[Journal of the Simplified Spelling Society, 1994/2 p21-24 later designated J17]
[See Journal and Newsletter articles, Pamflet 15 and Cut Spelling by Chris Upward.]

Err Analysis: som reflections on aims, methods, limitations and importnce, with a furthr demnstration. Part II.

Christopher Upward.

This articl in ritn in Cut Spelng (CS). Th first part apeard in JSSS J16 94/1, pp.29-33.

6. Mispelngs as symtms of th english spelng disese.

Mispelng analyses can be conductd for varius purposes. One reasn, as in th case of th NFER study, may be to try and mesur levls of litracy. Anothr, as with Wing & Baddeley, may be to shed lyt on th sycolojy of riting jenrly, perhaps (tho not in th case of Wing & Baddeley) askng wethr riters retreve spelngs primarily from ther auditry or ther visul memry, ie ar riters tryng to recreate wat they recal of th pronunciation of words, or ar they tryng to reproduce visul imajs? For spelng reformrs, howevr, th main purpos of mispelng analysis is to identify th dificltis that riters experience in tryng to comit a givn riting systm to paper. Wen Thorstad [1] showd english children making nearly 8 times as many mispelngs as ther italian countrparts, her findngs conveyd a powrful mesaj about th relativ dificlty of litracy in two languajs; a simlr mesaj arose from a study by th presnt authr [2] wich found nativ english-speakng students of jermn making nearly 7 times as many mispelngs in ther english mothr tong as in jermn, th foren languaj they wer studying.

For spelng reformrs to incorprate th lesns of mispelng analysis into ther proposals for improved orthografy implys an ergnomic vew of ther task wich contrasts strongly with th aproach adoptd by erlir jenrations of reformrs. Wen Walter Ripman and William Archer first publishd New Spelling in 1910, wen Daniel Jones and Harold Orton revised it in 1948 [3], and wen Laurence Fennelly updated it as New Spelling 90 in 1991 [4], they wer al startng from th premis that spelng reform means respelng words systmaticly acordng to ther deemd pronunciation. Th way traditionl english orthografy is mispelt is scarcely relevnt to this task of desynng a completely new orthografy. Since th mid-20th century an alternativ aproach of step-by-step reform has tendd to domnate thinkng on th subject: it looks for specific falts in th presnt orthografy, and sujests how som of these may be individuly corectd. Th main proposals along this path hav com from Axel Wijk [5], Harry Lindgren [6], and CS. It is in th identification and especialy in th prioritization of such specific falts that mispelng analysis may provide useful evidnce.

7. Simplifyng metaorthografic rules.

How shud we then interpret th errs found in th presnt study? We hav alredy noted that both in size and subject matr th corpus is smal and therfor canot lead to any definitiv conclusions; at best it may hav som indicativ and methodolojicl valu. Of th metaorthografic errs, aberant capitlization was by far th most serius, th gret majority of cases involvng arbitry use of capitl letrs in midsentnce or somtimes even midword. No spelng reform can reduce this problm; wat is required is simply that lernrs be traind to hyr standrds of scriptorial disiplin. Fewr errs involvd failur to use capitl letrs wher convention at presnt requires them. As is proposed for CS (se th CS Handbook, Chaptr 5), th presnt conventions for using capitl letrs in english cud be simplifyd to mach those in othr languajs, for instnce only requiring capitlization of th names of peple and places. Many errs wud be avoidd if th corect forms wer then England, english, America, americn, friday, october. A notebl area of uncertnty, regardng capitlization of th seasns (Summer or summer?) and points of th compass (South or south?), wud then also be larjly resolvd.

