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

Appendix A: Assumptions and technicalities

© 2015 Greg Brooks, CC BY http://dx.doi.org/10.11647/OBP.0053.12

A.1 Citation forms

This book is almost entirely concerned with the citation forms of words, that is, how they would be pronounced by people with RP accents who were asked to read them aloud from a list, and/or how the words’ pronunciations are transcribed in broad IPA in the Cambridge English Pronouncing Dictionary. However, quite a few words have what Carney (1994) calls ‘allegro’ and ‘lento’ pronunciations, that is in more rapid and less rapid speech, and both may well feature in their citation forms if a sufficient sample of people is polled. I cover a few variants of this sort (see especially section 6.10), but it would be impossible to cover all of them.

A.2 Phonemes

Phoneticians disagree profoundly about the acoustic existence of phonemes. However, for the purposes of analysing any alphabetic spelling system it seems to me that assuming the psychological reality of phonemes is inescapable, and I have proceeded on that assumption. One justification might be that there are no possible correlations between parts of letters and aspects of the acoustic signal. Another might be that otherwise it is difficult to imagine how alphabets came to be invented in the first place.

I have also assumed that long vowels take longer to pronounce than short vowels, even though the acoustic evidence shows this to be only partly true, if at all.

If phonemes are assumed to have some reality, how are they to be defined? The most basic definition is the one I offer in chapter 1: ‘distinctive speech sounds’, that is, differences in sound which make a difference to the meanings of words. Thus in English /b, p/ are phonemes because the words bad, pad (and many others) which differ in meaning differ in speech only in this respect. But a fuller definition would make it clear that phonemes exist in a dynamic system with others within (a particular variety of) a particular language.

So distinctions which are phonemic in English may not be in some other languages (e.g. /l, r/ are not separate phonemes in Japanese or Kikuyu), while distinctions that are not phonemic in English may be so elsewhere. For example, unaspirated and aspirated /k, kʰ, p, pʰ, t, tʰ/ are not phonemic in English (and are therefore difficult for monolingual speakers of English to tell apart without training) because the unaspirated versions occur only after /s/ (try holding a hand in front of your mouth and saying pin, spin and notice the puff after the /p/ in pin and its absence in spin) – but are phonemic in many languages of the Indian sub-continent (and are thought to have been so in classical Greek, where the six phonemes were written κ, χ, π, φ, τ, θ respectively (in modern Greek χ, φ, θ represent /x (as in Scots loch), f, θ/ respectively – and this shows where the values of two IPA symbols have come from).

A.3 Syllables

Though difficult to define rigorously, syllables are intuitively obvious – psycholinguists showed many decades ago that children can be taught very quickly how to count (or indicate by moving the right number of pebbles or other symbols) the syllables in words spoken to them by an experimenter – and that phonemes are much less intuitive and more difficult to count.

In strict linguistic terms, therefore, as just implied, only spoken words have syllables, and written words do not – but ordinary usage can be very confusing here since dividing and hyphenating words at line ends in print is called ‘syllabification’ (in Britain; ‘syllabication’ in North America). The word extra can be used to show the difference. If this word ever needed to be split between lines it would presumably appear as:

ex-

tra,

but its spoken syllables are /ˈek - strə/, with the two phonemes represented by the <x> belonging to separate syllables.

There are two reasons for insisting that only spoken words have syllables. First, imagining that written words have syllables further confounds already confused attempts to predict word stress from the written forms of words (see section A.10 below).

Secondly, even if it was thought useful to try to define syllables within written words, this would very quickly lead to problems. As the example of extra just demonstrated, it is often difficult, sometimes impossible, to say where the boundaries between written syllables are.

However, all of this poses a problem for the grapheme-phoneme sections of this book, chapters 9 and 10 – especially chapter 10 – because it is sometimes necessary even so to refer to syllables, and therefore (explicitly or by implication) to the spoken forms of words, including resorting to circumlocutions such as ‘the syllable containing /s/ spelt <ce>’. Where this pinches most is in sections 10.41-42, where I attempt to give general rules for the pronunciation of the vowel letters as single-letter graphemes in monosyllables and polysyllables respectively, and in section A.10 below, where I summarise the difficulties involved in trying to predict where the stresses fall on English words, given only their written forms. For my attempt to get round some of this with clear definitions see the heading of section 10.41 and the first paragraph of section 10.42.

A.4 Graphemes

The phoneme inventories of languages are mainly established by finding ‘minimal pairs’, spoken words which differ in only one sound segment but have different meanings; see again bad/pad above, and for another example (the few pairs of English words differing only in /θ/ v. /ð/ such as wreath/wreathe) see section 9.36. Some linguists try to establish the graphemes of an orthography similarly, that is by classifying all the letter shapes which differentiate written words with different meanings. For English, this would result in an inventory of about 50 graphemes – the upper- and lower-case versions of the 26 letters of the alphabet, plus 2 for the variant forms of lower-case <a, ɑ> and <g, g> (unless those were called ‘allographs’ by analogy with the allophone variants of phonemes), possibly minus a few for letters with graphically similar upper- and lower-case forms <C, c; K, k; O, o; P, p; S, s; U, u; V, v; W, w; X, x; Y, y; Z, z>, possibly plus some for ‘ligatures’ (joined letters) which used to be used in print (e.g. <æ> in words like ægis, Cæsar), and possibly plus a few for common abbreviations and punctuation marks <& ! , . @ ? : ; …> - but where would you stop? For example, are numerals graphemes? Also, this approach would signally fail to uncover any multi-letter graphemes, and by extension the feature I have labelled ‘dual-functioning’, both of which seem to me absolutely necessary in analysing English spelling.

I have instead taken the (to me) more common-sense approach of asking which letters and letter-combinations represent which phonemes (chapters 3 and 5), and then using the inventory of graphemes so established (chapter 8) to work back to phonemes (chapters 9 and 10).

A.5 Every letter belongs to a grapheme (almost)

It is commonly believed that English spelling has lots of ‘silent letters’, ‘magic <e>’ being the classic example, along with the first letter in word-initial clusters such as <kn, wr>. Well yes, but every alphabetic script is composed entirely of silent letters, if you think about it. What is meant is letters which might as well not be there, since the spelling would represent the same word-sound without them, e.g.

write

which could be

rite (and is, in another meaning)

honest

which could be

*onest

friend

which could be

*frend

beauty

which could be

*buty

or letters which at their position in the written form of a word do not correspond to a phoneme at that position in the spoken form but may nevertheless affect its pronunciation, e.g. ‘magic <e>’.

In my view the identification of silent letters is more a matter for spelling reformers than for teachers. Learners have to learn the current spellings, and have to develop ways of remembering non-obvious parts of the system - of which there are many more besides ‘silent letters’ (e.g. whether medial and linking /w/ and /j/ are represented in the spelling or not – see sections 3.8.7-8 and 9.0). It may be more helpful to learners to be asked ‘How do we write /r/ at the beginning of writing?’ and told ‘<wr> at the beginning of a word is pronounced /r/’ than to be told ‘The <w> at the beginning of writing is silent.’

