[Journal of the Simplified Spelling Society, J20, 1996/1 pp30-33]
[See also Adult Literacy and ALBSU.]
[Chris Upward: see Journals, Newsletters, Leaflets, Media, Pamflet 15, Book, Papers.]

Wat's th Problm - Spelrs or Spelngs?

Christopher Upward revews

Writing Skills:
a survey of how well people can spell and punctuate

London: The Basic Skills Agency (formrly Adult Literacy and Basic Skills Unit [ALBSU]), Decembr 1995, 17pp (unumbrd exept in Table of Contents), ISBN 1 85990 044 5.
Availbl fre from th Basic Skills Agency.

This revew folos on from revews of erlir ALBSU reports (JSSS, J8 1988/2, p32; J18 1995/1, p37), and is ritn in Cut Spelng. Thanks ar du to Leslie Morphy of th Basic Skills Agency for comments on a draft of th revew.

1. Context of th survey.

In its J18 1995/1 issu, JSSS revewd th presnt surveys precursr (Carol Elkinsmyth/John Bynner The Basic Skills of Young Adults London: Adult Literacy and Basic Skills Unit, 1994) and noted (p38, §3) that wile th c.22-year-old subjects wer asesd for ther readng and numeracy skils, her riting skils wer not evaluated. Th presnt 1995 survey gos som way towards filng that gap, altho th 980 subjects ar difrnt peple, constituting a representativ sampl of th population of England and Wales covrng ajes 16-60. Th riting skils examnd ar spelng (therby extendng a smal 1992 survey reportd on in JSSS J18 1995/1, pp40-41, §7), punctuation, and to a limitd extent th ability to compose gramaticl sentnces. Th National Foundation for Educational Research helpd desyn th tests, wich wer implmntd by Opinion Research Business.

JSSS has alredy discusd th importnce, aims and tecniqes of spelng-err analysis at som length in 1994 (Part I, J16 1994/1, pp29-33; Part II, J17 1994/2, pp21-24). That discussion was based on th asumtion that th err-anlyst has access to and can study th actul errs made. Th presnt survey of riting skils dos not provide us with such data, but merely with statistics for th percentaj of riters ho mispelt a limitd numbr of comn words. We canot therfor identify precisely wat caused th errs, tho we can sujest som probbl causes and draw som conclusions about th significnce of th survey and its relevnce to english spelng reform. One factr, as always, is th predictbility of th spelngs concernd, but, thanks to th smal numbr of test words used, we can also ask how far th frequency of each word in jenrl usaj, ie, th probbl familiarity of its orthografic form to th riters, may also hav afectd th outcom. Frequencis wer noted in terms of thousnds (designated 'k' belo) of ocurences in th COBUILD corpus of 17 milion words.

2. Dificlty machd against frequency.

Th subjects of th survey wer testd on th spelng of 20 words wich, altho in no way specialized, wud be likely to ocur in an employmnt environmnt. Each word was presentd in th context of a sentnce such as I am sorry to have missed you, or I am looking for accommodation, and they wer furthr categrized in four groups of five, graded by dificlty.

Group 1: sorry, come, please, have, my.


This first group consistd of very comn, mostly monosylabic words, but in no case was th spelng entirely straitforwrd (for instnce, thre of them endd in silent E), and al wer mispelt by at least one persn. Th esiest wer found to be my (tho confusebl with th patrns of mine, lie) and have, with its aberant E misleadngly sujestng a rym with shave. Trickir was come, with two aberant letrs (contrast home, rum), and about 10 subjects mispelt it. Sorry with th uncertnty of final Y (contrast I in horrid, IE in worried) and dubld R (contrast singl R in story, very, bury) trapd twice as many. Please with its arbitry EA digraf, its S for /z/, and its final silent E (contrast th numerus othr posbl spelngs rymng with it in these, trapeze, cheese, sneeze, frieze, seize, pleas, bees) tripd up thre times as many. Wile we may considr that th dificltis containd in please ar much gretr than in have, my, we may also note that COBUILD givs have (including haven't) a frequency of 86k and my of 46k, wile please (lematized to include inflectd forms and derivativs insofar as these displayd virtuly th same letr string) showd just over 3k and sorry only 2k; come (with derivativs) rated 25k, and th total frequency for this group of words amountd to c.160k. Th relativ familiarity of have, my, come may therfor hav reduced ther proneness to mispelng at least as much as th relativ (un-)predictbility of ther letrs.

