Open Book Publishers logo Open Access logo
  • button
  • button
  • button
GO TO...
Contents
Copyright
book cover
BUY THE BOOK

5. The phoneme-grapheme correspondences of English, 2: Vowels

© 2015 Greg Brooks, CC BY http://dx.doi.org/10.11647/OBP.0053.05

5.1 The general picture: the principal spellings of English vowel phonemes

This chapter can be summed up by saying that only five of the vowel phonemes of RP /æ, e, ɒ, aʊ, juː/ have highly regular spellings (80%+) wherever they occur, while none of the other 15 has a spelling accounting for more than 60% of its occurrences (though see section 5.4.3 for the possibility that /ɪ/ may also belong in the highly regular group).

The main regularities for all 20 vowel phonemes, plus /juː/, are summarised in Table 5.1, by position in the word. The letter-name vowels /eɪ, iː, aɪ, əʊ, juː/, plus /uː/, need to be analysed according to position within non-final vs final syllables (and then, within final syllables, according to two further, crossed dichotomies; see also sections 6.2 and 6.3), whereas all the rest need to be analysed according to position as initial, medial or final phoneme. The phonemes are therefore classified into short pure vowels, long pure vowels other than /iː, uː/, diphthongs other than /eɪ, aɪ, əʊ/, and the letter-name vowels plus /uː/.

Amidst the clutter of Table 5.1 various generalisations can be discerned:

  • five of the short pure vowels have a predominant spelling in initial position (/ʊ/ does not occur in this position, in RP)
  • the letter-name vowels and /uː/ have remarkable consistency in non-final syllables (with the notable exception of /iː/ in unstressed syllables)
  • the letter-name vowels and /uː/ mostly have split digraph spellings in closed final syllables (with the notable exceptions of /iː, uː/ in monosyllables)
  • there is a minor and scattered pattern before consonant clusters: in closed monosyllables the long vowel /ɑː/ and the letter-name vowels /aɪ, əʊ/ are spelt with the single letters <a, i, o> and the letter-name vowel /eɪ/ is spelt <ai> before /nt/; for /ɑː, eɪ/ this pattern extends to closed final syllables of polysyllables
  • digraphs with <-y> (<ay, oy>) tend to occur word-finally and to alternate with digraphs with <i> (<ai, oi>) elsewhere
  • similarly, digraphs with <-w> (<aw, ew, ow>) tend to occur word-finally and to alternate with digraphs with <u> (<au, eu, ou>) elsewhere
  • the biggest muddle is /ɔː/.

Table 5.1: Main spellings of the 20 vowel phonemes, plus /juː/,
by word position.

Vowels other than the letter-name vowels and /uː/

Vowel

Position

Initial phoneme

Medial phoneme

Final phoneme

Short pure vowels

/æ/

<a>

(does not occur)

/e/

<e>

(does not occur)

/ɪ/

<i>, but frequently <e> in unstressed syllables

<i>, but <a> in word-final unstressed /ɪʤ/, and frequently <e> in other unstressed syllables

(does not occur)

/ɒ/

<o>

<o>, but mainly <a> after /w/

(does not occur)

/ʌ/

<u>, but there are many examples with <o>

(does not occur)

/ʊ/

(does not occur)

<oo> in monosyllables ending in /d, k/, <u> elsewhere

(does not occur)

/ə/

<a>, with few exceptions

<a>, with many exceptions

<er>, with many exceptions

Long pure vowels other than /iː, uː/

/ɑː/

<ar>, but <a> before consonant clusters

(nothing predominates)

/ɜː/

(very rare)

<er>, but <or> after initial /w/

<er>

/ɔː/

<au, or>, but mainly <a> before /l/

<au, aw, or, ore>, but mainly <a> after /w/ and before /l/

<aw, or, ore>

Diphthongs other than /eɪ, aɪ, əʊ/

/ɔɪ/

<oi>

<oy>

/aʊ/

<ou>

<ou>, but <ow> before /l, n/ and vowel letters

<ow>

/eə/

<air>

<ar>

<are>

/ɪə/

(very rare)

<er>

<ear>

/ʊə/

(so rare and diverse that no generalisations are worthwhile)

The letter-name vowels, plus /uː/

Vowel

In non-final syllables

In final syllables

Closed

Open

In polysyllables

In monosyllables

In polysyllables

In monosyllables

/eɪ/

<a>

<ai> before /nt/, otherwise <a.e>, with many exceptions with <ai>

<ay>

/iː/

mainly <e>; many excep-tions with <i> in unstressed syllables

<e.e>, with many exceptions

<ee>, with many exceptions

<y>, with many exceptions

<ee>, with some exceptions

/aɪ/

<i>

<i.e>

<i> before consonant clusters, <igh> before /t/, otherwise <i.e>

<y>, with a few exceptions, mostly with <i>

<y>, with more exceptions than examples

/əʊ/

<o>

<o.e>

<o> before consonant clusters, otherwise <o.e>

<ow> in two-syllable words after /l, r/, otherwise <o>

<ow>

/juː/

<u>

<u.e>

<ue>

<ew>

/uː/

<u>

<oo> in stressed final /ˈuːn/, otherwise <u.e>

<oo>

<oo>

<ew>

5.2 Order of description

In sections 5.4-7 I set out the vocalic phoneme-grapheme correspondences between RP and British spelling, under the vowel phonemes listed in the order in which they appear in Table 5.1. Section 5.4 covers short pure vowels, section 5.5 long pure vowels other than /iː, uː/, section 5.6 diphthongs other than /eɪ, aɪ, əʊ/, and section 5.7 the letter-name vowels plus /uː/.

Under each vowel phoneme I deal with the spellings in this order:

1) The basic grapheme. In my opinion, each of the 20 vowel phonemes of English, plus /juː/, has a basic grapheme, the one which is most frequent and/or seems most natural as its spelling.

2) Other graphemes which are used to spell that phoneme with reasonable frequency.

3) Oddities, graphemes which are used to spell that phoneme only rarely.

4) Any 2-phoneme graphemes in which the phoneme occurs. (Almost all the 2-phoneme graphemes are also Oddities, but a few belong to the main system and are included there).

5) Any 3-phoneme grapheme in which the phoneme occurs. Both 3-phoneme graphemes are definitely Oddities.

Most entries end with Notes, and some have Tables.

By reasonable frequency here I usually mean at least 9% of the occurrences of that phoneme in running text. The reason for setting a generally higher criterion for vowel spellings than for consonant spellings (see section 3.2) is that vowel spellings are so much more varied. For the choice of 9% see in particular /ɔː/ spelt <au, aw> and /uː/ spelt <ew> (sections 5.5.3, 5.7.6), which definitely have to be considered parts of the main system of English spelling; and contrast (at 8%) /ʌ/ spelt <ou>, /ɜː/ spelt <ear> and /ɔː/ spelt <our> (sections 5.4.5, 5.5.2, 5.5.3), which equally certainly are Oddities and not parts of the main system. However, as with consonant phonemes, the dividing line cannot be absolute. I have ‘promoted’ four infrequent correspondences, /ɒ/ spelt <a> at 6%, /ɪə/ spelt <eer> at 8%, /juː/ spelt <ue> - percentage unknown, but clearly very low, and /uː/ spelt <ue> at <1% (sections 5.4.4, 5.6.4, 5.7.5-6), to the main system, where they obviously belong. In the case of /uː/ spelt <ue> this is largely because in the grapheme-phoneme direction the correspondences of <ue> are highly regular – see section 10.37.

Again, the frequencies are Carney’s text frequencies (see section 3.2), but for /ɪ, ɪə/ I take issue with them, and for /iː/ I dispense with them completely – see sections 5.4.3, 5.6.4, 5.7.2.

5.3 The main system and the rest

As for the consonant phonemes, under each vowel phoneme I separate the correspondences with graphemes into what I consider to be the main system and the rest. The correspondences I include in the main system are those which seem to me to operate as part of larger regularities, even though pretty rarely as absolute rules. For instance, there is a strong tendency in English spelling for the letter-name vowel phonemes /eɪ, iː, aɪ, əʊ, juː/, plus /uː/, to be written with the single vowel letters <a, e, i, o, u> in non-final syllables. Within the main system I include only the correspondences which seem to me to form part of these larger regularities. For the vowel phonemes these comprise the basic correspondences and the correspondences which have reasonable frequency as I’ve defined it above, plus a few with lower frequencies which have to be in the main system, but not the 2-phoneme graphemes (with a few exceptions), or the 3-phoneme graphemes and Oddities. Correspondences which have reasonable frequency are shown in 9-point type, the rest in 7.5-point.

5.4 Short pure vowels: /æ e ɪ ɒ ʌ ʊ ə/

N.B. Six of the seven short pure vowels do not occur word-finally, but /ə/ is frequent in that position.

5.4.1 /æ/ as in ash

Does not occur word-finally.

The main system

Basic grapheme

<a>

99%

e.g. cat

Other frequent graphemes

(none)

The rest

Oddities

1% in total

<ae>

only in Gaelic pronounced /ˈgælɪk/

<ai>

only in Laing, plaid, plait

<al>

only in salmon

<ei>

only in reveille

<i>

only in absinthe, impasse, ingenu(e), lingerie pronounced /ˈlænʒəriː/ (also pronounced /ˈlɒnʤəreɪ/), meringue,
pince-nez
, timbale, timbre

2-phoneme graphemes

(none)

5.4.2 /e/ as in end

Does not occur word-finally, and is rare before /ŋ/ - see section 3.8.2.

The main system

Basic grapheme

<e>

84%

e.g. pet

Other frequent graphemes

(none)

The rest

Oddities

16% in total

<a>

only in any, ate pronounced /et/ (also pronounced /eɪt/), many, Thames, first <a> in secretaria-l/t, second <a> in asphalt pronounced /ˈæ∫felt/ (also pronounced /ˈæsfælt/), and

- a few words ending <-ary> with the stress two syllables before the <a>, e.g. necessary, secretary, pronounced /ˈnesəseriː, ˈsekrəteriː/ (also pronounced /ˈnesəsriː, ˈsekrətriː/ with no vowel phoneme corresponding to the <a> - for the elided vowels in this and the next three paragraphs see section 6.10)

- a few adverbs ending <-arily>, e.g. militarily, necessarily, primarily, voluntarily pronounced /mɪlɪˈterɪliː, nesəˈserɪliː, praɪˈmerɪliː, vɒlənˈterɪliː/ with the <a> stressed (also pronounced either /mɪlɪˈteərɪliː, nesəˈseərɪliː, praɪˈmeərɪliː, vɒlənˈteərɪliː/ with /eə/ spelt <ar> and the <r> also a grapheme in its own right spelling /r/ - for dual-functioning see

section 7.1 - or reduced to four syllables as /ˈmɪlɪtrəliː, ˈnesəsrəliː, ˈpraɪmrəliː, ˈvɒləntrəliː/ with stress shifted one or two syllables forward, again no vowel phoneme corresponding to the <a>, and the vowel before /liː/ changed from /ɪ/ to /ə/)

- temporary pronounced /ˈtempəreriː/ (also pronounced /ˈtemprəriː/ with no vowel phoneme corresponding to the <o> and the <a> now spelling /ə/)

- temporarily pronounced /tempəˈrerɪliː/ (also pronounced either /tempəˈreərɪliː/ with /eə/ spelt <ar> and the <r> also a grapheme in its own right spelling /r/ - for dual-functioning see section 7.1 - or reduced to three syllables as /ˈtemprəliː/ with no vowel phonemes corresponding to the <o> or the <a> and the two /r/ phonemes reduced to one)

<ae>

only in aesthetic pronounced /esˈθetɪk/ (also pronounced /iːsˈθetɪk/), haemorrhage, haemorrhoid

<ai>

only in bouillabaisse, said, saith and (usually, nowadays) again(st)

<ay>

only in says

<ea>

6% in about 60 words – see Note

<ei>

only in heifer, leisure, seigneur

<eo>

only in Geoff(rey), jeopardy, Leonard, leopard

<ie>

only in friend

<u>

only in burial, bury

2-phoneme graphemes

(none)

3-phoneme grapheme

/eks/ spelt <x>

only in X-ray, etc.

Notes

There are only about 60 stem words in which /e/ is spelt <ea>, but no rule can be given to identify them, so here is a list: Beaconsfield; treacher-ous/y; bread, breadth, dead, dread, (a)head, lead (the metal, plus derivatives leaded, leading), meadow, read (past tense and participle), Reading (Berkshire), (al)ready, spread, (in)stead, steadfast, steady, thread, tread(le); deaf; breakfast; dealt, health, jealous, realm, stealth, wealth, zeal-ot/ous; dreamt, seamstress; cleanly (adjective, plus derivative cleanliness), cleanse, leant, meant; leapt, weapon; (a)breast; peasant, pheasant, pleasant; measure, pleasure, treasure; sweat, threat(en); breath, death; feather, heather, leather, weather; endeavour, heaven, heavy, leaven and other derivatives not listed. In my opinion very little would be lost if all these words were instead spelt with <e> – indeed, one spelling reform proposal is that the first change should be to spell all occurrences of /e/ with <e> and nothing else – but:

  • this might be difficult for some of the Oddities above
  • bred, led, red, lent, wether would become homonyms of the words already so spelt
  • various words would have to acquire unfamiliar letters to conform to other rules: *tretcher-ous/y, *bredded, *bredding, *dredded, *dredding, *hedded, *hedding, *ledded, *ledding, Redding (so spelt in first map, 1611), *(al)reddy, *reddying, *reddied, *reddies, *deff, *breckfast, *jellous, *zell-ot/ous, *weppon, *swetted, *swetting, *thretten(ed/ing).

    5.4.3 /ɪ/ as in ink

Does not occur word-finally, in my opinion/version of RP. For doubts about the percentages see Notes.

The main system

For all these categories see Notes.

Basic grapheme

<i>

only 61% if word-final /ɪ/ spelt <y> is allowed, but a lot more otherwise

e.g. sit. Regular in initial and medial positions

Other frequent graphemes

<y>

20% if word-final /ɪ/ spelt <y> is allowed, but a lot less otherwise

e.g. bicycle, crystal

<e>

only 16% if word-final /ɪ/ spelt <y> is allowed, but a lot more otherwise

e.g. diocese (first <e>), England, English, enough, entirety (both <e>’s), extreme (first <e>), pretty, scavenge (first <e>), stupefy, variety. Regular in certain suffixes

The rest

Oddities

at least 4% in total

- in stressed syllables

<ee>

only in breeches /ˈbrɪʧɪz/

<hea>

only in forehead pronounced /ˈfɒrɪd/

<ie>

only in sieve

<o>

only in women

<u>

only in business, busy

- in unstressed syllables

<a>

in about 250 words ending in unstressed word-final /ɪʤ/, which is mainly spelt <-age>, e.g. village, plus furnace, menace, necklace, octave, orange, signature, surface, spinach pronounced /ˈspɪnɪʤ/ and second <a> in character, palace. See Notes

<ai>

only in bargain, captain, chamberlain, chaplain, fountain, mountain, porcelain

<ee>

only in been when unstressed, cheerio /bɪn, ʧɪriːˈjəʊ/

<ei>

only in counterfeit pronounced /ˈkaʊntəfɪt/ (also pronounced /ˈkaʊntəfiːt/), forfeit, sovereign, surfeit

<ia>

only in carriage, marriage

<ie>

only in (hand/nec)kerchief, mischief, mischievous

<o>

only in pigeon (taking <ge> as spelling /ʤ/; compare pidgin)

<u>

only in lettuce, minute (noun /ˈmɪnɪt/, ‘60 seconds’), missus

<wi>

only in housewife (‘sewing kit’, pronounced /ˈhʌzɪf/)

2-phoneme grapheme

/ɪz/

spelt <s>

only, following an apostrophe, in regular singular and irregular plural possessive forms ending in a sibilant consonant (/s, z, ∫, ʒ, ʧ, ʤ/), e.g. Brooks’s (book), jazz’s (appeal), Bush’s (government), (the) mirage’s (appearance), (the) Church’s (mission), (the) village’s (centre), (the) geese’s (cackling). See /z/, section 3.7.8

Notes

Carney (1994: 135, 139, 161, 380, 430) states that, in the version of RP which he analysed (and which he and many other phoneticians prefer to call SBS, Southern British Standard, and Cruttenden (2014) now dubs GB, General British), /ɪ/ does occur word-finally and in that position is mainly spelt <y>. His percentages for the spellings of /ɪ/ are based on that analysis. But he also points out (especially on pp.134-5, 380) that many (especially younger) RP-speakers do not have word-final /ɪ/ in their accents, but (a short version of) /iː/ instead. And on page xxii he says that /ɪ/ does not occur in final open syllables, thus contradicting most of his other statements on this (cf. also his p.56).

Cruttenden (2014: 97; and cf. p.84), on the other hand, says: ‘Word-final unaccented /ɪ/ has now been replaced in all but the oldest GB speakers by /iː/ …, e.g. in copy ‘. I agree, and think that children learning to spell English are more likely to hear the final phoneme of, say, city as /iː/ rather than /ɪ/, and different from the definite /ɪ/ in the first syllable. Similarly, Mines et al. (1978, Table A-2, p.237) show that only about 1% of occurrences of /ɪ/ in their analysis were word-final (admittedly in the General American accent, but here the point is applicable to RP as well). I have therefore not followed Carney’s analysis, but Cruttenden’s, and count /ɪ/ as occurring only in initial and medial positions. This does mean, unfortunately, that I have not been able to use Carney’s percentages (which, oddly, Cruttenden, 2014: 113) retains) without reservation, and he does not provide enough information to re-calculate them (this would require knowing what proportion of <y>-spellings are word-final). This difference in analysing /ɪ/ also entails differences in the analysis of and percentages for /iː/ – see section 5.7.2.

<i> is regular in initial and medial position in both stressed and unstressed syllables.

Exceptions with <e>: The only words in which /ɪ/ is spelt <e> in stressed syllables are England, English, pretty and Cecily pronounced /ˈsɪsɪliː/ and therefore as a homophone of Sicily (Cecily is also pronounced /ˈsesɪliː/). Categories where <e> is the regular spelling of unstressed /ɪ/ are:

  • the past tense and past participle verb ending <-ed> spelling /ɪd/ after /t, d/, e.g. ousted, decided. N.B. Carney (1994: 135) says this ending is not included in his percentages
  • a few adjectives which are derived from or resemble past participles but have /ɪd/ rather than the expected /d, t/, but often with a different meaning, e.g. accursed, aged (/ˈeɪʤɪd/ ‘elderly’ vs /eɪʤd/ ‘having x years’), beloved (/bɪˈlʌvɪd/ ‘the loved one’ vs /bɪˈlʌvd/ ‘adored’), blessed (/ˈblesɪd/ ‘holy’ vs /blest/ ‘consecrated’), cragged, crooked (/ˈkrʊkɪd/ ‘untrustworthy’ vs /krʊkt/ ‘at an angle’), Crutched (Friars), cursed (/ˈkɜːsɪd/ ‘damnable’ vs /kɜːst/ ‘swore badly/put a hex on’), cussed (/ˈkʌsɪd/ ‘stubborn’ vs /kʌst/ ‘swore mildly’), deuced, dogged (/ˈdɒgɪd/ ‘persistent’ vs /dɒgd/ ‘followed’), fixed (/ˈfɪksɪd/ ‘persistent’ vs /fikst/ ‘mended’), horned (owl), jagged (/ˈʤægɪd/ ‘with sharp points’ vs /ʤægd/ past tense of jag), learned (/ˈlɜːnɪd/ ‘wise’ vs /lɜːnd/ regular past tense of learn), (bow/one-)legged, naked, ragged (/ˈrægɪd/ ‘torn, exhausted’ vs /rægd/ past tense of rag), rugged, sacred, supposed (/səˈpəʊzɪd/ ‘apparent’ vs /səˈpəʊzd/ past tense of suppose), wicked, wretched. In (ac)cursed, blessed, crooked, Crutched, cussed, deuced, fixed, wretched, not only does the /ɪ/ surface (see section 7.2) but the /t/ voices to /d/
  • the past participle verb ending <-ed> spelling /ɪd/ before adverbial <-ly>, e.g. advisedly, allegedly, assuredly, barefacedly, composedly, confusedly, deservedly, determinedly, fixedly, markedly, relaxedly, (un)reservedly, supposedly, unabashedly, unashamedly, undisguisedly, unrestrainedly. Again, in barefacedly, fixedly, markedly, relaxedly, not only does the /ɪ/ surface (see section 7.2) but the /t/ voices to /d/
  • the <ed> element in a very few nouns in <-ness> formed from past participles, e.g. preparedness, where not only does the /ɪ/ surface (see section 7.2) but also /r/-linking occurs (see section 3.6) and the <r> is both part of the grapheme <are> spelling /eə/ and a grapheme in its own right spelling /r/. For dual-functioning see section 7.1
  • the superlative adjective ending <-est>, e.g. biggest, grandest
  • the archaic second and third person singular verb endings <-est, -eth>, e.g. gavest, goeth
  • the noun plural and third person singular present tense verb endings /ɪz/ sometimes spelt <-es> after /s, z, ∫, ʒ, ʧ, ʤ/ - see the entries for those consonants in sections 3.6.6, 3.6.8, 3.7.3, 3.7.4, 3.6.2, 3.6.4. N.B. Carney (1994: 135) says this ending is not included in his percentages either
  • the unstressed noun suffixes /ɪs, lɪs, lɪt, nɪs/ spelt <-ess, -less, -let, -ness>, e.g. goddess, listless, booklet, madness. There are many nouns ending in unstressed /ɪs/ when it is not a suffix and is therefore not spelt <-ess>, e.g. furnace, menace, palace (all of which have alternative pronunciations in /əs/), diocese (which is exceptional in various ways – see under /iː/, section 5.7.2), justice, practice, mortice/mortise, practise, premise/premiss, promise, treatise and all nouns ending in /sɪs/, e.g. crisis. The set of words ending in stressed /ɪs/, all but one spelt with <-iss>, is very small: amiss, bliss, dismiss, diss, hiss, kiss, miss, p*ss, remiss, Swiss (exception: abyss)
  • the unstressed prefixes (Germanic) /bɪ/ and (Latin) /dɪ, ɪ, ɪks/ɪgz, prɪ, rɪ/ spelt <be-, de-, e-, ex-, pre-, re->, e.g. before, beholden, decline, deliver, effective, efficient, extreme, examine, precede, predict, regale, reject
  • the ending /ɪtiː/ when the previous letter is <i>, i.e. in anxiety, dubiety, gaiety, moiety, notoriety, (im)piety, (im)propriety, sobriety, society, variety, plus entirety, naivety, nicety, surety. The last four words are exceptions to the general rule that the ending /ɪtiː/ is spelt <-ety> only when the previous letter is <i>, otherwise <-ity>, e.g. nullity, paucity, including cases where this involves <e>-deletion or <y>-replacement or -deletion (e.g. scarcity, laity – see sections 6.4-6); but the more regular spellings *entirity, *naivity, *nicity, *surity would look odd, as would *layity. In entirety, surety, /r/-linking occurs (see section 3.6) and then the <r> is both part of the graphemes <ir, ur> spelling /aɪə, ʊə/ and a grapheme in its own right spelling /r/. For dual-functioning see section 7.1. The spelling of the ending /ɪtiː/ with <e> when the previous letter is <i> is one of the ways in which English spelling avoids the sequence <ii>, which appears to be tolerated only in alibiing, fasciitis, leylandii (and probably many other biological species names), Pompeii, radii, shanghaiing, Shiite, skiing, taxiing (all of which have an automatic intervening /j/-glide).

