[Journal of the Simplified Spelling Society, 1993/1 pp10-18 later designated J14]
[See Journal, Newsletter, Anthology, Bulletin articles, Personal View and web by Valerie Yule.]

Improving English spelling for readrs.
Th necessity for reserch.

Valerie Yule, Monash University, Australia.

It is posibl to improve English spelling for readrs, and omission of letrs surplus to readrs' needs cud facilitate fluency and comprehension in reading. These two statements must be demonstrated by reserch, not left to argument or assumption. This articl discusses th visual fonologicl and morfemic features to be considrd in designing optiml spelling for readrs. It emfasises th need to cater for th wide ranje of linguistic ability among English-speakrs and internationly, and outlines th design of some of th reserch reqired and extant to determin what spelling chanjes may be most qikly adjusted to by present readrs, and what chanjes cud improve reading skills most.

Spelling reformrs hav a hard time of it becaus present English spelling is jeneraly assumed to be unchanjabl - unlike almost anything else around them, even th languaj itself. But th whole field of reading is also ful of assumptions, many of them taken for grantd by spelling reformrs themselvs. This articl is not as simpl as it may seem in its revew of some long-lived assumptions.

Improving English spelling to improve reading.

Conservativs do not realise that English spelling needs to be improved to help readrs. Th spelling reform movement in jeneral does not eithr. Both sides tend to assume that lerning to read is th batlfield, and that once someone can read, th nature of th riting system is not a furthr issue.

Spelling reformrs focus on improving English spelling for lernrs and riters, who ar seen to sufr so gratly from traditionl orthografy (TO) (Yule 1991). Th stated aims of th Simplified Spelling Society (1992) are 'to bring about a reform of th spelling of English in th interests of ease of lerning and economy in writing.'

'And what about th workrs?' cries a voice from th bak.
What about th major users of th ritn word - th readrs?

Th importance of reading in spelling reform.

Conservativs ar, I think, correct in making ese of reading th top priority for a riting system, altho rong in arguing that th very features of English spelling that make it dificult to lern make it esyer to read, and so shud not be chanjed. Only a few riters ar actualy needed to keep a modrn economy and society operating, and computers can now handl th spelling mecanics for thm. But it is desirabl for everybody to be able to read, and almost evry literat person spends a good deal of work and lesure time reading th printd word. Lerning to read is a once-in-a-lifetime task (normaly), and if th end result is suficiently rewarding in status/ plesure/ oportunitys/ mony, and teaching is excelent, lerners wil make as great an efort to lern as sportsmen today to excel in sport, for th same reasons - as in 18th century Scotland, which had 90% literacy compared to England's estimated 16%, and modrn Japan (altho this is no good argument for leaving a poor riting system unimproved).

Some authrs also state boldly that an orthografy easy to lern might be too dificult to read or rite, or that th most economic riting might be very dificult to read or to lern (e.g. Frith 1980). Shaw's alfabet, probably th most economicl English spelling conceivable is very dificult to lern and even its designr found it hard to rite (Read 1973). A spelling system needs to be a 'best fit' for th needs and abilitys of all three categorys of user.

There is also th pragmatic point that powr resides with those who ar alredy literat, and they wil be most likely to accept spelling chanje if it promotes and facilitates reading with comprehension with miniml disruption to thir present habits.

Improving print for readrs.

Reserch-based knolej is continualy growing on how readrs can be helpd by th way riters organize and express their messajes. We hav a growing reserch base for knolej on how reading can be facilitated by improvements in layouts, paragrafing, fonts, print size, and colors for typeface and bakground (See for exampl th work of James Hartley at Keele). We kno that spacing words and th addition of punctuation ar definitly invaluabl inventions to aid readrs and there is also reserch on th most eficient direction for a script (e.g. Gray, 1956).

Th gratest gaps in our knolej about how to improve print for readrs ar at th most essential levl, in how riting systems themselves may facilitate or handicap. Indeed, reserch in spelling reforms is an exellent way to expand our knolej of how peple read, since at least 95% of th world's reading reserch availabl in English has been conductd with th assumption that TO is a fixd fact of life, as wel as current fact of English print. (A ruf estimat from th bibliografys of reading reserch now availabl on CD-Rom, such as ERIC and Silver Platter.)

Th folloing discussion of print for readrs wil assume continuation of th roman alfabetic script, since it is in line with intrnationl usaj, altho stil keeping, on th bak boilr of th mind, th posibility of some future revolutionry breakthru to a completely difrent typ of riting system that can cross languajes.

Reserch with an imediat practice orientation must considr:

a) Sales appeal. What typ of spelling reform cud current literats adjust to imediatly with case?

b) User appeal. What typ of spelling reform wud turn out to benefit readrs mor than th spelling they alredy kno, after an adjustment and practice period of say six months?

c) Future appeal. What typ of spelling reform wad benefit future readrs most, disregarding th retirements of those presently literat and maintenance of bakwrds compatibility?

A standrd spelling for readrs.

