Study, questioning government’s synthetic phonics evidence base, sets up questions for Ofsted

A detailed new academic study, which concludes that the government’s approach to the teaching of reading in England is “not sufficiently underpinned by research evidence” and is “failing children”, seems to raise fresh questions for the schools inspectorate, Ofsted.

The paper, by UCL Institute of Education researchers and based on a lengthy review of the existing international research base, comes out in favour of a “balanced” approach to the teaching of reading, with synthetic phonics taught alongside reading for meaning. https://bera-journals.onlinelibrary.wiley.com/doi/epdf/10.1002/rev3.3314

If the paper is correct that this is the finding that the most robust research evidence on phonics teaching points towards, then there is a clear conflict with the rules which Ofsted has set for how schools and teacher education institutions are currently inspected.

The research also includes a host of other interesting findings, including fresh analysis suggesting that England is an “outlier” in its teaching of reading compared to other English language countries. And the researchers suggest that phonics could be taught for a far shorter period in English classrooms than is the case now, based on research from overseas.

The detail: England’s internationally-unusual emphasis on discrete, synthetic phonics

Although this aspect seems to have attracted little media attention so far, this paper’s analysis of how England’s approach to the teaching of reading differs from that seen in other English-speaking countries seems important.

The paper categorises three approaches to teaching which have been advocated by those engaged in “reading wars” debates over how children best learn:

-Synthetic phonics: “a focus first and foremost on teaching children about phonemes and letters…at key moments in the teaching programme phonics teaching is separate from practising reading with whole texts. In the early stages of the approach in particular, whole text reading is required to be done by ‘decodable’ books,” featuring only or emphasising phonetically-regular words.

-Whole language: “a focus first and foremost on whole texts, ‘real’ books.”

-Balanced instruction: “a focus first and foremost on the balance between teaching based on use of whole texts and systematic teaching about the alphabetic code and also other linguistic features…lessons make explicit links between phonics teaching and other linguistic aspects using whole texts”.

The research uses a survey of primary school teachers in England, alongside textual analysis of national and state-level curriculum documentation from other English speaking countries, to establish, first, just how focused teachers in England are on the first approach, and, second, how this is unusual, internationally.

The teacher survey found 66 per cent of 634 responses on this question stating that they taught “synthetic phonics first and foremost”, with whole texts seen as the main emphasis only by one per cent. Some 27 per cent said their phonics teaching was “systematically combined with other emphases including reading comprehension” – suggesting a balanced approach.

The curriculum documentation analysis then confirmed how unusual such an approach is, when compared to other English-speaking nations.

Wyse and Bradbury analysed curriculum documents from five other countries – Australia, New Zealand, Ireland, the United States – for which they looked at documentation from the state of Massachusetts, as well as looking at the position for the nation as a whole – and Canada, where they looked specifically at documentation for the provinces of Quebec, Ontario and Alberta.

The research found that only England and Massachusetts had orientated their teaching towards an approach which emphasised “synthetic phonics” in the teaching of reading.

By contrast, the United States as a whole and Australia had “balanced instruction” systems – where phonics teaching was taught throughout with comprehension work.

In Ireland, New Zealand and the three Canadian provinces, a “whole language” approach, which seems the antithesis of England’s policy, was favoured.

The paper features some very interesting contrasts with the English approach, as revealed by these curriculum documents from overseas.

The Irish Primary Language Curriculum, for example, almost seems to caution against too much of an exclusive focus on phonics instruction at first – which is the favoured approach in England. In what seems to be a warning not to neglect comprehension work, it states: “Direct comprehension instruction, although a vital aspect of the reading process, can often be neglected, particularly in the infant classrooms where a great emphasis is placed on phonics”.

Wyse and Bradbury analysed these documents for how frequently the word stem “phon” – which of course features in the words “phonics” and “phoneme”, for example – appeared. In the English national curriculum, “phon” featured 78 times. This was twice as frequently as the next-highest region, which was Australia, on 37. In Quebec in Canada, it featured only twice, while there was no mention at all in New Zealand’s main curriculum document.

