[Journal of the Simplified Spelling Society, 28, 2000/2, pp12-15]
[See J28 article by John Downing, items about ita and further articles by Chris Upward.]

John Downing's i.t.a. Evaluation.

Chris Upward.

John Downing (1967) Evaluating the Initial Teaching Alphabet: a Study of the Influence of English Orthography on Learning to Read and Write. London: Cassell & Company Ltd, 327pp. Chris Upward here sumrizes in Cut Spelng John Downings report on th i.t.a. experimnt wich he directd. It represents th most importnt and extensiv reserch evr conductd into th practicl efects on litracy aquisition of simplifying english spelng. Previus items on th i.t.a. (beside John Downings articl on pp3-11 of this issue) that hav apeard in JSSS ar Threadgall R (1988) The Initial Teaching Alphabet: proven efficiency and future prospects. JSSS J7, 1988/1, pp18-19, and Threadgall R (1993) Remedial Education and a Consistent Alphabet. JSSS J15 1993/2, pp6-7.

1. Th problm: is th traditionl orthografy of english an importnt cause of dificlty in lernng to read?

Chaptr 1 discusses difrnt vews of th iregularity of english spelng. Som emfasize th inadequacy of th roman alfabet and propose extra letrs, eg, to replace digrafs (singl letrs insted of CH, SH, TH, etc) or for long vowls (especialy replacing 'majic E'). Othrs emfasize th many ways of spelng a givn sound and propose for instnce givng th same spelng to th vowls in vain, vane, vein. Othrs again wud remove redundnt letrs from th alfabet (eg, rite K, KW, KS insted of C, Q, X); or replace blok capitls by enlarjd versions of th smal forms, so danger wud be capitlized as danger, not DANGER. Like its mid-19th century ancestr Fonotypy, th i.t.a. was not desynd as a permnnt replacemnt for traditionl orthografy (t.o.), but as a transitionl initial lernng sceme.

Downing comnts that these ideas ar based on lojicl expectation, not on reserch evidnce of childrens dificlties; nor do they tel us wich problms cause most dificlty, and wich cause least.

2. A revew of previus investigations.

Chaptr 2 first revews reserch into th readng process carried out befor th mid-1960s. Som influential reserch asumed that, because skild adult readrs dont normly spel words out as they read, beginrs dont need to eithr, and that they therfor dont need to lern th sequence or sound values of letrs in words. One emnnt reserchr beleved children undr th aje of 7 incapabl of graspng th fonic principl. Such ideas implyd that spelng reform was unecesry. Downing furthr notes th lak of comparativ reserch into litracy aquisition in difrnt languajs (but see Downing, 1973.-cu). Experimnts purportng to sho that iregulr spelngs ar scarcely hardr to read than regulr spelngs ar refuted.

Th successful use of fonemic spelng systms in sevrl places in th northeastrn USA in th late 19th century is described, tho not al th details ar recordd (did they use Isaac Pitmans Fonotypy, or regulrized roman?). Furthr experimnts conductd in th UK in th erly 20th century, one using fonetic symbls and othrs using th SSSs Nue Speling, ar also described. Two side-efects of these scemes wer improved pronunciation and improved educationl and intlectul potential jenrly.

Downing finds that al these experimnts wer insuficiently controld and lakd objectiv testng - weaknesses he pland to avoid in evaluating th i.t.a.

3. Th Initial Teaching Alphabet: bakground and description.

Th idea for th i.t.a. experimnt orijnated in th House of Commons. In 1949 James Pitman, MP (Conservative, Bath), secndd th Private Members' Bill presentd by Mont Follick, MP (Labour, Loughborough), wich was narroly defeatd. Anothr oprtunity arose in 1953, wen they presentd a Bill calng for investigation of th litracy gains that myt result from scools using simplifyd spelng. Tho oposed by th Govrnmnt, this Bill was passd by 65 votes to 53. Th eventul outcom was Govrnmnt aproval for reserch into th efect of iregulr spelng on litracy standrds.

