This post began life in one form, and then further digging uncovered various things that meant I needed to rewrite it for clarity: there seem to be some misunderstandings circulating about the Phonics Screening Check, even among those (in fact especially among those) who are strong advocates for it.
Before I get into the details of the check, I should reiterate that I absolutely understand the value of an early and kindly-done assessment to make sure children are acquiring the skills they need for reading.
I also understand the value of placing restrictions on the scope of any test – but would note that thinking about which restrictions have been chosen can be instructive in relation to the assumptions of those creating the test and wishing to use its results.
In addition, I should say yet again that I see phonemic awareness, and an explicit focus on the building blocks of the sounds of our language, as a deeply valuable aspect of literacy teaching. Systematic Synthetic Phonics seems to me to have some very powerful characteristics which can scaffold the development of this awareness.
Ambiguity in the Check
1. Ambiguous graphemes
I was recently pointed to an article by Dick Schutz, Are “Leading Educationalists” Too Smart to be Dumb about Reading Instruction?. It attempts to counter arguments made by David Reedy, Andrew Davis, and many others together describing themselves as “a coalition of leading educationalists organized by the UK Literacy Association”, in an Open Letter .
The sections indented and in bold below are from the “Open Letter” and the italic represents his responses in each case.
With the pseudo-words, any plausible pronunciation is marked correct. So, for instance, children can decode “vead” in either of two ways: they can produce something rhyming with “bed” or with “seed” “veed” is not included in the Check, and none of the items actually in the Check have this ambiguity. “fot” is the first item in the 2013 Check, and children might possibly pronounce it as rhyming with “foot.” However, 97.1% of Yr 1 children pronounced the item consistent with the Alphabetic Code. 97.2% of Y2 children who re-took the Check pronounced the item consistent with the Code. The concern express by the coalition is nonsense, not the items in the Check.
However, with the so-called “real words”, the blend produced must match the sound of a real spoken word. Hence “blow” pronounced to rhyme with “cow” is unacceptable. Had “blow” not been classed as a real word, a response rhyming with “cow” would have earned a mark. “blow” has not been included in the administration of the Check. The Specifications for the construction of the Check preclude the posited ambiguity.
Schutz links to the Framework for the check, which illustrates how constrained it is. As a test of sounding and blending, of course, some constraints are logical.
However, the Check’s constraints do not exclude the ambiguous ‘ea’ digraph from the check: on p.12, section 3.2.2 there is a table which specifically lists it, with examples ‘head’ and ‘bead’. At the end of the table there are notes on the acceptable pronunciations of various words; the one for ‘head’ saying
In some regions the ‘ea’ in head is the same phoneme as in bead. The phoneme intended here is the same as the ‘e’ in bed.
There is the further note for ‘book’:
In some regions the ‘oo’ in book is the same phoneme as in room. The phoneme intended here is the same as the ‘u’ in put.
So although the Checks so far may not have included such ambiguities, the Framework assumes that at some point it will, and therefore the UKLA’s concerns in this regard would seem to be justified.
2. Ambiguous syllable stress
In a previous post, ‘Phonemic Capital’, I discussed the importance of syllable stress in creating a meaningful word out of a set of multisyllabic phonemes. Elizabeth Nonweiler commented to say that
There is no problem with which syllable to stress in nonsense words in the phonics check, as all the nonsense words have only one syllable.
I therefore removed the section I had written on this issue, because I thought it did not apply. However, I had not read her words carefully enough, not had I yet seen the detail of the Framework (the perils of wading in…), and in fact it turns out that this problem could indeed arise, since the Framework itself expects that it will. It states (p.9):
The two-syllable words assessed will be real words because of the difficulty of inventing polysyllabic pseudo-words with limited alternative pronunciations that can be scored reliably. This is an issue for two-syllable words because of the effects of stress placement on vowel pronunciation.
This passage needs unpicking. It is clear that it is expected or intended that two-syllable words could be included in the Check. In fact, the Framework states that they should be (p. 15):
Section 2 will contain..4 x two-syllable real words with different orthographical representations, one with five letters, one with six letters, one with seven letters and one with eight letters (4 words).
There seems to me to be inconsistency between Nonweiler’s assumption that excluding two-syllable words from the pseudo-words list will solve the problem, and the actual aims and content of the Check. The Framework does not see a binary distinction between pseudo- and real words. In fact it seems to be constructed on the assumption that it would be beneficial if children did not know the real words that they are to decode: that is, these real words would be no different from pseudo-words from the child’s perspective. The Framework says (p.8):
The real words will include between 40 per cent and 60 per cent less common words, which children are less likely to have read previously. Less common words are included so that the majority of children will need to decode using phonics rather than rely on sight memory of words they have seen before.
Therefore, it is possible within the remit of the Check that the two-syllable words could be unknown to the children, and that they would in those cases possibly be faced with ambiguity about stress. Further, in ‘Phonemic Capital‘ I discussed the issue of syllable stress in ordinary decoding (not just the Check), and said that in my experience with my own children, even with common, well-known words, syllable stress can be all-important to a child’s understanding of a word. The example I gave there was ‘began’, which is not a phonetically complex word, but which relies very much on correct stress for its pronunciation.
The notes on ‘head’ and ‘book’ show that it is expected that there is a correct pronunciation for these words: as Reedy et al. pointed out, this discriminates in favour of certain children for all sorts of complex reasons.