Group 2: apply, complain, would, writing, because


Th next group containd one very comn but tricky monosylabl (would, 52k) and four disylabic words, of wich only because (23k) was particulrly comn. Writing rates only 2k by itself, but if lematized its frequency rises to 8k. Apply, complain ar both only 1k, and th total for th hole Group is c.85k. Would is dificlt because of its near-uniqe (ie, shared only with could, should) vowl spelng combined with silent L. Because (if pronounced to rym with was) sufrs from th unusul valu of AU, th /z/ valu of S, and th silent final E, and is furthrmor uniqe in its structur (it is etmlojicly od too, th jermanic prefix be- being atachd to th french-derived base word cause). Writing sufrs from silent initial W and th loss of final E from write (it dos not rym with th paralel endng of benefiting). Apply has dubld P, unlike apologise in Group 3. Only complain apears unproblmatic - until one compares it with plane. As in Group 1, th two comnst words produced fewst errs, altho th longr, less comn because surprisingly saw only 2% mispelngs, wile th shortr, very comn would came off over twice as badly at 5%. Th rarer but less aberant apply, complain both scord 7% errs, wile writing did worst in th group by far with 11% rong. Despite som anomlis of detail, it again apears that familiarity carrid mor weit than predictbility in facilitating acurat spelng.

Group 3: apologise/-ize, unfortunately, allowance, receive, sincerely


These words ar jenrly longr, with 2, 3, 4 or 5 sylabls, and of much loer frequency: receive (lematized) just tops 3k, (un-)fortunate-ly almost reachs 2k, wile sincere-ly, apologise/-ize and frends ar belo 500; allowance itself is a mere 300, but its base word allow and derivativs giv 5k. Th total for th Group amounts to c.11k. Th spelng of these words may look less blatantly aberant than, say, come or would, but they harbr numerus sutl traps. Th singl P of apologise/-ize is uncomn beside th domnnt PP of appear, approach, appeal, as wel as apply in Group 2; its singl L difrs from its greek compatriot Apollo; its secnd O is an indetermnat shwa-vowl; its G sujests a spelng with J; and its I contrasts with Y in analyse/-yze; all of wich errs togethr wud perhaps produce *appollajyze. Th first sylabl of fortunate is homofnus with four; its U is jenrly asimlated to th preceding T, givng th sound of CH; th A is pronounced shwa, wich in secretly is E and in definitely is I; th E is silent; and th endng myt be rongly gesd at by analojy with incidentally; al of wich combined cud lead to th multipl mispelng *unfourchenitally. Allowance ofrs th hazrd of dubl L (allowed is ofn mispelt aloud), th dubly ambiguus -ANCE endng (contrast influence, immense, defence/defense), and th final silent E; *alouens myt therfor be a concevebl represntation. Receive is notorius for mispelngs with -IEVE (cf achieve, but also eve, leave, sleeve), but in adition th C is pronounced like S, and th final E is silent; *resiev wud therfor not be an altogethr unintelijnt atemt. Sincere likewise is comnly seen with S for C, and with EER for ERE (*sinseerly). In th event, allowance incurd only 15% errs, perhaps thanks to th high frequency of allow, but th rest wer mor than twice as err-prone, being mispelt by at least one riter in thre. Th scors wer, in asendng ordr, sincerely 33% mispelt, receive 37%, and apologise/-ize and unfortunately equal at 40%. Th relativly hy frequency of receive was clearly no defence against its notoriety, but th bad shoing of unfortunately is less esy to explain; sincere and apologise/-ize wil hav been significntly less familir to th riters.