Pitman and Follick disagreed on th aim of th simplifyd spelng. Follick was intrestd primarily in a spelng reform to benefit non-nativ speakrs (he ownd a chain of languaj scools), wile Pitman wantd to test a transitionl orthografy for th benefits it myt ofr beginng readrs. Downing notes that reserch myt hav shown ther was not necesrily a conflict between th two aims. But th disagreemnt was also persnl, as both Follick and Pitman wer keen to promote ther own scemes. Th dispute endd wen Follick died in 1958.

Downing comnts that th i.t.a. was in th end efectivly imposed by Pitmans persnality, influence and resorces. It was desynd as an amalgm of Fonotypy, th alfabet devised by his granfathr Isaac Pitman, and th SSSs New Spelling. Th i.t.a. was not based on sientific reserch into wat systm myt be most efectiv, nor on a reasnd choice of alternativs (Follicks sceme fel out of contention with its authrs deth, and Axel Wijks Regularized English was rejectd despite his passionat advocacy). Th i.t.a. was a clevrly desynd compromise between ful regularity and suficient simlarity to t.o. to facilitate transfer; but it did not necesrily represent th best of al posbilities.

To launch th i.t.a. in 1960 Pitman secured suport from th Ministry of Education, th Local Education Authorities Association, and th National Union of Teachers. Th reserch was to be conductd undr th auspices of th Institute of Education at London University and th National Foundation for Educational Research thru a Reading Research Unit atachd to th Institutes Department of Educational Psychology. Copyryt was renounced for th i.t.a., giving anyone th ryt to print and copy texts using th i.t.a. Th SSS meanwile distnced itself from th hole entrprise.

Pitman did not link th i.t.a. to a specific teachng method, such as look-and-say or fonics. His desyn was to aid lernrs transfer to t.o. by keepng som featurs of t.o., such as dubld letrs, T befor CH, th alternativs C/CK/K, vowl letrs representng shwa (or even comnly silent, such as th secnd E in every), and two or mor spelngs for as many as ten sounds. Furthrmor, som overlaps with t.o. cud ocur, as with i.t.a. union for t.o. onion. I.t.a. was desynd to represent forml platform speech, with som scotish overtones.

Pitman aimd to asist transfer by ensuring simlarity in th jenrl apearance ('configuration') of words in i.t.a. and t.o., especialy ther 'top coastline', ie, th sequence of letrs with and without asendrs. On th othr hand, th i.t.a. did reduce th numbr of alternativ spelngs for sounds, it replaced digrafs by ligaturs, and it removed 'majic E' (eg, hate became hæt, hope became hœp).

4. The first i.t.a. experimnt: aims, desyn and method.

Downing set great stor by sientific method (th i.t.a. was an 'experimnt'), and proposed nine 'hypotheses', to be proved or othrwise. Th hypotheses stated that, compared with children aquiring ther litracy skils thru t.o., children lernng thru i.t.a. shud:

1) lern fastr (since i.t.a. involvd less lernng)
2) read with fewr errs (since i.t.a. was mor regulr)
3) perform no worse aftr transfer to t.o. than previusly in i.t.a.
4) read t.o. with greatr acuracy, fluency and undrstandng
5) rite mor
6) hav a larjr vocablry
7) spel betr
8) solv problms unconectd with litracy betr
9) hav hyr atainmnts in mathmatics.

Overal, th purpos of th experimnt was to investigate th efect of t.o. on th process of lernng to read.

Paralel groups of children aquiring ther initial litracy using i.t.a. (th experimentl groups) and t.o. (th control groups) wer set up. Th two sets of groups wud be closely machd by varius criteria: they wud use th same readng books (th Janet and John series in t.o. and transcribed into i.t.a.); scool conditions wud be simlr; and evry efrt wud be made to prevent any Hawthorne efect distortng th results (ie, th control groups wud get as much special atention as th experimentl groups). Even so, a perfect mach wud not be posbl.

Wen scools wer invited to participate, in th first year (1961-62) too few enrold (only 20 experimentl groups and 33 control groups); but by 1963 th experimentl groups numbrd 87 (by 1966 nearly 1,800 scools in England wer using i.t.a.). Anothr erly dificlty was th insuficiency of i.t.a. readng material. Th teachrs, wethr teachng experimentl or control groups, wer to use ther norml teachng methods, wich in most cases ment beginng with 'look-and-say'; only about one scool in twenty taut fonics first. Th timing of transfer from i.t.a. to t.o. for each child was left to th teachrs. (Tho beginng with look-and-say, som i.t.a. children spontaneusly developd fonic skils, so improving ther test scors.) Parents wer fuly informd of th experimnt in advance.

