in older speakers, in the South Midland from southern Delaware into Texas and rarely in the rural West. The rarer /ɪ/, /ɑ/, and /ɜ/ appear sporadically but most commonly on the Atlantic and Gulf coasts.

c. barn, farm, car, etc. The vowel /ɑ/ occurs throughout the country (though variants have some differences in distribution: see 3.II.10). The rounded low-back vowel /ɔ/ sometimes occurs, in Delmarva and in parts of the rural West (as Utah).

d. poor, sure, moor, etc. Variation in words of this type appears to be no less complicated than each word’s historical development. The vowels /u/, /ᴜ/, /o/, /ɔ/, and /ɜ/ are the major alternants. Both /ᴜ/ and /ɔ/ can be found in all parts of the country, with /ᴜ/ and /u/ most frequent in the northeastern quarter, /o/ and /ɔ/ in the southeastern quarter, /ᴜ/ and /ɔ/ in much of the West. /ɜ/ is the least common vowel in this context, occurring mainly in the word sure (where it is quite general), and varying both by speaker and by social context.

e. horse, hoarse, morning, mourning, etc. /ɔ/ predominates in both word types, but /o/ is a major variant, /ɑ/ an infrequent one. A distinction between /ɔ/ in horse and morning and /o/ in hoarseand mourning occurs inconsistently along the East Coast (becoming more frequent in the South) and even less consistently in inland areas, such as central New York state, St. Louis, and perhaps the rural West. It appears to be retained mostly by older speakers. When no distinction is made between /o/ and /ɔ/, these words appear to have /o/ most strongly in the Northeast and the Southeast, with some currency in the rural West. /ɑ/ occurs sporadically, occasionally as a stigmatized feature, in the South Midland extending from North Carolina into Texas and the rural West; in the St. Louis area it is characteristic of urban speech.

f. learn, earth, service, etc. Constricted variants of /ɜ/ are usual here, but /ɑ/ also occurs sporadically, as in learn /lɑrn/, burn /bɑrn/, earth /ɑrθ/. It is less common today than formerly and occurs most frequently among older speakers and those with little formal education.

g. fire, tire, wire, etc. The diphthong /aɪ/ is the most frequent vowel in these words, although /ɑ/ also occurs frequently in the South Midland and adjacent areas, and occasionally in the rural West. /ɔ/ sometimes occurs in Delmarva.

h. flower, shower, hour, etc. The /aᴜ/ diphthong occurs most frequently in these words, though occasionally /ɑ/ is heard, especially in Pennsylvania, and /æ/ occurs occasionally in the South Midland.

3.1.2 Vowels before intersyllabic /r/.

Because of the strong influence on vowels of an /r/ following in the same syllable, in polysyllabic words (e.g., carry, horrible) vowel variation depends, in part, on whether the medial /r/ articulation is present at the end of the stressed syllable or at the onset of the next syllable only. (The possible effects of presence or absence of this “linking” sound apply to /l/ as well.) However, variants produced by other influences also occur in these word types.

a. carry, parent, merry, Mary, bury, etc. /ε/ occurs most frequently in these words overall, with some striking variation in word incidence and in regional patterning. In words like carry, parent,and marry /æ/ also occurs frequently, especially in those areas of the East Coast and the South which lack postvocalic /r/ in monosyllables, but also in the Inland North through the Great Lakes region to the West, where word incidence complicates description. The word parent, for example, appears to have /æ/ more frequently than other words of the type. In Mary, various, area, and several other words, /e/ is frequent in the South, occasional in some words in the North. In merry, bury, /ε/ predominates, but /ɜ/ also occurs sporadically, especially in such areas as Philadelphia and Texas.

b. forest, orange, tomorrow, etc. /ɔ/ probably predominates, but /ɑ/ is very frequent on the East Coast, extending westward through the southern half of the country, lessening in frequency in the rural West. Word incidence, however, complicates this generalization. Tomorrow and sorry, for example, have /ɑ/ generally; but some speakers in the North, particularly the western Great Lakes area (Wisconsin, northern Illinois) have /ɔ/ in those words. The word orange has /ɑ/ among some younger speakers who otherwise have /ɔ/.

c. hurry, curry, furry, etc. /ɜ/ predominates, but /ǝ/ also occurs, especially in areas of the East and South that lack postvocalic /r/ in monosyllables. The Inland North and West have both vowels.

3.I.3 Vowels before /l/.

a. milk, pillow, Illinois, etc. While /ɪ/ is the usual vowel in these words, /ε/ also occurs, especially in the Great Lakes area.

b. television, yellow, help, etc. /ε/ predominates, but /æ/ alternates, especially in the speech of those who have /ε/ rather than /ɪ/ in milk, pillow, Illinois, etc.

c. really, steel, wheel, etc. Although /i/ is the most common vowel, younger speakers throughout the West and speakers of all ages across the southern half of the country frequently have /ɪ/. The alternation does not affect all words of this type equally: a speaker who pronounces really as [ˈrɪlᵻ] will not necessarily pronounce steel as [stɪl].

d. sale, jail, mailer, etc. /e/ is most common, but a variant of /ε/ also occurs, especially among younger speakers in the West. Because of variation within the two sound units /e/ and /ε/, words such as sale and sell do not usually sound alike for those who have /ε/ in sale.

e. call, doll, fall, etc. /ɑ/ predominates, but /ɔ/ is a frequent alternate. These vowels are highly variable in nearly all contexts, making generalizations somewhat unreliable, but it can be said that /ɔ/ is especially frequent in the Boston area and also occurs irregularly, particularly among older speakers, from western Pennsylvania into the rural West.

f. pile, mild, child, etc. The /aɪ/ diphthong is usual in these words (with much variation within the unit itself; see 3.II.13), but the monophthong /ɑ/ also occurs in the South and South Midland.

g. howl, prowl, growl, etc. The /aᴜ/ diphthong is the most common vowel in these words, but occasionally /æ/ and /ɑ/ occur.

h. result, bulk, etc. /ǝ/ predominates, with /ᴜ/ also occurring fairly frequently in the South Midland and adjacent areas and sporadically in the West, especially in non-metropolitan areas. /ɑ/ occurs occasionally.

i. cool, fool, pool, etc. The tense vowel /u/ is most common, but the lax vowel /ᴜ/ also occurs occasionally, especially among younger speakers in the West.

3.I.4 Vowels before /n/.

a. pen, ten, mentor, etc. While /ε/ occurs most frequently, the higher vowel /ɪ/ is also very common, especially in the southern half of the country and in the rural West. For some speakers, /ɪ/ and /ε/ are not distinguished before /n/, while for others the coalescence is restricted to particular words. (The word been, which is regularly pronounced /bɪn/ in much of the country, is occasionally /bεn/ in the South even for those who have /ɪ/ in pen, men, send, etc.) /e/ and /æ/ also occur occasionally, especially in the South Midland, in such words as bench.

b. onion, country, un-, etc. /ǝ/ occurs most frequently, but the lower central vowel /ɑ/ also occurs, especially as a prefix, as in uncommon, uncertain, unequal, etc. /ɑ/ seems to be most frequent in the Mid Atlantic, Upper Midwest, and parts of the South. It results in spellings such as “oncommon” and has been stigmatized as a “wrong” pronunciation.

c. on, dawn, etc. Both /ɑ/ and /ɔ/ occur frequently, with /ɔ/ being more common in the East and South and /ɑ/ predominating in the Inland North and the urban areas of the West.

