Phonological history of English short A

The pronunciation of "short A" varies in English.

The development of the Early Modern English phoneme /aː/

Context

Late Middle English had two phonemes /a/ and /aː/, differing only in length. /a/ ("short A") was found in words such as cat [kat] or trap [trap], and also before /r/ in words such as start [start]. /aː/ ("long A") was found in words such as face [faːs], and before /r/ in words such as scare [skaːr].

As a result of the Great Vowel Shift, the "long A" [aː] of face was raised, initially to [æː] and later to [ɛː]. This process began in the 15th century. [æː] "seems to have been the normal pronunciation in careful speech before 1650, and [ɛː] after 1650". In a separate development, the [a] of trap was later fronted to [æ] (the value it retains in many accents today). This fronting was mostly confined to "vulgar or popular" speech in the 16th century, but gradually replaced the more conservative [a] in the 17th century, and was "generally accepted by careful speakers by about 1670". In recent years however, /æ/ in Received Pronunciation has lowered to a fully open [a].

These trends, allowed to operate unrestrictedly, would have left standard English without any vowels in the [a] or [aː] area by the late 17th century. However, this putative gap was filled by the following special developments:

• In two environments, Middle English [a] developed to [aː] rather than [æ]
  • Before non-prevocalic /r/ (e.g. in start, star; but not in carry), [a] developed to [aː] in all words
  • Before some fricatives, broadening happened inconsistently and sporadically
• Words that had Middle English [au] had a regular development to [ɒː] (for example, paw). However, before a nasal, such words sometimes instead developed to [aː] (e.g. palm).

The [aː] of the late 17th century has generally backed to [ɑː] in several varieties of contemporary English, for example in Received Pronunciation.

The following table shows some developments of Middle English /a/ in Received Pronunciation. The word gate, which derived from Middle English /aː/, has also been included for comparison.

		gate	cast	cart	cat	glad
Middle English		[ɡaːt]	[kast]	[kart]	[kat]	[ɡlad]
Great Vowel Shift	Phase 1	[ɡæːt]
	Phase 2	[ɡɛːt]
	Phase 3	[ɡeːt]
	Phase 4	[ɡeɪt]
Lengthening before /r/				[kaːrt]
Lengthening before /f,θ,s/			[kaːst]
Fronting of /a/					[kæt]	[ɡlæd]
Backing of /aː/			[kɑːst]	[kɑːrt]
/r/-dropping				[kɑːt]
bad-lad split						[ɡlæːd]
Lowering of /æ/					[kat]	[ɡla(ː)d]

The table below shows the results of these developments in some contemporary varieties of English:

		RP	NE	SCO	IRL	GA
Lengthening before /r/		✔	✔			✔
Lengthening before /f,θ,s/		✔
Fronting of /a/		✔			variable	✔
Backing of /aː/		✔			variable	✔
R-dropping		✔	✔
bad-lad split		✔
Lowering of /æ/		✔
Output for	gate	[ɡeɪt]	[ɡeːt]	[ɡet]	[ɡeːθ̠]	[ɡeɪt]
	cast	[kʰɑːst]	[kʰast]	[kʰast]	[kʰæst~kʰast]	[kʰæst]
	cart	[kʰɑːt]	[kʰaːt]	[kʰart]	[kʰaːɹθ̠~kʰɑːɹθ̠]	[kʰɑɹt]
	cat	[kʰat]	[kʰat]	[kʰat]	[kʰæθ̠~kʰaθ̠]	[kʰæt]
	glad	[ɡla(ː)d]	[ɡlad]	[ɡlad]	[ɡlæd~ɡlad]	[ɡlæd]
	gas	[ɡas]	[ɡas]	[ɡas]	[ɡæs~ɡas]	[ɡæs]

Development of short /a/ when not before /r/

In late Middle English, pairs such as cat, cart, were pronounced [kat], [kart] respectively, distinguished only by the presence or absence of [r]. However, by the late 17th century they were also distinguished by the quality and length of the vowel. In cat, the vowel had been fronted to /kæt/, while in cart it had been lengthened to /kaːrt/. This is the result of the development of Middle English /a/ in the environment of a following non-prevocalic /r/. It seems to have first occurred in the dialects of southern England in the early 15th century, but did not affect Standard English until the later 17th century. It has affected most varieties of contemporary English, which have distinct vowels in pairs such as cat, cart, although the original identity of the vowel is preserved in Irish English: [kʰæθ̠] and [kʰaːɹθ̠] or [kʰɑːɹθ̠].

Before intervocalic /r/, broadening did not generally take place: the vowel of carrot /kærət/ remained the same as that of cat: this is preserved in most modern varieties (but see the Mary–marry–merry merger).

Development before fricatives

Unlike lengthening before nonprevocalic /r/, which applied universally in Standard English, lengthening, or broadening, before fricatives was inconsistent and sporadic. This seems to have first occurred in the dialects of Southern England between about 1500 and 1650. It penetrated into Standard English from these dialects around the mid-17th century.

The primary environment which favored broadening was before preconsonantal or morpheme-final voiceless fricatives /f, θ, s/. The voiceless fricative /ʃ/ has never promoted broadening in Standard English in words like ash and crash. There is, however, evidence that such broadening did occur in dialects.

Once broadening affected a particular word, it tended also to affect its inflectional derivatives. For example, from pass ([pʰaːs]) there was also passing [ˈpʰaːsɪŋ]. This introduced broadening into the environment _sV, from which it was otherwise excluded (compare passage which is derived not from the English word pass but separately from French, and was never affected by broadening).

