BRITISH WORKER: I can’t work today. I’ve got diarrhoea.
AMERICAN BOSS: Diarrhea? That’s dreadful. You could have sent me a sick note.
BRITISH WORKER: But I can’t spell it.
Not many people can spell this particular word either in the British or in the somewhat simplified American fashion. Yet if you know how an English word is pronounced and roughly what it means, you ought to be able to write it down without much trouble. If you find that you can’t do that, then the writing system may well seem to be at fault.
However, ‘catastrophic’ is a severe word. Before rushing to condemn the whole system, you ought to see what English spelling sets out to do and the extent to which it is consistent in doing it.
In looking at spelling we need to keep sounds and letters quite separate. So letters are cited in angled brackets and symbols for speech-sounds are put between slant lines. The letters <said> spell /sed/ and <text> spells /tekst/.
People often foul things up when talking about spelling because they do not differentiate between letters and sounds. It is best to use ‘vowel’ and ‘consonant’ only for sounds. The words fight, cry, shop, axe, coin, caught each contain two consonants and one vowel.
If you want to talk about letters then say ‘vowel letter’ and ‘consonant letter’. Unusual re-spellings or mistakes are marked with an asterisk: *<stoopid>, *<langwidge>.
The myth that there are ‘five vowels’ in English refers to the vowel letters <a, e, i, o, u> of the roman alphabet. Depending on your accent, you will find about twenty vowels in English. Try collecting them by changing the vowel sound in a series of otherwise identical-sounding words, noting each new vowel that turns up. If you start with, say, lick, you can change the vowel sound and get a different word lack. Without bothering about the spelling, carry on ringing the changes of sound and you will turn up lock, luck, leak, lake, like. You will find the vowel of leak again in feel, but that frame will give you some new ones, as in fell, full, foil, fowl, foal, fall, and so on in other frames. Some twenty-four consonant sounds can be found by the same method: bet, debt, get, jet, let, met, net, pet, set, vet, wet, yet. A different frame will give you some new ones in sheaf, sheath, sheathe, sheet, and even chic (or sheik). Some less common consonants will be quite hard to find, such as the middle one in measure.
In a Garden-of-Eden alphabetic writing system, you would have a single letter for each speech-sound and one speech-sound for each single letter. Large numbers of English words appear to follow this strict pattern: best, dispel, dividend, film, frog, help, jam, limit, map, profit, rob, splint, tendril, win, yet, etc. But if we look further, we soon see that too many speech-sounds are chasing too few letters.
Most consonants, at least some of the time, may have a single-letter ‘alphabetic’ spelling: <b, d, f, g, h, j, 1, m, n, p, r, s, t, v, w, y, z>; /k/ has a choice of <c> or <k>. But there is often ‘divergence’, where one speech-sound has several different spellings and one spelling may stand for different speech-sounds. In spite of the available single-letter spelling <f>, the consonant at the beginning of foot has more complex spellings in physics, enough, offer. The <s> in easy represents /z/, the <u> in quick represents /w/ and the <f> in of represents /v/. The consonant at the beginning of yet, yellow also comes pre-packed as part of the vowel spelt <u(e)> in cue, cute, pure. The most divergent consonant is /k/, which has different spellings in catch, chemist, back, acclaim, chukker, key, quay, quite, Iraq and as part of the /ks/ in axe.
Six consonants do not have a single-letter spelling of their own and require at least two letters, such as <sh> or <ch>. These are the consonants found in the middle of the following words: method, bother, wishing, measure, patches and the consonant represented by <ng> in singer when no actual /g/is pronounced.
Five pairs of vowels can have single-letter spellings: <a> in scrap, scraping, <e> in met, meter, <i> in pip, piper, <o> in cop, coping, <u> in rub, ruby. There is also <y> in cryptic, cry, which duplicates the <i> spellings. The examples given in each pair represent a ‘short’ and a ‘long’ vowel.
For this letter-sharing to work, ‘markers’ are needed in some contexts to tell you which value the letter has. To get the long value of <a> in a single-syllable word, you have to add a marker <-e>, as in scrape. To get the short value before a suffix beginning with a vowel like <-ing>, you double a final consonant letter, as in scrapping. So, with this marking, a single vowel letter can be used with two values in scrap, scrape, scrapping, scraping.
Four consonants have unusual doubling. The normal doubling of /k/ in native words is <ck>: stoking, stocking; baker, backer. The consonant at the end of beach and batch has <ch> as its ‘single’ spelling and <tch> as its ‘doubled’ spelling. Similarly we have <g(e)> as a ‘single’ spelling and <dg(e)> as a ‘doubled’ spelling in cadge, cage. A doubled <vv> occurs only in slang words: navvy, luvvies, revving. So the vowel difference in words such as level, seven, devil and fever, even, evil is not marked.