3.I.5 Vowels before /š/, /ž/, /č/, /ǰ/.

a. fish, dish, itch, etc. /ɪ/ is most common, but /i/ also occurs, especially from Delaware and Maryland westward through the Ohio River Valley into northwestern Missouri; it also occurs occasionally in the rural West and in the South Midland. /i/ is most frequent among older speakers.

b. measure, treasure, fresh, edge, etc. /ε/ predominates, but /e/ also occurs. The distribution of /e/ is similar to that of /i/ in fish (above), except that it occurs more extensively in the South and West and is more frequent among speakers of all age groups.

c. bush, push, butcher, etc. /ᴜ/ predominates, but /u/ also occurs, especially in the South Midland and into the rural West. Use of /u/ seems to be receding. Infrequently, /ɜ/ occurs.

d. rush, mush, begrudge, etc. /ǝ/ is by far the most common vowel, but /ε/ also occurs, especially in the South Midland, and /ɜ/ occurs sporadically, especially among older, less educated speakers in the rural Ohio Valley.

e. gosh, wash, botch, etc. Both /ɑ/ and /ɔ/ occur frequently, with /ɑ/ being usual in the South, parts of the Northeast, and much of the West, and /ɔ/ occurring regularly in the South Midland. Occasionally the diphthong /ɔɪ/ occurs, especially in the Ohio River Valley and in parts of the South Midland. Pronunciations with intrusive /r/—[wɑrš], [wɔrš], [gɑrš], [gɔrš]—also occur, with widespread though sporadic evidence from Pennsylvania westward and southward (except in areas without postvocalic /r/), and occasional occurrence on the West Coast.

3.I.6 Vowels before /g/, /ŋ/.

a. egg, leg, keg, etc. /ε/ predominates, but /e/ is also frequent, sometimes alternating with /ε/ even within the speech of individuals. Across the North /ε/ dominates, becoming mixed with /e/ in the rural West; from the South Midland into the Southwest, /e/ is most frequent; in the South and the lower Ohio Valley, both vowels occur. Cag, with /æ/, was once very common for keg.

b. log, frog, fog, etc. Both /ɑ/ and /ɔ/ are current, but they vary more by word incidence than the vowels in other classes, making generalizations somewhat tenuous. On the whole, /ɑ/ is most frequent across the North and in the urban West; /ɔ/ is most frequent through the Ohio River Valley, the South Midland, and into the rural West. The South and eastern New England have /ɔ/ most frequently, with /ɑ/ occurring not infrequently. As with the low vowels in other contexts, younger speakers appear to be adopting the more fronted vowel.

c. long, song, strong, etc. Variation between /ɑ/ and /ɔ/ is complex, but /ɔ/ probably predominates in terms of both geographic range and use throughout different age groups. The unrounded /ɑ/ occurs on the Atlantic Coast from Virginia southward to Florida, and in the urban West. It also occurs from Pennsylvania westward, often varying with /ɔ/. The South Midland has /ɔ/, often varying with /ɑ/ in its western regions. In the North, /ɑ/ and /ɔ/ are mixed in eastern New England, /ɑ/ is frequent in the western Great Lakes area, and /ɔ/ is common further west.

d. ring, thing, think, etc. While /ɪ/ is most frequent, both /ε/ and /æ/ also occur, especially in the South Midland extending into Texas, with sporadic occurrences in the South and the rural West.

3.I.7 /æ/ and /ɑ/ (“broad a”).

aunt, rather, dance, path, etc. /æ/ predominates, with /ɑ/ (including both fronted [a] and retracted [ɑ˃] as variants) also occurring, especially in New England and on the Atlantic Coast. Word incidence appears to be highly variable in those areas that have [a] and [ɑ˃]. Word incidence is a stylistic factor among older speakers in much of the country, including Hawaii, where /ɑ/ may be used in aunt and rather in reading style, but /æ/ appears in conversational usage. Aunt and rather have frequent incidence of /ɑ/ (articulated as [ɑ˃]) sporadically down the East Coast to Florida, and (articulated as [a]) in New England, but are quite variable in incidence. Younger speakers appear to have /æ/ forms more frequently and regularly across the word set.

3.I.8 /u/ and /ᴜ/ before front consonants: /p/, /t/, /f/, /v/, /m/.

coop, root, roof, hooves, room, etc. These words exemplify a limited set that gives evidence of variation of /u/ and /ᴜ/. /u/ dominates in the usage of younger speakers and those in urban areas, although any word may prove an exception. The rural, the older, and the less educated often have variable usage, even for a single word. The less frequent vowel /ᴜ/ has its most frequent occurrence in the Northeast, sporadically west through the Great Lakes into Minnesota, and on the South Atlantic and Gulf Coasts into Texas, with scattered occurrences in the rural West (being most frequent there in Washington and Hawaii).

3.I.9 /ɑ/ and /ɔ/ after /w/.

wash, water, watch. These words have both /ɑ/ and /ɔ/, with variation among words and even word-forms (e.g., wash, Washington) being common. In water most of the East Coast and the South have /ɔ/. Other areas frequently have /ɑ/, but /ɔ/ (particularly when articulated as [ɒ]) is not infrequent. /ǝ/ occurs occasionally in Pennsylvania and New Jersey. In wash, /ɑ/ occurs along most of the East Coast (except for eastern New England) and throughout the South into Texas, and often in the West; /ɔ/ is frequent elsewhere; /ǝ/ occurs sporadically. Watchappears to have /ɑ/ more frequently than /ɔ/ everywhere, but backed or rounded forms occur sporadically.

3.I.10 /u/, /ju/, /iu/ after /t/, /d/, /n/.

tube, due, new, etc. Following an alveolar consonant, the vowel /u/, the palatalized glide /ju/, and the back-gliding vowel /iu/ all occur. /u/ is most common, occurring through most of the country. In addition to /u/, /ju/ also occurs frequently on the Atlantic Coast from Delaware and Maryland south into Florida, then westward into Arkansas and eastern Texas, as well as in the San Francisco area and in Hawaii. There is also scattered usage in the rural West. In New England, /iu/ occurs alongside /u/, though less frequently than formerly. It is most common in the speech of those with little formal education. The vowel [ɪu], probably to be considered a variant of /ju/, also occurs in word sets with o spellings. For example, do and too occur as [dɪu] and [tɪu] sporadically, not only in areas with /ju/ in due and tube, but also in the South and the rural West, even among speakers who lack the glide in new, due, etc.

3.I.11 Alternation of /ɔɪ/ and /aɪ/ before /l/, /n/, /s/, /z/.

joint, point, boil, oil, poison, hoist, etc. The most frequent vowel in these words is the diphthong /ɔɪ/. In parts of eastern New England, in the southern Appalachians, in scattered areas of the Mid and South Atlantic, and sporadically elsewhere, /aɪ/ also occurs. /aɪ/ is especially common among older people in folk speech and is also retained as a jocular form among some speakers of standard English. Word incidence varies, with /aɪ/ occurring more frequently in hoist than in other words, with heist perhaps being understood as a distinct word.