As far as word-division is concernd, probbly no spelng reform wud be able to help much, as th conventions ar always likely to be arbitry: wen shud we rite into, wen in to? But ther is no particulr justification for riting a lot as a singl word, as hapnd so ofn in th presnt survey. Lernrs hav to grasp that a simpl noun frase like any othr is involvd here, and that they shud no mor rite alot than they wud rite afriend. On th othr hand, th use of th apostrofe, and hence its misuse, cud be substantialy reduced by simplifyng th presnt rules. Again as proposed in th CS Handbook (Chaptr 5, Part 2), apostrofes can be dispensd with almost entirely for indicating posession (hence peoples wud be both th plural and th singulr posessiv form), and in contractions like dont (th latr omission was alredy practisd by George Bernard Shaw). If these importnt uses of th apostrofe wer abolishd, ther wud be far fewr errs like happen's, injustis's, are'nt. Howevr, pronoun/verb contractions wud stil need apostrofes to sho th difrnce between, say, he'd/head, she'll/shell, and by this criterion th dificlt distinction between it's/its wud stil need to be mastrd.

8. Th NS and CS aproachs.

Vowl sounds ar notoriusly hard to spel in english, and in th presnt study they constituted th larjst singl categry of mispelngs propr. Among th long vowls, th many alternativ ways of spelng th sound /i:/, as e, ea, ee, ei, ie, etc, ar comnly observd to cause particulr trubl, as was seen in our corpus in th forms acheive, corea (=career), meat (=meet), peice, resonable, wierd. Th spelng of /aι/ (Brain [=Brian], buy [=by], deiying/dieying/dieing [=dying], kaliedoscope, liabary [=library], paralized, sosity [=society]) and to a lesr extent of /eι/ (raisist [=racist], waist [=waste]) wer also shown to be trublsm. Isolated mispelngs of th othr long vowls wer seen with /o:/ in lonley (=lonely), and /u:/ in babon (=baboon), crewl (=cruel).

Th classic solution to these problms in th angloamericn spelng reform tradition is that of New Spelling (NS) [7], wich uses th basic long-vowl digrafs AE, EE, IE, OE, UU/UE. Th abov words myt then be spelt as folos [8]: acheev, kareer, meet, pees, reezonabl, weerd; Briean, bie, dieing, kaliedoscoep, liebrary, parraliezd, sosieety, raesist, waest, loenly, babuun, cruel. Advocats of NS hav always regardd these long vowls as representng th cor of th english spelng problm, and they hav few qualms about th major impact ther respelng has on th familir form of words and th apearance of text.

CS on th othr hand is constraind by th criterion of not making drastic chanjes to wat is familir, and therfor canot systmaticly regulrize th spelng of long vowls. CS only reduces thre of th abov spelngs, givng acheve, pece, reasnbl, wher redundnt letrs can be removed. But if CS dos not claim to provide a longterm ideal solution here, it at least larjly overcoms th perenial bugber of th 'i befor e exept aftr c' rule, creating th regularity of eve, leve, sleve, receve, beleve.

Hovrng somwher between long /i:/ and short /ι/ (dependng on accent) is th last vowl in words like happy, stupid, wich presnt conventions may require to be variusly and unpredictbly transformd from Y to IE etc, dependng on th gramaticl form of th word in any givn context. This kind of variation produced a hole bach of errs in th corpus: enemys, happyness, humanites, marride, showey, stupied, worring. For these NS has enemiz, hapines, humanitiz, marrid, shoy, stuepid, wurrying, as oposed to CS enmis, happiness, humanitis, marrid, showy, stupid, worrying.

Related to th long a-vowl problm ar spelngs with a foloing r, with th errs billionare, there (=eithr 'their' or 'they're'), unfare; for these, NS has bilyonaer, dhaer, dhae'r, unfaer, wile CS has bilionair, ther, they'r, unfair. Th /ɜ:/ vowl also causes dificlty, especialy for non-rotic speakrs for hom caught/court ar homofones and ho ar therfor prone to mispelngs like thort for thought, tho this err was not found in th presnt corpus. So we se th mispelngs addition (=audition), afull (=awful), thoght, for wich NS has audishon, aufool, thaut, and CS has audition, awful, thot. Th confusion of aloud/allowed is overcom in NS, wich has aloud for both, wile CS stil distinguishs aloud/alowd. Th mispelng inturperet anticipates NS inturpret, tho CS keeps interpret. A few anomlusly spelt short vowls wich also gave rise to mispelngs, namely dosen't, meny, wepans, freind/frend, apear in NS as duzn't, meny, weponz, frend, and in CS as dosnt, many, wepns, frend.