Then there are groups of words with /ɪ/ spelt <e> where no rule can be given:

  • the ending /ɪt/ is normally spelt <-it> (e.g. rabbit and about 200 other words) but is spelt <-et> in, e.g., ashet, brisket, budget, buffet (‘strike’), dulcet, facet, fillet, gannet, gullet, nugget, plummet, punnet, russet, secret, valet (also pronounced with /eɪ/) and about 150 other words, so all words with both endings just have to be learnt
  • the ending /ɪfaɪ(d)/ is spelt with <e> rather than <i> in just four words: liquefy, putrefy, rarefied, stupefy. These endings are normally spelt <-if-y/ied> (e.g. nullify, pacified), including cases where this involves <e>-deletion or <y>-replacement (e.g. amplify, jollify – see sections 6.4-5); liquefy has the alternative spelling liquify, but the more regular spellings *putrify, *rarified, *stupify would look odd
  • a ragbag of other words, e.g. allegation, employ, forest, hallelujah, integral (when pronounced /ˈɪntɪgrəl/, with stress on first syllable; also pronounced /ɪnˈtegrəl/, with stress on second syllable), kitchen, mannequin, regalia, subject (noun /ˈsʌbʤɪkt/, with stress on first syllable; the verb is pronounced /səbˈʤekt/), vinegar; first <e> in anecdote, antelope, barometer and all the instruments ending in <-ometer> (but not kilometre or other compounds of metre), celebrity, consecrate, eccentric, ellipse, elope, enamel, integrate, negate, neglect, scavenge, sequential; second <e> in elegant, elephant, elevate, peregrine, and many others.

Initial /ɪ/ spelt <y> is extremely rare, occurring only in the archaic word yclept (‘named’), and the names of the plant and essential oil ylang-ylang (also spelt ilang-ilang), the type of boat yngling, and the elements ytterbium, yttrium and the names Yvette, Yvonne. Only in yngling, yttrium is it stressed.

Other exceptions with /ɪ/ spelt <y> are all medial and mainly of Greek origin. No generalisations seem possible about contexts in which medial /ɪ/ is spelt <y> rather than <i>, so here is a list: abyss, acetylene, acronym, amethyst, analysis, analytic, aneurysm (also spelt aneurism), antonym, apocalypse, apocrypha(l), asphyxiate, beryl, bi/tri-cycle, calyx, cataclysm, catalyst, chlamydia, chlorophyll and a few other words ending in -phyll, coccyx, cotyledon, crypt(ic), crystal, cyclamen, cygnet, cylinder, cymbal, cynic, cyst, (ptero)dactyl, di/tri-ptych, dynasty (first syllable), eponym, etymology (second syllable), eucalyptus, glycerine, gryphon, gymkhana, gym(nast/ium), gyp, gypsum, gypsy (first syllable), hieroglyph, hydroxyl (last syllable), hymn, hypnosis, hypnotise, hypocrisy (first syllable), hypocrite, idyll, larynx, lymph, lynch, myriad, metempsychosis, nymph, onyx, oryx, oxygen, paralysis, paralytic, paroxysm, pharynx, phylactery (first syllable), physics, polygamy (second syllable), polymer, polyp, pygmy (first syllable), pyx, rhythm, salicylic, sibyl, strychnine, sybarite, sycamore, sycophant, syllabic, syllable, syllabub, syllabus, sylloge, symbol, sympathy, syndicate, synonym (first and last syllables), syntax, synthetic, syphilis, tryst, tyranny (first syllable). In my opinion, nothing but a source of confusion would be lost if all such words were spelt with <i>.

There are fairly clear rules for spelling word-final /ɪʤ/. When stressed, its regular spelling is <-idge> and there appear to be no exceptions, but there are very few words in this set: (a)bridge, fridge, midge, ridge. When unstressed, the regular spelling is <-age>, e.g. cabbage, disparage, garage pronounced /ˈgærɪʤ/, image, mortgage, village and about 250 other words. However, here there are several exceptions, with various spellings: carriage, college, (ac)knowledge, marriage, ostrich, privilege, sacrilege, sandwich pronounced /ˈsæmwɪʤ/ (and many other placenames with this ending – but I’m not dealing with placenames), selvedge, spinach pronounced /ˈspɪnɪʤ/, vestige, plus three words with, confusingly, the regular stressed spelling: cartridge, partridge, porridge (and the last of these is even more confusing because the alternative spelling porage has the regular unstressed ending).

5.4.4 /ɒ/ as in ox

Does not occur word-finally, or in US accents.

The main system

Basic grapheme

<o>

92%

e.g. long

Other frequent grapheme

<a>

6%

e.g. squash, wash, what. See Notes

The rest

Oddities

2% in total

<ach>

only in yacht

<au>

only in Aussie, Australia, Austria, because (also increasingly pronounced, unusually, with stressed /ə/), cauliflower, laurel, Laurence, sausage, plus a few words also pronounced with /ɔː/: auction, austere, caustic, claustrophobi-a/c, hydraulic,
(bacca)laureate

<e>

in about 20 more recent French loanwords, e.g. (the relevant <e>’s are in caps) ambiEnce, cliEntele, denouemEnt, détEnte, divertissemEnt, Embonpoint, Embouchure, En (suite), Enceinte pronounced /ɒnˈsænt/ (also pronounced /enˈseɪnt/), Enclave pronounced

/ˈɒŋkleɪv/ (more often pronounced /ˈeŋkleɪv/), Encore, Ennui, EnsEmble, EntEnte, Entourage, Entracte, Entrepreneur, Entree, Envelope pronounced /ˈɒnvələʊp/ (also pronounced /ˈenvələʊp/), gEnre, rapprochemEnt, rEntier. See Notes

<eau>

only in bureaucracy, bureaucratise

<ho>

only in bonhomie, honest, honour and derivatives

<i>

only in lingerie pronounced /ˈlɒnʤəreɪ/ (also pronounced /ˈlænʒəriː/)

<ou>

only in cough, hough, lough, trough

<ow>

only in (ac)knowledge, rowlock

2-phoneme graphemes

(none)

Notes

If we follow Crystal (2012: 131-2) and Upward and Davidson (2011: 176-9), ‘more recent’ in terms of loanwords from French means after the Great Vowel Shift, which began about AD1400 and was complete by about AD1600.

There is a reasonably strong tendency for /ɒ/ to be spelt <a> after /w/, however spelt – see Table 5.2, and cf. /ɔː/, section 5.4.3. The reason for putting this correspondence in the main system despite its percentage is that Carney will have excluded the two high-frequency function words what, was, and these are enough to make this a frequent correspondence.

Table 5.2: Spellings of /ɒ/ as <a> after /w/.

after /w/ spelt <u>

– always after /k/ spelt <q>

after /w/ spelt <w>

after /w/ spelt <wh>

(e)quality, quad and derivatives, quadrille, quaff, quag(mire) (also pronounced /ˈkwæg(-)/), quagga (also pronounced /ˈkwægə/), qualify and derivatives and associates, quandary, quant and derivatives, quarantine, quarrel, quarry, quash, quatrain, squab(ble), squad and derivatives, squal-id/or, squander, squash, squat

swab, swaddle, swallow, swamp, swan, swap, swash(buckling), swastika, swat, swatch, twaddle, wad, waddle, waddy, wadi, waffle, waft (also pronounced with /ɔː/), wallah, wallet, wallop, wallow, wally, walrus (also pronounced with /ɔː/), wampum, wan, wand, wander, wannabe, want, wanton, warrant, warren, warrigal, warrior, was, wash, wasp, wassail, wast, watch, watt, wattle

what

Exceptions (words in which /ɒ/ is spelt <o> not <a> after /w/)

quod, quondam

swop, swot, wobble, wodge, wog, woggle, wok, wombat, wonk, wonton, wop, wot

whop(per)

Other words in which /ɒ/ is spelt <a> are ambience, bandeau, blancmange (second <a>), bouffant, chanterelle, confidant(e), debutante, diamante (second <a>), fiance(e), flambe, flambeau, insouciance, jalap (first <a>), mange-tout, moustache (now mostly pronounced with /ɑː/ in RP), nuance, scallop (also pronounced /ˈskæləp/), seance, stalwart (first <a>), wrath pronounced /rɒθ/ (also pronounced /rɔːθ/). Elderly relatives of mine (born about 1880) would say /ˈɒlbət/ (‘Olbat’) when referring to Victoria’s consort.

For more detail on the absence of /ɒ/ in US accents see Cruttenden (2014: 127) and Carney (1994: 59).

5.4.5 /ʌ/ as in up

Does not occur word-finally in RP, and does not occur at all in local accents of the north of England.

The main system

For both categories see Notes.

Basic grapheme

<u>

63%

e.g. dulcimer, up

Other frequent grapheme

<o>

27%

e.g. above, monk

The rest

Oddities

10% in total

<oe>

only in does(n’t)

<oo>

only in blood, flood

<ou>

8% only in chough, Colclough pronounced /ˈkəʊlklʌf/ (also pronounced /ˈkəʊkliː/), country, couple, couplet, courage, cousin, double, doublet, enough, flourish, *hiccough (properly spelt hiccup), housewife (‘sewing kit’, pronounced /ˈhʌzɪf/), nourish, rough, slough (‘shed skin’), sough, souther-n/ly, touch, tough, trouble, young

2-phoneme grapheme

/wʌ/

spelt <o>

only in once, one

Notes

Many people from the North of England do not have this phoneme in their accents, but retain the earlier /ʊ/ (see next section) in most of the words in which RP has /ʌ/. So far from simplifying their task, this presents them with three principal ways of spelling /ʊ/: <o, oo, u>. Also, in some northern accents some words in which RP has /ʌ/ are pronounced with /ɒ/, e.g. one, among, nothing.

For spelling RP /ʌ/ some regularities can be stated:

  • <o> is regular before /θ, ð, v/: doth, nothing (is she sweet?); brother, mother, other, smother; above, coven, covenant, (dis/re/un)cover, covert pronounced /ˈkʌvɜːt/ (also pronounced /ˈkʊəvɜːt/), covet(ous), covey, dove, glove, govern(or), lovage, lovat, love, Lovell, oven, plover, shove, shovel, slovenly, windhover (exceptions: southern, guv)
  • <u> is regular before /b, d, g, ʤ, k, p, ∫, ʧ/: e.g. club, hub, public; bud, mud, shudder; buggy, juggle, luggage; budget, cudgel, judge; buxom, duck, luxury; abrupt, cup, supper; blush, thrush, usher; clutch, duchess, much (exceptions: amok, Cadogan, conjure (‘do magic’), pinochle, sojourn (also pronounced with /ɒ/), twopence, twopenny; country, couple, cousin, double(t), doubloon, touch, trouble).

Since there are no other useful generalisations it seems best to give a list of other words with /ʌ/ spelt <o>: accomplice, accomplish, become, borough, colour, colander (also pronounced with /ɒ/), Colombia (second syllable), come, comfort(able), comfrey, comfy, company, (en)compass, constable, coz, cozen, done, dost, dozen, dromedary, front, frontier, honey, London (first syllable), Monday, monetary, money, monger and its compounds, mongrel, monk, monkey, Monroe, Montgomery (twice), month, none, onion (first syllable), some, somersault, son, sponge, thorough, ton, tonne, tongue, won, wonder, worrit, worry. Some words which used to have /ʌ/ in RP now have /ɒ/ instead, e.g. combat, comrade, conduit, Coventry.

/wʌ/ also has 2-grapheme spellings, e.g. <wo> in wonder.

5.4.6 /ʊ/ as in pull

Occurs only medially (in RP), and never before /ŋ/ (in RP).

The main system

For both these categories see Notes.

Basic grapheme

<oo>

64%

e.g. hood, look

Other frequent grapheme

<u>

32%

e.g. cushion, push

The rest

Oddities

4% in total (probably an underestimate because Carney will not have counted could, should, would)

<o>

only in bosom (first <o>), wol-f/ves, wolfram, wolverine, Wolverhampton, woman

<or>

only in worsted (‘cloth’)

<ou>

only in courier, pouffe pronounced /pʊf/ (also pronounced /puːf/)

<oul>

only in could, should, would

2-phoneme graphemes

(none)

Notes

In RP (as distinct from local accents of the north of England, in which /ʊ/ is much more frequent) there are rather few words containing this phoneme, perhaps only about 80 stem words, plus a potentially much larger set of adjectives and nouns ending in /fʊl/ spelt <-ful>.

<oo> spelling /ʊ/ occurs in only about 28 stem words, namely four words which have alternative pronunciations with /uː/: food /fʊd, fuːd/, hoodlum /ˈhʊdləm, ˈhuːdləm/, room /rʊm, ruːm/, woofer /ˈwʊfə, ˈwuːfə/ (cf., as mentioned under Oddities, pouffe, though with a different grapheme), plus Chinook, forsook, foot, gooseberry /ˈgʊzbriː/, hoof (and its plural hooves), poof(ter), soot, woof (/wʊf/ ‘barking’; contrast woof /wuːf/ ‘weft’), wool and monosyllables ending in /d, k/: good, hood (plus its use as a suffix, e.g. childhood – but for hoodlum see above), stood, wood (and its derivative woodbine); book, brook, cook, crook, hook, look, nook, rook, shook, took (exceptions: could, should, would; pud, suk). The high percentage of <oo> spellings despite it occurring in so few words is due to some of those words having very high frequency.

<u> is regular everywhere except in the <oo> words and Oddities listed above. However, there are only about 57 stem words in this set in RP: ambush, Buddha, buffet /ˈbʊfeɪ/ (‘food’), bulbul (twice), bull, bullace, bullet, bulletin, Bullingdon, bullion, bully, bulrush (first <u>), bulwark (also pronounced with /ʌ/), bush, bushel, butch, butcher (but one of the teachers of English at my grammar school in the 1950s said /ˈbʌʧə/), cuckoo, (mea) culpa, cushion, cushty, cushy, ebullient (also pronounced with /ʌ/), fulcrum (both <u>’s), full, fulmar, fundi (/ˈfʊndiː/ South and East African English for ‘expert/skilled person’/in Britain, a member of the fundamentalist, uncompromising wing of the German Green Party; contrast fundi /ˈfʌndaɪ/, plural of fundus ‘inner corner of organ’), gerenuk, kaput, kibbutz, kukri, lungi, lutz, mullah, mush (/mʊʃ/, slang for ‘friend’), muslim, Musulman (twice), pud, pudding, pull, pullet, pulpit, push, puss, put, putsch, schuss, s(c)htum, shufti, sputnik, sugar, suk, Sunni, thurible, thurifer, thruppence, tuk-tuk (twice), umlaut (first <u>), Zumba, plus derivatives including Buddhism, bullock, fulfil, fully, ful(l)ness, fulsome, and the adjective/noun suffix /fʊl/ spelt <-ful> - there are at least 150 words so formed, e.g. beautiful, handful. Unstressed in that suffix and otherwise only in ambush, fulcrum (second syllable), fulfil, gerenuk, tuk-tuk (second syllable).

For elision of the /ʊ/ when /liː/ spelt <-ly> is added to adjectives in <-ful> to form adverbs see section 6.10.

5.4.7 /ə/ (the schwa vowel) as in the first sound
in about

The most frequent phoneme in spoken English.

The only short vowel which does occur word-finally, in my opinion/version of RP.

Occurs only in unstressed syllables, except in the increasingly frequent pronunciation of because as /bɪˈkəz/.

Rare before /ŋ/ - see section 3.8.2.

N.B. Where the schwa vowel is part of a diphthong it is dealt with elsewhere – see /eə, ɪə, ʊə, əʊ/, sections 5.6.3-5, 5.7.4. For the so-called ‘triphthongs’ /aɪə/ (which I analyse as a two-syllable sequence consisting of a diphthong plus schwa) and /aʊwə, ɔɪjə/ (which I analyse as two-syllable sequences consisting of a diphthong plus automatic intervening /w/- or /j/-glide plus schwa), see sections 5.7.3, 5.6.2 and 5.6.1 respectively. For this treatment of triphthongs, see also Cruttenden (2014: 153).

For all categories see also Notes, and for further guidance sections 6.7-9.

The main system

Basic grapheme

<a>

35%

e.g. about. Regular in initial and medial positions. Especially prevalent in initial position, where the only exceptions appear to be words formed from the Latin prefix ob- and its derivatives, e.g. obscure, obtuse, occur, offend, but medial position is much more variable

Other frequent graphemes

<o>

19%

e.g. Burton, obscure

<er>

15%

e.g. alter. Regular in word-final position and in the prefixes hyper-, inter-, per-, super- when not stressed on <er>. All these prefixes permit /r/-linking (see section 3.6) before stems beginning with a vowel phoneme, e.g. hyper-active, interactive, peroxide, supererogatory

<e>

13%

e.g. artery

Rare 2-phoneme grapheme

/əl/

spelt <-le>

6%

only word-final and only in this reversed spelling, e.g. able, possible. Though not very frequent as a correspondence for /ə/ this counts as part of the main system because of its higher frequency as a correspondence for /l/ - see section 3.6.5

The rest

Oddities

12% in total. Of all those listed, only <ar, or, ur> occur in both medial and final positions. None occur in initial position (see above under Basic grapheme, and Notes)

- in medial position

<ai>

only in certain, chieftain, coxswain, curtain, mainsail (second syllable), topsail, villain

<anc>

only in blancmange /bləˈmɒnʤ/

<ar>

regular in the suffixes /wəd(z)/ spelt <-ward(s)>, e.g. afterwards, backward(s), downward(s), forward(s), froward, inward, leeward, onward, outward, windward, and predominant in the ending /əd/ more generally – see Notes. Otherwise in an unpredictable ragbag of words, e.g. anarchy, awkward, bastard, billiards, blackguard pronounced /ˈblægəd/ (also pronounced /ˈblægɑːd/), bombardier, bulwark, coward, custard, dotard, gabardine, halyard, innards, lanyard, monarch, mustard, niggardly, orchard, scabbard, stalwart, steward, vineyard, wizard

<eau>

only in bureaucrat(ic)

<ei>

only in foreign

<eo>

only in bludgeon, curmudgeon, dudgeon, dungeon, gudgeon, luncheon, puncheon,
(e)scutcheon, smidgeon, sturgeon, surgeon, truncheon, widgeon. See Notes

<eu>

only in pasteurise pronounced /ˈpaːsʧəraɪz/

<i>

in a large number of adjectives ending in /əbəl/ spelt <-ible> where the stem without the /əbəl/ mostly does not sound like a real word, e.g. possible. See Basic grapheme <a> above, Notes and section 6.7. Also in a few adverbs ending <-arily> when not stressed on the <a>, which becomes elided (see section 6.10), so that the <i> in <-ily> spells /ə/, e.g. necessarily, voluntarily pronounced /ˈnesəsrəliː, ˈvɒləntrəliː/ (see also under /e/. section 5.4.2)

<ia>

only in fuchsia, miniature, parliament. In words like crucial, initial and in Christian I count the <i> as part of a digraph with the preceding consonant letter – see /∫, t∫/, sections 3.8.3, 3.7.2

<io>

only in cushion, fashion, marchioness, stanchion. In words like question, nation, lesion, vision, lotion, fusion I count the <i> as part of a digraph with the preceding consonant letter – see /t∫, ∫, ʒ/, sections 3.7.2, 3.8.3-4

<oar>

only in cupboard, larboard, starboard

<oi>

only in connoisseur, porpoise, tortoise

<or>

2% regular medially in prefix /fə/ spelt <for->, e.g. forbid, forget, forgive, forsake (but this is a very small set); otherwise rare medially, but cf. Deptford (and many other placenames with this element), Holborn, scissors, stubborn

<ou>

regular in adjectives ending in /əs/ spelt <-ous>, e.g. anxious, famous. Otherwise only in camouflage, doubloon, limousine, moustache, tambourine, vermouth pronounced /ˈvɜːməθ/ (also pronounced /vəˈmuːθ/)

