Slow Reading Was the Constant

What the research found

Languages differ in how neatly their spelling maps onto their sounds. Italian, Spanish, Finnish and Greek are “shallow”: letters and sounds line up closely, so a reader who knows the rules can sound almost anything out. English and French are “deep”: spelling is irregular, and the same letters can be read many ways. Researchers have long wondered whether dyslexia looks the same across these different systems, or whether a transparent language hides the difficulty.

This 2021 meta-analysis pooled the results of many separate studies to answer that question. Carioti and colleagues combined 79 articles, yielding 450 comparisons between dyslexic readers and typically developing readers of the same age. In total, 14,947 people were included, split almost evenly between those with dyslexia and those without. The studies covered both children (roughly 8 to 13 years) and adults (roughly 19 to 32 years), reading European languages of varying depth.

The first finding concerns accuracy, meaning whether words are read correctly. Here, the depth of the language mattered a great deal, and it interacted with age. Among children, the gap between dyslexic and typical readers was very large in deep orthographies (a standardised effect of about 2.38) and smaller, though still substantial, in shallow ones (about 1.53). In plain terms, English- and French-speaking dyslexic children fell further behind on reading words correctly than their Italian or Spanish peers did. Among adults the pattern softened and even reversed: the accuracy gap shrank in deep orthographies (about 0.83) and was somewhat larger in shallow ones (about 1.12). A standardised effect of this size describes how far apart two groups sit; values above 0.8 are conventionally called large.

The second finding is the one the authors treat as central. Reading speed, or fluency, separated the two groups everywhere. When reading was untimed, dyslexic readers were markedly slower (an effect of about 1.47). When reading was timed, the gap was larger still (about 1.82). Crucially, neither the depth of the language nor the age of the reader changed this much. The authors write that fluency is “the universal manifestation of reading deficits irrespective of age and orthographic depth,” and that reading speed “is a reliable index for discriminating between DD and control groups across European orthographies from childhood to adulthood.” Accuracy, in other words, can be masked by a transparent language or improved with age; slowness persists.

What it means for you and your child

The practical message is about what to watch. A child reading a shallow, regular language, or an older child who has had years of practice, may read words accurately and still be dyslexic. Accuracy can recover or be propped up by a forgiving spelling system. Effortful, slow reading is the more constant sign. If your child reads correctly but laboriously, one word at a time, or tires quickly and avoids reading, that pattern deserves attention even when accuracy looks fine on a spelling test.

It also helps to set expectations honestly. This study looked at European languages. English is among the deepest of them, which is why the accuracy gap in young English-speaking children here was so pronounced. That is not a reason for alarm so much as a reason to look early, because the difficulty is most visible in the years when accuracy is still being built.

Two cautions matter for how much weight to place on these numbers. First, this is a meta-analysis: it combines existing studies rather than running a new experiment, so it inherits their choices and their gaps. The authors are candid that the studies were unevenly distributed across ages and language types, especially for the timed-reading and naming-speed measures, and they say their results should be read “with caution” for that reason. They also drew only on work published between 2013 and 2018 and classified languages by depth alone, setting aside other features of writing systems. A different slice of studies might shift the figures.

Second, and more fundamentally, none of this establishes a cause. The meta-analysis shows a consistent difference between groups; it does not explain why slow reading occurs, nor does it prove that any single mechanism produces it. It cannot tell you what will help a particular child. What it offers is a well-supported description: across thousands of readers and several languages, speed was the thread that ran through every group. That is genuinely useful for knowing what to look for, and modest about everything else.

If you recognise your child in this, the sensible next step is a proper assessment by a qualified specialist, one that measures reading speed directly and not only accuracy. A single study, however large, is a signpost rather than a diagnosis. This one points clearly at one thing: watch the pace, not just the correctness.