A standrd spelling system is needed for all readrs, regardless of their reading stratejs. 'Spelling as you speak', without any standrd, has been advocated as a solution to riters' spelling problems [1], but this wud make a massiv increase in dificulty for readrs. Most peple cannot rite down acuratly how they or anyone else actualy speaks; and even when they can, th extreme diffences between dialects and continents wud hindr comunication. Th growing babelisation of spoken English is becoming an increasing concern, as nationl and local sentiments override th purpos of internationl comunication. (See, for exampl, articls in English Today, such as McArthur 1987 and Urdang 1990). Local spellings wud increase this centripetl disintegration of th English languaj.

A standrd spelling for readrs is also essential becaus fast automatic word identification depends upon th familiarity of th orthografy. This is esily proven. Try reading storys ful of misprints or in dialect spelling even when there ar no diacritics.

Curiusly, this automatic word recognition seems unafectd by chanjes in font or print-size. Salient words such as your name or special intrests jump out of a paje at yu, regardless of how they ar printd. In th Stroop test, when good readrs try to name colors of colord inks used to write th names of othr colors, they get confused by what th words say, wheras very poor readrs can name th colord inks without being distractd by what th print says. That is, word-recognition does not take place at th levl of th literal visual apearance of words, but from an abstraction, as a mor centraly located process. This is why th letrs and th spelling that make up orthografy matr, even tho their exact shape or dress does not.

This stil leves room for alternativ spellings, as they can become familiar too. Our dictionarys accept alternativ spellings for thousands of words, e.g. JAIL/GOAL, and SILVAN/SYLVAN. These ar imediatly recognised because of their familiarity. Mor words cud hav alternativ spellings within a standrd spelling system - but that standrd orthografy remains essential.

Expert readrs, averaj readrs, strugling readrs.

A spelling reform must take into acount that th needs of readrs of difrent ability wil difr to some extent. To some extent, what suits lernrs wil suit readrs, since many adult readrs remain basicly lernrs and almost evry readr must encountr unfamiliar words. But th reading stratejys of expert and weak readrs can be very difrent. Some 5% of th literat population - an estimat extrapolated from reserch in tertiary institutions - ar fast readrs who can process complex prose with ese and aparently skim pages. Expert readrs can mastr almost any form of script, and a spelling reform shud enable them to skim even fastr, harvesting meaning at a glance, to th limit of th inbilt information-processing capacities of thir nervus system. [2]

On th othr hand, a hi proportion of th English-reading population operates hardly above basic literacy. Since most of them do not read for plesure, they do not lern to read fast, and wil continualy be encountring words that they hav nevr read befor, or hav forgotn.

And at least 15% of th 'literat' population continue to strugl in reading as if they wer perpetual novices. Surveys of adult semi-literacy in English-speaking cuntrys ar periodicly luridly reportd in th press (see for exampl Internationl Literacy Year reports). Furthr, most readrs of English today ar not nativ English speakrs, and therefor need an English spelling that links th ritn word closely to th spoken word, so that they can increase their knolej of spoken as wel as ritn English thru their experience of reading, and use their knolej of th spoken languaj to help them in their reading.

Spelling is important for reading.

A comon belief in education is that spelling may be relevant to riting, and even to lerning to read, but not to reading. Th formost guru is Frank Smith (1982), altho Goodman (1982) also stresses that good readrs rely on prediction from context and th visual configuration of features of words that become automaticly recognised thru familiarity; reading is 'a psycolinguistic gessing game'. Howevr, in this game, th best readrs hav multipl stratejys to operate with while poorer readrs may hav only a few resorces or perhaps even only one. They esily sufr information overload if trying to operate mor than one stratejy at once. Trubl occurs with poor readrs when they can only use fonics without developing visual memory (slo poor reading) or can only rely on memory for sight words and gessing (inaccurat poor reading). A spelling reform to help all typs of poor readrs wud be suficiently fonemic, morfemic and consistent to make all stratejys esyer for poor readrs, to alow them to capitalise on mor tecniqes, and benefit mor from thir own particular strengths.

How can th spelling system maximise th accessibility of multipl stratejys in reading?

Fonemic spelling reform for readrs.

Some conservativs claim that it is important for readrs that English orthografy shud not be 'fonetic'. In this, of cours, they overlook th abilitys of readrs in th many languages that do hav a mor consistently fonemic spelling than English - and in fact, comparativ reserch on this matr is not esy to find. There ar several lines of argument put up against a 'fonetic' spelling system for English readrs. All can be found falacius, I think.

Gillooly (1973) has made an influential claim that a fonetic spelling reform wud slow down readrs, on th grounds that it wud force them into 'sounding out' in ordr to read. This claim ignors th fact that a printd word once familiar, is no longr 'sounded out', but recognised imediatly - and indeed for skild readrs, even th 'sounding out' is not th slo articulation it is for a lernr, but a fast mentl abstract process. Readrs in languajes with basicly fonemic spellings such as Italian ar not restrictd to fonologicl stratejys any mor than we ar.