Evidence from international reading tests – PISA, where children are assessed at the age of 15, and PIRLS, where they take place at the age of 10 – was then looked at to see if there was a correlation between the type of approach to the teaching of reading and test scores.

Wyse and Bradbury put more emphasis on the PISA results, arguing that they offer a longer-term perspective on how students fare, with the tests taken at least 10 years after individuals start learning to read.

The PISA tests seem to show more impressive results for those nations and states which have what the IOE paper characterises as balanced or “whole language” approaches, the paper argues, with some combination of Canada, Ireland and New Zealand tending to make up the top three of the nations covered in this study, in PISA reading assessments which have been carried out every three years since 2000.

Wyse and Bradbury also argue that the United Kingdom*’s fluctuating PISA reading scores over time, with the country’s score in the first round in 2000 still the highest on record, show a “positive correlation” in favour of the teaching approach of England’s first national curriculum from the early 1990s, compared to the more “structured phonics teaching” which followed the introduction of the National Literacy Strategy under New Labour in 1998.

PIRLS results have presented a contrasting picture, with England sometimes scoring above New Zealand and Canada, and an improvement in the latest two rounds of the tests – in 2011 and 2016 – having come for cohorts which had experienced an increase in the focus on synthetic phonics-although results in 2016 after this was intensified under the Conservatives were only marginally above the score for 2001.

Overall, Wyse and Bradbury are cautious, though, as to the emphasis they place on this international test data, arguing that “in the end they are not a sufficient way of determining which approaches to the teaching of phonics and reading are most effective in the curriculum”.

What the results do at least show, though, for those who might argue that England’s outlier approach to teaching reading is a better option than that used overseas, is that there seems as yet not much, in the data from PISA and PIRLs, backing up that view.

The research evaluation methodology

The final aspect of this study to flag up, then, is Wyse and Bradbury’s synthesis of existing research studies.

They carried out what the paper rather dauntingly terms a “systematic qualitative meta-synthesis” – in non-technical terms, this seems to amount to be a detailed, high-level review of existing research studies on the subject, or perhaps a meta-meta-meta analysis of the existing research.

The IOE research searched for existing studies, which themselves had synthesised individual study findings.

It then included them in its own analysis if they fulfilled a number of selection criteria, including that they had focused mainly on the teaching of phonics and reading, and that they had included quality assurance checks.

The Wyse and Bradbury paper sought to emphasise research conducted using randomised control trials, and among them, those which had used assessments of children’s reading carried out some time after the intervention, to judge whether it had had a sustained effect on children’s reading.

Having conducted this selection process, the Wyse and Bradbury findings then incorporated two existing large-scale evaluations of research, while looking in detail at papers included within a “systematic review” of 55 individual papers. This led them to reach conclusions on findings from research on this subject which were said to be sufficiently robust to be included.

The findings

Wyse and Bradbury first quoted from the two existing large-scale research evaluations – or “tertiary reviews” – which had met their selection criteria for inclusion.

One, published in 2019 and led by Professor Carole Torgerson of the University of York, said the IOE study, had “concluded that although there was evidence that phonics teaching was beneficial for young readers, the evidence did not support a ‘phonics only’ teaching policy because ‘many studies had added phonics to whole language approaches, balanced instruction is indicated’.”

This suggests to me that this study found support for phonics teaching, but based on the effectiveness of “balanced instruction” approaches.

A second “tertiary review,” published in 2020 – of which the lead author was Jeffrey Bowers of the University of Bristol – had found that “the…research [it included] provides little or no evidence that systematic phonics is better than standard alternative methods used in schools…the findings undermine the claim that systematic phonics is more effective than alternative methods including unsystematic phonics (including whole language).”

The IOE research then looked in detail at 55 research papers which had formed the basis of a “systematic review” of research in the field published in 2014 by Sebastian Suggate, of the University of Regensburg in Germany.

Wyse and Bradbury then focused on the 19 of these 55 studies which had assessed interventions aimed at “typical readers” – as they wanted findings which were widely generalisable, rather than the 36 papers which had looked at reading among “children with reading difficulties”.