In accordance with this idea I have adopted a ‘principle of exhaustiveness’ (first proposed by Albrow, 1972, and adopted by Carney, 1994): that is, as far as possible every letter in a word’s spelling should be allocated to one of the phonemes in its spoken form. So you will find that I have analysed <wr> as one of the spellings of /r/, <ho> as one of the spellings of /ɒ/, <ie> as one of the spellings of /e/, <eau> as one of the spellings of the 2-phoneme sequence /juː/, <ps> as one of the spellings of /s/ (as in psychology), etc., etc.

On the whole, this works well. However, in section 6.10 you will find a whole set of elided vowels, cases where a vowel letter never corresponds to a phoneme, even in citation forms (or only in very conservative or artificial ‘spelling guidance’ pronunciations), and where I consider it would be over the top to add sequences consisting of those vowel letters and the preceding consonant letters to the inventory of graphemes. This does lead to a fuzzy boundary on the category of graphemes, but it seems to me that complete consistency is unattainable here.

The impossibility of completeness is particularly visible in the case of odd spellings of place- and personal names. For example, most of the letters in Leicester, Worcester can be assigned to phonemes in their spoken forms /ʽlestǝ, ʽwʊstǝ/ (as can all the letters in the alternative spellings Lester, Wooster), except <ce>: if the principle of exhaustiveness is to be maintained, these letters would, I think, have to be combined with the <s> as a new grapheme corresponding to /s/ - but there is no warrant for a grapheme <ces> in the main vocabulary, so I have not added it to the inventory or used Leicester, Worcester as examples of /e/ spelt <ei> or /ʊ/ spelt <or> respectively, even though these are in the inventory. And if you believe in Cholmondeley-Featherstonhaugh as a genuine spelling of a double-barrelled surname pronounced /ʽʧʌmliːˈfænʃɔː/ you’d have to add /ʌ/ spelt <ol>, and puzzle over how to divide <onde> between /m, l/ with no principled way of allocating any of the letters to either phoneme. Worst of all, between the initial /f/ and final /ɔː/ of /ˈfænʃɔː/ I can see no way of getting the <s, n> to be part of the spellings of /n, ʃ/, which are in the opposite order to the letters.

A.6 Split digraphs

The classic case of a so-called ‘silent letter’ which does influence the pronunciations of words and cannot be removed without altering them is the ‘magic’ <e> in split digraphs – but how should split digraphs be defined? A first and superficially appealing definition would be ‘Cases of <a, e, i, o, u, y> followed by a consonant letter and stem-final <e> where the <e> indicates the letter-name (‘long’) pronunciation of the vowel letter’. This will not work, because:

1) the pronunciation of <y.e> is /aɪ/, which is the name of <i> (and not /waɪ/, the name of <y>)

2) <u.e> is pronounced not only as <juː> but also as <uː>

3) all the other split digraphs except <y.e> also have pronunciations other than the ‘letter-name’ one.

For my separate definitions of the six split digraphs I recognise see sections 10.4/17/24/28/38/40. Here I attempt to generalise them in this formulation:

A split digraph consists of stem-final <e> preceded by (usually) a single consonant letter (other than <h, j, q, r, w, x, y>) preceded by one of <a, e, i, o, u, y> where that letter is not preceded by a vowel letter and where the digraph is pronounced either as the name of the first letter of the digraph or as another long vowel or diphthong.

The last phrase covers the pronunciation of <y.e> and the /uː/ pronunciation of <u.e> as well as non-letter-name pronunciations of the other split digraphs.

The exclusion of <h, j, q> from occupying the ‘dot’ position in a split digraph is mentioned solely for completeness (and only here, and not in any of the relevant sections of chapter 10): there are not, and cannot be, any such sequences as <ahe, eje, iqe>, etc., in stem-final position. The exclusion of <r, w, x, y> from occupying the ‘dot’ position keeps out <are, ere, ire, ore, ure, yre; awe, ewe, owe; aye, eye> which need to be analysed as trigraphs to account for their correspondences, and <axe, exe> which need to be analysed as having <xe> as a digraph separate from the preceding vowel letter. (Other combinations, e.g. <iwe, oxe, uye>, do not occur).

Letter <g> as sole occupant of the ‘dot’ position is odd. There are no words ending <yge>, and very few ending <ege, ige, oge, uge> - see sections 10.17/24/28/38 – but there are hundreds ending <age>, many of which have neither of the split digraph pronunciations – see below, and see the entry for <a.e>, section 10.4, for the three competing pronunciations.

The restriction to one intervening consonant letter has to be relaxed to allow for words where there is clearly a split digraph according to the rest of the definition but there are two consonant letters or <gu, qu> forming a consonant digraph intervening. This extension covers varying numbers of words under <a.e, i.e, o.e, u.e, y.e> (and none under <e.e>), totalling about 64 in all. The full list of consonant digraphs which can occupy the dot position is <ch, gn, gu, ll, mb, qu, ss, th, tt>, and most words containing them are unusual. There are also just 13 stem words with <n, g> or <s, t> spelling separate phonemes intervening in <a.e> pronounced /eɪ/ which seem to me to fit the definition and which I have decided to include: arrange, change, grange, mange, range, strange (plus the derived forms estrange, exchange); baste, chaste, haste, lambaste, paste, taste, waste, plus four oddities: caste with <a.e> pronounced /ɑː/ surrounding <st>, and three words with <squ> occupying the dot position: bisque, odalisque with surrounding <i.e> pronounced /iː/, and brusque pronounced /bruːsk/ (also pronounced /brʌsk/, which requires an analysis not involving a split digraph) with surrounding <u.e> pronounced /uː/. The dot position cannot be occupied by any other multi-letter sequence, in my analysis.

The stipulation that the leading letter in a split digraph must not be preceded by a vowel letter is needed to rule out vowel digraphs, etc., which do not need the final <e> to indicate their pronunciations. This differentiates my analysis from that of Mountford (1998), who recognises the following 12 ‘split trigraphs’ with two vowel letters preceding the dot: <ai.e, au.e, ea.e, ee.e, ei.e, eu.e, ia.e, ie.e, oi.e, oo.e, ou.e, ui.e>, plus 10 more ‘split trigraphs’ with a consonant letter (counting not only <l. r> but also <w, y> as consonant letters) immediately preceding the dot: <al.e, ar.e, aw.e, er.e, ir.e, is.e, or.e, ow.e, oy.e, ur.e>, and even the following five ‘split four-letter graphemes’, all with two vowel letters and then a consonant letter preceding the dot: <ais.e, ear.e, ier.e, oar.e, our.e>. Some of these extended split graphemes were posited as far back as Cordts (1965). I have found none of them necessary in my analysis because all such cases yield instead to analyses with the letters before the dot forming graphemes in their own right, and the final <e> forming a di/trigraph with the preceding consonant letter(s).