Group 4: maintenance, immediately, necessary, occasionally, accommodation


This most dificlt group consistd of th longst words (3, 4 or 5 sylabls), with rathr lo frequency (lematized as apropriat, frequencis wer necessary 4k, immediate-ly 3k, occasional-ly 1.5k accommodat-ion 800, maintenance undr 500), tho th total (10+k) was scarcely belo that of Group 3. Th spelng hazrds they contain ar of th kind that afects th hole english languaj, rathr than being rare anomlis like come, would, receive. Four of th words contain dubld consnnts, 2 x CC, 1 x LL, 2 x MM and 1 x SS. Four contain post-accentul shwa befor L, or N, or R, twice befor N in maintenance, and once befor L in occasionally, befor N in accommodation, and befor R in necessary. Shwas also ocur befor othr letrs in immediately, necessary, wher they myt be difrntly spelt as in quietly, emissary (ie, givng *immedietly, *necissary). Th spelng of /s/ varis between C in maintenance, necessary, and SS in necessary, and th sound valu of C varis between /s/ in those two words and /k/ in occasionally. Th adverbial endngs ar pronounced virtuly th same in immediately, occasionally, but ar difrntly spelt. Th noun maintenance relates to th verb maintain, wich cud sujest (by th chomskyan principl of morfofonemic stability) th mispelng *maintainance. Of th 980 subjects, over 410 mispelt maintenance, immediately, necessary, wile over 540 mispelt occasionally and over 670 mispelt accommodation. No dout som of th mispelt words containd mor than one err, and th total numbr of errs cud even be substantialy gretr than th numbr of mispelt words. In th lyt of such figrs, it wud perhaps be surprising if any of th points of dificlty noted abov faild to atract at least one err, but th report dos not giv th evidnce. No particulr corelation between numbr of errs and frequency of words seems worth noting for Group 4.

Th politicl implications of err-rates over 50% ar worth reflectng on. They represent in efect a majority vote of a sampl of th population in favor of simplr spelngs for th words concernd. Shud not th majority principl somhow also aply in deciding how a languaj is spelt? Th implications of these percentajs shud giv dictionry-makers pause: ar they democraticly acountbl or not?

3. Ke statistics burid.

Th survey has no coments to make on th orthograficl implications of th abov results, but it dos anlyz th percentaj of mispelngs for each word by certn categris of subject: by sex, by aje-ranje, by employmnt status, by socio-ecnomic group, and by educationl qualification. It dos not, howevr, giv th results for Wales sepratly from those for England, altho ther ar grounds (initial litracy aquisition in a languaj with predictbl spelng) for thinkng welsh subjects myt hav performd betr. As one wud expect, th socialy and educationly disadvantajd performd least wel, indeed th results may even hav somwat overestmated nationl standrds, since th subjects wer intrvewd at home, so excluding th exeptionly disadvantajd homeless, as wel as servng prisnrs.

Curiusly, th survey dos not avraj out th results across al th words for th varius tables ar givn for th percentaj of mispelngs of each of th 15 words in th thre mor dificlt word groups (2, 3, 4), one table for men versus women, and anothr table for five aje-groups; but th readr is left to do th arithmetic needd to produce a composit figr for standrds by adng th percentajs for th 15 words and dividing them to produce an avraj. If we spend a few minuts doing th calculations, we ar able to say that women outperformd men not merely for evry word exept would, maintenance (men betr) and complain (sexs equal), but that overal women (27.9% mispelngs) outperformd men (32.7% mispelngs) by nearly 5%.

This failur to avraj th statistics for th seprat words is especialy surprising wen it coms to th analysis by aje-groups, since a distinct trend is therby blurd. A table hedd 'Younger and Older' givs th percentaj of mispelngs for each of th 15 words made by five difrnt aje-groups. Th overal avraj for each aje-group, once we hav workd it out, tels us that th worst spelrs by som distnce wer th yungst: th 16-24 year-olds avrajd 35.7% words mispelt; th next worst wer th oldst (55-60 year-olds), ho avrajd 32.8% mispelngs; th thre midl groups performd noticebly betr, tho with a slyt decline thru successiv jenrations, th 45-54 year-olds avrajng 27.0%, th 35-44 year-olds 27.5%, and th 25-34 year-olds 28.6% errs.