3.I.12 Vowels in weakly stressed final syllables.

a. haunted, bucket, houses, etc. In the final syllable of such words the front vowel /ɪ/ (often articulated as a retracted variant [ᵻ]) appears to predominate over mid-central /ǝ/. /ǝ/ is most common from the Mid Atlantic states westward to Ohio, alternating with /ɪ/ into the West and Southwest. The North, most of the South, and the West have /ɪ/ predominantly.

b. sofa, china, California, Florida, etc. Words of this type usually end in /ǝ/, but the vowel /i/ also occurs in folk speech (especially in northern New England and parts of the South Midland and South), and /ɜ/ (articulated as [ɚ]) also appears occasionally.

c. Missouri, Cincinnati, etc. Both /i/ and /ǝ/ occur in these place names, with widely mixed usage. In general, natives of these places tend to have /ǝ/, while outsiders tend to use spelling-pronunciations with /i/.

d. meadow, window, follow, tomato, etc. Words of this group may end in /o/, /ǝ/, or /ɜ/ (articulated as [ɚ]). In general, /o/ predominates in the North, North Midland, and West, while /ǝ/ is most frequent in the South Midland, South, and their settlement areas. Especially among folk speakers in the South Midland and South, /ɜ/ also occurs. In words like borrow, /i/ occurs alongside /o/ and /ǝ/. Like /ɜ/, /i/ occurs most frequently in folk speech in the South Midland and South.

3.I.13 Consonants.

Alternations between consonants tend to be somewhat less systematic in geographic distribution than those for the vowels, and often occur less regularly in all the words in a set.

3.I.14 Alternations among voiceless stops: /p/, /t/, /k/.

These consonants occasionally alternate with one another, though not always in regular or reciprocal patterns. In such words as ask, desk, and stark, /t/ can occur for usual /k/; the reverse is true in words such as turtle, blast, and credit, where /k/ sometimes occurs. Infrequently /t/ occurs where /p/ is usual in words such as wasp [wɔst]; and /p/ for /t/ has been observed in such forms as catnip [kæpnɪp]. The velar stop /k/ can alternate with the bilabial /p/, as in baptize [ˈbækˌtaɪz].

3.I.15 Alternations between voiced and voiceless consonant pairs: /p~b/, /t~d/, /f~v/, /s~z/, /θ~ð/, /č~ǰ/.

Alternations between voiced and voiceless counterparts occur with some frequency. In a word such as Baptist, for example, the pronunciations [bæbdᵻs(t)] and [bæbtᵻs(t)], rather than [bæptᵻs(t)], are common, especially in the South and South Midland. The use of /z/ in such words as greasy, blouse, (and perhaps absurd) is characteristic of most of the South, the Midland, and much of the Southwest, while /s/ in those words occurs regularly in the North (with a few exceptions, such as New York City) and on the West Coast. In the words with and without, both voiced /ð/ and voiceless /θ/ occur, with /ð/ generally dominating in the North and usage being mixed elsewhere. Some alternation between /f/ and /v/ takes place, as in nephew [ˈnεvju] andfascinator [ˈvæsɪnetɚ], but this seems to be a relic pattern occurring in scattered isolated communities rather than in broad regional patterns. The use of /č/ where /ǰ/ is usual, as in college[ˈkɑlɪč], occurs infrequently, being largely restricted to areas of dense foreign-language settlement.

3.I.16 Alternation between /k~kj/, /g~gj/.

In words such as car, garden, and cow, where the low vowel or diphthong is usually preceded by a velar stop, alternation between the stop and the stop plus a palatal glide occurs. The clusters /kj/, /gj/ occur most often in the South Midland and in parts of the South. Less frequently the clusters occur before front vowels, as in care and carry, often articulated as [kɪ] rather than [kj]. These clusters occur most often on the Atlantic Coast between the Potomac and Savannah Rivers, sporadically westward to the Mississippi River, and occasionally in the West.

3.I.17 Alternations among stops and fricatives.

Use of the stops /t/ and /d/ rather than the fricatives /θ/ and /ð/, as in thing [tɪŋ] and this [dɪs], is fairly widespread, especially in northern urban areas (where it is often characteristic of working-class speech), in the South (especially among Blacks), and in areas such as the Upper Midwest and the Southwest that have had dense settlement by foreign-language speakers. These alternations are common even within an individual’s speech. Others are much less common: use of /f/ for /θ/ and /v/ for /ð/ as in tooth [tuf] and with [wɪv] occurs occasionally, especially among conservative Black speakers; the use of /b/ where /v/ is expected, as in seven [ˈsεbṇ], [ˈsεbṃ] or drive [draɪb] also occurs especially frequently among Blacks; the reverse alternation, of /v/ for /b/, as in February [ˈfεvǝˌwεri], occurs much less frequently. Occurrence of the stop /d/ where the fricative /z/ is expected, as in business [ˈbɪdnᵻs], or wasn’t[ˈwʌdnt], is especially frequent in the South but is not restricted to that area.

3.I.18 Alternations among fricatives and affricates.

Alternation of /s/ for /š/ occurs most frequently before /r/, as in shrimp, shrink, and shrivel, with pronunciations [srɪmp], [srɪŋk], [ˈsrɪvǝl] occurring most often in the South and South Midland, often among Blacks. Scattered instances occur elsewhere. (A further substitution, yielding [swɪmp], [swɪŋk], [ˈswɪvǝl], also occurs most frequently among Blacks in the South.) In such words as measure, garage, beige, and television, /ž/ and /ǰ/ alternate, but stylistic and social variation are more prominent than regional patterning. In garage, both [gǝˈrɑž] and [gǝˈrɑǰ] are frequent, though the former pronunciation is often a learned one. Pronunciations of the type [ˈmeǰɚ], [beɪǰ], and [ˈtεlǝˌvɪǰǝn] occur most frequently among speakers with little formal education. The same is true for words such as rinse, lance, and wince, where /č/ can occur for /s/ in contexts following a nasal consonant ([rɪnč], [rεnč], [lænč], [wɪnč]); such forms illustrate the preservation of older pronunciations.

3.I.19 Alternations among fricatives and frictionless continuants.

Use of the frictionless continuant /w/ where the fricative /v/ is expected, as in varmint [ˈwɑ(r)mᵻ(n)t], very [ˈwεri], and November [noˈwεmbɚ] is a characteristic both of English Cockney speech and of creolized varieties of English; it occurs most frequently among Blacks in the South Carolina–Georgia Low Country, and also occurs occasionally in Hawaii. The reverse alternation, of /v/ for /w/ as in wax [væks] occurs, but rarely.

The articulation of /h/ before /w/, as in wheel [hwil], where [hwεr], why [hwaɪ], seems to be receding in favor of the pronunciations [wil], [wεr], and [waɪ]. Use of /hw/ is highly subject to style shifting, occurring much more frequently in reading and formal speech than in conversational speech. Regional patterns of use of /hw/ and /w/ were identifiable several generations ago but are much less so today, with use varying greatly within regions and even within the speech of many individuals. Retention of /h/ before /j/, as in humor and huge, also varies widely within regions, with /hj/ probably dominant. The New York City metropolitan area often has /j/, as do other parts of the East Coast, but this varies from word to word as well as within and across regions.

3.I.20 Nasal consonants.

Alternations among nasal consonants vary less by region than by phonological context, often being affected by contiguous consonants. In words such as length and precinct, for example, the usual /ŋ/ sometimes varies with /n/, as [lεnθ], [ˈpriˌsɪnt], under the influence of the following interdental and dental consonants. Similarly, conversation becomes [kɑmvɚˈsešǝn], and is sometimes further assimilated to [kɑmpɚˈsešǝn]. Use of /n/ for /ŋ/ in words with final -ing is widespread, though it occurs more regularly in the South and South Midland than elsewhere. Its obverse, use of /ŋ/ for /n/, occurs among speakers who, having been corrected for saying runnin’ rather than running, carry that “correction” to words that may seem analogous, but are not, so that words such as mountain and button become [maᴜntɪŋ] and [bʌtɪŋ]. Alternation of /n/ with /m/, as in palm [pɑn], [pæn] and mushroom [ˈmʌšˌrun] occurs sporadically; it is especially frequent among older speakers.