The letter-name pronunciations of <a.e, e.e, i.e, o.e, u.e> as /eɪ, iː, aɪ, əʊ, juː/, plus <u.e, y.e> as /uː, aɪ/, then represent the obvious pronunciations of the split digraphs.

However, under <a.e, e.e, i.e, o.e> I have added /ɑː, eɪ, iː, uː/ respectively as alternative pronunciations to cover large numbers of words with <a.e> pronounced /ɑː/, just four stem words with <e.e> pronounced /eɪ/, a moderately large set of words with <i.e> pronounced /iː/, and just six with <o.e> pronounced /uː/. With these extensions, five of the six split digraphs have two pronunciations each, the exception being <y.e>, which is only pronounced /aɪ/. It seems to me that these extra correspondences of the split digraphs need to be analysed in this way and not, for example, as <a> in massage, etc., spelling /ɑː/ and the <e> only forming a digraph with <g>.

It is noticeable that for <a.e, e.e, i.e> these extra correspondences derive from French spelling conventions. As Crystal (2012) and especially Upward and Davidson (2011) document, French words which entered the English language before about 1600 became anglicised in pronunciation and followed English spelling conventions, such as they were (there had been some inconsistencies in Anglo-Saxon (Old English) spelling – in particular it had no consistent system for distinguishing long and short vowels – and after 1066 Norman French scribes introduced others). But what I have in several places called ‘more recent’ borrowings of French words (that is, those which arrived after the completion of the Great Vowel Shift in about 1600) in almost all cases retained their French spellings and (approximations to) their French pronunciations, despite the fact that this has introduced new correspondences for some vowel graphemes and increased the number of inconsistencies.

One of the very few words introduced after 1600 which did acquire an anglicised pronunciation is blouse: if it had behaved like other ‘more recent’ borrowings it would be pronounced closer to French as /bluːz/ rather than as anglicised /blaʊz/. The process of assimilation can be heard at work in garage, mauve, doyen(ne) and foyer:

  • The General American pronunciation of garage as /gəˈraːʒ/ is closest to French, having only replaced /a/ with /ə/ in the first syllable and /ʁ/ with /r/ in the second. In RP, there are two pronunciations. In /ˈgæraːʒ/, the last two French phonemes have been retained, but the stress has shifted to the first syllable and the vowel in that syllable has shifted to /æ/. In /ˈgærɪʤ/, anglicisation is complete: the second syllable is now pronounced as in the great majority of polysyllabic words ending <-age>, and the only phoneme which is still pronounced as in French is the initial /g/;
  • In Britain, the pronunciation of mauve varies between French-like /məʊv/, with the English diphthong /əʊ/ replacing /o/, and /mɔːv/, with a fully anglicised vowel perhaps reflecting a ‘spelling pronunciation’ of <au>; the latter pronunciation is less usual;
  • Both doyen and doyenne have a French-like pronunciation /dwaɪˈjen/ and a mid-way pronunciation /dɔɪˈjen/ where the French 2-phoneme sequence /waɪ/ has shifted to /ɔɪ/ but the stress has remained on the second syllable; doyen (but not doyenne) also has the fully anglicised pronunciation /ˈdɔɪjən/ where the stress has moved to the first syllable;
  • Similarly, foyer has three pronunciations: /ˈfwaɪjeɪ/ with the French pure vowels /ɑ, e/ anglicised to diphthongs /aɪ, eɪ/ and the stress shifted to the first syllable, /ˈfɔɪjeɪ/ (which I first noticed in an RP speaker in November 2013) with the first diphthong further anglicised to /ɔɪ/, and /ˈfɔɪjə/ with the final vowel totally anglicised to schwa.

In large numbers of cases where the consonant in the dot position is <c, g>, and a small number where it is <v>, the <e> also forms a digraph with the consonant letter (for dual-functioning see section 7.1) spelling (and pronounced) /s, ʤ (or /ʒ/), v/ respectively. Traditionally, the <e> is said to ‘mark’ <c, g> as having their ‘soft’ pronunciations /s, ʤ/ (see also section A.8), and not their ‘hard’ pronunciations /k, g/. The alternative ‘soft’ pronunciation of <g> as /ʒ/ never seems to be noticed, but this is the least frequent phoneme in spoken English and its spellings are rarely taught explicitly. <ve> is different: the <v> would spell and be pronounced /v/ without the <e>, which is present purely as a strong spelling convention (see section 3.7.7) even when the <e> is not part of a split digraph.

When is a split digraph not a split digraph? Note the restriction to letter-names and alternative long vowels or diphthongs in my definition. There are also copious examples of words with final <e> preceded by a single consonant letter (other than <r, w, x, y>) preceded by <a, e, i, o, u> not preceded by another vowel letter where <a, e, i, o, u> have neither their letter-name pronunciations nor the alternative pronunciations listed above. For categories and lists see the exceptions mentioned under <a.e, e.e, i.e, o.e, u.e> in sections 10.4/17/24/28/38. (There are no such cases with <y.e>).

In general, these are words where what appear to be the ‘leading’ vowel letters in split digraphs are pronounced ‘short’. In most cases the ‘leading’ vowel letter is pronounced /ə/ or /ɪ/, for example in mortgage (and lots of other words ending in unstressed <-age>), purchase, accurate (and lots of other words ending in <-ate>), college, diocese, bodice (and several other words ending in <-ice>), engine (and several other words ending in <-ine>), mortise, practise, premise, promise, treatise, definite (and several other words ending in <-ite>), give, massive (and thousands of other polysyllabic nouns and adjectives ending in <-ive>), fulsome, handsome (and all the other adjectives ending in <-some>), welcome, purpose, lettuce, minute (/ˈmɪnɪt/, ‘60 seconds’). Examples with other short vowel phonemes are axe, have (when stressed) with /æ/, allege, annexe, clientele, cortege with <e>, compote, (be)gone, scone, shone with /ɒ/, and above, become, come, deluxe, done, dove, glove, love, none, shove, some with /ʌ/. It seems to me that all such cases, unlike those involving long vowels, diphthongs or /juː/, are more economically analysed as having the relevant short vowel spelt variously <a, e, i, o, u> and the word-final <e> forming a digraph with the intervening consonant letter. All of this admittedly produces a fuzzy boundary around split digraphs, and a possible source of confusion over some words which might have a split digraph pronunciation, or not, but it seems to me that complete consistency is not attainable in this area either.

For an attempt at a pedagogical statement of the split digraph rule see section 11.4.

A.7 Rhymes and phonograms (and rimes)

To keep this section simple to start with, let us define rhymes as those endings of one-syllable words which sound the same in more than one word, e.g. the /uːt/ sounds of boot, coot, hoot, toot, etc. Phonograms are the corresponding parts of written words, in this case <oot>; alternative terms for ‘phonogram’ are ‘rime’ (in that spelling) and ‘(word) body’. I do not use ‘body’ because it has more usual meanings in the language, and do not use ‘rime’ in this section because it is confusing to use homophonic terms for the corresponding parts of spoken and written words (and because there is already a word ‘rime’ meaning ‘hoar frost’).