As regards th unstresd centrl 'obscure' vowl shwa, ther is an aditionl dificlty. Not merely is th choice of letr larjly unpredictbl in traditionl spelng, but so is its position - and it is furthrmor somtimes not spelt at al. Th risks ar therfor multipl: th shwa may be spelt with th rong letr(s), as in acter, catorgery, closists, consios, favourate, independant, intelligant, listern, politicion, Sharan, sponcerd, wepans, alcaholic, crimenals, intelegent/intellegent, knowladgable, orphaniges, prejidice, proberly, corea, Farari, sucure; or th letr may be omitd as in alcholic, diffrent, famly, intrested/intrests, jewellry (cf americn jewelry), misrable; or a letr may be insertd altogethr unecesrly as in inturperet (or it may be placed in th rong position, as in tabel, tho no exampl of this was found in th presnt corpus).

For these words NS has aktor, kategory, kloesest, konshus, faevorit, independent, intelijent, lisen, politishan, Sharon, sponsord, wepons, alkoholik, kriminalz, nolejabl, orfanejes, prejoodis, probably, kareer, Ferari, sekuer; diferent, family, interested, jueelry, mizerabl, inturpret; and CS has actr, categry, closest, concius, favorit, independnt, intelijnt, lisn, politician, Sharon, sponsrd, wepns, alcoholic, crimnls, nolejbl, orfnajs, prejudice, probbly, career, Ferari, secure, difrnt, famly, intrest, jewlry, misrabl, interpret. It is clear that th CS use of sylabic L, M, N, R alows a numbr of mor ecnomicl and predictbl spelngs than NS, wich is surprisingly conservativ and inconsistnt in its treatmnt of shwa, not even harmnizing th trublsm -abl, -ibl sufixs.

With regard to consnnts, th problm of wethr to rite them dubl is resolvd by most spelng reform proposals with th rule that they ar nevr normly dubld. NS is almost entirely consistnt on this, since its regulrization of long vowls means that consnnts do not need to be dubld to sho a preceding short vowl. NS therfor has drugy, akademikaly, aloud, intelijent, eekwaly, realy, poluushon, tely, imatuer, anoi, bilionaer, kwestyonaer, apreeshyaet, droping, arae, posibl, profeshonal, sadnes, of (for off), with only embarrasing and to-morroe keeping RR. Altho CS simplifys most dubld consnnts, it canot always emulate NS. CS has to keep GG from druggie (pairs like tinny/tiny need to be distinguishd by consnnt dublng in CS); it preservs SS in most contexts (profession needs to be distinguishd in CS from adhesion, hardiness from sardines); and th of/off problm remains untuchd; but CS dos normly simplify RR. So CS rites druggi, academicly, alowd, intelijnt, equaly, realy, polution, telly, imature, anoy, bilionair, questionair, apreciate, dropng, aray, posbl, professional, sadness, off, embarasng and tomoro.

Th overlapng uses of th letrs Q, K, C, S, T, X, Z wer discusd in th CS Handbook (Chaptr 6, §1.3.1, §1.3.2, §1.4). With these consnnts too NS ofrs a far mor complete regulrization than CS can. NS has chek, thik, sponsord, konshus, kritisiez, injustises, sosiety, praktis, sukses, raesist, profeshon, seksist, Lester, muezishan, compared with CS chek, thik, sponsrd, concius, criticize, injustices, society, practis, success, racist, profession, sexist, Lestr, musician. Wud NS rite etcetera as etsetera, abreviated to ets? CS keeps etc.

Th remainng, mor mislaneus consnnt errs in th corpus becom NS asthma (th NS dictionry is unsure here, givng asma, astma as alternativs), brilyant, deesent, involvd, tramp, understand, vandal, koodn't, taebl, whaer, as against CS asma, briliant, decent, involvd, tramp, undrstand, vandl, cudnt, table, wher.

