[Journal of the Simplified Spelling Society, J20, 1996/1 pp12,13]
[Valerie Yule: see Book, Journals, Newsletters, Media, Anthology and Bulletins, Personal View 10 and web.]

Spelling Needs Reserch
and Reserch Needs Replication.

Valerie Yule.

Valerie Yule is a reserch psycologist, formerly clinicl child psycologist in hospitals and scools, lecturer, and reserchr in literacy and in the development of imagination, also author of Orthografy and Reading: Spelling and Society, an investigation of writing systems and of the feasibility of spelling chanjes in English.


Spelling reforms need reserch into how they meet the needs and abilities of users and lernrs. Without this assurance the most theoreticly impressiv scemes may prove useless. Argument is not enuf.

Reserch must also be replicated and extended to ensure that findings apply widely, and not only to small sampls. The problems and importance of these requirements ar ilustrated by the experience of the riter's own reserch into readrs' responses to surplus-cut spelling modifications.

1. The need to repeat experiments to ensure findings ar reliabl.

Statistics ar almost idolised in reserch in the behavioral siences. This is because findings from small sampls of peple ar then jeneralised to whole populations by calculating the probability that they are not simply chance results. Such inferential statistics ar cheapr than repeating experiments to make sure the findings stand up to repetition. Howevr, in human afairs the variabls ar oftn so complex and uncertain that replication is far mor convincing evidence than any singl study. This is the issue I woud like to call atention to now. I hav done what I can in arguing for reserch rather than argument in spelling reform, and in piloting experiments in reformd spellings and in testing responses to modified spellings. We now need replication and extension of existing reserch.

2. Problems of repeating experiments.

The problem is not just expense. A greatr problem is that publishing new reserch that makes new findings is mor important for reserchrs, to hav mor chance of publication, and indeed, of academic employment and promotion. There is litl kudos or reward for those who test the reserch of others by replication - unless it explodes some great myth of our time, and oftn not even then. For exampl, Milgram found that his sampls of American students woud obey experimentrs' ordrs to giv electric shoks to human victims. Australian reserchrs found that Australian student subjects refused do what their experimentrs askd, if it seemd to involv hurting others. But Milgram's gloomy conclusions ar stil reportd as the truth about human nature.

3. Recomendations for reserch jurnals.

I woud like to see reserch jurnals include, as regular features, notes about replications of publishd reserch - whether it has replicated wel, or not replicated at all. This woud hav four purposes:
i) to publicise the present status of erlier publishd findings
ii) to recognise and encuraj those necessary peple, replicaters - and so encuraj mor replicaters.
iii) It woud also be fruitful for theory and practice, since it coud point out the criticl conditions undr which a finding woud apply or not.
iv) It woud enable the orijinl experimentrs to say if they had been taken up rongly or misrepresentd, particulrly, as is not unknown, when those misrepresentations hav been made in the wider media and no public rebutl has been allowd.

4. Replication of reserch into surplus-cut spelling.

I woud like to support this call with ilustrations from my own long-term work on readrs' responses to spelling chanjes.

As far as I kno, no one has atemptd to replicate any of my pilot reserch on readrs' response to modified spellings, altho the materials for them hav been ofrd for some years. Some of my experiments hav been of necessity no mor than explorations, so that only speculations coud be made as to their implications. Others, eg, experiments in responses to tachistoscopic presentations of singl words, hav replicated with so many difrent word lists and typs of subjects as wel as having statisticl support, that I am fairly confident of the conclusions drawn from them - that in those conditions readrs ar not significantly disadvantajd on first meeting chanjed spellings of words.

5. Adaptation to spelling chanje with practice - criticly important reserch.

Howevr, the typ of reserch that it is most necessary to replicate concerns readers' adaptation to modified spellings in text with practice over time. Initial modifications in spelling reform need to be those that present readrs can acsept most esily, becaus they wil normaly be th arbitrs of chanje in a society alredy literat. John Beech (1983) made a study of how 13 subjects adaptd to reading a 'Regular Spelling' of his own design. In the 1980s a coleag and I carried out an unpublishd study of 94 subjects' adaptation to reading in surplus-cut spelling over a series of up to 20 sessions. In the event, this reserch was fataly handicapd by a combination of misfortunes that preventd its propr completion and pland follow-up. There had also been dificulties during the experiment in obtaining reliabl and completed responses from some of the 'poor readr' subjects in some of the hour-long sessions of reading practice when co-supervisors had not been availabl. Howevr, the experiment was able to sho no significant difrences between group post-test scors reading in the modified spelling and initial scors in conventionl spelling, and there was the intriguing finding that is worth pursuing, that poor readrs in the experimentl condition appeard to hav improved in reading in conventionl spelling, as wel as making subjectiv reports of greatr enjoyment of reading. Some aspects stil need further investigation.