In a phenomenon going back to Middle English, [f, θ] alternate with their voiced equivalents [v, ð]. For example, late Middle English path [paθ] alternated with paths [paðz]. When broadening applied to words such as path, it naturally extended to these derivatives: thus when [pʰaθ] broadened to [pʰaːθ], [pʰaðz] also broadened to [pʰaːðz]. This introduced broadening into the environment before a voiced fricative.

Broadening affected Standard English extremely inconsistently. It seems to have been favored when /a/ was adjacent to labial consonants or /r/. It is apparent that it occurred most commonly in short words, especially monosyllables, that were common and well-established in English at the time broadening took place (c. 1500–1650). Words of 3 or more syllables were hardly ever subject to broadening. Learned words, neologisms (such as gas, first found in the late 17th century), and Latinate or Greek borrowings were rarely broadened.

A particularly interesting case is that of the word father. In late Middle English this was generally pronounced [ˈfaðər], thus rhyming with gather [ˈɡaðər]. Broadening of father is notable both in two respects:

• its occurrence before an intervocalic voiced fricative [ð]
• its distribution in many accents that do not otherwise have broadening, such as those of North America.

The Oxford English Dictionary describes the broadening of father as "anomalous". Dobson, however, sees broadening in father as due to the influence of the adjacent /f/ and /r/ combined. Rather and lather appear to have been subject to broadening later, and in fewer varieties of English, by analogy with father.

The table below represents the results of broadening before fricatives in contemporary Received Pronunciation.

Environment	RP /æ/ as in TRAP ("flat A")	RP /ɑː/ as in PALM or FAther ("broad A")
_[f]$	carafe*, chiffchaff, gaffe, naff, riffraff	calf**, chaff*, giraffe, graph (telegraph, see above), half**, laugh**, staff
_[f]C	Daphne, hermaphrodite, kaftan, naphtha	aft, after, craft, daft, draft/draught**, graft, laughter**, raft, rafter, shaft
_[θ]$	hath, math (abbrev. for mathematics)	bath, lath*, path
_[θ]C	athlete, decathlon (pentathlon, biathlon, etc.), maths
_[s]$	alas*, ass (donkey), ass (term of abuse)*, crass, gas, lass, mass (amount), Mass (religious service)*	brass, class, glass, grass, pass
_[sp]	asp, aspect, aspen, aspic (jelly), aspirant, aspirin, Diaspora, exasperate*, jasper	clasp, gasp, grasp, hasp*, rasp
_[st]	aster, asteroid, astronaut (astronomical, etc.), bastion, blastocyst (blastopore, etc.), canasta, castanets, chastity, elastic*, fantastic, gastric, gymnastic, hast, Jocasta, mastic, masticate, mastiff*, mastitis, mastoid, mastodon, masturbate*, monastic, onomastic, pasta, pastel, plastic*, procrastinate, Rastafarian, raster, sarcastic, scholastic, spastic	aghast, avast, bastard*, blast, cast, caster, fast, ghastly, last, mast, master, nasty, past, pasteurize*, pastime, pastor, pastoral*, pasture, plaster, repast, vast
_[sk]	Alaska, Basque*, emasculate, gasket, Madagascar, mascot, masculine, masquerade*, Nebraska, paschal*, vascular	ask, bask, basket, cask, casket, flask, mask, masque*, rascal, task
_[sf]	blasphemy*
_[ð]	blather, fathom, gather, slather	father, lather*, rather
other (see below)		calve**, castle, fasten, halve**, raspberry

• * indicates that the other pronunciation is also current in RP.
• ** indicates that this word had late Middle English /au/ (possibly in addition to late Middle English /a/)
• Words in italics were first recorded by the Oxford English Dictionary later than 1650

In general, all these words, to the extent that they existed in Middle English, had /a/ ("short A" as in trap) which was broadened to [aː]. The exceptions are:

• half and calf, which had been pronounced with [half, kalf] in early Middle English before developing around the early 15th century to [hauf, kauf] by L-vocalization. In accents of England the development was subsequently the same as that in words such as palm (see below). The North American development to [æ] as in trap seems to be the result of shortening from [hauf, kauf] to [haf, kaf], although there is little evidence of this development.
• laugh, laughter and draft/draught, which all had [auχ] in Middle English. This first changed to [auf] (accepted in Standard English from about 1625, but earlier in dialects), and was then shortened to [af]. The subsequent development was similar to other words with [af], such as staff. The development of draft/draught is notable: in the 17th century it was usually spelled draught and pronounced to rhyme with caught, making clear its derivation from the verb to draw. The pronunciation with [f] was rare, and its use in current English is a historical accident resulting, according to Dobson, from the establishment of the spelling variant draft.

The words castle, fasten and raspberry are special cases where subsequent sound changes have altered the conditions initially responsible for lengthening. In castle and fasten, the /t/ was pronounced, according to a slight majority of 16th and 17th century sources. In raspberry we find /s/ rather than /z/.

The pattern of lengthening shown here for Received Pronunciation is generally found in southern England, the Caribbean, and the Southern hemisphere (parts of Australia, New Zealand and South Africa). In North America, with the possible exception of older Boston accents, broadening is found only in father (the success of broadening in this word alone in North America unexplained) and pasta (which follows the general pattern for recent Italian loanwords, cf. mafia). In the Boston area there has historically been a tendency to copy RP lengthening which perhaps reached its zenith in the 1930s but has since receded in the face of general North American norms.

In Irish English broadening is found only in father (which may, however, also have the FACE vowel). In Scottish and Ulster English the great majority of speakers have no distinction between TRAP and PALM (the Sam–psalm merger). In Welsh English Wells finds broadening generally only in father, with some variation. In the north of England, broadening is found only in father and usually half and master.