There are claims (apparently first made by Adams, 1990: 85, but first systematically investigated by Treiman et al., 1995) in the literature on teaching children to read and spell that many of the alternative spellings of vowel phonemes and the alternative pronunciations of vowel graphemes, considered to be unpredictable in isolation, are more predictable if phonograms and rhymes are considered as units. The claim was originally confined to monosyllables with CVC phonological structure, which led to monosyllables with no initial consonant, and therefore VC structure, and relevant polysyllables, being mostly overlooked. Although my analysis is focused almost entirely at the segmental (phoneme and grapheme) level, I have examined this claim (in its original version plus VCs and some polysyllables), and found it largely unconvincing.

In the spelling direction, it seems to me that there are no rhyme-phonogram correspondences in (C)VC monosyllables which would be worth teaching as units because all the ‘families’ of words are too small and/or have too many exceptions. However, in (C)VCC monosyllables there are just two that might be worth teaching:

/eɪnt/ spelt <-aint> in faint, paint, plaint, quaint, saint, taint (only exceptions: ain’t (sort of), feint). Despite applying to only 6 monosyllables (against 2 exceptions), this is probably worth teaching because it generalises to the final syllables of 6 polysyllables: acquaint, attaint, complaint, con/di/re-straint and (despite possibly being split across syllable boundaries) to non-final syllables in at least 3 more: maintain, plaintiff, plaintive. Score: 6-2 in monosyllables, 15-2 overall.

/əʊld/ spelt <-old> in bold, cold, fold, gold, hold, old, scold, sold, told, wold, plus the final syllables in behold, cuckold, blind/mani-fold, marigold, scaffold, threshold. The only stem word exception is mould (and this is spelt mold in the US), but there is possible confusion with several past tenses/participles, some of which are homophones of the stem words: bowled, doled, foaled, holed, poled, polled, rolled, soled, strolled, tolled. Taking in non-final syllables of polysyllables appears to add just one example, solder, and one exception, shoulder (soldier does not qualify as either because here <old> spells /əʊlʤ/). Score: in monosyllables, 10-10 in UK, 11-9 in US; overall, 18-11 in UK, 19-10 in US. Despite the poor score in monosyllables, this would probably be worth teaching when children are clear about spelling regular past tenses and participles with <-ed>.

For a clear example of a phonogram whose spelling is entirely predictable at the segmental level, and therefore not worth teaching as a unit, see the discussion of /iːz/ under /z/, section 3.7.8.

Conversely, for the final syllable /zəm/ which is always unstressed (and is therefore not a rhyme) and almost always spelt <-sm> (only exception: bosom), see section 3.5.4. To the latter might be added words ending /əʊs/ (see section 3.7.6). There are only four monosyllables with this rhyme: close (adjective/noun), dose, gross and the very rare surname Groce, but dozens of polysyllables (the ‘sugar’ words dextrose, glucose, lactose, sucrose (all of which have alternative pronunciations in /əʊz/), and the adjectives comatose, lachrymose, morose, verbose, etc.), the only exception without <-ose> being the verb engross. So ‘Word-final /əʊs/ is almost always spelt <-ose>’ is a reliable generalisation – but probably of very limited use to young children and their teachers.

In the reading aloud direction, it seems to me that there are just five phonogram-rhyme correspondences which would be worth teaching as units, three applicable mainly to (phonologically) (C)VC monosyllables and two to (C)VCC monosyllables:

<-all> pronounced /ɔːl/ in all, ball, call, fall, gall, hall, pall, small, squall, stall, tall, thrall, wall. The only words which are exceptions in RP are mall, shall, with /æl/ - and mall is /mɔːl/ in General American. These final syllables would have to be clearly distinguished from non-final syllables with <all>, e.g. alliance. Score: 13-2 in RP, 14-1 in General American.

<-ead> pronounced /ed/ in bread, dead, dread, head, lead (the metal), read (past tense and participle), spread, stead, thread, tread (exceptions: bead, knead, lead (verb), mead, plead, read (present tense), with /iːd/). This pattern generalises to breadth if we are feeling generous, and to the final syllables of two polysyllables: ahead, instead. If we are feeling even more generous we can add some polysyllables with non-final <ead>, of which some but not all are derived forms of the relevant monosyllables: already, meadow, Reading, ready, steadfast, steady, treadle (contrast reading, beadle). Score: 11-6 in monosyllables, 20-8 overall.

<-ind> pronounced /aɪnd/ in bind, blind, find, grind, hind, kind, mind, rind, wind (‘turn’), plus one stem polysyllable, behind (exceptions: rescind, tamarind, wind (‘stiff breeze’) with /ɪnd/). These final syllables would have to be clearly distinguished from non-final syllables with <ind>, e.g. indicate. Score: 9-1 in monosyllables, 10-3 overall.

<-old> pronounced /əʊld/ in bold, cold, fold, gold, hold, old, scold, sold, told, wold, plus mold in US spelling and the polysyllables behold, cuckold, blind/mani-fold, marigold, scaffold, threshold and (non-finally) solder (only exception: soldier with /əʊlʤ/). Score: in monosyllables, 10-0 in RP, 11-0 in General American; overall, 18-1 in RP, 19-1 in General American.

<-ook> pronounced /ʊk/ in book, brook, cook, crook, hook, look, nook, rook, shook, took plus polysyllables Chinook, forsook (exceptions: gook, snook, spook, stook, gobbledegook, all with /uːk/). Score: 10-4 in monosyllables, 12-5 overall.

This fairly meagre haul of reasonably reliable phonogram-rhyme correspondences would be worth adding to the similarly meagre haul of four reliable pronunciation rules for vowel graphemes analysed in chapter 11, but even together they fail to dent the importance of focusing almost all phonics teaching for reading on the segmental level. And phonics teaching for spelling should be even less influenced by the two possibly usable rhyme-phonogram correspondences listed above. For supporting evidence, see Solity and Vousden (2008: 490), who found that teaching onset-rimes would mean ‘there would have to be a fourfold increase in the amount of information children would need to learn to read material aimed at children, and an eightfold increase to move on to adult-directed text’.

A.8 Dual-functioning

In section 7.1 I deliberately sidestep an obvious question: why analyse any letters at all as belonging to more than one grapheme at the same time?

(Carney (1994: 37) was strongly opposed to ‘overlapping’ of graphemes, but did not analyse cases such as those I adduce here. However, he was rightly critical of Cordts (1965), who ‘for reasons not explained’ assigned some letters to two graphemes where this is not warranted, e.g. the <e> in cake to both <a.e>, which is essential, and to <ke> as a spelling of /k/, which is entirely unnecessary).