3.I.21 Alternation of /r/ and /ə̯/.

In words having an r spelling after a vowel in the same syllable (such as fire, course, board), the consonant /r/ varies with a vowel-like unsyllabic consonant /ə̯/ according to fairly distinct regional and social patterns. Therefore, the use of /ə̯/ (also understood as the loss or “dropping” of /r/) is one of the most easily recognized features of variation in AE. Postvocalic /r/ is regularly retained in all of the Midland and North (except for eastern New England and the New York City metropolitan area) and in the West and Southwest. It is also retained in some areas of the Atlantic Coast, as in Delmarva and parts of the Carolinas. On most of the Atlantic Coast, however, and through most of the South (with the exception of many areas in Florida), and sporadically elsewhere, postvocalic /r/ is not retained but occurs as /ə̯/. That is, words such as fort, charm, and fire are pronounced not as [fɔɚt], [čɑɚm], and [faɪɚ], but as [fɔə̯t], [čɑə̯m], and [faɪə̯], where the /ə̯/ is neither a part of the preceding vowel nor the nucleus of a second syllable. Within the geographic patterns outlined above, some social distinctions can be made. Particularly in northern cities with large numbers of Black residents, the /ə̯/ consonant occurs alongside /r/. In areas of predominant use of /ə̯/, the adoption of /r/ seems to be spreading among younger and better-educated speakers.

3.I.22 Reduction of consonant clusters.

In syllables with two or more contiguous consonants, the cluster is often “reduced,” or pronounced without the distinct articulation of each sound. In words such as ground, boiled, last, width, loaves, for instance, many speakers articulate only one final consonant, producing [graᴜn], [bɔɪl], [læs], [wɪθ], [loz]. Similarly, in words with three final consonant sounds, the clusters are often reduced to two or even one consonant: asked [æskt] may become [æst] or [æs]; wolves may become [wᴜlz] or [wᴜvz]; fifths may become [fɪθs] or [fɪfs]; and words such as wasps, desks,and fists may be reduced to [wɑsp], [dεsk], [fɪst], or even further to [wɑs], [dεs], [fɪs], or may have a vowel added between consonants to facilitate articulation, as [wɑs(p)ɪz], [dεs(k)ɪz], [fɪs(t)ɪz]. Cluster reduction can also occur initially and medially, as in scratch [kræč], library [ˈlaɪbεrɪ], government [ˈgʌvɚmǝnt], [ˈgʌvmǝnt]. While reduction of consonant clusters occurs in all regions, it is somewhat more frequent in the South and South Midland than elsewhere.

3.I.23 Excrescent sounds.

The introduction of sounds where there is no historic basis for them occurs frequently, with regional and social patterns varying from word to word. Such “intrusive” or “excrescent” sounds may be either vowels or consonants. They may occur initially, medially, or finally, and appear in such varied words as across [ǝˈkrɔst], anything [ˈεnɪθɪŋk], apron [ˈemprǝn], athlete [ˈæθǝlit],attic [æntɪk], balance [ˈbælǝnst], breakfast [brεkfrǝst], curl [kwɝl], earth [jɝθ], elm [εlǝm], family [ˈfæmblɪ], percolate [ˈpɝkjulet], and wash [wɔrš]. (See also Language Changes Especially Common in American Folk Speech.)

3.II Variation within Sound Units

This division treats each of the forty sound units, with the major variants, recognized in American English. No given variant can be exhaustively treated here: occurrence outside the stated limitations is possible, and not every speaker who fits the stated limitations necessarily has the given variant. To a certain extent, what one hears in a given place depends on who the speaker is and under what conditions he or she is speaking. And finally, some variants may be used by a majority of the speakers in one region, by a minority in another; estimates of probable proportions can only be general. All phonetic variation is probably governed by context—neighboring sounds, stress, grammatical category, usage level, even word meaning—as well as by the location and social characteristics of the speaker, but detailed, narrow studies would have to be done to establish these characteristics firmly. The following discussions are based on a wider network and represent broader patterns of use. They should provide the information necessary for a reader to infer the most likely pronunciations for particular regions.

3.II.0 The stressed vowels.

The key words given illustrate only some of the possible phonetic contexts for each vowel.

3.II.1 /i/ beat, dream, feel, etc.

The principal variants are [i], lengthened [i:], upgliding [ɪi], and ingliding [iǝ]. Though these variations are subtle, they show some regional and social patterning. [ɪi] occurs in all regions, but competes with the other variants in given areas. The ingliding vowel (when not produced by stress patterns) occurs most prominently in the South Carolina Low Country (approximately the eastern third of the state); it is receding in favor of [ɪi] and [i(:)]. Monophthongal [i(:)] is found in New York City, eastern Pennsylvania, sporadically across western New England, New York, Ohio, Michigan, and Illinois, more often in Wisconsin, Minnesota, and the Dakotas, and sporadically in the Northwest. It is occasional throughout the South Midland into Texas. It appears to be strongest in areas heavily settled by speakers of Germanic languages.

3.II.2 /ɪ/ bit, him, pill, etc.

The principal phonetic variants are [ɪ], lengthened [ɪ:], ingliding [ɪǝ], retracted [ᵻ], and lowered [ɪ˅]. [ɪ(:)] and [ɪǝ] are the most common, occurring throughout the country. South of a line from Philadelphia to Kansas City, [ɪǝ] is quite common except in coastal areas. Younger speakers in the Inland North also are making increased use of this form. In both areas, stress and phonological context variably control the use of [ɪǝ]. In the South into Texas, [ᵻ] occurs in scissors, sister, and several other words. Across the North, sporadic occurrence of a lowered [ɪ˅] (approaching [ε]) occurs, particularly but not exclusively in younger speakers.

3.II.3 /e/ bait, main, tail, etc.

The principal phonetic variants are [e], lengthened [e:], upgliding [eɪ], ingliding [eǝ], and upgliding [εɪ] with a lowered onset. [eɪ] predominates generally, but in some regions other variants are frequent. [e] and [e:] occur in nonurban areas of New England and westward through the Great Lakes area to the West Coast, including Alaska; they are particularly strong in Wisconsin, Minnesota, and the Dakotas. They also occur in eastern Pennsylvania and sporadically westward into the Great Plains. They compete with [eǝ] in eastern South Carolina, and may also occur in areas with concentrated non-English settlement, such as Hawaii. The upglide [εɪ] occurs in both Philadelphia and Pittsburgh and sporadically in central Pennsylvania and Ohio. It appears to be receding in favor of other variants.

3.II.4 /ε/ bet, men, help, etc.

The principal phonetic variants are [ε], ingliding [εǝ], raised [ε˄], lowered [ε˅], and upgliding [εɪ]. Throughout most of the country, [ε] is the most common, with other variants prevailing only in specific phonological contexts. In the southern half of the country as far west as New Mexico, however, ingliding [εǝ] tends to dominate, especially before /l/ and /r/. It also occurs sporadically in the North and seems to be increasing among young speakers. The raised variant [ε˄] occurs widely before nasal consonants (and sometimes alternates with the /ɪ/ sound unit; see 3.I.4). The lowered [ε˅] occurs mainly among young speakers in northern urban areas, in words like desk and best. The upglide occurs most often in the South and South Midland.