Words with /au~a/ followed by a nasal in Middle English

There was a class of Middle English words in which /au/ varied with /a/ before a nasal. These are nearly all loanwords from French in which uncertainty about how to realize the nasalization of French resulted in two varying pronunciations in English. (One might compare the different ways in which modern French loanwords like envelope or envoy are pronounced in contemporary varieties of English).

Words with Middle English with the vowel /au/ generally developed to [ɒː] in Early Modern English (e.g. paw, daughter). However, in some of the French loanwords, especially short words in common use, the /au/ instead developed to /aː/. As this development preceded the Great Vowel Shift, the resulting Early Modern English sound was [eː] rather than [aː], as in change. In the table below, these words are classified according to the lexical sets of John Wells.

Environment	TRAP lexical set	BATH lexical set	PALM lexical set	THOUGHT lexical set	FACE lexical set
_[m]$			alms, balm, calm, palm, psalm, qualm	shawm
_[mp]	champion, rampant, stamp*	example, sample
_[mb]	amber				chamber
_[mf]	pamphlet
_[nt]	ant*, lantern, phantom, rant, scant	advantage, aunt, can't, chant, grant, plant, slant, vantage		daunt, flaunt, gaunt*, gauntlet, haunt, jaunt?, saunter, taunt, vaunt
_[nd]	abandon, grand, random	command, demand, Flanders, remand, reprimand, slander		jaundice, laundry, Maundy
_[n(t)ʃ]	franchise	avalanche, blanch, branch, ranch, stanch, stanchion		haunch, launch, paunch, staunch
_[n(d)ʒ]	evangelist, phalange				angel, arrange, change, danger, grange, mange, range, strange
_[ŋk]	bank ("bench", "financial institution"), canker, flank, plank, ranco(u)r, sanctity
_[ŋɡ]	anger*, angle, strangle
_[ns]	ancestor, finance, ransom, romance	answer*, chance, chancellor, dance, enhance, France, lance, lancet, prance, stance, trance, transfer (trans-),		launce	ancient
Other	salmon	almond

* Not a French loanword

In some cases, both the /a/ and the /au/ forms have survived into modern English. For example, from Sandre, a Norman French form of the name Alexander, the modern English surnames Sanders and Saunders are both derived.

It is apparent that the development to the FACE lexical set (contemporary /eɪ/) was particularly common before /ndʒ/, whereas before velars only the development to the TRAP lexical set (contemporary /æ/) is found.

While these words were generally spelled with both ⟨a⟩ and ⟨au⟩ in Middle English, the current English spelling generally reflects the pronunciation, with ⟨au⟩ used only in words of the THOUGHT lexical set; one common exception is aunt.

Trap–bath split

The trap–bath split is a vowel split that occurs mainly in the southern and mainstream varieties of English in England (including Received Pronunciation), in the Southern Hemisphere accents of English (Australian English, New Zealand English, South African English), and also to a lesser extent in older Boston English, by which the Early Modern English phoneme /æ/ was lengthened in certain environments and ultimately merged with the long /ɑː/ of father. (Wells 1982: 100–1, 134, 232–33)

In this context, the lengthened vowel in words such as bath, laugh, grass, chance in accents affected by the split is referred to as a broad A (also, in the UK, long A). Phonetically, the vowel is [ɑː] in Received Pronunciation (RP); in some other accents, including many Australian and all New Zealand accents, it is a fronter vowel ([ɐː] or [aː]), and it may be a rounded [ɒː] in South African English. In accents unaffected by the split, these words usually have the same vowel as words like cat, trap, man, the short A or flat A.

The sound change originally occurred in southern England, and ultimately changed the sound of [æ] to [ɑː] in some words in which the former sound appeared before [f, s, θ, ns, nt, ntʃ, mpl], leading to RP [pɑːθ] for path and [sɑːmpl] for sample, etc. The sound change did not occur before other consonants; thus accents affected by the split preserve /æ/ in words like cat. (See the Variations section below for more details on the words affected.) The lengthening of the bath vowel began in the 17th century but was "stigmatised as a Cockneyism until well into the 19th century".

British accents

The presence or absence of this split is one of the most noticeable differences between different accents of English English. An isogloss runs across the Midlands from the Wash to the Welsh border, passing to the south of the cities of Birmingham and Leicester. North of the isogloss, the vowel in most of the affected words is usually the same short-a as in cat; south of the isogloss, the vowel in the affected words is generally long. (Gupta 2005)

There is some variation close to the isogloss; for example in the dialect of Birmingham (the so-called "Brummie") most of the affected words have a short-a, but aunt and laugh usually have long vowels. Additionally, some words which have /æ/ in most forms of American English, including half, calf, rather and can't, are usually found with long vowels in the Midlands and Northern England.

In northern English dialects, the short A is phonetically [a ~ a̠], while the broad A varies from [ɑː] to [aː]; for some speakers, the two vowels may be identical in quality, differing only in length ([a] vs [aː]) (Wells 1982: 356, 360). John Wells has claimed that Northerners who have high social status may have a TRAP-BATH split (Wells 1982: 134) and has posted on his blog that he grew up with the split in Upholland, Lancashire. AF Gupta's study of students at the University of Leeds found that (on splitting the country in two halves) 93% of northerners used [a] in the word "bath" and 96% of southerners used [ɑː]. However, there are areas of the Midlands where the two variants co-exist and, once these are excluded, there were very few individuals in the north who had a trap-bath split (or in the south who did not have the split). Gupta wrote, "There is no justification for the claims by Wells and Mugglestone that this is a sociolinguistic variable in the north, though it is a sociolinguistic variable on the areas on the border [the isogloss between north and south]".

