One Cluster Held Three Children

What the research found

Researchers in Greece set out to answer an old question: is developmental dyslexia one condition, or several? They studied 101 children with a confirmed dyslexia diagnosis, all in the 3rd to 6th grades of primary school. The children were aged 8 to 12, with a mean age of 11.15 years. Sixty-three were boys and thirty-eight were girls. All were native Greek speakers, diagnosed through official state centres.

Each child sat a wide battery of assessments, eighteen tests in all, drawn from familiar instruments such as the WISC-V and various tests of attention, memory and movement. The tests were grouped into seven domains: phonological ability, attention, memory, motor skills, processing speed, visual skills, and visual-motor skills. Phonological ability refers to how a child hears and manipulates the sounds within words, long thought to be the core weakness in dyslexia.

The researchers then ran a hierarchical cluster analysis using Ward’s method. Cluster analysis is a statistical way of sorting people into groups so that children within a group resemble each other more than they resemble children in other groups. The technique does not know anything about dyslexia; it simply finds the natural groupings in the numbers. It produced three clusters.

The three groups were very uneven in size. Cluster 1 held 39 children, about 39 per cent, and showed weaknesses across the board: memory, attention, processing speed, phonology, visual-motor and visual skills. Cluster 2 was the largest, with 59 children, roughly 58 per cent, and showed a narrower profile of difficulty in memory, motor and visual-motor skills. Cluster 3 held just 3 children, under 3 per cent, whose difficulties sat almost entirely in the motor domain. Children in Cluster 1 performed significantly worse than those in Cluster 2 on five of the seven domains.

It is worth dwelling on those numbers. The headline is that three subtypes emerged. The quieter finding is that one of them contained three children. A group that small cannot carry much statistical weight, and the authors treat it cautiously. What the study more firmly shows is a split between children with broad, multi-domain difficulties and children whose difficulties are more contained.

The authors are careful about what this means. The study is cross-sectional, a single snapshot in time, so it cannot tell us whether these profiles are stable as children grow or respond to teaching. They make no causal claims. They also note that Greek is a transparent language, where letters map cleanly onto sounds, which may make decoding easier and push difficulties into other areas such as processing speed. They state plainly that findings in Greek may not transfer to English or other languages, and they call for research across different writing systems.

What it means for you and your child

The useful idea here is one you may already sense: dyslexia does not look the same in every child. This study puts structure around that intuition. Most of the children fell into two recognisable patterns, one with wide-ranging weaknesses and one more focused on memory and movement. If your child has a dyslexia diagnosis, this is a reminder that the label is a starting point, not a full description. Two children with the same diagnosis may need quite different support.

That said, be careful how much you ask this study to bear. It describes 101 children at one moment, in one language, sorted by a statistical method that will always return some grouping if you ask it to. The clusters are real patterns in this sample. They are not proven categories that every dyslexic child belongs to, and they have not yet been shown to predict which teaching methods will work. The tiny third group, with only three children, is best read as a caution against treating any of these subtypes as fixed.

The authors themselves draw the sensible conclusion. They argue for individualised intervention that responds to each child’s particular strengths and weaknesses, rather than a single programme applied to everyone labelled dyslexic. Notice that this recommendation does not actually depend on the three subtypes being correct. It follows from the simpler and better-established observation that dyslexic children differ from one another. A careful, individual profile of your child’s memory, attention, processing speed and phonological skills is more informative than any subtype name.

Practically, this suggests a few things. When your child is assessed, look for a breakdown across several domains, not a single score. Ask which specific skills are strong and which are weak, because the study’s real lesson is that the weaknesses cluster differently in different children. If a tutor or school proposes an intervention, the relevant question is whether it targets your child’s actual profile, not which “type” of dyslexia they have been assigned.

Finally, hold this study in proportion. It is a thoughtful piece of exploratory work that adds weight to the view that dyslexia is heterogeneous. It does not settle the subtype debate, and it was not designed to. Because Greek is far more phonetically regular than English, the specific profiles here may not describe an English-reading child well at all. Treat the finding as a lens for asking better questions about your own child, and let a full, individual assessment do the rest.