It is also comonly confidently stated in textbooks on reading that 'tho reformd spelling may simplify th pronunciation of words, th cost wud be hi to skild readrs who get gramaticl and semantic information about words from their orthografic forms' (Gibson 1975, p 187). But a reformd spelling cud esily improve on TO's present shaky representation of gramaticl and semantic information, as consistent modifiers of a basic alfabetic principl of sound-symbol correspondence, e.g. participls cud stil conclude with -D/ED without needing to distinguish /d/t/ pronunciations, and plurals cud be concluded with -s, without needing to distinguish /s/z/ pronunciation.

Gibson states as confidently: 'Th principal basis for th eficient recognition of words is th intraword conditionl redundancys jenerated by orthografic rules. Fonecticly precise spelling wud remove thse important clues to eficient word perception.' [3] But cud a reformd spelling improve on these suposed benefits of TO for readrs? As described by Gibson herself (pp 190 ff) th 'intraword conditionl redundancys jencrated by orthografic rules' refers to information about letr constraints that reduce uncertainty and facilitate reading th word. These constraints ar conditionl rules about position of consonant clustrs, and numbr of vowels that can follo one anothr. As readrs gain experience, they ar increasingly able to make use of this knolej for speedy word recognition. However, th falacy of th anti-reformrs lies in not recognising that a consistent basicly fonemic English spelling system wud stil hav such constraints just as much, or even mor, than our present less predictabl TO. Orthografic redundancy, legal letr seqences, and pronounceability wud simply work mor closely togethr, and so mor eficiently.

To be universaly accessibl to readrs, a standrd alfabetic riting system must be suficiently decodabl for poor averaj readrs and overseas readrs, as wel as lernrs, to be able to work out unfamiliar words with acuracy. And unless we hav picture or symbol riting, that must mean a visibl relationship to th spoken languaj - i.e. to some degree fonemic. Altho not necesarily purely fonemic, as long as consistency is maintaind.

There is stil controversy over how much dialect speakrs ar disturbd in their reading by fonemic spelling that reflects a standrd pronunciation. Howevr, th evidence is that today internationly 'English-knoers' can undrstand British Receved Pronunciation or standrd American speech from film and broadcasting, and spelling that follows these two speech forms can also be understood as a very broad-band (diafonic rathr than fonemic) representation of anyone's speech, even when it is most closely connectd to those norms. [4] When comunication rathr than self-expression is th aim of any riting, then th wider th readrship that can recognise th spoken languaj that is representd th betr. This typ of uniformity is as sensibl as having standrd gajes for nuts and bolts, and standrd gajes for linkd railways.

Beginrs lerning to read must hav some sort of fonolojicl stratejy to help them work out new words, unless they hav unusualy good visual memorys (Yule 1992) - hence th major problem for yung def lernrs. For lernrs who practice with motivation, automatic recognition of th words qikly follows, and it is only novl words that absolutely reqire th fonologicl rute. A good readr sails ahed mainly on automatic recognition and clues from prediction, but fonology remains a bak-up, not only as th chief stratejy for novel words, but as th safest means of cheking th acuracy of gessing from miniml clues, and reinforcing visual recognition.

There is some evidence that fonology monitrs their reading anyway, even with lografic scripts. In th first place, a good readr uses short term memry to remembr th beginning of a sentence by th time th end is reachd, and this apears to hav a strong fonologicl mediation. Expert readrs ar also sensitiv to styl and th rythms of text, which suports th notion that they can 'hear' as wel as see what they read. In tests of languaj skils, including fonological skils, they usualy come out wel ahed of othr categorys of readr. Altho some say it is th reading experience that develops th fonolojical skils, th evidence sujests that th process is spiral, each factor encurajing furthr development of th othr.

That is, fonology does matr to all categorys of readr, and so an improved spelling shud atempt to remove impediments in th way of consistent fonologicl representation.

Howevr, I wud also claim that a completely fonemic spelling wud not be in th best intrests of readrs, any mor than in th best intrests of child lernrs (Yule, 1992). Alfabetic spellings shud maximise th advantaj of th alfabetic principl of linking th ritn languaj to th spoken, but consistent modifications of such a fonemic spelling may also help readability. Th need for modification derives from th nature of th English languaj itself. Much of th vocabulary is bilt up from base morfemes (units of meaning); all polysylabls contain unstressd vowels which present a problem for fonemic representation; there is a furthr problem with around nineteen English vowel sounds and only five Latin letrs availabl to represent them; and while th hi numbr of homofones (words that sound th same) is not th major issue that is sometimes stressd by oponents of fonemic improvements, they must also be considrd.

Fonology and representation of morfemes.