In some West Country accents of English English where the vowel in trap is realized as [a] rather than [æ], the vowel in the bath words was lengthened to [aː] and did not merge with the /ɑː/ of father. In those accents, trap, bath and father all have distinct vowels /a/, /aː/ and /ɑː/. (Wells 1982: 346–47).

In some other West Country accents, and in many forms of Scottish English, there is no distinction corresponding to the RP distinction between /æ/ and /ɑː/.

Trap–bath split in Received Pronunciation

In Received Pronunciation (RP), the trap–bath split did not happen in all eligible words. It is hard to find a clear reason why some changed and others did not. Roughly, the more common a word the more likely that the change from flat /æ/ to broad /ɑː/ took place. It also looks as if monosyllables were more likely to change than polysyllables. The change very rarely took place in open syllables, except where closely derived from another word with /ɑː/. Thus passing is closely derived from pass, and so has broad A /pɑːsɪŋ/: passage is not so closely derived, and thus has flat A /pæsɪdʒ/. Here is the set of words that underwent transition, and counterexamples with the same environment:

RP sets for the trap–bath split

Set Broad [ɑː~äː] Flat/short [æ] /-ðər/ father, lather, rather blather, gather, slather /-f/ calf, chaff, giraffe, graph, half, laugh, photograph, staff, telegraph Aphrodite, chiffchaff, Daphne, gaff(e), graphic, staph /-ft/ aft, after, craft, daft, draft/draught, graft, haft, laughter, raft, rafter, shaft kaftan, Taft Word-final /-θ/ bath, lath, path hath, math(s), polymath Word-final /-s/ brass, class, glass, grass, pass ass (donkey), bass (fish), crass, gas, lass, mass (amount), morass, sass /-sk/ ask, bask, basket, cask, casket, flask, mask/masque, rascal, task Asquith, gasket, mascot Word-final /-sp/ clasp, gasp, grasp, hasp, rasp asp Word-final /-st/ aghast, avast, bastard, blast, cast, caste, caster, contrast, fast, ghastly, last, mast, master, nasty, past, pasteurize, pastime, pastor, pastoral, pasture, plaster, repast, vast bast, iconoclast, p(a)ederast, hast Miscellaneous /-st/ castor, disaster, master, nasty, pastor, plaster Aston, astronaut, castigate, chastity, drastic, pasta, pastel, plastic, raster /-ʃ/ m(o)ustache all other words in this set Word-final /-v/ calve, halve, salve (note: [ɑːv] only when spelled ⟨alv⟩) chav, have /-mpəl/ example, sample ample, trample /-nd/ Alexander/Alexandra/Sandra, chandler, command, demand, Flanders, remand, reprimand, slander all other words in this set /-nt/ (ad)vantage, aunt, can't, chant, Grant, grant, plant, shan't, slant ant, banter, cant, fantasy, mantle, phantom, rant, scant Word-final /-ntʃ/ all words in this set (branch, planche, ranch, etc.) /-ns/ advance, answer, chance, chancellor, dance, enhance, France, glance, lance, lancet, prance, stance, trance Anson, cancer, fancy, ransom, romance [changed] castle, fasten, raspberry

For the words in the last row, subsequent sound changes have altered the conditions initially responsible for lengthening

There are some words in which both pronunciations are heard among southern speakers:

• the words Basque, bastard, chaff, Glasgow, lather, mass (church service)
• Greek elements as in telegraph, blastocyst, chloroplast
• words with the prefix trans-

Use of broad A in mass is distinctly conservative and rare now: other international fluctuations are both common, but with further complications.

While graph, telegraph, photograph can have either (now in Received Pronunciation, they have broad A), graphic and permutations always has a flat A.

Broad A fluctuates in dialects that include it; before s it is a more common alternative when in its common voiceless variant (/s/ rather than /z/) (in transfer [tʰɹɑːnsfɜː], transport [tʰɹɑːnspɔːt] and variants) than when it is voiced (thus translate [trænzleɪt], trans-Atlantic [tʰɹænzətlæntɪk]).

Social attitudes

Some research has concluded that many people in the North of England have a dislike of the /ɑː/ vowel in BATH words. AF Gupta wrote, "Many of the northerners were noticeably hostile to /ɡrɑːs/, describing it as 'comical', 'snobbish', 'pompous' or even 'for morons'." Whilst writing on a Labovian study of speech in West Yorkshire, KM Petyt wrote in 1985 that several respondents "positively said that they did not prefer the long-vowel form or that they really detested it or even that it was incorrect". However, Joan Beal said in a 1989 review of Petyt's work that those who disliked the pronunciation still associated it with the BBC and with the sort of professional jobs that they would aspire to.

Southern Hemisphere accents

Evidence for the date of the shift comes from the Southern Hemisphere accents, those of Australia, New Zealand, and South Africa.

In Australian English, there is generally agreement with southern British in words like path, laugh, class. But before N+consonant, as in dance, plant, most Australians use a flat A; aunt and can't, however, are invariably pronounced with a broad A. Phonetically, the broad A is [äː]. In Australia there is variation in the word castle, both pronunciations are commonly heard. For more information, see the table at Variation in Australian English. In South Australian English the broad A is usually used.

South African and New Zealand English have a distribution of sounds similar to that of RP.

