Few arguments in education have run as long or as hot. It has been called the reading wars, which is journalistic but not entirely unfair, and it has consumed careers.
The useful first move is to separate two questions that get argued as one: what the evidence shows about teaching decoding, and what should happen in a curriculum. The first is comparatively settled. The second is a question about values, priorities and time, and no amount of evidence settles it.
What the two positions actually were
In caricature, one side held that reading should be taught by explicit instruction in the relationships between letters and sounds, systematically, in a planned order. The other held that reading develops best through immersion in meaningful text, with children using context, pictures and sentence structure to identify words, and letter to sound knowledge picked up along the way.
The second position was not stupid, and it is worth understanding why intelligent people held it. It came from an accurate observation, which is that skilled readers do not appear to sound words out. It also came from a genuine concern that decontextualised drill would make reading joyless.
The error was inferring the beginner from the expert. Skilled readers recognise words instantly because they have automated a process they once performed effortfully. Watching the expert tells you where the learner is going, not how they get there.
What the evidence settled
Over several decades, reviews in several countries reached broadly the same conclusion: systematic explicit teaching of letter to sound relationships produces better word reading than approaches that leave those relationships implicit. This shows up in national reviews, in meta analyses, and in the evidence summaries maintained by bodies such as the Education Endowment Foundation.
Two qualifications are routinely lost.
The effects are concentrated on decoding. That is what phonics teaches and what it improves. Effects on reading comprehension are smaller and less consistent, which is exactly what you would expect given that comprehension depends on vocabulary and knowledge, as set out in how reading is learned.
And the effects are averages across trials, with a wide spread. That is a general property of educational research rather than a weakness of this finding, and it is discussed in effect size in plain words.
What the argument is now about
Very little of the current disagreement is about whether decoding should be taught explicitly. The live questions are elsewhere.
How much of the timetable. A programme can be well evidenced and still be given too much of the day, displacing the vocabulary and knowledge building that comprehension depends on.
What counts alongside it. Nobody serious argues for phonics instead of books. The argument is about the balance and about what is squeezed.
Fidelity and prescription. Programmes bought as packages tend to be implemented rigidly, which improves consistency and reduces the ability to respond to a particular child. This is a general tension in evidence based practice and not specific to reading.
Assessment. In England, the phonics screening check administered in Year 1, overseen by the Standards and Testing Agency, is a short assessment of decoding that includes invented words. Including invented words is methodologically deliberate, since a real word can be recognised without decoding. It is also the single most complained about feature, because it looks absurd to a parent.
Who decides. Under the argument sits a question about professional autonomy: whether a government should specify method, or specify outcomes and leave method to schools. That is a political question wearing an empirical costume.
Two mistakes to avoid
The first is treating the evidence for phonics as evidence for a whole curriculum. It is evidence about one component. A school that taught nothing but decoding would produce children who could pronounce text they did not understand.
The second is treating the limits of the evidence as a reason to reject it. That decoding instruction does not solve comprehension is not an argument against decoding instruction, any more than the fact that tyres do not steer is an argument against tyres.
What this means for a parent
Less than the volume of the argument suggests. Children learn to read in schools using a range of approaches, and most of them learn to read.
What is worth knowing is that the vocabulary and knowledge side is the part a home naturally contributes to, through conversation, books read aloud well above the child's own reading level, and talk about things that are not present. That is not a programme and it is not homework. It is the ordinary business of talking to a child, described accurately.
And if reading is a particular worry, the class teacher and the school's special educational needs coordinator are the people who can look at the specific child, which no article can.
Why the heat persists
Because the argument is not really about a method. It is about who is trusted to make decisions, about whether teaching is a craft or a delivery system, and about the fear on one side that children will fail unnecessarily and on the other that a rich subject will be flattened into drill.
Both fears are reasonable. Neither is resolved by another meta analysis, which is why more evidence has not ended the dispute and will not. Recognising which part of an argument is empirical is most of what it takes to follow it, and it is a habit we try to encourage throughout our evidence section.