Conservativs oftn cite Chomsky's opinion that English spelling is alredy 'optiml' for readrs becaus, he claims, it represents th underlying lexicl forms of th languaj, in that when related morfemes chanje pronunciation, th spelling wil, he claims, stil represent th same 'underlying fonology' (Chomsky 1970). Chomsky's theory has been exploded by th evidence not once, but many times (see Downing & Leong 1982, Yule 1991 for bibliografys of reserch) but it is stil quoted like any theory that justifies th status quo or a vestd interest - explodabl but unsinkable. His anecdotal evidence of related spellings such as NATION/NATIONAL, MEDICINE/MEDICAL, as wel as th singl letr vowel representation for both long and short vowels in polysylabic words such as EDUCATION/EDUCABLE and IMPOSE/IMPOSITION/IMPOSING - with IMPOSTER - as a derivative can be countrd by a mass of anti-Chomsky exampls in English spelling such as SPEAK/SPEECH, FIRE/FIERY, SUCCEED/SUCCESSION. Howevr, there is a case that English spelling might be improved for all categorys of user if it wer modified to resembl his ideal mor closely than Chomsky fancies alredy exists. In practice Chomsky's argument about representation of 'deep fonology' means stable representation of morfemes. Th argument is basicly that readrs can undrstand meaning mor qikly if words with similr related meanings share a comn spelling base, so that recognition of morfemic elements can lead strait into reading for meaning (e.g. COM= with, MEDI= related to medicin, ANTI= against). This relationship cud easily be improved in English spelling, if morfemes (units of meaning) wer consistently speld mor clearly and stably than they ar now. A morfo-fonemic spelling wud be basicly fonemic, but modified consistently to promote visibl representation of units of meaning, e.g. JELI/JELID, STABILITY/STABIL. However, reserch is needed to find th actual degree to which visibl identity needs be preservd. For exampl it is probably a matr of one-trial lerning to realise that FLY/FLIGHT/FLEW/FLIES (or even SLAY/SLAUGHTER/SLEW) hav related meanings - but cud a reformd spelling speed th process of reading for meaning if th words wer speld say FLY/FLYT/FLYS/FLU?

Improving vowel spelling for readrs.

TO has over 318 vowel spelling patrns for th 19 or so English vowel sounds. One argument that has been put up is that this adds to th distinctivness of words for practisd readrs, so that they ar imediatly distinguishd and recognised mor esily, since they hav mor distinctiv features. And so there is a value in spelling demons such as BEAUTY, MAUVE, FLEGM. Th oposing argument is that rationalised and consistent vowel spellings wud speed word recognition, in line with th primacy of th Word Freqency efect, that th mor freqently a word ocurs in print, th fastr it is recognised by readrs. On this argument, th mor freqently a spelling patrn ocurs, th fastr th recognition of a word containing it, if th whole-word freqency efect itself is held constant.

BUTY might ride on th bak of DUTY, MOVE on STOVE, FLEM on THM. Certanly analojy stratejys cud facilitate reading by bakwrd readrs, if all th vowel spelling patrns wer reduced to say only forty, which wud alow for distinctivness of final position vowel spellings, a grater degree of bakards compatibility with our heritaj of print, reduction of adjustments reqired for present readrs, and alternativ spellings for homofones when this is proved desirabl by reserch.

Almost evry spelling reform recognises representation of th 19+ English vowel sounds as th gratest dificulty in TO and problem for reform. Our most comn usajes for th singl letrs A E I O U do not mach their use on th Continent and in most othr alfabetic orthografys of th world - altho this is not a major problem in vew of human ability to adjust qikly and even automaticly to any consistent systematic chanje of values. Howevr, th distinction between long and short vowels in print is a major spelling problem in TO and a major tiklish problem for spelling reformrs. Dozens of difrent expedients to represent th long vowels hav been sujested and a wide disparity hav been incorporated into TO to add to th confusion.

In TO, dubld consonants and 'majic <e>' are clumsy and erraticly aplied expedients to retain stable morfemic representation, (e.g. MAT/MATTING, MATE/MATING). Modified spellings may atempt to introduce new letrs, or diacritics, or folio th practices of reform proposls such as World English Spelling, th Initial Teaching Alfabet, American Spelling and Nue Speling wich remove th clumsiness and inconsistencys but also th morfemic stability by using th digrafs AE, EE, IE, OE, and UE, of which two ar relativly unusual combinations and one is givn a new pronunciation. Upward's Cut Spelling is orijinl in th degree to which it completely supresses shwa indeterminat vowels, and by so doing clarifies th distinction between short and long vowels, e.g. HOP/HOPNG/HOPE/HOPING. This eliminates th need for dubld consonants, altho Upward retains th convention of using 'majic' silent letrs -E and -I that can indicate a long preceding vowel. Yule's moderat CS retains dubl consonants rathr than delete as radicly. J. H. Martin's introductry spelling system uses an optional macron to distinguish long vowels, but alows children in their own riting to make litl distinction between long and short vowels, as is their norml practice, as wel as evident in pijin spellings, wich fulfil a need for maximum simplicity and broad-band representation. Th neatest solution is clearly Upward's CS. But reserch must find whethr this is th most user-frendly. [5]