To me, these inconsistencies also raise the question: why have any real words at all? If what is being tested is a narrow mechanical skill, what is the rationale for muddying the analysis of results with real words?
In English, sounding and blending are not always in themselves enough to decode a word fully (see my earlier post in which I talk about ‘tweaking’ after blending, a need to which Debbie Hepplewhite in particular has drawn attention). If we think of GPCs as like Lego bricks, in English we always need to be aware that the colour of some of the bricks may only be decided as a result of the other bricks with which they are matched, and often as a result of which other groups of bricks precede and follow them, too.
The various notes and caveats in the Framework for the Check make it clear that ambiguity can be a problem in decoding English. Yet there is no mention of the only means by which some graphemes can be disambiguated: context.
So various points arise, from the details of the Framework document:
- The ambiguities of English orthography cause problems for creating a fair test of decoding words out of context.
- The only way of testing decoding according to the SSP definition would be to use purely single-syllable pseudo-words out of context.
- That is, the effectiveness of synthetic phonics as a method can only be properly tested by excluding a central feature of the English language: ambiguity of pronunciation (even to the exclusion of some very commonly-occurring words and GPCs).
If the check were a full test of children’s decoding of English, there would be no problem in including common ambiguous graphemes such as those in as bow/bow or read/read from the test.
The phonetically restricted nature of the check, and the ambiguities it acknowledges but does not solve, clearly demonstrate, to me, that synthetic phonics is not, on its own, enough for children to be able to decode. If all ambiguous graphemes were to be excluded from the test this would not solve the problem of their relative frequency in the language, as a result of which young readers are likely to (a) know their meaning and (b) come across them relatively regularly.
Although the Framework document mentions ambiguities of pronunciation and stress, it does not suggest any means by which children are supposed to disambiguate. So children are left untested on one of the core skills required to decode English: they are being asked, if they do not know the word, to take a guess at its pronunciation. Since guessing is something that SSP is designed to obviate, this is an uncomfortable situation.
If a test were to be created purely in relation to a child’s ability to recognize unambiguous phonemes and create word-like sounds out of them, this would clearly not be a full test of their decoding ability, but only of part of it.
So, what is the rationale for not testing the other necessary English decoding skills as well, such as the ability to choose the correct pronunciation – and hence the correct word – when reading GPC-ambiguous text? Why leave a child to guess? It seems especially inappropriate to do this to children who have been taught that ‘guessing’ a word is something that they absolutely should not do, and to ask them to do it in a situation where they are being tested by a potentially unfamiliar adult.
The words in the check are presented without context. It is very interesting if the actual content so far has excluded two-syllable words and ambiguous graphemes, since this would suggest tacit recognition that there are other things going on when a child (or anyone else) decodes such words.
None of us could know which pronunciation of bow/bow read/read object/object to choose without the guidance of the surrounding context of vocabulary and syntax, so by excluding context and also excluding words which require context for disambiguation, the check, if it has excluded these issues, implicitly recognizes that context is necessary for the decoding of certain words, even very common ones.
In order to decode ‘bow’ fully, a reader needs context. Is the actor taking a bow or tying a bow? The exclusion of such ambiguous words from the check seems to show that the check is not a test of the whole landscape of decoding ability, but merely one aspect of it. A very important (actually fundamental) aspect, obviously, but not the only one.
Given that the test seems tacitly to demonstrate the importance of context in decoding many English words, even commonly-occurring and otherwise simple ones, why does SSP appear to ban the use of context in teaching children to read, and why does the check not include a contextual element?
If it does not in fact exclude context, why do many people seem to think that it does? Note that the Framework document says that a phonic approach should be the ‘prime’ one, which implies that it might not be the only one:
Since this phonics screening check is a decoding check, only words that are phonically decodable have been included. It is expected that teachers will ensure that elements of early reading not assessed in this phonics screening check are also taught, such as reading and discussing books. The following statements indicate additional skills that children should possess by the end of Year 1 but that will not be included in the phonics screening check.
By the end of Year 1 children should:
- apply phonic knowledge and skill as the prime approach to reading unfamiliar words that are not completely decodable;
- read many frequently-encountered words automatically;
- read phonically decodable three-syllable words;
- read a range of age-appropriate texts fluently;
- demonstrate understanding of age-appropriate texts.
It is vital that children are given the opportunity to develop these skills throughout Year 1, in addition to developing the phonic decoding skills that are assessed in the phonicsscreening check.
The phrase ‘only words that are phonically decodable’ is interesting. How is ‘head’ phonically decodable to a child who does not know the word? Even if they do, how are they to know that there are not two possible pronunciations, as with ‘read’? And if a child ‘decodes’ the word with the correct pronunciation because they know it, they have done more than decode phonically: they have referred to their own store of vocabulary.
This seems to me to be a bit of a tangle, and there are some inconsistencies and ambiguities in the Framework, the Check, and the claims made about both (and about SSP too) which need a light to be shone onto them.
Answers, opinions and explanations gratefully received – I’m writing this blog in the spirit of enquiry and further understanding…. If an SSP practitioner could explain (unambiguously!) what the SSP orthodoxy is in relation to deciding between bow/bow, that might stop me blathering on about this. Maybe.