The government’s assessment of early reading, taken by hundreds of thousands of five- and six-year-olds in England every year, is not testing what it is supposed to test, research has concluded.

The phonics screening check, introduced by the coalition in 2012, is failing to assess the full range of phonic knowledge which the government-designed national curriculum says pupils should have.

Detailed analysis of the words which pupils have been asked to read in the check, alongside the pass mark, shows that youngsters can get through the assessment with only basic phonic knowledge, rather than with a full understanding of the phonics curriculum.

And, although billed as a test only of pupils’ ability to follow the rules of sounding out written letters into spoken words, in fact pupils need vocabulary knowledge to work out how to pronounce 40 per cent of the words. This means that the “phonics” check is not simply a test of phonics.

This has big implications for teaching, it is argued, with staff wrongly spending time teaching elements of phonic knowledge which will not be very useful either in the tests themselves or to the pupils in reading real books, when they might more usefully be spending more time building children’s vocabulary.

The findings come in a paper being presented to BERA today by Dr Jonathan Solity, of the education consultancy and teaching resource provider Optima Psychology, who is also an honorary research fellow at University College London, and Dr Cat Darnell, of the University of Birmingham.

The rules for the phonics screening check, taken by children towards the end of Year 1, state that pupils are to be assessed on their understanding of what are known as grapheme-phoneme correspondences (GPCs).

The pupils are presented with words – 20 real-words in each check, and 20 made-up or “pseudo words” – and then assessed on their ability to read letters and combinations of letters (graphemes) in each word, which they then turn into sounds (phonemes). They then blend individual phonemes together to make spoken words. So children might see the word “cat”, for example, then sound out the letters “c”, “a”, and “t” and then blend them together to make the spoken word.

The check’s rules specify 64 different letters or letter combinations – graphemes – on which children could be tested. Each one produces at least one possible sound, or phoneme, with some graphemes producing more than one phoneme. For example, the letter c could produce the sound the c makes in “cat”, or the one it makes in “cell”, so in this case one grapheme produces two possible grapheme-phoneme correspondences.

Overall, the check is meant to assess children’s understanding of 85 of these grapheme-phoneme correspondences. However, Dr Solity and Dr Darnell’s analysis of the first three years of the check – 2012, 2013 and 2014 – shows that 27 of the GPCs, or 32 per cent – were never tested in the words children were actually asked to sound out.

Only a few of the GPCs dominated the content of the words children were asked to sound out in the 2012-14 checks, with 15 GPCs – all of them associated with single letters such as the “t” sound produced by the letter “t” – accounting for two thirds of the GPCs assessed.

In 2012-14, children needed to pronounce 32 out of the 40 words (80 per cent) correctly to pass the test. However, the analysis shows that any pupil could achieve this score using only relatively simple phonic knowledge.

That is, they could score 90 per cent in 2012, 92.5 per cent in 2013 and 85 per cent in 2014 by only using their knowledge either of letters which the check’s rules say should only produce one sound, such as the letter “d” making the d sound in “dog”. Or, where a letter produces more than one sound, pupils only needed to rely on the most frequently-found sound, such as the “c” sound the letter c makes in “cat”.

In other words, the research found, pupils did not need to know the less common GPCs which the curriculum says they should understand, such as the long “e” sound that the letter e makes in the word “she” or the sound the letter combination “ou” makes in the word “you”, in order to pass the check. This is despite the national curriculum specifying that children should learn “alternative sounds for graphemes”, and the rules of the check itself stating that children should be able to decode some less frequent GPCs in order to pass.

“Children are able to successfully demonstrate their ability to decode phonically [ie to pass the test] even if they do not yet possess knowledge of multiple mappings or knowledge of less frequent GPCs,” the researchers found. Children “who may not possess knowledge of a large number of GPCs” can therefore pass the test.

Ministers strongly advocate the teaching of systematic synthetic phonics – children using only their knowledge of GPCs in order to sound out simple words – before they go on to tackle more irregular words which they simply have to remember how to pronounce.

However, the paper shows that in many cases in the “phonics check”, children actually have had to rely on their vocabulary knowledge to know how a word sounds, rather than only following phonic rules, in order to be marked as having sounded out the word correctly.

For example, the word “brown” featured in the 2014 check. But pupils had to use their vocabulary knowledge rather than phonics to pronounce the word correctly since the grapheme ‘ow’ can be pronounced in one of two ways to rhyme with ‘cow’ or ‘slow’ according to English phonic rules. It is only knowing the meaning of ‘brown’ that leads to the correct pronunciation.

One approach to the findings, admit the researchers, might be to change the phonics check so it more comprehensively assessed the content of what the national curriculum says pupils should be taught.

This, however, would be a mistake, they argue. Instead, the content of the check correctly reflects the reality that only a small number of GPCs dominate the English language.

Pupils should therefore spend time learning this small number quickly, and then move on to reading “real books” to build up the vocabulary which will help them decode words where the sounds made are less phonically predictable.

Instead, schools are potentially wasting time on teaching GPCs which do not actually feature much in the check, and, more importantly, which are of little use to children in reading for real. It is disadvantaged children who would most benefit from a greater attention on building vocabulary, rather than learning less frequent GPCs, it is argued, as those living in less “language rich” homes are exposed to far fewer words outside of school than other children are.

Some pupils, then, are struggling with reading having wasted too much time being drilled on less frequent GPCs, it is argued.

Dr Solity said: “This is not an anti-phonics argument. It is absolutely clear that children need to be taught phonics, and systematic synthetic phonics in particular.

“What we are questioning is whether it is worth teachers spending a great amount of time making sure pupils learn all 85 GPCs, rather than concentrating on the most frequent ones and then building pupils’ vocabulary.”

He added: “Reading standards are more likely to be improved, and literacy difficulties prevented, through teaching a small number of high utility GPCs and devoting the time currently spent teaching low frequency GPCs, to developing pupils’ language skills and vocabulary knowledge.”

“Is the Phonic Screening Check a Major Cause of Pupils’ Difficulties in Learning to Read?” is being presented to BERA by Dr Jonathan Solity on Thursday, September 15^th.

Further information from:

Warwick Mansell

BERA annual conference press officer

07813 204245

Email: warwickmansell@gmail.com

Notes for editors:

1 The authors have also analysed the content of the 2015 phonics check, and reached similar conclusions, though they have not yet produced a paper on this. Analysis of the 2016 check has yet to happen.

2 The paper says that 80 per cent of the 60 “real words” featured in the 2012-14 checks have more than one “plausible” alternative pronunciation, if pupils only used phonic rules to answer them. Yet only the pronunciation which is correct in real life will gain them a mark, meaning that pupils need vocabulary knowledge – the ability just to remember what the true pronunciation of the word is – in all these cases to be marked correct in this “phonics” check. With “real words” making up half of all words asked of pupils in the check, this means that 40 per cent of all the words contained in the test require vocabulary knowledge to work out which phonically plausible alternative is correct.

3 Examples of real words pupils were asked to read in the 2014 check include doom, crowds and fighters. Examples of pseudo words were vol, teg, froid and harnd. The full paper for 2014 is available here: http://bit.ly/2bXV1NB.

3 Dr Solity runs a company called Optima Psychology, for which Dr Darnell also works as an associate researcher. The firm offers teaching resources, including reading resources which support the teaching of phonics through encouraging children to read real books. The resources focus on the approach the researchers advocate in the paper’s conclusion, ie ensuring pupils master the most frequently-occurring GPCs.

4 The annual conference of the British Educational Research Association is being held at the University of Leeds from Tuesday, September 13th to Thursday, September 15th. More than 500 research papers will be presented during the course of the conference.