Reading Reform Foundation

Thank you for that clarification, Jenny.

I'm sure that I won't be the only person who has muddled the two.

More thoughts arising from Toots's mention of Appendix B in the Torgerson, Brooks and Hall review....

When I first read the review in 2006, I ticked quite a lot of things in Appendix B, particularly in the two complete paragraphs on p. 56. I also pencilled in a comment that ‘the pressure for “fast” arises at least partly because of the pressure to get children reading books a.s.a.p.’. What I had in mind was that people who favoured approaches other than s.p. at the time were arguably taking things even faster by expecting children to read whole books using several different cueing strategies from the start of Reception.

The reference to Brooks at the end of the ‘Only’ paragraph on p. 56 made me re-read the 2003 paper of his referred to (http://dera.ioe.ac.uk/4938/5/nls_phonics0303gbrooks.pdf) and also a chapter by him in the UKRA’s book Perspectives on the Teaching and Learning of Phonics (ed. Cook, published 2002). The book chapter was about phonemic awareness training and something that struck me afresh was the enthusiasm Brooks expressed back then for a Welsh initiative called ‘POPAT’ (Programme of Phonemic Awareness Training). This involved initial training in phonemic awareness without letters:

In 2002, Brooks wrote:POPAT seems to me to have the potential to speed up the learning of the alphabetic principle for many children, and to unlock it for children who are mystified by conventional phonics.

Brooks himself was monitoring it, so when I saw him at a conference the year after the book came out, I asked him whether he was writing up any results. He said he wasn’t, as the children had now done the Key Stage 1 assessment and the POPAT children had done no better than the rest. I was sorry that this was not publicised – it might have helped to persuade people that training phonemic awareness before starting to teach phonics doesn’t help. It illustrates the way that useful findings don't always see the light of day.

In his 2003 seminar paper, I was struck by this, on p. 13:

He wrote:Also, it is not as clear as it might be from the experimental literature that synthetic phonics is markedly more effective than analytic. As far as I am aware, there are rather few experiments, and all but one (Rhona Johnston and Joyce Watson’s study in Scotland) took place in the United States; confirmation is needed that the same result would apply in England.

In any case, before a randomised controlled trial was authorised the existing literature on synthetic vs. analytic would have to be thoroughly evaluated through a systematic review and meta-analysis; if this provided convincing evidence from several trials in English-speaking countries then it would be unnecessary to carry out an RCT in England. If the review failed to provide convincing evidence, there would then be a case for mounting a study of ‘naturally occurring variation’ in the teaching of phonics correlated with children’s progress.

Well, the review of existing literature was done 3 years later, with Brooks himself as part of the team (Torgerson, Brooks and Hall). The researchers concluded, on the basis of two USA studies and one UK study, that there was no ‘convincing evidence’ in favour of s.p., but this conclusion may not have been reliable, as Johnston, McGeown and Watson have pointed out – here’s the extract again:

Johnston et al. wrote:However, a meta-analysis, funded by England’s Department for Education and Skills (DfES), claimed that there was no clear outcome as to whether synthetic or analytic phonics was the most effective method (Torgerson, Brooks, & Hall, 2006), which may seem surprising in the context of the research by Torgesen et al. (1999) and Johnston and Watson (2004). There are various reasons for this null result. One of the three studies included in the meta-analysis was an unpublished study of kindergarten children, where the children were inappropriately trained on complex vowels, such as tape and rode (Skailand, 1971); these sorts of words are not suitable for early sounding and blending. An advantage was found for the analytic phonics group on the trained items, but not on the untrained words. However, the data on the reading of the trained words were used in the meta-analysis, whereas the National Reading Panel only analysed examined performance on untrained items. Torgesen et al.’s (1999) study was also included. This showed in the long term that the synthetic phonic method was more effective than embedded phonics but Torgerson et al. (2006) used data from a few months into this two and a half year study, when the embedded phonics group was briefly ahead in reading. This was because the synthetic phonics group was mostly learning phoneme awareness at this stage rather than phonics. The third study included was Johnston and Watson’s (2004) Experiment 2, and this also showed that synthetic phonics teaching led to much better reading skills than the analytic phonics method. (Reading and Writing: An Interdisciplinary Journal, 25, 2012).