Well, if you do not assign some letters to more than one grapheme at the same time, the correspondences for some phonemes become even more complicated than they already are, and the results seem to me counter-intuitive. For example, it is clear that in care there are two graphemes <c, are> spelling the two phonemes /k, eə/. But how should derived forms such as caring be analysed? There are now five phonemes /k, eə, r, ɪ, ŋ/ of which /k, ɪ, ŋ/ are obviously spelt <c, i, ng>. The /r/ is also obviously spelt <r> - but does this mean that /eə/ is now spelt only <a>? If so, should this analysis be extended to independently-occurring medial examples such as parent? Here there are six phonemes /p, eə, r, ə, n, t/, of which four /p, ə, n, t/ are obviously spelt <p, e, n, t>, leaving <a, r> to spell /eə, r/. Both here and in caring we could analyse the <a> as spelling /eə/ and <r> as spelling /r/ and add /eə/ spelt <a> to the list of correspondences - but then in scarce, scarcity /eə/ can only be analysed as spelt <ar>, so /eə/ spelt <ar> has to stay in the inventory.

The problems just with /eə/ ramify when we look at words like air, aeroplane, pair, pairing, mayor, mayoress, sombrero, scherzo, heir, heiress: /eə/ spelt <air, ayor, er, heir> must be in the inventory to account for air, pair, mayor, scherzo, heir, but if we rule out dual-functioning we have to add /eə/ spelt <ae, ai, ayo, e, hei> to the list of correspondences (and delete only /eə/ spelt <aer>), and also add <ayo, hei> to the list of graphemes (and delete only <aer>) to account for aeroplane, pairing, mayoress, sombrero, heiress. Similar considerations apply to other phonemes spelt with graphemes ending in <e, r, w, y> where I posit dual-functioning. I therefore conclude that my analysis is actually conceptually neater, and keeps the lists of graphemes and correspondences from growing even more enormous.

Also, dual-functioning is an economical factor in English spelling in another sense – without it we would always have to spell various adjacent phonemes separately, and in many cases the system would have no well-motivated way of doing this. For example, bolero, bowie, buoyant, hero, jury, parent might have to be spelt *balairro, *boewie, *boiyant, *heerro, *joorry, *pairrant. And many other single-function spellings would probably be even stranger and more complicated.

My dual-functioning analysis solves a problem to which Venezky (1970: 53; in the following quotations I have edited Venezky’s symbols into those used here) says ‘no realistic solution is possible’ without adopting his proposal for a set of graphemes he designates as ‘markers’. These are, for example, the letter <e> in clothe and pace. He correctly notes that ‘in each word [the <e>] marks two separate patterns. In clothe [it] marks the correspondences <o>/əʊ/ and <th>/ð/; in pace it marks <a>/eɪ/ and <c>/s/,’ and goes on: ‘The traditionalist is faced with a dilemma here; are the units <o.e/a.e> or <the/ce>? Or shall we take a fine razor and split <e> into two parts so that both alternatives can be taken?’ This attempted reductio ad absurdum is proved meaningless once we accept that a letter can belong to two graphemes simultaneously.

A.9 Graphemes containing apostrophes

There are four of these in my analysis: <ey’re> spelling /eə/ in they’re, <e’er> spelling /eə/ as a contracted form of ever either independently or in (e.g.) ne’er, where’er, <e’re> spelling /ɪə/ in we’re, and <ou’re> spelling /ɔː/ in you’re. Although these are all contracted forms and not stem words I have included them because their pronunciations are distinctive, and unpredictable from the uncontracted forms.

Along the way I considered two other possible graphemes containing apostrophes for inclusion: <n’> spelling /ən/ as in isn’t, etc., and <’s> as the regular singular and irregular plural possessive and the contraction of is/has. I decided against <n’> because it seemed neater to consider /ən/ in these contractions as spelt solely by the <n> (see section 3.5.5).

<’s> was in drafts of the book for a very long time, with three correspondences: /s/ after voiceless non-sibilant consonants, /z/ after vowels and voiced non-sibilant consonants, and /ɪz/ after sibilant consonants. It was the last of these that originally led me to include <’s>, on the grounds that it seemed a neat way of accounting for the 2-phoneme sequence /ɪz/ in this context; including this correspondence logically meant bringing in the other two. But I eventually took <’s> out on much the same grounds as for excluding <n’> - it seemed neater to consider /ɪz/ as spelt solely by the <s> (see section 3.7.8). Omitting this correspondence logically implied omitting the other two.

A.10 Word stress

In the sections of chapter 10 devoted to the vowel letters <a, e, i, o, u> as single-letter graphemes I included /ə/ as one of their pronunciations, accompanied by the comment ‘regular in unstressed syllables’ or ‘regular when unstressed’, and the words ‘stressed’ and ‘unstressed’ occur in many other places in that chapter and this Appendix. In doing so, I evaded what might be considered a key issue in deriving the pronunciations of English words from their written forms: how can you tell which syllables are stressed and which are not?

The short answer is that this is a hugely complicated topic which could not fit in this book and would require a large, separate volume.

Some would say that that volume already exists, in the shape of Chomsky and Halle’s (1968) famous study The Sound Pattern of English. But, technically, their rules for assigning stress, including and especially their Main Stress Rule, operate not on the written forms of English words but on abstract or ‘deep’ versions of them which, it is essential for present purposes to note, are phonological. Thus the entire analysis is strictly speaking irrelevant to the question of how to predict stress from written forms. Moreover, Chomsky and Halle’s system is so complex that it is much too unwieldy for pedagogical purposes.

As far as I’ve been able to discover, there are just two authors who have tackled the question of how to predict word stress in English from written forms. Wijk (1966: 124) says:

Though it is not possible to lay down any completely satisfactory rules for the stressing of English words, it should be emphasised that there are vast numbers of words which do not offer any difficulty at all in this respect.

and then proceeds (pp.125ff.) to present five principal categories of polysyllables for which he says it is possible to formulate a rule – but most do not seem well worked out, all except those I have adopted and numbered (1) and (2) below have copious exceptions, and all the rest tacitly assume that it is obvious how many syllables written English words have.

Dickerson (1978) cites several authors who have said, in effect, that it is impossible to predict the stress patterns of English words – but they were all referring to trying to predict the stress pattern from the spoken form, that is, from the sequence of full and reduced (schwa) vowels. (Rule 45 in Clymer, 1963/1996 ‘When the last syllable is the sound r’ [assuming this means a General American retroflex (‘r-coloured’) version of /ə/] ‘it is unaccented [=unstressed]’ appears to be a confused statement of the obvious fact that in English /ə/ is almost never stressed, but is wrong to imply that this is true only in word-final position.) But if you already know the sequence of full and reduced vowels in English words you already know the stress pattern, and if you already know the stress pattern you already know the sequence of full and reduced vowels, so the argument is circular. Native speakers of English usually already know both, and do not need to be able to deduce the stress pattern from the written form, except perhaps when we encounter an unfamiliar word – and then we may well be in the same boat as foreign learners.

