A whole new language

A highly politicised debate has been raging since the 1980’s about the relative merits of “whole-language” versus “phonics” in literacy teaching. Despite the best efforts of some to convince us that the debate is over, and that we are heading towards a synthesis embodying the best ideas from both sides, the debate is still very much alive.

It is a mystery why the debate should have become so politicised. What works best for young readers should be a scientific rather than a political question. This article is about looking at the debate from the scientific perspective.

For those new to the terminology, “phonics” refers to an approach based on first teaching children the connections between the sounds of words and the way they are spelled. Modern phonics is a systematic development of traditional approaches which teach the child to sound out written words. This works because most English spelling is approximately phonetic.

Whole-language methods were first developed in the 1970’s, and started to become widespread in the 1980’s. The essence of whole-language is its treatment of the written language as a new language that can be learned by directly associating whole printed words with whole spoken words and their meanings. The original advocates of whole-language in the 1960’s and 1970’s drew on the theories of the eminent American linguist Noam Chomsky. Chomsky argues that children are born with an innate knowledge of spoken language structures.

Millenia of human experience have shown that children do not need direct instruction to learn a new spoken language. What works best is immersion in a helpful, stimulating environment where the new language is used constantly. The whole-language movement thought that learning to read should be as natural a process as learning to speak. The key idea was that children would “construct” for themselves the knowledge they needed, given the right environment.

In its pure form, whole-language completely bypasses systematic phonics instruction, working on the assumption that students will work out for themselves the phonic relationships between letters and sounds, if that is the way that particular student needs to construct his knowledge of written language.

The whole-language approach also brought with it a rethink of the traditional teacher-student relationship, one that rejected rote learning in general. Phonics, based on mastering a set of sound-letter associations explained and reinforced by a teacher, was seen as an outdated teaching method. The whole-language advocates linked up with a broader “progressive” movement in education which questioned all teaching methods based on authority or repetition.

The movement has drawn its inspiration largely from big picture ideas in philosophy and linguistics, which are stated at a high level of abstract generality. That gives you considerable latitude for development of a program. And so it has turned out that, in practice, whole-language utilises a number of techniques and there is a lot of emphasis on the creativity of individual teachers and students in developing personalised approaches. Some whole-language supporters will recommend the use of phonics-style techniques embedded into a whole-language program, whereas others reject phonics totally. They all seem to agree on one thing though; phonics should not be seen as the essential foundation. At best it is one technique among several.

By the mid 1990’s whole-language had supplanted phonics as standard practice in most schools in the English speaking world. As it did so, it started to attract criticism. A body of evidence was accumulating to indicate that despite increased resources and supposedly better teaching methods, the literacy performance of Australian schoolchildren was not getting any better, and was even, by some measures, getting worse. Similar concerns emerged in other English speaking countries, including the USA, where the trend had started.

In Australia, things came to a head in 2004 following publication in The Australian of an open letter of concern from 21 academics, criticising the widespread use of whole language methods in Australian schools. This led to the then education minister commissioning an inquiry in November 2004. The committee was chaired by Dr Ken Rowe from ACER, and included a number of eminent literacy researchers and practitioners.

Their brief was to consider the merits of whole-language and phonics in literacy teaching. They reviewed and summarised relevant empirical evidence on literacy teaching from all over the world, placing extra weight on studies using randomised controlled trials.

The report was published in December 2005. It came straight to the point:

The evidence is clear,… that direct systematic instruction in phonics during the early years of schooling is an essential foundation for teaching children to read. (Executive Summary, p1.)

Turning to the question of whole language, the committee went on to say:

The Inquiry found strong evidence that a whole-language approach to the teaching of reading on its own is not in the best interests of children, particularly those experiencing reading difficulties. Moreover, where there is unsystematic or no phonics instruction, children’s literacy progress is significantly impeded, inhibiting their initial and subsequent growth in reading accuracy, fluency, writing, spelling and comprehension. (p. 2)

Rather than further describe the large body of empirical evidence referenced in the report to support these conclusions, we will look instead at learning to read from an information processing perspective.

Most researchers working from this perspective have concluded, like the Rowe inquiry, that the whole-language approach is more difficult for the child.

Why? Let’s begin at the beginning.

Most children come to school at about the age of five or six already fluent in spoken English. They have already learned the associations between the sounds and the meanings of anything up to about 5,000 root words. That is about 5,000 “pieces” of information, where a piece of information is just the association between one sign (the sound of the spoken word) and another sign (the meaning of the root word).