Some words are made up of several recognizable building blocks. The word reason is a single unit, while un+reason+able+ness consists of four. The son bit is not such a unit. English spelling often tries to give each of these building blocks a constant spelling. A good example is the verbal ending <-ed>. This sounds quite different in wished, begged, and wanted. If you think that they would be better spelt phonetically as *<wisht>, *<begd>, you are losing the advantage of a constant spelling for the regular past-tense ending.
You may think it awkward to have /s/ spelt differently in sent and cent. That may be, but the <c> spelling of both /k/ in electric and /s/ in electricity keeps the spelling of that unit constant.
The best examples of this principle are the long and short spellings of single vowel letters seen in word pairs such as:
| atrocious – atrocity | female – feminine | omen – ominous |
| austere – austerity | grateful – gratitude | reside – residual |
| chaste – chastity | legal – legislate | sole – solitude |
| crime – criminal | mine – mineral | supreme – supremacy |
In these pairs the basic long vowel is shortened when it comes three syllables from the end of the word. These shortened vowels do not require marking with a double consonant letter as *<omminous>, *<minneral> or *<sollitude>.
Keeping a constant spelling may involve the use of so-called ‘silent’ letters. The <g> does not represent /g/ in sign, but it does in derived forms resignation, signal, signature, signify. Similarly we have malign, malignant. Changing to *<sine>, *<maline> would spoil the visual link. Should we keep the <w> of two because twenty, twin, between are remotely related? Should shepherd be re-spelt as *<sheppard>, a regularized spelling when used as a name?
On the other hand the <g> of gnarled, gnat, gnash, gnaw, gnome and the <k> of knee, knife, knight, knock, know, knuckle are quite empty letters. They are the debris of history and are never pronounced in any derived word (except for acknowledge). It would be no loss to change to *<narled>, *<nat>, *<nife>, *<nuckle>, etc.
There are other important markers. The <-e> in bathe, breathe, loathe, wreathe not only marks the vowel as long but marks the last consonant as ‘voiced’ rather than the ‘voiceless’ one in bath, breath, loath, wreath. Other examples are lathe, lithe, swathe. Mouth and smooth used as verbs lack this marking.
The marker <-e> in browse, copse, lapse, please, tease, tense is used to prevent confusion with the plural forms brows, cops, laps, pleas, teas, tens. It marks the browse group as single units and as such is called ‘lexical <-e>’.
Some marking is needed to sort out the two distinct consonants represented by <g>. Before <a, o, u> we have /g/, as in gap, got, gum and the consonant spelt <j> in jam before <i, e> in gin, gem. The problem is that there are some exceptions with /g/ before <i, e>: gear, geese, get, giddy, gild, gilt, gimmick, girl, give. Some words however have used the letter <u> as a marker for /g/ in guess, guest, guide, guild, guilt, guise, guitar. Its use is not very consistent, since guard, guarantee do not need any <u> marker (cf. garden).
Words spelt the same but pronounced differently are called homographs: <minute> may be an adjective (‘a really minute insect’) or a noun (‘half a minute’). A minute steak has to be interpreted by the reader: either a very small steak or one cooked for a minute.
Words pronounced the same but spelt differently are called homophones: <vain>, <vane>, <vein>, or <foul>, <fowl>, or <meat>, <meet>, <mete>. These variant vowel spellings clearly make it harder for the writer, but it is often claimed that such divergence is not always a bad thing for the reader, since different words should look different on the printed page.
Even so, a good number of words are both homographs and homophones: sounding the same and looking the same. These are sometimes called homonyms. For instance, hamper represents two completely different unrelated words: either ‘a basket’ or ‘to hinder’. Quarry means either ‘a stone quarry’ or ‘a hunted animal’. You will find two or more very different words sharing each of the following forms: bark, bellows, bound, cricket, fine, firm, fit, flat, hail, last, leaves, pants, plane, quail, rest, rose, stable. Check in a good dictionary if in doubt. The intended sense is usually obvious from the context. If spelling reform reduced divergence, it would clearly add to the number of such homonyms: ‘a weather vane’, ‘a *vane in the wrist’ and Very *vane’.
A vowel may be weakened by lack of stress. Unless you are speaking very formally the highlighted vowel letters in the following words all spell the same indistinct /ə/ sound: about, asparagus, author, caravan, courageous, driver, polite, together. This obscure vowel has borrowed the name ‘schwa’ from Hebrew. As you can see, the spelling of /ə/ varies widely, since it reflects what the vowel would be in a stressed context. You find /ə/ equally in organ, political, president, but the spelling is prompted by the stressed vowels of organic, politics, presidential. The spelling of the basic units is constant.
Words borrowed from French have sometimes been altered by anxious academics looking beyond the French spelling to the distant Latin original. The words debt, doubt, were medieval borrowings of French dette ‘debt’, doute ‘doubt’ without a <b>. The ‘silent’ <b> was inserted in the sixteenth century to resemble the original Latin debitum, dubit-are, and to draw attention to the shared meaning of related English words derived from the same roots, such as debit, dubitative.
