Spelling

[chew8]

A friend who has been having a clearout has just returned a letter which I wrote to her during a visit to South Africa in August 1987. She and I have known each other since being at university together in England in the early 1960s, and we kept in touch when I went back to South Africa and taught there in the mid-1960s at the school mentioned below. She then joined the staff of the sixth-form college where I was teaching in the 1980s, and used to help me with my annual marking of spelling scripts – Schonell Graded Word Spelling Test B. This is an extract from the letter, where ‘school’ is the S.A. school and ‘college’ is the sixth-form college:

A friend who is now vice-principal of the school had organised the whole of the 5th form (aged 16+) to do the spelling test for me. As I expected, the results are better than those from [college] –the highest mark, among the 177 who did it, is 36 out of 70 (10 out of 70 at college, with a significant number between 20 and 35), and all but 4 of the school pupils have spelling ages over 12. 55.5% have spelling ages of 14 or over, as compared with under 43% of the 1986 college intake and about 46% of the combined 1984-6 college intakes. The weakest school students still spell fairly logically – e.g. cemetary, leasure, subtirranian, missilaneous, apperatus, exagerate – which sets them apart from the weakest college students who are so woefully lacking in strategy.

Back then, in 1987, I was very aware of the ‘woeful lack of strategy’ shown by some of the college students, but it would be another year or so before letters of mine to the TES elicited responses from people in England who told me about the prevalence of non-phonic teaching. Things then started to fall into place: I knew that primary schools in S.A. were still using the sort of code-based approach that had been used when I was at primary school there, and this fitted in with the differences I saw in spelling between 16-year-olds in S.A.and 16-year-olds in England. The S.A. students were the equivalent of a comprehensive-school group (i.e. average in terms of general ability), whereas the sixth-form college students were well above the national average in terms of GCSE results, and yet the S.A. students were better at spelling than the college students.

Jenny C.

[john walker]

Very interesting, Jenny! The question is then: how would the two sets of results compare with what you might expect/get today?
It would be intriguing to see what would happen if someone were to run the same test. Any takers?
Best,
John

[chew8]

Apologies: in copying from my handwritten letter, I wrote 'the highest mark, among the 177 who did it...' I should have written 'the lowest mark among the 177 who did it'. The lowest mark in the South African school (36) was over 3 times as high as the lowest at my college (10), all students being aged 16+.

It would be wonderful if someone could run the Schonell test now with that age-group. I did my last lot of testing in September 1999 and retired from the college the following summer. I toyed with the idea of asking if I could re-run the test in 2010, when the college entrants would have been students who had started infant school in 1999, but I decided against it, as there were factors such as changes in the intake (e.g. more or fewer students on non-academic courses) that could have affected the results. In any case, the students would have still been the products of the NLS rather than of the more phonics-orientated approach recommended in the 2006 Rose report.

Jenny C.

[Debbie Hepplewhite]

Jenny - your extensive testing of spelling made a really important contribution to revealing the long-term effect of pupils not being taught explicit phonics - and then you yourself went on to make an invaluable contribution to the debate about reading instruction for many decades - and still you continue to do so.

That contribution is also about the real need for objective testing, sustained testing, and the need to compare results in different settings and circumstances.

This is exactly what is required re the simple phonics screening at the end of Year One in England - and let us hope this regime is continued for the long-term to get to grips with the realities of teaching effectiveness for reading instruction.

Spelling requires an equivalent regime - and John and his Sounds-Write associates are right to have worked hard to establish objective and sustained testing of spelling - and should be highly commended for this effort and the results themselves.

Let's hope in the future that the DfE might establish the equivalent to the Year One phonics screening check for reading - for spelling purposes.

[chew8]

One problem I was always aware of when working with students aged 16+ was the difficulty of proving that there was a link between the way they had been taught to read and spell as beginners and the problems some of them had with spelling 10 years or so later. When I was teaching Surrey students aged 16+, I knew a bit about typical teaching at infant level, partly through tutoring a few children, partly from what I heard from parents and a couple of ed.psychs. whom I knew, and partly from reading about what was happening in teacher-training – e.g. Tom Gorman’s 1989 paper ‘What Teachers in Training Read about Reading’. But there was really only a smoking gun – I knew that first look-and-say and then ‘real books’ had been prevalent when my students had been beginners, but I didn’t know for sure that the strugglers had been taught that way – there was always a possibility that they hadn’t.