Mesurmnt of performnce in both th experimentl and control groups was based on varius factrs. Progress was mesurd by how many books in th basal readr series each child completed by certn dates. Th Schonell Graded Word Readng Test mesurd childrens ability to read singl words. Readng acuracy, fluency and comprehension wer mesurd by th Neale Analysis of Reading Ability. Th amount of riting produced in a week was quantifyd, and th numbr of advanced words used was taken as a mesur of vocablry. Othr tests used wer th Standish test of comprehension from silent readng and th Schonell Graded Word Spelling Tests.

5. Results of th first i.t.a. experimnt.

Chaptr 5 first describes th care with wich th reserch was conductd. Cheks wer made for a Hawthorne efect. Scools wer machd by orgnizationl and sociolojicl criteria, with many scools rejectd for failur to meet th criteria. Wher machng was less than perfect, th reserchrs ensured any advantaj normly lay with th t.o. and not th i.t.a. groups. Th numbr of children testd totald 660 each for th i.t.a. groups and th t.o. groups, thru reduced numbrs and subsets for som of th tests, numbrng 459, 433, down to 152 for i.t.a. and t.o.

Th i.t.a. lernrs made much fastr progress thru th basal readrs than did th t.o. lernrs: aftr 21/2 years 78% of th i.t.a. lernrs had passd beyond Book 5, but only 39% of th control group children had don so. This satisfyd Hypothesis 1 (se §4 abov).

The Schonell word recognition test produced an avraj scor of 18% in i.t.a. but only 7% in t.o. aftr one year, and aftr 11/2 years 34% and 15%. In th Neale test, th scors for acuracy, speed and comprehension wer respectivly 25, 26 and 7 for i.t.a., against 14, 25 and 5 for t.o. This satisfyd Hypothesis 2.

Howevr, Hypothesis 3 (i.t.a. lernrs shud perform no worse wen testd in t.o. than in i.t.a.) had to be rejectd. Even if th i.t.a. lernrs had transferd to t.o. as long as four months befor th test, ther performnce was stil worse than it had been in i.t.a. (not til th end of th third year was this setbak overcom). Th i.t.a. lernrs wer nevrthless superir to ther t.o. equivlnts by th great majority of tests, notwithstandng th setbak.

Th Schonell readng test (first 50 words of th ful 100) identifyd th most err-prone spelngs. Since th words wer listd in ordr of dificlty, th numbr of errs grew stedily thru th test; but som words stood out as causing mor errs than expectd. Such words, with ther position in th list, wer: 6 school, 12 road, 15 light, 19 people, 21 dream, 28 beginning, 29 postage, 30 island, 33 ceiling, 36 canary, 47 orchestra, 48 knowledge. Of these 12 words, 9 contain long vowls difrntly spelt in i.t.a., and at least 6 contain redundnt letrs. Downing noted that it was fonicly misleadng letrs rathr than th chanjed apearnce ('configuration', 'top coastline') of words that caused most dificlty.

Hypotheses 5, 6 and 7 wer satisfyd: th i.t.a. lernrs rote nearly 50% mor than th t.o. lernrs, they had a wider vocablry, and they spelt mor acuratly in t.o., tho here ther advantaj only became significnt in ther 4th year of scoolng. Th i.t.a. lernrs did not outperform th t.o. lernrs in jenrl problm-solvng (Hypothesis 8), and in mathmatics only marjnly (Hypothesis 9).

Th tests also showd wethr lernrs of hy, avraj or lo achevemnt particulrly benefitd from th i.t.a. Across th ranje of tests, th ablest lernrs benefitd soonst and most, th midl categry gaind less dramaticly, and th least able only benefitd aftr som delay. (Downing notes in §9.4 of his articl abov [p9] that th valu of th i.t.a. for lo ability lernrs may hav been undrstated.)