Of the 19 interventions looked at, one had assessed the effectiveness of an approach to the teaching of reading based on synthetic phonics; another six had assessed “other phonics” approaches, eight had assessed “balanced instruction” approaches, and the remaining four, “whole language”.

The IOE research then further narrowed down the field of studies to be considered, including removing ones which did not feature a reading comprehension element, since the end goal of the teaching of reading is for children to understand the meaning of texts.

It was left with eight studies which had demonstrated an effective intervention for the teaching of reading: four in the United States, one in Canada, on in Israel and two in Norway, with the Israeli and Norwegian studies having assessed interventions in the effectiveness of children being taught to read not in English, but in Hebrew and Norwegian, respectively. None of them was in England.

The IOE research favoured studies which had looked at interventions which had used teachers to instruct children in reading, rather than teaching assistants or volunteers. However, only two of the former type of studies were found.

Of the eight studies overall which had demonstrated an effective intervention for the teaching of reading, six were classed by Wyse and Bradbury as based on “balanced instruction”; one was a “whole language”, and one – one of the two Norwegian studies - was “phonics but not synthetic phonics”.

Among the conclusions Wyse and Bradbury make was that “our interpretation of the most robust research evidence…suggests that phonics teaching is likely to be effective if it is…carefully connected with the reading of whole texts, both decodable and real books, including a focus on reading for meaning, in all lessons”.

The research also suggests that children in England, in being taught through a different approach, have lost out.

Summing up the evidence of the paper, a press release put out by UCL IOE was headlined: “Government’s approach to teaching reading is uninformed and failing children.”

Analysis

The approach favoured by “the most robust research,” as set out by Wyse and Bradbury as above, is clearly, from the evidence within the paper and also from my own investigation of the approach now favoured by Ofsted – different from that encouraged by policymakers in England.

It is difficult to tell, as an outsider, from this research whether its conclusion based on “the most robust” evidence is the only one that could reasonably be reached, based on rigorous studies in the field. Indeed, the researchers themselves include a caveat that, for all that the searching process looks lengthy and the criteria for studies’ inclusion seems reasonable, they may have missed important research.

But what may stand out most powerfully is what the study concludes about whether in fact the research base is as robust as it might be, given the strength with which the Department for Education and Ofsted have been promoting a particular approach to the teaching of reading, seemingly in advance of much of the English-speaking world.

Wyse and Bradbury underline the fact that they were able to find no studies that met all criteria they had drawn up in advance for inclusion in this research as sufficiently rigorous and relevant to questions around the teaching of reading in England. Again, the criteria seem reasonable, as they included, for example, that studies must have “an experimental design with random allocation”; be “delivered by standard class teachers”; and “undertaken in England with the English language”.  

Strikingly, given this policy emphasis, and how long these “reading war” arguments have been going on, Wyse and Bradbury conclude: “Direct experimental comparisons for many aspects of teaching that need answers are not part of the research designs”.

Remarkably, again given how long this debate has been running, they also state: “The merits of decodable texts versus ‘real’ books and texts has not been researched using an RCT [randomised control trial] with longitudinal design.”

In other words, there has not been a study which tests, through an RCT, the reading comprehension of children who have been taught using mainly the “decodable texts” approach against those using “real” books, in which there is an investigation into which has been the better long-term impact on their reading ability.

If it is possible to wonder whether this paper’s suggestion that children appear to have lost out from the lack of a more balanced approach can be made definitively without a better research base – this critic puts forward a more trenchant critique https://fillingthepail.substack.com/p/has-synthetic-phonics-been-demolished - the overall argument that the policy push is not backed by thorough evidence seems well-grounded.

To put it another way, policymakers who were concerned to base policy on evidence would have ensured the study as described above was constructed.

Wyse and Bradbury also point to one of the Norwegian studies, which finds that “effective phonics and reading teaching” could be delivered over a period of only 9.1 hours – equating to only about four weeks of teaching of 30 minutes per day. The implication would appear to be that this was all that was needed to ensure mastery of phonics.