Another 7+ years have passed since the Torgerson et al. review was published but there has been no ‘study of naturally occurring variation' of the type that Brooks thought might be needed (unless the 2012 study by Johnston, McGeown and Watson counts as one). And even if such a study had been done and had been good enough for Brooks, would it have been good enough for Davis? It seems that Davis himself has been for many years in a position where he could have been pressing for people to do the sort of research he favours - I wonder whether he has done this.

Shortly after the Torgerson et al. review appeared, Jim Rose wrote:...notwithstanding the uncertainties of research, there is much convincing evidence to show from the practice observed that, as generally understood, ‘synthetic’ phonics is the form of systematic phonics work that offers the vast majority of beginners the best route to becoming skilled readers. Amongst other things, this is because it teaches children directly what they need to know... whereas other approaches, such as ‘analytic’ phonics, expect them to deduce them’ (Rose Review, 2006, para. 47).

The footnote to the paragraph refers to the definitions of synthetic and analytic phonics in the 2006 Torgerson et al. review, where a.p. is indeed defined as requiring children to ‘deduce’ or ‘infer’ letter-sound correspondences:

Torgerson et al. wrote:Analytic phonics refers to an approach in which the phonemes associated with particular graphemes are not pronounced in isolation. Children identify (analyse) the common phoneme in a set of words in which each word contains the phoneme under study. For example, teacher and pupils discuss how the following words are alike: teachers and children discuss how the following words are alike: pat, park, push and pen. Analytic phonics for writing similarly relies on inferential learning: realising that the initial phoneme is the same in [certain words] children deduce that they must write that phoneme with the letter <p>. (p. 13).

Statements that phonemes are not pronounced in isolation in a.p. have always puzzled me. If /p/, for example, is not pronounced in isolation by teachers or children, how do teachers know that children have identified exactly the right sound? In any case, I think that phonemes probably were pronounced in isolation in the a.p. condition in the Clackmannanshire study.

The definitions of s.p. and a.p. used by Torgerson et al. are those used by Brooks in his 2003 paper, which, in turn, are based on those of Strickland (1998). Strickland, by the way, is not an s.p. advocate (see her ‘Afterword’ to M.J. Adams’s 1990 book Beginning to Read), so Davis is wrong to imply that the tendency to use terms such as ‘phoneme’ loosely is a particular weakness in s.p. circles.

S.p. has had government backing since 2007, with Letters and Sounds embodying the recommendations of the Rose Review and made available to schools not already using, or prepared to start using, other suitable programmes. Mixed methods obviously continued, however, and results continued to be mediocre at best. I suspect that it was because of this that it was decided to offer matched funding and introduce the phonics screening check.

Jenny C.

I tried to post this on TES last week - without success:

Andrew Davis: Is it reasonable to ask you how much knowledge you have of teaching synthetic phonics? Have you even visited schools in your area, or further afield, where effective synthetic phonics is taught? Have you carried out a comparison with schools using multi-cueing strategies- phonics as an adjunct to Whole Language or mixing synthetic phonics with analytic phonics? What is of interest is the knowledge/data you have of schools whose teachers possess the professional knowledge to teach our alphabetic code systematically - ie synthetic or linguistic phonics trained teachers?

In London, the borough of Barking and Dagenham came 32 out of 32 boroughs( their senior advisor disapproves of synthetic phonics - within the last few weeks, Ofsted highlighted the inadequate performance of Barking and Dagenham schools}. To give you just two examples of authorities antithetical to synthetic phonics : Cambridge,, the primary school most hostile to synthetic phonics produced the lowest results in the City for SATs 2; Oxford, a city that eschewed Synthetic phonics for years, performed dismally in Sats 1 and 2.

In effect, lack of robust Synthetic Phonics teaching means that thousands of children can read only 2,000 + words. They are left bewildered by their lack of understanding about 'how reading works' and have their life-chances destroyed. The reason for the 10 minute Phonics Check is precisely to ensure that thousands don't drop through the net.

The widespread use of Synthetic Phonics in Tower Hamlets and Newham significantly improved their results over the last decade. The ARK academies use evidence- based instruction - in this case Synthetic Phonics for literacy – and has outstanding results. And so on…

Perhaps you would be prepared to name 50 or 100 schools, with high percentage of children on FSM, with good Sats 2 results having experienced instruction you approve of?

"Perhaps you would be prepared to name 50 or 100 schools, with high percentage of children on FSM, with good Sats 2 results having experienced instruction you approve of?"

