Tuesday 12 August 2014

Synthetic phonics and irregular word reading: Cause for concern?

There have been numerous concerns about the introduction of a phonics focused synthetic phonics approach to the teaching of reading in England.  One of the concerns often highlighted is that a phonics focused approach does not suit the opaque (i.e., irregular) nature of the English writing system.  For example, words such as ‘aisle’, ‘pint’, ‘yacht’ are all irregular and pose difficulties for children learning to read.  Indeed, of all the alphabetic writing systems, English is one of the most irregular and research has shown that the ease with which a child learns to read is related to the orthographic depth of the writing system (Ellis et al, 2004).  Therefore learning to read English is challenging.

However, the argument that a phonics focused synthetic phonics approach is unsuitable to teach children to read English is an argument from which I have seen no research evidence.  Examples have been given of children being unable to read ‘high frequency’ words such as ‘the’ – sounding out ‘t’’h’’e’ then blending them together and forming an inaccurate pronunciation.  However, to me, this is better than I’ve seen prior to the introduction of synthetic phonics, where the first three words of the reading test that I used to administer (British Ability Scales II) were ‘the’ ‘up’ and ‘on’, and read by a number of students as ‘Biff’ ‘Chip’ and ‘Floppy’.  Surely teaching children about the alphabetic nature of the writing system, illustrating that there is a relationship (albeit not perfect) between the letters and sounds, is better?

In my own research, I have found no evidence that relying on a phonological reading strategy impairs children’s ability to read irregular words.  In fact, I have found the opposite.  In two studies briefly described below, I have examined a) the extent to which relying on a phonological reading strategy influences irregular word reading and b) the skills supporting children’s irregular word reading.

In the first, we found that children aged 6 -8 who took a more phonological approach to reading (i.e., relied more heavily on using phonics rules to read) performed better on assessments of irregular word reading.  In this study, irregular words were selected to be unfamiliar (i.e., low frequency), thus requiring a strategy to read them as opposed to immediate recognition.  This study was carried out with 172 children who varied in their strategy use and there was clearly a very strong relationship between dependence of a phonological reading strategy (use of phonics rules) and performance on assessments of irregular word reading (r = .66, p<.001) and standardised assessments of reading (r = .65, p<.001). 

In the second study (McGeown, Johnston & Moxon, 2014), among 180 children aged 6-9, we found that children’s nonword reading skill (ability to decode using letter-sound correspondences) was a very strong and significant predictor of their ability to read irregular words.  Indeed, it was a stronger predictor than their frequency of reading or language skills, suggesting that direct instruction in decoding skills may be appropriate to support children’s ability to read irregular words.

So why would this be the case?  Rather than categorising regular and irregular words as different word types, my collaborators and I have argued that even irregular words contain regular elements that provide a cue to pronunciation.  This is the essence of Seidenberg’s theory of quasi-regularity (Seidenberg, 2005).  For example, words such as ‘aisle’ ‘pint’ and ‘yacht’ are not completely irregular and a child with good phonics knowledge and using a phonological reading strategy will be better able to make use of the regular elements of irregular words to read.  It will not be a perfect strategy – the English writing system will not allow it; however this extra cue appears to be helpful and supportive as children face the challenging task of reading irregular words.


 Ellis, N. C. Natsume, M., Stavropoulou, K., Hoxhallari, L., Van daal, V. H. P., Polyzoe, N., Tsipe, M-L., & Petalas, M. (2004). The effects of orthographic depth on learning to read alphabetic, syllabic, and
logographic scripts.  Reading Research Quarterly, 39, 438-468. doi:10.1598/RRQ.39.4.5.

 McGeown, S. P., Johnston,R. S., & Moxon, G. E.  (2014).  Towards an understanding of how children read and spell irregular words: the role of nonword and orthographic processing skills.  Journal of Research in Reading, 37, 51-64.  doi:10.1111/jrir.12007. 

McGeown, S. P., Medford, E., & Moxon, G. (2013).  Individual differences in children’s reading and spelling strategies and the skills supporting strategy use.  Learning and Individual Differences, 28, 75-81. doi: 10.1016/j.lindif.2013.09.013.

 Seidenberg, M.S. (2005). Connectionist models of word reading. Current Directions in Psychological Science, 14, 238242. doi: 10.1111/j.0963-7214.2005.00372.x


  1. Many thanks for this posting, Sarah, I've flagged it up on my PI forum here:


  2. I would suggest, Sarah, that we would all be better off dispensing altogether with notions of 'regular' and 'irregular'. 'Regular' really refers to one-to-one sound-spelling correspondences, after which everything else appears to fall into the category of 'irregular'. Actually, if we adopt the approach there are sounds (44 or so - you have forty-five in Scottish English!) and there are spellings (probably around 175 common spellings). Some spellings, such as ai for the sound /ie/ in 'aisle' are simply less frequently encountered but then we probably wouldn't be teaching it, or 'yacht' to very young children.
    The code, as you say, is complex, which is why it needs teaching from simple to more complex, starting with the one-to-ones. However, if we firmly anchor out teaching in the sounds of language (which children learn naturally) and teach children that the squiggles on the page are representations of those sounds, we have a system which is immediately understandable from a psychological point of view.
    All we need to do then is to teach it systematically and cumulatively, teaching children: conceptual understanding [We can spell sounds with one, two, three or four letters; that sounds can be spelt with more than one spelling; and, that many spellings represent more than one sound.]; the three key skills of blending, segmenting and phoneme manipulation; and all 175 spellings sorted and categorised according to sound.
    It should take around three years of schooling, at the end of which, in our experience at Sounds-Write, more than ninety per cent of children are already highly proficient readers and spellers, preparing them adequately for 'reading to learn'.
    I've written about how to do this a number of times on my blog, The Literacy Blog, which you can look at here http://literacyblog.blogspot.co.uk/, if you're interested. Sounds-Write have also trained nearly twelve thousand teaching practitioners in England, Wales, Northern Ireland and the Irish Republic, Australia, Zambia and a number of other countries, although we've never trained anyone in Scotland!

  3. Thank you John, I fully agree with what you have written. I also agree that making a distinction between regular and irregular words can be unhelpful; however the irregular nature of the English writing system is often used as an argument against phonics teaching. I wished to illustrate that this need not be the case. You may be familiar with different models of word reading – I believe that there is more research support for the quasi-regular model (i.e., the idea that the words vary in their regularity) than the dual route model (i.e. the idea that regular and irregular words are distinctly different word types). Anyway, was very interested in what you wrote and will certainly start reading your blog.

  4. Thanks, Sarah.
    I'm not at all convinced by the dual-route model myself. In fact, I've recently been reading Ashby and Rayner's 'Literacy Development: Insights from Research on Skilled Reading', which, I think undermines the case for the dual route.
    I wonder, too, what you thought of Sweller's 'Human Cognitive Architecture' and how his (and van Merrienboer, Paul Kirschner , et al) ideas can be applied to the teaching of reading. I've made a tentative stab at this in my latest blog post.

  5. I'm not familiar with these John, thanks for drawing my attention to both - will certainly look at. Working my way through your posts (very interesting and enjoyable to read!) so hope to get to that one soon.