The Memory Gap Closes at Reading Level

What the research found

Children with dyslexia often perform poorly on memory tasks. They struggle to hold a string of numbers in mind, or to learn a list of words. For years this has fed a tempting idea: that a weak memory is one of the underlying causes of the reading difficulty. This study set out to test that idea directly, and the design is what makes it worth reading closely.

The researchers studied 48 Italian children in three groups of 16. The first group had developmental dyslexia, with an average age of about 13. The second group was matched to them on chronological age: same age, typical reading. The third group was the clever one. These children were younger, around 10 on average, but they had been chosen because they read at the same level as the children with dyslexia. This is called a reading-level matched design. If you want to know whether a difficulty is a cause of poor reading or a by-product of it, you compare children who read equally well but arrived there by different routes.

The logic runs like this. A child with dyslexia and a typically developing younger child who read at the same standard have, roughly, the same amount of reading experience. If the child with dyslexia still shows a memory deficit against that younger peer, the deficit cannot simply be explained by less practice at reading. It points to something more fundamental. But if the two groups look the same, the memory problem is better understood as travelling alongside reduced reading experience rather than causing the difficulty.

The results were consistent across every memory task. On short-term memory, the span tasks that ask a child to repeat back a sequence, children with dyslexia scored well below their same-age peers, roughly 7.1 against 8.8 items (p less than 0.001). That is the familiar deficit. But against the younger reading-level group, there was no meaningful difference at all (p equals 0.29). The same pattern held for verbal learning over time. The dyslexia group recalled fewer words than same-age peers, about 7.8 against 9.9 (p equals 0.003), yet matched the younger readers closely at 9.0 (p equals 0.45). On visual and spatial learning, the three groups did not differ from one another at all.

In short, the memory gap that appears so clearly against same-age classmates simply closed when the comparison was made against children of the same reading level. The authors are careful in their own conclusion, writing that children with dyslexia and typical readers did not differ in memory once matched this way, and that the causal role of memory in reading impairment has not been established.

What it means for you and your child

If your child has been assessed and a report notes a weak working memory or short-term memory alongside dyslexia, this study offers a useful and reassuring reframing. The low memory score is real. Your child was not imagined to struggle. But the score is best read as a marker that tends to accompany dyslexia, not as the hidden engine driving it. When you compare like with like on reading, the memory difference thins out to nothing.

This matters for how you interpret advice. Programmes that promise to fix reading by training memory in isolation, with computer games that stretch how many items a child can juggle, rest on the assumption this study weakens. If the memory difference is largely a shadow cast by reduced reading experience rather than its cause, then drilling memory in a vacuum is unlikely to move reading much. The more direct route to better reading remains structured, systematic teaching of the sounds and structure of written language. That is where the evidence base is strongest.

There is also a quieter point about how we describe children. A same-age comparison makes a child with dyslexia look as though they carry a broad basket of deficits: reading, memory, learning. A reading-level comparison shows something gentler. On these tasks the child is performing much as any younger child at the same reading stage would. The difficulty is more specific, and more localised to reading itself, than the raw scores against classmates suggest.

Two cautions are owed, and the authors state them plainly. This is a small study, 16 children per group, and small samples can mislead in either direction. It also captures a single moment rather than following children over years, so it cannot fully settle questions of cause and effect over the course of development. One well-designed study narrows the range of plausible explanations. It does not close the question. The authors themselves call for larger samples and longitudinal work that would track the same children over time.

What you can take from it is a sense of proportion. A memory score on a report is a data point, not a diagnosis of a separate problem to be fixed on its own. Read alongside your child’s reading level, it often tells a simpler story than it first appears to. The steady, evidence-led work of teaching reading directly remains the thing most likely to help.