<ow>

only in Meadowhall (locally, in Sheffield), sorrowful

<u>

regular in unstressed prefix /səb/ spelt <sub->, e.g. subdue, subject (verb, pronounced /səbˈʤekt/), sublime, submerge, submit, subside, subsist, substantial; also in nouns ending in unstressed /əs/ spelt <-us>. Otherwise in, e.g., bogus, capitulate, cherub, commensurate, congratulate, conjugate, glandular, modular, naturist, petulan-t/ce, postulant, spatula

<ua>

in nouns, only in actuary, estuary, mortuary, obituary, sanctuary, statuary, voluptuary, when pronounced with /ʧəriː/ rather than /ʧʊəriː/ (see also under /ʧ/, section 3.6.2), plus casualty, February, victuals /ˈkæʒəltiː, ˈfebrəriː, ˈvɪtəlz/; also often in rapid pronunciation of adjectives like actual (see again /ʧ/, section 3.6.2), sexual and especially adverbs derived from them (see also section 6.10 on elided vowels)

<ur>

perhaps usual only in Saturday, surprise, but there are several words which may have either /ɜː/ (see section 5.5.3) or /ə/, e.g. liturgy, metallurgy, saturnine, surmise, surmount, surpass, survey (verb), survive

<y>

only in pyjama(s)

- in final position

<ah>

only in ayah, cheetah, fellah, haggadah, hallelujah, Hannah, loofah, messiah, moolah, mullah, mynah, pariah, purdah, (maha)rajah, Sarah, savannah, verandah, wallah and some other very rare words

<ar>

only in an unpredictable ragbag of words, e.g. altar, beggar, briar, burglar, cedar, cellar, cochlear, collar, columnar, curricular, familiar, friar, fulmar, globular, jugular, liar, linear, lumbar, lunar, molar, nuclear, particular, peculiar, pedlar, peninsular, planar, polar, popular, regular, scalar, scapular, scholar, sugar, titular, vicar, vulgar. Many such words permit /r/-linking, e.g. polarise, polarity – see section 3.6

<ere>

only in were when unstressed

<eur>

only in amateur, chauffeur (if stressed on first syllable and pronounced /ˈʃəʊfə/), grandeur. /r/-linking occurs in amateurish – see section 3.6

<or>

regular in nouns formed from verbs in <-ate>, e.g. administrator, agitator, alternator, commentator, creator, curator, dictator, elevator, incinerator, insulator, orator, spectator, including cases where the verb is rare, e.g. aviator, plus groups ending in /ktə, esə, ɪtə/ spelt <-ctor, -essor, -itor>, e.g. actor, conductor, constrictor, detector, reactor; aggressor, assessor, compressor, confessor, depressor, predecessor, possessor, professor, successor; capacitor, depositor, editor, inhibitor. Otherwise only in an unpredictable ragbag of nouns, e.g. advisor (also spelt adviser), camphor, conspirator, conqueror, contributor, conveyor, councillor, counsellor, distributor, donor, emperor, error, horror, incisor, inventor, languor, liquor, metaphor, pallor, pastor, phosphor, rotor, sailor, sponsor, squalor, stupor, suitor, survivor, terror, tormentor, torpor, traitor, tutor, plus <or> in rigor only in the Latin phrase rigor mortis

<ough>

only in borough, thorough

<our>

only in an unpredictable ragbag of words, e.g. arbour, ardour, armour, behaviour, candour, clamour, clangour, colour, endeavour, favour, fervour, flavour, glamour, harbour, honour, humour, labour, neighbour, odour, parlour, rancour, rigour (but <or> in the Latin phrase rigor mortis), rumour, saviour, splendour, succour, tumour, valour, vapour, vigour. In many of these words US spelling has <or>

<re>

only in an unpredictable ragbag of words, e.g. accoutre, acre, calibre, centre, chancre, fibre, goitre, litre, louvre, lucre, lustre, manoeuvre, massacre, meagre (contrast eager), mediocre, metre and its compounds, e.g. kilometre (contrast meter and its compounds, e.g. barometer), mitre, ochre, ogre, reconnoitre, sabre, saltpetre, sceptre, sepulchre, sombre, spectre, theatre, timbre. In many of these words US spelling has <er>. None of these words has a /r/ phoneme in the final syllable (in RP), but when a suffix beginning with a vowel is added, some lose /ə/ and have /r/-linking instead; e.g. centre /ˈsentə/ plus /əl/ becomes /ˈsentrəl/ (central) - see section 3.6. In accoutrement the schwa disappears and two phonemes surface: /r/ spelt <r> and /ɪ/ represented by the first <e> - see section 7.2. In acreage, massacreing, ochreous, ogreish /ˈeɪkərɪʤ, ˈmæsəkərɪŋ, ˈəʊkərəs, ˈəʊgərɪʃ/ /r/ also surfaces, but the schwa and /r/ seem to be represented by <e, r> in reverse order - see again section 7.2. Even more difficult to analyse is manoeuvrer if pronounced /məˈnuːvərə/, where no letter seems to spell the first schwa – but, as Gödel proved, no formal system can be both complete and consistent

<ur>

only in augur, femur, langur, lemur, murmur (second syllable), sulphur

<ure>

almost all examples of word-final /ʧə/ are spelt <-ture>, e.g. architecture, capture, caricature, conjecture, creature (contrast preacher, teacher), culture, curvature, departure (contrast archer, marcher), expenditure, feature (again

contrast preacher, teacher), fixture, fracture, furniture, future, gesture, juncture, lecture, legislature, literature, manufacture, miniature, mixture, moisture, nature, nurture (contrast lurcher, (re)searcher), pasture, picture (contrast pitcher), posture, puncture, rapture, rupture, scripture, sculpture, signature, stature, stricture, structure, temperature, texture, tincture, torture, (ad)venture, vulture. Other examples of final /ə/ spelt <-ure> include censure, conjure (‘do magic tricks’) pronounced /ˈkʌnʤə/, figure, injure, leisure, measure, perjure, pleasure, pressure, procedure, seizure, tonsure, treasure, verdure (cf. verger); also in azure pronounced /ˈæʒə, ˈeɪʒə/ (also pronounced /ˈæzjə, ˈeɪzjə, ˈæzjʊə, ˈeɪzjʊə/). See Notes

<yr>

only in martyr, satyr, zephyr

2-phoneme graphemes

For /aɪə/

spelt <ir, ire, yr, yre> and /waɪə/ spelt <oir>

see under /aɪ/, section 5.7.3

For /aʊwə/

spelt <hour, our>

see under /aʊ/, section 5.6.2

For /ɔɪjə/

spelt <oir>

see under /ɔɪ/, section 5.6.1

/əl/ spelt <l>

only in axolotl, dirndl, shtetl. See /l/, section 3.7.5

/əm/ spelt <m>

see /m/, section 3.4.4

/ən/ spelt <n>

see /n/, section 3.4.5

/jə/

(1) spelt <eu>

only in aneurism/aneurysm, pasteurise pronounced /ˈpaːstjəraɪz/ (also pronounced /ˈpaːsʧəraɪz/)

(2) spelt <u>

frequent in unstressed penultimate syllables of words of three or more syllables stressed on the antepenultimate syllable, e.g. amulet, angular, argument, calculate, chasuble, coagulate, contributor, corpuscular, distributor, emulate, fabulous, garrulous, immunise, inaugural, incubus, insula-r/te, jugular, manipulate, muscular, nebulous, particular,

penury, popul(o)us, querulous, regula-r/te, scapula(r), scroful-a/ous, scrupulous, stimul-ant/ate/us, succubus, succulent, tremulous, truculent, vernacular; also in antepenultimate syllable of copulation, population with stress on following syllable. Where the preceding consonant is /d, t/ the sequences /dj, tj/ affricate to /ʤ, ʧ/ (see sections 3.7.4, 3.7.2 and cf. pasteurise above), e.g. in (in)credulous, fraudulen-ce/t, glandular, modul-e/ar, nodul-e/ar, pendulum, sedulous; century, congratulate, fistula, flatulen-ce/t, fortunate, petulan-t/ce, postulant, postulate, saturate, spatula, titular

(3) spelt <ua>

in my analysis, only in January, valuable – but see the discussion of words with <u, a> under <u>, section 10.36

(4) spelt <ure>

only in failure, tenure and azure pronounced /ˈæzjə, ˈeɪzjə/ (also pronounced /ˈæzjʊə, ˈeɪzjʊə, ˈæʒə, ˈeɪʒə/)

3-phoneme grapheme

/waɪə/
spelt with a single grapheme <oir>

only in choir – one of only two 3-phoneme graphemes in the entire language

Notes

Under /ɪə/ in section 5.6.4 you will see that I disagree with Carney’s analysis of that phoneme and have therefore re-allocated a large number of words to /iː/ plus /j/-glide plus /ə/. However, this has not added any graphemes to the correspondences for /ə/. I have left Carney’s percentages for /ə/ unchanged on the assumption that the distribution of its correspondences within his analysis of /ɪə/ is broadly similar to that within his analysis of /ə/.

The articles a, the are pronounced /ə, ðə/ before consonant phonemes in running speech, and sometimes also when pronounced as citation forms – and therefore stressed, thus also counting as partial exceptions to the rule that /ə/ occurs only in unstressed syllables. But they also have the alternative citation forms /eɪ, ðiː/, which are not exceptions. Other function words which have /ə/ in running speech, e.g. to, was, were pronounced /tə, wəz, wə/, are never so pronounced as citation forms, which are instead /tuː, wɒz, wɜː/.

The reason for the wide range of spellings for /ə/ is that any vowel, however spelt for its full pronunciation, can be reduced to the non-distinctive schwa in an unstressed syllable. The default spellings are <a> in initial and medial positions and <er> in final position, and some guidance can be given for a few major categories, but there are very many words that just have to be learnt – see the Oddities above and the various ragbag lists there and in these Notes.

1. Initial position

Here the hugely predominant spelling is <a>, and this applies both to the native English prefix a- (historically derived from on), e.g. in abide, aboard, about, ahead, alight, aside, athwart, away, and to derivatives of the Latin prefixes ab-, ad-, e.g. in abrupt, abhor, abound, acclaim, accost, accuse, acquire, address, adhere, adopt, affirm, aggressive, allure, annul, appear, assure, attend, aver; also in some words of other origins, e.g. (Greek) anaemia, anathema, aroma.

The only set of exceptions appears to be words with /ə/ spelt <o> in the Latin prefix ob- and its derivatives, e.g. oblige, obscene, obscure, observe, obsess, obtain, occasion, occur, offend, official.

2. Medial position

Again the default spelling is <a>, though less strongly than in initial position. Some patterning can be seen in initial and final word elements, but very little otherwise in medial position.

2.1 Medial position in prefixes/initial elements

A few guidelines can be given for when a schwa here is not spelt <a>:

  • the prefixes /ˈhaɪpə, ˈɪntə, ˈsuːpə/ are almost always spelt <hyper-, inter-, super->
  • the unstressed prefixes /kən (and related forms), prə, tə/ are spelt with <o>, e.g. collect, collide, command, commit(tee), confess, connect, connive, connubial, contrast (verb, pronounced /kənˈtrɑːst/), corrode, corrupt; procure, produce, profane, profess(or), prolong; today, together, tomorrow
  • there are several words beginning <chloro-, micro-, mono-, phono-, photo-, saxo-> where the stress is on the first syllable and the schwa in the second syllable is spelt <o>.

2.2 Medial position in suffixes/endings/final syllables

The ending /ət/ in many nouns and adjectives is almost always spelt <-ate> - see the list of about 90 words under /t/ spelt <te>, section 3.5.7 (exceptions: chariot, idiot, patriot).

For those who say /ɪtɪv/ for words ending <-itive> the ending /ətɪv/ is always spelt <-ative>.

The adjectival ending /əbəl/ is mainly spelt <-able> in words where the unsuffixed form sounds like a real word, and mainly <-ible> where it doesn’t, but there are numerous exceptions (see section 6.7).

The adjective-forming suffix /əl/ is usually spelt <-al>, e.g. central, liberal, loyal, royal; (ar)boreal, cereal, corporeal, ethereal, funereal, marmoreal, sidereal, venereal; congenial, editorial, industrial, jovial, managerial, material, memorial, radial, remedial, serial and about 450 others ending in <-ial>. For the various spellings of final /əl/ see also sections 4.4.3 and 4.4.2-3.

There are fairly clear rules for word-final /əm/:

  • if preceded by /d/ the ending is usually the noun-forming suffix /dəm/ spelt <-dom>, e.g. kingdom, thral(l)dom, wisdom (exceptions: agendum, carborundum, macadam, madam, referendum, sedum, tandem)
  • if preceded by /z/ the spelling is almost always <sm> (only exception: bosom). See under /m/, section 3.4.4
  • if preceded by /s/ the ending is usually adjectival /səm/ spelt <-some>, e.g. handsome (exceptions: balsam, flotsam, jetsam; besom, blossom, buxom, hansom, lissom, ransom, transom)
  • otherwise word-final /əm/ is usually spelt <-um>, e.g. atrium, bacterium, compendium, delirium, gymnasium, medium, opium, potassium, radium, stadium, tedium and about 200 others ending in <-ium>, plus album, colosseum, linoleum, lyceum, mausoleum, maximum, museum, petroleum, rectum (exceptions: algorithm, rhythm; amalgam, bantam, bedlam, buckram, gingham, marjoram; anthem, emblem, item, problem, stratagem, system, theorem, totem; atom, axiom, bottom, custom, fathom, idiom, maelstrom, phantom, pogrom, symptom, venom).

There are fairly clear rules for word-final /əs/:

  • in adjectives the spelling is almost always <-ous>, e.g. famous and at least 2000 others (only exceptions: bogus, emeritus)
  • in nouns the spelling is almost always <-us>, e.g. abacus, anus, bonus, cactus, campus, caucus, census, chorus, circus, citrus, corpus, crocus, discus, exodus, focus, fungus, genius, genus, hiatus, hippopotamus, isthmus, litmus, lotus, octopus, onus, nucleus, radius, rhombus, stimulus, surplus, syllabus, Taurus, terminus, tinnitus, virus and hundreds more (exceptions: (some of which are also pronounced with /ɪs/): furnace, menace, necklace, palace, pinnace, populace, solace, surface, terrace; alias, bias, Candlemas, canvas, Christmas, Lammas, Martinmas, Michaelmas; carcase/carcass, purchase; canvass, trespass, windlass; purpose; porpoise, tortoise)
  • there seem to be only five pairs of adjective/noun homophones which differ only in the spelling of the /əs/ ending: callous/callus, mucous/mucus, populous/populace, rufous/Rufus, venous/ Venus, though of course nouns which are rank-shifted to modifier position before other nouns retain the <-us> spelling: chorus line, citrus fruit, litmus test
  • there seem to be only two words ending /əs/ which exist only as verbs: embarrass, harass (pronounced /ˈhærəs/ rather than the more recent /həˈræs/); the spellings of the other few verbs ending /əs/ are the same as the related nouns: menace; bias; purchase; canvass, trespass; chorus, focus.

The ending /əd/ is usually spelt <-ard>, e.g. awkward, bastard, blackguard pronounced /ˈblægəd/ (also pronounced /ˈblægɑːd/), coward, custard, dotard, halyard, lanyard, mustard, orchard, scabbard, steward, vineyard, wizard and see Oddities above for the suffixes /wəd(z)/ spelt <-ward(s)> (exceptions; method, period, synod).

The endings /ək, əp/ are usually spelt <-ock, -op>, e.g. bollock, bullock, buttock, hassock, hillock, mattock, pillock, rowlock (exception: bulwark); bishop, gallop, wallop (exceptions: catsup, chirrup, ketchup, stirrup, syrup).

In the suffix spelt <-ology>, the schwa after /l/ is always spelt <o>, e.g. biology, chronology.

In the suffix spelt <-ological>, the schwa before the first /l/ is always spelt <o>, e.g. biological, chronological, and the second one always <a>.

The ordinal numeral-forming suffix /əθ/ is always spelt <-eth> in twentieth,, ninetieth.

Beyond various words listed under the medial Oddities <ai, ei, eo, io, or> there are some fairly clear rules for word-final /ən/:

  • in the various endings pronounced /ʃən/, all words with <-si*n, -ti*n> have <o> for the schwa except Asian, Persian, Prussian, Russian, gentian, Titian; all words with <-ci*n> have <a> for the schwa except coercion
  • the spelling <-on> otherwise occurs mainly in nouns, e.g. bacon, Briton, button, carton, chameleon, cotton, galleon, halcyon, matron, melodeon, mutton, Odeon, person, piston, siphon/syphon, wanton, plus a set of words in <-ion>: accordion, aphelion, bastion, battalion, billion, bullion, carrion, centurion, champion, clarion, collodion, companion, criterion, dominion, ganglion, medallion, million, mullion, minion, oblivion, onion, opinion, pavilion, perihelion, pinion, rebellion, scorpion, scullion, stallion, union
  • the irregular past participle ending /ən/ (that is, when the ending is pronounced as a full syllable, namely after a consonant phoneme) is spelt <en>, e.g. (for)bidden, bitten, broken, chosen, eaten, fallen, forsaken, frozen, (for)given, hidden, (a)risen, spoken, stolen, swollen, (mis)taken, (a)woken, woven, written, even in fossilised forms where the stem verb is now regular or its past participle is disused or used only adjectively, e.g. beholden, bounden, brazen, cloven, drunken, graven, ((mis)be/ill-)gotten, laden, molten, proven, (bed)ridden, riven, rotten, (mis)shapen, shaven, shriven, shrunken, smitten, stricken, stridden, striven, thriven, (down)trodden
  • <en> also occurs in, e.g.; alien, dozen, even, flaxen, garden, golden, happen, heaven, listen, open
  • <an> occurs in the noun/adjective ending /ən/ in antipodean, caesarean, cyclopean, empyrean, epicurean, euclidean, European, galilean, Herculean, Jacobean, Linnaean, Manichaean, paean, pythagorean; plebeian; barbarian, comedian, grammarian, guardian, historian, pedestrian, reptilian, ruffian, thespian and about 200 other words ending in <-ian>
  • But the endings /ənt, əns, ənsiː/ have the variant spellings <- ant/-ent, -ance/ -ence, -ancy/-ency> - see section 6.8.

2.3 Otherwise in medial position

The default spelling is still <a>, e.g. sole <a> in buffalo, dynamo, seraph, theatre; first <a> in banana, bravado, farrago, mama, palaver, papa, staccato; second <a> in archipelago, balaclava, ballast, breakfast. Exceptions:

  • with <e> include artery, bolero (/ʽbɒlərəʊ/ ‘garment’), soviet, first <e> in coterie;
  • with <o> include abdomen, acrobat, aphrodisiac, bolero (/bəʽleərəʊ/, ‘dance’), cellophane, cenotaph, custody, daffodil, espionage, exodus, geographic, iodine, kaolin, lobelia, mandolin, mimeograph, parody, police, purpose, ricochet, second, theocratic, violate, vitriol; first <o> in creosote, stereophonic, tobacco; second <o> in broccoli, choreographic, colloquy, obloquy, rollocking.

And see again the medial Oddities, above.

3. Final position

The default spelling is <-er>. Examples include: amber, arbiter, auger, bitter, brother, cancer, character, chipper, chorister, clover, double-decker, eager, ember, knocker, ladder, laager, lager, lever, Londoner, lumber, mother, neuter, number, other, oyster, proper, slander, slender, sober, thunder, timber, tuber, water, yonder; all comparative adjectives, e.g. better, brighter, colder, dearer, easier, happier; most agentive nouns formed from one-syllable verbs, e.g. drinker, jumper, killer, roamer, runner, speller, viewer (exceptions: actor, sailor); many longer agentive nouns where <e/y>-deletion (see sections 6.4, 6.6) applies, e.g. astrologer, astronomer, biographer, commuter, diner, geographer, lover, philosopher, remembrancer, settler, subscriber; also words with the suffix <-ometer> (‘measuring device’), e.g. barometer, thermometer (contrast kilometre pronounced /ˈkɪləmiːtə/ to rhyme with metre and all its other compounds; however, kilometre is also pronounced /kɪˈlɒmɪtə/ to rhyme with all the words ending <-ometer>).