A typical argument against phonics is that children will end up spelling ‘phonetically’ rather than correctly. Yes – but if some spelling is going to be incorrect, as will always be the case, ‘phonetic’ misspellings are far preferable to bizarre misspellings where the letters on the page bear little relationship to the sounds in the word. The South African 16-year-olds I tested produced more correct spellings than my Surrey 16-year-olds and the errors of the weakest were phonically plausible (e.g. ‘equipt’, ‘equiped’), whereas the errors of the weakest Surrey students were bizarre (e.g. ‘acupet’ and ‘epitt’ for ‘equipped’).

I think the Year 1 phonics check will help, but the long-term impact on spelling won't be known for years and may never be known unless objective testing is done. The Vernon test was re-standardised in 2005-6 and this showed a slight improvement since the first standardisation in 1975 - the authors put this down partly to the NLS. If that test is re-standardised again in the next few years it may give us the sort of evidence we need - but will it be?

Jenny C.

[Susan Godsland]

Thanks for flagging up the Gorman paper, Jenny.

I've found a copy of it on the web and it's shocking reading:

What teachers in training are taught about reading.
Greg Brooks, Tom Gorman, Lesley Kendall and Alison Tate

http://www.nfer.ac.uk/publications/91007/91007.pdf‎ (1992)

[chew8]

Also very interesting, but not actually the same publication, Susan. I think the one you found was produced as part of an inquiry by CATE (Council for the Accreditation of Teacher Education) in the follow-up to the release of data by Martin Turner. The one I mentioned was by Gorman alone, and it may not be available on line - my copy was given to me, but I can't remember by whom. It appeared in 1989 and had, I believe, been requested by Sir Keith Joseph when he was Sec. of State for Education.

Jenny C.

[Susan Godsland]

I did wonder because of the date difference but presumed it was a re-print.

You're right, the Gorman 1989 paper isn't available, but I did find this by Gorman and Fernandes -equally shocking.

Reading in recession

www.nfer.ac.uk/publications/91141/91141.pdf‎

[chew8]

Yes indeed - I got a hard copy of that when it came out. Not long afterwards (1993, I think) another NFER publication, Spelling it Out, attacked my data on spelling. I have copies of the correspondence I had with the author (Greg Brooks) at the time.

Jenny C.

[Susan Godsland]

Spelling it out: The spelling abilities of 11- and 15-year-olds
Greg Brooks, Tom Gorman and Lesley Kendall
Research Report, January 1993

This report is based on a study of 1500 scripts collected during Assessment of Performance Unit Language Monitoring surveys between 1979 and 1988. Two general writing tasks (one narrative, one argumentative) were studied.

Free download:
http://www.nfer.ac.uk/publications/91163/91163_home.cfm

[chew8]

A key point was that the NFER had simply looked at the spelling in samples of children's free writing, whereas I had run exactly the same spelling test (same words, same time of year) every year for 9 years when the NFER report came out. I asked whether they had taken any account of the difficulty of the words the children had used in their writing - they hadn't.

Jenny C.

[Debbie Hepplewhite]

Jenny said:

I think the Year 1 phonics check will help, but the long-term impact on spelling won't be known for years and may never be known unless objective testing is done. The Vernon test was re-standardised in 2005-6 and this showed a slight improvement since the first standardisation in 1975 - the authors put this down partly to the NLS. If that test is re-standardised again in the next few years it may give us the sort of evidence we need - but will it be?

I do hope that there will be a lead on testing for spelling - and not just for 100% accuracy, but also for the type of analysis Jenny describes above.

If a steer is not taken by government, perhaps we should all work collaboratively to build on the same spelling tests that Sounds-Write commendably use across a number of schools?

[chew8]

Although analysis of types of errors gives a good indication of whether or not children have been taught phonics systematically, what the government and the general public will probably be more interested in is correct spelling. It would be good if it could be shown that more children are spelling more words correctly since the introduction of the Year 1 phonics check, but this would need something like another re-standardisation of the Vernon test

One couldn’t see whether there was more phonically plausible spelling among primary-school children unless one had a representative sample of spelling-test scripts from pre-check days, and I doubt whether anyone has those. I have 6,000+ scripts from 16-year-olds, but it will another 8 years before the first products of the phonics check are 16.

Jenny C.