Silent letrs ar jenrly omitd in reformd spelngs, tho in CS it has to be remembrd that by no means evry silent letr is also a redundnt letr, and only redundnt letrs ar cut out. For th mispelt words in this categry from our corpus, NS rites aloud (=allowed), sponsord, unfortuenetly, U'r, behiend, muuving, pouch, raesist, huu'z, els, noing, for wich CS has alowd, sponsrd, unfortunatly, u'r, behind, moving, pouch, racist, ho's, else, noing.

9. Desynng spelng for peples needs and abilitis.

Until th mid-20th century at least, most spelng reform proposals wer based on mor or less straitforwrd, but rathr academic, scemes for th consistnt represntation of th sounds of words. These hav had virtuly no influence on th actul developmnt of ritn english. Insofar as we can put names to any of th peple ho, by contrast, hav influenced spelng developmnt in english, they hav tendd to be concernd with th processes of litracy teachng, most notebly th 16th century scoolmastrs Mulcaster [9] and Coote [10], and in America som 200 years later, Noah Webster [11], ho set th americns on th road to a marjnly mor rationl spelng systm than that undr wich th british stil labor.

Th fact that these pedagojicl aproachs hav had a mor significnt impact than hav regulrizing academic scemes sujests that a pedagojicl aproach today may prove th mor productiv. Intlectuly stimulating (even intoxicating) tho al spelng reformrs find th task of desynng ther spelng systms, ther public motivation needs to be th public good.

They need to ask how ther exprtise can be practicly aplyd to publicly beneficial ends, wich may be defined in th brodst terms of improving ritn comunication, or in naroer terms such as making publishng mor cost-efectiv (a larjly ecnomic criterion) or litracy aquisition mor succesful (an educationl criterion). It is with th latr criterion in mind that we may usefuly recal Valerie Yules precept that spelng shud be desynd to meet peples needs and abilitis [12].

So wat can err analysis tel us about peples needs and abilitis? A negativ reply wud be that it merely tels us how stupid peple ar, in othr words wher they lak ability, or wher ther needs conflict with ther abilitis. But that implys we ar prepared to accept that lak of ability shud disqualify peple from ful litracy, wich is tantmount to sayng they do not need ful litracy. No socialy responsbl persn in advanced industrial societis can accept that conclusion today, wen such a premium is placed on th hyest posbl standrds of education for al. (In erlir centuris th educated classes comnly considrd that education for al was a positivly danjerus idea.) Th positiv reply is that err analysis tels us wher th spelng systm fails to meet peples litracy needs because it is not atuned to ther abilitis.

Th importnce of comparativ err analysis such as Thorstads between english and italian is that it proves th falt lies with th english spelng systm, and not with inate lak of ability on th part of lernrs. Much hyr standrds of litracy ar posbl in english too (th i.t.a. experimnt proved that), [13] provided we lern from th evidnce of wher lernrs ar stumblng in th presnt systm. Wenevr a particulr modl of car is found to sufr persistnt accidnts, th most meticulus err analysis is undrtaken to ensure that wherevr human err ocurd, wethr in th desyn or bildng of th vehicl or in its opration, th cause is identifyd and elimnated. Exactly th same principl shud aply to th accidnts of litracy. We can asume that al drivers wud wish to avoid accidnts if they cud, and we can asume al lernrs wud wish to avoid ilitracy if they cud. Th spelng reformrs task is to ensure they can.

Th smal sampl of mispelng analysis demnstrated in this paper indicated th importnce of simplifyng th spelng of long vowls, wich is somthing rathr neglectd by CS (exept insofar as long E is regularized in eve, leve, receve etc, and long I is regulrized as Y in sy, syn, syt, replyd, etc). Conversly, th abov analysis also showd how serius ar th problms of spelng shwa, wich is somthing rathr neglectd by NS. These ar lesns that hav been derived from err analysis and can be aplyd in th desyn of spelng reform.