Exceptions (in addition to the Oddities, above):

  • where the schwa vowel is spelt within a 2- or 3-phoneme grapheme: see those headings above
  • spellings with <e>: genre, macabre (which appear to be the only two words where final <-re> is pronounced /rə/ rather than /ə/ - contrast the Oddities in <-re> listed above), the (unstressed before a word beginning with a consonant phoneme), lasagne. There seem to be very few words in this set
  • spellings with <a>:

    1) agenda, arcana, automata, bacteria, corrigenda, criteria, curricula, data, desiderata, ephemera, erotica, errata, esoterica, exotica, fauna, flora, fora, insignia, juvenilia, maxima, media, memorabilia, memoranda, militaria, millennia, miscellanea, minima, opera, phenomena, prolegomena, pudenda, referenda, schemata, stigmata, strata, trivia, which etymologically are all Latin or Greek neuter plural nouns (though agenda, opera are now always singular in English, increasingly data, media are too, and bacteria, criteria are often used as singulars by people who don’t know that their singulars are bacterium, criterion)

    2) also in a set of exotic loanwords of three or more syllables stressed on the penultimate syllable, e.g. abscissa, alfalfa, alpaca, amenorrhoea, amoeba, anaconda, angina, angora, antenna, arena, aroma, aspidistra, aurora, balaclava, balalaika, ballerina, banana, bandanna, belladonna, bonanza, bravura, cadenza, candelabra, carcinoma, cassava, cavatina, cedilla, chim(a)era, chinchilla, chorea, cicada, concertina, conjunctiva, corona, cyclorama, diarrhoea, dilemma, diploma, duenna, emphysema, enigma, eureka, extravaganza, farina, felucca, flotilla, glaucoma, gonorrhoea, gorilla, granadilla, guerrilla, gymkhana, hacienda, hegira, hosanna, hydrangea, hyena, idea, iguana, indaba, influenza, koala, lactorrhoea, lacuna, liana, logorrhoea, Madonna, magenta, mahatma, manila, mantilla, mazurka, madeira, miasma, mimosa, nirvana, (o)edema, ocarina, omega pronounced /əʊˈmiːgə/, operetta, pagoda, panacea, panatella, panorama, pashmina, patella, patina pronounced /pəˈtiːnə/, penumbra, persona, pharmacopoeia, pianola, placenta, propaganda, protozoa, pyorrhoea, regatta, rotunda, rubella, saliva, sarcoma, savanna, scintilla, semolina, siesta, sonata, sultana, syringa, tapioca, tiara, toccata, tombola, trachea, umbrella, urea, urethra, vagina, Valhalla, vanilla, vendetta, veranda, verbena, verruca, viola (/viːˈjəʊlə/ ‘musical instrument’)

    3) also in a set of loanwords of two syllables stressed on the first syllable, e.g. alpha, asthma, aura, china, cobra, coda, coma, comma, contra, copra, delta, diva, dogma, drama, eczema, era, extra, fatwa, gala, gamma, geisha, guava, gurkha, halma, henna, hydra, junta, karma, lama, lambda, lava, lemma, libra, llama, magma, manna, mantra, nova, okra, ouija, panda, pasha, plasma, plaza, polka, pukka, puma, pupa, quagga, quota, rhea, rota, saga, schema, skua, soda, sofa, stanza, stigma, tantra, toga, trauma, tuba, tufa, tuna, tundra, ultra, villa, visa, vista, viva, vodka, vulva, yoga, yucca, zebra, zeugma

    4) and a further ragbag of words which fit none of those categories, e.g. algebra, ammonia, anaemia, anaphora, anathema, apnoea, area, azalea, begonia, camellia, camera, chlamydia, cholera, cinema, cithara, cochlea, copula, cornea, cupola, dyspnoea, (en)cyclopaedia, enema, formula, gondola, harmonica, hernia, hysteria, japonica, myopia, nausea, omega pronounced /ˈəʊmɪgə/, orchestra, parabola, patina pronounced /ˈpætɪnə/, peninsula, pergola, plethora, primula, replica, retina, salvia, scapula, sciatica, scrofula, sepia, stamina, swastika, taffeta, tarantula, tempera, utopia, vertebra, viola (/ˈvaijələ/ ‘flower/girl’s name’).

Any spelling for final /ə/ which ends in <r, re> allows /r/-linking, e.g. central, ethereal, managerial, terrorist, authority, authorial, favourite, calibration, fibrous, leverage, polarise, cigarette (with movement of the stress), vicarious, vulgarity, dictatorial, rigorous (with deletion of the <u> from the final syllable of the stem), theatrical, sulphuric, injurious, adventurous – see section 3.6.

5.5 Long pure vowels (other than /iː, uː/): /ɑː ɜː ɔː/

5.5.1 /ɑː/ as in aardvark

The main system

Basic grapheme

<ar>

60%

e.g. farther

Other frequent grapheme

<a>

34%

e.g. father. More frequent in RP than in other accents. Regular before consonant clusters, but also occurs elsewhere. See Notes

The rest

Oddities

6% in total

<aa>

only in baa, Baal, Graal, kraal, laager, naan, salaam

<aar>

only in aardvark, aardwolf, bazaar, haar

<a.e>

only in final syllables and only in about 30 (mostly more recent French) loanwords, namely ballade, charade, chorale, façade, gouache, grave (/grɑːv/, ‘French accent’), locale, morale, moustache, promenade (noun, ‘seafront path’; the verb with the same spelling, ‘walk at leisure’, is pronounced with /eɪ/), rationale, strafe, suave, timbale, vase, plus a set of words ending in /ɑːʒ/ spelt <-age>, e.g. badinage, barrage, camouflage, collage, corsage, decalage, décolletage, dressage, entourage, espionage, fuselage, garage pronounced /ˈgærɑːʒ/, massage, menage, mirage, montage, triage, sabotage (only exception to final /ɑːʒ/ spelt <-age>: raj). The <e> in chorale, locale, morale, rationale differentiates those words visually from choral, local, moral, rational

<ah>

only word-final and only in ah, bah, hookah, hoorah, kabbalah, Shah, whydah

<al>

only in calf, half; calve(s), halve(s), salve(s) (also pronounced /sælv(z)/); almond, almoner, alms, balm, calm, embalm, malmsey, napalm, palm, psalm, qualm

<are>

only in are when stressed

<arr>

only in bizarrery, carr, charr, parr

<arre>

only in barre, bizarre. /r/-linking occurs in bizarrery – see section 3.6

<arrh>

only in catarrh. /r/-linking occurs in catarrhal – see section 3.6

<as>

only in fracas

<at>

only in eclat, entrechat, nougat

<au>

only in aunt, draught, laugh(ter)

<ear>

only in hearken (also spelt, more regularly, harken), heart, hearth

<er>

only in Berkeley (the town in England), Berkshire, Cherwell, clerk, derby, Derby, Ker pronounced /kɑː/ (also pronounced /kɜː/), sergeant

2-phoneme graphemes

/wɑː/

See also Notes

(1) spelt <oi>

only in a few words more recently borrowed from French, e.g. bourgeoisie, coiffeu-r/se, coiffure, pointe, soiree, toilette

(2) spelt <oir>

mainly word-final and only in a very few words more recently borrowed from French, namely abattoir, boudoir, memoir, reservoir, voussoir; non-finally, only in avoirdupois. /r/-linking occurs in memoirist, noirish – see section 3.6

(3) spelt <oire>

only word-final and only in a very few words more recently borrowed from French, namely aide-memoire, conservatoire, escritoire, repertoire

(4) spelt <ois>

only word-final and only in a very few words more recently borrowed from French, namely avoirdupois, bourgeois (/z/ surfaces in bourgeoisie – see section 7.2), chamois (the animal, pronounced /ˈ∫æmwɑː/, as opposed to the leather made from its skin, pronounced /ˈ∫æmiː/, the latter also being spelt shammy), patois (contrast fatwa)

Notes

If we follow Crystal (2012: 131-2) and Upward and Davidson (2011: 176-9), ‘more recent’ in terms of loanwords from French means after the Great Vowel Shift, which began about AD1400 and was complete by about AD1600.

In RP, <a> is regular before consonant clusters, e.g.

  • (in monosyllables) aft, craft, graft, haft, raft, shaft (exception: draught); chance, dance, glance, lance, prance, trance; ranch; can’t, chant, grant, plant, shan’t, slant (exception: aunt); ask, bask, cask, flask, mask, task; clasp, gasp, grasp, hasp, rasp; basque, masque; blast, cast, caste, fast, last, mast, past, vast (other exception: alms);
  • (in final syllables of polysyllables) abaft, advance, enhance, avalanche, command, countermand, demand, remand, enchant, bergomask, aghast, contrast (noun and verb);
  • (in non-final syllables) macabre; padre; after, rafter; example, sample; chancel, chancery; revanchis-m/t; commando, slander; answer; basket, casket; bastard, caster, castor, disaster, flabbergasted, ghastly, master, nasty, pasta (also pronounced with /æ/), pasteurise, pastime, pastor, pasture, plaster (exceptions: aardvark, aardwolf, laughter, malmsey).

Otherwise, in non-rhotic accents such as RP no rules can be given for where /ɑː/ is spelt <a> rather than <ar>, so here are some lists of words where /ɑː/ spelt <a> occurs:

  • several words before medial /ð/ spelt <th>, e.g. father, lather, rather (exception: farther), and before final /f, s, θ/ spelt <-ff(e)/-ph, -ss, -th>, e.g. chaff, distaff, staff; giraffe; cenotaph, and graph and all its unsuffixed compounds: auto/cardio/ di/encephalo/epi/mimeo/para/photo/tele/tri-graph (exceptions: calf, half); brass, class, glass, grass, pass (exception: arse); bath, path (exception: hearth)
  • word-finally in bra, hoopla, Libra, (grand)ma, mama, (grand)pa, papa, schwa, spa (contrast several of the Oddities)
  • a large set of words, many of them loanwords, but all ending in a vowel phoneme and with stressed /ɑː/ spelt <a> in the penultimate syllable, e.g. armada, avocado, balaclava, banana, blasé (sometimes stressed on last syllable), bravado, bravo (sometimes stressed on last syllable), cadre, cantata, cascara, cassava, cicada, cinerama, cyclorama, desiderata, desperado, drama, farrago, finale, gala, Gestapo, guano, guava, gymkhana, iguana, incommunicado, karate, khaki, lager, lama, lava, legato, liana, literati, llama, llano, marijuana, mascara, meccano, nazi, pajama, palaver, panorama, pastrami, plaza, praline, pro rata, pyjama, safari, saga, salami, schemata, sonata, soprano, staccato, stigmata, strata, sultana, tiara, toccata, tomato, tsunami, virago
  • a final ragbag: adagio (second <a>), amen, banal (second <a>), castle, claque, corral, debacle, fasten, plaque, pajamas (second <a>), pyjamas (first <a>).

Words in which final /ɑː/ is spelt <-ar> allow /r/-linking, e.g. far away /fɑːrəˈweɪ/, sometimes with <r>-doubling, e.g. sparring – see section 3.6.

/wɑː/ has the 2-grapheme spelling <ua> in guacamole, guano, guava, iguana, suave.

5.5.2 /ɜː/ as in earl

The main system

For all these categories see Notes.

Basic grapheme

<er>

38%

e.g. berth, exert, herd, serf, sherd, tern, twerp; defer, infer, prefer, refer

Other frequent graphemes

<ir>

18%

e.g. birth, fir, whirl

<or>

17%

regular after initial /w/, e.g. word

<ur>

17%

e.g. fur, gurgle, surf, turn, urn

The rest

Oddities

10% in total

<ear>

8% never word-final, and only in dearth, earl, early, earn, earnest, earth, heard, hearse, learn, pearl, rehearse, (re)search, yearn

<ere>

only in were when stressed

<err>

in stem words only in err (for which see also section 4.3.2), but frequent in consonant-doubling before suffixes, e.g. preferred (see section 4.2)

<eu>

only in chauffeuse, coiffeuse, masseuse, milieu

<eur>

non-finally, only in secateurs; otherwise only word-final and only in about 12 recent loanwords of French origin, e.g. chauffeur (if stressed on second syllable, pronounced /ʃəʊˈfɜː/), coiffeur, connoisseur, entrepreneur, hauteur, masseur, poseur, provocateur, raconteur, repetiteur, restaurateur, seigneur and a few other rare words

<irr>

only in chirr, shirr, whirr

<olo>

only in colonel

<our>

only medial, and only in adjourn, bourbon (‘whiskey’), courteous, courtesy, journal, journey, scourge, and tourney pronounced /ˈtɜːniː/ (also pronounced /ˈtʊəniː)

<urr>

in stem words only in burr, purr, but frequent in consonant-doubling before suffixes, e.g. furry, demurring, occurred (see section 4.2)

<yr>

only in gyrfalcon, myrmidon, myrtle

<yrrh>

only in myrrh

2-phoneme graphemes

(none)

Notes

/ɜː/ is rare in initial position, and the dozen or so words in which it does occur are split between <ear> (earl, early, earn, earnest, earth), <er> (ermine, ersatz, erstwhile), <err> (only in err), <ir> (only in irk) and <ur> (urban, urbane, urchin, urge, urgent, urn).

<er> is the default spelling in medial and final positions. It is regular in hyperbole, interpret, superfluous, superlative and other words with initial /haɪˈpɜː, ɪntˈɜː, suːpˈɜː/ stressed on the second syllable and spelt <hyper-, inter-, super->; concern, discern, convert, revert and other words with the (Latin) elements <cern, vert>; confer, defer, prefer, refer with the (Latin) element <-fer>. Other examples: adverse, alert, averse, assert, berth, certain, commercial, conserve and derivatives, deserve, desert (/dɪˈzɜːt/, ‘abandon’, as opposed to /ˈdezət/, ‘arid area’), dessert, determine, disperse, epergne, eternal, exertion, exterminate, ferment, germ, gherkin, herb, herd, hermit, immerse, inert(ia), jersey, kerchief, kernel, merge, mercenary, nerd, observe, perfect, perk, permanent, person, quern, reserve, reverse, serf, serpent, serve, submerge, swerve, tern, terse, thermal, thermos, twerp, universe, verger, verse, vertigo.

<or> is regular after initial /w/ whether spelt <w> or <wh>: whortle(berry), word, work, world, worm, worse(n), worship, worst, wort, worth(y); otherwise only in attorney. Exceptions: were, whirl, whir(r).

<ir> is regular in the prefix /ˈsɜːkəm/ spelt <circum->, e.g. circumflex, circumstance, circumvent; also after /g, kw, θ/, as in gird, gird(le), girder, girl, girn, girt, girth (exceptions: gherkin, gurgle, regurgitate); quirk, quirt, squirm, squirt; third, thirst, thirteen, thirty (exceptions: thermal, thermos, Thursday). Otherwise <ir> occurs in an unpredictable set of words, e.g. besmirch, birch, bird, birth, chirp, circle, circus, cirque, dirk, dirt, fir, firm, firmament, first, firth, flirt, hirsute, irk, kirtle, mirth, shirk, shirt, sir, skirl, skirmish, skirt, smirk, stir, swirl, twirl, Virgo, virtual, virtue, virtuoso, virtuous, whir, whirl, zircon.

<ur> is regular:

1) in the (Latin) verb element <-cur> (‘run’) as in concur, (dis)cursive, cursor, excursus, incur(sion), occur, recur and more generally after /k/: cur, curb, curd, curfew, curl, curlew, curse, curt, curtail, curtain, curtsey, curve, scurf, scurvy (exceptions: colonel, courteous, courtesy, kerchief, kernel, kersey, kirtle, skirl, skirmish, skirt);

2) after /b, ʧ/ and after /s/ in initial syllables of polysyllables: auburn, burble, burden, burdock, burgess, burgher, burglar, burgeon, burgoo, burgundy, burlap, burlesque, burly, burn, burnet, burnish, burp, bursar, burst, disburse, hamburger, laburnum, suburb; church, churl(ish), churn; surface, surfeit, surgeon, surly, surmise, surmount, surpass, surplice, surplus, surveillance, survey, survive (exceptions: berg, berth, birch, bird, birth; chirp, chirr, concerto if pronounced with /ɜː/ rather than /eə/, serpent).

Otherwise <ur> occurs in an unpredictable set of words, e.g. absurd, appurtenance, blur, blurt, demur, disturb, diurnal, expurgate, frankfurter, fur, (re)furbish, furl, furlong, furlough, furnace, furnish, furniture, further, furtive, furze, gurgle, hurdle, hurl, hurt, hurtle, insurgent, jodhpurs, liturgy-y/ical, lurch(er), lurk, metallurg-y/ical, murder, murky, murmur (first syllable), nasturtium, nocturnal, nurse, nurture, purblind, purchase, purgation, purgatory, purge, purl, purlieu, purloin, purport, purse, pursu-e/it,
purvey
, regurgitate, return, Saturn, saturnine, slur, splurge, spur, spurn, spurt, surd, surf, taciturn, Thursday, turban, turbid, turbine, turbot, turbulent, turd, turf/ves, turgid, turkey, turmoil, turn, turnip, turquoise, turtle, urban, urbane, urchin, urge, urgent, urn. In some words where <ur> is not stressed it may be reuced to /ə/, e.g. purport, pursu-e/it, surpass.

Words in which final /ɜː/ is spelt with a grapheme which includes final <-r> allow /r/-linking (see section 3.6), e.g. murmuring, whirring, purring, sometimes with <r>-doubling (see section 4.2), e.g. conferring, occurring, demurral.

5.5.3 /ɔː/ as in awe

The main system

For all these categories see Notes.

Basic grapheme

<or>

25% (with <ore, ar>)

e.g. order, afford, for. See also Table 5.4

Other frequent graphemes

<ore>

only word-final, e.g. before, except in compounds of fore-. See also Table 5.4

<ar>

regular medially after /w/, e.g. ward. See also Table 5.3

<a>

29%

regular before /l/; otherwise only in water, waft pronounced /wɔːft/, wrath pronounced /rɔːθ/

<au>

9%

e.g. autumn, cause; word-final only in landau, Nassau. See also Table 5.4

<aw>

9%

never before /r/; e.g. awful, crawl, paw. See also Table 5.4

The rest

Oddities

28% in total

<al>

5% only in balk, calk, chalk, falconer, stalk, talk, walk

<augh>

2% only in aught, caught, daughter, distraught, fraught, haughty, (Mc)Naught(on), naught, naughty, onslaught, slaughter, taught

<aul>

only in baulk, caulk, haulm

<aur>

only in bucentaur, centaur, dinosaur (and the names of various dinosaur species, e.g. pterosaur), minotaur

<awe>

only in awe and derivatives other than awful

<oa>

only in abroad, broad, broaden

<oar>

2% only in boar, board, coarse, hoar, hoard, hoarse, oar, roar, soar

<oer>

only in Boer pronounced /bɔː/ (also pronounced /bʊə/)

<oor>

3% only in door, floor; also boor, moor, poor, spoor if pronounced to rhyme with door, floor

<orp>

only in corps (plural), pronounced /kɔːz/

<orps>

only in corps (singular), pronounced /kɔː/

<orr>

only in abhorred

<ort>

only in mortgage, rapport. /t/ surfaces in rapporteur – see section 7.2

<ough>

6% only in bought, brought, fought, nought, ought, (be-)sought, thought, wrought

<our>

8% only in bourne, court(esan), course, four, mourn, pour, source, your(s)

<ou’re>

only in you’re. See section A.9 in Appendix A

2-phoneme graphemes

(none)

Notes

Generalisations for /ɔː/ are weak because it has so many spellings. However, some are possible for instances of /ɔː/ after /w/ and before /l/. <ar> is regular in medial position after /w/, however spelt – see Table 5.3 and cf. /ɒ/, section 5.4.4. There are no words in which /ɔː/ is spelt <ar> without a preceding /w/.

Table 5.3: Spellings of /ɔː/ as <ar> after /w/.

after /w/ spelt <u> – always after /k/ spelt <q>

after /w/ spelt <w>

after /w/ spelt <wh>

quart(an/er/et/ic/ile/z); also, according to particle physicists, quark /kwɔːk/ (pronounced /kwɑːk/ by the rest of us)

award, dwar-f/ves, reward, sward, swarf, swarm, swart(hy), (a)thwart, towards, untoward, warble, ward, warden, warfarin, warlock, warm, warn, warp, wart; also war (only example in final position and therefore only one with potential /r/-linking (and concomitant <r>-doubling – see section 4.2), e.g. warring – see section 3.6)

whar-f/ves

Exceptions (words in which /ɔː/ is not spelt <ar> after /w/)

quorn, quorum, squaw, squawk

caterwaul, sworn, walk, wall, walnut, waltz, water, whorl, worn; also waft, walrus if pronounced with /ɔː/ rather than /ɒ/.

wall, walnut, walrus, waltz instead follow the generalisation about /ɔː/ before /l/, next

<a> is regular in all positions before /l/:

  • initial: albeit, alder, alderman, all, almanac (usually pronounced with /æ/), almighty, almost, already, altar, alter, alternate (with both stresses and meanings), although, altogether, always (only exception: awl, which is also pre-final – see below)
  • medial (except before final /l/): bald, balderdash, baldric, balsam(ic), balti, enthralment, falcon, false, falsetto, falter, halt, halter, instalment, malt, palfrey, palsy, paltry, psalter, salt, scald, thraldom (also spelt thralldom), walnut, walrus (also pronounced with /ɒ/), waltz (exceptions: assault, cauldron, fault, vault)
  • pre-final (= medial before final /l/; N.B. This is the only place in my entire analysis where I have found it useful to use the term ‘pre-final’): appal, ball, call, enthral, fall, gall, hall, pall, small, squall, (fore/in)stall, tall, thrall, wall (exceptions: caterwaul, haul, maul; awl, bawl, brawl, crawl, drawl, scrawl, shawl, sprawl, trawl, yawl; whorl). As this list shows, here /l/ after <a> is mostly spelt <ll>, the only exceptions being appal, enthral. On variation between <l> and <ll> see also section 4.4.7.

The only words in which /ɔː/ is spelt <a> other than before /l/ are waft if pronounced /wɔːft/, water, and wrath if pronounced /rɔːθ/.

Beyond this it is simplest to list spellings of /ɔː/ in <au, aw, or, ore> – see Table 5.4.

Many other examples of medial /ɔː/ spelt <or> before /r/ arise from suffixation of words ending in <-ore> (e.g. boring), when /r/-linking occurs – see section 3.6. In all these suffixed cases and in all the cases where medial /ɔː/ spelt <or> occurs before a vowel, the <r> is both part of grapheme <or> spelling /ɔː/ and a grapheme in its own right spelling /r/ (for dual-functioning see section 7.1). However, dual-functioning does not apply to /ɔː/ spelt <au> before /r/ since <au> already spells /ɔː/ without the following <r>.

The only example of <r>-doubling in final /ɔː/ spelt <or> appears to be abhorred, where the stress in the stem word is on the last syllable and <rr> arises from the main consonant-doubling rule – see section 4.2. In abhorrent there is both <r>-doubling and /r/-linking (see section 3.6) but the preceding vowel changes to /ɒ/ and <rr> spells only /r/.

