Has synthetic phonics been demolished?

I cannot believe we are all doing this again

---

Education research articles are rarely reported in the press. Yet some researchers have the right mojo. A new paper by Dominic Wyse and Alice Bradbury published in the British Educational Research Association’s Review of Education has so far generated a report in The Guardian that describes the paper as a ‘landmark study’, an op-ed, also in The Guardian by one of the authors and an op-ed in The Conversation by the other author. That’s quite a campaign. What’s the story?

It is about reading. Much to the consternation of education professors in recent years, England has pivoted towards requiring early teaching of the relationships between written letters and the sounds they represent - phonics.

Advocates of what has become known as ‘synthetic phonics’ do not believe that the only thing children need to be taught is letter-sound relationships. Many subscribe to the ‘simple view of reading’ that says that reading is the product of decoding - turning squiggles on the page into recognisable words - and comprehension - knowing what those words mean and understanding the situation they are describing. ‘Phonics only’ is not enough and nobody claims that it is.

Teachers may still read books aloud in order to build vocabulary and word knowledge, or simply to motivate children about books and stories. In a synthetic phonics programme, children also typically read ‘decodable’ books themselves. These contain only the letter-sound relationships that the children have so far been taught so that they are not expected to read anything they cannot decode. This contrasts with the common approach of using ‘levelled’ readers that control vocabulary and use repetition or pictures to enable children to guess unknown words.

Here is how Wyse and Bradbury describe decodable texts:

“In the early stages of the approach in particular, whole text reading is required to be done with ‘decodable’ books which are reading scheme/basal books with vocabularies controlled to enable repetition of key words learned during the phonics programme.”

This is not correct. Decodable books do not control vocabulary or focus on the repetition of key words - that’s a feature more typical of levelled readers.

So, why make such a mistake?

Essentially, the evidence for phonics is so strong at this point that it is hard to deny. Instead, critics tend to argue against the strawman of ‘phonics only’. Take this example from the section of Wyse and Bradbury that reviews previous research.

“Galuschka et al. (2014) found that phonics instruction was the most effective method for the reading and spelling performance of ‘reading disabled children and adolescents’ (p. 9). However, the description of phonics instruction in this [systematic review] included reading fluency, described as ‘repeated word or text reading practice’, so could be described as a balanced instruction orientation.”

Strong, suggestive evidence of the value of phonics is highlighted but is then spun as evidence for ‘balanced instruction’. What do the authors think children are doing when they read decodable books if not undertaking, ‘repeated word or text reading practice’? By this definition, synthetic phonics programmes are already ‘balanced’ and so there is no problem here.

Wyse and Bradbury conduct what they call a ‘systematic qualitative meta-analysis’ which is methodologically weird. Usually, when researchers conduct a systematic review, they search all of the academic archives for relevant keywords and then filter the papers they find using preset, ideally preregistered, criteria. Instead, Wyse and Bradbury have sourced their papers from other researchers’ systematic reviews. Or, rather, one review - Suggate, 2014. Although a lot less time-consuming, this is a problem because we now have multiple layers of selection criteria, including Wyse and Bradbury’s, that do not necessarily complement each other. And because Suggate’s results are already known, it gives too many researcher degrees of freedom, presenting a strong risk of bias in the findings.

The result is a mish-mash, with none of the papers meeting all of Wyse and Bradbury’s new criteria.

“The main finding of our [systematic qualitative meta-analysis] is that there is no study that has been carried out in England with typically developing readers that fits our rigorous criteria.”

Nevertheless, they confidently assert that:

“Our findings from analysis of tertiary reviews, systematic reviews and from the [systematic qualitative meta-analysis] do not support a synthetic phonics orientation to the teaching of reading: they suggest that a balanced instruction approach is most likely to be successful.”

In addition, Wyse and Bradbury conducted a survey of teachers in England. This was not a random sample and so there is probably no need to discuss it further. Nevertheless, it seems to show that most teachers in the sample, prior to their students sitting England’s phonics check, a national assessment of phonics knowledge, were using a synthetic phonics approach. I guess we are invited to infer that this is the mythical ‘phonics only’ that involves no reading of texts, but we’ve seen why such a view would be fallacious.

The final strand of Wyse and Bradbury’s analysis involves PISA and PIRLS - large scale international literacy assessments. Part of their argument is obscure and strikes me as potentially selective. However, one main thrust is that English-speaking countries that have higher PISA rankings that England, do not use a synthetic phonics approach.

PISA rankings are not a good use of PISA data. Many things vary between different states that could impact on reading scores other than the method of teaching reading - demographics, relative wealth, cultural priorities, population size, population homogeneity etc. A better comparison is to look at how a country has changed over time. Here are the reading trajectories for Canada and England:

Note that PISA assesses the reading comprehension skills of 15-year-olds - which will be impacted by both decoding ability and comprehension - that synthetic phonics only became prominent in England after the 2006 Rose review, and that the phonics check - which initially showed that a large number of students were failing to meet the standard - was only introduced in 2012.

When it comes to PIRLS, the most recent round arguably showed significant improvements in reading performance which we might tentatively associate with the introduction of the phonics check. So there’s that.

Wyse and Bradbury venture way beyond the limited evidence they have accumulated to make a number of claims, with the discussion section reading more as an extended opinion piece than a discussion of the evidence they have presented. For instance, on the basis of seemingly nothing, they argue that scripted lessons are likely to be ineffective. They also make a political point:

“In our view it is now apparent from the example of successive governments in England, of different political parties, that it is not appropriate for governments and their individual ministers of education to have the power to directly control curriculum and pedagogy.”

Who should have this power? Presumably some unaccountable committee made up of education professors. I disagree. I would rather be able to vote out those who impose educational failure on children.

Does such a coordinated attack on phonics matter? Few will read this Substack. Far more will read the articles in The Guardian and The Conversation. And around we all go again.

One thing is for certain. No longer can anyone claim that it is advocates of phonics who are fanning the reading wars in England.

Update: Dr Jen Buckingham has written a more thorough rebuttal that you can read here