North American accents

Most accents of American English and Canadian English are unaffected by the split. The main exceptions are parts of New England (see Boston accent), where the broad sound can be used in some of the same words as in southern England, such as aunt, ask, bath etc. ("aunt" though is unique, as the broad a pronunciation is found sporadically throughout the U.S., not only in New England)

Related, but distinct, phenomena include the following:

• The phonemic tensing of /æ/ in the accents of New York and particularly Philadelphia, in which tensing occurs specifically before [f, s, θ, n, m] (in New York tensing occurs in more environments; see section below).
• The drawled pronunciation /æ/ → [æə] in Southern accents; many South Midland, Appalachian English, and inland Southern speakers also raise the /æ/ in aunt, dance, plant to [ɛ] or [e].

The /ɑː/ phoneme

In Modern English, a new phoneme /ɑː/ developed that did not exist in Middle English. The phoneme /ɑː/ comes from three sources: the word father lengthening from /a/ to /aː/ for an unknown reason (thus splitting from gather); the compensatory lengthening of the short /a/ in words like calm, palm, psalm when /l/ was lost in this environment; and the lengthening of /a/ before /r/ in words like car, card, hard, part, etc. In most dialects that developed the broad A class, words containing it joined this new phoneme /ɑː/ as well. The new phoneme also became common in onomatopoeic words like baa, ah, ha ha, as well as in foreign borrowed words like spa, taco, llama, drama, piranha, Bahamas, pasta, Bach, many of which vary between /ɑː/ and /æ/ among different dialects of English.

Bad–lad split

The bad–lad split is a phonemic split of the Early Modern English short vowel phoneme /æ/ into a short /æ/ and a long /æː/. This split is found in some varieties of English English and Australian English in which bad (with long [æː]) and lad (with short [æ]) do not rhyme. (Wells 1982: 288–89, 596; Horvath and Horvath 2001; Leitner 2004).

The phoneme /æ/ is usually lengthened to /æː/ when it comes before an /m/ or /n/, within the same syllable. It is furthermore lengthened in the adjectives bad, glad and mad; family also sometimes has a long vowel, regardless of whether it is pronounced as two or three syllables. Some speakers and regional varieties also use /æː/ before /ɡ/, /ŋ/, /l/ and/or /dʒ/; such lengthening may be more irregular than others. Lengthening is prohibited in the past tense of irregular verbs and function words and in modern contractions of polysyllabic words where the /æ/ was before a consonant followed by a vowel. Lengthening is not stopped by the addition of word-level suffixes.

British dialects with the bad–lad split have instead broad /ɑː/ in some words where an /m/ or /n/ follows the vowel. In this circumstance, Australian speakers usually (but not universally) use /æː/, except in the words aunt, can't and shan't, which have broad /aː/.

Daniel Jones noted for RP that some speakers had a phonemic contrast between a long and a short /æ/, which he wrote as /æː/ and /æ/, respectively. Thus, in An outline of English phonetics (1962, ninth edition, Cambridge: W. Heffer & Sons) he noted that sad, bad generally had /æː/ but lad, pad had /æ/. In his pronouncing dictionary, he recorded several minimal pairs, for example bad /bæːd/, bade /bæd/ (also pronounced /ˈbeɪd/). He noted that for some speakers, jam actually represented two different pronunciations, one pronounced /dʒæːm/ meaning 'fruit conserve', the other /dʒæm/ meaning 'crush, wedging'. Later editions of this dictionary edited by Alfred C. Gimson, dropped this distinction.

Outside of England, can meaning 'able to' remains /kæn/, whereas the noun can 'container' or the verb can 'to put into a container' is /kæːn/; this is similar to the situation found in æ-tensing in some varieties of American English. A common minimal pair for modern RP speakers is band /bæːnd/ and banned /bænd/. Australian speakers who use ‘span’ as the past tense of ‘spin’ also have a minimal pair between longer /spæːn/ (meaning width or the transitive verb with a river or divide) and /spæn/, the past tense of ‘spin’ (/spæn/). Other minimal pairs found in Australian English include ‘Manning’ (the surname) /ˈmænɪŋ/ and ‘manning’ (the present participle and gerund of the verb ‘to man’) /ˈmæːnɪŋ/ as well as 'planet' /ˈplænət/ versus 'plan it' /ˈplæːnət/.

Apart from Jones's, dictionaries rarely show a difference between these varieties of /æ/.

/æ/ tensing

In the sociolinguistics of English, /æ/ tensing is a process that occurs in many accents of American English, and to some degree in Canadian English, by which /æ/, the "short a" vowel found in such words as ash, bath, man, lamp, pal, rag, sack, trap, etc., is raised and lengthened or diphthongized in various environments. The realization of this "tense" (as opposed to "lax") /æ/ varies from [æ̝ˑ] to [ɛə] to [eə] to [ɪə], depending on the speaker's regional accent. A common realization is [eə]—that is, a centering diphthong with a starting point closer than the vowel [ɛ] as in dress); that transcription will be used for convenience in this article. For the purposes of the chart below, the symbol "~" represents a continuous (rather than a phonemic split) system.