Although I have said above that /ɔː/ before /r/ is never spelt <aw>, we should remember the childish pronunciation of drawing as /ˈdrɔːrɪŋ/.

Words in which final /ɔː/ is spelt with other graphemes which include final <r> also allow /r/-linking, e.g. hoary, flooring, pouring – see again section 3.6.

Table 5.4: <au, aw, or, ore> as spellings of /ɔː/.

For other spellings of /ɔː/ see above.

Initial

Medial

Final

<au>

(before /r/)

aura, aural (also pronounced with /aʊ/), aureole, aureomycin, auricle, auriferous, aurochs, aurora.

See notes above Table

(not before /r/)

aubretia, auburn, auction (also pronounced with /ɒ/), audacious, audible, audience, audio, audit, auger, augment, augur, August, august, auk, aumbry, auspic-e/ious, austere,authentic,

(before /r/)

apatosaurus and many other dinosaur names, saurian, sauropod, taurine, Taurus, thesaurus.

See notes above Table

(not before /r/)

applaud, assault, astronaut, bauble, bauxite, caterwaul, caucus, caudal, cauldron, cause, caustic, cauterise, caution, clause, daub, daunt, debauch, exhaust, faucet, fault, faun, fauna, flaunt, flautist, fraud, gaudy, gaunt,

only in landau

author(ity), autis-m/tic, autograph, automatic, automobile, autonomy and many other compounds of <aut(o)->, autumn, auxiliary

gauntlet, gauze, glaucoma (also pronounced with /aʊ/), glaucous, haul, haunch, haunt, holocaust, hydraulic (also pronounced with /ɒ/), inaugurate, jaundice, jaunt, juggernaut, laud, launch, launder, laundry, marauder, maudlin, maul, mausoleum, nausea-a/ous, nautical, paucity, paunch, pauper, pause, plaudit, plausible, raucous, sauce, saucer, sauna, saunter, staunch, taunt, taut, vault, vaunt

<aw> (never before /r/)

awful, awkward, awl, awning

bawd, bawl, brawl, brawn(y), crawl, dawdle, dawn, drawl, drawn, fawn, gawp, hawk, hawser, lawn, mawkish, pawn, prawn, scrawl, scrawny, shawl, shawm, spawn, sprawl, squawk, tawdry, tawny, tomahawk, trawl, trawler, yawl, yawn

caw, claw, draw, flaw, gnaw, guffaw, haw, jackdaw, jaw, law, lockjaw, macaw, maw, paw, raw, rickshaw, saw, seesaw, slaw, squaw, straw, thaw, yaw

<or>

(before /r/)

oracy, oral, oration, orient (noun). See notes above Table

(not before /r/)

or, orb, orbit, orc, orchard, orchestra, orchid, ordain, ordeal, order, ordinary, ordnance, ordure, organ, organdie, organise, orgasm, orgy, ormolu, ornament, ornate, ornery, ornithology, orphan, orthodontist, orthodox and many other compounds of <ortho->, ortolan, orts

(before /r/)

aurora, authorial, borax, chlorine, choral, chorus, corporeal (second syllable), decorum, dictatorial, editorial, euphoria, flora, floral (also pronounced with /ɒ/), forum, glory, memorial, oratorio (third syllable), quorum, variorum. See notes above Table

(not before /r/)

abort(ion), absorb, absorption, adorn, afford, border, born(e), cavort, chord, chortle, cohort, consort, cord, cork, corm, corn, corner, cornice, corporal, corporeal (first syllable), corporation (first syllable), corpse, corset, corvette, disgorge, divorce, dork, dormitory, endorse, enormous, exorcise, extortion,

abhor, cantor, condor, corridor, cuspidor, décor, for, grantor, humidor, ichor, lessor, matador, mentor, mortgagor, nor, or, praetor, quaestor, realtor, tor, toreador, vendor

Table 5.4: <au, aw, or, ore> as spellings of /ɔː/. cont.

Initial

Medial

Final

force, forfeit, forge, fork, forlorn, form, forsythia, fort, forth, fortune, gorge, gormandise, gormless, gorse, horde, hormone, horn, hornet, horse, horticulture, important, inform, lord, lorgnette, morbid, mordant, morganatic, morgue, morning, morphine, morse, morsel, mortal, mortar, nork, normal, north(-ern/ly), perform, platform, porcelain, porch, porcupine, pork, porpoise, porphyry, portion, portico, portrait, record, remorse, report, resort, scorch, scorn, Scorpio, scorpion, shorn, short, snorkel, sorcerer, sordid, sorghum, sort, sport, stork, storm, suborn, support, sword, sworn, thorn, torc, torch, torment, torn, tornado, torpedo, torpid, torque, torsion, torso, tort, tortoise, torture, uniform, vortex, worn

<ore>

(occurs only in ore, which I classify as final)

only in compounds of fore-, of which there are 60+

adore, albacore, before, bore, carnivore, chore, claymore, commodore, core, deplore, encore, explore, fore, furore, galore, gore, herbivore, ignore, implore, lore, more, omnivore, ore, pinafore, pore, score, semaphore, shore, snore, sophomore, sore, spore, stevedore, store, swore, sycamore, therefore, tore, whore, wore, yore

5.6 Diphthongs (other than /eɪ, aɪ, əʊ/): /ɔɪ aʊ eə ɪə ʊə/

5.6.1 /ɔɪ/ as in oyster

The main system

Basic grapheme

<oi>

61%

e.g. boil. Regular in initial and medial positions. Never word-final (except in the Greek phrase hoi polloi)

Other frequent grapheme

<oy>

39%

e.g. boy. Regular in word-final position; rare elsewhere, but see Notes

The rest

Oddity

<aw>

only in lawyer, sawyer

2-phoneme grapheme
(counting the automatic /j/-glide as part of the first phoneme)

/ɔɪjə/ spelt <oir>

only in coir /ˈkɔɪjə/

Notes

<oy> is regular in word-final position, <oi> elsewhere. Exceptions:

  • <oi> word-finally: only in hoi polloi
  • <oy> non-finally: only in arroyo, boycott, coypu, foyer pronounced /ˈfɔɪjə/ (also pronounced /ˈfwaɪjeɪ, ˈfɔɪjeɪ/), gargoyle, groyne, hoyden, loyal, oyster (only occurrence in initial position), royal, soya, voyage. In arroyo, foyer pronounced /ˈfɔɪjə/, loyal, royal, soya, voyage the <y> is both part of <oy> spelling /ɔɪ/ and also a grapheme in its own right spelling /j/. For dual-functioning see section 7.1.

coir is the only word in the language with /ɔɪjə/ spelt with the single grapheme <oir>. Word-final /ɔɪjə/ also has the 2-grapheme spellings <oya, oyer> only in soya, foyer pronounced /ˈfɔɪjə/, the first word of the name of the ancient court known as Oyer and Terminer, and coyer, comparative of coy (‘Had we but world enough and time…’), and the 3-grapheme spelling <-awyer> only in lawyer, sawyer /ˈlɔɪjə, ˈsɔɪjə/. Effectively, therefore, all the example words mentioned so far in this paragraph rhyme. Medially, /ɔɪjə/ occurs in loyal, royal and their derivatives, and possibly nowhere else. It is also noticeable that within this (tiny) set of words, only coir itself does not contain <y>.

5.6.2 /aʊ/ as in ouch

The main system

Basic graphemes

<ou>

93% (with word-final <ow>)

e.g. about

word-final <ow>

e.g. allow

Other frequent graphemes

(none)

The rest

Oddities

7% in total

<aow>

only in miaow

<au>

only in ablaut, faustian, gaucho, gauleiter, glaucoma pronounced /glaʊˈkəʊmə/ (also pronounced /glɔːˈkəʊmə/), sauerkraut (twice), umlaut and the Greek letter name tau; also in aural if pronounced /ˈaʊrəl/ to distinguish it from oral /ˈɔːrəl/

<ough>

only in bough, doughty, drought, plough, slough (‘muddy place’)

pre-consonantal <ow>

6% e.g. brown

2-phoneme grapheme (counting the automatic /w/-glide as part
of the first phoneme
)

/aʊwə/

(1) spelt <hour>

only in hour

(2) spelt <our>

only in devour, flour, lour, our, ours, scour, sour and dour pronounced /ˈdaʊwə/ (which makes it a homophone of dower; dour is also pronounced /dʊə/). These words allow /r/-linking, e.g. floury, scouring – see section 3.6. Also see Notes

Notes

<ow> is regular:

  • in word-final position, e.g. cow (only exceptions: thou /ðaʊ/, archaic second person singular subject pronoun, thou /θaʊ/, ‘one thousandth of an inch/a thousand pounds/dollars’)
  • before a schwa vowel spelt with a vowel letter or digraph, i.e. only in bowel, dowel, rowel, towel, trowel, vowel; bower, cower, dower, flower, glower, power, shower, tower; coward, dowager (no exceptions), plus howitzer with /ɪ/ and prowess with /e/; this would also cover rowan in its Scottish pronunciation /ˈraʊwən/ (/ˈrəʊwən/ in England)
  • in most words ending in /aʊl/, namely cowl, fowl, growl, howl, jowl, owl, prowl, scowl, yowl (only exception to this sub-pattern: foul)
  • in most words ending in /aʊn/, namely brown, crown, down, drown, frown, gown, renown, town (only exceptions to this sub-pattern: (pro)noun).

<ou> is regular everywhere else. Exceptions (in addition to the Oddities above and <ow> subpatterns just listed): chowder, crowd, dowdy, powder, rowdy; cowrie, dowry; frowsty; blowsy (contrast blouse), bowser, browse, dowse, drowse, drowsy, frowsy.

/aʊwə/ also has the 2-grapheme spellings <-ower> in bower, cower, dower, flower, glower, power, shower, tower, <owar> in coward and <owa> in dowager. All of these words except coward allow /r/-linking, e.g. cowering, flowery – see section 3.6.

In the words with medial <ow> followed by a vowel letter listed above, the <w> also represents a /w/-glide between /aʊ/ and the following schwa (or /ɪ/ or /e/). In these words, therefore, the <w> is both part of the digraph <ow> spelling /aʊ/ and a grapheme in its own right spelling /w/ (for dual-functioning see section 7.1). The words within this set ending in <-ower> form perfect rhymes with the words ending in <-our> listed above, and these too seem to me to have an automatic /w/-glide – but the /w/ is not represented in the spelling. So for the /w/-glide the alternative spellings mean ‘Now you see it, now you don’t ‘. For more on that, see sections 3.8.7 and 9.0.

5.6.3 /eə/ as in air

For the two sets of percentages see Notes.

The main system

Basic grapheme

<are>

59%
(24%) (with <ar>)

only word-final, e.g. bare, care, fare, flare, hare, pare, stare, tare, ware. See Notes

Other frequent graphemes

<ar>

initially, only in area, Aries; never word-final. Regular medially, especially where word-final <e> is deleted before a suffix beginning with a vowel letter, e.g. caring, but there are also independent examples, e.g. adversarial, Aquarius, barium, commissariat, garish, gregarious, hilarious, malaria(l), multifarious, nefarious, parent, precarious, proletariat, Sagittarius, variegated, various, vary, and a fairly large set of nouns/ adjectives in <-arian>, e.g. agrarian, barbarian (2nd <ar>), centenarian and other age terms, egalitarian, grammarian, librarian, proletarian, utilitarian, vegetarian; in all these cases the <r> also spells /r/ (for dual-functioning see section 7.1). Medially before a consonant, only in scarce, scarcity. See Notes

<air>

28% (12%)

regular initially because of air and its compounds (see under <aer> below); medially, only in fairy, prairie (with dual-functioning <r> - see section 7.1) and (Scots) bairn, cairn, laird; otherwise only word-final and only in affair, air (again), chair, corsair, debonair, despair, eclair, fair, flair, hair, impair, lair, mohair, pair, repair, stair

<ear>

10% (4%)

only word-final and only in (for(e)-)bear, pear, swear, tear (‘rip’), wear

The rest

Oddities

3% (60% !!) in total

- non-final

<aer>

except in anaerobic, faerie, only initial and only in words where the morpheme air is followed by a vowel phoneme, namely several compounds of aero-, e.g. aerobic, aerodrome, aeroplane, aerosol, etc., plus aerate, aerial. In all these cases the <r> is both part of <aer> spelling /eə/ and a grapheme in its own right spelling /r/. For dual-functioning see section 7.1. Compounds with the spelling <air>, e.g. aircraft, airmail, are more numerous, and therefore (because there are so few other words beginning /eə/, namely area, e’er, and heir and its derivatives) <air> is the main word-initial spelling

<ao>

only in aorist

<eir>

only in theirs

<er>

only in bolero (/bəʽleərəʊ/, ‘dance’), concerto pronounced /kənˈʧeətəʊ/ (also pronounced /kənˈʧɜːtəʊ/), concierge, recherche, scherzo, sombrero. In bolero, sombrero the <r> is both part of <er> spelling /eə/ and a grapheme in its own right spelling /r/. For dual-functioning see section 7.1

- final

<aire>

only in a few polysyllabic recent loanwords of mainly French origin, namely affaire, commissionaire, concessionaire, doctrinaire, laissez-faire, legionnaire, millionaire, questionnaire, secretaire, solitaire. /r/-linking occurs in millionairess – see section 3.6

<ayer>

only in prayer pronounced /preə/ (‘religious formula’; contrast prayer pronounced /ˈpreɪjə/, ‘one who prays’)

<ayor>

only in mayor and derivatives. /r/-linking occurs in mayoral, mayoress – see section 3.6

<eah>

only in yeah

<e’er>

only in e’er, ne’er, where’er and a few other archaic contracted forms. See Section A.9 in Appendix A

<eir>

not counted (22% !!) only in their

<ere>

not counted (37% !!) (with <er>) only in ere, there,
(no)where and a few polysyllabic recent loanwords of French origin, namely ampere, brassiere, cafetiere, commere, compere, confrere, misere, premiere. /r/-linking occurs in thereupon, wherever, compering, etc. – see section 3.6

<erre>

only in parterre

<ey’re>

only in they’re. See Section A.9 in Appendix A

<heir>

only in heir. There is /r/-linking in heiress, inherit – see section 3.6, and in inherit /h/ also surfaces – see section 7.2

2-phoneme graphemes

(none)

Notes

If we follow Crystal (2012: 131-2) and Upward and Davidson (2011: 176-9), ‘more recent’ in terms of loanwords from French means after the Great Vowel Shift, which began about AD1400 and was complete by about AD1600.

<are> is regular word-finally (and would be more so if there, where were spelt *thare, *whare – but that would destroy the parallelism with here), <air> initially, <ar> medially.

Scarce, scarcity are the only words in which /eə/ spelt <ar> occurs before a consonant and the <r> is only part of the grapheme <ar> (hence more logical spellings for them would be *scairce, *scaircity, on the model of bairn, cairn, laird); in all other cases <ar> occurs before a vowel and the <r> is both part of <ar> spelling /eə/ and a grapheme in its own right spelling /r/ - for dual-functioning see section 7.1.

Similarly, in all the patterns with word-final <r(e)> (which is every word-final pattern listed above except <eah>), there is potential /r/-linking (and dual-functioning) before a suffix beginning with a vowel phoneme (e.g. staring, repairing, wearing) or a following word beginning with a vowel phoneme – see section 3.6. Examples before a following word: prayer of intercession, mayor of Sheffield, ne’er a hope in h*ll, misere ouverte, they’re arriving, heir apparent.

Carney (1994: 110-1) points out that text frequencies for /eə/ differ vastly according to whether function words are included or not: ‘The three words where, there and their account for more than half the raw text frequency of /eə/. In this book I have almost exclusively used his function-words-excluded frequencies, but at the head of this entry and against a couple of the Oddities I have shown, for interest, both those and (in brackets) the very different frequencies when function words are included.

5.6.4 /ɪə/ as in ear

For why the percentages are double Carney’s see Notes.

For all medial and final occurrences see also Table 5.5.

The main system

Basic grapheme

<ear>

56%

(except in afeard, arrears, beard, bleary, weary and the half-exception ear) only word-final, e.g. appear, hear

Other frequent graphemes

<ere>

24% (with <er>)

only word- final, e.g. interfere, mere, sincere. Carney’s (original) percentage excludes here, which would skew the figures (cf. /eə/, just above)

<er>

except in era, only medial, e.g. hero, series. In all cases the <r> functions also as the spelling of /r/ - see section 7.1

<eer>

8%

except in eerie (where <r> functions also as the spelling of /r/ - see section 7.1), only word-final, e.g. beer

The rest

Oddities

12% in total

<eir>

only in weir, weird

<eyr>

only in eyrie (where <r> functions also as the spelling of /r/ - see section 7.1)

<e’re>

only in we’re. See Section A.9 in Appendix A

<ier>

never initial; medially, only in fierce, pierce, tierce; otherwise only final

<ir>

only in emir, fakir (can be stressed on either syllable, and is also pronounced with /ə/), kir, kirsch, nadir pronounced /ˈneɪdɪə, næˈdɪə/ (also pronounced /ˈneɪdə/), souvenir, tapir

2-phoneme graphemes

(none)

Notes

Carney (1994: 190) posits two sources of this phoneme in RP:

1) cases where there used to be a /r/ consonant following an /iː/ vowel. A letter <r> remains in the spelling (usefully for speakers with rhotic accents), but in RP the /r/ has disappeared except when /r/-linking occurs (see section 3.6);

2) cases where there never was a /r/ phoneme but an /iː/ has combined with a following /ə/.

I accept the first category but not the second. Carney does say (ibid.) that the second category ‘for some speakers may still represent disyllabic /iː/ plus /ə/’, and I think this is the case in my accent and that of many other RP speakers. For example, on Carney’s analysis the expression Stay, dear and the word stadia would both be analysed as pronounced /ˈsteɪdɪə/, with two syllables, but I think only the former is so pronounced and that stadia is pronounced /ˈsteɪdiːjə/, with three syllables (and an automatic /j/-glide before the final schwa – see section 3.8.8). I have therefore assigned almost all occurrences with <r> to /ɪə/ (for the exceptions see below), and all occurrences without <r> instead to /iːjə/.

Fortunately, unlike the situation with /ɪ, iː/, Carney provides just enough information to re-calculate the percentages for /ɪə/ without the second category, and the results (which are double the percentages given by Carney) are shown above.

<eer> may seem a more ‘basic’ spelling of /ɪə/ than <ear> but accounts for a much smaller percentage of its occurrences.

Curiously, Carney does not list <ier> or any of the words containing it in his treatment of /ɪə/, presumably because none occurred in his corpus.

The only words in which /ɪə/ occurs initially seem to be ear (where it is the whole word and therefore also final), eerie, era and eyrie.

Spellings with <r> which I believe belong to /iːjə/ rather than to /ɪə/ are few in number, and restricted to:

1) adjectives in <-ear, -iar>: cochlear, linear, nuclear, familiar, peculiar;

2) comparative adjectives in <-ier>, e.g. easier, happier.

In all these cases I believe the ending has two syllables.

A serendipitous outcome of my analysis of /ɪə/ is that all its occurrences in polysyllables are stressed, wherever in the word they occur, except that fakir, frontier, nadir can be stressed on either syllable and belvedere is often stressed on the first syllable.

Medially, the predominant spelling is <er>, and in polysyllables there are no exceptions; finally, the predominant spelling is probably <eer> in lexical frequency but definitely <ear> in text frequency because most of the words in which it occurs have high frequency. However, because no clear guidance can be given on which spelling of /ɪə/ occurs in which words in final position, lists for both medial and final positions are given in Table 5.5.

In all the patterns listed above which occur word-finally (which is all of them except <er, eyr>), there is potential /r/-linking before a suffix beginning with a vowel phoneme (e.g. hearing, sincerity, beery) or a following word beginning with a vowel phoneme – see section 3.6. Examples before a following word: hear and obey, beer and skittles, we’re off!

Table 5.5: Spellings of /ɪə/ in medial and final positions.

medial

final

<ear> afeared, beard

<ear> blear, clear, dear, drear, ear, fear, gear, hear, near, rear, sear, shear, smear, spear, tear (‘moisture from eye’), year; appear, arrear

<eer> beer, cheer, deer, jeer, leer, peer, queer, seer, sheer, sneer, steer, veer; auctioneer, Brexiteer, career, charioteer, commandeer, domineer, electioneer, engineer, gazetteer, mountaineer, muleteer, musketeer, mutineer, pamphleteer, pioneer, privateer, profiteer, scrutineer, veneer, volunteer

<eir> weird

<eir> weir

<er> adherent, cereal, coherence, coherent, ethereal, funereal, hero, inherent, managerial, material, perseverance, serial, series, serious, serum, sidereal, venereal, zero; also frequent when words in <-ere> are suffixed, e.g, interfering

<ere> mere, sere, sphere; ad/co/in-here, austere, belvedere, cashmere, interfere, revere, severe, sincere

<ier> fierce, pierce, tierce

<ier> bier, pier, tier; bandolier, bombardier, brigadier, cashier, cavalier, chandelier, chevalier, clavier, corsetier, frontier, fusilier, gondolier, grenadier, halberdier, vizier

<ir> kirsch

<ir> fakir, kir, nadir, souvenir

In all the words with <er> in this column the <r> is both a grapheme in its own right spelling /r/ and part of the grapheme <er> spelling /ɪə/. For dual-functioning see section 7.1.

All the words in this column have the potential for /r/-linking – see section 3.6, some with change of vowel, e.g. sincerity.

5.6.5 /ʊə/ as in rural

This phoneme is so rare in RP that it would be futile to identify a basic grapheme, so I have just listed 1- and 2-phoneme graphemes. In all cases see Notes.