Aje-group16-2425-34 35-4445-5455-60
% rong35.728.6 27.527.032.8

It is posbl, of corse, that these results ar a quirk of th relativly smal sampl (an avraj of undr 200 subjects per aje-group), but inevitbly we find ourselvs lookng for causes of th variations. Thus we may speculate that th education of th oldst aje-group very likely sufrd from th erlir scool-leving aje and th social turmoil (wartime evacuation, etc) of th 1940s, and ther myt may also be som efect of ajing. In terms of litracy policy, howevr, th decline in standrds shown by th yungst aje-group is th most striking findng. Ther hav been repeatd alegations in recent years that th litracy standrds of yung peple hav seriusly deteriorated, yet this claim has also been contestd, and th evidnce has hithrto been tantlizingly inconclusiv, restng mor on anecdote than on relybl statistics comparing standrds of successiv jenrations. From th data givn in this report we ar able to derive such figrs - and if they ar representativ, they ar surely alarmng. They imply that th spelng-acuracy of 16-24 year-olds is over 7% worse than that of peple ten years oldr, over 8% worse than those twenty years oldr, and nearly 9% worse than those thirty years oldr. Or, to put it a difrnt way, todays 16-24 year-olds mispel over 25% mor words than do ther eldrs. It has been sujestd that spelng-acuracy improves with aje and experience, but can such a factr explain th dramaticly reduced acuracy of th yungst jenration? Mor plausbl reasns ar perhaps a jenrl weaknng of readng habits in favor of television (se JSSS J19 1995/2, p7, §5, on th IEA litracy survey), and a chanje in teachng methods away from fonics toward look-and-say and of atitudes away from authoritarian prescription toward fre expression. Th Basic Skills Agency tels us (persnl comunication) that it is "wary about making firm judgements about standards declining on the basis of this research", altho they "do tend to think that there was a period in schools when rather less concentration was paid to 'secretarial' skills than was perhaps desirable".

Th survey reports mor briefly on performnce in two othr 'secretarial skils' (as th National Curriculum describes them). Regardng punctuation, over 52% of subjects wer jujd to hav a poor undrstandng especialy of th rules for use of th apostrofe, but th yungst aje-group did not perform noticebly worse than th midl groups. On th othr hand, in th riting task (form-filng), mor in th yungst aje-group performd poorly or very poorly than in th othr aje-groups.

Th survey thus dos not hylyt th crucial figrs and issus concernng spelng, wich remain burid in its vivid and colorful diagrams and colums of percentajs. It tentativly concludes that "the results suggest that more people in the youngest and the oldest age-groups had difficulties than in the other age-groups" and "more younger people, some of whom have only left school recently, seemed to have difficulties than those ten or twenty years older", but th only lesn it draws is to say that "the results reinforce the need to improve standards" because job-aplications containng mispelngs ar ofn autmaticly rejectd. Spelng reformrs wil inevitbly feel that th most importnt questions rased by th survey hav been alowd to go begng.

4. Implications for spelng reform.

Efectiv tho th survey was in drawng media atention to poor standrds of riting (especialy spelng) in England and Wales, it scarcely begins to considr th implications of its findngs. It takes a naro vew of th importnce of good riting skils, wich is seen only in terms of avoidng rejection of job-aplications, and not in terms of enhanced powrs of comunication, nor as suportng readng skils and thus enhancing litracy standrds and therby th potential for th individuls educationl advancemnt jenrly.

Ther is also ambiguity in th surveys presntation of spelng dificlty. On th one hand, certn spelngs ar described as dificlt, but on th othr hand certn peple ar described as findng them dificlt. So dos th problm lie with th spelngs, with th spelrs, or with both? Wy ar only som spelngs dificlt, and wy do only som peple find them dificlt? To crak this conundrm, we need to examn th spelngs themselvs, and find out wy som peple trip over som of them (and indeed wy most peple trip over occasionally and accommodation).

Wy som peple succeed wher othrs dont, regardless of th dificlty inherent in any particulr spelng, is a question of individuls abilitis and education. As with al skils, th levl of proficiency acheved in litracy skils depends on a combnation of aptitude and trainng. Ther now seems to be wide agreemnt that th fashnbl trainng methods of recent decades wer misconceved, and that maxmizing litracy skils depends on systmatic developmnt of fonic undrstandng. But even if deficiencis in recent trainng (wich may partly explain th jenrationl difrnces observd) ar rectifyd, we must stil expect lak of aptitude (seen in its extreme form in dyslexia) to limit th litracy levls achevebl by som lernrs. So much for th dificltis orijnating in spelrs themselvs.