A fair enough question Geraldine,I look forward to Andrew Davis's reply.

It's a pity you were unsuccessful with your message, Geraldine. Do you know why?

Just one thing: are 'Sats 1' and 'Sats 2' recognised abbreviations? People might understand better if you called them 'Key Stage 1 Sats' and 'Key Stage 2 Sats' to make it clear that you don't mean Levels 1 and 2 in the Sats.

Jenny C.

Earlier in this thread, Maizie referred to a TES thread: http://community.tes.co.uk/tes_opinion/ ... 2219857=26

I’ve been following this on and off. One thing that occurs to me is that it’s very easy for people to be at cross-purposes in debates of this type. It seems possible that Msz and Andrew Davis are at cross-purposes on the matter of how real words are used in Sounds-Write – perhaps things will become clearer if the debate continues.

Jenny C.

Toots wrote:

chew8 wrote: Good decoding is an easy skill to acquire in most other alphabetically-written languages because the first sound which children are taught to associate with each grapheme is usually the sound which will apply in all words they encounter from then on. This is clearly not the case in English, but isn't the sensible solution to take extra time and care over the teaching of decoding in English? That could ensure far less prevalence of the problem that ‘failing to decode accurately stands in the way of reading for meaning’.

What does this extra time and care consist of? There are certain problems in English orthography which will not be solved by extra emphasis on phonics for longer. It's not a case of there being more to learn than in shallow orthographies but a case of there being some aspects that cannot be learnt through phonics alone. The differences between the orthographies is not a simple quantitative one, where there is more of the same to learn, but a qualitative one, where there are different things to learn about the language.

There are more possible grapheme phoneme correspondences in English than in some phonetically simpler languages aren't there? So there is more to learn than in some languages. Correct me if I am wrong please because I have not researched this.

There are more possible grapheme phoneme correspondences in English than in some phonetically simpler languages aren't there? So there is more to learn than in some languages. Correct me if I am wrong please because I have not researched this.

A hell of a lot more, Volunteer! Take Spanish, for example: in Spanish there are around between 22 and 24 sounds. It's difficult, as it is in English, to be completely accurate in this because a 'language' is comprised of a number of overlapping varieties. However, there is general agreement on this, just as there is general agreement that there are around 44 sounds in English. So, from the start, Spanish contains less than half the sounds there are in English. There are in Spanish about (something over) 30 ways of spelling the sounds; whereas in English, as we generally agree, there are around 175 common ways of spelling the 44 or so sounds.
While it's true to say that Spanish and Italian are fairly straightforward, as far as I know many other European languages are equally easy to teach and learn. Even if they are comprised of more sounds than, say, Spanish or Italian, the orthographic representations of the sounds are relatively, in comparison with English, transparent. In other words, unlike in English, there is generally one way of spelling a sound rather than many ways; and conversely, many spellings represent one sound and not many sounds.
This is precisely the reason why teaching successful decoding in English, the key to unlocking the whole curriculum, requires good teaching, and takes time and effortful practice.

Compare these two charts of Spanish and Danish (the latter being quite a complex code too):

http://alphabeticcodecharts.com/other_languages.html

...with the English alphabetic code and its spelling/pronunciation alternatives:

http://alphabeticcodecharts.com/free_charts.html

But, the English language has many words with even further code than we see on the charts, but such words can be addressed in a more incidental way or as 'common' words flagged up when required:

http://www.phonicsinternational.com/Pro ... nglish.pdf

The better we as teachers (and parents who in effect are 'teachers') know and understand our languages' codes and, in the case of the English language, all those more unusual spellings, the better teachers we can be.

That is why I have made a commitment to always provide free alphabetic code charts as I think that 'people' are entitled to such organised information about their language.

volunteer wrote:

Toots wrote:

chew8 wrote: Good decoding is an easy skill to acquire in most other alphabetically-written languages because the first sound which children are taught to associate with each grapheme is usually the sound which will apply in all words they encounter from then on. This is clearly not the case in English, but isn't the sensible solution to take extra time and care over the teaching of decoding in English? That could ensure far less prevalence of the problem that ‘failing to decode accurately stands in the way of reading for meaning’.

