Guest Blog: Letter Position Processing and Learning to Read
With only 26 letters to represent all possible words, it is inevitable that words overlap considerably with each other. The more words that children can read, the more likely it is that a word will overlap in letters with other words, meaning that sensitivity to both letter identity and letter position is critical. Relatively little is know about how children learn to process letter position information. We asked ReadOxford visitor Dr Yvette Kezilas to tell us more.
Reading a passage of text requires the mind to be somewhat sensitive to the position of letters within words. Without sensitivity to letter position information, words would frequently be misread. Stop might be misread as “pots”, “post”, “tops” or “opts”, making the experience of reading unenjoyable at best, and incomprehensible at worst. Whilst letter position processing needs to be sensitive enough to enable the fine discrimination between words that look similar, words like smile and slime for example, it is also flexible, enalbing tehse wrods to be raed with apparenlty litlte congitive effrot. This fine balance between precise and flexible letter position processing is thought to be a hallmark feature of skilled reading.
Mastering this balance between precise and flexible letter position processing does not always come easily. It’s normal for young children to occasionally shuffle letters when they read, but some children do so far more frequently than their peers. This reading profile is known within the research community as “letter position dyslexia” (LPD), and is thought to be caused by a hyper-flexibility in the processing of letter positions within words (Friedman & Rahamim, 2007; Kezilas, Kohnen, McKague, & Castles, 2014; Kohnen, Nickels, Castles, Friedmann, & McArthur, 2012). As a result, children with LPD frequently misread anagrams. For example, they might misread diary as “dairy”, or vice versa. This can compromise comprehension, if the word diary is interpreted as “something that comes from a cow”.
At present, there is a lot of confusion as to what exactly goes wrong in LPD. This is because it is yet unclear as to how children in general process letter position, and how this might change as they transition from novice to skilled readers. In fact, there are two somewhat contradictory developmental patterns within the literature. Some research suggests that letter position processing in typical readers becomes less flexible with development (Acha & Perea, 2008; Castles, Davis, Cavalot, & Forster, 2007; Kezilas, McKague, Kohnen, Badcock, & Castles, in press; Perea & Estevez, 2008); whereas other research suggests that letter position processing becomes more flexible with development (Lété & Fayol, 2013; Ziegler, Bertrand, Lété, & Grainger, 2014).
The aim of my research is to resolve these mixed findings within the literature, and hence to clarify how letter position processing typically changes over the course of reading development. The hope is that this information will then provide us with critical insights into how and why children with LPD struggle to deal with letter position.
One potential reason for the mixed findings is that changes in letter position processing over the course of development are driven by changes to more than one component of the reading system. Most studies that have found a developmental change in letter position processing have concluded that this is driven by a change in the visual encoding of letter position information (e.g., Perea & Estevez, 2008; Ziegler et al., 2014). However, visual encoding does not occur in isolation. In fact, the reading system is highly dynamic – not only does visual encoding have ‘bottom-up’ effects on later processes in the system such as word retrieval, but later processes also have ‘top-down’ effects on the visual encoding of text (e.g., Reicher, 1969; Wheeler, 1970; see here for discussion of the word-superiority effect, nicely illustrating top-down influences). It could therefore be the case that changes in letter position processing over the course of development are driven, at least in part, by changes in components of the reading system that occur after the initial encoding of letter position information.
A specific hypothesis that we have recently tested is that changes in letter position processing might be influenced by changes in the size of a child’s written word vocabulary. When a child encounters a word for the first time (e.g., cat), they must sound out the letters that make up the word in order to retrieve its meaning (e.g., /c/-/a/-/t/). As the child becomes more and more familiar with the word, an orthographic representation (or memory trace) for the written word is formed, and the word can then be recognised more automatically. As children’s reading vocabulary increases, the letters (and position of letters) within a reader’s orthographic representations must become better specified in memory to ensure that words do not get confused with one another.
One possibility is that older readers code letter position more accurately than younger readers because they have well-specified orthographic representations that they can draw upon to aid them in accurate letter position coding. To test this hypothesis, we looked at how children and adults code letter position within words (e.g., slime), and pseudowords (e.g., blire; see Kezilas, Kohnen, McKague, Robidoux, & Castles, in press, for a full description of the task). If a person’s orthographic representations assist them in letter position coding, then letter position should be coded more accurately for words (for which people have orthographic representations) than for pseudowords (which do not have orthographic representations).
In our study, we found that children were equally good at coding letter position in words and pseudowords, indicating that their orthographic representations had no influence on their ability to code letter position. In contrast, we found that adults were far better at coding letter position information in words than in pseudowords. These findings indicate that changes in letter position processing observed over the course of development are driven, at least in part, by the development of well-specified orthographic representations.
The results from this study are just one small piece of the puzzle. By continuing to isolate the various components of the reading system that are thought to be involved in letter position processing, I hope to provide a clearer understanding of how letter position processing develops in typical readers. Only once we have this information will we be able to fully understand why letter position processing develops atypically in children with LPD.
