Many thanks to Mandy Nayton and colleagues from Australia’s Dyslexia-SPELD Foundation for organising a fantastic conference. Three packed days, more than 500 delegates and 84 speakers, all united by a common theme: what are the factors that underpin children’s language and literacy development and how can this knowledge be used to build effective education for all children?
I was fortunate to have the opportunity to present two talks. In my keynote I talked about reading comprehension and in particular, the oral language foundation that supports reading comprehension and its development. My slides are here. I began by setting out that while reading comprehension is clearly multi-faceted and complex, a big stumbling block for many children is with reading words. Put simply, if a child can’t access language via print, reading comprehension will suffer. That being said, everyone knows that there is much more to reading comprehension than being able to read words. In case anyone needs convincing, consider the case of poor comprehenders, the term used to describe children who appear to read well (they read aloud with age-appropriate levels of accuracy and fluency) but nevertheless struggle to understand what they have read (for review, Nation, 2005). For these children, difficulties with reading comprehension are not a downstream consequence of poor word-level reading. By studying poor comprehenders we can therefore examine the cognitive and linguistic factors that underpin reading comprehension, relatively undisturbed by factors that influence word reading accuracy.
To highlight a few salient points, based on follow-up discussions and questions:
1. Memory and language comprehension. I discussed a paper by Hua and Keenan (2014) which for me beautifully highlights the complex interplay between verbal memory and language comprehension. In this elegant and simple experiment, children read short stories. After each, they were invited to retell what they could remember and after this, they were asked a series of questions about the story. Some of the questions required an inference to be made. Others were literal, in that the answers could be found directly in the text.
All children found inference questions harder than literal ones and not surprisingly, poor comprehenders answered fewer questions correctly than skilled comprehenders. Comprehension performance was intimately tied to story retell in two interesting ways. First, inference questions placed a high load on memory: to answer inference questions correctly, more premises from the story needed to be recalled than for literal questions. Hua and Keenan found that if the relevant premises had been recalled, children were equally able to answer inference vs. literal questions. Second, poor comprehenders were worse at story recall: they remembered fewer premises than the skilled comprehenders. However, when the key premises were recalled, the poor comprehenders were able to answer the questions as well as the skilled comprehenders — be they inference or literal.
So, comprehension is constrained by text memory. Or is it? The important question now becomes: why is it that poor comprehenders are less able to remember key elements of the story? A likely (in my view) answer to this question is that they fail to remember it because they failed to understand it properly. Poor memory for verbal information is poor language is poor verbal memory. What’s chicken and what’s egg is far from clear, as discussed in some of our previous papers (e.g., Nation et al., 1999; Pimperton & Nation, 2010). On this view, reading comprehension is intimately linked to verbal memory as both are dependent on proficiency in oral language and the differences in knowledge that this brings about.
2. Oral language in children with poor reading comprehension. Here I considered the question of cause and effect. We know that poor comprehenders (selected on the basis of poor reading comprehension, despite age-appropriate levels of word reading) tend to show low levels of oral language. Listening comprehension and vocabulary for example, tend to be below age-expected levels. Plausibly however, this might be a consequence of their poor reading comprehension, rather than a cause. If poor comprehenders read less, downstream consequences for oral language are to be expected, especially when children move from learning to read to reading to learn. I described a retrospective study that helps unpack cause and effect (Nation et al., 2010).
We had the opportunity to follow a group of children from reception through to Year 3. Having identified those children who fitted the poor comprehender profile at that time point, we looked backwards in the data to see what their language skills were like before they learned to read. If limitations in oral language constrain reading comprehension development, there should be evidence of language weaknesses before the onset of reading comprehension failure. If however children grow into their oral language weaknesses because of poor reading, these should emerge over time, rather than being present from the outset. The data in our study were clear in showing that children who went on to be poor comprehenders (in terms of poor reading comprehension) showed oral language weaknesses (relative to control children) at school entry and onwards. It’s important to note that this study is quite small, and that differences in oral language were mild-to-moderate rather than severe. However, the patterns replicate those seen in other longitudinal datasets (Catts et al., 2006; Elwer et al., 2013). The conclusion from these studies seems clear: low levels of oral language earlier in development put children at risk of reading comprehension failure down the line. In turn however, it is highly likely that poor reading comprehension will itself then lead to reduced opportunities to learn vocabulary and language via reading. In the words of Keith Stanovich, a Matthew effect where the rich get richer and the poor get poorer.
3. Implications for intervention. If oral language weaknesses place children at risk of later reading comprehension failure, a strong prediction follows: intervention that improves oral language should also lead to improvements in reading comprehension. This is an important topic that deserves its own blog. In the meantime however, suffice it to say that there is evidence from RCTs to support this prediction (see Clarke et al., 2010; Fricke et al., 2013).
4. Finally, I briefly considered the question of how early to go. It’s perfectly possible to measure language in infants and toddlers. Can doing so help us identify which children might be at risk for later difficulties with reading and thus highlight the way for targeted intervention in the early years? In our own study, we found a relationship between vocabulary assessed in infancy and later reading and language proficiency in primary school (Duff et al., 2015; see here for overview). Importantly however, this relationship was quite modest suggesting that language skills are not stable in the early years. Some children are late talkers yet catch up in the pre-school years; others might make a good start but then plateau. What is clear is that children who start school with low levels of oral language are at severe risk for reading difficulties. Targeting intervention at oral language at this time point makes sense, and can be effective.
Early identification and targeted intervention were hot topics for discussion at the DSF conference and there was clear consensus on the continuities between low language and risk for reading failure. More generally, the drums were banging for evidence-based education, nowhere louder than in a symposium discussing this in the context of how to improve Australia’s tail of underachievement where Pam Snow reminded us of the term eudugenic failure, as discussed by Greg Ashman.
My other talk was about word reading development and the factors that promote children’s transition from novice to expert. Some of the theoretical background to this talk is described in this short paper. We are looking at how children acquire new words via reading experience and asking what promotes optimal learning. Our work is in its infancy and we’ll be reporting more about it in a future blog.
In sum, a fabulous conference: search #DSFconf on twitter to learn more.
**Some of the references cited are open access and freely available but some aren’t. Please email me directly via my .ox.ac.uk address if you’d like a copy of any of the papers cited in this blog.