What does this extra time and care consist of? There are certain problems in English orthography which will not be solved by extra emphasis on phonics for longer. It's not a case of there being more to learn than in shallow orthographies but a case of there being some aspects that cannot be learnt through phonics alone. The differences between the orthographies is not a simple quantitative one, where there is more of the same to learn, but a qualitative one, where there are different things to learn about the language.

There are more possible grapheme phoneme correspondences in English than in some phonetically simpler languages aren't there? So there is more to learn than in some languages. Correct me if I am wrong please because I have not researched this.

It is not simply a case of English having more grapheme phoneme correspondences to learn, although that is true. It is also the case that some graphemes can represent more than one phoneme (in English) and some phonemes can be represented by more than one grapheme.
This means that applying phonics alone cannot reliably give you the correct pronunciation of many English words, but applying phonics can give you correct pronunciation of Spanish words. It doesn't matter how well you know the English 'code', or how many GPCs you know, this will remain the case. Something else beyond phonics is needed for accurately decoding English. This is what I mean by a qualitative as well as a quantitative difference between English, a deep orthography, and Spanish, a shallow orthography. It means that SP works well in Spanish, but not so well in English.

I’m sure none of us would deny that decoding and encoding present far more problems in English than in most other languages. Let’s stick to the decoding/reading, side of things, however, as that seems to be what is bothering the critics of synthetic phonics.

I think we would also all agree that context may have to come into play where ambiguity needs to be resolved: do I put the stress on the first syllable of ‘progress’ or the second? Do I pronounce ‘does’ to rhyme with ‘toes’ or ‘buzz’? Does ‘tear’ in the sentence I am reading rhyme with ‘dear’ or with ‘pear’? In hearing the reading of hundreds of young children over the years, however, I have found instances of this sort of thing cropping up fairly infrequently – certainly not enough to undermine my belief that the prime need of beginners is to learn to read words by sounding out and blending,

Jenny C.

Yes, it does occur, but as Jennie says it's extremely infrequent. Young children become v. adept at self correcting: if they say 'tearing /teering/ the paper, they quickly correct. For reading, it's only the c.l% who need v.v. gradual exposure to all 170+ Correspondences. Spelling may be another kettle of fish.

Spelling is another kettle of fish - it requires much more word-specific knowledge than reading. Again, though, I don't think that this would be disputed by synthetic phonics advocates, and the fact that they teach spelling systematically no doubt also helps with the word-specific knowledge needed for reading.

Jenny C.

toots said:

It is not simply a case of English having more grapheme phoneme correspondences to learn, although that is true. It is also the case that some graphemes can represent more than one phoneme (in English) and some phonemes can be represented by more than one grapheme.
This means that applying phonics alone cannot reliably give you the correct pronunciation of many English words, but applying phonics can give you correct pronunciation of Spanish words. It doesn't matter how well you know the English 'code', or how many GPCs you know, this will remain the case. Something else beyond phonics is needed for accurately decoding English. This is what I mean by a qualitative as well as a quantitative difference between English, a deep orthography, and Spanish, a shallow orthography. It means that SP works well in Spanish, but not so well in English.