/æ/ tensing in North American English accents:
Environment		Example words	Boston & Rhode Island	Baltimore & Philadelphia	Canada & Northwest USA	Cincinnati (traditional)	General USA & Midland USA	Inland North USA (Great Lakes)	New York City & Yat	Southern USA	Upper Midwest USA	Western USA
Consonant following /æ/	Syllable type
/r/	open	arable, arid, baron, barrel, barren, carry, carrot, chariot, charity, clarity, Gary, Harry, Larry, marionette, maritime, marry, marriage, paragon, parent, parish, parody, parrot, etc.; this feature is determined by the presence or absence of the Mary-marry-merry merger	tense [ɛə~æ] or lax [æ]	lax [æ]	tense [ɛ~ɛə]	tense [eə]	tense [eə~ɛə~æ]	tense [eə]	lax [æ]	[æʲə]	tense [ɛ~ɛə]	tense [ɛ~ɛə]
/m/, /n/	closed	Alexander, answer, ant, band, can (the noun), can't, clam, dance, ham, hamburger, hand, handy, man, manly, pants, plan, planning, ranch, sand, slant, tan, understand, etc.; in Philadelphia, began, ran, and swam alone remain lax	tense [eə~ɛə~æ]	tense [eə]	tense [ɛə~æ]				tense [eə]		tense [eə~ɛə~æ]	tense [eə~ɛə~æ]
	open	amity, animal, can (the verb), Canada, ceramic, family (varies by speaker),, gamut, hammer, janitor, manager, manner, Montana, panel, planet, profanity, salmon, Spanish, etc.		lax [æ]					lax [æ]
/ɡ/	open	agate, agony, dragon, magazine, ragamuffin, etc.	lax [æ]		tense [e~ɛ~æ]	lax [æ]	lax [æ]				tense [eɪ]	lax [æ]
	closed	agriculture, bag, crag, drag, flag, magnet, rag, sag, tag, tagging, etc.							tense [eə]
/b/, /d/, /dʒ/, /ʃ/, /v/, /z/, /ʒ/	closed	absolve, abstain, add, ash, bad, badge, bash, cab, cash, clad, crag, dad, drab, fad, flash, glad, grab, halve (varies by speaker), jazz (varies by speaker), kashmir, mad, magnet, pad, plaid, rag, raspberry, rash, sad, sag, smash, splash, tab, tadpole, trash, etc. In NYC, this environment, particularly, /v/ and /z/, has a lot of variance and many exceptions to the rules. In Philadelphia, bad, mad, and glad alone in this set become tense. Similarly, in New York City, the /dʒ/ set is often tense even in open syllables (magic, imagine, etc.)	tense [ɛə~æ] or lax [æ]		lax [æ]	tense [eə]					lax [æ]	tense [ɛə~æ] or lax [æ]
/f/, /s/, /θ/	closed	ask, bask, basket, bath, brass, casket, cast, class, craft, crass, daft, drastic, glass, grass, flask, half, last, laugh, laughter, mask, mast, math, pass, past, path, plastic, task, wrath, etc.		tense [eə]
all other consonants		act, agony, allergy, apple, aspirin, athlete, avid, back, bat, brat, café, cafeteria, cap, cashew, cat, Catholic, chap, clap, classy, diagonal, fashion, fat, flap, flat, gap, gnat, latch, magazine, mallet, map, mastiff, match, maverick, Max, pack, pal, passive, passion, pat, patch, pattern, rabid, racket, rally, rap, rat, sack, sat, Saturn, savvy, scratch, shack, slack, slap, tackle, talent, trap, travel, wrap, etc.		lax [æ]		lax [æ]			lax [æ]
Footnotes  1) Nearly all American English speakers pronounce /æŋ/ somewhere between [æŋ] and [eɪŋ], though Western speakers specifically favor [eɪŋ]. 2) The NYC, Philadelphia, and Baltimore dialects' rule of tensing /æ/ in certain closed-syllable environments also applies to words inflectionally derived from those closed-syllable /æ/ environments that now have an open-syllable /æ/. For example, in addition to pass being tense (according to the general rule), so are its open-syllable derivatives passing and passer-by, but not passive.

Distinction between phonemic and non-phonemic /æ/ tensing

/æ/ tensing has two possible forms: either "non-phonemic" or "phonemic" (split). In General American, for example, the word man can be pronounced on a continuum from the lax-vowel  [mæn] to the tense-vowel  [meən], though the latter pronunciation is much more common. However, both vowel qualities are considered possible variations (allophones) of the single "short a" phoneme in man. Therefore, General American uses a continuous system in which a tensed allophone does not demonstrate that a new phoneme has splintered off from the original.

In some dialects, though, including the New York City and Mid-Atlantic ones (such as spoken in Philadelphia), the variations of "short a" can actually become entirely distinct phonemes, so that using a tense vowel rather than a lax vowel can potentially change the meanings of words or phrases. For instance, in the Mid-Atlantic dialects, the surname Manning can only be pronounced with a lax vowel as [ˈmænɪŋ]; if it is pronounced tensely as [ˈmeənɪŋ], it may be perceived as an entirely different word: the verb manning (as in "he was expertly manning the vehicle"). Therefore, these dialects have a phonemic split of the "short a." This relationship between two words (like Manning and manning) that differ in only a single sound is known as a minimal pair. Here are further examples of minimal pairs of the short a, using the Mid-Atlantic and General American accents for reference as, respectively, phonemic and non-phonemic accents:

Example words		Mid-Atlantic U.S.	General U.S.
calf e.g. The calf was born today.	caf e.g. Students must eat in the caf.	[kʰeəf] versus [kʰæf]	both, typically:  [kʰæf]
halve e.g. A knife can halve the bread in two.	have e.g. She might have fun.	[heəv] versus [ˈhæv]	both, typically:  [hæv]
manning e.g. He was manning the control panel.	Manning e.g. We met the Manning family.	[ˈmeənɪŋ] versus [ˈmænɪŋ]	both, typically: [ˈmeənɪŋ]
madder e.g. He's madder than a rabid dog.	matter e.g. Discuss this matter further.	[ˈmeəɾɚ] versus [ˈmæɾɚ]	both, typically:  [ˈmæɾɚ]
mass e.g. I grabbed a mass of clay.	Mass e.g. She works at Mass General.	[meəs] versus [mæs]	both, typically:  [mæs]
plan it e.g. We'll plan it after breakfast.	planet e.g. The planet orbits the Sun.	['pʰɫeənɪt] versus ['pʰɫænɪt]	both, typically:  ['pʰɫeənɪt]

Phonemic /æ/ tensing systems

In the Mid-Atlantic dialect, including Baltimore and Philadelphia, but also in the traditional dialects of Greater New York City, New Orleans (Yat), and Cincinnati (amongst older speakers), the tense /eə/ is an entirely separate phoneme from /æ/ (in Labovian linguistic variable notation, the phonemes are represented as "aeh" and "ae" respectively), since certain minimal pairs can be found.