Unstressd vowels ar a notabl feature of spoken English, and their erratic spelling is a notabl feature of ritn English. They ar sometimes omitd, as in TABLING, speld with < e > as eqivalent to ə shwa as in LABELLED, spelled acording to a forml pronunciation that maches related words as in METAL/METALLIC, CIVIL/CIVILITY, or quite unpredictably, as in SCHOLAR/ACTOR/WAITER/MURMUR/TAPIR. Most spelling reformrs ar unconsiusly influenced by TO in representing shwa, but atempt to make th spelling less of a gessing game. Cut Spelling cuts a numbr of Gordian knots in completely cutting out schwa spellings, as in METL, CIVIL, SCOLR, ACTR, WAITR, MURMR, simultaneusly deleting th spelling problem, clarifying th pronunciation, and shortning th words to save time, enerjy, ink, paper and mony. To th degree that it clarifies th pronunciation and shortns words, Cut Spelling shud therfor benefit readrs. Since it deletes letrs, rathr than substituting othrs, it shud hardly disturb them, since th apearance of th word remains basicly unchanged. Upward givs th analojy of 'shorn lams', which stil remain recognisabl as lambs.

Howevr, when CVC clarification of pronunciation is replaced by consonant clustr in mor radicl deletions such as SPELNG, EVANJLISM, or INSRECTION, wil it benefit or handicap readrs? Or may it both benefit expert readrs by helping them to speed up, and handicap trubld readrs by complicating their decoding and impeding their vocabulry development? This is where reserch is essential befor setting up a final modl.

Th homofone argument against fonological representation for readrs is only a side-issue. It has oftn been argued that a completely fonemic spelling wud not suit readrs of th English languaj becaus there ar so many words that sound th same (homofones) and that these must all be speld difrently to avoid confusion. It is, howevr, esy to point out that 90% of th time homofones ar red in a context that automaticly indicates th sense in which they ar being used - for exampl, in this paragraf, no readr wud hav hesitated over th posibi ambiguity of SIDE, ISSUE, SOUND, MUST, POINT, OUT, TIME, SENSE, BEING, OVER. There ar probably only a few dozen words, if that, which may continue to reqire visual distinction becaus they ar oftn used contiguusly, e.g. TO, TOO, TWO, and FOR, FOUR, FORE. Many current heterografs may prove not realy necessry, including perhaps, from this paragraf, Th/THe, BEEN/BEAN, WUD/WOOD, NOT/KNOT, THER/THIR, SO/SEW/SOW, BEE/BE, ALL/AWL, RED/READ, NO/KNOW.

Visual recognition of words in reading.

It is widely suposed that visual recognition of words enables fastr reading (on analojy with th speed of light, and of seeing) than fonologicl decoding (speed of articulation, or even speed of sound or of hearing), altho there is some evidence that for th skild readr both these processes may take place closer to central brain processes than to actual sight, hearing or speech. Comprehension processes may act as fast on what one may 'hear in one's hed' as on what one can 'see in one's mind's eye', becaus neithr hav externl fysical time constraints. Al visual orthografic features that readrs might use include recognition of orthografic regularitys (i.e. recurring spellings for segments within words; th grater salience of consonants over vowels; th importance of th initial letrs of a word, folloed by th final ones, with medial letrs posibly even ignord; th top half of th line of print conveying mor than th botm half; and idiosyncratic configurations - special configurations that make some words stand out, e.g. GIRAFFE. Fast readrs may be like sports stars in ability to chanje tak and pik th best strategy as they go along.

Th value of none of these stratejys wud be afectd by spelling reform, as soon as readrs had adjustd to th chanjed apearances.

Economy of word length as an advantaj for readrs.

Short words are esyer to read than long words, othr things being eqal. There is anecdotl evidence that Finnish is sloer to read than English despite its fonemic consistency, becaus its averaj word-length is sevn letrs. Howevr, if all words ar reduced to about th same short length, text looks less readable and I think this cud be proven. 'Shorter words make faster reading' does not apply if all th words ar short, despite th copy-editr's misleading title givn to an editd New Scientist articl on deleting surplus letrs for a practicabl spelling reform (Cut Spelling 1, Yule 1982). In continuum text there apears to be an advantaj in a mixture of word lengths. In English, th function words tend to be 1-3 letrs and th content words tend to be longr, which hilights th most important vocabulry, and adds to th variety of visibl distinguishing features, thus creating visual distinctions that ar similr to th Japanese use of complex kanji complementd by simpl sylabic kana.

This probabl advantal of a 'mixd economy' cud be tested. For exampl a design for a pilot experiment compares reading three texts of cognitiv dificulty and similr zoologicl content. One paragraf consists of short words about bears, foxes and birds, th second uses only longer words about hippopotamuses, rhinoceroses and elefants, and a third is constructd with a 'normal mix of word lengths'.

Cut Spelling.

A spelling improvement by deletion of letrs surplus to representation of pronunciation or meaning has benefits for lernrs and riters that hav been pointd out elswhere. For readrs th advantajs ar economy in word length - there is less to process or to skim, and there is less intrusiv clutr for both decoding and automatic word recognition.