1-phoneme graphemes

<eur>

only in pleurisy, where the <r> also spells /r/. For dual-functioning see section 7.1

<oor>

only word-final and only in boor, moor, poor, spoor pronounced /bʊə, mʊə, pʊə, spʊə/ (also pronounced /bɔː, mɔː, pɔː, spɔː/). There is /r/-linking in, e.g., boorish – see section 3.6

<our>

only in amour, bourbon (‘biscuit’), bourgeois(ie), bourse, contour, detour, dour pronounced /dʊə/ (also pronounced /ˈdaʊwə/), entourage, gourd, gourmand, gourmet, houri, mourn (e.g. in mourning pronounced /ˈmʊənɪŋ/ to distinguish it carefully from morning /ˈmɔːnɪŋ/), potpourri if we take the second <r> as spelling /r/, tour, tourney pronounced /ˈtʊəniː/ (also pronounced /ˈtɜːniː/), tournament, tourniquet, troubadour, velour. There is /r/-linking in, e.g., touring – see section 3.6 - and in houri the <r> is both part of grapheme <our> and a grapheme in its own right spelling /r/. For dual-functioning see section 7.1

<ur>

never word-final; initially, only in urtext; otherwise only medial, e.g. injurious, insurance, juror, jury, luxuriance, luxuriant, luxuriate, luxurious (pronounced /lʌgˈʒʊəriːj-əns/ənt/eɪt/əs/), prurien-t/ce, rural, usurious; also centurion, durable, (en)during, duress, maturity pronounced /senˈʧʊəriːjən, ˈʤʊərəbəl, (ɪn)ˈʤʊərɪŋ, ʤʊəˈres, məˈʧʊərɪtiː/, i.e. with /tj, dj/ affricated to /ʧ, ʤ/. In all cases except urtext the <r> is both part of <ur> spelling /ʊə/ and a grapheme in its own right spelling /r/ - for dual-functioning see section 7.1

<ure>

only word-final, e.g. abjure, adjure, assure, brochure, caricature (also pronounced with final /ə/), conjure /kənˈʤʊə/ (‘summon with an oath’), cynosure, embouchure, endure pronounced /ɪnˈʤʊə/, ensure, insure, mature pronounced /məˈʧʊə/, overture (if final syllable is pronounced /ʧʊə/ rather than /ʧə/), sure

2-phoneme graphemes

/jʊə/

(1)

spelt <eur>

only in Europe (where the <r> is also a grapheme in its own right spelling /r/ - for dual-functioning see section 7.1) and liqueur pronounced /lɪˈkjʊə/

(2)

spelt <ur>

never word-final; initially, only in urea and various words derived from or cognate with it, e.g. Uranus pronounced /ˈjʊərənəs/ ‘urine us’ (also pronounced /jəˈreɪnəs/ ‘your anus’), urethra, uric, urine, urology; medial examples are bravura, curate (both the noun ‘junior cleric’ pronounced /ˈkjʊərət/ and the verb ‘mount an exhibition’ pronounced /kjʊəˈreɪt/), curie, curious,

furious, fury, mural, purify, purity, security, spurious; also centurion, durable, (en)during, duress, maturity pronounced /senˈtjʊəriːjən, ˈdjʊərəbəl, (ɪn)ˈdjʊərɪŋ, djʊəˈres, məˈtjʊərɪtiː/, i.e. with /tj, dj/ NOT affricated to /ʧ, ʤ/. In all cases the <r> is both part of <ur> spelling /jʊə/ and a grapheme in its own right spelling /r/ - for dual-functioning see section 7.1

(3)

spelt <ure>

only word-final, e.g. allure, coiffure, cure, demure, endure pronounced /ɪnˈdjʊə/, immure, inure, lure, manure, mature, ordure pronounced /məˈtjʊə, ɔːˈdjʊə/, overture (if final syllable is pronounced /tjʊə/ rather than /tjə/), photogravure, pure, secure, sinecure, Ure; also in azure pronounced /ˈæzjʊə, ˈeɪzjʊə/ (also pronounced /ˈæzjə, ˈeɪzjə, ˈæʒə, ˈeɪʒə/)

Oddities

All the correspondences for this phoneme are Oddities

Notes

This phoneme is rare and getting rarer in RP, and may eventually disappear. Its rarity means percentages for graphemes would be misleading, as would treating /jʊə/ as a separate phoneme from /ʊə/ in parallel fashion to separating /juː/ from /uː/. Many words in which /ʊə/ used to occur now have /ɔː/ instead. For instance, the word your used to be pronounced /jʊə/ (and still is, in some accents), but in RP is now /jɔː/ – and cure, liqueur, mature and pure are now often heard as /kjɔː, lɪˈkjɔː, məˈʧɔː, pjɔː/ in up-market accents. But words like curious, fury, injurious, juror, jury, prurient, rural, spurious seem to be resisting the change to /ɔː/. Check your own pronunciation of the words listed in this section.

Carney (1994: 194-5) also classifies as examples of /(j)ʊə/ many words in which letter <u> is followed by a spelling of /ə/ (e.g. cruel, jewel, usual). I analyse these instead as being pronounced with /(j)uː/ and /ə/ constituting a separate syllable (and an automatic intervening /w/-glide). It is noticeable that all these words end in a consonant phoneme.

In phonologically similar words which end in a vowel phoneme (which here is always /ə/) it seems to be agreed that the ending is /ˈ(j)uːwə/. Words in these (very small) groups are:

  • (with /ˈuːwə/) brewer, sewer (/ˈsuːwə/, ‘foul drain’, as opposed to sewer /ˈsəʊwə/, ‘one who sews’), interviewer, viewer (in the last two the /j/ glide is spelt <i> – contrast the next group);
  • (with /ˈjuːwə/) ewer, fewer, hewer, newer, renewer, skewer (cf. the homophone skua). A few derived forms also have /ˈuːwə/, e.g. doer (‘one who does’), two-er (‘conker which has broken two others’).

In my analysis a few /(j)ʊə/ v. /ˈ(j)uːwə/ minimal pairs seem possible, e.g. Ure/ewer, dour/doer, tour/two-er – but the phonological difference is minute (and some phoneticians would say non-existent).

5.7 Letter-name vowels: /eɪ iː aɪ əʊ juː/, plus /uː/

5.7.1 /eɪ/ as in aim

The main system

For all these categories see Notes.

Basic grapheme

<a.e>

38% (76%
in monosyllables)

regular in closed final syllables, e.g. dilate, make, take, ache, champagne

Other frequent graphemes

<a>

27%

regular in non-final syllables of stem words, e.g. agent, bacon, labour

<ay>

18%

regular in open final syllables (= in word-final position), e.g. chardonnay, day, display, way; never initial; rare medially but cf. always, claymore, mayhem, nowadays

<ai>

12%

regular before /nt/, e.g. paint (only exceptions: ain’t (sort of), feint); never word-final

The rest

Oddities

5% in total

<ae>

only in brae, Gaelic pronounced /ˈgeɪlɪk/, maelstrom, reggae, sundae

<ah>

only in dahlia

<aigh>

only in straight

<ais>

only in palais

<ait>

only in distrait, parfait, and trait pronounced /treɪ/ (also pronounced /treɪt/)

<alf>

only in halfpence, halfpenny

<ao>

only in gaol

<au>

only in gauge

<aye>

only in aye (‘ever’)

<e>

only in 60+ more recent loanwords mainly from French where French spelling has <é>, namely (in non-final position) debris, debut, decor, eclair, ecru, elan, ingenu, precis; first <e> in debacle, debutante, decalage, decolletage, denouement, detente, elite, ingenue, menage, regime, seance, (Greek) heter/hom-ogeneity pronounced /hetər/hɒm-əʊʤɪˈneɪjɪtiː/ (usually pronounced /hetər/hɒm-əʊʤɪˈniːjɪtiː/), (Old English) thegn and (Hawaian) ukulele; (word-finally) abbe, attache, blase, cafe (also pronounced with /iː/), canape (also pronounced with /iː/, hence the invitation I once received to a party with ‘wine and canopies’), cliche, communique, conge, consomme, coupe, diamante, fiance, flambe, frappe, glace, habitue, macrame (derived from Turkish), manque, outre, retrousse, risque, rose (‘pink wine’), roue, saute, soigne, souffle, touche, (Amerindian/Spanish) abalone, (Italian) biennale, finale, latte, (Greek) agape (‘love feast’), (Spanish/Nahuatl) guacamole, (Japanese) anime, kamikaze and (Mexican Spanish) tamale; final <e> in (French) emigre, expose, naivete, protege, recherche, resume (‘c.v.’), retrousse (KiSwahili/Spanish), dengue and (Turkish) meze.

There is an increasing tendency to spell the French loanwords in this list, within English text, with <é>, thus signalling their status as not-yet-fully-assimilated loanwords (and my spell-checker keeps inserting <é> where I don’t want it to) – but it could also be argued that this is yet another spelling complexity for native English speakers to cope with, especially since the Compact Oxford Dictionary recognises such forms as flambés, flambéing, flambéed, which on the other hand suggests that <é> is becoming a grapheme of English – if it does, where would it fit in the alphabet and therefore dictionaries?

<ea>

only in break, great, steak, yea, Yeats

<ee>

only word-final and only in about 13 more recent loanwords where French spelling has <ée>, namely corvee, dragee (‘sugar-coated sweet’) pronounced /ˈdrɑːʒeɪ/ (also pronounced /ˈdreɪʤiː/), entree, epee, fiancee, levee (‘reception or assembly’, also pronounced with /iː/), matinee, melee, nee, negligee, puree, soiree, toupee. The tendency to use <é> is growing here too

<e.e>

only in crepe, fete, renege, suede, Therese /kreɪp, feɪt, rɪˈneɪg, sweɪd, təˈreɪz’/

<ei>

only in about 15 words, namely abseil, apartheid, beige, deign, feign, feint, heinous pronounced /ˈheɪnəs/ (also pronounced /ˈhiːnəs/), lei (only example in an open syllable), obeisance, reign, rein, reindeer, seine, sheikh, skein, surveillance, veil, vein

<eigh>

only in eight, freight, heigh, inveigh, neigh, neighbour, sleigh, weigh, weight

<er>

only word-final and only in a few more recent French loanwords, namely atelier, croupier, dossier pronounced /ˈdɒsiːjeɪ/ (also pronounced /ˈdɒsiːjə/), foyer pronounced/ˈfwaɪjeɪ, ˈfɔɪjeɪ/ (also pronounced /ˈfɔɪjə/), metier, rentier

<es>

only in demesne

<et>

only word-final and only in about 20 more recent French loanwords, namely ballet, beret, bidet, bouquet,

buffet (‘food’), cabaret, cabriolet, cachet, cassoulet, chalet, crochet, croquet, duvet, gilet, gourmet, parquet, piquet, ricochet, sachet, so(u)briquet, sorbet, tourniquet, valet pronounced /ˈvæleɪ/ (also pronounced /ˈvælɪt/). /t/ surfaces (see section 7.2), always with change of preceding vowel, in balletic with /e/, parquetry, valeting with /ɪ/

<ey>

never initial; medially, only in abeyance, heyday; word-finally, only in bey, convey, fey, grey, hey, lamprey, obey, osprey, prey, purvey, survey, they, whey

<ez>

only in laissez-faire, pince-nez, rendezvous

2-phoneme graphemes

(none)

Notes

If we follow Crystal (2012: 131-2) and Upward and Davidson (2011: 176-9), ‘more recent’ in terms of loanwords from French means after the Great Vowel Shift, which began about AD1400 and was complete by about AD1600.

<a> is regular in non-final syllables of stem words – see section 6.3. Exceptions (in addition to derived forms, e.g. daily, gaily, playing, and those listed among the Oddities above): aileron, attainder, caitiff, complaisant, dainty, daisy, gaiter, liaison, maintain, plaintiff, plaintive, raillery, raisin, traitor, wainscot, all with <ai>.

<a.e> is regular in closed final syllables, including not only the large number of mono- and polysyllables with a single final consonant phoneme spelt with a single letter, but also:

  • five words with two consonant letters forming a digraph representing a single consonant phoneme separating <a.e>: ache, champagne, bathe, lathe, swathe
  • the small group of words ending in /eɪnʤ/ spelt <-ange>: arrange, change, grange, mange, range, (e)strange, with two consonant phonemes separating <a.e> (no exceptions)
  • the small group of words ending in /eɪst/ spelt <-aste>: baste, chaste, haste, lambaste (which has the variant form lambast, with /æ/), paste, taste, waste, still with two consonant phonemes separating <a.e> (exceptions: only waist as a stem word, but confusion is possible with a number of past-tense verbs, namely based, chased, paced).

The only monosyllable in which /eɪ/ is spelt just <a> without another vowel letter (and irregularly before a doubled spelling) is bass /beɪs/ ‘(player of) large stringed instrument’/‘(singer with) low-pitched voice’.

<ay> is regular in word-final position in both mono- and polysyllabic stem words (for exceptions see the Oddities above), and rare elsewhere in stem words: the only medial examples seem to be claymore, mayhem, and there are none in initial position. However, medial <ay> spelling /eɪ/ does also occur in compound words, e.g. always, hayfever, maybe, playground, and frequently before suffixes, e.g. playing.

Fortunately, there are no occurrences of word-final /eɪ/ spelt <ai>, thus reducing the possibility of confusion with <ay>, but all words with /eɪ/ spelt <ai> are still exceptions, either to the prevalence of <a> in non-final syllables or to the prevalence of <a.e> in closed final syllables.

The only useful sub-rule is that <ai> is regular before /nt/, e.g. (in monosyllables) faint, paint, plaint, quaint, saint, spraint, taint (only exceptions: ain’t (sort of), feint); (in polysyllables) acquaint, attaint, complaint, con/di/re-straint, plus Aintree, dainty, maintain, maintenance, plaintiff, plaintive (the last six words being apparently the only examples before /nt/ in non-final syllables of stem words), but it is not predictable by rule elsewhere. The rule that <ai> is regular before /nt/ is one of only two cases where the spelling of a rime/phonogram is more predictable as a unit than from the separate phonemes and there are enough instances to make the rule worth teaching – see section A.7 in Appendix A.

About half (by text frequency) of /eɪl, eɪn/ spellings in closed monosyllables have <-ail, -ain>, e.g. ail, bail, fail, flail, frail, grail, mail, hail, jail, pail, quail, rail, sail, snail, tail, trail, wail (also cf. Braille); brain, chain, drain, fain, gain, grain, main, pain, plain, rain, sprain, stain, strain, swain, train, twain, vain, wain and the irregular past participles lain, slain (see third paragraph below), but this means that these groups have maximum confusability with words in <-ale, -ane>, etc., e.g. (and listing just a few that are homophones of words in <-ail, -ain>) ale, bale, male, hale, pale, sale, tale, whale; gaol; fane, lane, mane, pane, plane, vane, wane; feign, reign; rein, vein, and all these words have to be learnt individually.

This is also true of:

1) polysyllables with <ail, ain(e)>: (in non-final syllables) aileron, daily, gaily, raillery; attainder, wainscot; (in final syllables) assail, avail, curtail, detail, entail, entrails, prevail, retail, travail, wassail; ascertain, chilblain, cocaine, contain, disdain, domain, entertain, explain, migraine, moraine, obtain, pertain, plantain, quatrain, remain, terrain

2) the ragbag of other words (mono- and polysyllables) with <ai>, where there is also considerable potential for confusion: (in non-final syllables) caitiff, complaisant, daisy, gaiter, lackadaisical, liaison, raisin, traitor; (in final syllables) afraid, aid, aide, aim, aitch, arraign, bait (contrast bate), baize (contrast bays), braid (contrast brayed), braise (contrast brays), campaign (contrast champagne), (de/ex/pro-)claim, cockaigne, faith, gait (contrast gate), liaise, maid (contrast made), maim, maize (contrast maze), malaise, mayonnaise, plaice (contrast place), praise, raid, raise (contrast raze), staid (contrast stayed), staithe, strait (contrast straight), traipse, waif, waist (contrast waste), wait (contrast weight), waive (contrast wave), wraith and the irregularly-spelt (but not irregularly-pronounced) past tenses and participles laid, paid (see next paragraph).

As pointed out under /d/, section 3.5.2, the spellings laid, paid are irregular; the regular spellings would be *layed, *payed. But for the irregular past participles of lie (‘be horizontal’), slay the spellings lain, slain seem preferable to *layn, *layen, *slayn, *slayen, and can perhaps be counted as extensions of the general <y>-replacement rule – see section 6.5.

There seem to be no words ending in /eɪ/ spelt <er> taking suffixes beginning with vowel letters, and therefore no /r/-linking (section 3.6). If so, this is the only such category.

5.7.2 /iː/ as in eel

For the absence of percentages see Notes.

The main system

For all these categories see Notes.

Basic grapheme

<ee>

e.g. eel (virtually the only occurrence in initial position), beech, bee, see

Other frequent graphemes

<e>

e.g. ether, lever, be. Regular in non-final syllables of stem words. In closed monosyllables, apparently only in retch

<ea>

e.g. each, beach, sea

<i>

e.g. chic (only example in a closed stem monosyllable), alien, litre, ouija, safari

<y>

almost entirely word-final, where it is the regular spelling in polysyllables, e.g. city, plus rare medial examples, e.g. caryatid, embryo, halcyon, polysyllable

Rare graphemes

<e.e>

rare in closed monosyllables; regular in closed final syllables of polysyllabic words, e.g. complete, discrete, grapheme, phoneme

<ie>

never initial. In non-final syllables only in chieftain, diesel. Otherwise only in final syllables: (closed monosyllables) brief, chief, fief, field, fiend, frieze, grief, grieve, lief, liege, mien, niece, piece, priest, shield, shriek, siege, thief, thieve, wield, yield; (open monosyllable) brie (only); (closed final syllables of polysyllables) achieve, aggrieve, Aries, belief, believe, besiege, hygiene, relief, relieve, reprieve, retrieve, series, serried, species; (open final syllables of polysyllables) aerie, anomie, auntie, Aussie, birdie, bogie, bolshie, bonhomie, boogie, bookie, bourgeoisie, bowie, brassie, budgie, caddie, calorie, camaraderie, chappie, collie, commie, conscie, cookie, coolie, coterie, cowrie, curie, darkie, dearie, eerie, eyrie, gaucherie, genie, g(h)illie, girlie, goalie, hoodie, laddie, lassie, lingerie pronounced /ˈlænʒəriː/ (also pronounced /ˈlɒnʤəreɪ/), luvvie, menagerie, movie, nightie, organdie, pixie, prairie, reverie, rookie, quickie, specie, stymie, sweetie, talkie, zombie. For ‘<i> before <e> except after <c>’ see section 6.1

The rest

Oddities

<ae>

only in (in non-final syllables) aegis, aeon, aesthet-e/ic, anaemi-a/c and other words ending /ˈiːmiːjə, ˈiːmɪk/ spelt <-aemi-a/c>, anaesthetist, archaeolog-ical/ist/y, Caesar(ian), caesura, encyclopaedia, faeces,

haemoglobin, Linnaean, Manichaean, mediaeval, naevus, paean, palaeolithic, praetor, quaestor. Many of these words have alternative spellings in <e>, especially in US spelling; (in final syllables – always open; no examples in closed final syllables) algae, alumnae, antennae, formulae, larvae, personae, pupae, vertebrae

<ay>

only finally and only in quay and compounds of day (birthday, holiday, Sunday, yesterday, etc.), except heyday, midday, nowadays, today, workaday, which have /eɪ/, as does holidaying

<ei>

only medial and only in: (in non-final syllables) ceiling, cuneiform, disseisin, (n)either pronounced /ˈ(n)iːðə/ (also pronounced /ˈ(n)aiːðə/), heinous pronounced /ˈhiːnəs/ (also pronounced /ˈheiːnəs/), inveigle, plebeian; (in final syllables) caffeine, casein, codeine, conceit, conceive, counterfeit (also pronounced with /fɪt/), deceit, deceive, perceive, protein, receipt, receive, seize. For ‘<i> before <e> except after <c>’ see section 6.1

<eo>

only in feoffee, feoffment, people

<ey>

except in geyser pronounced /ˈgiːzə/, only final and only in abbey, alley, attorney, baloney, barley, blarney, blimey, cagey, chimney, chutney, cockney, comfrey, coney, donkey, dopey, flunkey, fogey, galley, gooey, hackney, hockey, homey, honey, jersey, jockey, journey, key, kidney, lackey, malarkey, matey, medley, money, monkey, motley, nosey, palfrey, parley, parsley, pokey, pulley, storey, tourney, turkey, valley, volley

<i.e>

only in closed final syllables, but in at least 70 words – see Table 5.6 and the note below it

<is>

only finally and only in chassis, commis (chef), coulis, debris, precis, verdigris pronounced /ˈvɜːdɪgriː/ (also pronounced /ˈvɜːdɪgriːs/), vis-à-vis (last syllable)

<it>

only finally and only in esprit, petit mal, wagon-lit

<oe>

only in non-final syllables and only in amenorrhoea, amoeba, apnoea, coelacanth, coelenterate, coeliac, coelom, coenobite, coenocyte, diarrhoea, dyspnoea, foetal, foetid, foetus, gonorrhoea, lactorrhoea, logorrhoea, oedema, oenology, oesophagus, oestrogen, oestrus, phoenix, pyorrhoea, subpoena, plus

onomatopoeia, pharmacopoeia if <ia> is taken as spelling /iːjə/. Many of these words have alternative spellings in <e>, especially in US spelling

<ois>

only in chamois (the leather, pronounced /ˈ∫æmiː/ (also spelt shammy), as opposed to the animal from whose skin it is made, pronounced /ˈ∫æmwɑː/)

2-phoneme graphemes

(none)

Notes

The reason for the absence of percentages here is my re-allocation of word-final <y> to /iː/ rather than /ɪ/ (see section 5.4.3), and of many of Carney’s /ɪə/ words to /iːjə/ (see section 5.6.4). Carney doesn’t give enough information on either set of words to calculate the effect of these re-allocations on the percentages for /iː/.