I have to say, Toots, that you wouldn't be making statements such as the above if you took the trouble to find out exactly what good quality phonics programmes teach.
Of course it isn't simply a case of English having more sounds and more graphemes. And, yes, it the case that (not some but large) numbers of graphemes can represent more than one grapheme. Where you are wrong is in your assumption that 'applying phonics alone cannot reliably give you the correct pronunciation of many English words'. It can if (a) one has an extensive enough knowledge of the code, and (b) the word one is decoding is within one's lexical repertoire.
Bearing in mind that all words can be decoded because all words in English have been assigned spellings, even though the spellings may be less frequently encountered, it is obvious that good programmes teach an extensive code knowledge. However, that in itself is not sufficient. One also has to teach the necessary skills, which is why in our Sounds-Write programme, we, as well as teaching segmenting and blending skills, teach phoneme manipulation, the ability to take out a sound in a word and substitute another. For example, if a child reads 'steak' as 'steek' or 'stake', because <ea> can be /ee/, /e/, or /ae/. We teach them to be able to take out the wrong one and put in the correct one. This, as Jenny pointed out above is done in the context of the sentence. Thus, in the situation in which the child reads 'Last night I ate a tasty steak' in one of the wrong ways, their brain is likely to be telling them that 'Last night I ate a tasty stek' doesn't make sense. Neither does, 'Last night I ate a tasty stake'. When those don't work/make sense, the child tries the other one. However, in order to be able to try the other, the child has to have been taught that <ea> can be /ee/, /e/ or /ae/ in the first place. And this is the task of a good quality phonics programme.
There is also the issue of conceptual understanding, which we think is enormously important. Children need to know that many spellings can represent different sounds. Is this a problem? Of course not! Long before children enter school, they are already familiar with the idea that something can 'stand for' something else. So, conceptually, it is easy to point out that the letter <o> can be /o/ and it can be /oe/. Later, as their code knowledge improves/becomes more extensive, they'll learn that it can be /oo/ and /u/. At first, their efforts to perform this skill, which embraces all these other aspects, is clunky. But, like all skills, the more practice the young learner gets, the better they become. As they read more and gain more experience of reading, their brains become better and better at spotting patterns in the language, which is why my brain defaults to /o/ when I see the spelling <a> following the spelling /w/ ('wan', 'wasp', 'swan', etc, etc.) but instantly switches to /or/ in 'walk'.
This is why even fluent, highly skilled readers are able to read words (especially place names they have never seen before) in a number of different ways and, in the case of a place name, then ASK a local what it's called. In Milton Keynes, there is a place called 'Broughton'. In this instance, the <ough> spelling represents the sound /or/; whereas, a few miles up the road in Northamptonshire, there is also a place called 'Broughton', in which the spelling <ough> represents the sound /ow/. How did I know? By asking someone is this place name this, or is it that?
If the word isn't within a person's vocabulary, what happens then? Well, if they can't decode, they haven't a hope in hell of reading the word. If they can, they can make a reasonable stab at it even if some spellings are ambiguous. The word will be learned in context, which is how we learn so much new vocab, and the person can consult a decent dictionary to find out the pronunciation and stress, or ask a more knowledgeable other if the word needs to be in their spoken vocabulary.
Apologies for such a long-winded answer but the devil is in the detail.

The devil is indeed in the detail.

One detail not flagged up in your account is that if a reader has to use context and sense in order to decode a word, they are not using phonics. Another is that if a reader has to ask a local or consult a dictionary about pronunciation they are not using phonics. I have no problem with either strategy, which are necessary, but you can't claim them for phonics.

This is phonics - lifted from your website:

Our approach teaches the conceptual understanding needed to become an effective reader:

that letters are spellings of sounds: visual language is a representation of spoken language
that a spelling can contain one, two, three, or four letters - examples are: s a t, f i sh, n igh t and w eigh t
that there is more than one way of spelling most sounds: the sound 'ae', spelt as <a-e> in 'name', can be represented as <a> in 'table', <ai> in 'rain', <eigh> in 'eight', <ay> in 'play', and so on
that many spellings can represent more than one sound: <ea> can be the sound 'e' in 'head', 'a-e' in 'break', or 'ee' in 'seat'
Within this conceptual framework, we teach the factual knowledge required to become an effective reader and speller: the approximately 176 spellings that represent the 44 or so sounds in English, starting with the most simple one-to-one correspondences.

There is no mention of these other strategies (using context/ information from others) in this account. You have also missed out the detail (?) that, in order to be an effective reader you have to be able to understand the content of what you are reading, thereby equating decoding with reading. Additionally, you claim in your post that without decoding a person cannot read an unknown word. They can, actually, because context provides clues for meaning. Of course, in practice, a person coming across an unknown word will have a phonic stab at it and use the context, but there is nothing to say that were the word unpronounceable it could not nevertheless be understood and recognised at a later time or different location.

I would also question your thesis about the decoding skill readers bring to words such as 'was' and 'walk'. The brain switch you describe is just as likely to be a result of the reader recognising the word than an instantaneous process of recognising that after 'w' 'a' spells /o/ but not if it's part of 'alk' in which case 'al' has to be decoded as /or/. Many researchers have described the notion of a lexicon of known words, that are recognised on sight, however originally decoded. A self-teaching mechanism has also been described where the reader is able to decode new words not in the light of their phonic knowledge but in the light of their word knowledge.

These details show that learning to read and spell English is not simply a matter of exhaustively learning the GPCs. And one might say, "Thank goodness for that". The GPCs are difficult to learn, remember and apply when decoding, and can positively stand in the way of building up the fluency needed for ease of reading. So thank goodness there are other useful routes which compensate for the difficulties of English orthography.