The paper adds that this “would be very much shorter than the 2014 national curriculum requirements in England which specify phonics teaching from the Early Years Foundation Stage (for children from age four) and then in the national curriculum for children aged five to seven. Only when children are in year 3 (age seven to eight) do the non-statutory requirements note that ‘At this stage, teaching comprehension should be taking precedence over teaching word reading directly’.”

Again, it might be too much to say that the evidence as discussed above supports an immediate move towards a radical reduction in the length of phonics teaching – although it could also be questioned why the current DfE approach should be the starting point. But what stands out again, at least as I understand the argument as put forward by Wyse and Bradbury, is the absence of evidence in favour of the current amount of time devoted to phonics teaching in England.

The research also includes many other interesting findings. Read Write Inc, the synthetic phonics programme offered by the former headteacher and DfE adviser Ruth Miskin, now seems to be dominant in English classrooms, the survey found.

And the government’s phonics screening check, introduced for key stage 1 pupils in 2012 by the former schools minister Nick Gibb, appeared not to be popular among those survey respondents registering an opinion.

Results from the check are also included in data systems used by Ofsted, despite 68 per cent of respondents to the government’s consultation on the introduction of the assessment being opposed to this.

Ofsted

The Wyse and Bradbury paper characterises Ofsted as one of the mechanisms through which the government has sought to enforce its emphasis on synthetic phonics. It states that “a range of associated means to ensure compliance to synthetic phonics teaching were…implemented by the government,” for example Ofsted setting out in its current inspection framework what inspectors look for in the teaching of reading.

Ofsted clearly puts great emphasis on synthetic phonics. In 2018, Education Uncovered sought a response from the inspectorate on the research basis of statements in its controversial “Bold Beginnings” report on early years teaching, which had focused on “systematic synthetic phonics”.

This had stated that: “All primary schools should make sure that the teaching of reading, including systematic synthetic phonics, is the core purpose of the Reception Year.”

Under what initial teacher education providers should do, Bold Beginnings states that they should: “Devote a greater proportion of their training programme to the teaching of reading, including systematic synthetic phonics as the [Ofsted’s bold type] route to decoding words.”

However, when I asked Ofsted in 2018 for the research base on which these statements rested, it pointed me towards a study which had in fact not found in favour of synthetic phonics, but that a “variety of approaches” had been found to be effective.

As Wyse and Bradbury point out, great stress is placed by Ofsted, in the way it currently inspects primary schools through the inspection framework in place since 2019, on systematic synthetic phonics. Its particular emphasis seems to tie in with the IOE paper’s definition of the synthetic phonics approach, Wyse and Bradbury’s definition of synthetics phonics including the statement that: “In the early stages of the approach in particular, whole text reading is required to be done by ‘decodable’ books.”

In its current Inspection Handbook**, in its stipulations as to how inspectors are to inspect early years provision, Ofsted states: “In Reception, staff teach children to read systematically by using synthetic phonics and books that match the children’s phonic knowledge [my italics].”

Under its criteria for judging whether a school is to be rated good or better, describing what good or outstanding provision looks like, the handbook states: “Reading books connect closely to the phonics knowledge pupils are taught when they are learning to read.”

Given that the Wyse and Bradbury paper suggests there is no good evidence to support this particular approach, I asked Ofsted to point me towards its evidence base.

The inspectorate said: “Our education inspection framework is underpinned by extensive research, which points to the importance of teaching phonics until children can decode automatically. Pupils who struggle to decode unfamiliar words quickly fall behind their peers and struggle to catch up.”

That looked fairly unspecific. But Ofsted did publish its own research review, in January 2019. This sets out the evidence on which it says the current inspection framework rests. So I looked at that.

This document rightly flags up the criteria referring to phonics in the current inspection handbook. But it seems to me that there is no explicit link made between the specific type of phonics teaching which inspectors are now required to look for, and the research base.

Specifically, again Ofsted’s inspection handbook mentions that “reading books connect closely to the phonics knowledge pupils are taught when they are learning to read”, while not mentioning, in this section on learning to read, anything about the need for children to be given access to authentic texts.

Again, this seems to correspond to Wyse and Bradbury’s stipulation that, in synthetic phonics programmes, “In the early stages of the approach in particular, whole text reading is required to be done by ‘decodable’ books”: texts in which children are exposed only or mainly to phonetically regular words.