The serious implication here is that it can be difficult to work out from the spelling the pronunciation of a word you have never heard people say, especially if it is unusual, and easy to get some words wrong. For example, I have heard (or heard of) people pronouncing cotoneaster as /ˈkɒtəniːstə/ (‘cotton-easter’) rather than /kəˈtəʊniːjæstə/, machete as /məˈʧiːt/ (‘muh-cheat’) rather than /məˈʃetiː/, oesophagus as /əʊˈwiːsəʊfægəs/ (‘oh-wee-so-fag-us’) rather than /iːˈsɒfəgəs/, Yosemite as /ˈjəʊzəmaɪt/ (‘yoh-zuh-might’) rather than /jɒˈsemɪtiː/, and (possibly the most classic case) misled as /ʴmɪzǝld/ (like the verb mizzled).

There are two words spelt forestage: the obvious two-syllable one pronounced /ˈfɔːsteɪʤ/, ‘the part of a theatrical performance area between the curtain and the orchestra pit’, and with a morpheme boundary after fore-; and a three-syllable word from medieval English law with a morpheme boundary after forest-, pronounced /ˈfɒrɪstɪʤ/ and meaning either ‘a duty payable to the king’s foresters’ or ‘a service paid by foresters to the king’. Similarly, forestride, with a morpheme boundary after forest- and therefore more often spelt as two words, pronounced /ˈfɒrɪstraɪd/ with three syllables, was briefly the name of a bus service in the Reading area; but it could be misread (and I did so misread it) as having a morpheme boundary after fore- and the two-syllable pronunciation /ˈfɔːstraɪd/, and as perhaps meaning a specially determined way of walking (but there is no trace of such a word in the dictionaries). All of this again illustrates the need, which I stated right at the start of this book, to look up the pronunciation of whole words in a good pronouncing dictionary (that is, one which uses IPA symbols).

Dickerson took what appeared to be a novel approach to the problem of deducing how English polysyllables are stressed from their written form. Unfortunately, I found it impossible to adopt. This is because he assumes that non-native learners of English know what a syllable is:

To assign major stress to a word, only two syllables are relevant. One is called the Key Syllable, the other the Left Syllable, namely, the syllable immediately to the left of the Key.

(Dickerson, 1978: 138)

This also assumes that learners know where syllables in written English words begin and end and (as with Wijk) how many syllables there are in written English words. These are huge assumptions and, as I found when I tried to work them through, very awkward to specify in detail, mainly because they covertly assume that the learner already knows the spoken form of the word – which is precisely what Dickerson says he is trying not to assume.

For example, it is true that the great majority of two-syllable words in English are stressed on the first syllable (as pointed out by Clymer, 1963/1996, rule 30) – but how is a reader who does not already know (for example) the words blase, dais to deduce from their written forms that they have two syllables when all other words ending <-ase> or containing <ai> between two consonant letters are monosyllables? Even if that prediction were possible, trying to make a usable rule for stress on two-syllable words would still entail listing hundreds of such words which are instead stressed on the second syllable, including dozens of cases where verbs are stressed on the second syllable and identically-spelt nouns/adjectives are stressed on the first. Indeed, Hunnicutt (1976) showed that a computerised version of the Chomsky-Halle rules could not assign the correct stress to such pairs.

Another serious problem for any attempt to deduce the stress pattern of English words from their written forms arises from the existence of the elided vowels analysed in section 6.10 – it is hardly ever possible to deduce that a particular vowel letter in medial position (as opposed to stem-final ‘silent’ <e>) represents no phoneme at all and therefore isn’t even a candidate for taking the stress. Consider, for example, afferent (three syllables, first <e> pronounced /ə/) and different (two syllables, first <e> elided). And in some words a particular vowel may be elided or not, often depending on accent, for example migratory, pronounced either as /ˈmaɪgrətriː/ (three syllables, stress on first, <o> elided) or as /maɪˈgreɪtəriː/ (four syllables, stress on second, no elision) or as /maɪgrəˈtɔːriː/ (four syllables, stress on third, no elision).

So here I have merely stated a few useful rules for determining which syllable in polysyllables has main stress. The three most useful rules for predicting main stress are:

1) Virtually all words ending in <ea, eo, eou, eu, ia, io, iou, iu> followed by a single consonant letter or none and with at least one vowel letter earlier in the word have the stress on the syllable preceding <ea, eo, eou, eu, ia, io, iou, iu>, including all the hundreds of words ending <-tion> (as mentioned in rule 36 of Clymer, 1963/1996) and all those containing the five medial graphemes (other than <sh>) pronounced /ʃ/

2) Virtually all words ending in <-ience, -iency, -ient, -(s)sive>, have the stress on the preceding syllable

3) Virtually all words ending in <-ic(al)> have the stress on the preceding syllable. For exceptions see the last four paragraphs of section 10.22.

Otherwise, there are only rules covering small numbers of cases, such as:

4) Almost all words ending in <-ator> have the stress on the <a> if they have three syllables, e.g. creátor, curátor, dictátor, spectátor, otherwise on the syllable two before the <a>, e.g. admínistrator, ágitator, áviator, cálculator, cómmentator, ínsulator. Only exceptions: consérvator, conspírator, órator, prédator, sénator, which all have <a> pronounced /ə/ and stress on the syllable before that

5) The grapheme <air> is always stressed in polysyllables except in corsair (usually stressed on first syllable), millionairess (where the feminine ending <-ess> is usually stressed instead), mohair (always stressed on first syllable)

6) All words ending in <-eer, -esce, -esque, -ique> have stress on the final syllable

7) The grapheme <ier> is always stressed in polysyllables, except that frontier can be stressed on either syllable (but there are lots of words where <i, er> are separate graphemes)

8) All words ending in <-tte> have stress on the final syllable except etiquette, omelette, palette, which have stress on the first syllable

9) Almost all words ending in <-oon> have stress on the final syllable except forenoon, honeymoon, pantaloon, which have stress on the first syllable, and those ending in <-zoon> (‘living thing’), in which the ending has two syllables and is stressed on the first <o>.

There is also one helpful rule for where main stress does not fall:

10) The vowel letters as single-letter word-final graphemes in polysyllables are hardly ever stressed (Clymer, 1963/1996, rule 32 is a subset of this applying only to two-syllable words ending in a consonant letter followed by <y>). Strictly speaking this does not apply to word-final <e> as part of a split digraph, but that is of course never stressed either since there is no word-final vowel phoneme in such cases. There are very few exceptions, all of which are disyllables (and none at all with <i, u>): mama, papa; blase (which can also be stressed on the first syllable), manque, outre, risque; lasso; ally (if stressed on the second syllable, as the verb sometimes is), ap/com/im/re/sup-ply, defy, deny, descry, espy, July, rely. This rule implies that:

  • in all other two-syllable words with a single vowel letter as the word-final grapheme (that is, those with only one other vowel grapheme earlier in the word), the stress falls on that other vowel grapheme (= the first syllable)
  • a single-letter word-final vowel grapheme is never stressed in words of more than two syllables (except perhaps in unassimilated loanwords, e.g. Italian omertà), but this is of no help in predicting where the stress does fall in those longer words.