Before the eighteenth century, subtle was generally spelt <suttle>, just like regular scuttle, even by authors such as Milton, a well-known Latin scholar. In spite of this the present spelling with an empty <b> was adopted to match Latin subtilis, though the <b> has never been pronounced in English.
Such interference is often inconsistent. The <p> of receipt links it to receptacle, reception, but deceit lacks a <p>, in spite of deception. The <c> spelling of the early French loan grocer is a regular English spelling (racer, slicer), so why not have gross spelt *<groce> on the lines of race, truce, slice?. As it is, gross is the only English word in which <oss> does not sound as it does in boss, cross, doss, dross, floss. Ironically, the regular *<groce> was a common medieval spelling that did not survive.
A similar mismatch is the French <gn> spelling of align, alignment. The base form in English is <line> and not, as in French, *<ligne>. However, some dictionaries do allow the common-sense spelling <aline>, <alinement>. The <b> of crumb, crumby is only pronounced in crumble. Interestingly, when used as slang, crummy will have a straight phonetic spelling. Dummy likewise comes from dumb.
The Old English of the Anglo-Saxons has given us our basic stock of words: life, death, earth, heaven, sun, moon, day, night, black, white, broad, narrow, teach, learn, seek, find, eat, drink, food, meat, fire, wood, tree, eye, knee, hand, foot and so on.
Since medieval times English has adopted cultural loanwords from French. The early ones included attach, certain, chance, conquer, courage, language, money, place, pleasant, royal, strange, sure, tender, value, and even a word as common now as very, which at first meant ‘true’.
Modern loanwords from French come with their present French spelling and a close approximation to French pronunciation: collage, entourage, piquant, pirouette.
Technical terms for use in science are often derived from Latin or Greek. Aqueduct, subaquatic are Latinate counterparts in meaning to ordinary English waterway, underwater. Similarly, Greek elements make up scientific terms such as photosynthesis, polyglot, pyromania.
The <-rrh(o)ea> of diarrhea (‘through-flow’) recurs in other Greek-based words such as catarrh (‘down-flow’), seborrhoea (‘grease-flow’). Scientists have to learn a mini-language of such elements. When such terms escape into common use they often cause spelling problems for the ordinary person, as we saw at the outset.
That leaves a whole array of loanwords that are variously ‘exotic’: kayak is from Eskimo, felucca is from Arabic by way of Italian. The now familiar tobacco comes from Arawak, an American-Indian language.
These various subsystems are often marked by their own peculiar spelling correspondences. If you know a yucca to be an exotic plant, you will not spell it *<yucker>. The <ch> of chief, an early French loan, has the same sound as in native cheap, cheese. The modern loan chef retains its present French value of <ch> (like the <sh> of shop), as do chauffeur, charade. The spelling is not altered to *<shef>. This same <ch> will also spell /k/ in Greek-based words such as character, chemist, synchronic. Similarly, is a ‘Greek’ spelling for /f/, as in diaphragm, philosophy, phobia, symphony.
Borrowing foreign spellings along with foreign loanwords is not the only way of doing it. In Swedish, for example, foreign loans are usually spelt with ordinary Swedish spelling. So French loans coiffure, pirouette are spelt in Swedish as <koaffyr> and <piruett>. If we decided to impose a single uniform system of regular alphabetic spellings and ignore the origin of words, markers of cultural origin would be lost. Would that matter?
English spelling has to cater for a wide range of English accents, which differ in their goodness of fit with present spelling conventions. If you pronounce <w> and <wh> the same in witch, which; weather, whether; wine, whine; then you have to learn by rote which individual words have <wh->. If, as in much of southern England, you pronounce court, cores, floor, formerly, source without an /r/ and hence the same as caught, cause, flaw, formally, sauce, you have to learn which individual words have an <r> and which do not. Most Scottish, Irish and American speakers have kept their /r/ in all positions and so have a spelling advantage here.
The spelling system has to cater as best it can for phonetic differences between speakers. If people were encouraged to spell as they spoke, there would emerge a number of different written dialects of English.
Like flies in amber, English spelling has preserved a continuous record of cultural activity by borrowing foreign spelling conventions along with the borrowed words. The spelling of phlegm tells you that it is a scientific term and that it is related to phlegmatic. But for those who are struggling towards literacy, it might be better to spell it *<flem>.
For detailed references on topics such as spelling reform, spelling and dialect, the spelling of names, types of spelling mistake, homophones and homographs and an analysis in detail of spelling correspondences, see Edward Carney, A Survey of English Spelling (London: Routledge, 1994). A short practical textbook by the same author is English Spelling (London: Routledge, 1997) in the series Language Workbooks.