This is an oversimplification of course. In reality there are many complexities. A single sound can have multiple meanings, and a single meaning can have multiple sounds, and as linguists and philosophers will tell you, meanings themselves are anything but simple. Nevertheless, this number of word-meaning associations gives a rough benchmark for the complexity of learning a new language.

Now consider a six-year-old learning to read English using a whole-language approach.

The emphasis is on reading whole words in a meaningful context, so the child will not be drilled in the ways particular letter groups are typically pronounced. In some whole-language programs, there will be an “embedded” phonics component. This is an indirect approach to phonics which will not usually occupy too much time, especially at the front end of the program. However, the front end is precisely where we should expect phonics to be most effective, as we will see.

As you would expect, students vary in the learning strategies they adopt. Some quickly pick up the rules relating letters and sounds by induction from examples. They start to do what whole-language says they should: construct their own model of the relationships between letter groups, sounds, and meanings.

But not everyone does this easily. Struggling students may construct only a fragmentary phonics model or maybe even none at all. In these cases it could mean that in effect the student has to learn a whole new language - about 5,000 new pieces of information, as he is not making proper use of the information that written words in English have an internal structure which can be correlated with their phonetic structure.

It is possible to learn to read this way. There are countries where the written language is not phonetic, like China. Many children learn second languages at school. This shows that learning 5,000 new associations is not impossible for normal kids.

Not impossible, but nor is it easy. It is particularly hard for disadvantaged children.

Now consider a child learning to read using phonics. Here the idea is not to learn a whole new language, but to learn a relatively small number of rules which will allow the child to decode from an encoded form (written English) into it’s “clear” form (spoken English).

First you have to learn to break down spoken English into its phonetic components. It is a matter for debate how many separately identifiable phonemes are used in English. The International Phonetic Alphabet identifies 107 basic sounds, most of which are not used in English. In fact it is usually assumed there are about 40 phonemes in American English.

These 40 phonemes then have to be associated with letters and letter groups. There are only 26 letters, but there are many commonly encountered letter groups which have quite distinctive pronunciation in English. Phonics programs usually assume up to about 120 commonly encountered letters or letter groups.

Learning these associations gets you only so far. Written English is basically phonetic, but the rules are very context dependent. Each letter group can have several different ways of being pronounced, and each sound can have different ways of being spelled, all depending on context. This increases the complexity of the task, meaning that the child may need to learn not just 120 new associations but maybe several hundred new associations. And and top of that, there are still individual exceptions which have to be learned as well.

That’s not easy, but it’s still a lot easier than learning 5,000 essentially unstructured and unrelated pieces of information, which is the problem faced by some whole language learners.

Based on the information processing principles outlined above, phonics should help the beginning reader make more rapid initial progress than would be possible with whole-language. The beginner has many fewer pieces of information to learn in order to make a probably correct translation of simple written English into spoken English. And because of the emphasis in phonics on a simple, standard method, we should expect it to have a higher success rate with disadvantaged students, provided of course that they have the basic letter and sound recognition abilities required to use the method.

This kind of analysis is not new, and is widely accepted by cognitive scientists who study language acquisition. It appears to be supported by the best available scientific evidence. Rigorous explications of an information processing model of phonics have recently been developed and tested experimentally by researchers.

This is why most people who study literacy from a cognitive science perspective conclude that phonics should be used as the foundation technique for beginning readers. Phonics gives them the quickest possible start, by leveraging all they already know about spoken English, and giving them some easily learned rules they can use to decode written English. And “quick” wins bring the confidence that reading is something they can make progress on and eventually master. As any parent or teacher knows, confidence is a vital element with young children.

It also seems pretty clear that once reading is well established, motivated readers will start to outgrow the phonics based strategies that got them started in the first place. The “decoding” method works, but it is too slow for experienced readers, who will soon start using whole word recognition anyway.

Therefore some elements of the whole-language approach may be useful in taking children with well established phonics skills to the next levels of achievement.

For all that, the clear conclusion, warranted by both theory and empirical evidence, is that phonics should be taught first, as the foundation technique. This is especially important for children from disadvantaged backgrounds, provided of course they already speak English. There is evidence that it is precisely these children who have been most badly affected by the shift away from phonics during the 1980’s and 1990’s.

It is a sad irony that whole-language educators, while professing their concern for equity, are in practice limiting opportunities for some of the most disadvantaged.