A comparison of th value of simpl spellings as against mor complex and less predictabl spellings was undertaken by a VDU lexicl decision experiment, using words that hav mor than one spelling acceptd in dictionarys. 'Press th blak butn if yu imediatly recognise this as a real word, regardless of any misspelling.' Howevr, th findings wer that familiarity was what matrd most for fast recognition of words, not their length or economy. It was th dictionry's preferd spelling, its first listing, that was recognised faster and with most certainty. Nevrthless, three subjects who had had a limitd exposure to 'Cut Spelling' wer mor likely to respond fastr to th simplr of th alternativ spellings - and this clu needs to be folloed up. If replicated, it shows that Cut Spelling principls can be taken on bord very qikly, and jeneralised qikly to othr vocabulry. This wud also demonstrate clearly that as a stratejy to teach a spelling reform, a jeneralisabl principl gets results mor rapidly than simply lists of words.

Readrs' response to spelling chanjes.

Immediat adjustment. A key experiment by Yule & Greentree (1986) compared th speed with which machd grups of subjects red for meaning sevral paragrafs of texts in five difrent spelling modes, with comprehension tests. Difrences between reading in norml spelling and in both versions of Cut Spelling wer not significant, folloed by a control text for Cut Spelling in which othr letrs had been deleted. A morfofonemic spelling on Chomskian principls rankd next, while a fonemic spelling, World English Spelling, was significantly slowr to read than all th othrs.

Substitution of letrs was clearly a mor disruptiv visibl chanje than deletion of 'superfluus letrs'.

Testing 'Cut' spellings for readrs.

A ranje of pilot experiments (Yule 1991) call for replication, since replication is mor convincing in this field than probability statistics. They seek to test readrs' speed and comprehension on first reading in a moderat Cut Spelling. Som of thes experiments wil be publishd in detail, but replication is also desirabl with mor eficient electronic tecnology than was availabl to th riter, to determin which forms of letr deletion ar helpful or disadvantajus for readrs, and th principls that wud explain and predict efects. Their designs wil be briefly outlined.

Subject grups reqired include readrs who ar naiv, novice or practisd in Cut Spellings, and at difring levls of reading ability. Control groups red TO or othr reformd spellings.

Paper and pencil experiments use a series of set texts for difrent typs of operations at varying levls of reading complexity, and in both TO and Cut Spellings. Subjects can be testd on speed and comprehension of reading in each spelling mode, or askd to mark all 'misspellings' that they notice while reading th passajes - when CS spellings such as ACOMODATE pass most subjects' observation, they ar surely fit for oficial jenral use. Or subjects may be askd to underline CS spellings they aprove, or to rerite th passaj in th spelling they wud like to hav.

Tacistoscopic or VDU experiments using TO and CS plus control seudowords ar scord for speed, acuracy, and with note taken of misreadings. Singl words can be presentd for lexicl decision in a Latin square design ('Is this a word - taking no acount of any misspelling'); Frases can be presentd ('Does this make sense? ' e.g. 'HAPPY BIRTHDAY' or 'JUMPING CARPET') and sentences reqiring Yes/No keyd responses to qestions of fact, such as 'AR CHOCOLATS, USUALY GREEN?' In this typ of experiment so far, TO remains significantly superior to CS for naiv subjects, but some typs of letr deletion apear mor imediatly acceptabl than othrs. Th potential of CS is seen in th far grater dificulty subjects hav in recognising words which hav control letr deletions. Some letrs in words apear to be surplus; othr letrs most definitly ar essential.

One experiment explors th posibility that good readrs with very fast word recognition might hav a mentl 'form of the word' that is not necesarily TO, but may follo basic spelling structure without surplus letrs, and so be an objectiv template for Cut Spellings. A qestion here of cors, is that individuals may difr in their mentl 'form of th word' acording to their erliest experiences of th word. This qestion cud be tested using maskd priming tecniqes, which cud hone down what forms wud then be most helpful in print, and take acount of th difring needs of good and poor readrs. A pilot test of this qestion with a limitd sampl of words did not suport this proposition - th ansr apeard to be that visual apearance was stil th major factr in th form of th word for averaj readrs. Howevr, one exelent readr who had experience of CS befor provided support for th hypothesis, and this result reqires replication. If not a chance efect, it cud be determind whethr th finding cud be atributed to superior reading skill or to th CS experience. Maskd priming experiments need to be repeatd with a wider range of words, and include grups of very good readrs, poor readrs, and readrs with prior practice in CS.

Can readrs use analojicl spelling stratejys in adapting to spelling chanjes?

Th ansr is yes, acording to a pilot lexicl decision experimentl series in which readrs who had been testd with one list of words, were next testd with a list of othr words using analogus TO or CS spellings - e.g. LITL/SKITL or READR/TAILR. Findings need replication but indicate that readrs may make rapid adjustment to CS.

Th effects of practice.