3.II.5 /æ/ bat, sand, pal, etc.

The principal phonetic variants are [æ], a raised and nasalized [æ̃˄], ingliding [æǝ], and upgliding [æɪ]. Although [æ] occurs in all parts of the United States, most areas also have at least one of the other variants in some phonological environments. The upgliding [æɪ], as in ash and bag, occurs most frequently south of a line from Delaware through southern Missouri, being less common in coastal areas than inland. Scattered instances occur in New England and the rural West. Raised and nasalized variants are common from New England and New York City westward to the Pacific Coast (including San Francisco). This feature appears to be spreading among the young, even in areas outside those listed above. In the South the raised variant without nasalization can be heard. The ingliding diphthong [æǝ] occurs occasionally in the South and South Midland.

3.II.6 /u/ boot, fool, moon, etc.

The principal phonetic variants are [u], upgliding [ᴜu] and [ʌu], ingliding [uǝ], and centralized [ʉ]. The upgliding diphthong [ᴜu] is the most widespread variant. In parts of the northern Midwest, eastern Pennsylvania, the Georgia–South Carolina coast, and sporadically elsewhere, the monophthong [u] occurs, often being attributable to foreign-language influence. Younger speakers in these areas tend to use this form less frequently than their parents. [uǝ] and [ʌu] occur in the South and several adjacent areas: the former especially near Charleston, South Carolina, the latter throughout the South, especially in the east. The centralized [ʉ] occurs in the South, the lower Midwest, and the rural West, and seems to be increasing among younger speakers.

3.II.7 /ᴜ/ good, pull, push, etc.

The principal phonetic variants are [ᴜ], ingliding [ᴜǝ], frontgliding [ᴜɪ], and centralized [ᵾ]. The most widespread variant is [ᴜ]. The inglide occurs frequently throughout the South Midland, the lower Midwest, and the rural West and occasionally in New England and the South. The centralized variant is heard frequently in the South Midland and adjacent areas of the Midwest and the Southwest, and is also found in the rural West. Particularly before /š/, as in push, the front-gliding variant [ᴜɪ] occurs, being most frequent in parts of New England and the South.

3.II.8 /o/ boat, rose, gold, etc.

The principal phonetic variants are an upgliding [oᴜ], ingliding [oǝ], upgliding [ɞᴜ], monophthongal [o], and mid-back [ɵ] (traditionally called the “New England short o”). Upgliding [oᴜ] occurs through most of the country. The monophthong [o] occurs most frequently in the Upper Midwest, along the Georgia–South Carolina coast, and sporadically elsewhere, especially where foreign-language settlement has been dense. It is becoming a conservative pronunciation. The ingliding variant occurs most often in coastal South Carolina, Georgia, and Florida, but also in eastern Virginia and the southernmost part of Maryland. Formerly found chiefly in the Baltimore-Philadelphia area, western Pennsylvania, eastern Ohio, and northeastern North Carolina, the upglide [ɞᴜ], with centralized start, seems to be gaining currency, occurring sporadically across the North and into the West. It appears to be increasing particularly among younger speakers. The New England [ɵ], by contrast, has been receding rapidly and is becoming rare even in northeastern New England, where it was still common a generation ago.

3.II.9 /ɔ/ caught, bought, August, etc.

Not all speakers have the raised low-back rounded vowel /ɔ/. In areas in which it occurs, it often contrasts with /ɑ/ in such word pairs as cot–caught /kɑt/–/kɔt/ and tot–taught /tɑt/–/tɔt/. For those who do not make the distinction between /ɑ/ and /ɔ/, the retracted variant [ơ] of the /ɑ/ vowel may occur where other speakers have /ɔ/. The variants of /ɑ/ and /ɔ/ are among the most complex pronunciation variables in AE, governed by phonetic context and by social and geographic factors, and involving the separation or nonseparation of word classes. See 3.III.4.

The principal phonetic variants of /ɔ/ are a low-back round [ɔ] and a raised variant [ɔ˄], a lower low-back round [ɒ], upgliding diphthongs [ơɔ] and [ɔᴜ], and an ingliding [ɔǝ]. [ɔ] is most common, occurring widely throughout the East and the South and in much of the North. In eastern New England both [ɒ] and [ɔ] are current; in the New York metropolitan area [ɔ˄] is common; [ɒ] is frequent in western Pennsylvania and into Ohio. The diphthongal variants occur throughout the South into Texas and are occasional in the rural West.

3.II.10 /ɑ/ father, stock, college, etc.

The principal phonetic variants are fronted [a] and [a˂], central [ɑ], and retracted [ơ]. [a] occurs most frequently from New England across the Great Lakes area to the West Coast. It is most frequent in those areas before a spelled r as in car, yard, scarf, but is also common in other phonological environments, especially among younger speakers. [a˂], even approaching [æ] for some speakers, is especially common in New England. The central [ɑ] occurs in at least some phonological contexts over most of the United States but often alternates with one or more of the other variants, usually controlled by phonetic context. The retracted variant [ơ] is frequent on the East Coast, although the controlling contexts differ from north to south. Where postvocalic /r/ is not articulated, a lengthened [ơ:] is often heard in words like car [kơ:] in metropolitan New York, Virginia and adjacent areas of Maryland and North Carolina, and coastal South Carolina and Georgia. Where postvocalic /r/ is articulated, [ɒ], a variant of /ɔ/, can also be heard, especially in the South Midland and the rural West. It is most frequent there among older and poorly educated speakers. In other phonological contexts, [ơ] also occurs frequently in western Pennsylvania and sporadically elsewhere.

3.II.11 /ǝ/ cut, sun, judge, etc.

The principal phonetic variants are [ʌ], raised [ʌ˄], ingliding [ʌǝ], and upgliding [ʌɪ]. [ʌ] prevails throughout the North and is fairly common in parts of the South. Raised and ingliding variants are also found in the South Midland and South. The upgliding [ʌɪ] is a positional variant, occurring before /š/ and /ǰ/ (as in brush, judge) in parts of the South. In unstressed position, as in the first syllable of about, the symbol [ǝ], corresponding to stressed [ʌ], is used.

3.II.12 /ɜ/ bird, sermon, further, etc.

The principal phonetic variants are a fully constricted [ɝ] (articulated with the tongue curled back), a slightly constricted [ɜ̣], an unconstricted [ɜ], and an unconstricted upgliding diphthong [ɜɪ]. The constricted variant prevails throughout most of the Inland North, the Midland, and the West. The unconstricted vowel is generally confined to eastern New England, the New York City metropolitan area, tidewater Virginia, and coastal South Carolina and Georgia (areas that lack postvocalic /r/ in general). In the areas where unconstricted [ɜ] prevails, a partially constricted variant may also occur sporadically. The upgliding diphthong [ɜɪ], which occurs in metropolitan New York and in parts of the South, is sometimes stigmatized, having been characterized in literature as a Brooklyn lower-class feature. While for most speakers it is not homophonous with [ɔɪ] (as suggested by such spellings as boid and toid for bird and third), it is sufficiently different from [ɝ] and [ɜ] to sound strange to many Americans. In unstressed position, as in father, mother, the symbols [ɚ] and [ǝ] are used, corresponding in articulation and distribution to stressed [ɝ] and [ɜ].

3.II.13 /aɪ/ nice, ride, wire, etc.