Unlike most of the split digraphs, <e.e> is not very frequent – in Carney’s analysis (excluding final /iː/ spelt <y>) it accounts for only 3% of spellings of /iː/, and for only 27% even in monosyllables, and percentages counting in final /iː/ spelt <y> would be even lower. It is the second rarest of the split digraphs, the rarest being <y.e>.

The regular spellings of /iː/ are:

  • in open and closed monosyllables: <ee>
  • in open final syllables (= stem-finally) in polysyllables: <y>
  • in closed final syllables of polysyllables: <e.e>
  • in non-final syllables, especially of stem words: <e>, but there are large numbers of exceptions with <i>.

In open monosyllables the regular spelling is <ee>, as in bee, fee, flee, free, gee, ghee, glee, knee, lee, pee, scree, see, spree, tee, thee, three, tree, twee, wee. Exceptions: be, he, me, she, the (when stressed), we, ye – but these are all function words, which don’t have to obey the Three-Letter Rule (see section 4.3.2); flea, lea, pea, plea, sea, tea; key; ski; brie.

In closed monosyllables Carney (1994: 157-8) lists 108 words with <ea>, 87 with <ee>, and 30 with minority spellings. However, those with <ee> seem more frequent, e.g. beech, cheek, cheese, deep, feed, feek, feet, geese, green, keep, meet, need, seem, seen, sleep, sleeve, sneeze, speech, speed, street, week, wheel. Also, analysing <ea> as the regular spelling here would seem odd, given that <ea> has several other correspondences, some of them with long lists of words, while <ee> has hardly any and they are all rare. I therefore take <ee> to be most regular spelling of /iː/ in closed monosyllables. For exceptions see Table 5.6, plus chic, retch, seize.

In open final syllables of polysyllables the regular spelling (in my analysis, as against Carney’s) is <y>, e.g. city, pretty. Exceptions include:

  • those listed under the rare grapheme <ie> and the Oddities <ae, ay, ey, is, it, ois> above
  • aborigine, acme, acne, adobe, anemone, apostrophe, bocce, catastrophe, coyote, dilettante, epitome, extempore, facsimile, (bona) fide, hebe, hyperbole, Lethe, machete, menarche, minke, nepenthe, oche, posse, psyche, recipe, reveille, sesame, simile, stele, strophe, tagliatelle, tsetse, ukulele, vigilante
  • a few words in <-e> where pronunciation of the final phoneme varies between /iː/ and /eɪ/: abalone, cafe, canape, finale, forte, furore, guacamole, kamikaze, karate
  • one word in <-ea>: guinea
  • all the words ending in <-ee> indicating ‘person to whom something is done’ (all with final stress), e.g. addressee, amputee, appointee, assignee, conferee, debauchee, dedicatee, deportee, divorcee, draftee, employee, enrollee, examinee, grantee, inductee, internee, interviewee, invitee, legatee, lessee, licensee, mortgagee, nominee, parolee, patentee, payee, referee, trainee, transferee, trustee, vestee
  • a ragbag of other words ending in <-ee>, including: (with initial stress) apogee, coffee, dragee (‘sugar-coated sweet’) pronounced /ˈdreɪʤiː/ (also pronounced /ˈdrɑːʒeɪ/), filigree, fricassee, gee-gee, jubilee, kedgeree, levee (‘reception or assembly’, also pronounced with /eɪ/), lychee, manatee, pedigree, perigee, Pharisee, prithee, puttee, Sadducee, spondee, squeegee, standee, suttee, thuggee, toffee, trochee, yankee; (with medial stress) committee; (with final stress) absentee, agree, attendee, banshee, bargee, bootee, buckshee, chickadee, chimpanzee, decree, degree, devotee, dungaree, escapee, goatee, grandee, guarantee, jamboree, marquee, refugee, repartee, rupee, settee, truckee
  • a further ragbag of mostly foreign words ending in <i>: anti, bikini, broccoli, chilli, confetti, deli, ennui, graffiti, khaki, kiwi, literati, macaroni, maxi, midi, mini, muesli, mufti, nazi, potpourri, safari, salami, scampi, spaghetti, sari, semi, shufti, stimuli, taxi, tsunami, umami, vermicelli, wiki, yeti, yogi.

In closed final syllables of polysyllables the regular spelling is <e.e>. For exceptions in <ei> see the Oddities, and for those in <ea, ee, ie, i.e> see Table 5.6. There are apparently just five exceptions with <i>: ambergris, aperitif, batik, massif, motif, and one with <e>: harem. There is also a small group with /iːz/ spelt <-es>, namely some plural nouns of Greek origin with the singular ending /ɪs/ spelt <-is>, e.g. analyses (/əˈnælɪsiːz/, the singular verb of the same spelling being pronounced /ˈænəlaɪzɪz/), apotheoses, axes, bases (/ˈæksiːz, ˈbeɪsiːz /, plurals of axis, basis; axes, bases as the plurals of axe, base are pronounced (regularly) /ˈæksɪz, ˈbeɪsɪz/), crises, diagnoses, emphases, exegeses, nemeses, oases, periphrases, synopses, theses and all its derivatives, plus (Greek singulars) diabetes (also pronounced with final /ɪs/), herpes, litotes, pyrites, (a stray Greek plural with singular in <-s>) Cyclopes, and (Latin plurals) amanuenses, appendices, cicatrices, faeces, interstices, mores, Pisces, testes.

A very odd word that is relevant here is dioceses. In its singular form diocese, pronounced /ˈdaɪjəsɪs/, each phoneme (except the automatic /j/-glide) can be related to a grapheme, provided the final /s/ is analysed as spelt <se>. But the plural has the two pronunciations /ˈdaɪjəsiːzɪz, ˈdaɪjəsiːz/. In the former, again each phoneme (except the /j/-glide) can be related straightforwardly to a grapheme, provided we accept that the first <e> spells /iː/, both <s>’s spell /z/ (the first being voiced despite being voiceless in the singular – cf. the other words of Greek origin just listed), and the second <e> spells /ɪ/. But in the second pronunciation it seems as though /iːz/ is spelt <-eses> and I am at a loss to know which letters to relate the two phonemes to – perhaps more rational spellings would be *diosis (singular), *dioses (plural), which would bring both forms into line with those listed above, and with all other words with final /sɪs/, which are all spelt <-sis>, despite neither *diosis nor *dioses having a genuine Greek etymology.

The five major possibilities in closed final syllables of polysyllables and in closed monosyllables are shown in Table 5.6. There appear to be no useful rules suggesting when spellings other than <e.e> in closed final syllables of polysyllables and <ee> in closed monosyllables occur – all the other words just have to be learnt.

In non-final syllables of stem words the letter-name spelling <e> (see section 6.3) predominates, especially in word-initial position, where the only exceptions appear to be aegis, aeon, aesthete, eager, eagle, easel, Easter, easy, either pronounced /ˈiːðə/, oedema, oenology, oesophagus, oestrogen, oestrus. In medial syllables <e> still predominates, e.g. beauteous, completion, European, Jacobean, lever, phonemic, simultaneous, spontaneous and thousands of others.

Table 5.6: <ea, ee, e.e, ie, i.e> as spellings of /iː/ in closed final syllables.

Regular spelling

In polysyllables: <e.e>

In monosyllables: <ee> on the basis of the argument above

Exceptions (in addition to those listed under the Oddities <ei, i, ie> and in the paragraphs above this Table)

<ea>

impeach;

peace; beach, bleach, breach, each, leach, peach, pleach, preach, reach, teach

bead, knead, lead (verb), mead, plead, read (present tense)

leaf/ves, sheaf/ves

league

beak, bleak, creak, freak, leak, peak, sneak, speak, squeak, streak, teak, tweak, weak, wreak

anneal, appeal, conceal, congeal, repeal, reveal;

deal, heal, leal, meal, peal, seal, squeal, steal, teal, veal, weal, zeal

beam, bream, cream, dream, gleam, ream, scream, seam, steam, stream, team

demean;

bean, clean, dean, glean, jeans, lean, mean, quean, wean

cheap, heap, leap, neap, reap

decease, decrease, increase, release;

cease, crease, grease, lease

appease, disease;

ease, pease, please, tease

leash

beast, east, feast, least, yeast

defeat, entreat, escheat, repeat, retreat;

beat, bleat, cheat, cleat, eat, feat, heat, leat, meat, neat, peat, pleat, seat, teat, treat, wheat

heath, sheath, wreath

breathe, sheathe, wreathe

bereave

cleave, eave, greave, heave, leave, sheave, weave

<ee>

exceed, proceed, succeed; genteel; esteem, redeem; boreen, canteen, careen, colleen, dasheen, lateen, nanteen, sateen, tureen; discreet

(regular, e.g. beef, deep, feed, green, seem, week)

<e.e>

(regular, e.g. complete, discrete, grapheme, phoneme)

breve, cede, eke, eve, gene, glebe, grebe, meme, mete, Pete, rheme, scene, scheme, Steve, swede, Swede, theme, these

<ie>

achieve, aggrieve, belief, believe, besiege, hygiene, relief, relieve, reprieve, retrieve, series, species

brief, fief, field, fiend, frieze, grief, grieve, lief, liege, mien, niece, piece, priest, shield, shriek, siege, thief/ves, thieve, wield, yield

<i.e>

caprice, police; pastiche; prestige; fatigue, intrigue; automobile, imbecile; chenille; regime; benzine, brigantine, brilliantine, chlorine, cuisine, dentine, figurine, gabardine, guillotine, iodine, latrine, limousine, machine, magazine, margarine, marine, mezzanine, morphine, nicotine, opaline, phosphine, quarantine, quinine, ravine, routine, sardine, strychnine, submarine, tagine, tambourine, tangerine, terrine, trampoline, tyrosine, vaccine, wolverine; antique, boutique, critique, mystique, oblique, physique, technique, unique; cerise, chemise, expertise, valise; odalisque; pelisse; artiste, dirigiste, modiste; elite, marguerite, petite; naive, Khedive, recitative

fiche, niche, quiche; clique, pique; bisque; suite

However, there are also at least a thousand exceptions – see under the Oddities <ae, ei, ey, oe> above, plus:

  • with <i>:

    1) before <a> spelling /ə/ with automatic intervening /j/-glide (Carney would place these words under /ɪə/): ammonia, anaemia, bacteria, begonia, camellia, chlamydia, (en)cyclopaedia, hernia, hysteria, media, myopia, salvia, sepia, utopia; amiable, dutiable, enviable, variable; congenial, jovial, managerial, material, memorial, radial, remedial, serial and about 450 others ending in <-ial>; barbarian, comedian, grammarian, guardian, pedestrian, ruffian, thespian and about 200 others ending in <-ian>; dalliance, luxuriance, radiance, variance; suppliant, radiant, suppliant, variant; alien; audience, convenience, ebullience, experience, obedience, prurience, salience; expediency, leniency; convenient, ebullient, expedient, lenient, obedient, prescient, prurient, salient, sentient, subservient, transient; soviet; twentieth, etc.; period, sociological, axiom, accordion, bastion, battalion, bullion, carrion, centurion, clarion, collodion, ganglion, medallion, mullion, scorpion, scullion, stallion, chariot, patriot; commodious, compendious, curious, dubious, felonious, melodious, odious, previous, scabious, serious, studious, tedious and about 100 others ending in <-ious>; atrium, bacterium, compendium, gymnasium, medium, opium, potassium, radium, stadium, tedium and about 200 others ending in <-ium>; genius, radius; also second <i> in amphibious, bilious, billion, brilliancy, brilliant, criteri-a/on, delirium, editorial, fastidious, hilarious, historian, histrionic, idiom, idiot, industrial, juvenilia, memorabilia, millennia, oblivion, omniscience, omniscient, perfidious, perihelion, reptilian, resilience, resilient, trivia, vitriol, third <i> in incipient, initiate (noun), insidious, insignia, invidious, militaria;

    2) before other vowel phonemes with automatic intervening /j/-glide: ap/de-preciate, associate (verb), audio, calumniate, caviar, foliage, luxuriate, medi(a)eval, negotiate, orient (verb), oubliette, patio, radio, ratio, serviette, studio, trio, verbiage, viola (/viːˈjəʊlə/ ‘musical instrument’); also first <i> in conscientious, liais-e/on, orgiastic, partiality, psychiatric, speciality, second <i> in histrionic, inebriation, insomniac, officiate, superficiality, vitriolic, third <i> in initiate (verb)

    3) before a consonant phoneme other than /j/: albino, ballerina, cappuccino, casino, cliché, concertina, farina, frisson, kilo, Libra, lido, litre, maraschino, merino, mosquito, ocarina, pinochle, piquant, scarlatina, semolina, visa; also first <i> in kiwi, martini, migraine, milieu, second <i> in bikini, incognito, libido;

  • with other main-system graphemes: beacon, beadle, beagle, beaker, beaver, creature, deacon, feature, heathen, meagre, measles, queasy, reason, season, sleazy, squeamish, teasle, treacle, treason, weasel; beetle, cheetah, feeble, freesia, gee-gee, geezer, needle, squeegee, sweetie, teeter, wheedle; chieftain and other compounds of chief-, diesel; caryatid, embryo, halcyon, polyandry, polysyllable, polytechnic and many others with poly-.

    5.7.3 /aɪ/ as in ice

The main system

For all these categories see Notes.

Basic grapheme

<i.e>

40% (70% in
monosyllables)

regular in closed final syllables (except in monosyllables before /t/ and consonant clusters; only other exception: mic /maɪk/, short for microphone), e.g. bike, sublime

Other frequent graphemes

<i>

42% (with <ie
(see Oddities), y>)

regular in non-final syllables, e.g. item, word-finally in polysyllables, e.g. alkali, and in monosyllables before consonant clusters, e.g. child

<y>

e.g. beautify, by, cycle, psyche, sky; regular word-finally

<igh>

13%

only in about 26 stem words (see section 10.25). Regular in monosyllables before /t/, e.g. sight. In non-final syllables, only in blighty, righteous, sprightly. Word-finally, only in high, nigh, sigh, thigh

The rest

Oddities

5% in total

<a>

only in majolica pronounced /maɪˈjɒlɪkə/ (also pronounced /məˈʤɒlɪkə/), naif, naive, papaya

<ae>

only in maestro, minutiae

<ai>

only in ailuro-phile/phobe, assegai, balalaika, banzai, bonsai, caravanserai, Kaiser, naiad, samurai, shanghai

<ais>

only in aisle. See Notes

<aye>

only in aye (‘yes’), aye-aye

<ei>

only in deictic, deixis, eider(down), eidetic, eirenic, either, Fahrenheit, feisty, gneiss, heist, kaleidoscope, meiosis, neither, poltergeist, seismic, stein

<eigh>

only in height, sleight

<ey>

only in geyser pronounced /ˈgaɪzə/ (usually pronounced /ˈgiːzə/)

<eye>

only in eye

<ia>

only in diamond

<ie>

only word-final, e.g. pie (see Notes), except for suffixed forms after <y>-replacement (see section 6.5), e.g allied, supplies

<ir>

only in iron pronounced /ˈaɪjən/ (but the Scots pronunciation /ˈaɪrən/ has retained the /r/ and has more regular correspondences)

<is>

only in island, isle(t), lisle, viscount. See Notes

<oy>

only in coyote

<ui>

only in duiker, Ruislip

<ye>

only word-final and only in bye, dye, lye, rye, Skye, stye

<y.e>

only in final syllables and only in: (monosyllables) byte, chyle, chyme, cyme, dyke, dyne, gybe, gyve, hythe, hype, rhyme, scythe, skype, style, syce, syne, thyme, tyke, type; (polysyllables) acolyte, analyse, anodyne, azyme, breathalyse, catalyse, coenocyte, condyle, dialyse, electrolyse, electrolyte, enzyme, formaldehyde, leucocyte, neophyte and at least 14 other words ending in /faɪt/ spelt <-phyte>, paralyse, phagocyte, proselyte, spondyle, about 20 derivatives of style, troglodyte, and at least 20 derivatives of type. In my opinion, all these words (except gyve) could be spelt with <i.e> without loss

2-phoneme graphemes

/aɪə/

(1)
spelt <ir>

only medially and mainly where <-e> has been deleted from words in the next category, e.g. aspiring, desirous, expiry, spiral, tiring, but there are a few independent examples, e.g. biro, giro, pirate, virus. In all cases the <r> is both part of <ir> spelling /aɪə/ and a grapheme in its own right spelling /r/. For dual-functioning see section 7.1

(2)
spelt <ire>

only word-final and only in ac/in/re-quire, admire, a/con/in/per/re/tran-spire, attire, desire, dire, empire, entire, (expire, fire, hire, (be/quag-)mire, quire, saltire, samphire, sapphire, satire, shire, sire, spire, e)squire, tire, umpire, vampire, wire. Many of these words allow /r/-linking, e.g. inspiration, satirical, spiral, wiring – see previous paragraph and section 3.6

(3)
spelt <yr>

only medial and only in empyrean, gyroscope, papyrus, pyrites, pyromaniac, thyroid, tyrant, tyro, tyrosine. In all cases the <r> is both part of <yr> spelling /aɪə/ and a grapheme in its own right spelling /r/. For dual-functioning see section 7.1. Words in which <y, r> are separate graphemes include dithyramb(ic), myriad, porphyr-y/ia, tyranny, syringa, syringe, syrup, all with the relevant <y> spelling /ɪ/

(4)
spelt <yre>

only word-final and only in byre, gyre, lyre, pyre, tyre. In my opinion these words could be spelt with <ire> without loss, as tire already is in US English. Some of these words allow /r/-linking – see section 3.6 – e.g. pyromaniac, and (with change of vowel and <r> spelling only /r/) lyrical

/waɪ/ spelt <oy>

only in foyer pronounced /ˈfwaɪjeɪ/ (also pronounced /ˈfɔɪjeɪ, ˈfɔɪjə/), voyeur

3-phoneme grapheme

/waɪə/ spelt with a single grapheme <oir>

only in choir – one of only two 3-phoneme graphemes in the entire language

Notes

The regular spellings of /aɪ/ are:

  • in non-final syllables, and in monosyllables before consonant clusters: <i>
  • in monosyllables before /t/: <igh>

    in closed final syllables (except in monosyllables before consonant clusters and /t/): <i.e>

  • word-finally: <y>.

<i> is regular in non-final syllables (see section 6.3), but for:

  • exceptions listed under the Oddities <a, ae, ai, aye, ei, ey, ir, is, oy, ui> and the 2-phoneme grapheme /waɪ/ spelt <oy>, above, plus Blighty, righteous, sprightly (also spelt spritely because of its derivation from sprite)
  • exceptions with <y>: asylum, aureomycin, cryostat, cyanide, cycle, cyclone, cypress, (hama)dryad, dynamic, forsythia, glycogen, gynaecology, hyacinth, hyaline, hybrid, hydra, hydrangea, hydraulic, hydrofoil, hydrogen and various other compounds of hydro-, hyena, hygiene, hygrometer, hymen, hyperbole and other compounds in hyper-, hyphen, hypothesis and other compounds in hypo-, lychee, myopic, nylon, psyche and almost all its derivatives (exception: metempsychosis, with /ɪ/), pylon, stymie, thylacine, thymus, typhoid, typhoon, typhus, xylophone, zygote and derivatives.

<i.e> is regular in closed final syllables of polysyllabic words, e.g. alive, archive, capsize, combine, concise, decide, entice, exercise, oblige, senile, sublime. Exceptions: see the Oddities listed above under <y.e>, plus alight, behind, delight, Fahrenheit, fore/hind/in-sight, indict, paradigm, remind, uptight, watertight and suffixed forms after <y>-replacement (section 6.5), e.g. allied, supplies.

In closed monosyllables:

  • <i> on its own appears to be regular before consonant clusters: child, Christ, mild, ninth, pint, whilst, wild and the group with /aɪnd/ spelt <-ind>: bind, blind, find, grind, hind, kind, mind, rind, wind (‘turn’). Possible extension: If the context were defined in terms of letters, aisle, climb, isle, lisle could be added. Exception under either definition: heist
  • <igh> is regular before /t/: bight, blight, bright, fight, flight, fright, hight, knight, light, might, night, plight, right, sight, slight, tight, wight, wright. Exceptions: height, sleight; bite, cite, kite, mite, rite, site, smite, spite, sprite, quite, white, write; byte
  • <i.e> is regular around other single consonant phonemes, e.g. fine, hive, ice, knife, like, lime, mile, mine, pipe, prize, ride, rise, including the small group with /ð/ spelt <th>: blithe, lithe, tithe, writhe. Exceptions: see the Oddities listed above under <y.e>, plus aisle, climb, isle, lisle (but for these four words see two paragraphs above), mic, stein.

In open final syllables of polysyllabic words <y> is regular, e.g. in 130+ words with the suffix <-fy>, e.g. beautify, and in ally, ap/com/im/re/sup-ply, defy, deny, descry, espy, July, multiply (verb), occupy, prophesy, rely. Exceptions: assegai, shanghai; aye-aye; a fortiori/posteriori/priori, alibi, alkali, alumni, alveoli, foci, fundi (plural of fundus), fungi, Gemini, gladioli, rabbi and a few more rare words.