6. Discussion of th results of th first i.t.a. experimnt.

Chaptr 6 first restates th two main aims of th reserch: to find how big a problm t.o. represents for lernrs, and wat dificlties arise at transfer from i.t.a. to t.o.

Th reserch had methodlojicl limitations, concernng variabls (same books used, but teachrs and teachng methods varied), th tests (do Schonell and Neale basicly test th same thing?), and statisticl analysis. Experimentl groups wer testd in 33 scools, control groups in anothr 33, and paralel i.t.a./t.o. classes in 8 scools. Perhaps th results wer distortd by using volunteer teachrs. Perhaps i.t.a. improved th skils of th teachrs. Great care was taken to avoid posbl causes of bias in favor of i.t.a., and syns of a Hawthorne efect wer sot, but non found. If anything, th i.t.a. side of th experimnt was at a disadvantaj (eg, in som cases shortaj of books, yungr lernrs, worse scool bildngs).

Othr mesurs of achevemnt wer as follos: by Term 5 th i.t.a. lernrs wer on avraj on Book 4 of th basal readrs, th t.o. lernrs stil on Book 2; by th Schonell test, i.t.a. lernrs then recognized 21/2 times as many words as t.o. lernrs. In fre composition, i.t.a. removed th limitation of t.o. that children cud only use words hos spelng they new or cud ask, and th i.t.a. children consequently rote mor and used a wider vocablry. Th greatr dificlty of t.o. is seen in th difrnt spelng of th vowl in I, my, night, like, buy, climb, eye, die, wich in i.t.a. is spelt with ligaturd ie in evry case.

Th most able lernrs gaind most from i.t.a., tho that did not mean that weakr pupils did not also gain (even wen ther gains wer minml, teachrs comntd how much mor purposfuly and independntly th least able workd in i.t.a.). A jenrl lesn was that less able lernrs needd mor time to mastr th i.t.a. befor transfer to t.o.

Th results wer confirmd from smalr experimnts by independnt reserchrs. Sister Johns experimnt (se §6 of John Downings article on p7 abov) was especialy striking, shoing how th cognitiv powrs of i.t.a. lernrs gaind mor widely than just in terms of litracy.

Regardng 'readng rediness', this was not only a matr of syclojicl rediness, but also dependd on externl factrs, one being th riting systm. It may be that children ar redy to aquire litracy skils erlir using a mor regulr riting systm. In terms of lernng sycolojy, th success of i.t.a. is du to th reduced 'cognitiv confusion' pupils experience, compared with t.o.

Th advantajs of th i.t.a. wer clearly demnstrated; but that did mean they wer du specificly to th i.t.a. caractrs. It may be that any regulrized riting systm wud hav givn equal benefits, and a systm that had been betr desynd myt hav givn even greatr benefits than th i.t.a.

Infnts teachrs felt th i.t.a. was worthwile for th imediat benefits lernrs enjoyd, regardless of longterm benefits for ther subsequent performnce in t.o. Teachrs ofn thot pupils made th transfer scarcely noticing th chanje, but th tests showd they sufrd a real setbak that lastd for som time aftr. Yet by most tests they stil outperformd th t.o. lernrs. Furthrmor, th setbak was not permnnt, being overcom by th end of th third year.

Regardng James Pitmans aim to keep a degree of simlarity between i.t.a. and t.o. spelngs, som i.t.a. forms wer identicl in t.o. (eg, milk) and othrs wer very close (eg, tree), but othrs involvd considrbl chanjes. On transfer lernrs at first misred many long vowls, as in road, train, people, dream, shoing that fonic iregularity posed a bigr problm at transfer than chanjed configuration. Particulr problms wer reveald in th tests from interference as between i.t.a. shoe and t.o. show, from th silent letrs in school, island, gnome, and from A in enabled, safety, wandered, AI in regain, C in ceiling, centre, G in imagine, I in final, IE in brief, IGH in frightened, OR in work (confused with walk), OW in now, PH in nephew, T in action. Many errs wer also made with age, attractive, canary, fruit, huge, magician, these, whom. Downing beleved much more reserch was needd to desyn a systm for optmm transfer. Transfer may also hav been mor dificlt because teachrs lakd experience and had not yet found th most efectiv methods; and they may hav been temtd to transfer too erly, rathr than wait til th children wer fluent in i.t.a.