The diphthongs typically vary by both onset and glide elements. The principal phonetic variants of /aɪ/ are upgliding [aɪ], [ɑɪ], [ơɪ], and [ʌɪ] (for onset varieties) and ingliding [aǝ] and [ɑǝ] (for glide varieties, also called “weakened diphthongs”). The lengthened monophthong [a:] (sometimes called a “flattened” diphthong) also occurs for /aɪ/. Both [aɪ] and [ɑɪ] (particularly the former) are widespread, occurring across the United States. [ơɪ] occurs most frequently on the East Coast, especially in Philadelphia and eastern North Carolina but occasionally elsewhere. The “fast” diphthong [ʌɪ] (starting from a more central position, so having a shorter glide) is most common on the East Coast, and is occasional along the Canadian border. It is especially frequent among conservative speakers in northeastern New England, tidewater Virginia and adjoining parts of Maryland and North Carolina, and coastal areas of South Carolina, Georgia, and Florida. There are scattered instances in the Gulf states. This variant occurs most often before voiceless consonants. The ingliding variants are more widespread, occurring especially frequently in the South and South Midland, and also in the lower Midwest and the Plains states as well as the Rocky Mountain area. Monophthongal [a:] is heard most frequently throughout the South and South Midland west to New Mexico, but is not restricted to that area. It occurs most often before /r/ and /l/ (as in wire, mile), before other voiced consonants (as in wide, five), and in final position (as in high, my, try).

3.II.14 /aᴜ/ about, loud, flour, etc.

The principal phonetic variants are [aᴜ], [ɑᴜ], [æᴜ], and [ʌᴜ]; additionally, diphthongs with fronted [ᵾ] as the second element are heard, and monophthongal [a:] also occurs. The major variant is [aᴜ], with [æᴜ] occurring next most frequently and probably increasing among young speakers. [ɑᴜ] is heard most frequently in the Upper Midwest and in other places that have had substantial Germanic-language settlement. Centralized [ʌᴜ] occurs most often among conservative speakers on the East Coast, and is especially frequent (occurring in all social groups) in tidewater Virginia. It is also an occasional variant in the North near the Canadian border. In the Southeast, diphthongs having centralized [ᵾ] as the second element are also heard. Flattened or monophthongal [a:] occurs, especially in parts of the Midland and Midwest; it is particularly common before /r/ and /l/.

3.II.15 /ɔɪ/ boy, moist, soil, etc.

The principal phonetic variants are [ɔɪ], [oɪ], [ɒɪ], [ɔǝ], and [ɔ:]. The major variant is [ɔɪ], occurring throughout the country. [oɪ], having a higher onset element, occurs most frequently on the East Coast, especially in the Philadelphia area, and is occasional in the South. The diphthong [ɒɪ], with a low onset vowel, is a conservative variant, heard especially among older speakers and limited in its word incidence. Ingliding [ɔǝ] and monophthongal [ɔ:] occur frequently in the South and Midland, being especially common in contexts preceding /l/, as in oil [ɔ:l] or [ɔǝl].

3.II.16 The consonants are on the whole less variable, at least in geographically generalizable ways, than the vowels. Their variability often is limited to specific words or tends to be stylistic rather than regional. Their influence on contiguous vowels, however, is great, as can be seen in the preceding sections. Two consonants, /l/ and /r/, have particularly strong influence over nearby vowels and some consonants and have some vowel-like characteristics themselves. Because they are very complex, they are treated in greater detail than the other consonants, but even so the descriptions outline only their major characteristics. For each sound unit, key words illustrating its occurrence in initial, medial, and final position are provided.

3.II.17 /p/ poor, spring, sleep, etc.

The principal phonetic variants are [p], aspirated [p‘], unreleased [p¯], and partially voiced [p̬]. In initial position, the aspirated variant is usual, except when followed by /r/ or /l/ (as in preach, place), in which case [p] often occurs. As the second element in a prevocalic consonant cluster (as in spring, spigot), [p] is usual. In final position, [p] is most common, but it alternates frequently with unreleased [p¯]. [p̬] also occurs infrequently, most often in intervocalic position, as in stopper, popple.

3.II.18 /b/ birds, labs, job, etc.

The principal phonetic variant is [b]. Occasionally [b̥], a variant without full voicing, occurs, especially in final position and especially in areas that have had dense foreign (particularly German) settlement. A third variant, [β], articulated with some friction rather than complete stoppage, occurs infrequently; it is most likely in such words as February, seven, and eleven, where /b/ and /v/ alternate for some speakers (see 3.I.17).

3.II.19 /t/ tap, stop, pet, etc.

The principal phonetic variants are [t], [t‘], [t¯], and [t̬] (whose occurrences parallel those of the corresponding variants of /p/), and a flapped [ɾ], a dentalized [t̯], and a glottal stop [ʔ]. The voiced variant and flap usually occur intervocalically, especially in rapid speech in words such as water, city, Saturday. The dental stop, which occurs most frequently as the initial sound of an unstressed syllable (as in hunter [ˈhʌnt̯ǝ(r)]) has widest occurrence in the New York metropolitan area. The glottal stop, heard in words like mountain [ˈmaᴜnʔṇ], bottle [ˈbɑʔḷ], and eighteen[ˈeɪʔˈtɪin], is usually in syllable-final but not word-final position. It occurs sporadically throughout the country, with somewhat greater frequency in urban areas such as Boston, New York, and New Orleans.

3.II.20 /d/ dog, pads, lad, etc.

The principal phonetic variant is [d]. A partially devoiced variant [d̥] also occurs occasionally; like [b̥], it occurs especially in areas of heavy foreign-language settlement.

3.II.21 /k/ cat, sketch, talk, etc.

The principal phonetic variants are [k], [k‘], and [k¯] (whose occurrences parallel those of the corresponding variants of /p/ and /t/), and [x], articulated with velar friction. The [x] variant usually occurs in final position and is most common in areas of Germanic settlement.

3.II.22 /g/ get, dragged, dog, etc.

The principal phonetic variant is [g]. A partially devoiced variant [g̥] also occurs occasionally and, like [b̥] and [d̥], is most common in areas of heavy foreign-language settlement.

3.II.23

The fricatives and affricates have little variation, usually being articulated [f], [v], [θ], [ð], [s], [z], [š], [ž], [č], [ǰ]. The primary exceptions are that the voiceless sounds can be articulated with partial voicing, [f̬], [θ̬], [s̬], [š̬], [č̬], and each of the voiced sounds has a partially devoiced variant, [v̥], [ð̥], [z̥], [ž̥], [ǰ̥], that occurs most often in areas of heavy German settlement. In addition, a fronted variant [š˂] occasionally occurs for /š/, particularly before front vowels and before /r/ as in shrink and shrimp. The /ž/ sound has a unique limitation: it occurs regularly in medial and final position but occurs in initial position only in proper names and loanwords (Zsa Zsa, genre).

3.II.24

/m/ mad, bump, rum

/n/ now, dent, ran

/ŋ/ skunk, spring, etc.

The principal articulations of the nasal consonants are [m], [n], [ŋ]. Each can occur as the syllabic nucleus, as in bottom, button, Washington, and in postvocalic position each can be assimilated, resulting in the nasalization of the preceding vowel, as in jumping [ˈǰʌ̃pɪŋ], mantle [mæ̃tḷ], skunk [skʌ̃k]. The velar nasal /ŋ/ occurs only in medial and final positions.