Ofsted’s summary of research states: “Studies show that explicit and systematic teaching of the manipulation of phonemes (the smallest unit of sound in a language) and phonemic awareness (the ability to identify phonemes in written words) is crucial and should be continued until children can automatically process this information.

“Children’s reading development is also aided by a literature-rich environment and practice in reading authentic literature and familiar materials.”

Yet, while Ofsted indicates that exposure to authentic texts can help children’s reading development – and by implication in Ofsted’s view has been found by research to do so – this does not feature in its inspection handbook. That is, the words “authentic” or “real texts” or “real books” are not mentioned in the handbook. And, while the handbook does indicate that children should have some exposure to stories that “develop pupils’ vocabulary, language comprehension and love of reading,” again what it says about how phonics should be taught seems to emphasise phonically regular texts.

The handbook states that, when inspectors are making judgments on the quality of education, they will look to see that “the sequence of reading books shows a cumulative progression in phonics knowledge that is matched closely to the school’s phonics programme.”

The lack of emphasis on authentic texts within initial reading teaching, which is one element of effective programmes identified by the Wyse and Bradbury research, seems to match the situation found in schools as reported in its survey, with only 27 per cent of respondents stating that “my phonics teaching is systematically combined with other emphases including reading comprehension”.

This all may sound incredibly technical and arcane. But it is surely very important to the detail of the child’s learning experience at school. Should children during that very early stage of learning to read be taught mainly through “decodable” books, with little exposure to more authentic texts, or should there be a more balanced approach?

Ofsted inspections do seem to favour the former, although it is the Wyse and Bradbury claim that there is not good evidence for this. And, indeed Ofsted’s research review does not seem to flag up, as far as I can see, evidence for the specific approach its inspection systems now seem to promote.

Summing up their argument, Wyse and Bradbury state: “Although there remains on doubt that phonics teaching in general is one important component in the teaching of reading, the research certainly does not suggest the complete exclusion of whole language teaching…the most effective interventions relevant to our research questions carefully connected the reading of whole texts with the teaching of phonics and other relevant aspects within all lessons.”

It is remarkable, given that England has a government and now an inspectorate promoting what seems to be a very specific approach to the teaching of reading, that there is not a more definitive research base behind it.

Allegations of weaknesses in the research base are not new

This seems to have been a complaint going back at least 15 years, despite successive governments having since put ever greater emphasis on this particular approach to the teaching of reading.

In 2018, Education Uncovered reported how the Torgerson et al research review as mentioned above had found “little evidence” in favour of any one approach to the teaching of phonics. My piece highlighted the fact that Torgerson and a colleague, Greg Brooks, had advocated back in 2006 that a large randomised controlled trial was carried out before any new policy sought to change the teaching of phonics in schools.

However, instead a review carried out for the-then Labour government by the former senior Her Majesty’s Inspector Sir Jim Rose concluded that “synthetic phonics is more effective,” and, as Torgerson and Brooks said: “A decision was evidently taken in central government to roll synthetic phonics out nationally.”

This relationship between policymakers and research in this field is not new, then, although Ofsted’s position in enforcing particular teaching approaches may be.  

Overall, Wyse and Bradbury, who conclude that the influence of individual politicians over education policy needs to be reduced, are also surely right to argue that significant – and internationally unusual – stipulations by government and its agencies on the way that children should be taught to read really ought to be made on the basis of a proper research base.

If there truly are gaps in the research as indicated in this paper, this should be seen by the public as a deeply concerning finding.

*Results specific to England, rather than the UK, are reported in the Wyse and Bradbury paper as only having been reported in PISA tests taken from 2015. UK results are something of a proxy, given that most pupils in the UK are in England.

**Ofsted’s current inspection handbook for initial teacher education arguably puts even greater emphasis on synthetic phonics. It includes as one of the criteria for a provider being judged inadequate the statement that “EY [early years] and primary training does not ensure that trainees only learn to teach decoding using systematic synthetic phonics as part of early reading”.