Spellings need to be testd not only for readrs' imediat adjustment, in experiments which ar qite simpl to do, but also by mor dificult and expensiv reserch that can explor how much reading practice wud be reqired for subjects to read (or rite) betr in CS than in TO, and what th ceiling of eficiency might be. Litle exposure may be needed after all, as Beech's experiment with his Regular Spelling sujests (Beech 1983). In an experiment in th efects of reading practice conducted by Yule & MacKay (1987), 41 norml and poor readrs read a series of forty texts in an experimentl Cut spelling for an hour each weekday for a month, with a similar control grup of 41 subjects reading in TO. This pilot experiment incurrd some weaknesses thru practicl exijencies - for exampl individuals operated their own stop-watches - but post-experimental findings wer that diferences in final reading rate and comprehension between th control and experimental grups wer generaly not significant. There was an intriguing indication, both objectivly and subjectivly, that some of th poor readrs who had practisd CS by reading th texts with surplus letrs omitd, improved their ability and intrest in reading in norml spelling, wheras th poor readrs in th TO control grup did not. It was unfortunat that circumstances prevented th pland follo-up to test for long-term efects. An experiment like this reqires replication, with follo-up over time.

All such furthr experiments reqire replication that include subjects who hav had prior experience in CS. Findings shud be compared acording to subjects' degree of experience and practice and prior levl of reading skil, and there shud be some comparisons with othr typs of posibl reform - e.g. New Spelling 90.

Testing readrs' comprehension.

Gillooly (1976) and othr critics of fonemic spelling hav claimd that it wud force readrs to use fonemic stratejys permanently in reading, and be forced to 'bark at print' without 'reading directly for meaning'. This is manifestly untru, becaus readrs of a fonemic spelling hav th dubl advantaj of being able to decode new words as wel as th acqiring automatic recognition of familiar vocabulary. Most of th grat readrs of the past few hundred years wer taut by fonic methods, but this did not hold them bak from developing speedy reading stratejys.

Howevr, it is qite posibl that th nature of a spelling system may make it esyr or hardr to get th meaning qicly. Japanese claim that their ideographs make reading for meaning fastr and esyr than a script which is a linear sequence representing sounds - e.g. a train travelr can read th name of an aproaching station soonr in its kanji version than from th acompanying linear hirogana or roman alfabetic notices.

Research on reading a reformd spelling must test comprehension of meaning in oral and silent reading. Now most comprehension tests consist of esily markd multiple-answer qestions, allowing up to 50% gessing, or qestions supply a structure from which gessing is posibl, e.g. 'What colour were the tentpegs?' 'Why did the men want to get to Katmandu?' Mor valid tests of comprehension take longr to mark, but giv mor insight into what sort of purposful comprehension is going on. One test consists of open-ended qestions to be answered in riting if posibl oraly if not, with instructions such as 'What is the gist of this paragraph? List three facts or ideas from this paragraph.'

This sounds simpl when the paragraf is only around 150 words and content is simpl. Th test does not even force th readrs to focus on qestions th examinr might focus on. It tests th readrs' own personal taks. Yet my experience with this typ of comprehension qestion shows that even undrgraduats ar distressingly liabl to errors and misinterpretations in what they read, because they do not read acurately. This is a serius problem in practicl afairs - and we see th disastrus efects evry day in th way even journalists misinterpret simpl facts thru 'careless reading'. Much reading by weak readers is just 'going thru th motions'. To what extent may TO be to blame for their error-condoning reading stratejys?

Anothr feature to be investigated is why many peple, regardless of whole-word, fonics or whole-language training, cannot answer such comprehension qestions aftr first reading, even tho th paragraf may hav been very short and simpl, and readers kno in advance that they wil be askd for three ideas or facts from it. These readrs have to read th paragraf twice, as if once to get th words and again to get th consecutiv thought. (Even expert readrs of cours hav to reread when th thought or expression is complex.) Cud a reformd spelling help these readers to get as far as th meaning on th first time around?

Spelling reform for all categorys of reader.

Any spelling reform shud not ignor th fact that th bulk of th English speaking population and a hi proportion of international readership wd nevr hav th linguistic capacity to be super-readrs in English. A spelling reform that does not cater for their needs is missing its point, howevr theoreticly perfect it may be. It is particulrly desirabl that our present social subclass of iliterats and semiliterats shud be able to read esily, and so hav mor chance of being a prosperus part of th whole society. It is posibl that a two-levl spelling system may be necesary, as in Israel or even Greece, to cater for th diference between those who can read very wel because of their hi verbal intelijence and visual memory for print, and those whose functioning intelijence cud be gratly promoted by their reading, but who ar not above th average in basic capacity. It may prove desirable for example, to print tertiary-levl texts in economicl CS if it promotes th fastest procesng, and material for th general public in mor moderat CS if that givs a mor imediat key to th spoken languaj.