Similarly, Clymer’s (1963/1996) rule 35 ‘when ture is the final syllable in a word, it is unaccented’ (more accurately, ‘when <-ture> is word-final, it is unstressed’) is true, helpful if the word has only two syllables (e.g. picture), but useless for determining where the stress falls in longer words (e.g. furniture). And Clymer’s rule 31 ‘If a, in, re, ex, de, or be is the first syllable in a word, it is usually unaccented [=unstressed]’ poses huge problems. If it is meant to refer to prefixes, there is no way for anyone without deep etymological knowledge to tell when these word-beginnings are prefixes and when they are not. But if it is meant to refer to all words with these beginnings then it would be necessary to specify that each of them (except <ex-, in->) has to be followed by at least one consonant letter and all of them then by at least one vowel letter that is not ‘magic <e>’; and even then a quick scan of a dictionary reveals that there are far too many exceptions for the rule to be useful.

So, beyond the few definitely or possibly useful rules given above, the task of predicting word stress from the written forms of English words also awaits another study. That study would have to avoid the assumption, which I have knowingly perpetrated/perpetuated in section 10.42 and in the ‘rules’ I have stated above, that readers can tell from the spelling of English words how many syllables their spoken forms contain.

Appendix B: Pedagogically selected lists of phoneme-grapheme and grapheme-phoneme correspondences

© 2015 Greg Brooks, CC BY http://dx.doi.org/10.11647/OBP.0053.13

These lists are intended to be much more useful to teachers and to writers of early reading books than the full lists of correspondences in chapter 8. Similar tables (also largely devised by me) appeared in the Notes of Guidance to Letters and Sounds (DfES, 2007). As far as possible I have ensured that all words within the 1000 most frequent words in English whose correspond-ences are not covered by the major correspondences are listed in the right-hand columns of Tables B1-2 and in Tables B.6 and B.8. My source for the 1000 most frequent words was: http://en.wiktionary.org/w/index.php?title=Category:1000_English_basic_words&pagefrom=stamp#mw-pages [last accessed 20/8/2012].

For guidance on the phonetics underpinning the application of these lists in phonics teaching see Burton (2011), which also contains versions of these lists.

Table B.1: The phoneme-grapheme correspondences of British English spelling, by RP phoneme, 1: Consonants.

Phoneme

Grapheme(s)

As in …

Common words with rare graphemes for the phoneme

Basic

Other

/b/

b

bb

bed rabbit

<bu> build buy

/k/

c

ck k q ch

come back look queen Christmas

<cu> biscuit

/tʃ/

ch

tch

children match

<t> nature picture <ti> question

/d/

d

dd ed

dad teddy called

/f/

f

ff ph

from off elephant

<ft> often soften <gh> cough enough laugh rough tough

/g/

g

gg

get jogging

<gh> ghost <gu> guess guy

/h/

h

horse

<wh> who whole whose

/dʒ/

j

dg dge g ge

just budgie bridge giant orange

/m/

m

mm

my mummy

<mb> climb lamb thumb

<me> come some

<mn> autumn column

/n/

n

nn

now dinner

<gn> gnome sign

<kn> knife knock knot know

<ne> done engine none

/ŋ/

ng

n

sing sink

<ngue> tongue

/p/

p

pp

pen apple

<ph> shepherd

/r/

r

rr

red berry

<rh> rgyme rhythm

<wr> write wrong

/s/

s

c ce se ss

sit city once horse grass

<sc> science scissors <st> castle Christmas listen whistle

/ʃ/

sh

ti

ship station

<ch> machine <ci> special

<s> sugar sure

<ss> issue pressure tissue

<ssi> permission

/ʒ/

si

vision

<s> measure pleasure treasure usual

/t/

t

tt ed

but little looked

<pt> receipt <th> Thomas <tw> two

/θ/

th

thing

/ð/

th

that

<the> breathe

/v/

v

ve

very have

<bv> obvious <f> of

/w/

w

u

went queen

<wh> what when (etc.) wheel whistle white

/wʌ/ spelt <o> once one

/j/

y

yellow

<i> onion view

/z/

z

s se ze zz

zoo is please sneeze puzzle

<si> business <ss> scissors

Table B.2: The phoneme-grapheme correspondences of British English spelling, by RP phoneme, 2: Vowels.

Phoneme

Grapheme(s)

As in …

Common words with rare graphemes for the phoneme

Basic

Other

/æ/

a

and

/ə/

a

e er o

a the butter button

<ar> sugar <i> possible <our> colour favour honour <re> centre <ure> nature picture

/eɪ/

a.e

a ai ay

came bacon paint day

<aigh> straight <ea> break great steak <eigh> eight <ey> they

/eə/

air

are ar

fair fare parent

<ear> bear pear tear wear <ere> there where <eir> their

/ɑː/

ar

a

far ask

<al> half <are> are <au> aunt laugh <ear> heart

/e/

e

ea

went bread

<a> any many <ai> again(st) said <ay> says <ie> friend

/iː/

ee

e ea ey ie y

see he beach key field city

<e.e> these <eo> people <i.e> police

/ɪə/

eer

ear er ere

cheer hear hero here

<ier> fierce

/ɜː/

er

ir or ur

her girl word fur

<ear> early earth heard learn <ere> were <our> journey

/ɪ/

i

e y

is England gym

<a> language sausage <o> women <u> business minute

/aɪ/

i.e

i igh y

like I night my

<ei> either <eigh> height <eye> eye <ye> goodbye

/aɪe/ spelt <ir, ire, yre> biro fire wire tyre

/ɒ/

o

a

not was

<au> because sausage <ho> honest honour <ou> cough

/əʊ/

o

o.e ow

so bone blow

<oa> approach boat <oh> oh <ough> although

/ɔɪ/

oi

oy

boil boy

/ʊ/

oo

u

book put

<o> woman <oul> could should would

/uː/

oo

ew u u.e

too blew super rule

<o> do to two who <oe> shoe <o.e> lose move prove whose <ou> you <ough> through <ue> blue true <ui> fruit

/ʊə/

oor

ure

poor sure

<our> tour

/ɔː/

or

a ar au aw ore

for all warn sauce saw before

<augh> caught naughty <oar> board <oor> door floor <ough> bought brought fought ought thought <our> course four your

/aʊ/

ou

ow

out down

/aʊə/ spelt <hour> hour /aʊə/ spelt <our, ower> flour flower

/ʌ/

u

o

but some

<oo> blood flood <ou> country couple double encourage enough rough tough trouble young /wʌ/ spelt <o> once one

Table B.3: The phoneme-grapheme correspondences of British English spelling, 3: 2-phoneme sequences frequently spelt with single graphemes.