In these accents there has thus been a phonemic split. Nevertheless, the distribution between /æ/ and /eə/ is largely predictable in the Philadelphia, Baltimore, and New York regions: In Philadelphia and Baltimore, tense [eə] occurs in closed syllables before the /n/, /m/, /f/, /θ/, and /s/, as well as the words mad, bad, and glad. In New York, tensing occurs in all those environments in addition to before voiced stops and /ʃ/. Lax [æ] usually occurs before /ŋ/, /l/, and voiceless stops, and also usually occurs in open syllables regardless of the following consonant. The word avenue normally has tense [eə] (unlike average, etc.).

In Philadelphia, tensing in some lexical items before /l/ and nontautosyllabic nasals has been reported. Here are further examples that are true for Philadelphia and Baltimore, as well as New York City:

Tense [eə]		Lax [æ]
man	[meən]	hang	[hæŋ]
ham	[heəm]	pal	[pʰæɫ]
laugh	[ɫeəf]	lap	[ɫæp]
bath	[beəθ]	bat	[bæt]
pass	[pʰeəs]	passage	[ˈpʰæsɪdʒ]

The main Philadelphia/Baltimore exceptions to the above generalizations are:

1. When a vowel-initial word-level suffix is added to a word with tense [eə], the vowel remains tense even though it has come to stand in an open syllable:

   mannish has [eə] like man, not [æ] like manage

   classy has [eə] like class, not [æ] like classic

   passing has [eə] like pass, not [æ] like Pasadena

2. When a polysyllabic word with [æ] in an open syllable gets truncated to a single closed syllable, the vowel remains:

   caf (truncation of cafeteria) has [æ], not [eə] like calf

   path (truncation of pathology) has [æ], not [eə] like path 'way, road'

   Mass (truncation of Massachusetts) has [æ], not [eə] like mass

3. Function words and irregular verb tenses have lax [æ], even in an environment which would usually cause tensing:

   and (a function word) has [æ], not [eə] like sand

   ran (a strong verb tense) has [æ], not [eə] like man

[eə] is also used in these accents before intervocalic /r/ in words like dairy and Mary and in non-rhotic varieties of these accents in words like square and scarce (which rhymes with glass for many non-rhotic speakers).

The phonemic tensing of /æ/ is similar to the "broad a" phenomenon of certain other dialects. The environment of "broad a" overlaps with that of /æ/ tensing, in that it occurs before voiceless fricatives in the same syllable and before nasals in certain environments; and both phenomena involve replacement of the short lax vowel /æ/ with a longer and tenser vowel. However, the "broad a" is lower and backer than [æ], while the result of /æ/ tensing is higher and fronter.

It is also related to the bad–lad split of some Southern British and Australian dialects, in which a short flat /æ/ is lengthened to [æː] in some conditions. The most significant differences from the Philadelphian system described here are that bad–lad splitting dialects have the "broad a" phenomenon, so the split can't occur there; that 'sad' is long; and that lengthening can occur before /ɡ/ and /l/.

In Webster's Third New International Dictionary (1961; Springfield, Mass: Merriam-Webster Inc.), the Mid-Atlantic tense /æ/ (written with \aa(ə)\, the lax /æ/ being \a\) is shown at individual entries as a variant pronunciation; for instance, the pronunciation of can "container" is \'kan, -aa(ə)n\. In the 11th (2003) edition of Merriam-Webster's Collegiate Dictionary, which is partly derived from the Third unabridged, the distinction is discussed in an introductory section on pronunciation but ignored elsewhere in the text. The editors justify their decision by maintaining that "this distinction is sufficiently infrequent that the traditional practice of using a single symbol is followed in this book" (p. 34a).

Non-phonemic /æ/ tensing systems

Universal /æ/ tensing

In accents that have undergone the Northern cities vowel shift, the phoneme /æ/ is raised and tensed in all possible environments. The most clear example of this is with the Inland North dialect, spoken in such areas as Chicago, Cleveland, Detroit, Buffalo, Rochester, and Syracuse.

/æ/ tensing before /m/ and /n/

Most American and Canadian English speakers, at the very least, display an /æ/ that is raised (tensed) and diphthongized before any m or n sound, as for example found in camp, man, ram, pan, ran, clamber, Sammy, etc., while being otherwise lower and laxer, without splitting the "short a" into two contrasting phonemes as the New York, Baltimore, Philadelphia, and Yat accents do. A common form of this is what William Labov calls the "nasal system", in which /æ/ is raised and tensed exclusively and severely before nasal consonants, regardless of whether there is a syllabic or morphemic boundary present. The nasal system is found in several separate and unrelated dialect regions, including the southern Midwest, northern New Jersey, Florida, and parts of Canada, among others, but it is most prominent—that is, the difference between the two allophones of /æ/ is greatest, and speakers with the nasal system are most concentrated—in eastern New England (including in the Boston accent).