[john walker]

As we are all aware, this is an area fraught with problems.
Problem 1 is which test do we use. If we use the same test, using the same words, every year, inevitably teachers will teach to the test. This is why I like Dennis Young's Parallel Spelling Test, which allows teachers to use different sets of words so that the same test isn't given twice. However, as Jenny has pointed out, the test hasn't been re-normed for a long time and should perhaps be brought up to date.
Problem 2 is the question of what is being tested. Are we to test Dolch words, or should the criterion be to test what has been taught. If we were to follow the latter, then we'd probably agree to test CVC words, followed by CVCC, CCVC, CCVCC, and so on. Nevertheless, not everyone would be happy with this because, as we know, L&S makes the explicit teaching of words of greater complexity (CVCC, CCVC, etc.) optional. Some teachers would also want to throw in double consonants and some consonant digraphs, such as <sh>, <ch>, and so on, at an earlier stage than others.
Whatever we think about 1 and 2, problem 3 is: how do we go about testing different ways of spelling sounds? Do we test just one way of spelling sounds? Two? Three? Clearly teachers will jump up and down if we test code knowledge which hasn't (yet?) been taught.
When we decided, strategically, to lay more emphasis on testing spelling, we did so for a principled reason: we believed (and still do) that spelling gives a far more accurate picture of children's literacy than reading. Fortuitously, it is also much easier to test, in that a test can be delivered to a whole class in one fell swoop and it is easy to mark. That isn't to say that we don't think it's a good idea to test reading.
A lot of heads of schools involved in helping us to collect data were able, after a number of years, to be able to see at a glance how their cohorts compared. Obviously, although cohorts differ slightly from year to year, heads realised that there shouldn't really be major shifts in results, which then also gave an indication of how well teachers were teaching. The tests were also an encouragement for teachers to 'keep up to the mark'. [Of course, the problem with this last point is that once results between schools are compared, testing starts to become high stakes and then results are more likely to become skewed.] another advantage in using a test, any test, is that it keeps in mind that the teaching of reading and spelling is a big deal and that it is the foundation on which all else that takes place within the primary school is based.
What really surprised us when we started collecting data was how few schools were using any kind of normed and standardised tests. Every judgement they made about pupils' reading and spelling abilities seemed to be based on opinion. Unfortunately, this is still so often the case. What also surprised us was that when marking the spelling tests we sent out, teachers seemed to be more strict (over anomalies) than we were and there was hardly any evidence at all of cheating!
We have begun to think about this again recently and I have asked a number of schools to start testing, using the Parallel Spelling Test. For us though, it is enormously expensive to conduct on the same level that we did in the beginning. This is because one of our team [David Philpot] spent a lot of his working time sending out tests, lists of words to use that year, and clear instructions on how to conduct the tests. Then, there was all the data on the chronological ages of each individual child, the double marking, and so on. It was truly a labour of love and it costs a lot of money to be able to do it. David's reputation as an educational psychologist and his integrity were also widely respected - which was certainly a factor in Wigan.
As I've said before, I'd also love to see what would happen if the Schonell were to be used again and compared with Jenny's results. I wish now I'd kept the vast numbers of tests results we collected on boys arriving in secondary school at the beginning of what we now call Year 7. I remember very well how the results got worse year after year and that this made decisions about who needed interventions and who didn't very tricky, particularly when more boys were dropping below the 9:0 to 9:6 levels. This was generally considered to be the level at which children could at least function within the first year of their secondary schooling; yet there were so many dropping below the nine years level that we didn't have the resources to be able to offer extra help. Only the very most needy were given support. And that, was pretty much whole language in those days!

[chew8]

If one wants to show whetherwnational standards have gone up or down, then the test needs to have past national norms which can be used for comparison. I believe that phonics-taught children do better than mixed-methods children on any kind of test - more GPCs will have been taught explicitly, and even if things have not been taught explicitly (e.g. higher-order GPCs) children who have been taught in a strongly code-based way are more likely to have absorbed spellings via their reading.

The advantage of the Vernon is that there is a fairly recent standardisation (2005-6) which shows what standards were like not too long before the Y1 check was introduced. The researchers who did that re-standardisation were able to say that the National Literacy Strategy had probably made a difference - children who had been through primary school in NLS years did better than the original 1975 standardisation sample, but children who had been through primary school pre-NLS did not. If the Vernon were re-standardised yet again on children who had and had not taken the Y1 phonics check, it might show a before-and-after pattern.

The disadvantage of the Vernon is that I think the 2005-6 norms may be suspect for children below the age of 8 or 9 - e.g. the children used in the standardisation sample were supposed to be nationally representative, but the Y3 children had a raw score well below what was supposed to be 'normal' for that age-group. I had a lot of correspondence about this with the researchers several years ago, but got nowhere. Still, I suppose that if they re-standardised again, they'd look at raw scores again, and if the average raw score of Y3 children was higher than in 2005-6, that would mean that standards had risen.

Jenny C.