Readng ajes by th end of th third year wer 8 years 9 months for i.t.a. lernrs, and 8 years 4 months for t.o. lernrs. Aftr transfer, al ability levls of i.t.a. lernrs improved fastr than th t.o. lernrs, but abler lernrs fastr than th less able. A relativ weakness of th i.t.a. groups was th tendncy not to dubl consnnts in t.o. - perhaps because ther fonic skils gave no reasn to do so.

7. A secnd i.t.a. experimnt to provide mor rigrus control over certn variabls.

A secnd i.t.a. experimnt was begun in 1963 to examn abov al th variabl of teachr skil and persnality, and also th efect of classroom visitrs and externl publicity.

For th secnd experimnt two paralel classes wer set up in 16 scools, with th same teachr responsbl for litracy teachng in both, using i.t.a. for one class and t.o. for th othr. Thus any variabl arising from teachr persnality wud be removed. Both th i.t.a. groups and th t.o. groups containd som 550 children in al.

This aranjemnt created varius dificlties, such as unequal time spent in th two classes, difrnt teachng patrns, difrnt expectations, difrnt teachr prefrnces (most teachrs preferd using i.t.a.). Sevrl scools faild to maintain th conditions required for th thre years, and wer excluded from th experimnt. So wile th variabl of teachr persnality was controld, som othrs wer not.

Th same tests wer aplyd as in th first experimnt, but th results wer rathr difrnt. Altho most i.t.a. groups made fastr progress initialy, ther advantaj was ofn not significnt. In th t.o. tests taken in th midl of th secnd year th i.t.a. lernrs performd slytly less wel than th t.o. lernrs, and tho they regaind ther initial advantaj at th end of th secnd year, it was less markd than in th first experimnt, and was reversed at th end of th third year. Not even th most able i.t.a. lernrs in th end significntly outshon ther t.o. countrparts, as they had so strikingly don in th first experimnt. Th secnd experimnt nevrthless confirmd th two main findngs of th first experimnt: litracy skils ar mor esily aquired in i.t.a. than in t.o., but th transfer from i.t.a. to t.o. sets lernrs progress bak.

Th Schonell reading tests sujestd th same conclusion as in th first experimnt: Pitmans idea that a words jenrl configuration wud determn how esily lernrs transferng from i.t.a. wud be able to read its t.o. form was not validated; on th contry, dificlties wer caused far mor by difrnces of sound-symbl corespondnce. Beside th words alredy found to be hard to recognize in th first experimnt, th folloing wer noted in th secnd experimnt: something, biscuit, crowd, angel.

So wy did th secnd experimnt not demnstrate th gains made from using i.t.a. so decisivly as th first experimnt? First, al th scors, in both i.t.a. and t.o., tendd to be loer in th secnd experimnt. No externl factr (eg, greatr social problms) was found to acount for this; but it was suspectd th teachrs wer teachng at a mor relaxd pace. Certnly th date of transfer from i.t.a. to t.o. (decided at th discretion of each teachr) tendd to be significntly later in th secnd experimnt, so that fewr i.t.a. lernrs wer familir with t.o. wen they first took th t.o. tests. Th report on th first experimnt had surmised that a later date of transfer to t.o., wen mor pupils wer fluent in i.t.a., myt help; but th secnd experimnt sujestd th oposit. Posbly ther was a negativ efect if th habits of readng and riting i.t.a. became too ingraind befor lernrs confrontd t.o. A furthr factr that may hav adversly afectd performnce in th secnd experimnt was that teachrs disliked being split between two classes, using i.t.a. with one and t.o. with th othr.

8. Conclusions and recmendations.

Conclusions
Th first i.t.a. experimnt showd how lernrs benefitd from i.t.a. Th secnd experimnt, investigation. Most tests showd transfer to t.o. causing a setbak (no procedurs for minmizing it wer availbl), and slo lernrs needd perhaps to continu with i.t.a. beyond 3 years. One dificlty of i.t.a. for beginrs was that 'b/d' wer at first mirr-imajs, as in t.o.; but this was soon rectifyd by givng 'd' a short tail. Th tests showd t.o. as a serius cause of dificlty for lernrs, impeding readng, riting and vocablry developmnt, and that a simplifyd, regulrized orthografy reduces cognitiv confusion by clarifying sound-symbl corespondnces.