More widespread among speakers of the Western United States, Canada, and southern Midwest is a "continuous system," which also revolves around "short a" before nasal consonants, but does not have quite as extreme a raising of the tongue as the "nasal system" does. Most varieties of General American English fall under this category. This system resembles the nasal system in that /æ/ is usually raised and tensed to [eə] before nasals, but instead of a sharp divide between a high, tense allophone before nasals and a low, lax one before other consonants, allophones of /æ/ occupy a continuum of varying degrees of height and tenseness between those two extremes, with a variety of phonetic and phonological factors interacting (sometimes differently in different dialects) to determine the height and tenseness of any particular example of /æ/.

In the Southern United States, the pattern most characteristic of Southern American English does not employ /æ/ tensing at all, but rather what has been called the "Southern drawl": /æ/ becomes in essence a triphthong [æjə]. However, many speakers from the South do still have the nasal /æ/-tensing system described above, particularly in Charleston, Atlanta, and Florida; and certain speakers from the New Orleans area have been reported to have a system very similar to the phonemic split of New York.

Additional /æ/ tensing before /ɡ/ and /ŋ/

For some speakers in Canada and in the North-Central and Northwestern United States, a following /ɡ/ (as in magazine, rag, bags, etc.) or /ŋ/ (as in bang, pang, 'gangster, angler, etc.) tenses an /æ/ as much as or more than a following nasal does; in much of the Midwest and some of the Pacific Northwest, this extends to the point that /æ/ even merges with /eɪ/, so that bag, for example, rhymes with plague or vague.

See also

• Phonological history of the English language
• Phonological history of English vowels

References

• Dobson, E.J. (1968). English pronunciation, 1500–1700 2. Oxford: Clarendon Press. OCLC 310545793. 
• de Jong, Gea; McDougall, Kirsty; Hudson, Toby; Nolan, Francis (2007), "The speaker discriminating power of sounds undergoing historical change: A formant-based study", the Proceedings of ICPhS Saarbrücken, pp. 1813–1816 
• Roach, Peter (2011), Cambridge English Pronouncing Dictionary (18th ed.), Cambridge: Cambridge University Press, ISBN 9780521152532 

Trap–bath split

• Wells, John C. (1982). Accents of English. Cambridge: Cambridge University Press. ISBN 0-521-22919-7. (vol. 1). ISBN 0-521-24224-X (vol. 2)., ISBN 0-521-24225-8 (vol. 3). 
• Gupta, A. F., Baths and becks, English Today 81, Vol. 21, No. 1, pp21–27 (2005).
• Kortmann, Bernd; Schneider, Edgar W.; Burridge, Kate; Mesthrie, Rajend; & Upton, Clive (Eds.). (2004). A handbook of varieties of English. Berlin: Mouton de Gruyter.
• Horvath, Barbara M. and Ronald J. Horvath. (2001). Short A in Australian English: A geolinguistic study. In English in Australia, ed. D. Blair and P. Collins, 341–55. Amsterdam: John Benjamins

Bad–lad split

• Leitner, Gerhard. (2004). Australia's Many Voices. Berlin: Mouton de Gruyter. ISBN 3-11-018194-0. (vol. 1). ISBN 3-11-018195-9 (vol.2). 
• Wells, John C. (1982). Accents of English. Cambridge: Cambridge University Press. ISBN 0-521-22919-7. (vol. 1). ISBN 0-521-24224-X (vol. 2)., ISBN 0-521-24225-8 (vol. 3). 

æ-tensing

• Benua, L. 1995. Identity effects in morphological truncation. In Papers in optimality theory, ed. J. N. Beckman, L. Walsh Dickey, and S. Urbanczyk. UMass Occasional Papers 18. Amherst: GLSA, 77–136.
• Ferguson, C. A. 1972. "Short a" in Philadelphia English. In Studies in linguistics in honor of George L. Trager, ed. M. E. Smith, 259–74. The Hague: Mouton.
• Kahn, D. 1976. Syllable-based generalizations in English phonology. Ph.D. dissertation, UCLA. Reproduced by the Indiana University Linguistics Club.
• Labov, W. 1966. The social stratification of English in New York City. Washington, D.C.: Center for Applied Linguistics.
• Labov, W. 1972. Sociolinguistic patterns. Philadelphia: University of Pennsylvania Press.
• Labov, W. 1981. Resolving the Neogrammarian controversy. Language 57:267–308.
• Labov, W. 2005. Transmission and Diffusion.
• Labov, William, Sharon Ash, and Charles Boberg (2006). The Atlas of North American English. Berlin: Mouton de Gruyter. ISBN 3-11-016746-8. 
• Trager, G. L. 1930. The pronunciation of "short a" in American Standard English. American Speech 5:396–400.
• Trager, G. L. 1934. What conditions limit variants of a phoneme? American Speech 9:313–15.
• Trager, G. L. 1940. One phonemic entity becomes two: The case of "short a". American Speech 15:255–58.
• Trager, G. L. 1941. ə ˈnəwt on æ ənd æ˔ˑ in əˈmerikən ˈiŋɡliʃ. Maître Phonétique 17–19.
• Wells, J. C. 1982. Accents of English. 3 vols. Cambridge: Cambridge University Press.
• Wood, Jim. 2011. Short-a in Northern New England. Journal of English Linguistics 39:135-165.

External links

• Sounds Familiar? – Listen to examples of regional accents and dialects on the British Library's 'Sounds Familiar' website, including an audio "bath" map of the UK