Wen it coms to identifyng th dificltis inherent in som spelngs, th analyses (wich ar based on extensiv experience of err-analysis) givn in §2 abov ar intendd to explain wat it is that makes som spelngs mor dificlt than othrs. Undrlyng them al is th alfabetic principl. Th fenomnl success of th alfabet as a riting systm around th world over thousnds of years is du to th simpl device of using th letrs to represent speech-sounds. Wen a languaj uses th letrs in this way to spel its words, ther ar no dificlt spelngs (exept insofar as long words may require mor careful atention than short words), and hy standrds of litracy ar esily acheved. But wen a languaj uses letrs unpredictbly, then dificlt spelngs and consequent litracy problms ar th inevitbl result. It is to minmize this danjer that most languajs hav in th 20th century modrnized ther spelng systms to keep them as closely alynd with th alfabetic principl as is practicbl. English has not systmaticly implmntd th alfabetic principl for nearly a thousnd years now, a histry of neglect wich has produced myriads of dificlt spelngs in modrn english.

But ritn english is not bound to sufr dificlt spelngs for evr: they can be made esir. Th Cut Spelng (CS) used for this revew takes a considrbl step in that direction, mainly by removing redundnt letrs. Th 20 words on wich th 980 subjects wer testd apear as folos:
We may confidntly predict that if these CS forms had been th target spelngs testd, th litracy scors acheved wud hav been much hyr. Yet CS by no means represents th ultmat in simplicity - a mor radicl reform than CS cud acheve much mor. Of th words unafectd by CS, sorry, because, sincerely, complain myt perhaps becom sori, becoz, sinsirli, komplaen, wile th CS forms cud undrgo furthr chanjes to produce kum, pliiz, wwd, rytng, apolojyz, unfortiunatli, alouens, resiiv, maintnns, nesesri, okaezhnli, akomodaeshn. Ther is no need for my to be chanjed. Clearly th abov mor radicl respelngs cause such an upheval to th visbl form of english that ther introduction cud not be contmplated in th short term, but th fact that respelngs sujest themselvs for 19 out of th 20 words testd is itself eloquent testmny to th deeply unsatisfactry natur of th presnt spelng of english.

5. Conclusion.

We may conclude by drawng atention to an ambiguity in th surveys subtitle "A survey of how well people can spell and punctuate". Th word can is misleadng here: experience with mor predictbl riting systms shos that peple can spel a gret deal betr than they do at presnt using th traditionl orthografy of english. Th survey reports on todays disml standrds, but says nothing about futur posbilitis. Th Basic Skills Agency shud extend its horizons.

6. Aftrword: Difficulties with Basic Skills.

If th Basic Skills Agency has yet to aknolej th ke lesn of its riting skils survey, this may be partly because of othr targets it has been concentrating on of late. Thre months befor publishng th Writing Skills survey, it reportd (Difficulties with Basic Skills by John Bynner and Jane Steedman, 80pp) on a mor substantial pece of reserch. In th latr booklet it examns th social causes of th litracy and numeracy problms reportd in its 1994 survey The Basic Skills of Young Adults (se §1 abov). It concludes that th most importnt factr determnng proficiency in basic skils for yung adlts is ther childhood bakground, wich if unfavorabl tends to produce a cycl of lo achevemnt thru successiv jenrations. This unsurprising findng is bakd up by som quite sofisticated statistics, and leads to th cal for mor suport for th home and for th yungstrs concernd.

That ther cud be somthing fundmently rong with th very medium thru wich scools atemt to instil litracy skils jenrly is not mentiond. Ther is no dout that far too many children ar seriusly handicapd in ther education by varius kinds of deprivation, and that this handicap shud be adresd. At th same time, spelng reformrs wil need to go on remindng th Agency and othr oficial bodis concernd with litracy standrds that deprivation is not th only handicap afectng those standrds. Th caos of english spelng handicaps al lernrs, and to concentrate on those ho ar most severely afectd is like advocating dietry improvemnts only for th most severely malnurishd wen th hole population is sufrng from malnutrition. Giv special asistnce to th most deprived, and we may expect them to benefit. Simplify th spelng (probbly at less cost), and we help evryone.

Back to the top.