In open monosyllables the most frequent spelling is <y>: by, cry, dry, fly, fry, my, ply, pry, scry, shy, sky, sly, spry, spy, sty, try, why, wry, plus buy, guy (taking <bu, gu> to be digraphs spelling /b, g/); this set numbers 20 words. Exceptions (which number 24): aye, eye, I; die, fie, hie, lie, pie, tie, vie; bye, dye, lye, rye, stye; high, nigh, sigh, thigh; and the Greek letter names chi, phi, pi, psi, xi. A possible subrule might say that <ie> is regular after a single consonant letter, but this is a very small set, containing only the seven words die, fie, hie, lie, pie, tie, vie, and setting up this rule would cause problems for the grapheme-phoneme correspondences of <ie>.

The words aisle, island, isle(t), lisle, viscount are among the oddest in English spelling, with <(a)is> spelling /aɪ/ and the <s> having no consonantal value. Isle, lisle might have yielded to an analysis with /aɪ/ spelt <i.e> and the intervening <sl> spelling /l/ - but there is no other warrant for a grapheme <sl>, or for the ‘split trigraph’ <ai.e> which this analysis would have produced for aisle (whereas there is another warrant for the grapheme <ais>, in palais – see under /eɪ/, section 5.7.1). Also, this analysis would not have fitted the other words listed (or those with /iː/ spelt <is>, see section 5.7.2).

/waɪə/ also has the 2-grapheme spellings <wire> in wire and <-uire> in (ac/re)quire. And /aɪə/ has 2-grapheme spellings, e.g. <-iar> in liar, <-ier> in drier, <-yer> in dryer, flyer, <-igher> in higher. A possibly useful sub-rule is that word-initial /daɪə/ is always spelt <dia-> (derived from a Greek prefix) except in dire itself and diocese.

The <y> in coyote, foyer, voyeur is both part of the digraph <oy> spelling /(w)aɪ/ and also spells /j/ on its own. For dual-functioning see section 7.1.

5.7.4 /əʊ/ as in oath

The main system

For all these categories see Notes.

Basic grapheme

<o>

59%

regular in non-final syllables, e.g. focus, finally in polysyllables (except in two-syllable words after /l, r/), and in closed monosyllables before a consonant cluster

Other frequent graphemes

<o.e>

16%

(72% in
monosyllables)

regular in final closed syllables (except in closed mono-syllables before a consonant cluster), e.g. bone, remote

<ow>

18%

regular finally in two-syllable words after /l, r/ and in open monosyllables

The rest

Oddities

8% in total

<aoh>

only in pharaoh

<au>

only in chauffeu-r/se, chauvinis-m/t, gauche, hauteur, mauve, saute, taupe

<eau>

only word-final and only in bandeau, beau, bureau, chateau, flambeau, gateau, plateau, portmanteau, rondeau, tableau, trousseau and a few other very rare words. For the plurals of these words see /z/, section 3.6.7, and <x>, section 9.41

<eo>

only in Yeo, yeoman, Yeovil

<ew>

only in sew, sewn, Shrewsbury plus shew(ed), shewn (archaic spellings of show(ed), shown)

<oa>

only in (non-final syllables) gloaming; (closed final syllables of polysyllables) approach, cockroach, encroach, reproach; (closed monosyllables) bloat, boast, boat, broach, cloak, coach, coal, coast, coat, coax, croak, float, foal, foam, gloat, goad, goal, goat, groan, groat, hoax, loach, load, loa-f/ves, loam, loan, loath, loathe, moan, moat, oaf, oak, oast, oat, oath, poach, roach, road, roam, roan, roast, shoal, soak, soap, stoat, throat, toad, toast, woad; (finally) cocoa, whoa

<oat>

only in boatswain pronounced /ˈbəʊsən/ (also pronounced /ˈbəʊtsweɪn/)

<oe>

except in throes, only word-final and only in aloe, doe, floe, foe, hoe, oboe, roe, schmoe, sloe, toe, woe. See also sections 4.3.2 and 6.6

<oh>

only in doh, kohl, Oh, ohm, soh

<ol>

only in folk, Holborn, holm, yolk and old-fashioned pronunciation of golf as /gəʊf/

<oo>

only in brooch

<ore>

only in forecastle pronounced /ˈfəʊksəl/(also pronounced /ˈfɔːkɑːsəl/)

<os>

only in apropos

<ot>

only in argot, depot, entrepot, haricot, jabot, matelot, potpourri, sabot, tarot, tricot. /t/ surfaces in sabotage, saboteur – see section 7.2 – where the <o> spells /ə/

<ou>

only in boulder, bouquet pronounced /bəʊˈkeɪ/ (also pronounced /buːˈkeɪ/), mould, moult, poultice, poultry, shoulder, smoulder, soul

<ough>

only in dough, furlough, (al)though

<owe>

only in owe

2-phoneme graphemes

(none)

Notes

<o> is regular in non-final syllables, and the only exceptions I can find in stem words are boulder, bouquet pronounced /bəʊˈkeɪ/, chauffeu-r/se, chauvinis-m/t, hauteur, gloaming, poultice, poultry, shoulder, smoulder, yeoman, Yeovil – though there are many more in derived forms, e.g. mould-er/y, moult-ed/ing – see section 6.3.

For nouns ending in <-o> which do or do not add <es> in the plural see section 6.6.

For ‘linking /w/’ and a few cases in which a preceding <o> reduces to /ə/ see section 3.8.7.

The group of stem monosyllables with final /əʊld/ spelt <-old> is one of only two cases where the spelling of the rime/phonogram is more predictable as a unit than from the correspondences of the separate phonemes, and there are enough instances to make the rule worth teaching; see section A.7 in Appendix A. The only monosyllabic stem word exception in British spelling is mould, and even that is spelt mold in the USA. The pattern generalises to the polysyllables listed above, plus solder. But this rule applies only to stem words, and they would have to be clearly distinguished from the past tenses/participles doled, foaled, holed, poled, polled, rolled, tolled.

For final syllables of stem words see Table 5.7.

Table 5.7: Spellings of /əʊ/ in final syllables of stem words.

N.B. The regular (default) spellings are shown in 9 point, exceptions in 7.5 point.

In polysyllables

In monosyllables

Closed

<o.e>, e.g. chromosome, remote

Extension: Just one word with a 2-letter spelling of the word-final consonant phoneme, namely cologne

Exceptions: approach, cockroach, encroach, reproach; control, enrol, extol, patrol; behold, cuckold, blind/mani-fold, marigold, scaffold, threshold (see also paragraph below Table); revolt; almost

Before a consonant cluster: <o>, e.g. bold, cold, fold, gold, hold, old, scold, sold, told, wold (see also paragraph below Table); bolt, colt, dolt, jolt, volt; don’t, wont, won’t; ghost, host, most, post

Exceptions: boast, coast, oast, roast, toast, mould (hence the more consistent US spelling mold), moult. Also exceptions in phonetic (but not orthographic) terms are coax, hoax, but cox, *hox or *coxe, *hoxe would not work, the first two because they would suggest the wrong vowel sound, the last two because <x> never occupies the ‘dot’ position in split digraphs – see section A.6 in appendix A.

Before a single consonant phoneme: <o.e>, e.g. bone (72% of spellings in monosyllables)

Extension: Five words with 2-letter spellings of the word-final consonant phoneme, namely brogue, rogue, vogue, toque, clothe

Exceptions: gauche, mauve, taupe; sewn, shew-ed/n; bloat, boat, broach, cloak, coach, coal, coat, croak, float, foal, foam, gloat, goad, goal, goat, groan, groat, loach, load, loaf, loam, loan, loath, loathe, moan, moat, oaf, oak, oat, oath, poach, roach, road, roam, roan, shoal, soak, soap, stoat, throat, toad, woad (for coax, hoax see above); kohl, ohm; folk, yolk; boll, droll, poll, roll, scroll, stroll, toll; holm; comb; both, loth, quoth, sloth, troth; brooch; soul; bowl; blown, flown, grown, known, mown, own, show-ed/n, sown, thrown

Open

In two-syllable words after /l, r/: <ow>, namely bellow, below, billow, callow, fallow, fellow, follow, hallow, hollow, mallow, mellow, pillow, sallow, shallow, swallow, tallow, wallow, whitlow, willow, yellow; arrow, barrow, borrow, burrow, farrow, furrow, harrow, marrow, morrow, narrow, sorrow, sparrow, yarrow

Exceptions: aloe, cello, furlough, tableau; bureau, burro, pharaoh, tarot

In longer words and other two-syllable words: <o>, e.g. gecko, gizmo, Leo, manifesto, potato, Scorpio, tomato, Virgo

Exceptions: bandeau, chateau, flambeau, gateau, plateau, portmanteau, rondeau, trousseau; cocoa; oboe; apropos; argot, depot, sabot, tricot; although; bestow, bungalow, elbow, escrow, furbelow, meadow, minnow, shadow, widow, window, winnow

<ow>, namely blow, bow (goes with arrow), crow, flow, glow, grow, know, low, mow, row (‘line, use oars’), show, slow, snow, sow (‘plant seed’), stow, throw, tow

Exceptions: beau; sew, shew; fro, go, lo, no, so; whoa; doe, floe, foe, hoe, roe, sloe, throe, toe, woe; doh, soh; dough, though; owe

5.7.5 /juː/ as in union

The only 2-phoneme sequence to which I accord quasi-phonemic status – see sections 1.12 and 2.4.

The main system

Basic graphemes

<u>

82% (with <u.e>)

regular in non-final syllables, e.g. pupil, union; word-final only in coypu, menu, ormolu, parvenu

<u.e>

regular in closed final syllables, e.g. attribute, mute, use

Other frequent grapheme

<ew>

15%

never initial; in non-final syllables, only in newel, Newton, pewter, steward; otherwise, only in final syllables and only in (closed) hewn, lewd, mews, newt, thews; (open) clerihew, curfew, curlew, few, hew, Kew, knew, mew, mildew, nephew, new, pew, phew, sinew, skew, smew, spew, stew, view, yew; from this (admittedly short) list, <ew> appears to be regular word-finally in monosyllables – see Notes

Rare grapheme

<ue>

percentage not known
– see Oddities

appears to be regular word-finally in polysyllables – see Notes

The rest

Oddities

3% in total (including <ue>)

<eau>

only in beauty and derivatives

<eu>

only in various words of Greek origin, e.g. eucalyptus, eucharist, eudaemonic, eugenic, eulogy, eunuch, euphemism, euphorbia, euphoria, eureka, eurhythmic, euthanasia, leukaemia, neural, neurone, neurosis, Odysseus, Pentateuch, Perseus, pneumatic, pneumonia and other words and names.

derived from Greek πνευ̑μα pneuma (‘breath’) or πνεύμων pneumon (‘lung’), pseudo and all its derivatives including (colloquial) pseud, therapeutic, Theseus, zeugma, plus various very rare words; plus (non-Greek) deuce, euchre, Eustachian, feu, feud(al), neuter, neutr-al/on, teutonic; only instances in monosyllables are deuce, feu, feud, pseud; word-final only in feu

<ewe>

only in ewe, Ewell, Ewelme

<ui>

only in nuisance, pursuit

<ut>

only in debut. /t/ surfaces in debutante – see section 7.2

<uu>

only in vacuum pronounced /ˈvækjuːm/

2-phoneme graphemes

All the graphemes in this section are 2-phoneme graphemes

Notes

All the spellings listed above are used to spell /juː/. <eau, ewe, ut, uu> are used only to spell /juː/, while the rest are used to spell both /juː/ and /uː/ – see next section.

<u> is the regular spelling of /juː/ (and /uː/) in non-final syllables (see section 6.3), e.g. pupil, union. Exceptions: see the polysyllables listed under <ew> and the Oddities <eau, eu, ui> above.

How can <u> function as the regular spelling of both /juː/ and /uː/ in non-final syllables without causing confusion? Because there are hardly any minimal pairs, words kept apart in meaning solely by the presence or absence of /j/. The only pairs I’ve been able to find in non-final syllables are beauty/booty, bootie (but not bootee, with stress on second syllable) and pewter/Pooter – and note that none of these words has <u> as the relevant spelling (and the last word is an invented surname). Similarly, I can find no minimal pairs separated only by the presence or absence of /j/ in the final syllables of polysyllables, and only a few such pairs/sets among monosyllables, namely beaut (Australian slang), butte/boot; cue(d/s), queue(d/s), Kew/coo(ed/s); cute/coot; dew, due when pronounced /djuː//do; ewe, yew, you/Oo(h)!; feud/food; few/phoo; hew(s), hue(s), Hugh(es/’s)/who(se); hewn/Hoon; Home pronounced /hjuːm/, Hu(l)me/whom; lewd/looed; lieu/loo; mew/moo; mewed/mood; mews, muse/moos; mute/moot; pew(s), Pugh(s)/poo(s), Pooh(’s); pseud/sued (to me, these are /sjuːd, suːd/ respectively, though for many speakers they are homophones, in one pronunciation or the other); puke/Pook; pule/pool; use (verb)/ooze. Some people with Welsh accents distinguish threw, through as /θrjuː , θruː/, but for most speakers these are both /θruː/. Again, it is noticeable that none of these words has <u> as the relevant spelling, though some have <ue, u.e>.

The names Hugh, Hughes, Lamplugh, Pugh are the only words containing the grapheme <ugh> (the exclamation ugh contains two graphemes, <u, gh>), but because it occurs only in names I have not added <ugh> to the inventory of graphemes.

<u.e> is regular in closed final syllables of polysyllables, e.g. attribute (only exceptions: pursuit, vacuum) and in closed monosyllables, e.g. mute, use (exceptions: deuce, feud; hewn, mews, newt, thews; Ewell, Ewelme).

<ue> is only word-final and found only in (monosyllables) cue, hue, queue; (polysyllables) ague, argue, avenue, barbecue, continue, curlicue, ensue, imbue, pursue, rescue, retinue, revenue, revue, value, venue. Despite the shortness of the list just given, <ue> appears to be the regular spelling in word-final position in polysyllables (exceptions: curfew, curlew, mildew, nephew; coypu, menu, ormolu), and therefore qualifies as part of the main system. Also, as a grapheme <ue> has only two pronunciations (see section 10.37), and one of them is /juː/.

However, in word-final position in monosyllables <ew> appears to be regular (see the list above). Exceptions: ewe; cue, hue, queue.

/juː/ also has 2-grapheme spellings, e.g. in adieu, view, yew, you where /j/ is spelt <i, y> - see under /j/, section 3.7.8, and /uː/, next. But the 1-grapheme spellings listed above predominate, especially <u>. Here, /j/ is not spelt separately but subsumed in the 2-phoneme spelling.

5.7.6 /uː/ as in ooze

The main system

For all these categories see Notes.

Basic grapheme

<oo>

39%

e.g. ooze, booze, zoo. Regular in closed monosyllables, e.g. zoom and about 80 other words; also regular in polysyllables both word-finally, e.g. bamboo, and in the stressed ending /ˈuːn/, e.g. afternoon, baboon; rare elsewhere

Other frequent graphemes

<u>

27% (with <u.e>)

regular in non-final syllables, e.g. rudiments, super

<u.e>

regular in closed final syllables of polysyllables, e.g. intrude, recluse

<o>

15%

only in 11 stem words and their derivatives – see Notes. Carney’s percentage excludes do, to, who, which would distort the figures

<ew>

9%

regular word-finally in monosyllables, e.g. blew

Rare grapheme

<ue>

<1%

except for gruesome, muesli and Tuesday pronounced /ˈʧuːzdiː/, only word-final and only in accrue, blue, clue, construe, (resi/sub-)due, flue, glue, imbrue, issue, rue, slue, sprue, sue, tissue, true. See Notes

Frequent

2-phoneme sequence

/juː/, with 10 spellings – see previous section

The rest

Oddities

10% in total

<ee>

only in leeward pronounced /ˈluːwəd/ (also pronounced /ˈliːwəd/)

<eu>

only in rheum(ati-c/sm), sleuth, plus adieu, lieu, purlieu pronounced /əˈdjuː, ljuː, ˈpɜːljuː/ with <i> spelling /j/ (lieu is also pronounced /luː/)

<ieu>

only in lieu pronounced /luː/ (also pronounced/ljuː/)

<oe>

only in canoe, hoopoe, shoe

<o.e>

only in combe, lose, move, prove, whose /kuːm, luːz, muːv, pruːv, huːz/ and gamboge pronounced /gæmˈbuːʒ/, plus derived forms. See Notes

<oeu>

only in manoeuvre

<ooh>

only in pooh

<ou>

7% only in

(in non-final syllables) accoutrement, acoustic, bivouac, boudoir, boulevard, bouquet, boutique, carousel, cougar, coupon, coulomb, coulter, coupe, coupon, croupier, crouton, embouchure, goujon, goulash, insouciance, louvre, moussaka, oubliette, outré, ouzo, pirouette, rouble, roulette, routine, silhouette, soubrette, soufflé, souvenir, toucan,
toupee
, troubadour, trousseau, voussoir

(in closed final syllables) ampoule, barouche, canteloupe, cartouche, (un)couth, croup, douche, ghoul, group, joule, mousse, recoup, rouge, route, soup, troupe, wound (‘harm’)

(finally) bayou, bijou, caribou, frou-frou, marabou, sou, you

<oue>

only in denouement, moue

<ough>

only in brougham, through

<oup>

only in coup

<ous>

only in rendezvous

<out>

only in mange-tout, ragout, surtout

<oux>

only in billet-doux, roux

<ui>

only in bruise, bruit, cruise, fruit, juice, recruit,
sluice
, suit

<uu>

only in muumuu (twice)

Other 2-phoneme graphemes

(none)

Notes

On <oo> see also Notes under /ʊ/, section 5.4.6.

All the spellings listed above are used to spell plain /uː/. Those beginning with <o> are used only to spell plain /uː/, while those beginning with <e, u> (except <ee>) are also used to spell /juː/ – see previous section.

No rules can be given for when /uː/ is spelt <o> because it occurs only in the following 11 stem words: (monosyllables) do, to, tomb, two, who, womb; (polysyllables) caisson pronounced /kəˈsuːn/, canton (‘provide accommodation’, pronounced /kænˈtuːn/), catacomb, lasso, zoology, plus derivatives including cantonment, lassoing, whom, derivatives of zoology with initial <zoo-> (Greek, ‘living thing’) forming two syllables pronounced /zuːˈwɒ/ if the second syllable is stressed, otherwise /zuːwə/, derived forms of the very few words in which /uː/ is spelt <o.e> (see Oddities), e.g. approval, movie, removal, and the proper nouns Aloysius /æluːˈwɪʃəs/, Romania, Wrotham /ˈruːtəm/.

<u> is the regular spelling of /uː/ (and /juː/) in non-final syllables, e.g. rudiments, super – see section 6.3. Exceptions (in addition to derivatives of words with /uː/ spelt <o, o.e>, e.g. cantonment, lassoing; ap/dis/im/re-prove, approval, movie, remove, and among the Oddities above): brewer, jewel, sewage, sewer (‘foul drain’); bazooka, booby, boodle, boogie, boomerang, booty, canoodle, coolie, doodle(bug), googly, hoodoo, hoopoe, kookaburra, loony, moolah, noodle, oodles, oolong, poodle, voodoo, plus Aloysius, Romania, Wrotham. For how <u> functions as the regular spelling of both <uː> and <juː> in non-final syllables, see previous section.

In closed final syllables of polysyllables:

  • the stressed ending /uːn/ is mostly spelt <-oon>: afternoon, baboon, bassoon, buffoon, cartoon, cocoon, doubloon, dragoon, festoon, harpoon, lagoon, lampoon, macaroon, maroon, monsoon, octaroon. Exceptions: caisson pronounced /kəˈsuːn/ (also pronounced /ˈkeɪsən/), canton (‘provide accommodation’, pronounced /kænˈtuːn/)
  • otherwise the regular spelling is <u.e>, namely in include, intrude and various other words in <-clude, -trude>, plus peruque, abstruse, recluse, peruse, brusque /bruːsk/ (also pronounced /brʌsk/), etc. For exceptions see Oddities, plus vamoose.

Exceptions to the rule that <oo> is the regular spelling in closed monosyllables are:

  • with <u.e>: spruce, truce; ruche; crude, prude, rude; luge; fluke; rule, tulle; brume, flume, plume; prune, rune; jupe; ruse; brute, chute, flute, jute, lute
  • others: rheum, sleuth; shrewd, strewn; tomb, whom, womb; combe, lose, move, prove, whose; croup, douche, ghoul, group, joule, louche, mousse, rouge, route, soup, troupe, wound (‘harm’), youth; ruth, truth; bruise, bruit, cruise, fruit, juice, sluice, suit.

In word-final position the most frequent spellings are <-oo> in polysyllables, <-ew> in monosyllables:

  • polysyllables: ballyhoo, bamboo, buckaroo, cockatoo, cuckoo, didgeridoo, hoodoo, hullaballoo, kangaroo, kazoo, shampoo, taboo, tattoo, voodoo. Exceptions: adieu, purlieu; cashew, eschew, purview, review; lasso; caribou, marabou; ecru, guru, jujitsu, juju, impromptu; accrue, construe, imbrue, issue, residue, subdue, statue, tissue, virtue
  • monosyllables: brew, chew, crew, dew pronounced /ʤuː/, Jew, screw, shrew, strew, view, yew and the irregular past tenses blew, drew, flew, grew, slew, threw. Exceptions: lieu (/luː/); do, to, who; boo, coo, goo, loo, moo, shoo, too, poo, woo, zoo; sou, you; flu, gnu (taking <gn> as spelling /nj/, though gnu could alternatively be analysed as having (like gnat, gnaw, etc.) /n/ spelt <gn> and /juː/ spelt <u> - take your pick); blue, clue, due pronounced /ʤuː/, flue, glue, rue, sue, true.

Despite the rarity of <ue> as a spelling of /uː/ it has to be counted as part of the main system because as a grapheme (see section 10.37) it has only two pronunciations, both frequent, and one of them is /uː/.