Recmendations.
Of th 10 recmendations, No.8 is by far th most detaild.

1) Th i.t.a. shud be mor widely used, but teachrs need trainng, and th i.t.a. shud be improved.
2) Transfer to t.o. shud not be rushd, especialy for weakr lernrs.
3) Th best time for transfer needs to be reserchd.
4) Th first i.t.a. books wer transcribed from t.o. New i.t.a. books using wider vocablry ar needd.
5) Methods of teachng and of transfer shud be reserchd.
6) Th trainng of litracy teachrs, especialy in i.t.a., needs investigating. They may lak th requisit linguistic and syclojicl nolej.
7) New litracy tests shud be developd specific to i.t.a., intrnationly if posbl.

8) Th i.t.a. shud be seen not as an ideal, but as exemplifying a 'simplified and regularized writing system (s.r.w.s.).' Reserch and desyn of an improved s.r.w.s. can take place in an experimentl sycolojy labratry, without extensiv fieldwork like th i.t.a. experimnt. Ther is a conflict between th ideal s.r.w.s and th demands of transfer and implmntation. Transfer is esiest if th s.r.w.s. is close to t.o.; and closeness to t.o. makes it mor acceptbl to teachrs, parents, publishrs and printrs, and to othr english-speakng cuntries, including those in Africa and Asia.

Th i.t.a. experimnt reveald weaknesses in its desyn. It is not as simpl as it cud be (havng C, K, CK for th same sound is confusing), and its special caractrs put peple off. It wud do wel to abandn th letr C, dubld consnnts, reverse Z (zess) and som alternativ spelngs (teachrs wer disapointd that i.t.a. did not hav one-to-one sound-symbl corespondnce). Ther is no evidnce those complications aid transfer, and they confuse both lernrs and teachrs. Th special caractrs ar poorly desynd for handriting. Difrnt colors or diacritics cud betr distinguish th two values of TH or OO. Simpl ligaturs for th digrafs CH, SH, TH, NG, EE, AU, OE, insted of th special i.t.a. caractrs, wud carry over directly into cursiv handriting. Th caractr 'æ' was hard to rite (ligaturd 'ai' or 'ay' wud be esir). Ther is no evidnce that any of th special caractrs was beneficial.


Without th special caractrs, th grafemes cud be just th comnst t.o. represntations for each sound; and transfer wud mean lernng th alternativ spelngs used for those sounds in t.o. Transfer cud then also be gradul, perhaps teachng th alternativ spelngs for one sound at a time insted of confrontng th hole of t.o. at once. An s.r.w.s. that used roman letrs wud by defnition be closer to t.o., it wud look less stranje, it wudnt require a new typ-face. Th results of labratry experimnts cud be fed into th teachng process by regulr buletns. An intrnationl comitee wud make recmendations to educators and publishrs in difrnt cuntries. If th s.r.w.s wer primarily for initial litracy, th comitee cud be domnated by educators; if it wer for permnnt spelng reform, wider intrests wud need to be representd.

9) Th orijnl british group of i.t.a. lernrs shud be folloed up over a period of 10 years (not just 4 years, as in th orijnl experimnt), to discovr longr term efects. Th i.t.a. soon came to be widely used in th USA, Canada, Australia and elsewher and those cuntries cud benefit from th exprtise developd by th Reading Research Unit at London University in th corse of th experimnt. (A footnote records that for lak of finance this Unit had to be closed down befor th evaluation was publishd, making any furthr reserch problmatic.)

10) Th chief findngs of th i.t.a. experimnt (t.o. creates great dificlties for litracy aquisition; transfer to t.o. from i.t.a. causes a setbak) both lend suport to cals for reserch into spelng reform.

Reference.

ed. Downing, J (1973) Comparative Reading, Cross-National Studies of Behavior and Processes in Reading and Writing, New York: The Macmillan Company.

Back to the top.