3.II.25

/h/ high

/w/ wide, white, twenty

/j/ yesterday, funeral, etc.

These consonants have very little variation in articulation, being pronounced [h], [w], [j] in most instances.

3.II.26

/r/ run, pretty, spring, three, farther, car, etc.

/ə̯/ farther, car, fort, ear, mare, etc.

In initial and medial positions, as in rice, spruce, /r/ is usually articulated [r], an alveolar frictionless continuant. In medial position following /θ/, however, as in three or thrift, a flapped variant [ɾ] also occurs. It is in postvocalic position, as in car, fort, farm, ear, that /r/ has significant variation in articulation. It can be pronounced, like [r], with full retroflexion or turning back of the tongue tip, as [ɚ]; with only partial retroflexion [ɜ̣]; or with no retroflexion at all, in which case it sounds like the weakly stressed vowel /ǝ/ but is sometimes considered a vowel-like unsyllabic consonant /ə̯/ (articulated [ə̯]). Alternatively, if there is no retroflexion, the preceding vowel may simply be lengthened. In general, the West, Southwest, Inland North, and Midland have full retroflexion. There are a few pockets in those areas, however, where variants with partial retroflexion can be heard, as in San Francisco and Seattle (among conservative speakers) and sporadically in the Upper Midwest. Most of the Atlantic Coast and all of the South are the regions where postvocalic /r/ is most regularly “dropped” or pronounced without retroflexion, though the feature also occurs elsewhere, particularly in northern urban areas where large numbers of Blacks have moved. However, even in the speech of a single individual, this articulation is not necessarily consistent, and lack of retroflexion can alternate with full or partial retroflexion. When postvocalic /r/ is also in word-final position (as in ear, hair, door), speakers in those regions which regularly “drop” the /r/ usually have articulations such as [ɪiə̯], [hεə̯], [doə̯]. In the South and among some Blacks elsewhere, however, lengthening of the preceding vowel may occur instead. This is a conservative feature, which is sometimes socially stigmatized. In both the Northeast and the South, young speakers appear to have less r-dropping than older speakers, indicating that fuller retroflexion may gradually spread through areas where /r/ is now regularly lost.

In the Northeast, where loss of postvocalic /r/ is the norm, when final /r/ is followed by a word beginning with a vowel, as in the phrase mother and, full retroflexion of /r/ usually occurs. In the South, this “linking” /r/ may also be heard but is less frequent than in the Northeast. It is probably the analogical source of the intrusive /r/ that occurs in phrases like idea of, which becomes [aɪˈdiɚˌǝv]; this occurs most often in the Northeast.

Intervocalic /r/ exhibits a pattern related to that of postvocalic /r/. In words like hurry, hairy, marry, horrible, and during, some speakers have what could be described as two r’s, while others have only one. That is, the first type of speaker ends the first syllable with a postvocalic /r/ and begins the second syllable with an initial /r/, as in hurry [ˈhɝrɪ]. The second type ends the first syllable with a vowel, with /r/ occurring only in the second syllable, as [ˈhʌrɪ]. The “one r” pronunciation is most frequent along the East Coast and in the South; the “two r” pronunciation dominates through most of the rest of the country, though the Great Lakes region and the far West show mixed usage.

The influence of /r/ on preceding vowels is very strong, tending to pull vowels to the center of the mouth. Thus the “two r” speakers have neutralized some vowel distinctions kept by other speakers; for example, “two r” speakers usually have the pronunciation [ˈhεɚrɪ] for both Harry and hairy, while “one r” speakers distinguish between [ˈhærɪ] and [ˈhεrɪ]. Similarly, in words like forest and horrible, many “two r” speakers maintain the historical low-back /ɔ/, [ˈfɔɚrᵻst], [ˈhɔɚrᵻˌbǝl], while others have fronted, lowered, and unrounded the vowel to /ɑ/, [ˈfɑrᵻst], [ˈhɑrᵻbǝl].

3.II.27 /l/ light, blue, milk, ankle, fool, etc.

The principal variants are [l] (articulated near the center of the mouth), [ļ] (a “light” variant pronounced toward the front of the mouth), and [ƚ] (a “dark” variant articulated toward the back of the mouth). The light /l/ usually occurs in contiguity with front vowels or with bilabial or labiodental consonants (as in leaf, place, flee), while the dark /l/ is usually contiguous with a back vowel or a velar consonant (as in bowl, gloom, ankle). Occasionally [ƚ] occurs initially, as in love [ƚʌv], particularly in the Northeast and most especially in the Philadelphia area. Additionally, a completely vocalized variant [ɯ], which sounds very much like the vowel [ᴜ] or the consonant /w/, sometimes occurs in words such as milk [mɪɯk], twelve [twεɯv], and elbow [ˈεɯboᴜ]. This vocalization of /l/, which occurs most often in the South and the upper Ohio Valley, is analogous to the articulation of postvocalic /r/ in words such as fire [faɪə̯] and poor [pᴜə̯]. And just as /r/ can be “lost” in words such as far [fơ:], /l/ can be lost in such words as help [hε:p] or [hεǝp] and wolf [wᴜ:f]. In both cases, the loss is usually compensated for by a lengthened or ingliding vowel. The loss of /l/ is less frequent than that of /r/. It seems to occur most often in the South and among Blacks, but its regional and social patterns are less distinct than those for the loss of /r/, and its word incidence is less regular.

Intervocalic /l/, like intervocalic /r/, has two usage types. The first type closes one syllable with [ƚ] and begins the next with [l], as in cooler [ˈkuƚlɚ]. The second type ends the first syllable with a vowel and begins the next with [l], [ˈkulɚ]. In general, the East Coast, South, South Midland, and interior West have the “one l” usage, the interior Northeast and the North Central states have the “two l” type, and the coastal West has mixed usage.

Like /r/, /l/ has a strong influence on preceding vowels. Diphthongs are “flattened” or monophthongized more frequently before /l/, as in file [fa:l], than elsewhere. Lax vowels tend to alternate with tense, as when steel, pail, fool are pronounced [stɪl], [pεl], [fᴜl] rather than [stil], [pel], [ful]. (This pattern is most common in the West, but is not restricted to that area.) Additionally, lower lax vowels alternate with higher lax vowels before /l/, as when milk is pronounced [mεlk] and Philadelphia becomes [ˌfɪlǝˈdælfɪǝ].

3.III Summary of Major Regional Variants

This section takes some of the more significant pronunciation variations and summarizes them in terms of their regional and social distributions.

3.III.1 Presence or absence of postvocalic /r/.

Probably the most readily noticed feature of variation in AE, this feature is nevertheless a linguistically and socially complex matter. Most speakers of AE have a retroflex /r/ (articulated with the tip of the tongue turned back) following vowels in the same syllable, as in car, floor, farm, work, course. Most network radio and television professionals also have this feature. In most of the South and along much of the East Coast, however, postvocalic /r/ is not articulated as /r/ but rather as a vowel-like unsyllabic /ə̯/. But within these broad areas there is significant variation in practice; not all speakers “drop” the /r/, nor do all speakers follow the same practices in dropping it. Variation within these regions is affected by age, social class, and word incidence, and pronunciation within the speech of particular individuals also varies. Map 1 illustrates the areas (1) where medial postvocalic /r/ (as in course, work) is regularly dropped, and (2) where medial postvocalic /r/ is weakened, or articulated with less than full retroflexion of the tongue.

3.III.2 Weakened variants of diphthongs.