2-phoneme sequence

Grapheme(s)

As in …

2-grapheme spellings for same sequence

Basic

Other

/əl/ (only word-final)

le

little

animal label pencil carol beautiful

/juː/

u

eau ew ue u.e

union beauty few argue cute

view you

/ks/

x

box

banks tricks politics

N.B. The 2-phoneme sequence /kw/ is almost always spelt <qu> and should also be taught as a unit.

Table B.4: The grapheme-phoneme correspondences of British English spelling, 1: Single graphemes frequently pronounced as
2-phoneme sequences.

Grapheme(s)

2-phoneme sequence

Other phonemes

As in …

eau ew u ue u.e

/juː/

(too many to list)

beauty few union argue cute

le (only word-final)

/əl/

little

x

/ks/

box

N.B. The 2-grapheme sequence <qu> is almost always pronounced /kw/ and should be taught as a unit.

Table B.5: The grapheme-phoneme correspondences of British English spelling, 2: Major correspondences for consonant graphemes.

Grapheme(s)

Phoneme(s)

As in …

Basic

Other

b bb

/b/

bed rabbit

c

/k/

/s/

come city

ce

/s/

once

ch

/tʃ/

/k/

children Christmas

ck

/k/

back

d dd

/d/

dad teddy

dg(e)

/dʒ/

budgie bridge

ed

/d/

/t/

called looked

f ff

/f/

from off

g

/g/

/dʒ/

get giant

ge

/dʒ/

orange

gg

/g/

jogging

h

/h/

horse

j

/dʒ/

just

k

/k/

look

l ll

/l/

leg ball

n

/n/

/ŋ/

now sink

ng

/ŋ/

sing

nn

/n/

dinner

p pp

/p/

pen apple

ph

/f/

elephant

q

/k/

queen

r rr

/r/

red berry

s se

/s/

/z/

sit is horse please

sh

/ʃ/

ship

si

/ʒ/

vision

ss

/s/

grass

t tt

/t/

but little

tch

/tʃ/

match

th

/θ/

/ð/

thing that

ti

/ʃ/

/tʃ/

station question

u

/w/

queen

v ve

/v/

very have

w

/w/

went

y

/j/

yellow

z ze zz

/z/

zoo sneeze puzzle

Table B.6: The grapheme-phoneme correspondences of British English spelling, 3: Minor correspondences for consonant graphemes.

Grapheme(s)

Phoneme(s)

As in …

bu

/b/

build buy

bv

/v/

obvious

ch ci

/ʃ/

machine special

cu

/k/

biscuit

f

/v/

of

ft

/f/

often soften

gh

/f g/

cough enough laugh rough tough; ghost

gn

/n/

gnome sign

gu

/g/

guess guy

i

/j/

onion view

kn

/n/

knife knock knot know

mb me mn

/m/

climb lamb thumb; come some; autumn column

ne

/n/

done engine none

ngue

/ŋ/

tongue

o

/wʌ/

once one

ph

/p/

shepherd

pt

/t/

receipt

rh

/r/

rhyme rhythm

s ssi

/ʃ/

sugar sure; permission

s

/ʒ/

measure pleasure treasure usual

sc

/s/

science scissors

si

/z/

business

ss

/ʃ z/

issue pressure tissue; scissors

st

/s/

castle Christmas listen whistle

t

/tʃ/

nature picture

th tw

/t/

Thomas two

the

/ð/

breathe

wh

/h w/

who whole whose; what when (etc.) wheel whistle white

wr

/r/

write wrong

Table B.7: The grapheme-phoneme correspondences of British English spelling, 4: Major correspondences for vowel graphemes.

Grapheme(s)

Phoneme(s)

As in …

Basic

Other

a

/æ/

/eɪ ɑː ɒ ɔː ə/

and bacon ask was all about

a.e ai ay

/eɪ/

came paint day

air are

/eə/

fair fare

ar

/ɑː/

/eə ɔː/

far parent warn

au aw

/ɔː/

sauce saw

e

/e/

/iː ɪ ə/

went he England the

ea

/iː/

/e/

beach bread

ear eer ere

/ɪə/

hear cheer here

ee ey

/iː/

see key

er

/ɜː/

/ɪə ə/

her hero butter

ew

/uː/

blew

i

/ɪ/

/aɪ/

is I

ie

/iː/

field

i.e igh

/aɪ/

like night

ir

/ɜː/

girl

o

/ɒ/

/ʌ əʊ ə/

not some so button

o.e

/əʊ/

bone

oi oy

/ɔɪ/

boil boy

oo

/uː/

/ʊ/

too book

oor

/ɔː/

/ʊə/

door poor

or

/ɔː/

/ɜː/

for worm

ore

/ɔː/

before

ou

/aʊ/

out

ow

/aʊ/

/əʊ/

down blow

u

/ʌ/

/ʊ uː/

but put super

u.e

/uː/

rule

ur

/ɜː/

fur

y

/aɪ/

/ɪ iː/

my gym city

Table B.8: The grapheme-phoneme correspondences of British English spelling, 5: Minor correspondences for vowel graphemes.

Grapheme(s)

Phoneme(s)

As in …

a

/e ɪ/

any many; language sausage

ai ay

/e/

again(st) said says

aigh

/eɪ/

straight

al are

/ɑː/

half; are

ar

/ə/

sugar

au

/ɑː ɒ/

aunt laugh; because sausage

augh

/ɔː/

caught naughty

ea ey

/eɪ/

break great steak; they

ear

/eə ɑː ɜː/

bear pear tear wear; heart; early earth heard learn

e.e eo

/iː/

these; people

ei

/aɪ/

either

eigh

/eɪ aɪ/

eight; height

eir

/eə/

their

ere

/eə ɜː/

there where; were

eye

/aɪ/

eye

ho

/ɒ/

honest honour

hour

/aʊə/

hour

i

/ə/

possible

ie

/e/

friend

i.e

/iː/

police

ier

/ɪə/

fierce

ir ire

/aɪə/

biro fire wire

o

/ʊ ɪ wʌ/

woman; women; once one

o oe o.e

/uː/

do to two who; shoe; lose move prove whose

oa oh

/əʊ/

approach boat oh

oar

/ɔː/

board

oo

/ʌ/

blood flood

oor

/ɔː/

door floor

ou

/ɒ uː ʌ/

cough; you; country couple double encourage enough rough tough trouble young

ough

/əʊ uː ɔː/

although; through; bought brought fought ought thought

oul

/ʊ/

could should would

our

/ə ɜː ʊə ɔː aʊə/

colour favour honour; journey; tour; course four your; flour

ower

/aʊə/

flower

re

/ə/

centre

u

/ɪ/

business minute

ue ui

/uː/

blue true fruit

ure

/ə/

nature picture

ye

/aɪ/

goodbye

yre

/aɪə/

tyre