Th title of this articl referd to 'err analysis' in jenrl, and not only to mispelng analysis. Equaly importnt is th recordng, catalogng, and analysis of errs of readng. Readrs, like riters, stumbl wen ther is som kind of obstacl in ther path, and those obstacls apear not yet to hav been subjectd to th same degree of analysis as hav th obstacls that face riters. One of th objections most ofn rased against CS is that it is feard readrs may hav dificlty decoding its long consnnt strings. Thus CS may be very helpful to riters wen it irons out th arbitry vowl variations between th unstresd sylabls of, say, equivalent, excellent, sibilant, insolent by riting them al with -LNT, as equivlnt, exlnt, siblnt, inslnt, or of permanent, covenant, dominant, continent, consonant by riting them al with -NNT, as permnnt, covnnt, domnnt, contnnt, consnnt; but, it is askd, arnt th vowl letrs necesry for readrs in decoding th sylabic structur of such words? Here is wher err analysis of readng is needd. Valerie Yule has don som pioneerng experimnts [14], but much mor requires to be undrtaken.

Cud reformd spelng systms be desynd on th basis of err analysis alone? Errs certnly do not provide any imediat solutions to spelng problms. They may hylyt wat is rong with th curent systm, but they do not tel us directly wat chanjes ar needd. Taken en mass, they represent a heterojeneus colection of scriptorial junk, th detritus produced by (typicly) yung minds in ther strugl with intractbl material. It is th recurent patrns wich th corpus displays that hav to be interpretd if ther significnce is to be undrstood. But even wen th interpretation has been don, and conclusions drawn, that dos not necesrly mean that a resultng optmm systm for avoidnce of those patrns of err can be proposed for introduction at once. Th next staje is th testng of th proposed simplifyd spelngs, not only on lernrs, but on al categris of user. Th reasn wy th i.t.a. cud not be considrd as a jenuin systm of reformd spelng, despite its overwelmng success as a medium for litracy aquisition by beginrs, was that it was tecnlojicly and syclojicly incompatbl with th existng cultur of litracy.

Err analysis is probbly in its infncy. We need far mor infrmation about th performnce of readrs as wel as of riters, of beginrs as wel as of oldr lernrs, of adults as wel as of children, of non-nativ speakrs as wel as of nativs. Such infrmation can increse our confidnce that our proposed improvemnts to th spelng of english as th primary world languaj wil truly serv th needs and abilitis of al its users.


[1] Thorstad, Gwenllian (1991) 'The effect of orthography on the acquisition of literacy skills' in British Journal of Psychology, 82: 527-37. See JSSS articles.

[2] Upward, Christopher (1992) 'Is traditionl english spelng mor dificlt than jermn?' in Journal of Research in Reading, Vol.15 No.2, September 1992, pp82-94. Summary J13 1992/2 p22.

[3] Ripman, Walter & Archer, William New Spelling, London: Sir Isaac Pitman & Sons Ltd, 6th edition revised by Jones, Daniel & Orton, Harold (1948).

[4] Fennelly, Laurence (1991) New Spelling 90, SSS Pamphlet No.12.

[5] Wijk, Axel (1959) Regularized English, Stockholm: Almqvist & Wiksell.

[6] Lindgren, Harry (1969) Spelling Reform - a New Approach, Sydney Australia: Alpha Books.

[7] Th 1940s version of NS is used for ilustration here in prefrnce to th brodly simlr NS90, as th formr is mor consistnt and a dictionry is availbl. NS90 dos howevr hav som advantajs over th oldr version.

[8] These forms ar taken or deduced from: ed. Walter Ripman (1941) A Dictionary of New Spelling, published on behalf of the SSS by Sir Isaac Pitman & Sons, London.

[9] Mulcaster, Richard (1582) The First Part of the Elementarie.

[10] Coote, Edmond (1596) The English School-maister.

[11] Webster, Noah (1783) A Grammatical Institute of the English Language, subsequently republishd over 400 times undr th title The American Spelling Book.

[12] Yule, Valerie (1986) 'The Design of Spelling to Match Needs and Abilities' in Harvard Educational Review, August 1986, pp278-307.

[13] Downing, John (1967) Evaluating the Initial Teaching Alphabet, London: Cassell.

[14] Yule, Valerie (1993) Orthography and Reading: Spelling and Society, PhD thesis for Monash University, Victoria, Australia, publishd by Ann Arbor, Michigan: UMI Dissertation Information Service, Order Number 9231850, Part III 'Experiments in Orthographic Modification'.

Back to the top.