About the Author
Dr Yvette Kezilas completed her doctorate at Macquire University’s ARC Centre of Excellence in Cognition and its Disorders. She is now a Research Fellow in the School of Psychological Sciences at Melbourne University. She’s currently an Academic Visitor with ReadOxford, working on several experiments to explore the development of letter position coding further. With Ascen Pagan and Kate Nation, she’s exploring letter position processing in silent text reading, by monitoring children’s eye movements. With Yaling Hsiao and Kate, she’s using the Oxford Children’s Corpus to identify developmentally sensitive orthographic neighbourhood statistics. Her visit is made possible by study awards to Yvette from the UK Experimental Psychology Society and the CCD, and a grant to Kate from St John’s College. We thank the EPS, the CCD and St John’s College for supporting our collaboration.
Hi Yvette,
This is a very interesting and very important topic. Thank you for your posting.
You refer to development of letter positioning – and I was very interested in the observation that adults response to letter positioning in pseudo words did not match letter positioning in real words.
Could this be partly explained by changes in teaching approach over the years?
Your post does not refer to possible influences of teaching approach and content but I suspect that the teaching, not just the ‘development’ could be a significant factor in this topic.
I suggest that children form a reading (decoding) reflex not just as a consequence of development but as a consequence, at least in part, of the prevailing teaching when they are in the beginning stages of learning to read and later on – dependent not just upon the child’s natural tendencies (genetic and environmental) but also influenced by prevailing teaching methods (for example, whole word, whole language, mixed methods, systematic synthetic phonics with no multi-cueing guessing strategies).
Not only could the initial teaching of reading influence how children perceive and analyse the printed word, but also the prevailing practices of children’s reading as they years roll on.
For example, how much is the child left to read silently and without supervision?
How much is the child asked to ‘read aloud’ and how does the teacher step in to comment, or support, or teach – and what is the approach and professional understanding of the teachers, and anyone listening to the reading aloud as time marches on – to the decoding process?
We are all aware of the reading battles about the method and content of reading instruction and, I suggest, too often these are not considered when the emphasis in on ‘developmental’ processes.
For example, the professional knowledge, beliefs and practices of teaches of reading, and supervisors of reading, vary enormously to this day – even in England where, for example, systematic synthetic phonics has been promoted heavily for nearly a decade. Nevertheless, a three year survey by the NFER and the Reading Recovery practices based in the Institute of Education indicate that teachers do not share a common understanding nor common practices for teaching reading and supporting long-term reading.
You mention the decoding of pseudo words and this brings a case in point for illustrating how differently teachers in infant and primary schools (and others) differ hugely in their understanding and attitudes about the reading process. A number of teachers notably argued that their children who were ‘better readers’ made errors in pseudo word reading because they ‘wanted to make sense’ of the words and therefore read them as real words. This is linked directly to your topic of letter order as, in some words more than others, children reversed letter order of the pseudo words to say them as real words.
There are a number of explanations for this state of affairs. One is very simply that it is easy-enough to do this even as adults (noted in your posting differences between words such as ‘diary’ and ‘dairy’ are easy to understand when reading at a glance). Another explanation is that it is a feature of sounding out and blending the sounds (phonics decoding) that to attain the real word from this process, there is often a need to ‘tweak’ or ‘modify’ the pronunciation.
A further explanation, however, is that the ‘better readers’ may be using multi-cueing guessing strategies more routinely than phonics decoding to lift the words off the page, which leads to a ‘stab’ at words words are rather than a totally accurate ‘all-through-the-printed-word’ approach. I am suggesting, then, that children with a more multi-cueing approach (whether taught or by natural default) are more likely to be less attentive to letter order, and they are more likely to be less accurate readers – and possibly spellers too.
Sadly, this issue of reading reflex for beginner readers, and for longer term reading, is not fully thrashed out in teacher-education (I suspect). Nor is the issue of ‘word-skipping’ as a common factor of silent reading.
‘Word skipping’ whereby the silent reader can actually get the gist of a new word in print that is not in their spoken vocabulary, but the reader does not bother to decode the word to come up with a pronunciation (silently) – therefore cannot add that word to spoken language – is not one I hear ever, ever discussed other than when I seem to bring it up.
This issue of needing to give new words encountered in print (which are not in existing oral language) a pronunciation in order to add it to spoken language is a clear demonstration of the need to have phonics knowledge and skills for reading for life. Thus, phonics is not merely baby stuff, it is adult stuff, but adults don’t realise this – neither do most of the teaching profession and teacher-training profession (I suspect).
This means that any learner who is presenting as a weak reader generally, and perhaps there are genetic tendency-to-muddlement traits in the family, will no doubt have difficulties with letter order in words. And one of the answers to that is extremely high quality phonics teaching and not being taught to through multi-cueing reading strategies, which ironically are the prevailing strategies provided through many high-profile intervention programmes! So, the very children who most need to be kept from guessing through many cues are the same children who receive such provision.
Muddling letter order may well be a development issue when children are left to their own devices but could be addressed with the very best evidence-informed teaching.
Sadly, the adults themselves in the teaching and training fields are not yet on the same page.
Warmest regards,
Debbie
Sorry about typos – typed in a great hurry!