Since then I hav carried out a further series of experiments and pilot experiments in Australia (see reports in the JSSS). As I now no longr hav access facilities to continue these, I am returning to look at individual difrences in the data, testing ocasionl subjects individualy on some of the paper and pencil tests, and looking at raw data in the longitudinl study that I did not previusly analyse because too much was missing. In the final fully supervised sessions, for which complete data wer availabl, experimentl subjects' reading rates in 'surplus-cut' spelling wer not significantly difrent from those of control subjects and their own initial rates in conventionl spelling. However, the data for all eighteen practice sessions ar not complete. Some protocols hav missing sections, or ar scrappy because some subjects showd exessiv variability in response, while others did not fill in complete responses for every text they read. But group findings for the subjects reading in modified spelling during many of the practice sessions wer that quite oftn they read mor slowly than the control group. Was this contrast of reading rate with the post-practice test a factor of the experimentl situation? Woud the missing data hav chanjed these apparently inconsistent findings?

The few anecdotal reports availabl 18 months later indicated that some poor readrs continued their improvement in reading skil and intrest aftr their extended practice in reading in surplus-cut spelling - what of those who coud not be contacted? Mor longitudinl studies ar needed.

6. Difrent findings for difrent groups.

Another study (Yule & Greentree, 1986) compared readrs' initial responses to difrent typs of modified spelling with reading in conventionl spelling. Strong confirmation of the relativ dificulty of difrent typs of spelling modification when first encountrd is provided by ocasionl tests I hav givn other subjects individualy. I expectd this since the statisticl findings wer repeatd over three groups of subjects and with difrent texts. Howevr, in this ocasionl testing, I hav found that while a good many individuals replicate the finding of rapid adjustment to norml reading speed, most subjects pikd up mor sloly than the Scottish subjects. Why hav Australian subjects been slower in adaptation to text reading in surplus-cut spelling? Wer difrences in conditions important? Ar there difrences in the subject sampls or populations? Woud students from Scottish scools 10 years later, following the pedagojicl chanjes ther, now respond difrently?

One speculation is whether difrences in how peple wer taut to read afect their adjustment to improved fonemicity in spellings. That is, does it afect adaptation to surplus-deleted spellings if readrs wer orijinly taut - or lernt - by fonic methods, See-and-Say, or Languaj-Experience? There is evidence that for realy fluent readers, their latent fonics stratejies improve their skil, whatevr visual and orthografic stratejies they hav also developd (Ch.5, Yule 1991). Cutting surplus letrs might therfor help readrs with a fonics base among their skils mor than those whose basic stratejies wer formd by predominantly visual methods of teaching such as Look-and-Say or the Languaj Experience Aproach.

Most or all of the subjects in the Scottish experiments woud probably hav been taut by fonics or eclectic methods, as the Whole-word and Languaj-Experience tecniques had only recently begun to suplant them for the yungr jeneration then stil in primary scools; but in Australia many of the subjects almost certnly startd with look-and-say typ teaching. They woud therefor be less likely to hav an undrlying fonics stratejy that coud benefit as much and as quikly from surplus-cut spellings.

7. Typs of experiments now needed.

Many experiments for testing responses to surplus-cut spellings ar simpl and cheap - they require only pencil and paper. Others require computers, which can program tachistoscopic presentations and carry out the statisticl analyses. Sofisticated experiments with tecniqes such as maskd priming ar also posibl and desirabl.

The most valuabl experiments woud use subtitles on video and television. This has many advantajes. Mass testing woud be posibl, and chanjes coud esily be made in the modified spellings presentd as the experiment progressd. Vewers' subjectiv responses on the relativ ese of chanjes and objectiv mesures of facilitated reading woud then help to identify the spelling principls and exampls that realy help lernrs and users to read texts. With continued exposure, vewrs woud becom familiar both with spelling chanjes and the idea of spelling chanjes. The informd opinions of the great English-speaking public and internationl EFL readrs coud then swing in support of spelling improvement against the elitist reargards that woud deny them mor efectiv means of languaj comunication in print and on screen.


Beech, J. 1983. The effects of spelling change on the adult reader. Spelling Progress Bulletin, 23.1.11-18.

Upward, C., Fletcher, P., Hutchins, J. & Jolly, C. 1990. A Handbook to Cut Spelling. Birmingham, U.K.: Simplified Spelling Society.

Yule, V. 1986a. The design of spelling to match needs and abilities. Harvard Educational Review, 56: 278-297.Chapters 1, 10.

Yule, V. & Greentree , S. 1986b. Readers' adaptation to spelling change. Human Learning 5: 229-241. Ch.9.

Yule, V. Orthography and Reading: Spelling and Society. Doctoral thesis, Monash University, 1991.

Note: this articl is ritn in a miniml surplus-cut spelling which is based on studies of 'what th market wil bear' when new readrs encountr cut spellings for th first time. It is therefor pragmatic rather than completely consistent. As it is, even a 4.3% cut in surplus letrs greatly reduces th clutr than handicaps lernrs and confuses spelrs.

Back to the top.