Th 'average reader' and th 'backward reader' and th 'dyslexic' may still retain to some extent caracteristics of learnrs - constantly needing to decode vocabulary if its apearance and even meaning ar unfamiliar thru novelty or their own poor visual memory. Reserch must determin whethr th dificulty they hav in working out consonant clusters in TO wud also apply to CS, in words such as PERMNNT and HORR. Th qestion of whethr consonant clustrs wud continue to be a base in automatic recognition of familiar words also needs to be setld. Experience with pijin spellings sujest that vowel interspersion aids readability for averaj readrs, e.g. TABEL is easyr than TABLE. It may be that aclimatisation may be easyr for consonant clusters in very comn words, just as irregulr spellings caus least dificulty in everyday reading in th most comon words. Longr and less familiar words may stil require a crutch - to be less intimidating as wel as intrinsicly casyr to decode.


Research can attempt to improve th spelling system to benefit readrs. A reformd spelling that benefits learnrs and riters wil not sacrifice its advantages for them if it is givn consistent modifications in ordr to maximise th advantages of readers. A reformd spelling shud enable readrs to utilise all existing reading stratejys for grater acuracy, speed and betr understanding in reading than is jeneraly developed at present with TO. Reading can be transformd from a 'psycolinguistic gessing game' to an 'eficient psycolinguistic process' that wud be even mor fun, because th readr nevr loses th game. As th peple ar able to read mor, and therefor kno mor, and reason mor, th functioning inteligence of th population is effectively improved. And they can develop tastes that extend beyond th present levl of our mass-readership tabloids.

Copies of th author's experimental designs and materials mentiond in this articl ar available to replicate experiments.


Beech, John. 1983. The effects of spelling change on the adult reader. Spelling Progress Bulletin. 23, 1. 11-18.

Chomsky, Noam. 1970. Phonology and reading. In H. Levin & J. P. Williams, (Eds.) Basic Studies on Reading. NY: Basic Books.

Downing, John. & Leong. C-K 1982. The psychology of reading. NY: Macmillan.

Frith, Uta. 1978. From print to meaning and from print to sound, or how to read without knowing how to spell. Visible Language 12: 43-54. Chapter 5.

Frith, Uta. 1979. Reading by eye and writing by ear. In P. Kolers, M. Wrolstad & H. Bouma (Eds). The processing of visible language. Vol 1. NY: Plenum Press.

Goodman Kenneth. S. 1982. Language and Literacy: the selected writings of Kenneth S. Goodman. Vol II. Reading, Language and the Classroom Teacher. F. V. Gollasch (Ed.) London: Routledge & Kegan Paul.

Gray, William S. 1956. The teaching of reading and writing: an international survey. Paris: UNESCO.

McArthur, Tom. 1987. The English languages? English Today. 11.9-13.

Read, Kingsley. 1974. Personal communication. And see 'Devising the Shaw Script'. Spelling Progress Bulletin. 6.2.11. 1966.

Smith, Frank. 1971, 1972, 1982. Understanding Reading. 3rd edition. NY: Holt Rinehart & Winston.

Simplified Spelling Society. 1992. Personal View documents.

Urdang, Laurence. 1990. On observing World English. English Today. 46.1. 11-16.

Yule, Valerie. 1982. Spelling as technology. Rewritten by John Bell and published as 'Shorter words mean faster reading! in New Scientist 96 1335. 356-7.

Yule, Valerie. & Greentree, Stephen. 1986. Readers' adaptation to spelling change. Human Learning 5: 229-241.

Yule, Valerie. 1986. The design of spelling to match needs and abilities, Harvard Educational Review, 56: 278-297.

Yule, Valerie & MacKay, Ken. 1987. Practice effects for adults and poor readers in reading text in a modified spelling. (Unpublished manuscript).

Yule, Valerie. 1991. Orthografy and Reading. Spelling and Society. In Dissertations Abstracts., publication number 9231850.


[1] A futur articl on spelling reform for riters discounts th claim that 'spelling as yu speak' wud be betr for them than standrd conventions.

[2] There may be a limit to how fast even these most expert readrs can read for comprehension, regardless of riting system. There may be a ceiling for how much information th human nervus system can take in qikly, and there ar also grat individual difrences in this capacity. Howevr, this factr of bilt-in limitation is one that shud be taken acount of by th riters seeking to comunicate, rather than by any atempt to restrict th structure of th riting system itself. Remember that th QWERTY keybord was deliberatly designed to slow down th typist so that th keys of a typriter wud not jam - and now wen this problem is obsolete, we ar stil stuck with QWERTY.

[3] Here we again meet th usual asumption that th only reform posibl for English spelling is pure fonemic.

[4] Nevertheless, spelling reformrs shud be careful about assuming that we all share a comn English languaj that is represented in their own proposals. Their own eforts usualy identify their own linguistic origjins, wich is usualy RP for th Simplified Spelling. Society, but, for exampl, Leeds for Reg Dean's Britic, and Australian for anything I rite myself.

[5] My personal preference wud be CS1 with a vowel system that alowd two alternativ spellings for currently digrafd medial vowels, particulrly in short words, and a further one for final vowels in longr words in texts for lerners and weaker readers.

Back to the top.