The “wide” diphthongs, /aɪ/ (bite), /aᴜ/ (bout), /ɔɪ/ (boy), all have varying degrees of weakening of the second element, to the point of becoming “flattened” diphthongs (that is, lengthened monophthongs). Thus, the sound unit /aɪ/ may be articulated as [aɪ] (the full diphthongal upglide), [a^ɪ] (a weakened upglide), [aǝ] (an inglide), or [a:] (a flattened variant). The diphthongs /aᴜ/ and /ɔɪ/ have corresponding variants. Some phonological environments encourage weakening more than others: voiced consonants following the diphthong are more likely to promote weakening than are voiceless consonants, and /r/ and /l/ encourage it more than other voiced consonants. Further, different areas of the country (and different social groups within the areas) have weakened variants of the diphthongs under different conditions. In final position (as in by, high, try), the South, South Midland, and part of the Southwest frequently have both weakened and flattened variants. The Plains states, Rocky Mountain area, and southern parts of the Midwest have the weakened variants [a^ɪ] and [aǝ], most frequently before voiced consonants, especially /r/ and /l/. The South tends to have monophthongal variants of /aɪ/ before voiced consonants; the South Midland has them in even more environments. In other parts of the country, if /aɪ/ is weakened at all it is most likely to be so before /l/.

The /aᴜ/ diphthong is less likely to have monophthongal variants than /aɪ/. However, it too is most likely to weaken before /r/ and /l/, and before voiced rather than voiceless consonants. The geographic area of most frequent occurrence includes northern Maryland and southern and western Pennsylvania, but extends basically from the upper Ohio Valley southward into the Southwest (especially before /r/ and /l/), with occasional usage elsewhere in the East.

Although /ɔɪ/ is less consistently weakened than the other two diphthongs, weakened or flattened variants (such as [ɔǝl] or [ɔ:l] for oil) occur across the South and South Midland into the Southwest.

Map 2 shows the approximate areas of (1) weakening and flattening of /aɪ/ in word-final position (as in by, high), and (2) weakening of /aɪ/ before a final voiced consonant (as in ride, mile, fire).

3.III.3 Diphthongized variants of monophthongs.

Sound units that are usually monophthongs may be articulated as diphthongs in any of several patterns. The most common is the addition of the inglide, as in pit [pɪǝt], bell [bεǝl]. This is most frequent in the South Midland and adjacent areas and in the South, but it does occur occasionally in western New England and westward in a slightly “faster” version (that is, with shorter and weaker articulation of the inglide). Parts of the South and the South Midland and adjacent areas also have upgliding variants of /æ/, /ε/, and /ǝ/, especially before /š/, as in ash[æɪš], mesh [mεɪš], and mush [mʌɪš]. The same upglides occur in some parts of the South and South Midland in other phonological environments as well, as in bag [bæɪg], egg [εɪg]. Also in the South Midland and in adjacent western areas, less frequently in the South, and as far north as Philadelphia in the East, /ɔ/ has diphthongized variants [ơɔ] and [ɔᴜ], as in dog [dơɔg], [dɔᴜg], loft [lơɔft], [lɔᴜft], August [ˈơɔgǝst], [ˈɔᴜgǝst]. Map 3 shows the area of most frequent diphthongization of one vowel, /ɔ/, before voiceless consonants (as in loft).

3.III.4 Alternations among vowels.

Some systematic alternations among vowel sets can be described in terms of tongue height and front-to-back articulation in the mouth. That is, if two vowels can both occur in a particular environment, one part of the country may regularly favor the lower over the higher vowel, while in another context speakers of one region may use a front vowel while others prefer a back vowel. Lower vowels are most frequent before /r/ and /l/, though they can also occur in other contexts: the articulation of Coors (brand name) as /kɜz/ ([kɝz]) rather than /kᴜrz/ and of there as /ðɑr/ rather than /ðɛr/ is heard most often in the South Midland and adjacent western areas; milk and pillow become /mεlk/ and /ˈpεlo/ most often in the Great Lakes region; the South and parts of the West are most likely to have lowered vowels in feel, sale, and school, yielding /fɪl/, /sεl/, /skᴜl/; the lowering of /ɪ/ to /ε/ or even to /æ/, as in thing /θεŋ, θæŋ/, occurs most often in the South Midland but also sporadically in the South and the West.

The use of a higher vowel than the norm occurs most frequently in the South Midland and parts of the South and rural West before nasal consonants, as in pen /pɪn/, hem /hɪm/, and length/lɪŋθ/. A similar alternation, that of /ε/ for /æ/ as in bat, occurs especially in the Northeast and the South. Before /r/, where the vowels /i~ɪ/, /e~ε/, /u~ᴜ/, /o~ɔ/ alternate in pairs, the higher vowel occurs most frequently in the South while the lower vowel is generally heard more often in the North. Much of the West has the lower vowel, but both can be heard frequently.

Variation between back vowels and more centralized vowels is most obvious in AE in the occurrence of /ɔ/ and /ɑ/. The low vowels as a group are highly variable in terms of their basic structure, their phonetic conditioning, social articulation, and the word sets that manifest a given sound, and the distinction between “short o” words like cot, tot, and rot, and au or ou words like caught, taught, and wrought is especially problematic. In general, AE has tended to front, lower, and unround the vowel in historical “short o” words from [ɔ] to [ɑ]. However, for some speakers, especially those in the northern tier of states, the fronting has gone to [a], even approaching [æ] in some cases. For others, the shift has been to a lower, rounded variant [ɒ]. This last occurs in eastern New England, western Pennsylvania, occasionally on the southern coast, and sporadically across the country into the Rocky Mountains and the Southwest. Before /r/, /ɔ/ is general (except for frequent /o/ forms in the South). However, some speakers have /ɑ/ here, too: the East Coast and the South have [ɑ] in words like forest [ˈfɑrᵻst], horror [ˈhɑrǝ], while some historically related but noncontiguous areas in Delaware, Maryland, North Carolina, southern Missouri, west Texas, and rural Utah have [ɒ] or [ɑ] in monosyllabic words (such as fork, horse, born) as well. The other environments are highly variable, even for the individual speaker.

A similar process of fronting of the vowel appears to be taking place for words like taught, sought, wrought. The South and much of the North generally maintain /ɔ/ in these words, so that there is a distinction between caught /kɔt/ and cot /kɑt/. In much of the West and in parts of the Southwest and Midwest, however, /ɑ/ occurs in caught, with the result that caught and cot sound alike. This is especially prevalent among younger speakers. In eastern New England and western Pennsylvania the vowel of caught is articulated [ɒ], but since the vowel of “short o” words is also [ɒ], the words sound alike in these regions too. Thus cot–caught word pairs may have contrasting low vowels, may have a single low vowel, or may vary by word or word subsets. Map 4 shows the general area where the distinction between /ɔ/ and /ɑ/ is not maintained, that is, where cot and caught sound alike.

3.III.5 Strictly speaking, each feature of pronunciation has a unique pattern even when considered in a general way, that is, when minimizing complexities of social groups, geographical discontinuities, and phonological environments. Some features, however, have similar enough distributions that, taken together, they form a recognizable pattern of overlap. Each of the maps in this section represents such a pattern. It is important to remember that the “boundaries” for each feature are necessarily fuzzy, and that individual speakers within each area may vary in their use of particular features that make up the pattern. Furthermore, since language is always changing, the patterns must be seen as representing the situation at a particular time and not as a static picture of what will always be.