This blog post is not another round in the well-known reading wars! Okay, now that that is out of the way, what is this new study actually about? Well, while there is quite a lot of evidence that explicitly and systematically teaching the connections between letters and sounds, or systematic phonics, is an important part of effective early reading instruction, there is something else to know. We know somewhat less about whether it also matters exactly how that phonics is taught.
The new study by Paul Gioia, Johannes Ziegler, and Jérôme Deauvieau, published in Learning and Instruction , investigates this question. What I really like is that, for once, the research takes place not in an English-speaking country but in France. That is no small detail, as French has a much more regular relationship between letters and sounds than English. Incidentally, the authors themselves warn that we should not simply generalise their findings to other languages and education systems.
In their study, the researchers distinguish between strict phonics and mixed phonics. What does this entail? With strict phonics, beginning readers only read texts that use grapheme-phoneme correspondences they have already explicitly learned. This means they do not have to memorise words whose letter-sound correspondences they do not yet know. With mixed phonics, teachers apply this principle less consistently. They may, for example, ask children to recognise or memorise words they cannot yet fully decode.
It turns out that this difference is quite real in France. The researchers analysed the 34 most commonly used reading methods. Together, these account for 95% of the methods used in French classrooms. The degree of text decodability ranged from 42% to 100%. The number of words children had to memorise before they had learned all the relevant letter-sound correspondences varied from 0 to 44.
In a next step, the researchers linked a large survey of teachers to the results of the French national assessments. The final analyses included 6,440 teachers and over 90,000 pupils. This allowed them to examine both reading fluency and reading comprehension halfway through the first grade and again at the beginning of the second grade.
And the results are remarkably consistent! Pupils who received strict phonics performed better than pupils who received one of the mixed variants. Midway through the first grade, the difference for reading fluency was approximately 0.11 to 0.16 standard deviations. For reading comprehension, it was 0.13-0.19 standard deviations. At the beginning of the second grade, the differences were somewhat smaller. But they were still present: 0.10-0.11 standard deviations for reading fluency and 0.08-0.10 for reading comprehension.
Experts will have immediately noticed that these are not gigantic effects. But do not forget what is being compared here. This is not phonics versus no phonics, hence the first sentence in this blog post. All groups received systematic phonics. There is another key point. It is about differences in how phonics is taught. And in that case, a difference of around a tenth to almost two tenths of a standard deviation is surely not negligible.
The researchers also examined whether the difference was the same for all children. As is so often the case in education, it was not. The weaker the children’s prior knowledge of the alphabetic principle at the beginning of first grade, the greater the advantage of strict phonics.
For reading comprehension, for example, the estimated difference between strict and mixed phonics among children with strong prior knowledge was approximately 0.17 standard deviations halfway through first grade and 0.09 at the beginning of second grade. However, among children with very weak prior knowledge, this rose to 0.28 and 0.21 standard deviations, respectively!
When the researchers examined the socioeconomic backgrounds of the schools, a similar pattern emerged. In relatively advantaged schools, for example, the difference in reading fluency between strict and mixed phonics was only 0.08 standard deviations halfway through the first grade and just 0.04 at the beginning of the second. In relatively disadvantaged schools, the difference was 0.17 and 0.23 standard deviations, respectively!
It therefore seems that more consistent phonics is associated with better reading performance. This link is strongest among pupils who start with the weakest pre-reading skills. But why would that be the case? The authors offer a possible explanation that sounds quite logical to me and reminds me of the cognitive load theory.
Suppose you are just learning to read. You first learn a number of letter-sound correspondences and subsequently receive only words and texts in which you can actually use that knowledge. The message you receive time and again is then quite consistent. Letters stand for sounds, and by combining those sounds you can read words.
With a mixed approach, on the other hand, a beginning reader can receive different messages at the same time. Sometimes you have to decode a word, but sometimes you have to recognise or remember a word even though you have not yet learned the necessary correspondences. That is not necessarily a problem for a child who already has enough prior knowledge. But… for a child who has just begun to master the alphabetic code, such an additional inconsistency could make it more difficult to develop a stable decoding strategy.
To be clear, the researchers have not demonstrated this in their study! It is what they see as a possible theoretical explanation for the observed pattern. Their study does not measure exactly which cognitive processes are responsible for the difference. We need further experimental research for that.
And it’s also important to keep in mind that this is an observational study; the researchers didn’t randomly allocate teachers to either strict or mixed phonics. It is therefore still possible that teachers who choose strict phonics differ in other ways from those who do not.
I think this is difficult to rule out, and also the researchers take this possibility quite seriously. For instance, they controlled for the pupils’ initial scores, class-level characteristics, the school’s socioeconomic background, and the teacher’s experience. They also examined whether strict-phonics teachers might simply spend more time on reading instruction. Instructional time did make a small difference. However, it did not explain the observed association. After controlling for instructional time, the difference remained.
They conducted yet another intriguing additional analysis. Maybe the teachers who intentionally picked a strict phonics approach were just a different type of teacher. Hence, among the group of strict phonics teachers, the researchers compared those who had chosen this approach on their own with those who had used it without choosing it. The researchers found no difference in their students’ reading achievement.
That makes the finding more robust, but of course it still does not turn the study into a randomised experiment. The authors themselves are clear about this as well: no firm causal conclusions should be drawn from this study. But it is certainly interesting!